Secrets of Minor-Piece Endings John Nunn
B. T. Batsford Ltd, London
First published 1995 © John Nunn 1995 ISBN 0 7134 7727 X British Library Cataloguing-in-Publication Data. A catalogue record for this book is available from the British Library. All rights reserved. No part of this book may be reproduced, by any means, without prior permission of the publisher.
Ty peset by John Nunn and printed in Great Britain by Redwood Books, Trowbridge, Wilts for the publishers, B . T. Batsford Ltd, 4 Fitzhardinge Street, London W lH OAH
A BATSFORD CHESS BOOK Editorial Panel: Mark Dvoretsky, John Nunn, Jon Speelman General Adviser: Raymond Keene OBE Managing Editor: Graham Burgess
Contents Introduction 1 ltJ+� v lLi 1.1 ltJ+� v lLi 1.2 lLl+b� v lLi 1.3 lL!+c� v lLi 1.4 ltJ+� v lLi 2 ltJ+� v i. 2.1 ltJ+� v i. 2.2 lLl+b� v i. 2.3 lL!+c� v i. 2.4 ltJ+� v i. 3 i.+� v i. 3.1 i.+� v i. 3.2 i.+b� v i. 3.3 i.+c� v i. 3.4 i.+� v i. 4 i.+� v lLi 4.1 i.+� v lLi 4.2 i.+b� v lLi 4.3 i.+c� v lLi 4.4 i.+� v lLi 4.5 Underpromotion 5 2i. v lLi Index of Players and Composers
5 7 8 38 58 74 86 88 106 124 138 151 152 166 181 195 206 207 223 236 248 255 265 282
Important note:
Readers should devote particular attention to this page,be
cause throughout the book I will use two common chess symbols in a non standard way. The symbols concerned are the exclamation mark '!' and the question mark'?'. These symbols are normally bestowed in an arbitrary way by authors; some scatter them lavishly through the text, while others dole them out sparingly. However, in this book these two symbols have precise definitions which depend only very slightly on the author's whim. The exclamation mark '!' after a move means that this move is the only one not to change the result of the position. For example, in a position which is winning for White,
' 1 i.a3 ! ' implies that i.a3 is the only move to preserve the
wm. The use of the word 'only' needs to be clarified. I consider a move to be unique (and therefore deserving of an exclamation mark), even if there are other moves that lose time by repeating the position.
A couple of moves are awarded a '! ! '. This means not only that the move is
unique in the above sense,but also that it is exceptionally beautiful and unex pected. If there is only one legal move, I usually do not award it an '!', except if I wish to emphasise a particular point,for example that there is a long sequence of'only' moves. The use of '?' is easier to explain: a move receives a question mark if it changes the result of the position. Of course a move can only change the result of a position for the worse. By adopting these special meanings,a great deal of information can be pre sented in an economical form. One other point deserves explanation. Beneath each diagram you will find symbols such as'+/-'. Once again this is a method of displaying information compactly. The symbol before the slash '/' gives the result of the position with White to move. This result is given from White's point of view,so that a plus before the slash means that White to play wins,an equals sign means White to play draws and a minus means that White to play loses. The symbol after the slash gives the result with Black to move,from Black's point of view. Thus the symbol above,'+/-',means that White wins whoever moves first. To take an
other example, '= /-' means that White to play draws, but with Black to play White wins. Sometimes we only give the result with one player to move, for example
' I=' means that Black to play draws. Normally this is because the po
sition makes no real sense with the other player to move, for example Black's king might be in check. One final abbreviation is the use of
Endings.
'ECE' for the Encyclopaedia of Chess
Introduction This final book in my endgame trilogy continues the theme of Secrets ofRook Endings and Secrets of Pawnless Endings. Secrets of Rook Endings repre sented an innovation in the world of chess literature; for the first time, the power of the computer was linked to a human interpreter to produce a book in which every move was guaranteed to be accurate, but at the same time the moves were explained in a way comprehensible to average players. Secrets of Rook Endings provoked some strong responses; these were mostly positive, but there were a few players who viewed the production of such a 'Final En cyclopaedia' with dismay, because it meant that a part of chess was forever frozen in silicon. As was inevitable, a very few misprints crept into Secrets of Rook Endings (these will be corrected in a reprint), but basically the promise of the book was kept. The second volume, Secrets of Pawnless Endings, dealt with most of the important four- and five-man endings without pawns. These included such important over-the-board endings as 1'.+J. v J:[ and 'ii'+J. v 'ii'. Two endings with purely minor pieces, J.+i!Ll v lLl and 2J. v i!Ll, were omitted for space rea sons. Although many of the chapters were of great interest for over-the-board players, some sections, for example the one covering 'ii'+i!Ll v 'ii', were primar ily of interest to theoreticians and study composers. Curiously enough this chapter contained some of the most beautiful positions in the whole book, proving that beauty and function don't always go together. This final volume marries aesthetics and utility. The five endings analysed are not only of great practical value, but also offer scope for creativity and in ventiveness. These endings are i!Ll+� v i!Ll, i!Ll+� v J., J.+� v J. , J.+� v lLl and 2J. v i!Ll. I had originally intended including J.+lD v lLl and 2i!Ll v � .but on further examination I decided to exclude these endings. J.+lD v lLl has been omitted due to its almost non-existent practical value, although a small sample of this ending may be found at the end of Chapter 4. The reasons for omitting 2i!Ll v � may be found at the start of Chapter 5 . Each of the first four endings above is in effect an amalgamation of 24 sub-endings, because there are 24 topologically distinct locations for the white pawn (one reviewer of Se crets of Rook Endings wondered what had happened to the e-, f-, g- and h pawns). Please bear this in mind if you think that these endings have received a disproportionately large amount of space. The final chapter on 2J. v lLl also includes some general discussion of end game databases and the human interpretation of them.
6 Introduction These five endings include most of the important five-man endings not covered in the two previous volumes. The only remaining ending of great practical importance, 'ii'+8 v •.will not be covered in this series (again see Chapter 5 !). One special type of position deserves a more detailed explanation, namely the reciprocal zugzwang. In a normal zugzwang, it doesn't matter much who is to move, because the superior side usually has a waiting move with which he can pass if it is his turn to move. This is not so in a reciprocal zugzwang, which may be defined as a position in which whoever moves first has to weaken his position. In an ending such as .i.+8 v .i., where only one side has winning chances, we may be even more precise. Assuming that White has the bishop and pawn, a reciprocal zugzwang is a position in which Black to play loses, but White to play can only draw. In other words, not only is Black in zugzwang if it is his turn to move, but White to play has no waiting move to maintain the zugzwang, so he is also in zugzwang if it is his turn to move. Re ciprocal zugzwang positions often have an importance far out of proportion to their relatively small numbers. The ending of 2i v lLJ would, if it arose in a game, be profoundly influ enced by the 50-move rule. Despite this, I have ignored the consequences of this rule. A discussion of exactly which positions take more than 50 moves to win would take far too long, and would be of little importance in an over-the board situation, where optimal play is the exception rather than the rule. Moreover, FIDE has modified this rule a number of times in the recent past, and a lengthy discussion of such a mutable rule has little point. Finally, I would like to acknowledge the help I have received in writing this book. First of all there are the three CD-ROMs created by Ken Thompson, who not only constructed the databases but also responded to various requests for further information. Secondly comes the invaluable van der Heijden study da tabase. I would also like to thank Graham Burgess, the dedicated Batsford Chess (sorry, Managing) Editor for proof-reading, and Petra Fink for both proof reading and compiling the index, which covers all three endgame books. However, the biggest thank you goes to all the people at ChessBase, for producing ChessBase for Windows and the accompanying CD-ROM access module. The whole book was written using these tools and it is no exaggera tion to say that without the assistance of ChessBase these three endgame books simply would not have been possible. John Nunn London, August/995
1
lZJ+�
v
lZJ
Of the four endings with a ltJ or .t and a pawn against a ltJ or .t, this is the most complex. Knights are tricky pieces at the best of times, but in a simpli fied endgame situation the peculiar move of the knight has a dramatic effect on the play. The move of the knight is more committal than the move of any other chess piece (here I am excluding the pawn). If a bishop moves from d4 to e5, then it retains the possibility of moving to al , b2, c3, f6, g7 or h8 on the following move. True, it has given up the chance to move along the a7-gl di agonal, but it has retained half the options which were previously available. The rook, queen and king also retain some of their previous options when they move. Moving a knight, on the other hand, entails totally abandoning the pos sibilities which were available on the previous move. If a knight can move to a particular square, X say, and it then moves to a different one, the knight is now two squares away from X and there is no route shorter than simply re versing back the previous move. This is an inevitable consequence of the fact that a knight move always changes the colour of the square the knight is standing on. In common parlance, 'You can't lose a tempo with a knight' . As one might expect from this discussion, zugzwangs arise frequently in the ending ofliJ+l!. v ltJ. Of the four endings with minor piece plus pawn v mi nor piece, liJ+l!. v ltJ has by far the highest number of reciprocal zugzwangs a massive 4 1 28 in all. Perhaps because of its complexity, theoretical analysis of this ending has remained at a basic level. Theoreticians have largely con centrated on the simplest cases, namely those in which the pawn is on the sixth or seventh rank. This is a logical course, since one cannot understand positions with the pawn further back until one has laid the foundation by an alysing the positions which might arise after a pawn push. One unhappy con sequence has been that some tricky over-the-board examples have either not been analysed, or have been analysed rather poorly. Whilst there is no space in this book to develop completely new theories for each endgame, I hope to offer a more systematic approach than hitherto. As this book is directed mainly at the practical player, I will concentrate most effort on natural situations of greatest value for over-the-board play. l . l : liJ+al!. v ltJ 1 .2: liJ+bl!. v ltJ 1 .3: ltJ+cl!. v ltJ 1 .4: liJ+dl!. v ltJ
8 38 58 74
8 lLJ+LS
v
lLi
I will adopt this method of dividing up the material in all four of the minor piece plus pawn v minor piece endings. While there is a definite connection between the sections, the four cases are sufficiently different to deserve inde pendent treatment.
1 . 1:
ttJ+� v ttJ
Knights are notoriously bad at fight ing against rook's pawns, so one might guess (correctly) that this case offers the defender the least drawing prospects. Indeed, a far-ad vanced pawn supported by the king is almost always a winning combina tion. We will begin our analysis with the white pawn on a7, and gradually work our way back down the board. The first point to mention regard ing the ending of lD+LSa7 v lLi is that there are some drawing positions even if the defender loses his knight. The first of these is quite obvious, but it can arise from a number of endings and deserves a diagram of its own.
1
=I-
(1): This is a position of recipro cal zugzwang. If it is Black to move, then White wins with 1 ...�c7 2 lLig3 �c8 3 lL!f5 c:Ji;c7 4 lL!e7 (or 4 lL!d6). Black cannot move to c8, so he has to let White's king out of the corner. It is easy to see that with White to play, the position is drawn. White could only win by covering c8, for example, when Black's king is on c7. This requires that the WlL! and B� are on the same coloured squares with Black to move. This is not the case after 1 lL!g3 or 1 lL!t'2, and since both the WlL! and B� change square colour on every move, it can never be the case. A mathematician would call this a parity argument. Although this nor mally refers to the distinction be tween odd and even, the dichotomy between dark and light squares is equivalent. Parity rears its ugly head several times in the ending of lLJ+[!, v lLJ. Note that the initial position of White's knight is unimportant - only the square colour matters. If the knight starts on a light square, then White to play draws; if it starts on a dark square, then White to play wins. (2): This is the other key position without a black knight. White cannot make progress, with or without the move. Black oscillates with his king between b7 and a8; if White ap proaches with his king, then Black is
tiJ+� v tjJ 9
tiJa7 is drawn, no matter how far away the black king is. The following position features a white knight sacrifice.
2
=I=
stalemated, while otherwise White cannot improve his position. Note that White's knight could be on b5 or c8 without making any difference to the assessment. These positions are relevant to the ending of tiJ+�a7 v tiJ, because they imply that in certain situations Black may even sacrifice his knight. Dia gram 2 also emphasises how b7 is a key square. If Black's king can oc cupy b7, the result is a draw, even without a black knight, while if White's king can settle on b7, he is virtually guaranteed a win, even with a black knight! Note that these drawing positions are only valid with the pawn on a7; move them down a rank, so that the pawn is on a6, and White wins who ever moves first. It is not only Black who has the chance to sacrifice his knight. The pawn on a7 is unique in that it is the only pawn position which can win against a blockading enemy knight (in the absence of the enemy king). For example, the ending �+�a6 v
/F. Richter, 1960 3rd Pr., Ceskoslovensky Sach 3
(3): For the sake of clarity, I have taken the position after two moves of Richter's study. Black must play l . . .tiJe5 ( l ...tiJf4 2 tiJf7 tiJg6 3 �d3 wins by marching to f5), but White's next move is far from obvious. The natural idea is to attack the e5-knight with the king, but if White plays his king to d4, Black draws by ... tiJf3+ and ... tiJg5. The square e3 is also un available, on account of ...tiJg4+ and . . .tiJf6, so Black's knight is actually creating a barrier that runs b6, c6, c5, c4, d4, d3, e3, f3, f2. Trying to sneak round the edge of the barrier is sim ply too slow, as Black's king can race towards the h-pawn by a direct route. The solution is the surprising 2 tiJg6 ! (White mustn't delay, since 2 �c3? �a2 ! 3 tiJg6 tiJxg6 ! 4 �d4 �b3 ! 5 �e4 �c4 ! 6 �f5 tiJh8 ! 7 �f6 �d5 !
8 �g7 'it.>e6! draws; 2 l'Df7?t'Dxf7! is also wrong since there is a new bar rier running e4, e5, e6, d6) l'Dxg6 3 c,iid3! (the squares e4 and f5 are both safe, so Black cannot erect a barrier) �b2 4 �e4 ! �c3 5 'it.>f5 ! l'Dh8 6 c,ilf6! 'it.>d4 7 c,ilg7! and Black is one tempo too slow. Other composers have tried to show similar knight sacrifices, but with less success, for example Gur genidze won a Special HM (Sha1ch maty v SSSR, 1 97 1 ) for a study which reaches the position W�g2, l'Da8, l!,a7 v B�bl, l'De2 after three moves. Gurgenidze's analysis con tinued 1 ...l'Df4+ 2 �f3 l'Dd5 3 l'Db6! l'Dxb6 4 �e4 'it.>c2 5 'it.>d4! 'it.>b3 6 �c5 and so on. Quite apart from be ing anticipated by Richter, Gurgen idze's study permitted an alternative win by 2 �f2 l'Dd5 (2...l'De6 3 �e3 �c2 4 �e4 l'Dc5+ 5 c,iid5 l'Dd7 6 �d6) 3 l'Db6! l'Dc7 (3 ... l'Dxb6 4 �e3! 'it.>b2 5 �d4! �b3 6 �c5! l'Da8 7 �c6 wins as in Richter's study) 4 'it.>e3 'it.>c2 5 c,iie4 �b3 6 �e5 �a3 7 l'Dd5 l'Da8 8 'it.>d6 c,ii a4 9 �c6 fol lowed by �b7. Halberstadt produced an interest ing, if unsound, study based on a re lated idea. (4): Halberstadt gave l �f5 and now: 1 ) 1. 'it.>e3 2 �e6! (a surprising finesse, but 2 �e5? ltJb6! reaches a reciprocal zugzwang with White to move; after 3 l'Dc2+ �d3! 4 t'Db4+ �c4! 5 l'Dd5 l'Da8! 6 �d6 �b5! 7 l'Dc7+ �b6! White loses his pawn, while 3 �e6 �d4 allows Black to approach with his king) l'Db6 (Black ..
+I=
4
Halberstadt, 1951 2nd HM, L'ltalia Scacchistica must prevent �d6, so this is forced) 3 �e5! and now Black finds himself on the wrong end of the reciprocal zugzwang. 3... l'Da8 loses to 4 l'Dc2+ �d3 5 �d5 heading for b7, while 3 ...�d2 4 �d4! followed by 'iti>c5 wins. 2) 1. �xel 2 �e6 (the simplest win; Halberstadt gave 2 �e4 l'Dc7 3 �d3 �dl 4 �c4!, which is also ef fective) l'Db6 (forced, or else �d6-c6 wins) 3 �e5 (White has to avoid the d5-d6-d7-e7 barrier) �d2 (by mov ing to b6, Black has given up the op tion of setting up a different barrier with ... l'Dc7) 4 �d4! followed by �c5 wins. 3) 1. �d2 2 �e5 l'Dc7 3 l'Df3+ �c3 4 l'Dd4 ! ltJas 5 �d5 ! �b4 6 �c6! �a5 7 �b7! t'Db6 8 ltJc6+ �b5 9 l'De5 is fairly easy. The dual in line 2 is one problem with Halberstadt's study, but a far more serious flaw is that White can win more simply with 1 �f7 �e3 (1 ...�xel 2 �e7 l'Db6 3 �d8 and 4 ..
••
�c7) 2 �e7 �d4 3 �d6 �e4 (or 3 . . .�c4 4 �c6) 4 tbc2 �f5 5 tbb4 and tbd5-c7. Despite this, Halberstadt deserves credit for discovering the reciprocal zugzwang in line 1 . Vitaly Halber stadt (1 903-67) was a specialist in endings of minor piece and pawn v minor piece, and we will meet his name many times in this and later chapters. Perhaps this was a rather unfortunate example to start with, because it was one of the few cases in which his analysis contained ma jor errors. He was normally one of the most accurate endgame analysts. Of course, one is not forced to ac cept a sacrifice:
Shaigarovsky, 1991 Tidskrift for Schack (5): White cannot prevent Black's pawn advancing to h2, but then the real battle starts: 1 tbf7 ! h3 2 tbg5 ! h2 3 tbe4+ ! �d3 (3 . . . �d4 4 tbf2 tbc5 5 �f5 ! tbd3 6 tbh 1 ! �e3 7 �g4 ! is a straightforward draw) 4 tbg3 ! (not 4 tbf2+? �e2) tbc5 5
�f5 ! (5 �e5? loses to 5 ... tbe4 ! 6 tbh 1 �e3 ! 7 �f5 �f3, followed by . . . tbg3) tbe4 6 tbh1 ! (after 6 tbxe4? �e3 ! 7 tbg3 �f3 ! 8 tbh 1 �g2 ! White's king is too far away from f2) tbf2 (an excellent winning try; 6 ...�e3 7 �g4 ! tbf2+ 8 �g3 ! draws as in the note to Black's third move) 7 tbg3 ! (offer declined! 7 tbxf2+? �e2 ! 8 tbh 1 �f3 ! is winning for Black) and Black is unable to make progress. His king cannot approach, and after 7 ... tbe4 White just returns to h l . The following position shows typical strategy when both kings are near the pawn:
+I= 6 Kralin, 1973 (end of study) Bth HM, Kommunist
(6): Black to play can draw by 1 . . .�g5, so suppose that White moves first. He wins by 1 �f7 ! �g5 2 tbf6 ! (forced, since 2 tbg3? �h6! 3 �g8 tbe7+ ! is an immediate draw) �h6 (if Black were to play he would be in zugzwang, so White must lose a move) 3 �g8 ! (no other move will
do; Black must be kept out of g7) CiJe7+ 4 �f8 ! l:iJg6+ 5 �f7 ! (White has lost a tempo while all the time maintaining control of g7) l:iJe5+ 6 �g8 ! l:iJg6 (Black's moves are all forced) 7 l:iJg4+ �g5 8 CiJe5 and the pawn promotes.
chosen a few of the most interesting ones.
=I-
8
Original
7
+I= Chernev, 1961 Practical Chess Endings
(7): Black is threatening perpet ual check by . . . CiJd7+ and ... CiJb6+, so White's first move is forced: 1 CiJf6 ! (this is a position of reciprocal zugzwang with Black to play) CiJa8 ( l . . . ..ti>c5 2 �b7 �b5 3 CiJd5 wins at once) 2 CiJd5 (the quickest method, although there are other winning lines; White should avoid 2 ..ti>xa8? �c7 ! drawing as in diagram 1) ..ti>d7 (Black's knight is paralysed, so he must prepare to counter ..ti>xa8 by . . . ..ti>c8) 3 �b7 �d8 (3 ... ..ti>d6 4 ..ti>xa8) 4 CiJb6 CiJc7 5 �c6 and wins. We have already encountered two positions of reciprocal zugzwang; there are 2 1 4 reciprocal zugzwangs in the ending of CiJ+�a7 v CiJ. I have
(8): Firstly suppose that Black is to play. After l . . ...ti>d6 ( l .. .CiJc7+ 2 ..ti>b6 ! ..ti>d6 3 ..ti>b7 ..ti>d7 4 l:iJc3 fol lowed by CiJd5 wins for White) 2 CiJb4 (a second position of reciprocal zugzwang) 'iti>d7 (2 . . . 'ifi>c7 3 'ifi>a6! ..ti>d6 4 ..ti>b7 ..ti>d7 5 CiJd5 'iti>d8 6 CiJb6 l:iJc7 7 'ifi>c6 is a win we have seen be fore) 3 CiJd5 'iti>d6 (3 . . . ..ti>c8 4 �c6) 4 CiJf6 (a third reciprocal zugzwang) Wc7 (4 ...CiJc7+ 5 ..ti>b6! llia8+ 6 ..ti>b7 is an immediate win) 5 ..ti>a6 ! (re ciprocal zugzwang number 4) �c6 6 CiJe8 ! (reciprocal zugzwang num ber 5 - note that White's move was forced in order to prevent ... CiJc7+ and . . . ..ti>b7) CiJb6 7 l:iJg7 (number 6) CiJa8 (or 7 ... ..ti>c7 8 CiJf5 CiJa8 9 CiJe7 ..ti>d6 10 ..ti>b7 !) 8 CiJe6 (number 7 this is the key position, because now Black is forced to move his knight at a moment when White can respond with a check) CiJb6 9 l:iJd4+ �c7 10 CiJf5 (number 8) CiJa8 (or 10 ... 'ifi>c6
ll'i+,0, v lb 13
1 1 ll'ie7+ �c5 1 2 �b7) 1 1 ll'ie7 and reciprocal zugzwang number 9 fi nally finishes Black off. He cannot play l l . . .ll'ib6 because of 1 2ll'id5+, while any king move allows 'it>b7. The position at move 5 is basi cally a parity situation. Black can either oscillate with his knight, or play his king up and down between c6 and c8. Neither of these options allows him to lose a tempo. White is also severely constrained; he cannot move his king at all, and his knight, by itself, cannot lose a tempo. In this situation the result of the game de pends largely on whether the tempi are odd or even (in this particular case there is the additional tactical point that White must be able to pre vent . . .ll'ic7+ when Black's king is on c6). With White to play the situation is familiar. White can reach all the re ciprocal zugzwang positions aris ing in the above analysis, but with the wrong player to move. A sample line is 1 ll'ib4+ ( 1 'it>a6 'it>c6 ! 2ll'ib4+ 'it>c7 ! and 1 'it>a5 'it>c6 2 'it>a6 ll'ic7+ are no better) 'it>d6 ! 2 'it>a6 'it>c7 ! 3 ll'id5+ 'it>c6! 4 ll'ie7+ 'it>c7 ! 5 ll'if5 'it>c6 and White is not making pro gress. The following game was decided by the reciprocal zugzwang from the previous diagram. (9): White was to play in the game and, as we known from diagram 8, the result depends on parity. In this case the parity operates in White's favour, and Black should not have been able to reach a winning zug zwang. The game continued 1 'it>f3
9
=I+ Purtov-Mikhalevski Budapest 1 993
ll'ie5+ 2 'it>f2 ! ll'ig6 3 'it>f3 ll'ih4+ 4 'it>f2 ! 'it>g4 (Black realises that he cannot make progress using only his knight, so he tries to triangulate with his king in order to lose a tempo) 5 ll'ig3 (the simplest draw is by 5 'it>e2; White can simply oscillate with his king between e2 and f2, meeting . . . 'it>g4 by 'it>e3 and . . . �h3 by �f2) 'it>g5 6 ll'ihl �f4 7 ll'ig3 (it wasn't too late to adopt the afore mentioned plan with 7 'it>e2) �g4 8 ll'ihl ! 'it>h3 (Black has succeeded in his ambition of losing a tempo with his king; in fact the position is still drawn for tactical reasons, but White didn 't realise the danger) 9 ll'ig3? (9 'it>fl ! was the only move; thanks to the awkwardly placed h4-knight, Black cannot prevent an annoying check on f2, e.g. 9 . . .ll'if5 10 lt'if2+ ! 'it>g3 l l ll'ie4+! �f3 1 2ll'id2+)ll'ig6 (now the parity is different and it is White who ends up in zugzwang) 10 ll'ihl (10 'it>f3 ll'ie5+ 1 1 'it>f2 ll'ic4 1 2 lt'ih 1 ll'id2 i s reciprocal zugzwang)
14 lb+.0.
v
lb
lbe5 1 1 lbg3 lbf7 1 2 lbhl lbg5 (an other reciprocal zugzwang) 1 3 �el ( 1 3 lbg3 lbe4+) �g2 14 lbf2 lbe4 1 5 lbhl �f3 0- 1 . The fundamental nature of the re ciprocal zugzwangs given in dia gram 8 becomes clearer if one looks at a couple of positions with the kings further away from the pawns.
move, for example after 1 �b2 �d2 ! 2 �b3 �d3 ! 3 �b4 �d4 ! 4 �b5 (4 lbf6 �e5) �d5 ! 5 lbf8 �d6 ! 6 �a6 �c7! drawing. Our final position with lb+.0.a7 v ttJ is particularly subtle.
+I=
11
Original =I-
10
Original (10): It turns out that this position also depends largely on parity. With Black to play, l . . .�d2 ( l . . .lbc7 is weaker, as White can win directly by 2 lbf6 �d2 3 �b2 �d3 4 �b3 �d4 5 �b4 �e5 6 �c5 lba8 7 �c6) 2 �b2 ! �d3 3 �b3 ! �d4 4 �b4 ! �d5 5 �b5 ! �d6 6 lbf8 (it is essential to prevent Black losing a move by . . . �d7-c7; it follows that 6 lbf6 also wins, but 6 lbg5? only draws after 6 . . . <;j;>d7 !) �c7 (6 ... �d5 7 �a6 �c6 8 lbe6 ! transposes) 7 �a6! �c6 8 lbe6 ! leading to diagram 8 after White's 8th move. With White to play, the critical zugzwangs all arise with White to
(11): White must move his knight is such a way as to prevent the imme diate march of Black's king towards the a-pawn, but the right square isn't immediately obvious: 1 tt:lc3 ! (not 1 tt:lf6? tba8 ! 2 �d2 �f4 ! 3 �d3 �e5 and the king gains a vital tempo, nor l tt:lc5? �e3 2 �dl tt:la8 !, followed by ...�d4; note that 1 tt:lc5 ? tt:la8? is a mistake, since 2 �d2 ! heads off Black's king and wins after 2 ...�f4 3 �d3 �e5 4 �c4 �d6 5 �b5 ! �d5 6 tt:ld7 �d6 7 lbf8 winning as in dia gram 8) tt:la8 ( l ...�e3 and l . . .�f4 lose to 2 tt:ld5+, so this is forced) 2 �d2! <;j;>f4 3 �d3 ! �e5 4 �c4 ! �d6 5 �b5 ! �d7 (an excellent defensive try; after 5 . . .�c7 6 �a6! the recipro cal zugzwang works against Black, and he loses after 6. . . �c6 7 tt:lb5 !
lb+.0. v lb 15
lbb6 8 lbd4+ �c7 9 lbc2 lba8 10 lbb4 �c8 1 l lbd5 ! ; the move played is an attempt to lose a tempo by tri angulation) 6 llJd5 ! (White's favour able knight position enables him to thwart Black's plan; 6 �a6? �c7 ! would turn the reciprocal zugzwang against White) �d6 (if White's knight were not controlling c7, then Black would regain the parity by 6 . . . �c7, but this is impossible when White's knight is on d5 or a6) 7 lbf6 (the same idea is repeated; Black would like to move to d7, but this is impossible) �c7 8 �a6 ! (the parity is in White's favour and he wins as in diagram 8) �c6 9 lbe8 ! lbb6 10 lbg7 lba8 1 1 lbe6 winning. Moving the pawn back to a6, we reach the most important theoretical positions in the ending of lb+� v lb. Reti established the basic princi ples, and since his time there have been few developments. First of all, let's look at Reti's analysis. (12): White wins quickly if Black is to move: 1 . . .�b4 2 �b6 �c4 (2...�a4 3 lbc3+) 3 lbc3 lbd6 4 �c7 lbe8+ 5 �c6 and the pawn advances. With White to play the win is more complex, and involves transferring the move to Black: 1 lbc5 ! �b4 ( l ... lbd6+ 2 �c7 lbb5+ 3 �c6 trans poses to the main line) 2 �b6 lbd6 (2 . . .�c4 3 lbe4 �b4 4 lbc3 lbd6 5 �c7 wins; note the trick 4 lbc3 - this is a very common idea in lb+� v lb and we will see this 'double deflec tion' idea time and time again) 3 lbe4 (Reti's win is not the quickest, but it is certainly the simplest) lbc8+
. . . -�· . . �- . . . -·· . . • ·ltj- . . . • • • • . . . +/-
12
Reti, 1929 Shakhmaty (3 . . .lbc4+ 4 �b7 �b5 5 a7) 4 �c7 �b5 (4 . . .lba7 5 �b7 lbb5 6 �b6 �c4 7 lbc3 is again a double deflec tion; 4 ... �a5 5 lbc5 ! lba7 6 �b7 ! lbb5 7 lbe4 transfers the move to Black) 5 �b7 �a5 6 lbc5 ! lbd6+ 7 �c7 lbb5+ 8 �c6 lba7+ (8 . . .�b4 9 lbe4 �a5 10 �b7 is the same) 9 �b7 ! lbb5 10 lbe4 (now White has achieved his aim and Black loses quickly) �b4 1 1 �b6 �c4 1 2 lbc3 winning. But White doesn't always win when he is supporting the pawn with his king: (13): White to play wins imme diately by 1 �b8, so suppose that Black moves first. He can draw by very precise play: 1 ...�a5 ! (the only move; 1 . . .�c5? 2
b8 and 1 ... lbc4? 2 �b7 lba5+ 3 �b6 lbc4+ 4 �c6 lba5+ 5 �d7 lbc4 6 lbd5+ are los ing) 2 �b8 lbb5 ! 3 �b7 lbd6+ ! 4 �a7 (White has transferred the move to Black; is it zugzwang?) lbf7 ! (the point is that after any other
16 lb+8
v
lb
Reti's positions have wide-rang ing application in ltJ+a8 v ltJ end ings, and many positions eventually reduce to them. The following posi tion has a longer solution, but it is not especially complicated if we bear the above principle in mind.
13
+I=
Reti, 1929 Reti's Endspielstudien knight move, for example 4 ... lbc4, White wins by 5 �b8 ! lbe5 6 �b7 !; the only good reply to 'iPb8 is ... lbd8, which both prevents �b7 and stops a7 on account of . . . lLic6+) 5 lLie6 (5 �b8 liJd8 !) 'iPb5 (not 5 . . . liJd6? 6 lLic5 ! lLib5+ 7 �b7 ! , winning as in diagram 12, but Cheron correctly points out an alternative draw by 5 . . .lLie5 6 �b7 lLic6 ! 7 lLic5 �b5 ! 8 lLib3 liJd8+ !) 6 liJd4+ �a5 ! 7 lLlc6+ �b5 ! 8 lLib4 liJd8 ! (reciprocal zug zwang with White to play) 9 �b8 (knight moves are also met by . . .lbc6+) lLic6+ ! drawing. The key difference between these two positions is that in diagram 1 2 White was able to maintain his king on b7, whereas in diagram 13 Black could always meet by 'iPb7 by a knight check, driving the king onto an inferior square. This provides us with the useful general principle that a pawn on a6 supported by a stable king on b7 virtually guarantees a win.
+I=
14
Original (14): Black to play draws by 1 . . . �d5 2 �b5 �d6, followed by . . . �c7, so we suppose that White moves first. 1 'it>b5 (not 1 lLic4? lLia7 ! 2 lLib6 lLic6+! and the king cannot advance to b6; note that if White does not play 2 lLib6, then Black can continue with 2... �d5) �d5 ( l ...lLia7+ allows a simpler win by 2 �b6 ! lLic8+ 3 �c7 lLia7 4 �b7 lLib5 5 lLia4 liJd6+ 6 �c6 lLic8 7 lLib6 lLia7+ 8 �b7 lLib5 9 lLia8 lLid6+ 10 'iPb8) 2 lLic4 ! (preventing . . . �d6) �d4 (there is nothing better; after 2 . ..lLia7+ the king advances) 3 lLia5 ! (White finds the only way to improve the position of his knight; after . . .'iPd5 White must be ready to cover d6 again)
lli+� v lli 17
�d5 (once again, a check makes life easier for White: 3 ... llid6+ 4 'it>c6 llic8 5 'it>c7 �c5 6 llic4 llia7 7 'it>b7 ! llib5 8 llia3 ! llid6+ 9 'it>b8 llic8 10 llic4 llie7 1 1 'it>b7) 4 llib7 �d4 (if now 4 . . .llia7+, then 5 'it>b6 ! llic8+ 6 �c7 llia7 7 llid8 llib5+ 8 'it>b6 llid6 9 llic6 �e6 transposes into the main line) 5 �c6 llia7+ 6 'it>b6 llic8+ 7 �c7 llia7 8 llid8 �d5 (8 ... 'it>c5 9 llie6+ 'it>b5 10 �b7 ! llic8 1 1 llic7+ loses at once) 9 'it>b7 llib5 10 'it>b6! llid6 ( 1 0 ... �c4 1 1 lllc 6 llid6 12 llie7 llib5 1 3 llif5 �b4 14 llid4 llid6 15 �c7) 1 1 llic6 'iite6 ( 1 1 . . .llic4+ 12 �c7) 1 2 �c7 (the key position White wins by a king triangulation) llib5+ 1 3 'it>b7 �d7 ( 1 3 ... �d6 14 llid4 llic7 1 5 a7 is a simple win) 14 llie5+ �d6 (or 14 . . . 'it>d8 1 5 'it>b6 llic7 16 a7 �c8 17 'it>c6 ! llia8 1 8 llic4) 15 'it>b6 llic7 16 a7 llia8+ 17 �b7 ! llic7 1 8 llic4+ 'it>d7 1 9 llib6+ �d8 20 �c6 and White finally wins. There are 230 positions of recip rocal zugzwang in the ending of lli+�a6 v lli. Here is one of the most interesting. (15): At first this doesn't look at all like a position of reciprocal zug zwang, but Black to play can't keep White's king out of the corner per manently: 1 . . .'it>c7 ( 1 . . .'it>c8 2 'it>c6 ! �b8 3 llib5 ! 'it>c8 4 a7 'it>d8 5 'iitb7 �d7 6 llic3 �d8 7 llid5 wins, while l ...llic7 2 a7 ! llia8 3 llib5 transposes to the main line) 2 llib5+! 'it>d7 (2 . . . 'it>c8 3 'iitc6 �b8 4 a7+! �c8 5 llid6+ and 2 ...'it>b8 3 �c6! 'it>c8 4 a7 are straightforward) 3 a7 (this is an other position of reciprocal zug zwang) �d8 (there is nothing better)
=I-
15
Original 4 'it>c6 �c8 5 llid6+ 'it>d8 6 llic4, fol lowed by llib6, and White wins. White to play has nothing better than the line 1 llib5 ( 1 llic4 'it>c7 2 �b5 'it>b8), but after l . . .llic7 ! 2 a7 llia8 ! we have arrived at the above reciprocal zugzwang with White to move; after 3 �d5 llib6+ ! the draw is absolutely clear. Moving on to the case with the pawn on a5, the following position is just diagram 12 shifted down a rank. White wins with or without the move, so it is a target position for White. The analysis is very similar to diagram 1 2, but owing to its funda mental importance I will give full details. (16): In fact it is much easier to win when Black moves first; White to play must start with a nine-move procedure designed to lose a tempo: 1 llic4 ! 'it>b3 (after 1 . . .llid5+ 2 �c6 llib4+ 3 �c5 we have transposed into the main line at move seven) 2 �b5 ! (if White moves his knight,
18 ttJ+8 v ttJ
. . ·�� .
• (f� n �
� �
• ;/:;&; ��
�t ?;::%
;'� /�
.
w<
• .•�tt /?!:?W • /,/,///,
•
•M�� w��1};�;;;:;?�w� m �� r:r�� �� �.;7 �0J:t ���·. . .. � ;�
. •
�J?;
� (!{:
:�;;
-
;/� •
,�:;w
�;;· �1t
?�
16
. •
• •
+!-
Original then Black replies 2 ... �a4, threaten ing 3 . . . llld 5+, and the knight has to return to c4 to defend the pawn) llld5 (2 . . . 'ifi>c3 3 llle3 �b3 4 lllc 2 llld5 5 'it>c6 llle7+ 6 �b7 lllf5 7 a6 llld6+ 8 'it>c6 lllc 8 9 cJitc? llla7 10 llld4+ �a4 1 1 rJitb7 is winning; note the double deflection at move 4) 3 llle5 (threat ening llld 3, followed by �c6, when . . . 'iti>a4 may be met by a6 without al lowing . . . lllb4+) lllc 7+ (3...lllc 3+ 4 �c5 'ifi>a4 5 a6 lllb5 6 �b6 reaches a position we analysed earlier) 4 �b6 llld5+ (4...�a4 5 lllc4 ! llld5+ 6 �c6 lllb4+ 7 �c5 transposes) 5 �c6 �a4 (after 5 . . . lllb4+ 6 'iPb5 llld5 White has gained time to execute his threat: 7 llld3 lllc 3+ 8 �b6 llld 5+ 9 cJitc6 llle7+ 10 �d7 llld5 1 1 a6 and White wins) 6 lllc4 ! lllb4+ 7 �c5 llla6+ (7 ... llld3+ 8 �b6! lllb4 9 llle3 trans poses, while after 7 . . . �b3 8 llle3 Black has the choice between trans position to the main line by 8 ... �a4 9 'it>b6 or losing to a double deflection after 8 . . . �a3 9 �b5 �b3 10 lllc 2 see the note to Black's second move)
8 �b6 ! lllb4 9 llle 3 (finally White has transferred the move to Black; the rest is relatively easy) �b3 10 �b5 �c3 11 lllc 2 (the double de flection also works when Black's king is on c3) llld5 12 �c6 llle7+ 1 3 'iii>b7 (now the pawn advances) lllf5 14 a6 llld6+ 15 �c6 lllc 8 1 6 �c7 llla7 17 llla3, followed by �b7, and White wins. Reti' s diagram 13 is virtually un changed when it is shifted down by a rank, giving us the position W�a6, lllc6, /ja5 v B�b3,llld 5. If Black is to play, he draws by 1 . . .'iPa4 ! 2 �b7 lllb4 ! 3 �b6 llld 5+! 4 rJita6 lllf6 !, just as before. Many other position can be solved using the analysis of diagram 16.
+I=
17
Averbakh, 1980 (17): Black threatens to draw by 1 . . .llld5+ and 2 ... lllb4+, so White's first move is forced. After 1 lllf4 ! (in fact this is a reciprocal zugzwang) �b3 2 �b5 �c3, Averbakh gave a win starting with 3 llld3 (a slightly different double deflection), but, in
tt:J +.0, v tt:J 19
view of our earlier analysis, the sim plest line is 3 lt:Je2+ 'it>b3 4 lt:Jd4+ �c3 5 lt:Jc2 transposing to the pre vious diagram at move 1 1 . The conclusion is very much the same as with the pawn on a6. If Black's king is behind the pawn, then a formation with the pawn on a5 and king on b6 is winning, except if Black can immediately check the king away. Sometimes, however, ac curate play is necessary.
18
+I= Cheron, 1955 Journal de Geneve
(18): Black to play draws by l . . .�f6, so suppose that White is to move: 1 �g6! lt:Je7+ ( l . ..lt:Jh4+ 2 'it>g7 lt:Jf5+ 3 'it>f7 transposes, while l . . .'it>e6 2 lt:Jg3 lt:Je7+ 3 'it>g7 ! forces the immediate advance of the pawn) 2 'it>g7 ! lt:Jf5+ 3 'it>f7 ! and now: 1 ) 3 lLJd6+ 4 �g6! lt:Jf5 5 lt:Jg3 ! (forcing Black's knight away from the h6-square) lt:Jh4+ 6 'it>g7 lt:Jf3 7 h6 (we have now reached a situation analysed earlier in this section) lt:Jg5 8 lt:Je2 lt:Je6+ 9 'it>f7 lt:Jg5+ 10 'it>g6 ...
lt:Je6 l l lt:Jgl lt:Jf8+ 1 2 �g7 ! lt:Je6+ 13 'it>g8 and the pawn advances. 2) 3 'it>d6 (3 . . . 'it>d5 is met the same way) 4 lt:Jg3 lt:Jh6+ 5 'it>g6 ! (5 'it>g7? lt:Jg4 ! is too slow, as it takes two moves to force the knight away from g4) lt:Jg4 6 �g5 lt:Je5 7 �f5 00 8 �f6 ! lt:Jh6 9 lt:Je4+ �c6 (9 . .�d5 10 �g6) 10 'it>g6 lt:Jg4 1 1 lt:Jf6 lt:Je5+ 12 'it>g7 and wins. 3) 3 tt:Jh6+ 4 �g6 ! lt:Jg4 (or 4 . . . lt:Jf5+ transposing to line 1) 5 'it>g5 ! lt:Jf6 6 h6 ! lt:Jh7+ 7 �g6! (the seventh 'only move' in a row) lt:Jf8+ 8 'it>g7 (now we are in familiar terri tory) 'it>f5 9 lt:Jg3+! �g5 1 0 lt:Je4+ ! 'it>f5 1 1 lt:Jd6+ �g5 1 2 lt:Jf7+ and White can safely take the knight. Before moving on to a couple of original positions, here are two prac tical examples of this ending. The following position demonstrates that Black has better drawing chances when his king is to the side of the pawn, as opposed to being behind the pawn. The reason is that in some variations Black has the chance to play his king in front of the pawn, and the necessity of preventing this restricts White's options. (19): I.. .'it>e6! (not l . .. �e4? 2 lt:Je7 lt:Jg7 3 h6 ! lt:Je6+ 4 �f6 lt:Jf8 5 lt:Jc6 tt:Jh7+ 6 'it>g7 lt:Jg5 7 lt:Jd8 'it>f5 8 lt:Jf7 lt:Je6+ 9 'it>g8 ! wins for White) 2 lt:Jf4+ 'it>e5 ! 3 tt:Jd3+ 'it>e6 ! (after 3 . . . 'it>e4 4 lt:Jf2+ 'it>e5 5 lt:Jg4+ ! we have transposed into the game) 4 lt:Jf2 (White cannot make any pro gress by checking, so he tries to improve the position of his knight; however, this gives Black time to re lease his king) 'it>e5? (a surprising er••.
.
...
20 lb+l!. v lb
I=
19
Barcza-Hecht, 1962 ror, as 4 ... lbe7 intending 5 ... c:j;f7 is a simple draw) 5 lbg4+ ! c:j;e6 6 c:j;g6 ! (this is a reciprocal zugzwang; note that White could not have played c:j;g6 at move 4, because after check ing White's king to f8 Black would be able to play ... c:j;f6) lbh4+ 7 c:j;g7 ! lbf5+ 8 c:j;f8 ! (a second reciprocal zugzwang) c:j;d6 (8 ...�d7 loses simi larly : 9 c:j;f7 c:j;d8 IO lbe3 lbd6+ 1 1 c:j;g6 c:j;e7 1 2 h6 ! c:j;f8 1 3 h7 ! lbt7 14 lbf5) 9 c:j;f7 c:j;d7 10 c:j;f6 lbe7 1 1 lbe3 c:j;d6 (or 1 1 ... c:j;e8 1 2 c:j;g7 ! lbc6 1 3 h6) 1 2 c:j;f7 ( 1 2 rJ;;g7 was faster, but this is good enough) lbc6 1 3 lbc4+ c:j;d7 14 h6 lbd8+ 1 5 c:j;f6 1-0. Generally speaking, if Black's king is in front of the pawn then the result is a draw. However, so helpless is the knight when confronted by an a-pawn that there are a few excep tional positions which are lost. (20): In the game it was White to play, and he won as follows: 1 lba6+ ! c:j;c8 ( 1 . . .c:j;a8 2 lbc7+ ! c:j;b8 3 a6 and the pawn runs through) 2 lbc7 ! c:j;d7 3 a6 lbc6 4 lbd5 c:j;d6 5
20
+I= Scheve-Estorch, 1 905
lbe3 c:j;d7 6 lbc4 lbb4 7 a7 ! lD
llJ+8 v lLl 21
=I-
21
Original lLlc3 'it>e5 9 'i1o>b5) 3 l2Jd3 ! 'i1o>f3 4 lLlb4 ! 'i1o>e3 5 'i1o>e5 ! 'i1o>d2 6 'i1o>d4 �e2 (Black's knight is miserably placed) 7 'i1o>c5 and wins. B2) 1...lLlc7 2 lLlf2 'ittg3 (2...lLla6 transposes to line B1) 3 lLld3 'ittf3 4 'i1o>e5 'i1o>e3 5 lLlb4 wins as in line B1. B3) 1...lLlb4 2 lLlf2 'ittg3 3 l2Jd3 and Black cannot take the knight, so White gains a tempo just as in the above lines. B4) 1 ...'i1o>h5 (after this Black cannot meet lLlf2 by ...�g3) 2 lLlf2 'i1o>h6 3 lLld3 'i1o>g7 (the king tries to find another way to the queenside) 4 1r>e6 lLlc7 + 5 d7 lLlb5 6 a6 'ittf6 7 lLlb2 'i1o>e5 8 'i1o>c6 lLla7+ 9 'ittb7 lLlb5 10 'i1o>b6 lLld6 11 lLlc4+ lLlxc4+ 12 'ittc6 winning. B5) 1...'i1o>h3 (the trickiest vari ation, because 2 lLlf2+? can be met by 2 ...'it>g3 as in line Wl below) 2 'ittf4 ! (this is the flaw with ...�h3 Black cannot meet 'i1o>f4 by ...'i1o>h5) 'i1o>g2 (or 2...lLlb4 3 lLlf2+ g2 4 l2Je4 fl 5 lLlc3 lLla6 6 'i1o>e5 'i1o>f2 7 'i1o>d5 'i1o>e3 8 'i1o>c4 'ittf4 9 b5 lLlb8 10 'i1o>b6
lLld7+ 11 �c6 lLlb8+ 12 �b7 lLld7 13 lLla4, followed by a6, and White wins) 3 lLlg3 ! 'i1o>f2 4 lLlf5 (the knight escapes just in time; now White can keep Black's king trapped on the second rank) 'i1o>e2 5 'i1o>e4! 'i1o>d2 6 'i1o>d4! 'i1o>c2 7 'i1o>c4! 'i1o>b2 8 'i1o>b5 lLlc7+ 9 'itc6 lLla6 10 Wb6 lLlb4 11 lLld4 'i1o>c3 12 lLlc6 lLld5+ 13 'i1o>b7 lLlf6 14 a6 and wins. W1) 1 'it>e5 (the immediate king march to the queenside fails because the h I-knight is in danger of being trapped) �g4 (but not l...g5? 2 lLlf2 'i1o>g6 3 �d6 'i1o>f7 4 'i1o>c6 'i1o>e8 5 'i1o>b6 lLlb4 6 lLld3) 2 'i1o>d5 (2 lLlf2+ 'i1o>f3 3 lLld3 'i1o>e3 is line W3) 'i1o>f3 (2...'i1o>f4? 3 lLlf2! 'i1o>e3 4 lLle4 pre vents Black's king moving to d3 and thereby wins) followed by ...�g2 and ...'i1o>xhl, drawing because the knight can stop the pawn by itself. W2) 1 �f4 (this allows Black's king to slip away via the back rank) 'i1o>h5! 2 lLlg3+ (or 2 lLlf2 'i1o>g6 3 'i1o>e5 'i1o>f7 ! 4 'i1o>d6 e8 ! 5 l2Jd3 'i1o>d8 ! 6 'i1o>c6 'i1o>c8 7 'i1o>b6 lLlc7 with a draw) 'i1o>g6! 3 'i1o>e5 �f7! 4 'i1o>d6 �e8! 5 l2Jf5 'i1o>d8 ! 6 lLle7 lLlb4 and Black de fends. W3) 1 lLlf2 'i1o>g3 ! (gaining a vital tempo) 2 l2Jd3 (or 2 lLle4+ 'ittf3 ! 3 'i1o>e5 lLlb4 ! and White has no reason able way to meet the threat of 4...lLlc6+; note that 3...lLlb8? would be less effective because of 4 'i1o>d5 ! 'i1o>f4 5 lLlc3 'itf5 6 'i1o>d6 lLla6 7 'i1o>c6 'i1o>e6 8 'i1o>b6 lLlb8 9 'i1o>b7 lLld7 10 lLla4 and wins) 'i1o>f3 ! 3 'itte5 'i1o>e3 ! 4 lLlf4 lLlb8 ! (here, too, Black uses the tacti cal point that White cannot easily prevent ...lLlc6+; not 4...lLlc5? 5 'i1o>d5
22 tll +Ll:i v tLJ
tlla 6 6 tlle 6 tllb8 7 �c5 �e4 8 tlld4 tlla6+ 9 'itb6 tllb4 10 tllc2 tlld5+ 11 rJi>b7 winning, nor 4...�f3? 5 tlld 5 �g4 6 'itd6) 5 tlld5+ �d3 ! 6 �d6 rJi>c4 and Black's king approaches the pawn, forcing the draw. In certain cases the win can be quite long and complex. In the fol lowing fascinating position White can only promote his pawn after a subtle zugzwang battle.
'iti>e2 (3...rJi>e4 4 �d6! �d4 5 tllb 2! reaches the zugzwang position in the main line) 4 tllb2 ! we reach a critical position. The subsequent play will make no sense unless I let the reader in on a little secret: the position with W'itd6 v B�d4 (other pieces as they are now) is reciprocal zugzwang. Therefore both sides struggle to avoid placing their kings on the cor responding squares d4 (for Black) and d6 (for White), because the op ponent would instantly (or at least after a little thought) occupy the other square, placing the unfortunate victim in zugzwang. There are now two lines: 1) 4 'itd2 (in this case the re ciprocal zugzwang is completely avoided) 5 'itd6! �c3 6 �c5 ! tlla6+ (6...�b3 7 �b5 tlld5 8 tlld3 tllc7+ 9 'itb6! tlld5+ 10 �c6 tlle7+ 11 �d7 tlld 5 12 a6) 7 rJi>b5! tllc7+ 8 rJi>b6 ! tlld5+ 9 �c5 tllc7 (9 ...tllb4 10 tlldl+ �b3 1 1 tlle3 �a4 12 �b6 wins just as in diagram 16) 10 tllc4 rJi>b3 11 tlle5 �a3 12 �b6 tlld5+ (12 ...'ita4 13 tllc 4! tlld5+ 14 �c6 tllb4+ 15 �c5 is diagram 16 again) 13 'itc5 'ita4 14 tllc 4! tllb4 15 'itb6 ! transposing yet again to diagram 16. 2) 4 �e3 5 �d7! (certainly not 5 �d6? 'itd4 ! and White has fallen into the reciprocal zugzwang) 'itd4 (it wasn't possible to avoid this for long, for example after 5...�e4 6 �c7 ! - again avoiding d6 - Black has the choice between 6...'ite5 7 tlld3+ ! and 6 ...�e3 7 �b6 �d4 8 �b5 tlld5 9 tlla4 as in the main line) 6 �d6! (now Black has to make a concession) tlla6 (6...�c3 7 'itc5 as ...
+/-
22
Original (22): White to play wins quite easily by 1 'itd6 'itg4 2 tllf2+ 'itf5 3 tlld3 and the pawn will soon move forward. With Black to play White can only win by very accurate play. After l...�g4 (l...�g5 allows White to win without any special finesses, for example by 2 'itd6 'itf4 3 tllf2 �e3 4 �c5 tlla6+ 5 �b6 tllb4 6 �b5 tlld 5 7 �c5 tllc7 8 tllg4+ �d3 9 'iti>b6 tlld5+ 10 �b7 tllb4 1 1 tlle 5+! 'iti>e4 12 tllc 6!) 2 tllf2 +! (White must extract his knight from its offside situation) 'itf3 (2 ...�f4 3 �d6 trans poses to the previous note) 3 tlldl !
...
tb+�
in line 1) 7 �c6 ! tDb4+ 8 �b5 tbd5 9 tba4 (White's knight has moved from the h-file to the a-file, and now serves to keep Black's king at bay) tbc7+ 10 �b6! tbd5+ 11 �c6 tbb4+ 12 �b5 tbd5 (White's triangulation has gained a tempo) 13 tbb6 tbc7+ 14 �c6 tba6 15 tbd7 (once again the knight and king co-operate to create a barrier against Black's king) tDb4+ 16 �b5! tbd5 17 tiJbS tbc7+ 18 �b6 tbd5+ 19 �b7 and the pawn ad vances, leading to a straightforward win. The final position with the pawn on a5 is not only of great importance (because it often arises from posi tions with the pawn further back), but it is also one of the most beautiful 5-man endings I have ever seen.
+I-
23
Original (23): If it were Black to move, White's pawn would be able to ad vance immediately. With White to play the win is far more complex, and involves transferring the move to Black. The manoeuvre by which
v
tD 23
White accomplishes this requires extreme accuracy - 10 consecutive 'only' moves: 1 �b8! (the first step is to transfer the knight from c5 to e5, but White cannot accomplish this immediately since he has to keep Black's king bottled up; after 1 tbd3? �d7! the position is a draw) tbd5 (preparing to meet 2 a6? by 2...tbb4! 3 a7 tbc6+!) 2 tbd3! (by defending b4 White threatens a6, so Black has no time to move his king) tbc7 3 tbe5! (this covers c6, so Black cannot defend by 3...tbd5 4 a6 tbb4) tba6+ 4 �a7! (the start of a six-move king tour which ends up on b7; after 4 �b7 tbc7 White loses time, be cause he can only win by returning to b8) tbc5 (4...tbc7 5 �b7 trans poses into the main line at move 9) 5 �b6! tba4+ 6 'iii>b7 ! (White still hasn't lost a tempo, because his king manoeuvre �b7-b8-a7-b6-b7 has taken an even number of moves, but he has misplaced Black's knight) tbc5+ 7 �c6! (White's big chance arrives; this triangulation loses a move) tba6 (7 ...tbe6 8 a6) 8 �b6! tbc7 (8 ... tbb8 9 �b7) 9 �b7! (now White is close to home; he just has to transfer the knight back to c5) tbe6 (9 ... tbb5 10 a6 tbd6+ 11 �c6 tbc8 12 tbf7+ �e7 13 �b7) 10 tbd3! (d3 is a very versatile square for the knight; by preventing ...tbc5+ White threatens a6 and again Black has no time to move his king to d7) tbc7 11 tbc5 (mission accomplished) tDb5 (11...tbeS 12 a6 tbd6+ transposes) 12 a6 tbd6+ 13 �c6 tbc8 14 tbe6+ �e7 15 �c7 tbd6 16 tbd4 tbe8+ 17 �b8 and White finally wins.
24 lll +L!. v lb
Now we move the pawn back to a4. Most of the interesting over-the board examples may be found in this section. Lest anyone should believe that 5-man endings are straightfor ward, in the following pages we will see two World Champions throw away half a point. First of all, let's see what happens when Reti's position is shifted back another rank. We already saw, in diagrams 12 and 16, that this posi tion is harder to win when White is to move. This tendency is even more pronounced when the pawn is on a4; indeed with White to move the posi tion is drawn and so it is a position of reciprocal zugzwang.
-
·'.. :::J!f:
• • · /$; • . . �� . . . �� . .r
A�� ��,';/,ii
•• -
///'//�
f� • ;.:{:X; �?fa: •.
• �·· bf::::::: • . ·ef� r.-- �� • • :;:�"LJ bf� �:a . ;��) ef� . :aef' �?; �.".·.·;;�. ;
%
=/-
24
Original (24): With White to play, the es sential difference lies in the variation 1 lllc3 llld4+! 2 'it>c5 lllb3+ 3 �c4 llla5+ 4 'it>b5 lllb7 5 llle4 'iii>b3 6 lllc5+ lllxc5! 7 a5 lllb7! 8 a6 llld6+! and the knight reaches c8, halting the pawn. Of course, this didn't work in the earlier diagrams. However, with
Black to play the result is the same as before: White wins by 1...'iPb2 2 'it>b4 �c2 3 lllcl (double deflection) llld4 4 �c5 llle6+ (or 4...lllf 5 5 a5 llle7 6 a6 lllc8 7 llle2 'iPb3 8 �b5 'it>a3 9 llld4 llld6+ 10 'it>c6 lllc8 1 1 'it>c7) 5 'it>b6 lllf4 6 a5 llld5+ 7 'it>b7 lllb4 and now: 1) 8 llle2? 'iii>b3 9 lllf4 deserves a special mention because it is an in teresting draw in its own right. Black must not play 9 ...'iti>a4?, when 10 'it>b6 ! reaches a position of recipro cal zugzwang with Black to move. Instead Black must steer clear of the mined square a4 and play 9 ...'iii>a3 !. If White waits by 10 'iii>c 7, then Black counters with 10...'it>b3 ! and White cannot make progress. 2) 8 llla2! llld3 9 �b6 is correct, and now the white pawn marches through. Using this analysis, it is possible to solve many positions with Black's king and knight behind the pawn, for example:
tb+.0. v lb 25
(25): Black to play draws by l ...ll'id4+, so we assume that White moves first. After 1 ll'if3 ! there are two lines: 1) 1 �b2 2 �b4 �c2 3 ll'iel+ �b2 4 ll'id3+ �c2 (4 ...�a2 5 ll'icl+ wins) and now the simplest win is by 5 tbc 1 transposing into the previous diagram. 2) 1 c;t>c2 (the best defence, not analysed by Averbakh) 2 �b4 �b2 3 �5 (this covers c6, and so prevents Black's defence 3...ll'id4) 'it>a2 (or 3 ..�c2 4 ll'id3 ll'id4 5 'it>c4 ll'if5 6 �b4+ �b2 7 �c5 &De7 8 a5 &Dc8 9 a6 �b3 10 tbct5 �a4 11 �c6 �a5 12 !ix7 !) 4 ll'id3 (now Black is forced to move his knight) ll'id4 5 tbc1+ �b2 6 �c5 �a3 7 a5 ! �a4 8 a6 ll'ib5 9 �b3 ll'ia7 10 'it>b6 ! ll'ic8+ 11 'it>b7 ! �6+ 12 'it>c7 ll'ib5+ 13 'it>b6 ll'id6 14 ll'id4 ll'ic8+ 15 'it>b7 ! and White wins. The conclusion is that in this type of position (W.0.a4 and B'it> defend ing from behind) White wins in the situation W�b4 v B'it>b2, but with W�b5 v B'it>a3 the result depends on subtle zugzwang considerations. Now we move on to a selection of over-the-board examples. The first position leads into the analysis of diagram 24. (26): The first phase of the ending was played very accurately by Laz arevic: 1 c;t>f5 ! ll'ih5 (1...ll'ie2 2 ll'ie6 �g3+ transposes to the next note) 2 �6 ! c;t>f3 (after 2 ...ll'ig3+ 3 'it>g4! � 4 ll'ig7 !, the lines 4...ll'if2+ 5 �5! ll'ie4 6 �g6 and 4...�d4 5 �f4 ! �f6 6 �f5 ! ll'ie4 7 ll'ie8 'it>e3 8 ll'if6 �g3+ 9 'it>g4 ! �f2 10 ll'id5 'it>g2 11 ...
...
.
26
+!=
Lazarevit-Vreeken Split wom OL 1964 ll'ie3+ c;t>f2 12 ll'ifl are winning for White) 3 �g5 ! ll'ig3 4 ll'ic5 ! (a recip rocal zugzwang with Black to play) c;t>g2 (the line 4 ...c;t>f2 5 'it>g4 c;t>g2 6 ll'id7 'it>f2 7 ll'if6 'it>g2 8 ll'ie8 'it>f2 9 ll'id6 'it>g2 10 ll'if5 ll'ie4 11 'it>f4 ll'if6 12 �g5 ll'id7 13 h5 also wins for White) 5 'it>g4 'it>f2 6 ll'id3+ 'it>g2 7 ll'if4+ 'it>f2. At this point White could have played 8 ll'id5 'it>g2 9 ll'ie3+ �f2 10 ll'ifl transposing to diagram 24 after White's third move. Instead of this, the game continued inaccu rately, with White missing several chances to win quickly: 8 ll'ih5 ll'ie2 9 ll'ig7 ll'ig3 10 ll'if5 ll'ie4 11 'it>f4 ll'if6 12 c;t>g5 ll'ih7+ 13 �g6 &Dfs+ 14 �f7 tbct7 15 h5 ll'ie5+ 16 'it>f6 ll'ig4+ 17 'it>g5 �f3 18 ll'ih6 (18 ll'ie3 would have won very rapidly) ll'if2 19 ll'if5 ll'ih3+ (Black decides not to re turn to g4, but it makes no difference to the result of the game) 20 'it>f6! �g4 c20...ll'if2 21 'it>g6 ll'ig4 22 c;t>g5 ll'ie5 23 ll'id6 ll'ig4 24 ll'ie4 ll'ie5 25 'it>f5 ll'ig4 26 ll'if6 is also losing)
26 lLJ+� v lLJ
2 1 h6 li:)g5 22 �g6 (22 li:)d4 �h5 23 �g7 �g4 24 �g6 �f4 25 li:)f3 was a more efficient winning line) li:)e6 23 li:)e3+ �h4 24 �f6 li:)f8 25 �t7 �h5 26 li:)g4 li:)h7 27 �g7 ! li:)g5 28 li:)f6+ ! �h4 29 li:)e4 tbe6+ 30 �g8 li:)f4 3 1 h7 li:)g6 32 �g7 �h5 33 li:)f6+ 1-0. With Black to play this position is drawn, but accurate defence is nec essary : 1 . . . li:)d5 ! 2 �f5 li:)f4 ! 3 li:)h3 (or 3 li:)e6 li:)xe6! 4 �xe6 �f4 ! ) li:)h5 ! 4 �g5 (after 4 �g4 li:)g7 ! 5 li:)g5 �d4 White is too slow to shift the knight from g7) li:)g7 ! (but not 4 ...li:)g3? 5 �g4 ! li:)e4 6 li:)g5 ! li:)f6+ 7 �f5 ! li:)h5 8 li:)e6 ! transposing into the White to play analysis after White's second move) 5 �g6 (or 5 li:)f4 �f3) �f3 ! and Black can attack the h-pawn with his king. In the following position, accu rate play by Portisch notched up the full point. At one or two moments he could have won more quickly, but this is a minor quibble. (27): In the game it was Black to move (White to play draws immedi ately by 1 �d2). It is often more im portant to keep the opposing king at arm's length than to push the pawn: l . . .�c3 ! (after l . ..a4? White draws by 2 �d2 ! a3 3 li:)d3 ! li:)f4 4 li:)c5+! �b2 5 li:)a4+ !) 2 lt:)c6 (or 2 li:)d3 a4 3 li:)cl a3 4 li:)a2+ �b2 5 li:)b4 �b3 6 li:)d3 �c3 7 li:)cl li:)f6 8 �e2 �c2 9 li:)d3 li:)d5 and wins) a4 ! 3 li:)a7 (3 li:)d4 li:)g3 4 li:)b5+ �b4 5 li:)d4 �c4 6 li:)c2 �c3 ! 7 li:)a3 li:)f5+ 8 �e2 �b3 9 li:)bl �b2 10 li:)d2 li:)d6) li:)f6 (3 . . . li:)g3 was faster) 4 li:)b5+ �b4 5 li:)d4 �c4 6 li:)c2 �c3 ! 7 li:)d4
=I+
27
Hort-Portisch Wijk aan Zee 1975 li:)d5+ 8 �e4 li:)b6 (this inaccuracy extends the win by four moves; 8 ...li:)c7 9 li:)e2+ �c2 10 li:)d4+ �b2 was quickest, since li:)b5 is now im possible) 9 li:)e2+ (Black wins even after the toughest defence 9 li:)b5+ �b4 10 li:)d4 �c4 1 1 li:)c2 li:)d5 1 2 �f3 c3 1 3 li:)a3 �b2 1 4 li:)c4+ �b3 1 5 li:)d2+ �b4 1 6 li:)bl li:)c3) d2 0- 1, since 10 lt:)d4 a3 1 1 li:)b3+ �c2 1 2 li:)d4+ �b2 forces the pawn home. This was an example of high qual ity over-the-board play, but the same cannot be said for the next three ex amples. (28): White was to play in the game; there are two main possibili ties: l) 1 �c7? (Alekhine's chosen move throws the win away) li:)c4 ! 2 li:)d3 and now: la) 2 �? (and certainly not 2 . . . �t7? 3 li:)e5+!) 3 li:)e5 ! li:)a5 4 li:)f3 ! (a tricky move, which is typi cal of li:)+� v li:); White threatens ••.
.!LJ+�
+I=
28
Alekhine-Van den Bosch Nauheim 1936 tlld 2 followed by �b6 trapping the knight, so Black is forced to play his knight away from a5) tllb3 (or 4...tllc4 5 tlld 2 tlle3 6 a5 tlld5+ 7 �d7 tllb4 8 tlle4 �fl 9 tllc5 �f8 10
v
tLJ 27
6 a5 �e7 7 �c6 tllc 8 8 �c7 tlld6 9 a6 �e6 10 .!LJc4) 2 tlle3! �e7 3 �c6 ! (not 3 �c7? tlle8+! 4 �c6 �d8!, nor 3 a5? �d7 ! 4 a6 .!LJc8+! 5 �b7 .!Dd6+! 6 �b8 tllb 5 ! 7 tllc 2 �c6 ! 8 tlld4+ �b6 ! with a draw in both cases once again keeping the king out is the most important factor) tllf 7 4
••.
1h-1h.
2) 1 .!Dd5 (not the only winning move) �fl ( 1...tllc4+ 2 �b5 tlld6+ 3 �c5 tllb7+ 4 �b6 tlld6 5 tlle3 cov ers c4, and White wins after 5...�f6
29
-/+
Simagin-Botvinnik USSR Ch 1955 (29): 1...tlld 2 (good enough to win, but 1...�f5 2 tllf3 �f4 3 tllh4 �g4 would have been simpler, for example 4 tllg2 tllc3+ 5 �b3 tlld5 6 �c4 �g3 7 .!Del h4 or 4 tllg6 .!LJc5+ 5 �b4 tlld7 6 �c3 �g5 7 tlle7 h4 ! and Black wins in both cases) 2
28 ttJ+L'!.
v
tLi
1 ) 2 'it>f5? (ttJ+L'!. v tLi proves too much for another world champion) 3 'it>c3 ! tLie4+ 4 'it>d4! (White's king is close enough to draw) lLig5 5 tLld3 'it>g4 6 lLie5+ 'it>f5 7 tLld3 'it>g4 8 lLie5+ 'it>g3 9 lLig6 ! tLie6+ 10 'it>e3 lLif8 1 1 tLixf8 h4 12 tLle6 h3 13 lLig5 •••
1f2- 1h
2) 2 'it>f4! 3 lLig6+ (now 3 'it>c3 may be met by 3 . . .lLif3 ! 4 lLig6+ 'it>g5 !, and the pawn advances imme diately) 'it>f5 4 lLih4+ 'it>g4 5 lLig6 (5 lLig2 lLifl ) tLlf3 6 'it>c4 'it>g5 ! 7 lLif8 h4 8 lLie6+ 'it>g4 9 e2 lLid3 1 3 'it>xd3 'it>f3 ! and Black wins. Somewhat surprisingly, Black wins even if White moves first, for example 1 'it>b3 'it>f5 2 tLlf3 lLig5 3 tLlh4+ 'it>g4 ! 4 lLig2 (4 lLig6 lLif7 5 'it>c3 'it>g5 6 lLif8 h4 is even easier) lLif7 5 'it>c3 'it>g3 6 lLie3 tLlh6 7 'it>d2 h4 8 'it>el 'it>f3 9 tLic4 'it>g2 ! 10 tLie3+ 'it>gl 1 1 lLid l h3 1 2 lLif2 h2 with a simple win. The following struggle was even more fluctuating. (30): 1 'it>g5 ? (this loses a vital tempo in the race to the queenside; 1 'it>f6 was better, when both 1 . ..lLif3 2 'it>e6 lLie5 3 'it>d6 lLid3 4 'it>c6 'it>c4 5 'it>b6 'it>b4 6 tLic3 ! and 1 . . . 'it>c4 2 'it>e5 lLif3+ 3 'it>d6 tLld4 4 lLib2+ �b3 5 tLld3 ! 'it>c3 6 lLic5 ! lLib3 7 lLia4+ ! 'it>b4 8 lLib2 ! lead to a draw) and now: 1 ) 1 lLif3+ (the simplest win) 2 'it>f5 lLie5 ! 3 'it>e6 lLid3 4 'it>d7 'it>c4 ! (but not 4 . . . lLic5+? 5 'it>c6 !) 5 'it>c6 'it>b4 ! 6 lLib6 lLif4 ! (reciprocal zug zwang with White to play) 7 'it>c7 'it>b5 8 'it>b7 lLie2 9 'it>c7 lLic3 10 'it>b7 ...
...
30
=I
Minero-Ruge Copa Latinoamericana 1992 tLldl 1 1 'it>c7 lLie3 12 'it>b7 lLic4 1 3 tLld5 'it>c5 14 lLic3 'it>b4 15 lLid5+ 'it>b3, followed by the advance of the pawn. 2) 1 lLig2 (the move played in the game maintains the win, but makes it much more complex) 2 'it>g4 lLie3+ ! 3 �f3 tLld5 ! (ruling out 'it>e2) 4 'it>f2 with two lines: 2a) 4 'it>c4 5 'it>el 'it>b4 ! 6 lLib2 'it>b3 7 tLld3 'it>c2 ! 8 lLic5 lLic3 9 'it>fl lLie4 ! 10 lLia4 lLif6 ! (an idea we have seen before; the threat is ... lLid7 fol lowed by . . .'it>b3) 1 1 lLib6 tLld7 ! 1 2 lLic4 a4 ! 1 3 'it>e2 'it>b3 ! 14 tLld2+ 'it>c3 15 'it>d l lLic5 16 lLibl + 'it>b2 17 tLld2 lLie4 1 8 lLic4+ 'it>c3 transposing to diagram 23 after White's 7th move. This would be a really tough line to find over the board. 2b) 4 tLic3? (the game continu ation) 5 lLib2 lLie4+ 6 'it>e2 'it>c3 7 lLia4+ ! 'it>b4 8 lLib6 (8 lLib2 would also be a draw) 'it>b5 9 lLid5 'it>c4 10 lLie3+? (for no obvious reason White plays his knight away from Black's ...
...
...
ttJ+,0, v ttJ 29
pawn; 10 ltJb6+ would be a draw) 'it>d4 ! (an excellent reply - White is prevented from improving his king position) 1 1 ltJc2+ 'it>c3 ! 12 ltJa3 (a blunder, but even after 12 'it>dl 'it>b2 ! 1 3 ltJd4 a4 14 ltJc2 'it>bl ! White should lose, because we have trans posed to diagram 23 after White's third move) �b3 0- 1 because White is unable to avoid the exchange of knights. There are a mere 1 85 reciprocal zugzwangs in the ending of ltJ+,0,a4 v ttJ. As usual, I have selected one for the reader to marvel at.
but these lose a vital tempo if White simply runs his king to the queen side. If Black plays 1 . . .ltJa5, then White's king gains access to d6, which helps him head off Black's king. The situation is simpler if White moves first; a king move al lows .. .'it>f6 while a knight move gives Black immediate access to g6. The detailed Black to move analy sis runs: B 1) 1 'it>g4 2 ltJe6 'it>f3 (2 ... ltJa5 3 'it>d5 'it>f5 4 ltJd4+ ! 'it>f6 5 'it>c5 ltJb7+ 6 �c6 ltJa5+ 7 'it>b5 ltJb7 8 ltJb3 'it>e5 9 'it>c6 ltJd8+ 10 'it>b6 ltJf7 1 1 a5 'it>d5 12 a6 ltJd6 13 ltJd2 ltJc8+ 14 'it>b7 ltJd6+ 15 'it>c7 ltJb5+ 1 6 'it>b6 ltJd6 17 ltJbl ltJc8+ 1 8 'it>b7 ! ltJd6+ 19 'it>b8 ltJb5 20 ltJc3+ and the pawn promotes) 3 'it>d5 'it>e3 4 ltJd8 (an attractive move) ltJa5 5 'it>c5 'it>d3 6 'it>b4 ! ltJc4 7 ltJc6 (this is a reciprocal zugzwang) ltJd6 8 a5 ! and the pawn advances. B2) 1. ltJa5 2 ltJe6+ ! 'it>g6 3 �d6 ! 'it>f7 (3 . . . 'it>f5 4 'it>d5 ! 'it>f6 5 ltJd4 ! transposes to line 2a in dia gram 30) 4 ttJd4 'it>e8 (4 ...'it>f8 5 'it>c5 'it>e8 again transposes to line 2a in diagram 30) 5 'it>c7 ltJc4 6 ltJc6 'it>f8 7 ltJe5 ! ltJa5 8 ltJf3 ! ltJb3 9 ltJd2 ! ltJc5 10 a5 ! 'it>e7 1 1 'it>b6 ! ltJd7+ 1 2 'it>c6 ! �d8 1 3 ltJc4 ltJb8+ 14 'it>b7 ltJd7 15 ltJe5 ltJc5+ 16 'it>c6 trans poses to the position after White's 7th move in diagram 23. B3) 1 �h6 (1 ... 'it>h5 2 'it>d5 'it>g5 is the same) 2 'it>d5 ! 'it>g5 (2 . . . ltJa5 3 ltJe6 'it>g6 4 ltJd4 'it>f7 leads to line 2a in diagram 30) 3 'it>c6! ltJa5+ 4 'it>b6 (4 'it>b5 ltJb7 5 'it>b6 ltJd6 6 'it>c6 ltJc4 is slower by two moves) ltJc4+ ...
.•
=I-
31
Original (31): With White to play the posi tion is drawn, but if Black moves first White can promote his pawn in 36 moves. This result appears para doxical, but the detailed analysis be low reveals the logic hidden behind it. Black's main defensive plan is to move his king towards a8 via g6, f7 and e8. lf Black moves first, then the only king moves to preserve the op tion of ... �g6 are ...�h6 and ... 'it>h5,
...
30 tLJ+.0-
v
tLJ
(4 . . . tDb3 loses after 5 lDrl7 followed by tbc5) 5 'iii>c5 tDa5 6 tDe6+ ! Wf5 7 'iii>d5 ! (7 tbd4+? is tempting, but Black can draw by continuing 7 ...'it>e4 ! 8 tDc6 lDb3+ 9 'it>c4 tbd2+! 1 0 'it>b5 'iii>d5 ! 1 1 a5 'iii>d6 ! 1 2 a6 rJic7 ! ) rJif6 8 lbd4 ! 'iii> f7 9 �c5 ! and once again we have transposesd to line 2a in diagram 30. With White to play, 1 tbe6+ (1 tbd7 is met by the obvious 1 . . . 'it>g6, while 1 'iii>d5 'iii>f6 2 'iii>c6 tba5+ ! 3 rJib6 tDb3 4 tbd7+ 'iii>e6 5 tbc5+ tDxc5 ! draws because of Black's ex tra tempo) 'iii>g6 ! 2 'iii>d5 'iii>f6 ! 3 tDd4 �e7 ! 4 'it>c6 tba5+ ! 5 'iii>b6 tbc4+ ! 6 'it>b5 tDd6+ 7 'it>c6 tDc4 reaches a po sition in which White cannot make progress. Now we move White's pawn back to a3. These positions are so compli cated that I have not been able to find any non-trivial position which was correctly analysed in pre-database times. First of all, let us once again move Reti's position back another rank (thus continuing the sequence from diagrams 12, 16 and 24): (32): Perhaps surprisingly, the latest shift does not alter the result; the position is again reciprocal zug zwang. With Black to play the analysis runs 1 ...'iii>b l 2 'iii>b3 ! 'iii>c l 3 lbg2 tDd3 (or 3 ... 'iii>b l 4 lbe3 lbd3 5 tbc4 ! 'iii>a l 6 'iii>c3 and now 6... tDc5 7 'it>b4 tba6+ 8 'it>b5 tDc7+ 9 'iii>c6 tDe6 1 0 a4 is a relatively straightforward win, while 6 . . .tDf4 7 a4 'iii>a2 8 a5 tbd5+ 9 'it>d4 tDc7 10 'it>c5 'it>b3 1 1 tbe5 transposes to line 1 of diagram
=I-
32
Original 22 after White's 1 1 th move) 4 tDe3 ! 'it>d2 5 tbc4+ ! 'iii>e2 6 lDb2 with two possibilities: 1 ) 6 llJe5 7 Wb4 ! (but not 7 a4? We3 ! 8 a5 �d4 9 a6 tDc6 ! - once again we emphasise that improving one's king position is more impor tant than pushing the pawn) tbd7 8 'it>b5 ! (after 8 a4? 'it>e3 ! 9 a5 Wd4 ! 10 'it>b5 tbb8 ! 1 1 tbc4 'iii>d5 ! we arrive at a reciprocal zugzwang with White to play) 'it>e3 9 'iii>c6 ! tDb8+ (9 . . tDe5+ 10 'it>d5 tbd7 1 1 'iii>d6 tDb8 12 'it>c7 is the same) 10 �c7 tDa6+ 1 1 �b6! tDb8 12 a4 'iii>d4 13 'it>b7 (not 13 a5? 'iii>d5 14 tbd3 'iii>d4 ! 15 lDb4 'it>c4 ! 16 tbc6 tbd7+! - note that 14 ...'iii>d6? in this line would have lost to 15 tbc5) tbd7 14 a5 ! tDc5+ 15 'it>b6 tLJct7+ 16 'iii>c6 tDc5 17 'iii>b5 'it>d5 18 tDtl3 lbe6 1 9 �b6 and wins. 2) 6 tbc5+ 7 'it>b4 tbb7 (7 ...tDd7 transposes to line 1 ) 8 'iii>b5 ! 'it>d2 9 a4 'it>c3 10 'iii>b6 ! tbd6 1 1 'it>c6 ! tDc8 1 2 'iii>c7 ! lbe7 13 a5 ! tDd5+ 14 'iii>c 6! tDb4+ 15 'iii>b5 tbd5 16 tba4+ 'iii>d4 17 lDb6 transposing to the position ...
.
...
ttJ+,0, v ttJ 31
after White's 1 3th move in line 2 of diagram 22. With White to play the situation is less complex. The winning line of diagram 16 fails since after 1 tlJc2 1Lld3+ 2 �c4 tiJb2+ 3 �c3 tlJa4+ 4 �b4 tlJb6 ! 5 tlJe3 �b2 ! 6 tlJc4+ ! 1Lixc4 ! 7 a4 lbe5 8 �b5 lbd7 Black's knight gets back and he draws com fortably.
is 5 tiJd6 �d4 6 �g4 ! tlJh7 7 �f5 ! �d5 8 tlJe4 tt:Jf8 9 h6 �c6 10 �f6, but Paoli's move doesn't change the result) tlJh7 reaches the critical posi tion. Now Paoli gave 6 h6? �e4 !, which is indeed drawn but, as we have said so often, an improved king position is usually more important than pushing the pawn. White can win by 6 �f4 ! �c4 7 �e5 tlJf8 8 h6 tlJd7+ 9 �e6 tt:Jf8+ 10 �f7 tlJh7 1 1 tlJh4 tlJg5+ 1 2 �e7 tlJh7 1 3 tlJf3 �d5 14 �f7 followed by �g7. 2) 2 �g3? and now: 2a) 2 tlJe4+? (given by Paoli) 3 �f4 ! tlJf6 and now: 2al) 4 tlJg3+? �f2 ! 5 h4 ttJ
+I
33
Paoli, 1982 Grzeban Jubilee (33): The diagram position arises after three moves of introductory play. After 1 �g2 ! tiJd2, Paoli gave 2 h4? as drawing and 2 �g3 ! as win ning, but in fact he had these exactly the wrong way round. Here is the correct analysis: 1) 2 h4! (this is the only move to win) tlJe4 3 h5 tlJg5 (White also wins after 3 ...tlJf6 4 h6! �d3 5 �f3 ! �c4 6 �f4 �d5 7 �g5 tlJh7+ 8 �g6 ! tt:Jf8+ 9 �g7 tlJe6+ 10 �f6 �f8 1 1 tlJg3 tlJh7+ 1 2 �f5 �d6 1 3 �g6 tiJf8+ 1 4 �f7 tt:Je6 1 5 tlJe4+) 4 �g3 ! �d3 5 �g4 (the quickest win
.•.
32 ttJ+.0, v ttJ
34
=I=
Podgaets-Tal Alma-Ata 1969 (34) : The position is a draw, even with White to move. After 1 �g6, as played in the game, Black has two straightforward drawing lines: 1) 1 tlJe7+ 2 �f6 tlJd5+! 3 �g5 �c7 4 h4 �d6 5 h5 �e6 6 h6 �f7 is safe. 2) 1 �c7 2 h4 �d7 3 h5 �e7 4 �g7 (4 h6 �f8 5 h7 tlJe7+) tlJd8 ! 5 tlJd4 (5 tlJc5 tlJf7 !) tlJf7 ! (reciprocal zugzwang with White to play) 6 tlJf5+ �e6 ! 7 tlJe3 �e7 ! (reciprocal zugzwang again) 8 tlJg4 (White can not make progress as the parity is wrong) �e6 9 �f8 tlJg5 ! 10 h6 �f5 ! drawing. However, in the game Tai chose a third move: 3) 1 tlJe5+? 2 �f6! tlJd3 (the best defence is 2 ...tlJd7+ 3 �e6 �c8 4 h4 !, but even then Black loses after 4 ... tlJf8+ 5 �f7 tlJd7 6 h5 or 4 ...�d8 5 h5 ! �e8 6 h6 ! tlJf8+ 7 �f6 ! tlJd7+ 8 �g7 tlJf8 9 tlJd4 �e7 10 tlJf3 �e8 1 1 tlJg5 �e7 1 2 tlJh3 tlJe6+ 13 �g8; after 2 . . . tlJf3 White has a neat win ...
by 3 tlJd4 ! ttJh4 4 �g5 tlJg2 5 tlJc2) 3 h4 tlJf4 4 tlJc5+? (Podgaets throws away a relatively easy win by 4 �f5 tlJh5 5 �g5 tlJg3 - if 5 ...tlJg7, then 6 tlJd4 - 6 tlJd2 �c7 7 �g4 tlJe2 8 h5 tlJd4 9 h6 tlJc6 10 h7) �c6! (having been let off the hook, Tai makes no mistake) 5 tlJd3 ttJh5+ ! 6 �g6 tlJg3 ! 7 tlJt1 �d6 8 tlJhl tlJe2 ! 9 �f6 tlJf4! 1 0 tlJg3 �d7 1 1 �f7 �d6 1 2 tlJe2 tlJxe2 ! 1/z- 1/2. There are just 92 positions of re ciprocal zugzwang with tlJ+.0,a3 v tD. We will not examine any now, but readers will find one of the more interesting positions in the analysis of diagram 37 below. With the pawn on the second rank, I have not been able to find a single correctly analysed composed position, so readers will have to make do with an incorrect one !
...
...
35
+/= Selman, 1966 Schakend Nederland
(35): Selman's solution ran 1 tlJa2+ tlJxa2 2 h4 tlJc l 3 h5 tlJe2 4
lb+� v lb 33
b6! lbf4 5 �g8 ! (not 5 g6 lbd5 2 h4 ! is also effective) lbd5 2 d4 3 �2+ ! �e5 4 h5 ! lbe7+ 5 �g7 ! �f5+ 6 'it>n ! lbd6+ (or 6 ... 'it>d6 7 �g6 �e5 8 lbg3 lbh4+ 9 cj;g7) 7 �g6! lbf5 8 lbg3 lbh4+ 9
36
=I+ Yandemirov-Berelovich Russia 1992
3 lbc4+ �c5 and 3 �e6 d6! lbd l 6 lba2 ! �b5 7 �d5 ! ) 4 'it>d6 a4 5 lbc5 a3 ! 6 'it>d5 a2 7 lbb3 c2) 0- 1 . White's first move i n the follow ing position is quite remarkable. The play depends on a succession of re ciprocal zugzwangs. (37): It is simpler to consider the White to move situation first. I am sure that the vast majority of over the-board players would choose 1 a4, even though this throws away the win. The point is that after 1 a4? ( 1 'it>b7? lbf3 2 a4 lbd4 3 a5 lbb3 also
34 lD+.0.
v
lD
+/-
37
Original leads to a draw) lDf3 ! (reciprocal zugzwang) 2 a5 lDd4 ! (again recip rocal zugzwang) 3 a6 lDb5 ! White is in a third reciprocal zugzwang. After 4 'itb7 (there is nothing better; 4 lDc2 �c6! 5 lDd4+ �b6! rounds up the pawn) lDd6+ ! 5 �b6 lDc8+ ! he cannot make progress. White cannot break the chain of reciprocal zug zwangs leading to the position after Black's third move because he can not lose a tempo with his knight, while a king move would either block the pawn or set White up for . . . lDc5+ (in reply to a6). Once again play is dominated by parity consid erations. Now we can see how White to play does actually win. The only chance to lose a tempo is on the very first move, by playing 1 a3 ! . Black's reply is irrelevant; in any case White will reply 2 a4, and the parity shoe is on the other foot: 1. . .lDf3 (the other lines are 1... �d8 2 a4 ! lDf3 3 a5 ! lDe5 4 �b7 ! - not 4 a6? lD
l . ..�d6 2 a4 ! lDf3 3 a5 ! lD
=I-
38
Original (38): One of the most mysterious reciprocal zugzwangs I have ever seen. The first point to note is that Black's knight is paralysed; if White is allowed to play a4 free of charge
lLJ+.0.
then he is almost always winning. On the other hand, it is hard for White to make progress. If he moves his king to the queenside, then Black's king is free to head for a8. White has two possible winning chances. The first is to use his king and knight to stalemate Black's king on h 1, forcing the a3-knight to move. The second idea is to improve his knight position while keeping Black's king bottled up in the lower right-hand corner, and eventually aim for a formation with his king and d3 and knight on c4. Playing lLJc4 with the king on d3 wins if Black's king is on f4 or g4, but only draws if Black's king is on f5 or g5. It follows that this second plan is only effective 1f Black's king is unfavourably placed. It turns out that if Black is to play, White can threaten the first plan; to avoid being stalemated Black has to play his king to the first rank, and then White can succeed with the second plan. If White moves first, the parity is wrong and by accu rate play Black can avoid defeat. Suppose Black moves first. Then be may continue: 1 ) 1...lLJc4 (we will not analyse this at every move, as allowing the pawn forward is obviously a major concession) 2 a4! lLJa5 3 lLJd3 �g2 4 �f4! �h3 5 lLJe5 �h4 6 lLJf3+ �h5 7 lLJd2 �g6 8 e5! �f7 9 �d6! �b7+ 10 �c6 lLJd8+ 1 1 �b6 win lllDg. 2) 1...�hl (after this Black 's king is trapped on the first rank, and White can execute the second plan) 2 �g3 �gl 3 lLJd3 �fl 4 �f3! �gl
v
lLJ 35
5 �e3 �g2 6 lLJe5 �g3 7 �d3 f4 (Black is one tempo too slow) 8 lLJc4! lLJb5 9 a4! lLJc7 10 �d4! �f5 1 1 �c5 ! lLJa6+ 1 2 �b6 lLJb4 13 a5 �e6 1 4 �b5 lLJd5 15 �c5 lLJc7 1 6 c6 lLJa6 1 7 lLJb2 lLJb4+ 1 8 �c5 lLJa6+ 1 9 �b6 lLJb4 20 lLJd3 and White wins. 3) 1 . �g2 2 lLJd3! (this is recip rocal zugzwang) �h2 3 lLJf4 �gl 4 �f3! wins as in line 4a. 4) 1 . .�gl 2 �f3! and now: 4a) 2...�hl 3 lLJd3 �h2 4 lLJf4 �gl 5 �e2 �h2 6 �t2 (here the �d3 plan is too slow, for example 6 �d3? �g3! 7 lLJd5 �g4! 8 lLJe3+ �g5 ! and Black is in the drawing zone) �hl (now that Black's king is bottled up, White can transfer his knight to f3) 7 lLJe6 �h2 (7 ... lLJc4 8 a4 �h2 9 lLJg5 transposes) 8 lLJg5 lLJc4 (after 8...�h l 9 lLJf3 Black has to move his knight under even less favourable circumstances, and then White wins by 9 ...lLJc4 10 a4 lLJa5 1 1 lLJd2 �h2 1 2 �e3 lLJc6 1 3 �e4 �g3 14 lLJb3 lLJb4 15 a5 lLJa6 16 �d5) 9 a4! (White has achieved his aim) lLJa5 10 �e3 �g3 1 1 lLJe4+ �g4 1 2 �d4! �f3 1 3 lLJd2+ �e2 14 �c3 lLJc6 15 lLJb3 �e3 16 �c4 �e4 1 7 �c5 lLJd8 18 lLJd2 + �e5 1 9 lLJc4+ �e6 20 �b6! winning. 4b) 2...�h2 3 lLJe2! (not 3 lLJd3? �h3! and the king slips out) �h3 4 lLJg3! (White covers h5 just in time) �h4 (4...�h2 5 lLJf5 �h3 6 �f4! is a transposition) 5 �f4! �h3 6 lLJf5 ! (White's knight is now more active, and the king and knight can act to gether to harass Black 's king) �h2 (or 6... �g2 7 �e4, and after 7...t2 ..
.
36 llJ+� v llJ
8 �d3 ! �f3 9 li:Jd6 �f4 10 ll:Jc4 ! White is in time, while 7 . . . �h3 8 �f3 transposes to the main line) 7 �e3 �h3 (Black must move up the board, or else the �d3 plan wins, for example 7 . . .�g2 8 �d3 'iti>f3 9 li:Jd6 �f4 10 ll:Jc4 ! li:Jb5 1 1 a4 ! ; note how White has used the auxiliary threat of �d3 and ll:Jc4 to force Black's king to an unfavourable square; now White can return to the first plan) 8 �f3 (not 8 �d3? �g4! 9 llJd6 'iti>g5 ! 10 ll:Jc4 li:Jb5 ! 1 1 a4 ll:Jc7 ! 12 'iti>d4 �f6! 1 3 �c5 �e7 ! 14 �c6 li:Ja6 ! 15 �b6 li:Jb4 ! 16 a5 �d7 and Black holds on) �h2 9 li:Je7 (the knight is heading for d4) with a final branch: 4bl) 9 �hl 10 li:Jd5 �h2 1 1 li:Jf4 is line 4a. 4b2) 9 �gl 10 li:Jd5 �h2 (or 10 . . . �fl 1 1 �e3 ! �g2 1 2 �d3 �f3 1 3 li:Jb6 'iti>f4 14 ll:Jc4 ! and White wins) 1 1 li:Jf4 is again line 4a. 4b3) 9 �h3 10 ll:Jg6 ll:Jc4 (now Black has no choice, because after 1 0 . . . �h2 1 1 li:Jf4 we are again in line 4a, where White executed his stalemating plan) 1 1 a4 ! ll:Ja5 1 2 �f4 ! (and not 1 2 �e4? �g4) �g2 (or 12 . . . li:Jb3 1 3 li:Je5 �g2 14 li:Jd3 �h3 15 �e4 �g4 16 �d5 ll:Ja5 17 �c5 �f5 1 8 �b5 li:Jb7 19 �b6 li:Jd6 20 li:Jb2) 1 3 li:Je5 'iti>f2 14 �e4! �e2 1 5 �d4 ! li:Jb7 (or 15 . . .�d l 16 �c5 �c2 17 �b4 li:Jb3 1 8 li:Jd3 li:Jd4 19 �c4 li:Jf5 20 li:Jb4+ �b2 21 a5 li:Jd6+ 22 �d5 wins) 1 6 li:Jd7 ll:Ja5 17 �c3 li:Jb7 18 �b4 �e3 19 li:Jc5 and wins. White to play cannot make sig nificant progress, e.g. 1) 1 li:Jd3 'iti>g2 ! and now White finds himself on the wrong end of a
reciprocalzugzwang; a knight move allows . . .�f2. while 2 �f4 �h3 again allows the king to escape up the h-file. 2) 1 li:Je2 �g2 2 'iti>f4 �f2 3 li:Jd4 �g2 4 li:Jf5 �f2 ! 5 ll:Jg3 'iti>el 6 �e3 ll:Jc2+ draws. 3) 1 �3 �h3 ! 2 li:Jd3 'iti>h4 3 �f4 �h5 4 �f5 �h6 5 �f6 'iti>h7 6 �f7 �h6 and Black's king escapes. I will end, as has become tradi tional, with the longest win in the ending of ll:J+� v ll:J, indeed in the whole ending of ll:J+� v llJ (48 moves):
...
...
...
+I-
39
Original (39): White to play wins easily af ter 1 a4 ll:Jc6 2 �b7 ll:Ja5+ 3 �b6 ll:Jc4+ 4 �c5 ll:Ja5 5 'iti>b5 li:Jb3 6 llJe6 �f2 7 'iti>b4, so we assume that Black moves first. The analysis is quite complex: l . . .li:Jc6 ( l ...�f2 2 a4 llJc6 3 'iti>b7 ! llJd4 4 �b6 li:Jb3 5 ll:Je6 �e3 6 ll:Jc5 is relatively easy) 2 �b7 (White can also play 2 li:Jf5, but this leads back into the main line after
lb+l!. v lb 37
2 ...�f2 3 �b7 ! lba5+ 4 �b6! lbc4+ 5 �c5 ! lbb2 6 lbd6!) lba5+ 3 �b6 ! .&:4+ 4 c5 drawing) lbd5+ (or 8 ... �f3 9 lbc4 ! 00+ 10 �d4 ! lbb4 1 1 a4!) 9 c5 ! lbb2 when 16 � 1 ! is a neat move which is win ning after 16 ...lbd3+ 17 �c4 lbe5+ 1 8 �d5 lbd7 19 �d6 lbb6 20 lbe3+ �g5 2 1
lbb2 22 a5 ! lDd3+ 23 �b5 ! lbc5 24 lDd7 wins) 13 �c3 ! (again recipro cal zugzwang) d2 22 c 1 23 lbd7, White wins as in dia gram 32. 2) 20...�e2 (Black tries to avoid falling into the diagram 32 situation) 2 1
38 tb+l!.
v
tLi
to this point White's moves have been essentially unique, except for some minor variations in move-or der; this must be one of the hardest known winning lines in a natural po sition) �d5 33 �b5 tbc6 (33 ...�e6 34 tLlc5+ �e7 35 �b6 �d6 36 tLlb3 �d7 37 �b7!) 34 a6! (the rest is easy, but I will include it for com pleteness) tba7+ 35 �b6! tbc8+ 36 �c7 ! tba7 37 lLid8 tLlb5+ 38 �b6 tbd6 39 tbc6 �e6 40 �c7 tLlb5+ 41 �b7 �d7 42 tbe5+ �d6 43 �b6 tbc7 44 a7 lLld5+ 45 �b7 tbc7 46 tbc4+ �d7 47 tLlb6+ �d6 48 lLld5 and the pawn promotes.
1.2:
ltJ+b� v ltJ
defender can force an immediate draw, for example by giving perpet ual check. The following position is typical.
40
+/-
Kling, 1867 The case of the b-pawn differs sig nificantly from the previous section. The knight is notoriously helpless against an edge pawn, and not sur prisingly the defender's drawing chances increase as the pawn is moved towards the centre of the board. There is also the concrete point that a lone knight can stop a b pawn on the seventh rank, even if it is supported by the king (in the ab sence of the enemy knight, of course). It follows that knight sacri fices are much less likely to succeed and they play a relatively small role in this section. Despite the improved drawing prospects, an advanced pawn still commands a great deal of respect. When the pawn stands on the sev enth rank, supported by the king, then the position is won except if the
(40): White can drive Black's knight away from d7 by attacking it with his own knight. This forces Black to reply ...tbe5, in order to meet b8W by ...tbc6+. Then White plays �a8, forcing ...tbc6. Black no longer has the possibility of a fork, so a further knight attack on c6 forces promotion of the pawn. One solution runs l tLlb4 (not l tLlb8? tbc5!) �c5 ( l .. .�c7 2 lLld5+ �d6 3 tbf6 tbe5 4 bSW+ is check, while 1 ...�e6 2 lLid3 is the same as the main line) 2 lLid3+ �d5 (or 2 ...�c4 3 lLif4 �c5 4 tbg6, again heading for f8) 3 lLif4+ (this is not the only winning plan; 3 tLlb2 �c6 4 tba4 �b5 5 tLlb6 tbe5 6 �a8 tLlc6 7 tbd7 is equally effective) �d6 4 tbg6 �d5 (4 ... �c5 5 tbf8 tbe5 6 �as tbc6 7 tbe6+ �d6 8 lLid8 also wins) 5 lLlf8 tbe5 6 �as tbc6 7 tbd7 �d6 S tLlb6
.!LJ+� v .!LJ 39
�c7 9 .!Od5+, followed by 10 .!Ob4, and the pawn advances. Exceptions can only occur if the attacker's forces are badly placed in lbe initial position:
• • • • . - . � • • • • • • • • • • • • • • • • . -�· · · . � � � ,, � � . .
41
�
=I+
Mamedov, 1979 (end ofstudy) Comm., �Shakhmaty 1979 (41): In this position the condi tions are especially favourable for the defender. His king is close to the enemy pawn, while the enemy knight is far away. Nevertheless, White only just draws: 1 .!Df5 ! (threat 2 .!Oh4) 'iPhl ( 1 ...'iPh2 2 .!Oh4 draws at once) 2 .!Og3+ ! 'iPh2 3 �1 + ! 'iPh3 4 'iPf2 ! (just in time) �5 (to prevent White's threat of �3) 5 'iPgl (5 .!Od2 .!Df4 6 .!Of3 ! also draws, but the composer's line is more attractive) .!Df4 6 .!Og3 ! (this is a position of reciprocal zugzwang; for the White to play analysis, see the following diagram) 'iPxg3 stalemate. Conversely, there are exceptional positions in which the attacker can win, even though the defender's king is in front of the pawn. Once again, a
miserable initial position is usually to blame.
• •ttJ••• m m m B • • ••• • • • = • • • • • • • • • • • • • • • • +I=
42
Olmutsky, 1960 (end of study) Revista Romana de Sah (42): Black to play draws in vari ous ways, for example by 1 ...'iPf7 2 'iPh6 .!Oe7 3 .!LJ
...
40 .!LJ+�
v
.!LJ
again) 5 .!Oc6 ! (threat .!Oe7) .!Oe6 (5 ....!Dd7 6 .!Oe5+ ! ) 6 .!Dd8+ ! .!Oxd8 7 'iii>h7 ! and the pawn promotes. One interesting case arises when White's knight is on b8. In this case both sides have to watch out for dan gerous checking possibilities, and quite subtle play can arise. Here is an example:
White to move may try: 1) 1 .!0d7? (certainly not 1 'iti>g3? .!Of5+ 2 �g4 .!Od6 !) .!Oc6 ! (the prob lem now is that White's king cannot move to e6 or f7 because of ....!Od8+, so Black's knight has set up a bar rier) 2 �g4 �e3 3 �f5 'iii>d4 ! 4 .!Oe5 (4 'it>e6 .!Od8+) .!Ob8 ! 5 'it>e6 'it>c5 ! 6 .!Od7+ �c6 and the pawn falls. 2) 1 �g4! (intending 2 �h5 and 3 �g6, but also threatening 2 .!Od7 .!Oc6 3 'ifi>f4) and now: 2a) 1 �d3 2 .!Oa6 ! .!Oc6 3 .!Ob4+ ! is the typical tactical point men tioned above. 2b) 1. ..'it>d2 2 .!Od7 ! (not 2 �f4? .!Oe6+ ! 3 �e5 .!Oc5, while 2 �h5? 'iti>c3 ! 3 �g6 'iti>c4 ! leads to a draw af ter 4 �f6 'it>b5 ! 5 .!Od7 .!Oc6 ! or 4 .!Od7 .!Oc6 ! 5 tLJe5+ .!Oxe5+ !) .!Oc6 3 'ifi>f4 �c3 (Black is unable to play 3 . . . �d3, so White's king can pene trate to e4 and d5) 4 'iti>e4 ! 'it>b4 5 'iti>d5 ! �b5 6 'iti>d6 ! .!Ob8 7 'iti>c7 ! .!Oa6+ 8 'iti>c8 'iti>c4 9 .!Ob6+ �b5 1 0 .!Oa8 �c6 1 1 .!Oc7 and wins. 2c) 1 'iti>e3 (meets the threat of 2 .!Od7, but the king is heading into a dead end) 2 'iti>h5 (not 2 �g5? .!Oe6+! 3 'iti>f6 .!Oc5) �e4 (or 2 ... �d2 3 'iti>g6! 'iti>c3 4 'it>f6, when White wins after 4 . . . 'it>c4 5 .!Od7 .!Oc6 6 .!Oe5 + ! or 4 ...'iti>b4 5 .!Oa6+) 3 'it>g6 and Black's king cannot move towards the pawn, for example 3 . . . 'iti>d5 4 .!Oa6 ! .!Oc6 5 .!Ob4+ ! wins. Thus White's king has time to cross to the queenside, with a simple win. There are 279 positions of recip rocal zugzwang in the ending of .!LJ+�b7 v .!O. Here is one of the most interesting: •••
43
+/=
Fritz, 1974 Magyar Sakkvilag (43): The squares g3, f4 and g5 are out of bounds to the white king, because Black can give a knight check and then attack the pawn. Thus White's main threat is 'it>g4-h5g6-f6, eventually supporting the pawn. This might appear rather slow, but Black is also restricted because 1 . . . 'iii>d 3, for example, is met by 2 .!Oa6 .!Oc6 3 .!Ob4+. The squares d5 and e5 are similarly mined. Black to play has enough time to approach White's pawn, for example l . . .'iti>d2 2 'it>g4 'it>c3 3 .!Od7 .!Oc6 4 'iii>f5 'iii>b4 5 'iii>e4 'iii>b5 6 'iii>d5 .!Ob8 ! 7 'it>d6 'iti>a6 draws.
.•.
lll +� v lLJ 41
Now suppose that the pawn is on b6. If Black's king is trapped in the corner by the enemy king, then the position is lost except if Black has an immediate perpetual check.
=I-
44
Original (44): Black to move must either allow 'ili>b5, or else play l ...�c5, which blocks c5 and therefore al lows 'ili>a6 by White. The main line runs 1 . . .�d4 ( I .. . 'ili>d5 2 llld3 'ili>c6 3 �a6 ! and l ...'ili>c5 2 'ili>a6 'ili>c6 3 'ili>a7 �b5 4 llld 3 both lead to a similar conclusion) 2 'ili>b5 'ili>d5 (or 2 ... lllb8 3 �b6 'ili>c4 4 �c7 llla6+ 5 'ili>c8 'ili>c5 6 llld 3+ 'ili>b5 7 lllf4, heading for c7, and White wins) 3 llld3 (cover ing c5, and thereby threatening to play 'ili>a6) 'ili>d6 (3 ... lllb8 4 'ili>b6 'ili>d6 S �5 lllc6 6 llla6 'ili>d7 7 lllb4 lllb8 8 �a7) 4 'ili>a6 ! 'ili>d5 (4...'ili>c7 5 'ili>a7 ! wins) 5 �a7 'ili>c4 6 lllf4 'ili>c5 7 lllg6 wins by heading for f8, as in diagram 40. When White is to play, he can only move his knight, but after 1 �2 (or 1 llla2 'ili>d5 2 lllb4+ 'ili>d6) the knight loses contact with d3, and after 1 . . .'ili>d5 ! 2 'ili>b5 (or else 2 . .'ili>c6 and White cannot reply 'ili>a6 because the square c5 is not covered) 'ili>d6 ! followed by ...'ili>c7 Black draws eas ily. .
45
+/-
Averbakh, 1955 (version) (45): If Black is to move then White wins easily by 1 ...lllg7 2 llld6 lllf5 3 'ili>f8, so White's task is to transfer the move to Black. He can not lose a move with his knight, so the win involves triangulating with the king. Playing 'ili>f8-e7-f7 leaves g8 undefended for a moment, so White must first manoeuvre his knight so as to drive Black's king back into the corner: 1 llld 6 lllg7 2 llle4 llle8 3 'ili>f8 (now White is ready to meet ... 'ili>g8 by lllf6+, so the trian gulation can begin) lllg7 4 �e7 'ili>g8 (there is nothing better, since after 4 . . . lllf5+ 5 'ili>f7, Black has only one more check) 5 lllf6+ ! 'ili>h8 6 �f7 lllf5 7 llle4 lllg7 8 llld6 llle 8 9 'ili>f8 and wins. In general, positions analogous to Reti's diagram 1 2 are drawn with a
42 lll + � v lll
b-pawn. The following position re veals the reason why:
llld4+ �c4 ! 5 lllc6 ll'lb3 ! 6 b7 lllc5 ! is also sufficient for a draw. The results are similar if White moves first, for example after 1 ll'lf7 Black should not delay with the transfer of his knight to a5, since 1 . . . �c4? loses to 2 llle5+ ! �d4 3 ll'ld3 ll'lb3 4 �b5 ! followed by b7. After 1 .. .ll'lb3 !, on the other hand, play is essentially the same as when Black moves first.
=I=
46
Original (46): Suppose that Black is to play. The key to the defence is to transfer the knight from the insecure square c5, where it is attacked by White's king, to a more stable defen sive position on a5. If White were able to match this manoeuvre by transferring his king to a6 then he might still win, but this proves to be impossible. Thus Black reaches a position with Bllla5 v W�c7, when the knight cannot be dislodged, for example 1 .. .ll'lb3 ! (the only move since after 1 . . .�c4? 2 llle4 ! the lines 2 . . . lllb 3 3 ll'ld2+ !, 2 . . .llle6 3 �d7 ! �b5 4 b7 ! and 2...llla6 3 b7 ll'lb8+ 4 �c7 ll'la6+ 5 �c8 are all winning for White) 2 llle6 (2 llle4 llla5+ ! 3 �c7 �b5 is similar, while if White plays 2 �b7, intending to meet 2 ...llla5+? with 3 �a6 ! , then Black just goes back with 2 . . . lllc5+) llla5+ 3 �c7 and now Black's simplest defence is 3 ...�c4 4 llld8 �b5, but 3 ...�b5 4
=I-
47
Original (47): This situation could not arise with an a-pawn (but it could have arisen from the previous dia gram after 1 . . .ll'lb3 ! 2 �b7 llla5+? 3 �a6 !). It is reciprocal zugzwang. White's king is paralysed because 1 �a7 is met by 1 ...�b5 2 ll'le6 lllc6+ 3 �b7 llla5+ 4 �c7, reaching the drawn formation of the previous dia gram. On the other hand 1 llle6 is met by 1 ...�a4 !, when 2 �a7 fails to 2 . . . lllc4 3 b7 llla5 ! . Since the parity is wrong, White cannot prevent Black oscillating between a4 and b4. Of course, if Black moves first the
tei+l!.
parity works against him and White wins by means of 1 .. .�a4 ( 1 .. .lCib3 2 �e6 lCia5 3 tC!f4 is the same) 2 tC!e6 �b4 3 tC!f4 tC!c6 (or 3 . . . �a4 4 tC!d3 �6 5 b7 !) 4 b7 ! �c5 5 tC!d3+ �d5 6 �b5 �d6 7 �b6 ! lCib8 8 teics tC!c6 9 tC!a6 �d7 10 tbb4. Here is a practical example of the above situation:
48
=I+
v
tCi 43
move played doesn't spoil anything) �f2 1 3 �g6? (a horrible blunder; 1 3 �h5? tC!f5 1 4 lCig6 �e3 is also lost, but after 13 �f6! Black would be un able to make progress as 13 ...tC!g2 is met by 14 tbf5) lCig2 ! 0- 1 , because 14 tbf5 tbf4+ followed by 15 . . . g2 wins. To summarise, when Black's king approaches the pawn from behind, W�c6 v BtC!c5 is generally drawn, W�c7 v BtC!a5 and W�a7 v BtC!c5 are almost always drawn, but W�a6 v BteiaS depends on parity. The situation in which Black's king tries to approach the pawn from the side is more complex. If White can interpose his own king between the enemy king and the pawn, then he has good winning chances, but extremely accurate play may be re quired, as in the following position:
Gillen-Harding Dublin Telecom 1991 (48): After 1 lllli4 ! we have the re ciprocal zugzwang situation from the preceding diagram. Black tries manoeuvring with his knight, but he cannot change the parity: 1 . . .lCid4 2 �g6 (2 �g5 is equally good) lCie6 3 �h4! (again reciprocal zugzwang) �f4+ 4 �g5 ! tbe6+ 5 �h5 ! tbc5 6 �g5 (now White uses his king in stead) tbe4+ 7 �h5 ! tC!f2 8 �g5 � 1 9 �h5 tC!e3 10 �g5 (finally Black decides to move his king round to f2, but we know from dia gram 46 that this doesn't help) �h2 1 1 �h5 �gl 12 �g5 ( 1 2 �g6 �f2 1 3 �g5 is perhaps simpler, but the
+I=
49
Original (49): The winning continuation is both confusing and complex. In the course of the play White repeat edly blocks the b-pawn with his king:
44 lb+8 v lb
1 �d5 ! �e7 (the most natural and the best defence; White wins more easily after 1 . . .llla5 2 lbc6 lllb7 3 lbd4 �e7 4 �c6! llla5+ 5 �c7 �f6 6 lbc6 lllb3 7 �d6 or 1 . . .llld8 2 �d6 �f5 3 1;c7 llle6+ 4 �c8 lllc5 5 llld 3) 2 �c6! llld6 (2 ...llld8+ 3 �c7 llle6+ 4 �b8 transposes into line 3 below at move 14) 3 1;c7 ! (threaten ing either 4 lllc4 or 4 lllf7 ) lllb5+ (3 . . . llle 8+ 4 �b7 llld6+ is identi cal, while after 3 . . . llle4 4 llld3 llld6 5 lbc5 lllb 5+ 6 �b7 �d6 7 �a8 lllc7+ 8 �b8 llld5 9 llle4+ ! �c6 the pawn advances and White wins by 1 0 b7 ! �b6 1 1 lllc 3 lllb4 12 �a8 llla6 13 llld5+ �b5 14 �a7) 4 �b7 ! (the first paradoxical obstruction of the pawn; after 4 �b8 llld6 White must return to c7) llld6+ (there is nothing better; 4 . . . �d6 loses to 5 lbc4+ 1;d5 6 �a8 �c6 7 b7 ! lllc7+ 8 �a7 lllb5+ 9 1;b8 lllc7 10 llle5+ �b6 1 1 llld7+ �c6 12 1;c8, and 4 ...1;e6 5 lllc4 leads to the same con tinuation) 5 1;a6! and now: 1 ) 5 �e8 6 lllc4 ! llle4 7 1;a7 lllc5 8 llle5 ! �e7 9 �a8 ! 1;d6 10 llld3 ! �c6 1 1 �a7 ! llld7 12 b7 ! and wins. 2) 5 1;e6 (Black cannot im prove his position, so he just waits) 6 lllc4 ! llle4 7 �b7 ! (the second self obstruction) �d7 (7 . . . �d5 8 �a8 ! lbc5 9 llla5 ! �d6 10 lllb3 ! �c6 1 1 �a7 ! transposes) 8 �a7 (the remain der of this variation depends heavily on reciprocal zugzwangs) lllc5 9 llla5 ! (the first reciprocal zugzwang) �c8 (9 . . . �e6 10 �a8 ! �d7 1 1 lllb3 �c6 12 �a7 ! llld7 13 b7 ! wins) 10 lllb3 lllb7 11 �a8! (we have arrived
at the second reciprocal zugzwang) llld8 12 lbc5 ! (and the third) lllc6 1 3 b7+ 1;c7 1 4 llla6+ 1;b6 1 5 lllb4 and the pawn slips through. 3) 5 lbe4 (transferring the black knight to c5 puts up the greatest re sistance) 6 lllc4 ! lllc5+ (6 ... �e6 7 �b7 ! as in line 2) 7 �b5 ! lllb7 8 �c6 ! llld8+ 9 �c7 ! llle6+ 10 1;b7 ! (once again the king has to step in the way) lllc5+ ( 1 0. . .1;f6 1 1 1;c6 �e7 12 llle5 llld8+ 13 �c7 llle6+ 14 1;b8 transposes to the main line, while 10 . . .�d8 1 1 llle5 lllc5+ 1 2 1;a7 1;e7 1 3 1;a8 !, followed by llld 3, advances the pawn) 1 1 1;c6! (the 1 1 th consecutive 'only move') llle6 12 llle 5 llld8+ 1 3 �c7 llle6+ 14 �b8 lllc5 (or 14 ...�d6 1 5 1;a7 ! lllc5 16 llld3 !) 1 5 �a8 ! 1;d6 16 llld3 ! 1;c6 17 �a7 ! llld7 18 b7 ! and wins as in diagram 40. The following practical example is rather similar to the preceding po sition. ...
...
...
-!+
50
Petrov-Aronin Cheboksary 1950
lb+�
(50): It was White to play in the game, but despite this the position is lost: 1 lbg2 (1 'iPb4 'iPe4 2 lbg2 lbf7 3 'iPc3 lbe5 covers d3, so that White does not have lbe l + and lbd3+ in response to . . . 'it>f3; Black wins after 4 'iPd2 'it>f3 ! 5 lbh4+ 'iPf2 6 'iPc2 �f3) lbf5 2 'iPc4 'iPe4 ! 3 'iPc3 'iPf3 (3 . . . lbd4 is exactly the previous dia gram, but the move played is a much simpler win) 4 lbel + 'it>e2 5 lbg2 �fl ? (a serious error after which Black's king gets stuck on f1 ; he could have won by 5 . . . 'iPf2 6 lbf4 �f3 7 lbd3 'iPe3 8 lbe l lbh4 9 'iPc2 �e2 10 lbd3 lbg6 1 1 lbc l + 'it>e3, followed by the pawn's advance) 6 �f4 ! lbe7 (Black cannot improve his king position because 6 . . .'it>f2 is met by 7 lbh5 ! g2 8 lbf4 ! , while 6. . . 'iPe l 7 'it>d3 leads nowhere) 7 �d2? (White could have drawn with accurate play: 7 'iPd3 lbg6 8 lbd5 ! �f2 9 lbe3 ! lbe5+ 10 'it>e4 ! lbc4 1 1 �5 ! and now l 1 .. .g2 12 lbf4 ! is an immediate draw, while I l . ..lbd2+ 1 2 'iPf5 leads to diagram 46) lbg6 ! (Black makes no mistake the second time round) 8 lbh3 (losing at once, but even 8 lbd5 'iPf2 9 lbe3 lbe5 would have lost after 10 lbd5 'iPf3 or 1 0 lbd 1 + �f3 1 1 lbe3 lbc4+ 12 �xc4 g2) g2 9 'iPe3 lbe5 0- 1 . Finally, one should always watch out for unexpected tactical points: (51): After 1 lbe4 'it>e2 ( l ...'it>el 2 �5 ! b2 3 lbd3+ is an immediate draw, while after l .. .b2 2 lbd2+! �e2 3 lbbl ! 'iPd3 4 'iPa4 'iPc2 5 0.a3+! 'iPc l 6 'iPb3 ! Black's knight is trapped) the situation appears hope less, but White saves the day with a
51
v
lb 45
=I+ Korteling, 1938 Iijdschrift v. d. KNSB
surprising defence: 2 'iPa4 ! b2 3 'it>a3 ! bl'lli' 4 lbc3+! lbxc3 stalemate. There are 492 positions of recip rocal zugzwang in the ending of lb+�b6 v lb. Here is a simple but amusing example from this collec tion.
•.• . . � .. � BttJB B B �·� . . . • • • • • • • • • • • • •• • • • • • • • =!-
52
Original (52): It isn't immediately obvious that Black can't move his king, but in fact l . . .'iPa3 (or l .. .'iPb2) fails to
46 lb+L!,
v
lb
2 llla5 llld6 (or 2 . . .lllf6 3 lllc 4+! �b4 4 llle 5 !) 3 lllc 4+! and 1 . ..�al (or 1 . . .�bl ) takes the king one square further away, losing after 2 lllc5 llld6 3 �g7! �b2 4 �f6 ! �a3 5 1Pe6 ! . Finally, 1 . ..lllf6 2 lllc5 ! lets the pawn through immediately. With White to play, a king move allows 1 .. .lllf6+ and 2 . . .llld 7, while 1 lllc 5 (or 1 llld8 llld6 ! 2 �g7 �b3 3 �f6 �b4) llld6 ! 2 �g7 �a3 ! 3 �f6 �b4 ! is a simple draw. We shall need the following ele mentary reciprocal zugzwang posi tion later on.
Just as with the a-pawn, there has been no attempt to systematically in vestigate positions with the pawn further back, so the following selec tion is inevitably somewhat subjec tive. We start with one of the very few serious analytical attempts, by the noted study composer Pogosi ants. It turns out that Pogosiants got virtually every variation wrong, but the positions which arise are so in structive that it is worth spending some time on his position.
+I=
54 =I-
53
Original (53): If Black is to move, he loses after 1 . ..�f8 (or 1 . . .llle l 2 �f6) 2 �f6 ! , because 2 . . .lllf4 may be met by 3 g7 with check. When White is to play, 1 �f6 lllf4 ! , 1 lllh6 llle 3 ! 2 g7 lllg4+ ! and 1 llld6+ �f8 ! 2 �f6 lllf4 are simple draws. This leaves 1 �e4, but then l . . .�f8 leads to a second reciprocal zugzwang; the only new idea is 2 �f3, but then 2 . . . llle l + ! 3 �g4 llld3 ! 4 �g5 lbe5 draws.
Pogosiants, 1977 3rd HM, Bulletin Problemistic (54): We have taken the position after a couple of introductory moves. Black is to play (White to move wins at once by 1 �e6). Pogosiants ana lysed this position in detail and con cluded that White wins, but the position is drawn and the study is un sound. In order to meet the threat of �e6-f7, Black's first moves are forced. 1 ... 2 �e6
llle7+! lllg6!
tll + t::. v tll 4 7
Now there are two main lines, 3 �f6 and 3 ltJc6.
A)
3 �f6
Pogosiants believed that this move throws away the win, but the position is drawn in any case.
3
...
tllf4?
Here the composer's move is a mistake. The best defence against the formation with W�f6 and W£::.g5 is to have the black knight on h4. Then White can only dislodge the knight by playing his own knight to f3 or f5 (not to g6 because of ...lllf3). Black will reply to either of these moves with . . . tllg 2, when the pawn still cannot advance because Black can play . . . lllf4 and ... tllh5+. There fore, White's best chance is to play tllf5 to force . . . tllg 2, and then con tinue with �e5. This seals Black's knight out of the game; the crucial question then is whether Black's king is close enough to catch the pawn. After 3 . . . tllh4 ! (3 . . . lllf8? 4 �f7 ! loses at once) 4 tllc4 �b7 5 tlld6+ (5 tlle5 �c7 6 lllf3 tllg 2! 7 �e5 �d7 8 g6 �e7 draws comfort ably) �c7 6 lllf5 tllg 2! 7 �e5 �d7 8 g6 we reach a situation in which dia gram 53 is of value, since we already know that 8 ... �e8 ! puts White in zugzwang and draws.
4 �fS!
4 tllc 6? does nothing to meet the threat of 4 . . .tllh 3! 5 g6 tllf4! 6 g7 tllh5+ !. 4 ... tllg2 The alternatives are: 1 ) 4...tllhS 5 g6 'iiib8 6 �g5 tllg7 7 �f6 lllh5+ 8 �f7 �c7 9 tllc 4! �d7 10 tlle5+ �d6 1 1 tllg4 �c7 1 2
tllf6 tllg 3 1 3 �e6 and White i s win ning. 2) 4...llJdS and now: 2a) 5 tllc6? (Pogosiants' move doesn't win) �b7 ! 6 tlld4 �c7 (but not 6...ltJc7? 9 �e5 tlle8 8 �e6 �c7 9 �e7 tlld6 10 g6 winning) 7 �e5 tlle7 ! 8 tllf5 tllg6+ ! (8 . . . �d7? 9 tllxe7 ! �xe7 10 g6!) 9 �f6 tllf4 ! and here the composer's analysis simply stops with the implication that White is winning, although the position is drawn. 2b) 5 �eS! tlle7 6 tllc6 ! (an at tractive winning move) tllg6+ 7 �f6 ! tllf4 8 �f5 ! tllh5 9 llld4 �b7 10 tlle6 (threat �g4) tllg 3+ 1 1 �g4 and the pawn moves forward.
5 tllc4? Here Pogosiants missed a win by 5 �g4 ! tlle 3+ 6 �f4 ! tllg2+ (the continuation 6 ... tlld5+ 7 'iti>e5 ! tlle7 8 tllc6 ! is line 2b of the previous note) 7 �g3 ! tlle3 8 g6 lllf5+ 9 �f4 tllg7 10 �e5 lllh5 1 1 tllc4 �b7 1 2 tlld6+ �c7 1 3 tlle4 �d8 14 tllg 3 tllg7 15 �f6 tlle8+ 16 �e6. After 5 tllc4?, Pogosiants only an alysed the extremely weak defence 5 ... �b8?, when White wins immedi ately by 6 �g4! �c7 7 g6. The cor rect line is 5 ... tllh4+ ! 6 �f6 (6 �g4 tllg6 !) �b7 7 tlld6+ �c6 8 lllf5 tllg 2 !, transposing into the note to Black's third move.
B)
3 tllc6
In this alternative winning at tempt White tries to improve his knight position before closing in with his king. Pogosiants believed that this move would lead to victory. 'iti>b7 3 ...
48 lLJ+.0.
v
lLJ
Not 3 . . . lLJf4+? 4 �f5 lLJd5 5 g6 with a simple win.
4 rM7
lDh4
The composer's main line is enough to draw, and there was even a second adequate defence: 4 . . .lDf4 5 lDd8+ and now:
Not 8 . . . lLJg2? 9 g6 ! lDf4 10 g7 !, and h5 is covered. 9 lLJfS lLJg2! 10 'it>eS
56
55
I=
1 ) S...�c8? 6 lLJe6 ! lDh5 7 �e8 �b7 8 g6 �c8 9 rl;e7 lLJg3 10 lDd4 lLJh5 1 1 lLJe2 rl;b7 1 2 �f7 'it>c7 1 3 lLJf4 and White wins. 2) Pogosiants didn' t analyse the drawing move S ... �b6!. The point is that after 6 lDe6 lDg2 ! White cannot play 7 g6 (7 �f6 lDh4 ! is no help) because of7 . . . lDh4 ! 8 g7 lDf5 ! . This line didn't work after 5 . . . �c8? be cause then the pawn promoted with check. s lLJd4 �c8 6 'it>f6 lLJg2 B lack can also draw by 6. . . �d8 7 lLJf5 (7 lDf3 lLJg2! 8 g6 lDf4 ! 9 g7 lLJh5+) lLJg2 ! transposing into the main line. 7 lLJe2 lDh4! �d8 8 lLJg3
=I=
The critical moment arrives. Po gosjants only analysed 1 0 . . . lDe l ? 1 1 g6 ! lLJd3+ 1 2 �e4 ! lDc5+ 1 3 �d5 ! lLJd7 1 4 g7 lDf6+ 1 5 �e6 and 10. . . �e8? 1 1 g6 ! �f8 1 2 �f6; he therefore concluded that White wins. However, we have the advantage of knowing the reciprocal zugzwang of diagram 5 3 . From this it is obvi ous that 10 ...�e8? is a serious error, because it walks straight into the zugzwang. Black can draw by a sub tle triangulation: 10 . . . �d7 ! (waiting for the pawn to advance) 1 1 g6 (there is nothing better) and only now 1 1 ...�e8 !, when it is White who falls into zugzwang. The following three positions rep resent a personal selection. In the first White's king must manoeuvre with care so as to reach the critical reciprocal zugzwang with Black to move.
lb+l!.
...
...
+I=
Original (57): Black to play draws easily by 1 . . .�d3 heading for White's pawn, so we can assume that White is to play. The winning line runs 1 �d6 ! (not l lllf6? �d3 nor l lllc3+? �d3 and B lack's king is too close to the pawn) lllc4+ (even after the infe rior defence 1 ...�d3 White must still play accurately: 2 �c6 ! lllc4 3 llld6 lbe5+ 4 �c7 �d4 5 b6! llld3 6 �c6 llle5+ 7 �b5 �d5 8 b7 llld7 9 lllc 8 �e6 10 �c6 with a simple win) 2 �d5 ! (2 �c5? appears more natural, but after 2 ...�d3 ! White arrives at the reciprocal zugzwang with the wrong player to move; Black can hold the draw after 3 llld6 lllb2 ! 4 �b4 �d4 ! 5 b6 llld3+ ! 6 �b5 lllc5 ! 7 �c6 lllb3 !) �d3 (White wins more easily after 2 . . . llla5 3 b6 �d3 4 �d6! �c3 5 �c5 ! lllb 3+ 6 �b5 ! �d4+ 7 �a6 or 2 . . .lllb6+ 3 �c6 �c4 4 llld 6 llla5+ 5 �c5 lllb 3+ 6 �c4 llld2+ 7 �c3 lllf3 8 lllc4 lllg5 9 b6 llle4+ 1 0 �d4 lllf6 1 1 llle 5) 3 �c5 ! (putting Black in a reciprocal zugzwang) and now:
lb 49
1 ) 3 lllb2 (Black has no king moves, so the knight has to give up control of b6) 4 lllf2 +! ..ti>e2 (4 ...�e3 loses immediately after 5 lbd 1 + ! ) 5 �d4 ! (this retreating move is hard to find) llla4 6 llle 4! and now: la) 6 �el (6 . . . �fl is met the same way) 7 lllc3 lllb6 8 �c5 ! llld7+ 9 �c6 llle5+ 10 �d5 llld7 1 1 llla4 �d2 1 2 �d6 and wins. lb) 6 lllb6 7 llld6 llla4 (7 ...�f3 8 lllc4 ! llla4 9 lllb2 transposes) 8 lllc4 (this position is also recipro cal zugzwang) �f3 (B lack can only move his king) 9 lllb2 lllb6 10 �c5 lllc8 1 1 lllc4 �g4 12 llld6 and the knight is forced away. 1 c) 6 �f3 7 lllc3 lllb6 8 ..ti>c5 llld7+ 9 �c6 llle5+ 10 �d5 llld7 1 1 llla4 �f4 1 2 �d6 llle5 1 3 lllb2 lllf7+ 14 �c7 and the pawn has a free run home. 2) 3 lllaS 4 llld6 lllb 3+ 5 �d5 �c3 6 b6 ! llla5 7 �c5 ! is relatively easy. 3) 3 llles 4 llld6 ! �c3 (the line 4 . . . llld7+ 5 �c6 ! llle5+ 6 �c7 �d4 7 b6! llld3 8 �c6 wins after 8 ... lllc5 9 llle4 lllb 3 10 �b5 �d5 1 1 b7 llld4+ 12 �b6 or 8 . . .llle5+ 9 �b5 transposing into the note to Black's first move) 5 �d5 ! llld7 (5 . . .llld3 6 b6 lllb4+ 7 �c5 ! llla6+ 8 �b5 lllb 8 9 b7) 6 �c6 ! llle5+ 7 �c7 ! �d4 (7 . . . llld3 8 b6 ! �d4 transposes) 8 b6 ! llld3 9 �c6 llle 5+ 10 �b5 (we have once again reached the note to Black's first move) �d5 1 1 b7 llld7 12 lllc8 �e6 1 3 �c6 llle5+ 14 �b6 llld7+ 15 �a7 llle5 16 lllb6 lllc6+ 1 7 �a8 �d6 18 lllc 8+ and White wins. .•.
57
v
...
...
...
50 ttJ+.0. v ttJ
The following diagram depends on various positions of reciprocal zugzwang.
1 �a3? �e2 2 �b3 �d3 are all drawn) and now: 1 ) 1 �dl (trying to head for b4) 2 �c3 �cl 3 �d4 �c2 4 �c5 ! �b3 5 ltJe5 ! (after this Black, rather sur prisingly, can't avoid a fatal zug zwang) �a3 (5 ...�a4 6 ltJc6 results in an immediate reciprocal zug zwang; 5 . . . �a3 prepares to meet 6 ltJc6? by 6 . . . �a4 ! , but White can change track) 6 ltJd7 (now a4 is out of bounds) �b3 7 ltJf6 �a4 (7 ...�a3 8 ltJd5, followed by �c6 and ltJb6, and White wins) 8 ltJd5 ! �a5 9 �c6 and Black is in a terminal re ciprocal zugzwang. 2) 1 �e2 (now White can gain the opposition) 2 �c2 ! (reciprocal zugzwang again; note that 2 ltJb2? fails to 2 . . . ltJb6 3 �b4 ltJd7 ! 4 �c4 �d2 ! 5 �d5 �c3 and Black's king reaches the pawn) �f2 (after 2 ...�f3 3 �d3 ! �f4 4 �d4 ! White can exe cute his plan without interference) 3 �d3 �f3 4 �d4 �e2 5 �c5 �d3 6 ltJd6 and White is threatening 7 �c6, so Black must play 6 . . .ltJc7, but White wins easily once the pawn advances to b6. 3) 1 �f2 (this subtle move ap pears to be the perfect defence, be cause after 2 �c3 ? ltJc7 ! White loses the pawn, while after 2 �c2? �e2! White is on the wrong end of a recip rocal zugzwang; however, the move has a subtle flaw) 2 ltJb2 ! ! (this unique winning move is incredibly hard to find) and now White threat ens simply �c4-c5, so Black must react: 3a) 2 �e2 3 �c4 ! �d2 trans poses to line 3c . ...
...
Original (58): Black to move has a com fortable draw by 1 . . . �e2 2 'iti>b3 'it>d3, so we may assume that White moves first. The first point to note is that White's king isn't very well placed. Playing �a5 is a dead end, because the king cannot advance further. Various squares are out of bounds, for example b4 (because of ...ltJc7 followed by ...ltJd5+), c3, d5 and e3 . This means that the only path for the White king to reach a8 is via b3, c2, d3, d4 and c5. It might seem that the position will depend only on the op position of the two kings, but in this case the position would be a draw because Black can maintain the op position. However, there is a deep fi nesse which tips the balance in White's favour: 1 �b3 ! (this is recip rocal zugzwang; the lines 1 �b4? ltJc7 ! 2 b6 ttJd5+ !, 1 �a5? �e2 and
...
..•
ll.'i+.0. v ltJ 51 3b) 2 ll.'ib6 (now Black's knight cannot move to c7, so b4 becomes possible) 3 �b4 ll.'id7 4 c4 �e3 5 �d5 (in comparison with the line l . . �e2 2 ll.'ib2? ll.'ib6, Black's king 1s one square further away from the queenside, and here White wins after 5 ... �d2 6 �c6) ll.'ib6+ 6 c6 ll.'ic8 7 �c5 cli>f4 8 ll.'ic4 ! and wins. 3c) 2 �e3 3 �c4 ! 'iii>d2 (3 . . .'iii>e4 4 �c5 ) 4 �d4 (this is the most ac curate move; Black's king is kept away from the action long enough for White to improve the position of his knight) �c2 5 ll.'ic4 ! �b3 6 �c5 ! !i:x7 (there is no choice as 6...a4 is impossible) 7 b6! ll.'ie6+ 8 �d6 ! &18 9 �d7 ! ll.'ib7 1 0 c7 ll.'ic5 1 1 �5 'iii>a4 1 2 ll.'id7 ll.'ia6+ 1 3 c8 and White wins. Readers who love paradox will not be disappointed by the following position. ...
.
...
+I=
59
Original (59): As we shall see, Black to play draws comfortably by 1 . . . �g2, so we may assume that White is to
play. At first sight a win appears out of the question, because after the ob vious continuation 1 e5 'iii>g2 2 �d4 �f3 3 �c4 e4 4 'iii>b4 ll.'ib6 ! Black's king is too close to permit White any winning chances. Never theless, the win is there, but it starts with a completely unexpected move - the white king heads directly away from the pawn: 1 �g3 ! (in fact this position is reciprocal zugzwang; 1 'iii>f3? �h2 2 ll.'ie3 �h3 ! transposes to the following note, while 1 ll.'ie3 'iii>h 2 ! also allows the black king out of the box) �gl 2 ll.'ie3 ! (a second reciprocal zugzwang; not 2 �f3? 'iii>h2 3 ll.'ie3 �h3 ! 4 ll.'ic4 'iii>h4 5 ll.'ib2 ll.'ixb2 and Black's knight can stop the pawn) and now Black faces a problem. White's first two moves have bottled Black's king up in the hl -corner, and if Black now plays 2 . . . h l White reveals the main thrust of his play - with the king on g3, the manoeuvre ll.'ic4-b2 is pos sible, since Black's knight cannot stop the pawn after ... ll.'ixb2. Once Black's knight has been displaced to b6, White can head to the queenside with his king because two factors op erate in his favour. Firstly, the king will arrive on c5 with gain of tempo, and secondly the knight on b2 pre vents Black's knight returning to a4, so the knight has to move to a poor square allowing further harassment by White's king. Black's other alter native is to play 2 ... ll.'ib6, pre-empt ing White's knight manoeuvre. Here is the detailed analysis: 1 ) 2 �hl 3 ll.'ic4 �gl 4 ll.'ib2 ll.'ic5 (or 4 . . . ll.'ib6 5 �f3 and White ...
52 tt:'i+.0.
v
ttJ
wins after 5 . . . ti'id5 6 tt:'ia4 �h2 7 'iti>e4 or 5 ... 'iti>h2 6 �e4 �g3 7 �d4 ! 'iti>f4 8 �c5 ti'id7+ 9 �c6 tt:'ie5+ 10 �d6) 5 'iti>f4 (accuracy is required; 5 b6? would throw away the win after 5 . . . �fl ! 6 �f3 �el ! 7 �e3 tt:'ib7 ! 8 �d4 �d2 ! 9 tt:'ic4+ �c2 ! 10 �d5 �c3 ! 1 1 �c6 tt:'id8+ ! 1 2 �c7 tt:'ie6+ ! 1 3 �c8 tt:'ic5 ! 14 tt:'ie5 �b4 ! 15 tt:'id3+ �b5) �f2 6 'iti>e5 ! �e3 7 �d5 ! tt:'ib7 8 b6 �d2 9 �c6! tt:'ia5+ 10 �b5 ! tt:'ib7 1 1 tt:'ic4+ 'iti>d3 (in con trast to the previous bracket, Black's king is on d3 and not c3; this makes all the difference) 1 2 'iti>c6 ! ti'id8+ 1 3 �c7 tt:'ie6+ 1 4 �d6 ti'id8 15 ti'ia5 and White wins. 2) 2 tt:'ib6 3 �f3 ! (now this is the correct plan, because with the knight on b6 White will gain a tempo when his king arrives on c5) tt:'ia4 (or 3 ... 'iti>h2 4 �e4 ! �g3 5 �d4 tt:'ia4 6 tt:'ic4 �f4 7 tt:'ib2 tt:'ib6 8 �c5 win ning) 4 ti'id5 ! (White improves the position of his knight while keep ing his grip on Black's king) �fl 5 �e3 (White can also play 5 �e4, but this allows the additional defence 5 . . . ti'ic5+) �el 6 �d4! (threatening both tt:'ic3 and �c4-b4) 'iti>d2 7 �c4 ! tt:'ib2+ 8 �b3 ! tt:'id3 9 b6 ! tt:'ic5+ 10 �c4 ! (but not 1 0 �b4? tt:'ib7 ! 1 1 �b5 �d3 1 2 �c6 ti'ia5+ and here Black draws) tt:'ib7 1 1 tt:'ib4 ! tt:'id8 1 2 �b5 ! tt:'ib7 1 3 �c6 ! tt:'id8+ 1 4 �c7 tt:'ie6+ 1 5 �d7 tt:'ic5+ 16 �c6 and wms. Apart from the minor dual possi bility at move 5, White's first 1 3 moves were unique. There are 279 positions of recip rocal zugzwang with tt:'i+.0.b5 v tt:'i.
With the pawn further back still, we are in almost totally unexplored territory. The following position at tempts to extend the stalemate of diagram 5 1 , but unfortunately it is unsound.
...
60
=I= J. Van den Ende, 1951 Tijdschrift v. d. KNSB
(60): We have taken the position after a couple of introductory moves. Black continues with 1 . ..g5 ( l ...�f5 2 tt:'ie4 draws) and now the com poser intended 2 'iti>g6 g4 3 'iti>g5 (or 3 �h5) g3 4 �g4 (or 4 �h4) g2 5 �h3 ! g l 'it' (or .l:t) 6 tt:'if3+! tt:'ixf3 stalemate. This is indeed the sim plest draw, but White has an alterna tive method: 2 tt:'ic4+ �f4 3 �f6! g4 4 tt:'ie5 ! g3 5 tt:'ig6+! �g4 6 tt:'ie5+! �h3 (alternatively, 6 ...�h4 7 tt:'ig6+! �h3 8 tt:'if4+ ! 'iti>g4 9 tt:'ig2 ! ti'if5 10 tt:'iel ! tt:'ih4 1 1 �e6 ti'if3 12 tt:'ig2! leads to a draw) 7 tt:'id3 ! tt:'ie2 8 tt:'ie l ! and Black cannot make progress. Although over-the-board exam ples are few, interesting positions do sometimes arise.
ltJ+.0, v ttJ 53
61
=I= Frolov-Psakhis Groningen Open 1993
(61): Although White won the game, the result should have been a draw: 1 ltJd4 �b4 2 ltJe2 'it>c5? (this move leads to disaster; Black could have held the game by either 2 . . .�b5 3 ltJf4 ltJg5 ! 4 'it>f5 ltjf3 5 ltJd3 ltJh2 6 g5 ltJf3 ! or 2 . . . ltJg5 3 'i!tf5 ltJf3 ! 4 �f4 ltJh4 ! 5 g5 �c5) 3 ltJf4 ! ltJf2 (Black cannot play 3 . . . ltJg5 because of 4 ltJe6+, so he cannot prevent the pawn's advance) 4 g5 ! 'iii>c6 5 �e6 �c7 6 g6 ltJg4 7 g7 ltJh6 8 ltJd5+ �d8 9 l0e7 1-0. Here is one of the 91 positions of reciprocal zugzwang with ltJ+.0ib4 v �(62): First of all, suppose that Black is to move: 1 . . .'iii>h2 ( l ...itJb6 2 b5 ltJd5 3 ltJg5 ! h2 4 'iii>f3 ! trans poses to the main line) 2 ltJg5 ! (a second reciprocal zugzwang) ltJb6 (2 . . .ltJf6 3 b5 ltJd5 4 'iti>f3 ! trans poses) 3 b5 ! (another reciprocal zug zwang) ltJa4 4 'it>f3 ! reaches a fourth reciprocal zugzwang, and this time a critical one for understanding the
=I-
62
Original diagram position. Now Black may play: 1) 4 �gl 5 ltJe4 ! �fl 6 �e3 ! �el 7 �d4 ! itJb6 8 �c5 ! ltJa4+ 9 'iii>c6 ltJb2 10 'iii>b6 'it>e2 1 1 �c5 'it>d3 1 2 ltJf2+ transposes to the position after White's 4th move in line 1 of diagram 57. 2) 4 ltJb6 5 ltJe4 (5 'iii>f4 ltJa4 6 ltJe4 is another possibility) 'iii>h 3 (5 ...ltJd5 exposes the knight to attack and allows White to win by 6 �e2 'iii>g2 7 �d3 ! itJb6 8 �d4 'iii>f3 9 ltJc3 'iii>g4 1 0 �c5 ! ltJd7+ 1 1 'it>d6 ltJb6 12 ltJe4 ltJc4+ 1 3 'iii>c5 ltJe5 1 4 ltJd6 ltJd7+ 15 �c6 ltJe5+ 16 �d5 ltJd7 17 ltJc4 �f5 1 8 'iii>d6) 6 'i!tf4 ! ltJa4 (or 6...�g2 7 �e5 'iii>f3 8 itJf6 ! �e2 9 ltJd5 ! ltJa4 10 'i!td4 ! �d2 1 1 �c4 ! itJb2+ 1 2 'iii>b3 ltJd3 1 3 b6 ! ltJc5+ 14 �c4 ! ltJb7 15 ltJb4 ! 'i!te3 1 6 �d5 is winning for White) 7 �f5 ! �h4 (7 ...�g2 8 'i!te5 'i!tf3 9 'i!td4 ! ltJb6 10 ltJc3 loses more quickly) 8 �e6 ! (great care is still necessary; after 8 'i!te5? �g4 ! we arrive at a reciprocal zugzwang with White to play, for ...
...
54 tLJ+,0, v tLi
example 9 �d6 �f5 ! 10 tLlc5 tLlb6 ! 1 1 tLld7 tLic8+ ! 1 2 �c7 tLle7 ! leads to a draw) �g4 (there is nothing better since 8 ... �h5 9 �d6! trans poses) 9 �e5 ! (White's triangulation has put Black in zugzwang) �h5 (9 ... �f3 10 �d4 ! transposes to the note to Black's 7th move, while 9 ... tLlb6 10 �d6 �f4 1 1 �c6! tLlc8 1 2 �c7 wins for White) 1 0 �d6! �g6 1 1 tLlc5 ! tLlb6 12 �c6 ! tLlc4 1 3 tLld3 ! � f7 14 �c5 ! tLl a5 15 b6! �e7 1 6 tLle5 ! tLlb7+ 1 7 �c6! lbd6 trans poses to the position after Black's second move in diagram 49. The logic behind this analysis is somewhat obscure. The main point is that Black's knight is better on a4 than on b6 (because a knight on b6 means that White's later �c6 will gain a tempo), while Black's king would like to edge up the h-file to, say, h4, when it is in a position to make a run for b8. In the position af ter 4 �f3 ! above, Black to play must make a concession, by either playing the knight to b6 or putting his king in reverse gear. When White moves first the re verse logic operates. After 1 b5 (or 1 tLlg5 �h2 ! 2 tLle4 tLlb6 3 �f3 �h3) �h2! 2 tLlg5 tLlb6! and now: 1 ) 3 �e3 �g3 ! 4 �d4 �f4 5 �c5 tLla4+ ! 6 �b4 tLlb6 ! draws; White's knight must stand on e4 for this idea to have a chance of working, or else Black's king can head back via g3. 2) 3 �f3 tLia4 (White is in zug zwang and must allow Black's king extra freedom) 4 tLle4 (4 �f2 tLlb6!) �h3 ! 5 �f4 �h4 ! 6 �f5 �h5 ! 7 �e6 �g6 Black's king is one square
ahead of the corresponding Black to play lines, and this allows him to draw after 8 �d6 �f7 9 tLlc5 tLlb6 10 �c6 tLlc4 1 1 tLld3 �e7 12 tLlb2 tLle5+ 1 3 �c7 tLld7 14 tLlc4 �e6. Here is one of the few practical examples with the pawn on b3:
+I= Gligoric-F. Olafsson Bled/Zagreb/Belgrade Ct 1959 63
(63): Black's knight is very poorly placed, since it can easily be driven away by tLlc4. After 1 b41 Black failed to notice the threat and played l .. .�g5 . The continuation was 2 tLlc4 ! tLlbl 3 �d4 ! 1-0, since 3 ... �f6 4 �d3 �e6 5 �c2 �d5 6 tLle3+ ! wins the knight. However. even after the superior defence 1 . ..tLlb l , White wins all the same by 2 liJd l liJd2 (or 2 . . . ltJa3 3 liJb2 liJbl 4 b5) 3 b5 tLlb3+ 4 �d6 �g6 5 b6 tLla5 6 tLlb2 �f7 7 �d7 �f6 8 tLltl3 �f5 9 �c7 �e6 10 tLlb4 !. Black to play draws by l .. .�g5 2 b4 �f6 3 tLlc4 tLlc2 ! 4 b5 �e7 ! and his king arrives back in time.
llJ+�
The following diagram is based on some reciprocal zugzwangs with �+�b3 v llJ.
v
llJ 55
b5 ! (reciprocal zugzwang 3) liJb6 (after 3 . . . �g2 4 �e2 !, Black can choose between transposing to the main line by 4 . . . llJb6 5 �e3 ! or playing 4 ... �h3 5 �d3 ! �g4 6 �d4 lLib6 7 �c5 ! llJa4+ 8 �c6 lLib2 9 llJd2, when White wins) 4 �d2! (re ciprocal zugzwang number 4) �g2 (4 . . . llJa4 5 �d3 transposes to the main line after 5 ... llJb6 6 �d4 �el 7 �c5 or 5 . . . �el 6 �c4 lLib6+ 7 �c5, while 4 . . . llJc8 is more or less the same) 5 �e3 ! �fl 6 �d4 ! and now: 2a) 6 �e2 7 liJd6 is a rapidly fatal reciprocal zugzwang, for exam ple 7 . . . llJa4 (7 . . . �f3 8 llJc4 ! llJa4 9 lLib2 liJb6 10 �c5) 8 llJc4 reaches yet another reciprocal zugzwang and now any black king move loses control of d3, and therefore Black can no longer meet liJb2 by . . .llJxb2 followed by ...llJd3 and ...llJb4. 2b) 6 �el 7 �c5 ! (not 7 llJd6? �e2 ! and White has fallen into zug zwang) llJa4+ 8 �c6 lLib2 9 �b6 �e2 10 �c5 d3 (or 10 ... llJa4+ 1 1 d4 �f3 1 2 llJc3 liJb6 1 3 c5 lLid7+ 14 �c6 llJe5+ 1 5 �d5 liJd7 16 llJa4) 1 1 liJf2+ transposes to the position after White's 4th move in line 1 of diagram 57. If Black moves first he is no better off, because 1 . . . llJf4 and 1 . . . �g2 are both met by 2 b4 ! , and we quickly transpose into the above analysis. There are a modest 5 1 reciprocal zugzwangs with llJ+�b3 v llJ. ...
+I-
64
Original (64): This position depends on a whole thicket of interwoven recipro cal zugzwangs. White to play may try: 1) 1 b4? (certainly the natural move, but it fails) llJf4 ! (this is a re ciprocal zugzwang with White to play) 2 b5 liJd5 ! (reciprocal zug zwang number 2) 3 �d2 lLib6! reaches reciprocal zugzwang num ber three, and this one is critical for the whole analysis. White to play cannot avoid making a concession to the black king by giving access to either el or f2. After 4 �d3 �e l ! 5 �d4 �d 1 ! 6 �c5 llJa4+ ! 7 �c6 �b2 ! 8 �b6 (8 liJd6 lLia4 !) �c2 ! 9 �a5 lLic4+ ! Black's king is close enough and he draws. 2) 1 b3! lLif4 ( l . . .�g2 2 b4 ! lLif4 transposes to the following note) 2 b4 ! (reciprocal zugzwang 2) lLid5 (2 ... �g2 3 b5 ! liJd5 transposes) 3
•••
Finally, there are two noteworthy studies with the pawn on b2. (65): Black's knight is trapped, and the question is whether or not
56 lLJ+l1 v lLJ
B•B B B B B B B B B B B � B B B B B B B BtiJB B B �� • . • . . �� B•B B B :0 ,, ri
+I=
65
Halberstadt, 1951 Suomen Shakki White can capture it with his king. Rather surprisingly, the answer does not depend on the opposition. White cannot win by playing �b4-c4-d3c2, because Black's king simply ad vances to b4. The key to success is the transfer of White's knight to an other square controlling d2, namely e4; this maintains the grip on Black's knight, while opening up an easier route (via b3) for White's king. White can play: 1 ) 1 b5? �c7 ! (this is recipro cal zugzwang) 2 �c5 �b7 ! (this is too; White cannot play lDc5+ and lDe4 because his king is blocking c5) 3 lLJa5+ (White transfers his knight to c4, but this is less effective) a6 4 lDc4 �b7 ! 5 lLJd6+ (it seems that the transfer to e4 has been ac complished, but in this position Black can advance his king) 'ilta6 ! 6 lLJe4 �a5 ! 7 �c4 lDa3+ ! and the knight escapes. 2) 1 'ilta4? �d7 2 �b4 'iltc6 ! and White cannot make progress.
3) 1 �b6? �b8 ! 2 �c6 �c8 3 �d5 �c7 ! 4 �c4 �b6! and again White cannot set up the winning for mation. 4) 1 �b4! (the key move; the threat of �c4-d3 forces Black's king to advance, and then White can choose the appropriate square on the fifth rank for his own king) �c7 (af ter l . ..'iltb7 2 lDc5+ �b6 3 lDe4 ! White wins more quickly) 2 �b5 ! (now Black is in zugzwang) �b7 (or 2 . . .�d6 3 �c4 ! �e5 4 �d3 ! �d5 5 �c2 and Black's king is one square too far away from b4) 3 lDc5+! �c7 4 lDe4 ! (White has transferred his knight without allowing Black's king to become too active) �d7 5 �b4 'ili>e6 6 �b3 �e5 7 �a2 ! (but not 7 �c2? lDa3+ !) �xe4 8 �xbl ! �d4 9 �a2 ! �c5 10 �a3 ! and wins.
+I= 66 Voja and Nestorescu, 1951 1/2 Pr., Revista Romana de Sah
(66): Black to play draws easily enough by l ...lDd6+, so suppose that White moves first. There are two al ternatives:
lll +� v lLJ 57
1 ) 1 g4? and now: l a) 1...Wc4? (the composers give this as drawing, but in fact it loses) 2 g5 (2 'ite5 lllc5 3 g5 ! is also possi ble) lllc 5+ 3 'ite5 ! llld7+ 4 'itd6! (not 4 'itf5? lllf8 ! 5 lllg4 'itd5 6 'Of6+ 'itd6 ! 7 lllh7 llle6 8 g6 'ite7 with a draw) lllf8 5 We7 ! lllg6+ 6 �6! lllf4 7 'itf5 and now: l a l ) 7...lLJds 8 lllg 8 'itd4 9 g6! 'Oe3+ 1 0 Wg5 llld5 l l lllf6 llle7 12 g7 ! 'ite5 13 lllg4+ ! 'it>e6 14 lllh6 ! (this is reciprocal zugzwang) 'ite5 1 5 /Of5 lllg8 1 6 Wg6 winning. la2) 7 lllg2 8 'itg4 llle 3+ 9 'itf3 00 10 g6 llle7 1 1 g7 ! 'itd5 12 'itf4 :iie6 1 3 Wg5 ! and wins. l a3) 7...lllh S (the two composers stopped their analysis here, assum ing that the position is draw) 8 We4 (Black can neither prevent g6 nor bring his king closer) lllg7 (8 ... 'itc5 9 lllf5, followed by Wf3-g4, wins the knight) 9 g6 lllh 5 1 0 We5 'it>c5 1 1 �f5 , followed by lllg 3 and White wins. l b) 1...llld6+ (this move rescues lhe study) 2 'ite5 lllc 4+! 3 We6 'itd4 4 g5 llle5 ! 5 lllf7 lllg6 ! 6 Wf6 lllh4 7 �6 Wd5 8 lllf5 lllg 2! and Black draws. 2) t lLJr7? lllc5+ 2 Wf5 Wd4 3 g4 :iid5 4 g5 llle6 draws. 3) 1 WdS ! 'itd3 2 g4 ! (not 2 lllf5? �8 ! 3 'ite5 lllf7 + 4 'itf6 'ite4 ! 5 �g3+ We3 !) 'ite3 3 g5 ! 'itf4 4 g6 ! �6 (Black is reduced to a desperate defence) 5 'itxd6! (5 g7? llle8 !) 'itg5 6 g7 ! Wf6 7 lllf5 ! (the only move; 7 g81i'? and 7 g8l:t? deliver stalemate, while 7 g8.t? 'itg7 loses a piece) and White wins. •••
There are 85 positions of recipro cal zugzwang with lll+�b2 v lll . For reasons of space, I will just mention one of the more interesting examples and leave the proof as an exercise for the reader: WWc6, lllg2, �b2 v BWd3, llla8. Perhaps surprisingly, the longest win with lll+b� v lLJ starts with the pawn on b3 and not b2, as one might expect. The situation is similar with the c- and d-pawns. The vast major ity of positions with the pawn on b2 are drawn, so the positions which are won tend to be special situations in which White has an immediate win for some reason. The longest win (to pawn promotion or knight capture) is actually 42 moves:
+I=
67
Original (67): With Black to play, any legal move draws, but perhaps the sim plest is 1 . . . 'itg3 2 b4 Wf4 bringing the king back to the defence. White to play must continue very precisely in order to win:
58 lb+�
v
lb
1 �c3 ! (not 1 b4? lbf2 and now 2 b5 lbe4 ! 3 b6 lbd6 ! loses the pawn, while 2 �c3 lbe4+ ! 3 �d4 lbd6! 4 lbd7 �g3 5 lbf6 �f4 is a simple draw) lbg3 ( l . . .lbf2 2 �d4 is much worse) 2 �d4 ! lbf5+ (2 . . . lbe2+ 3 �c4 ! transposes to the main line) 3 �c5 ! lbg3 4 �d5 ! lLle2 5 �c4 ! lbg3 6 lbc6 ! (not 6 b4? lbe4 !) lbe4 7 �d4 ! lDe5 1 2 b5 lbd6 1 3 b6 lbc4 is a draw; 10 �c6? is also met by 10...�f4) lLic7 1 1 b4 ! and now: 1) 11 �f3 12 �d6! lbe8+ 1 3 ..t>c6 ! (compared to the line 1 0 b4? �f4 ! , Black's king has been de flected from f4 to f3, depriving him of the defence with ...�e5 and . . . lbd6) ..t>e4 14 b5 ! �e5 15 lbc8 ! lbg7 l 6 lbb6 ! (another tricky move; 1 6 b6? lbe6 ! is only a draw) lbf5 ( 1 6 . . . lbe6 17 lbd7+! �f5 18 b6 is quite hopeless, while after 16 . . .�e6 17 lbc4 ! Black can choose between 17 ...lbf5 transposing to the main line or 17 ... �e7 18 b6 lbe6 1 9 lbe5 lbd8+ 20 �c7 lbe6+ 1 9 �b8 lbc5 22 �a8 ! �d8 23 lbd3 winning for White) l 7 lbc4+! ..t>e6 1 8 b6 lbe7+ 19 ..t>b5 ! �d7 20 lbe5+! �c8 1 9 �a6! 'iti>d8 2 2 lbf7+! �c8 2 3 �a7 lbc6+ 24 �a8 ! lba5 25 lLie5 �d8 26 �b8 lbb3 27 �a7 lbc5 28 lbd3 ! and wins. 2) 11 lbaS 12 lbc8 ! (not 12 b5? lbb6 ! 1 3 �d6 �f4) lbc7 13 �d6 ! lbb5+ 14 �c5 ! lbc7 15 �c6! lbe6 ...
..•
16 �d6! lbd4 17 �d5 ! lbb5 1 8 lbd6 ! lbc7 + 1 9 �c6 ! lbe6 20 lbb5 ! (once again White must play this paradoxical move blocking his own pawn) �g4 (20. . .�f4 1 9 �d5 ! lbf8 22 �d6 lbg6 23 lba3 wins for White) 1 9 lbc7 lLld4+ 22 �d5 lbe2 23 �e4 ! lLlc3+ 24 �d3 ! lbd 1 25 b5 lbb2+ 26 �d4 ! lba4 27 lba8 ! (on move 1 1 Black's knight visited a8; now it is White's tum) �f5 28 �c4 ! �e6 29 �b4 ! lbb2 30 b6 ! lbd3+ 29 �b5 ! lLie5 32 b7 ! lLid7 33 lbb6 lbb8 34 �c5 �e5 35 lba4 �e6 36 �b6 �d7 37 �a7 �c7 38 lbc3 lbc6+ 39 ..t>a8 ! �b6 40 lbd5+ �c5 41 lbe7 and wins.
1.3:
�+cLS v �
In many ways, there is little to add to the situation with the b-pawn, so we concentrate on those cases which display new features. When the pawn is on c7, the com ments at the start of section 1 .2 still apply. Diagram 40 shifted to the right remains a win, by the same method. Rather surprisingly, com posers have paid a lot of attention to this ending, and the following posi tions are particularly instructive. (68): This position is lost for White if it is shifted one square to the left. Here, however, White can draw by making use of the a2-square for his knight: 1 lLic 1 ! �f5 2 �g3 ! �e5 (the best try, as Black has no good move after 2 ...�e4 3 �f2!) 3 �f2 (but not 3 �g2? lbb5 4 � �d4! 5 lba2 �c4 6 �e3 �b3 7 lLic 1 + �b2
lll +� v lll 59
=I Tjavlovski, 1961 3rd Comm., Ceskoslovensky Sport
=I+ 69 Joita, 1965 (end of study) 3rd Pr., Revista Romana de Sah
8 llld 3+ �bl and Black wins) �e4 (if this position is shifted one file to the left, it becomes reciprocal zug zwang) 4 llla 2! (the saving resource) lllb 3 5 �e2 ! �d4 (another critical moment) 6 lllc l ! (a typical defensive idea; White should avoid 6 'it>f3? l0c5 7 �e2 llle4 ! 8 �f3 lllc3 9 lllc 1 llld5 10 �f2 �c3 1 1 �e2 �b2 win ning for Black) �c4 (or 6 . . . �c3 7 llla2+ ! �b2 8 lllb4 !) 7 llla2 ! (not 7 �e3? �c3 ! 8 llld3 lllc 5) llld4+ 8 �d2 ! �b3 9 lllc 3! (accurate to the end; 9 lllc 1 +? �b2 ! 1 0 llld3+ �bl ! 1 1 lllc l lllf3+ loses) �b2 (9 . . . lllf3+ 10 �c 1 ! llle5 1 1 �d2) 10 llld 1 + �bl 1 1 lllc3+ ! and White gives per petual check. (69): Black to play wins easily by l .. .�g4 2 �c4 �f3 and so on, so as sume that White moves first. He draws by 1 �d5 ! (not 1 �c4? �g4 2 �d3 �f3 3 llld2+ �g2 !) �h4 (after l . ..�g4 White replies 2 �e4! reach ing a position of reciprocal zug zwang; this implies that White must
avoid playing his king to e4 when Black can reply . . . �g4) 2 �e5 ! (2 �e4? loses after 2 . . . �g4 ! 3 llld2 llld6+ ! 4 �e3 �g3 ! 5 �d4 �g2 6 �e5 lllb5 7 �e4 lllc3+ 8 �d4 lllb l 9 lllc4 �gl 10 llle3 �2) �g5 3 lllh2 ! lllg 3 4 �d4 ! �f4 5 �d3 ! llle4 6 �e2 ! �g3 7 lllf3 ! �g2 8 llle l+ �gl 9 lllf3+ �g2 10 llle l+, with the same draw as in the previous dia gram. The situation in which Black's king is trapped on the a-file is unique to the c-pawn. The following posi tion is the key to such situations. (70): After 1 c7 ! lllb4+ ( 1 ...lllc5+ 2 �b6 ! llld7+ 3 �c6 ! lllb6 4 llld 6 �a7 5 lllc4 lllc8 6 �d7 wins easily) the win is surprisingly complex: 2 �a5 ! lllc6+ 3 �b6! llle7 4 �a6! lllc 8 (4 . . .l0c6 is a good trick, but fails to 5 c8l:.+ lllb8+ 6 �a5) 5 �4 (the next step is to transfer the knight to c5 in order to control b7 and thereby imprison Black's king) llle7 6 lllb3 lllc8 7 lllc5 llle7 (or 7 . . .llld6
68
60 tll +� v tLJ
70 +I= Selesniev, 1939 (version) '64 ' 8 �b6 tllc 8+ 9 'iti>c6 tllb6 10 tlla4 tllc 8 1 1 'iti>d7 and wins) 8 'iti>a5 (now White can move his king, but it is necessary to be careful, for example after 8 �b5? 'iti>a7 ! White falls into a reciprocal zugzwang) �a7 (there is no choice since 8 ... tllc8 9 'iti>b5 �a7 1 0 �c6 ! allows White's king to penetrate) 9 �b5 ! (now Black is in zugzwang) tllf5 (9 ... tllc 8 1 0 'iti>c6 !) 1 0 �c6 ! tlle7+ 1 1 'iti>d6 ! (the final trap is 1 1 �d 7? tlld5 ! ) tllc8+ 12 'iti>d7 tllb6+ 1 3 �d8 and wins. (71): After 1 c7 ! tlle7 2 tlld6 ! a position arises in which both kings are restricted. White would like to support the c-pawn with his king, but b4 and a4 are out of bounds, the lat ter because of ... tlld5 followed by . . . tllb6+. Therefore White must try to penetrate via c4. At first sight Black should have few problems preventing this, but if Black's king is on the c-file White can play 'iti>a4, be cause he promotes with check. Since d4, e4 and e3 are also out of bounds,
+I= Halberstadt, 1937 2nd HM, Tijdschrift v. d. KNSB 71
Black must aim for . . .'iii>e2-f3-f4-e5, but it turns out that this plan is too slow. The detailed analysis runs 2 ...'iii>e2 (2 . . . 'iti>c3 3 'iti>a3 ! 'iti>d3 4 'iti>b3 'iti>e2 5 �c4 and 2 ... �c2 3 �a3 ! �c3 4 �a4 ! are simpler) 3 'iti>b3 ! (but not 3 'iti>a3? �f3 ! , drawing after 4 'iii>a4 tlld5 ! or 4 'iti>b3 'iti>f4 !) and now: 1) 3...'iti>d3 4 �a3 (an attractive move) �e2 (or 4 ...'iii>c 3 5 �a4! 'iii>d3 6 �a5 'iti>e2 7 'iti>a6 'iii>f3 8 tllf5 tllc8 9 'iti>b7) 5 tllf5 (now that White has transferred his king from a2 to a3 with gain of tempo, he can win by forcing Black's knight to c8) tllc8 6 �b4 'iti>f3 7 �c5 'iii>g4 8 tlld6 tlle7 9 tllc4 �f5 10 'iii>d6 'iii>f6 1 1 'iii>d7 'iii>f7 12 tlles+ winning. 2) 3 �f3 4 'iti>c4 'iti>e3 (4 ...'iti>g4 5 �c5 'iti>g5 6 tllc4 'iti>f6 7 �d6) 5 'iii>c5 'iti>f4 6 'iti>d4 tllc6+ (Black cannot pre vent White's king penetrating to e6) 7 'iti>d5 and wins. There are 23 1 positions of recip rocal zugzwang with tll+�c7 v tll. ••.
lb+L!.
The following position depends on one of them.
•ttJ• • • • D • • • • • • • • • • • • • • • ••• •@ • • • • • .4'). •
v
lb 61
(Black's king is one square further away so this line is now winning) 'iti>d3 6 �f4 �c4 7 'it>e5 ! 'itc5 8 'it>e6! lbc8 9 'itd7 ! lbb6+ 1 o Wd8 'iti>d4 1 1 lbc6+ 'it>c5 12 lbb8 followed by lbd7. We end the £!.c7 section with a straightforward but amusing tacti cal point:
+/-
72
Original (72): After 1 . . .lbf2+ White has to choose his square with care: 1) 2 'itth4? lbe4 ! (now White is on the wrong end of a reciprocal zug zwang; in particular, any king move loses the pawn) 3 lbe7 (forced, but Black's king is close enough to the pawn) lbd6 ! 4 �g5 �c4 5 'iti>f4 'ittc5 6 'itte5 lbc8 ! 7 'iti>e6 'iti>b6 8 lbd5+ �c6! drawing. 2) 2 'iti>g3! (paradoxically, this gains a tempo) lbe4+ 3 �h4 ! (not 3 �4? lbc5 ! winning the pawn, nor 3 �g2? 'iti>e2 ! and White's king cannot move to h3; finally, attempting to creep along the first rank by 3 'iti>h2? �e2 ! 4 �g2 'iti>d3 ! 5 'it>fl lbd2+ 6 �el lbe4 doesn't work because d i is out of bounds) 'iti>e2 (Black's king must move away from the pawn since 3 ... 'iti>c3/d4 4 lba7 ! and 3 ...'ite3 4 lbb6! lose at once, while 3 ...'iti>c2 4 t'LJe7 ! lbd6 5 'iti>g5 is the same as in the main line) 4 lbe7 ! lbd6 5 'ittg5
=I 73 Lurye and Mitrofanov, 1986 (end of study) Comm., Molodoi Leninets
(73): White draws by 1 lbb2 ! (not 1 lbb6? �dl !) intending 2 lbc4, when Black will be unable to avoid the capture of his pawn. Neither 1 . ..lbe6 2 lbc4 ! nor 1 . ..lbd5 2 lbc4 ! helps, but what about l ...'itbl ? Then the tactical point is revealed: 2 lbc4 ! c 1 'ii' 3 lba3+ ! 'it>al 4 lbc2+ ! drawing by perpetual check. Now suppose that the pawn is on c6. If diagram 47 is shifted one square to the right (W'iti>b6,lbh5,l'!.c6 v B'it>c4,lbb5), it remains reciprocal zugzwang, and the analysis with
62 lb+l!.
v
lb
Black's king behind the pawn is the same as with a b-pawn. On the other hand, shifting diagram 49 results in a draw with either player to move, re flecting the generally increased drawing chances when the pawn is nearer the centre.
Diagram 50 remains a win when it is shifted to the right (W�c5, llle3 v B�e5, lDd6, l!.g3), but the recipro cal zugzwang of diagram 53 is de stroyed by the shift (because in the position W�d5, lbe5, M6 v B�d8, lllf2 , Black to play can draw by us ing the h-file with 1 .. .lllh 3!). Most of the new situations with a c-pawn arise when Black's king is trapped on the edge of the board. Averbakh discovered that even the following position is drawn:
=I=
74
Original (74): This is diagram 49 shifted to the right. It isn't immediately clear why the addition of the a-file helps the defence, but the following line demonstrates its importance: 1 �e5 �f7 ! 2 �d6 lDe6! 3 �d7 lllc5+ 4 �c7 llle6+ 5 �b6 �f8 (Black has several ways to draw, and this is probably not the safest choice, but it does show up the key difference) 6 llld4 lllf4 ! 7 �b7 llld 5 ! 8 lllf5 cli>f7! 9 �b8 �e6 10 llle 3 and Black was finished in the analogous situation from diagram 49, whereas here he can draw immediately by 1 0. . . lllb4 1 1 c7 llla6+. In other words, Black draws because White cannot dis lodge a knight on d5, even when his king is on b8.
=I=
75
Averbak.h, 1955 (75): Suppose Black moves first: l . . .�h7 ! 2 lDc4 (2 lllf3 tllxf3 ! 3 f7 llle5) and now: 1 ) 2 �h8? 3 lbd6 �g8 and now Averbakh continues with 4 �e8? �h7 ! 5 �d7, giving Black a second chance to draw by 5 ... �g6. How ever, White can win easily by 4 lbe4! lllf7 5 �e8 ! (this is a reciprocal zug zwang) llle5 6 lllg 5 ! and the second reciprocal zugzwang finishes B lack. 2) 2 �g6 3 llld6 �h5 ! (a tricky defensive move; Black must prevent ...
...
lll +l!. v lLJ 63
White promoting with check) 4 llle4 !Of3 ! (the point is that 5 f7 is met by 5 ...llle5 !) 5 'iii>e6 (after 5 lllc5 , Black can draw by 5 . . .'iii>g4 ! 6 llle6 llle5 ! , but not 5 . . .'iii>h4? 6 llld3 lllg5 7 lllf4 �g4 8 llle6 ! lllf3 9 'iii>d6 and the pawn advances) 'iii>g6 ! (returning is correct) 6 f7 llld 4+! 7 'iii>e7 lllf5+ ! 8 �e8 lllg 7+! 9 'iitf8 lllf5 ! drawing. White is no better off if he is to move. B lack answers 1 lllf3 with 1 . . .lllf7 ! , while after 1 'iii>e8 'iii>h7 White cannot play 2 lllf3 because of 2 ...llle4 ! . It i s worth noting Black that also draws if his knight starts on h6, for example 1 e8 (or 1 lllg4 lllf7 2 �e8 llld6+) 'iii>h7 ! 2 d7 h8 3 �d8 'iith7 4 'iii>e8 'iii>g 8 5 cJie7 lllf5+ 6 �e8 lllg7+ and so on. However, the following diagram makes it clear that B lack loses if bis knight is trapped on the corner square.
his knight emerges via b6. White to play wins by 1 'iitd6 ! (keeping the king out of c5) 'iii>a7 (or 1 . . .'iii>a5 2 llld5 'iii>a4 3 'iitd7 'iii>b 3 4 'iii>c8 c4 5 llle7 lllb6+ 6 'iii>b7 'iii>b5 7 c7 'iii>a5 8 lllf5 cJib5 9 llld6+ c5 1 0 lllc4 lllxc4 1 1 c8'if+) 2 llld5 ! (now that Black's knight is trapped, White can penetrate with his king) 'ittb 8 3 'iitd7 ! (White must take care not to move his knight and allow Black to escape from a8) 'iii>a7 4 'iii>c8 ! 'iii>a6 5 'ittb 8 ! 'iii>b5 6 'iii>b7 'iii>c5 7 llle7 lllb6 8 c7 b5 9 lllf5 'iii>a5 10 llld6 and wins.
77
+I=
76
Cheron, 1949 (76): Black to play can draw by 1...'iii>c5 ! 2 llla4+ 'iii>b 5, whereupon
I= Kremenietsky-Razuvaev Moscow Ch 1981
(77): After 1 . ..lllh8 ! (not 1 .. .lllf8 2 f7), Black can draw despite his poor knight position because the white knight doesn't occupy the powerful e5-square (as was the case in the previous diagram): 2 llld6 'iii>h5 ! 3 'iii>e6 'iii>g 6! (but not 3 ...'iith4? 4 lllf7 ! lllg6 5 llle5 lllh8 6 'iii>e7 'iii>h5 7 'iit f8 'iii>h6 8 'iii>g8 'iii>g5 9 'iii>g7 f5 10 llld7 lllg6 1 1 f7 and White wins) 4 cJie7 cJih5 ! (4 ... 'iii>h7? 5 lllf7 ! lllg6+
64 tll + � v tLJ
6 �e8 �g8 7 tllh6+ 'it>h7 8 tllg4 tllh8 9 tlle5 ! 'it>h6 10 'it>e7 �h5 1 1 'if.?f8 wins as in the previous note) 5 �e6 'it>g6 ! 6 tllf5 'it>g5 7 tlld4 'it>h6? (too passive; 7 . . .�f4 8 tllc6 'if.?g5 would have drawn) 8 tllf3 ! (White seizes his chance) tllg6 9 �d7 (this loses time; 9 'if.?f7 ! tllh8+ 10 �e7 tllg6+ 1 1 �e8 tllh8 12 'it>d7 �h5 1 3 �e7 �g6 1 4 'if.?e6 ! 'if.?h6 1 5 tlle5 �h7 16 'if.?e7 �g8 17 'it>e8 ! �h7 1 8 'if.?f8 would have won more directly) tllh8 (had Black played 9 . . .'if.?h5, White could only have won by going into reverse with 10 'it>e6 ! �h6 1 1 �f7) 1 0 tlle5 ? (a serious error; 10 �e7 tllg6+ 11 �e8 would have won as before) �g5 ! 1 1 �e6 'if.?h5? (the same error again; 1 1 . ..'it>f4 ! 12 tllf7 tllg6 ! would have drawn, because Black's active king position prevents �e7-f8 by White) 12 �e7 1-0, since 1 2 . . . �h4 1 3 �f8 �g5 14 'if.?g7 ! �f5 1 5 tlld7 tllg6 16 f7 wins easily. The following two over-the-board examples again show how easy it is to go wrong in practical play. (78): White to play wins by 1 tlld 5+ ! �b7 2 �e7 tllf3 (or 2 . . .�c6 3 tllf4 ! �b7 4 tlle6 tllf3 5 �d6) 3 tllf4 tlle5 4 �e6 tllc6 (4 . . .tllf3 5 �f5) 6 �d6 tlld8 7 'it>e7 tllc6+ 8 �e8 tlle5 9 tlld3 ! and the pawn ad vances. In the game Black was to move. Play continued 1 ...tllf7+? (missing a draw by 1 ... 'if.?b7 ! 2 'it>e7 and now not 2 ...�c7? 3 tlld5+ �c6 4 tllf4 ! �b6 5 tlle6 tllf3 6 'if.?d6, but 2 ... 'it>c6 ! 3 tlle4 tllf3 ! 4 �e6 tlld4+ ! 5 �e5 tllf3+! 6 'if.?f4 tlld4 !) 2 'if.?e6? (White could have won by 2 �e7 !, transposing
+I=
78
Yusupov-Vogt Altensteig 1993 into the game) tllg5+ ! 3 �d6 (3 'if.?f5 tlln ! 4 tlle4 'if.?c7 5 tllg5 tlld6+ ! 6 �e6 �c6! is no better) tllf7+? (giv ing White a second chance, which he gladly accepts) 4 'it>e7 ! tlle5 5 'if.?e6! tllc6 (or 5 ...tllf3 6 tlle4) 6 �d6 tDd8 7 tlld5+ (missing a simpler win by 7 tlle4 tllf7 + 8 �e7 tlle5 9 'it>e6 tllc6 10 'it>d6 tlld8 1 1 tllg 5) �b7 8 'if.?d7 tllf7 9 tllf4 tlle5 + 10 'it>e6 tllc6 1 1 �d6 tDtl8 1 2 'if.?d7 (once again, there was a simpler route to victory: 12 �e7 tllc6+ 1 3 �e8 tlle5 14 tlld3 !) tlln 1 3 tllg6 tllg5 14 �e7 �c6 15 tllf4 ! (White gets there in the end) �b7 16 tlle6 tllf3 17 'if.?d6 tllh4 1 8 f7 tllg6 19 tllf4 tllf8 20 'it>e7 tllh7 2 1 tlld5 1-0. (79): Black to play draws easily by 1 ...tDd4. In the game it was White to move, and after 1 c6 Black made a fatal error by l . . �b6?, losing to 2 tllc4+ ! 'it>a7 (2 . . . 'if.?c5 3 tlla3 ! tlla7 4 c7 ! also wins) 3 tDd6 ! tllc3 4 c7 00 5 tllb5+ ! 1 -0. However, we already know from the analysis of diagram .
liJ+�
79
=I= Psakhis-Sermek Groningen Open 1993
v
ljJ 65
liJd5+? �d6! 2 c7, when the stale mate defence 2 . . .liJc6 ! 3 �b6 �d7 ! 4 liJf6+ �c8 ! leads to a draw) �d6 2 c7 ! liJc6 3 �b6! 'iii>d7 (or 3 ...liJe7 4 liJf4 ! liJcS+ 5 �b7 ! �d7 6 liJd5 liJd6+ 7 �b8 ! �e6 8 liJb6 'iii>f6 9 liJc4) 4 liJf8+! 'iii>c8 5 liJg6 ! (this is a reciprocal zugzwang) liJa7 (5 . . .�d7 6 liJe5+! liJxe5 7 �b7 !) 6 liJe7+! �d7 7 �b7 ! �xe7 8 �xa7 ! 'iii>d7 9 'iii>b7 and promotes. The following study depends on a position of reciprocal zugzwang.
75 that Black should avoid having his king driven into the corner, and following this rule he could have drawn by I . . . 'iii>b4 ! 2 liJf7 'iii>c5 3 �g5 �d5. We continue with a couple of at uactive studies.
- - - -
!� .
�
� -
.
.
-�- - ,- , �- - . . � . - � ,I m m m - - - a m m m 80
+/= Truliko, 1981 3rd Comm., UV CSTV
(80): White wins by means of Y'CI")' precise play: 1 liJe6+! (but not 1
81
+/= Hanneman, 1961 1idskriftfor Schack
(81): Black to play draws, for ex ample by 1 . . .�e3. It is not at all ob vious that in order to win, White has to block the pawn with his own king. The solution runs 1 �g7 ! (but not 1 liJd7? liJg5 ! 2 �g7 liJe4 3 f7 liJg5 !) liJg5 (after 1 . . .�e3 2 liJd7 ! liJg5 3 'iii>g6 ! �f4 4 liJc5 Black is in recipro cal zugzwang and loses after 4 ...�g4 5 liJd3 �h4 6 liJe5 ) 2 liJh7 ! liJe6+ 3 �f7 ! liJd4 4 liJg5 (not 4 �g6? liJc6 ! 5 �f5 liJd8 ! 6 liJg5 �d3 7 �e5 �e3 !
66 �+� v �
and here it is White who is in recip rocal zugzwang, for example 8 �d6 �f4 ! is a draw) and now Black has to choose where to move his king: 1) 4...�d3 loses because it al lows White a later knight check: 5 �e7 �f5+ 6 �e6! �h6 7 �f7 �g4 8 �e5+ ! and 9 f7. 2) 4. �e3 appears better, but now that Black has committed his king, he cannot lose a tempo as in the note to White's 4th move, and thereby falls into reciprocal zug zwang: 5 �g6! (not 5 �e7? �f5+! 6 �e6 �h6 ! , nor 5 �g8? �f5 ! 6 �f7 �d4 7 �h7 �d5 8 �g6 �h4+ with a draw in both cases; it is worth noting that CMron recommended 6. . . �f4? 7 �h7 ! �g3 8 �h6! �d6 in the lat ter line, overlooking the simple tac tic 9 �f5+!) �c6 6 �f5 ! �d8 7 �e5 ! (reciprocal zugzwang with Black to move) �d3 8 �d6 �d4 9 �e6+ and wins. The following diagram is not only a fascinating reciprocal zugzwang, it is also worth analysing because some useful practical positions arise in the analysis. (82): First of all suppose that Black moves first: B l ) 1 ...�c7 2 �b6! and now: B l a) 2 ..�dS+ 3 �c5 ! �c7 4 �d6 �a8 (4 ... �b5+ 5 �d7 �e2 6 �a3 !) 5 �d7 �f2 6 lbc3 �e3 7 �a4 �d4 8 �c8 ! �c4 9 �b8 �d5 10 �b7! �d6 11 �b6 and wins. B lb) 2 lLJe8 3 �b7 ! (the ma noeuvre �b7-c8-d7 occurs repeat edly in the analysis) �f2 (3 . . . �e2 loses the same way) 4 �c8 ! �6+ 5 �d7 �b5 6 �a3 ! �c3 7 �e6, to be ..
.
.••
=I-
82
Original followed by the advance of White's pawn. B2) 1...�f2 ( l . ..�fl 2 lbc3 ! lbc7 3 �b6 �e8 4 �b7 �f2 5 �c8 is similar, but worse for Black) 2 �c3 ! �c7 3 �b6 ! �e8 (after 3 . . .�e6 4 �e4+! �e3, not 5 �c5? �f4 ! draw ing, but 5 �g5 ! �f4 6 �c5 ! �g6 7 �d6 and wins) 4 �d5 ! �e2 (4 ... �f3 5 �c5 is worse, since e4 and g4 are out of bounds because of �f6+, so Black loses after 5 ... �g3 6 �f6 lbc7 7 �b6 �e6 8 �e8 �g4 9 �g7) 5 �c7 ! (but not 5 �c5? �d3 ! because now White is in reciprocal zug zwang, and Black draws after 6 �6 �c7 ! 7 �b6 �e6! 8 �h5 �c4 9 �g7 �f4 ! ; this reciprocal zugzwang is useful to know) �g7 (or 5 ...�f6 6 �c5 ! �e4+ 7 �d5 �c3+ 8 �c4 �d2 9 �5 tl)e4 10 c7) 6 �c5 ! �d2 (after 6 . . .�d3 7 �b5 ! �e8 8 �d5! the position is reciprocal zugzwang, so the black king must avoid d3) 7 �d5 ! �e6+ (7 . . .�e8 8 �f6 ! �c7 9 �b6) 8 �c4 ! (this is again reciprocal zugzwang) 8 . . . �c2 (after 8 ...�cl
lll +� v lLJ 67
Black's king is too far away, and White wins by playing 9 �d3 �b2 I 0 �e4 �b3 1 1 �e5 !) 9 lllb4+ �b2 (9 ... �d2 10 llla6 is the same) 10 �6 �a3 1 1 �d5 ! and White wins. With White to play, it is hard to imagine that playing �b5 or lllc 3 weakens his position, but that is in deed the case: W l ) 1 lllc3 (this fails because Black can gain an important tempo with ... �d2; 1 llla3 �d2 is harmless) �7 ! and now: W l a) 2 lllb5 llle 8 3 �b6 �d2! 4 �c7 lllg7 ! 5 �c5 �c2 ! (this is the key move; we saw in line B2 that if Black's king is on d2 or d3, he inevi tably falls into zugzwang; here, how ever, Black's king has time to move to the other side of the d-file barrier) 6 llld5 llle6+ ! 7 �c4 �d2 (or 7 . �b2, again with reciprocal zug zwang) and now White is on the wrong side of the reciprocal zug zwang. W l b) 2 �b6 llle8 ! 3 �b7 'itid2! (gaining a tempo by attacking the knight) 4 llld5 (4 lllb5 �d3 ! is the same) �d3 ! and draws. W2) 1 �bS �e2 ! (not 1 . .�f2? 2 �c5 lllc7 3 lllc 3 llla8 4 �d6 �e3 5 �d7 �d4 6 �c8 ! lllb6+ 7 �b7 ! and White wins; the importance of ...�e2 is that Black needs to be able to cover c3 in one move, and thanks to White's �b5, which rules out the trick with lllc 3+, Black's king can bead in the right direction) 2 �c5 �7 ! 3 �d6 (3 lllc 3+ �d3 ! 4 lllb5 �6+ 5 �d6 �c4 ! draws) llla8 ! (not 3 . lllb5+? 4 �d7 ! �d3 5 llla3 ! lllc3 6 �d6 llle4+ 7 �e5 and White wins) ..
.
..
4 �d7 �d3 ! (this is the key move which was made possible by Black's l ...�e2; White's knight is kept away from c3) 5 �c8 lllb6+! 6 �b7 llld5 ! and, now that lllc 3 is impossible, Black draws comfortably. There are 432 positions of recip rocal zugzwang with lll+�c6 v lll . Positions with the c-pawn further back have hardly been analysed at all, so I will give a round-up of inter esting positions and present a few new discoveries. The following is a simple but at tractive composition, in which Hal berstadt again demonstrates his skill with minor-piece endings.
••• • •• • • • • .lb. • • B � • B . � . . � � � � B B • B • • • • B B B B 83
+I= Halberstadt, 1952 Les Lettres Franfaises
(83): Both knights are effectively immobilised, so this position can be resolved by the method of corre sponding squares. White's ultimate aim is to reach the reciprocal zug zwang �e6 v �e8 with Black to move. It follows that f6 and f8 also
68 lLJ+�
v
lLJ
correspond (this is just the opposi tion), and that e5 and f7 correspond (since ... c;i;>e8 is met by c;i;>e6 and ...c;i;>f8 by c;i;>f6) d5, which is adjacent to e5 and e6, must correspond to a square adjacent to f7 and e8; this has to be f8. Now we have the solution: 1 c;i;>e4 ! (not 1 c;i;>e5? c;i;>f7 !, nor 1 c;i;>d5? c;i;>f8 ! 2 c;i;>e6 c;i;>e8 !) c;i;>f8 (after I .. .c;i;>g7, the reply 2 c;i;>f5 ! threatens 3 c;i;>e6 and so forces 2... rJ;f7, but then 3 c;i;>e5 ! gains the correspondence) 2 c;i;>d5 ! (seiz ing the correspondence; not 2 c;i;>f5 ? lLJe7+ !) c;t>n 3 c;i;>e5 ! c;i;>f8 4 c;i;>f6 ! c;i;>e8 5 c;i;>e6 ! and wins. It is interesting to note that Black to play draws by l . .. c;i;>g7 ! 2 c;i;>e5 (or else ...c;i;>f6) c;Pf7! . As usual, practical examples only serve to emphasise the difficulty of conducting unfamiliar endings over the board.
=I
84 L. Cooper-Morris British Ch 1989
(84): White's king is actively po sitioned to the rear of the pawn, so
he should have no trouble drawing this position. In the early stages there are several drawing alternatives at each move: 1 lLJh6 lLJg3 2 lLJf7 c;i;>e3 3 lLJg5 lLJe2 4 'it>f5 lLJc3 5 c;i;>e5 (it would have been safer to take the chance to transfer the knight to a more active position by 5 lLJf7 'it>f3 6 lLJe5+ c;i;>g3 7 rl;g5; now the manoeu vre 7 . . . ll\bl 8 'it>f5 lLJd2 9 c;i;>e6 lLJf3 is not dangerous, since 10 lLJd3 cap tures the pawn) lLJbl 6 c;i;>f5 lLJd2 7 lLJf7 (White cannot prevent Black advancing the pawn) lLJf3 and now: 1) 8 c;i;>g4? (a fatal error) lDh2+ ! 9 c;i;>h3 f3 ! 1 0 lLJd6 f2 ! 1 1 ll\f5+ �f3 ! 1 2 lLJg3 lLJg4! (very accurate play by Black) 1 3 lLJfl lLJe3 14 lLJh2+ c;i;>e2 1 5 ri;h4 lLJf5+ 1 6 c;i;>h3 lLJd4 1 7 ri;g2 0- 1 . 2) 8 ll\h6 (one of many drawing moves) lLJh2 9 c;i;>e5 f3 10 ll\f5+! c;i;>d2 ( 1 0 . . . �d3 1 1 c;i;>f4 f2 1 2 lLJg3 ! 'iird4 1 3 lLJfl is a typical line) 1 1 �e4 f2 1 2 lLJg3 ! (Black is to move in a reciprocal zugzwang) 'it>el 1 3 'it>f4 ! (a second reciprocal zugzwang) ri;d 1 14 c;i;>e3 ! ri;el 15 'it>f4 ! and Black cannot make progress. (85): Cheron's analysis ofminor piece endings was not up to the same standard he displayed with rook end games. For example, despite analys ing this relatively simple position for one and a half pages, he errone ously claimed that 1 'iii>d8 lLJc3 2 c6 ! (the exclamation marks and ques tion marks are mine) lLJb5 doesn't win, on account of the lines 3 ll\e6? rl;a7 and 3 'it>d7? ri;a7? (or 3 ... lLJc6). However, 3 lLJe2! (threat 4 ll\d4) does the trick by setting up a position
lLJ+�
85
+I= Cheron, 1955 Journal de Geneve
ofreciprocal zugzwang. White wins after 3 . . . 'itta7 (3 . . .lDc7 4 lDd4 ! { sec ond reciprocal zugzwang } lDd5 5 lllb5 ! is a third, and final, reciprocal zugzwang, while 3 . . . �a8 4 'ittd7 ! is the same as the main line) 4 �d7 ! �b6 (or 4 . . . lDa3 5 c7 lDc4 6 lDd4 Ille s+ 7 �d8 lllf7+ 8 �c8 llle5 9 lllb 5+ 'ittb6 l O 'ittd8 lllf7+ 1 1 ri;e7 and wins) 5 lllc 3 ! llla7 6 c7 ! with an easy win. It also follows that his line 3 �d7? ri;a7? is wrong, since White can again win by 4 lDe2 ! ; thus 3 ..t>d7 should be met by 3 ...lllc7 !. Cberon's intended solution also wins: 1 ri;d7 lllc 3 2 c6 ! lllb5 3 lDe6 ! �a7 4 llld4 ! lllc 3 5 c7 llld5 6 lllc6+ �b6 7 c8'if. We end with a remarkable recip rocal zugzwang, chosen from the 203 with lll+�c5 v lll. (86): It certainly appears unlikely that any black king move makes his position worse, but this is indeed the case. The reason is that Black will meet the threat of c6-c7 by playing
v
lLJ 69
=I-
86
Original his knight to d5 . Then White would like to dislodge this knight by lDe2f4, but if Black's king is on g5 this is impossible. On the other hand if Black plays . . .�f5 or ... 'ittg 4, his knight can be disturbed by lLJe2-c3, because after ...tDxc3, c7 lDb5 White will promote with check. The de tailed analysis runs: B 1 ) 1. lllc3 ( 1 .. . �f6 2 c6 lllc3 3 lDe2 ! is the same) 2 c6 ! ri;f6 3 llle2 ! lDd5 4 lllf4 transposes to line B4d below. B2) 1. �h6 2 c6 lllc 3 3 lLJe2 lLJd5 4 lllf4 leads to B4c. B3) 1 ..�S 2 c6 wins at once as the pawn promotes with check. B4) 1 llle3 2 c6 ! llld5 3 llle2 ! (this is also reciprocal zugzwang) and now: B4a) 3 . �g6 4 lllf4+ ! deflects the knight. B4b) 3 �S (3 ...�g4 is also met by 4 lllc 3) 4 lllc 3 ! wins, as the de fence 4 . . . lllxc3 5 c7 lllb5 doesn't work when the pawn promotes with check. ..
..
.
...
. .
.••
70 tLJ+l'!. v tLJ
B4c) 3 �h6 4 lDf4 tDc3 5 ..t>a6 and the pawn advances. B4d) 3 � 4 lDf4 ! tbc3 5 �b6 tba4+ 6 �a5 tbc5 7 c7 tDb7+ 8 �b6 tDd6 9 �c6 tDc8 10 �b7 tbd6+ 1 1 �b8 with an easy win. B4e) 3 tDr6 4 �c7 ! �f5 (alternatively, 4 ... tDd5+ 5 �d6) 5 �d6! tDe8+ 6 �e7 tbc7 7 �d7 ! tbd5 8 �d6! tDf6 (8 ...�e4 9 tbg3+! �d4 1 0 lDf5+! �e4 1 1 tDe7 tDc3 1 2 �c5 !) 9 tbg3+ �g6 (9 ...�g4 10 �e6 tDe8 1 1 tbe4 tDc7+ 1 2 �d7 is the same) 1 0 ..t>e6 tbe8 1 1 tDe4 tDc7+ 12 �d7 tDb5 ( 1 2 . . . tbd5 1 3 �d6) 1 3 tDd6 tDc3 14 �e6 and wins. With White to play, a knight move is too committal, while b7 is actually the best square for White's king when it comes to avoiding forks: W l ) 1 tbe2 ( 1 lDf3+ �f6 ! and 1 tDh3+ �f6 ! 2 c6 tbc3 ! 3 lDf4 �e7 4 �b8 tDb5 ! are also drawn) tbe3 ! 2 tDc3 (2 c6 tDd5 ! is line W2) tbc4 ! 3 ..t>b8 tDe5 draws. W2) 1 c6 tbc3 ! 2 tDe2 (the alter natives are no better: 2 �b8 tDd5 ! 3 tDe2 tDb4, 2 �a6 tD
...
...
Composers have had no more luck with the pawn on c4:
=I= 87 Vlasenko and Tolstoi, 1969 (end of study) 1 st HM, Vserossiski
(87): The composers intended that their study should conclude 1 tDc3+ �d4? 2 ..t>f3 ! �xc3 3 f5 ! and the pawn cannot be stopped. Unfor tunately, Black can draw by l . ..�e3 ! 2 f5 (or 2 tDd 1 + ..t>e4 ! reaching a po sition of reciprocal zugzwang; the position is drawn after 3 f5 �e5 4 �g5 tbg3 !) tDf2+! 3 �h5 (3 ..tg5 tDh3+ 4 �h6 lDf2) tbd3 ! (but not 3 . . . tbh3? 4 f6 ! lDf4+ 5 ..t>h6 ! and wins) 4 f6 tDe5 ! 5 ..t>g5 �d4 and Black's king runs back. There are 57 reciprocal zug zwangs with tD+l'!.c4 v tD. Here is one: (88): This reciprocal zugzwang is relatively easy to understand. Black to play must either move his knight, allowing an immediate tactical dis aster, or his king must move away from f6, when White can send his own king to the queenside, as Black is deprived of a later defence based on ...�f6-e6:
lll +.0i
=I-
88
Original B l ) 1 ...�8 2 �d6 ! (Black can longer play . . .�f6) �e8 (Black can play his king in front of the pawn, but although this is normally a comfortable draw, here Black loses because of the terrible position of his knight) 3 c5 ! �d8 4 lllc 6+! �c8 S �7+ ! �b7 6 llld5 ! �c8 7 �c6! �b8 8 �d7 ! �b7 9 c6+ �b8 1 0 �d8 ! �a7 1 1 �c8 and White wins � in diagram 76. B2) 1...lllc7+ 2 �d7 llla6 3 llld3 ! followed by �c8-b7 traps the enemy knight. B3) 1 ...�h6 (here the king is just IOO far away, so White has plenty of nme to improve the position of his knight) 2 llld 3 �g7 3 lllb4 �f8 4 �5 �e8 5 �d6 �d8 6 �c6 win ning. B4) 1...�g8 (this puts up the most resistance) 2 �d6! lllb6 (this was impossible in line B 1 due to the reply llld 7+) 3 c5 ! lllc 8+ (3 . . . llla4 4 c6! lllc 3 5 �e6 lllb5 6 �d7 �g7 7 �4 �g6 8 llld6) 4 �c7 ! llle7 (or 4 . . llla7 5 �b8 lllb 5 6 c6 ! �f8 7 80
.
v
lll 71
lllc4 ! �e7 8 llla 3 ! llld4 9 c7 ! lllc6+ 10 �c8 ! �e6 1 1 lllc2 'it>e7 12 lllb4 llla7+ 1 3 �b8 ! �d7 14 �b7 ! and White wins) 5 �d8! llld5 6 c6 ! �g7 (6 . . .�f8 7 �d7 �g7 8 �d6 lllc 3 9 'it>e6 wins as in the note to Black's third move) 7 lllg4 ! �ii 8 llle3 ! lllf4 9 �d7 ! llle6 1 0 llld5 ! lllc 5+ 1 1 �d6 ! llle4+ 12 �e5 ! llld2 1 3 c7 ! lllc4+ 14 �d4! llld6 15 �c5 ! 'it>e6 1 6 �c6 lllc 8 17 lllb6 and finally wins. In this variation White had to find 15 consecutive 'only' moves. W l ) 1 cS (gives Black's knight access to b5) lllc 7+! 2 �d6 (2 �d7 llla6) lllb 5+! 3 �d7 llld4 draws. W2) 1 llld3 �f8 ! 2 lllb4 lllb6 ! 3 c5 llla4 ! 4 c6 �e8 ! 5 c7 lllc 5+ ! 6 'it>d6 lllb 7+! 7 �c6 llla5+ 8 �b6 'it>d7 ! and draws. W3) 1 llld7 lllc 7+! 2 �d6 llla 6! 3 lllc5 lllb4 ! 4 llle6+ �ii 5 c5 �e8 is safe for Black. W4) 1 �d6 (this is the key line) lllb 6! 2 c5 lllc 8+! 3 �c7 (or 3 �d7 llla7 ! 4 lllc6 lllb5 !) llle7 4 �d8 lll
72 lb+� v lLJ
+I=
89
(90): The solution runs 1 lDf4 ! (first reciprocal zugzwang) lDh7 2 lLJe6 ! �h8 3 f3 ! (second reciprocal zugzwang; not 3 f4?, when the parity is wrong in the subsequent recipro cal zugzwang positions and Black draws by 3 ... 'itg8 ! 4 f5 �h8 ! 5 �n lLJg5+ ! 6 lDxg5 with stalemate) �g8 4 f4 ! (third reciprocal zugzwang) �h8 5 �n (5 lLJf8 is equally effec tive) and wins since with the pawn on f4, White can meet 5 ... lLJg5+ by 6 fxg5 !.
Original by pushing the pawn) 5 �c4 ! (this is reciprocal zugzwang; not 5 c4? �e5 ! and White has to release Black's knight) �e5 (or 5 . . . �f4 6 �d5) 6 �d3 ! �d5 7 c4+ ! �d6 8 �d2 �c5 9 �c3 ! �d6 10 lDf3 �c5 l 1 lLJd2, fol lowed by �b2, and wins the knight. With the pawn on c2, only Hal berstadt's analysis is worth repro ducing: +I= Halberstadt, 1961 5th HM, teskoslovensky Sach 91
+I=
90
Halberstadt, 1952
(91): White's only winning hope is to exploit the poor position of Black's knight. 1 �f4 ! (for the mo ment there is no way out via d2 be cause of the reply lLJe4+) �g6 (after 1 . . .lDh2 2 lDc4 �e6 3 �g3 lLJfl + 4 'itg2 or 1 . . . �e6 2 lLJc4 ! �f6 3 �f3 White wins more easily) 2 lLJe4 ! (the straightforward plan of lLJc4 fol lowed by �f3-g2 no longer works, for example 2 lLJc4? �h5 ! 3 �f3 lLJh2+! 4 �g3 lLJg4 ! drawing) �n
tD+8
(2...�h5 loses to 3 tDg3+! tDxg3 4 fxg3 !, so the king has to retreat) 3 �f3 ! (threatening �g2, so Black's reply is forced) �e6 4 tDc5+! (not 4 �g2? �e5 !) �f5 (4...�d5 loses to 5 �b3 ! followed by �g2) 5 tDb3 ! (this appears to let the knight escape, but White rounds it up in any case) �h2+ 6 �g3 ! tDg4 7 tD
� � � W;Pd � � � B B B BtzJ • • • • • • • • -�· . . . -�· . • • • • • • • •• +I=
92
Original (92): 1 tDg5 ! (White's first prior ity is to improve the position of his knight; 1 c5? tDg6! 2 �e4 tDe7 ! 3 �5 tDc6+ 4 �d6 tDa7 ! 5 �c7 �g2 leads to a draw after 6 tDg5 tDb5+! 7 � tD
v
tD 73
1 ) l �g2 2 �e4! tDg6 3 tDe6 tDe7 4 tDd4 ! (not 4 c5 ? tDc6 5 tDd4 tDa7) tDc8 5 tDb5 ! (advancing the pawn is still premature: 5 c5? tDa7 ! 6 �d5 �f2 only draws) tDe7 6 c5 tDc6 7 �d5 ! tDb4+ 8 �d6 ! �f3 9 tDc3 ! �f2 10 tDd5 ! tDc2 1 1 �e5 ! tDa3 1 2 c6 tDb5 1 3 tDc3 tDc7 14 �d6 tDa8 15 tDa4 and White is win ning. 2) l...tDg6 2 tDf3 ! (2 c5 ? �g2) tDf4+ (2. . .�g2 3 �e4 ! tDf8 4 c5 ! tDe6 5 c6! �f2 6 tDd4 tDc7 7 �f4 transposes into the main line, while 2 . . .tDf8 3 c5 ! tDe6 4 c6! tDc7 5 �e4 �g2 6 tDd4 is fundamentally the same) 3 �e4 ! tDe6 4 �d5 ! tDf4+ (4... tDc7+ 5 �c6! tDa6 6 tDel tDb4+ 7 �b5 tDa2 8 c5 tDc3+ 9 �c4 tDa4 10 c6 ! wins) 5 �d6 ! tDd3 6 tDe5 ! tDc 1 7 �d5 ! tDb3 8 tDf3 ! with a fur ther branch: 2a) 8...�g2 9 tDd4 ! tDcl 10 �e4 ! tDa2 1 1 tDc2 ! tDc3+ 1 2 �d4 ! tDa4 1 3 tDe3+ �f3 14 tDd5 ! �g4 1 5 tDc3 ! tDb6 1 6 c5 ! tDc8 1 7 �e5 ! tDe7 18 �d6! tDf5+ 19 �e6 tDd4+ 20 �d5 ! tDc2 21 tDa2! tDe3+ 22 �e4 tDc4 23 c6 and wins. 2b) 8...ttJcl 9 tDel ! tDb3 10 tDc2 ! �gl ( 10 . . . �g2 1 1 tDd4! transposes to line 2a, while 10 . . .�h2 1 1 tDd4 ! tDc 1 12 �e4 ! is basically the same as the main line) 1 1 tDd4 ! tDcl 12 �e4! tDa2 1 3 tDc2 ! tDc3+ 14 �d4 ! tDe2+ ( 14 . . . tDa4 1 5 tDa3 ! �f2 16 tDb5 ! �f3 17 tDc3 ! is exactly analogous to line 2a) 15 �e3 ! tDc3 (or 1 5 . . . �fl 16 tDb4 tDc3 17 tDd5 tDbl 18 c5) 16 c5 ! (the pawn finally edges ahead) tDb5 ( 1 6 ... �fl 17 c6 tDb5 transposes) 17 c6 ! �fl 18 tDd4 tDc7 •••
74 tLJ+� v tLJ
1 9 We4 Wf2 20 Wf4 tDa8 (20... We 1 2 1 We5 ! Wd2 22 Wd6 ! tDas 23 tDb5 Wd3 24 tDc7 tDb6 25 Wc5 tDa4+ 26 Wb5 tDc3+ 27 Wb4 !) 2 1 We5 We3 (2 1 . . .'it>g3 22 tDe6 is the same) 22 tDe6 Wd3 23 Wd5 ! We3 24 Wc5 We4 25 Wd6 ! Wd3 26 tDc7 tDb6 27 Wc5 tDa4+ 28 Wb5 tDc3+ 29 Wb4 ! tDa2+ 30 Wa5 tDc3 3 1 tDe6 tDe4 32 c7 tDd6 33 Wb6 tDcs+ 34 Wb7 tLJe7 35 tDf4+ We4 36 tDg6 tDf5 37 Wc6 and White either promotes or captures the knight next move.
1.4:
lt:J+di0i v lZJ
Previous analysts have discovered little that is truly new with the d7pawn. The high point of pre-data base analysis is undoubtedly the following study by Halberstadt, surely one of the most attractive po sitions ever composed with tD+� v tLJ.
93
+I= Halberstadt, 1938 J/2, Pr., Gros
(93): Black to play draws imme diately with l ...tLJe5, so suppose that White moves first. If White's king were hidden away at, for example h 1 , then he would win easily even if Black were to move. The squares a8, b7, a6, b5 and d5 are out of bounds because of a white knight check, . . . Wb6 (or . . . Wb4/c3) may be met by tDc7 and tDd5+ and ...Wc5 is met by tDd6 and tDe4+. Thus a8, b7, a6, b6, b5, c5, d5, b4, c4 and c3 are all forbidden. This means that if the black king is on a7, then he is re stricted to playing ... Wb8-a7, be cause these two squares form an island which is cut off from the rest of the board. Black is even worse placed with his king on c6, because he is in immediate zugzwang. If the king is on a5, then it can eventually escape via a4, b3 and c2, but even from h l White's king will arrive to support the pawn long before Black is out of the cage. In the diagram position White cannot win straight away because his own king is exposed to black knight checks, but by carefully withdraw ing his king out of the knight's range he can force Black into zugzwang. The alternatives are: 1) 1 tDc7? tDf6 ! 2 tDd5+ tDxd5+ ! taking the knight with check. 2) 1 Wc3? Wc5 ! 2 tDd6 tDf6 ! 3 tDe4+ tDxe4+ ! is similar. 3) 1 Wa4? tDc5+ 2 Wa3 tDe6 3 Wb4 Wc6 4 tDf6 tDg7 is drawn. 4) 1 Wb3 ? tDc5+ ! 2 Wc4 tDe6 ! 3 tDf6 tDg7 ! 4 Wd5 Wc7 ! 5 We5 Wc6 6 tLJd5 Wd7 ! 7 Wf6 tDe6 ! 8 tDb6+ We8 ! 9 tDc4 tDc7 1 0 tDe3 tDe6! 1 1
lb+8 v lb 75
lbf5 d7 ! 1 2 �f7 lbd8+ 1 3 f8 lbe6+ ! and Black draws by standard means. 5 ) 1 �a3! and now: 5a) 1...�a7 ( l . . .�a5 2 lbc7 lbf6 3 �b3 puts Black in zugzwang) 2 �b2 �b8 (Black is completely tied up) 3 �c3 �a7 4 lbd6 lbf6 5 �d4 ! �b6 6 e5 and wins. 5b) 1 .�c6 2 �a2 ! (2 �b2? lets Black free himself by 2 ...lbe5 3 lbd6 lbc4+!) and this second paradoxi cal move puts Black in a fatal zugzwang. White wins after 2 ... �b6 3 lbc7 lbf6 4 lbd5+ ! or 2 ... �c5 3 lbd6 ! lbf6 4 lbe4+ ! . The two retreating moves by White's king create a powerful im pression. Mistakes can occur, even in ap parently simple situations. ..
B B B B B B B B• �- B B B B B Btt:JB B B B B B B B B B m B B B B�B B -/+
94
Szulc, 1945 Sach (94) : Black to move wins by i ...lbc3 2 lbe3 �g6 3 �a5 g5 and White cannot play �b4 because of ...lbd5+. The composer believed that
White to play could draw by 1 �b7 (not 1 �b6? lbb2 ! 2 lbe3 lbc4+ nor 1 �a5? lbb2 2 lbe3 lbc4+) lbb2? (my question mark) 2 lbe3 ! �g6 3 �c6! �g5 4 �c5 �f4 5 �d4 ! �f3 ! 6 lbd 1 ! lbxd 1 7 �d3 ! . However, the small finesse 1. . lbf2! wins for Black. The point is that after 2 lbe3 �g6 3 �c6 g5 ! 4 cli>d5 (or 4 �b5 �f4 ! 5 lbd5+ �f3 6 lbc3 lbe4 7 lbd 1 e2 8 lbb2 00 9 �c4 lbd3 10 lba4 �e l 1 1 lbc3 lbf4) �f4 ! 5 �d4 �f3 ! 6 lbd l Black's knight is defended, and so he can win by 6 ... e2! 7 lbc3+ �el . The following simple study fea tures a neat tactical point. .
95
=I+ Stolk, 1943 Iijdschrift v. d. KNSB
(95): Black to play wins easily by 1 . ..�b3 2 �bl lbd5 and 3 ... lbc3, but White to play draws with the trick 1 lbc3 ! �b3 2 lbbl ! and there is no winning promotion for Black. The following position is based on one of the 185 reciprocal zug zwangs with lb+8d7 v lb.
76 lll +� v lLl
For example, diagram 40 shifted two files to the right (W�c7, llic6, £!.d7; Bc;t>f6, llifl) remains a win, whoever moves first. Diagram 70 shifted to the right (Wc;t>b6, lllc5 , l!.d6 v B�b8, llie3) is an even easier win, because there are no stalemate tricks when Black's king is not in the corner. The following position reveals a significant difference with the d pawn. +I=
96
Original (96): White to play can try: 1) 1 �bl? (1 �b2? llic4+ ! is a blunder) �c3 ! 2 �a2 (or 2 �al �c2 3 �a2 �c3 ! 4 �bl c;t>b3 5 �cl �a4 6 �c2 �b5 ! 7 �d3 �b6 and Black's king approaches the pawn) c;t>c2 ! 3 �al (3 c;t>a3 loses the pawn) c;t>b3 followed by ...�a4, with a draw as before. 2) 1 �dl! (this position is recip rocal zugzwang) �e3 (after t . ..c;t>c3, the winning line is an exact reflec tion) 2 �c2 �e2 3 �b3 �f3 4 �b4 �g4 5 lDc6 llifl 6 �c5 �f5 7 �b6 �e6 8 �c7 with a straightforward win. Black to play draws by l . ..'iii>c 3 ! (not l . . .�e3? 2 'iii>c2 �e2 3 �b3 winning as above) 2 �d 1 �d3 (2 ... �b3 3 c;t>e2 �a4 ! 4 �d3 �b5 ! 5 �d4 c;t>b6 ! also draws) 3 �e 1 �e3 ! 4 �fl 'iii>f3 ! 5 �gl �g3 ! , keeping White's king imprisoned. Moving on the case of the pawn on d6, many of the results from ear lier sections apply in this situation.
=I=
97
Original (97): If this position is shifted either one or two files to the left, then it is reciprocal zugzwang (see dia gram 4 7), but here Black to play draws by making use of the extra space to the left of the pawn. After 1 . . .llia6! the draw is quite clear; White cannot advance the pawn, and otherwise Black can either return to c5 or start checking with ...llib8+. The following practical example is very interesting: (98): The game concluded 1 llic8+? �d8? 2 e7+ ! �d7 3 �e5 ! 1-0. Averbakh correctly pointed out
lb+� v lb 77
+I-
98
�g6 (or 2 ... lba4 3 e7 ! �f7 4 �c6 ! and wins) 3 e7 �f7 4
Kan-Goldenov USSR 1 946 that Black could have drawn by l .. .�f6! 2 e7 �f7! 3 'ifiid6 (or 3 'ifiic 5 �3+ 4 �d6 'it>e8 !) lbc4+ 4 �d7 �5+ ! , but failed to notice that White could have won the move be fore. The decisive line runs 1 lbf5+ ! �f6 ( 1 . . .�d8 2 e7+ f6 ! is reciprocal zug zwang, and therefore winning for White, while l . . .�e8 2 e7 �f7 3 �c5 ! transposes to the main line) 2 c7 �f7 3 �c5 ! (a remarkable move; not 3 �c6? lbc4 and Black draws) �4+ (3 ... lbd3+ 4 �d6 ! wins be cause, unlike the play after 1 lbc8+?, there is a threat of lbg7; 3 . . .�e8 4 �d6 lbc4+ 5 �e6 ! is also winning for White) 4 �c6 ! (reciprocal zug zwang) �e8 (there is no defence) 5 �6! 'it>f7 6 lbg7 and wins. Curiously, White wins even if Black moves first, for example I �f6 ( 1 ... �d8 2 �e5 'it>e7 3 lbc8+ �s because ...�f6 is no longer pos sible, while l . ..lba4 2 lDfS+! is the same as the following note) 2 lbf5 ! ...
-/+ Dobrescu, 1973 1st Comm. Rumanian Ch
99
(99): We have taken the position after one move of this 'White to play and draw' study. After 1 lbg2 ( 1 �b4 e2 ! transposes) e2 2 lbe l (2 �b4 e4 3 �b3, the com poser continued his analysis with 3 ...lbf3? 4 �c2! �e3 5 lbg2+! (not 5 lDd3? lbes 6 lbe1 �f2 1 �d2 lbc4+> �f2 6 lbf4 ! and White draws. Un fortunately, the study is unsound because B lack can win by 3 . . . �e3
78 tb+.0,
v
tD
4 �c2 tDh3 5 �c3 �f2 6 tbc2 tDf4 7 �d4 tDg2 8 tbb4 �e l 9 tbd3+ �d l 10 �e4 �d2 1 1 �d4 tDh4 12 �e4 tbf3 . The following position is baf flingly complex, as it depends on a network of four linked reciprocal zugzwangs.
method of making progress is by 2 tbc5 (2 �f6 tDb8 ! 3 �e7 �g4 4 �d8 �f5 5 �c7 �e6 is even less promis ing) tbb8 ! (this is a second recipro cal zugzwang, Z2) 3 �f5 �g3 ! , but in fact this is a third reciprocal zug zwang (Z3). Why is this position reciprocal zugzwang? Because after 4 tDa6 tbd7 ! 5 �e6 tbb6 ! we reach a key position:
+I=
1 00
Original (100): Black to play draws com fortably after 1 . . .�g3 2 �g5 �f3, since 3 d7 may be met by 3 ... tbe5, so assume that White moves first. 1 �g6! The first point to note is that the immediate 1 d7 doesn't work, be cause after 1 . . .�g3 White's king is imprisoned (2 �h6 and 2 �g5 both fail to 2 . . . tDe5 !). Playing 1 tDc5? fails to 1 . . .tbe5, and again White's king cannot move to g5 or h6. There fore White must move his king. The next important point is that the posi tion with �g5 v �h3 is reciprocal zugzwang (we will call this Z l ) . To see why this is so, look at the posi tion after 1 �g5? �h3. White's main
=I=
101
Original This type of position arises sev eral times with the black king on dif ferent squares. White has two ways of making progress; he can either try tDc7-a8 (Black generally meets tDc7 by ...tDa4), or tbb4-d5 (tbb4 aims to prevent . . . tDa4 because of the reply tbd3). After 6 tbb4, Black replies 6 . . . �f3 ! and we have a fourth recip rocal zugzwang (Z4) (see diagram 102 on the following page). Black to play would lose because he can only move his king, but the available moves all fail: l ...�e4 loses to 2 tbd5 ! tba4 3 tDc3+!, while l . . .�f4 and l . ..�e3 lose directly to
lb+�
=I-
1 02
Original 2 lbd5+ ! . 1 . . . 'iii>g4 fails to 2 lbd5 �4 3 lbe3+ 'i&i>f4 4 'i&i>d5 and, fi nally, l . . .'i&i>g3 loses to 2 lbd3 �g4 3 lbe5+ 'i&i>g5 4 lbc4 lba4 5 �d5 . Therefore, Black has to move his king away from f4, and then White wins by 1 . . .'iii>f2 2 lbd5 ! lba4 3 lbf6 �5+ 4 'i&i>d5 ! and the pawn ad vances. But the line discussed above (after 6 lbb4 'i&i>f3 !) is diagram 1 02 with White to play, and after 7 lbd5 lba4 ! 8 lbf6 'iii>f4 ! (this is yet another recip rocal zugzwang, ZS) 9 �d5 lbb6+ ! lO 'i&i>c5 'i&i>e5 ! Black's king is close enough to draw, while after 7 lbd3 �e4 8 lbb2 'i&i>d4 ! Black can prevent �4. Returning again to diagram 1 0 1 , White can also try the lbc7-a8 idea. ln this case Black can draw by 6 lllc7 �4 ! 7 'i&i>e7 lbb6! 8 'i&i>e8 'iii>f4 ! and Black is just in time to meet lbd5+ by ... 'i&i>e5. However, if Black's king were on the second rank he would be loo slow to attack the d-pawn and White would win.
v
lb 79
It follows from this analysis that diagram 1 0 1 is lost for Black when his king is on the second rank (be cause of lllc 7) or when his king is on f3 (since he is in zugzwang after White's lbb4). Working backwards, we can now understand why Z3 (W'i&i>f5, lbc5, �d6 v B'i&i>g3, lbb8) is reciprocal zug zwang. Black to play must either put his king on f3, play his king to the second rank, or move his king to the h-file. The first two lose to lba6 fol lowed by 'i&i>e6, forcing a winning version of diagram 1 0 1 in either case. Not surprisingly, the h-file isn't the ideal location for Black's king and a modification of the same method wins (the detailed analysis is given below). It turns out that Z2 is closely re lated to Z3. If Black is to play, then after 1 . .. 'i&i>g3 2 'iii>f5 he is in Z3 at once, while after l . . . 'i&i>g2 2 �f6 he must fall into Z3 or Z4 fairly quickly. Z l and Z2 are virtually the same, since the interpolation of the moves lbc5 and ...lbb8 changes very little. These are the general principles governing the position, so now the concrete analysis should appear somewhat less confusing. We have not analysed 1 �g4? so far, but after 1 . . .lbb8 2 �f5 'i&i>g2 3 lbc5 'i&i>g3 ! 4 lba6 lbd7 ! 5 'i&i>e6 lbb6 ! 6 lbb4 'iii>f3 ! we have reached Z4 with White to play, so the position is a draw. 'i&i>h3 1 ... l . . .lbb8 2 lbc5 ! and l .. .'i&i>g2 2 �f6 ! lbb8 3 lbc5 ! will transpose to later analysis.
80 �+� v �
2 �gS! By means of a king triangulation, Black has been forced into Z l . 2 �b8 Sooner or later Black will have to play . . . �b8, but it doesn' t matter much when he plays it. After 2 ... �g2 (2 . . .�g3 3 �f5 ! �b8 4 �5 ! leads directly to Z3) 3 �f6 ! White threat ens to win by d7, so 3 . . . �b8 is forced, when we transpose into the main line after 4 �c5 ! . 3 �S! This is Z2. �g2 3 ... 4 'iti>f6! �g3 After 4 ... �f3 5 �a6 ! �d7+ 6 �e6 ! �b6 7 �b4 ! Black has fallen into Z4. S �fS! We have arrived at Z3. The re maining analysis is given in outline only, as the most important points have been covered in the earlier dis cussion. �3 s Or: 1 ) S 'iPn 6 �a6 ! �d7 7 �e6 ! �b6 and now: l a) 8 �b4? �f3 ! (Z4) 9 �d5 (9 �d3 ..t>e4 10 �b2) �a4 ! 10 �f6 �f4 ! (Z5) 1 1 �d5 �b6+ ! 12 �c5 �e5 ! draws. l b) 8 �c7 �a4 (8 . . .�f3 9 �a8 ! �a4 10 �d5 !) 9 �e7 �b6 10 �e8 �e3 1 1 �a8 ! �c4 1 2 d7 ! �d6+ 1 3 �e7 wins. 2) s �h4 6 �a6 ! �d7 7 �e6 ! �b6 8 �b4 �g5 9 �d3 �g6 10 �e5+ �g7 1 1 �c4 �a4 1 2 �d5 �f7 1 3 �c6 �e8 14 d7+ �d8 15 �e5 promotes the pawn. ...
...
•.•
...
3) s �g2 6 �a6 ! �d7 7 �e6! �b6 8 �c7 ! �a4 9 �e7 wins as in line lb. 6 lDa6! 7 �e6! 8 �b4! This is Z4. After 8 �c7 �a4 9 �e7 �b6 White can still win by going into reverse, but 10 �e8? throws away the win completely af ter 10...�e4. � 8 The alternatives 8 . . . �e4 9 �d5 ! �a4 10 �3+ !, 8 ...�g4 9 �d5 �4 10 �3+ �f4 l l �d5 and 8 ...�g3 9 �d3 �g4 10 �e5+ �g5 1 1 �c4 �a4 12 �d5 are no better. 9 lbdS! lLJa4 10 �f6 �S+ 11 �dS! and White's pawn advances, with an easy win. There are 347 reciprocal zug zwangs with �+�d6 v �...
...
With the pawn on d5, once again Halberstadt made the only worth while discovery. (103): White must pay heed to the warning notice 'Take care - recipro cal zugzwangs about' . The two alter natives are: 1) 1 �S? �a7 ! (now White is in reciprocal zugzwang) 2 �f6 (White cannot land a �d7 cheapo, e.g. 2 �d7 �xd7 ! 3 e6 �b6 ! 4 e7 �c8 ! 5 e81i' �6+! or 2 �g5 �b6! 3 �7+ �xd7 ! 4 e6 �c5 5 e7 �e6+ 6 �f6 �c7 ! and the pawn is stopped) �b6! 3 �g7 �xc5 ! 4 �xf8 ..t>d5 ! drawing. 2) 1 �gS! �a7 (Black's only chance is to aim for a counterattack
lb+�
·- - - . - - - - - . �,� • . • � � . � d . d . d � �� - - I B B B . - - ,
v
lb 81
V,
+/= Halberstadt, 1949 Ist Comm., 1ijdschrift v. d. KNSB 103
against the knight on c5, because l ...c:i;b8 2 �h6 c:i;c7 3 �g7 ! wins easily) 2 c:i;f5 ! (not 2 �h6? �b6 ! 3 �7+ lbxd7 ! 4 e6 lbf6) and now Black is the one in the reciprocal mgzwang. White wins after 2 ...�b8 3 �d7+ lbxd7 4 e6 (here the pawn promotes with check) or 2 . . .�b6 3 �7+! lbxd7 4 e6 ! lbc5 (and here &lack's king blocks b6) 5 e7 ! lbb7 6 *e5 and wins. There are 1 3 1 reciprocal zug zwangs with lb+�d5 v lb. The next position also depends on reciprocal zugzwangs. (104): Black to play draws com fortably by l . ..c:i;f5 2 d5 lbg4 3 d6 �5, so assume that White moves first. The alternatives are: 1) 1 dS? c:i;f5 ! 2 d6 lbg4 ! 3 lbc4 � ! 4 llle3 + �g5 and draws. 2) 1 �e6? c:i;f4 ! (now White is on dM: wrong end of a reciprocal zug zwang) 2 d5 lbg4! (a second recipro a.l zugzwang) 3 d6 lbe5 ! and again
+I=
104
Original we have reciprocal zugzwang. The parity is wrong for White to break up Black's defensive formation. 3) 1 c:i;f6! (triangulation does the trick; now Black is to move in the reciprocal zugzwang) and now: 3a) 1 c:i;g3 2 �f5 ! lbg4 3 lbfl + ! c:i;h4 (or 3 ...�f3 4 d5 ! lbh6+ 5 �e6 !) 4 d5 ! lbh6+ 5 �e6 �g5 6 d6 �g6 7 d7 and wins. 3b) 1. c:i;hS 2 d5 lbg4+ 3 �e6 ! �g6 4 llle4 ! (4 d6? lbf6!) lbh6 5 d6 lbf7 6 d7 ! winning. 3c) 1. c:i;f4 2 �e6 (now Black is in zugzwang) lbg4 3 d5 ! (and again) lbe5 4 d6 ! (a third time) lbd3 5 lbb3 ! lbe5 6 lbc5 and the fourth zugzwang finishes Black off, since 6 . . .lbg6 7 �f6 leaves Black helpless. ...
.•
••
There are very few winning posi tions with the pawn on d3 or d2; in deed White can only hope to win if Black's knight is virtually impris oned in the initial position. Two po sitions by Halberstadt illustrate the exceptional circumstances in which
82 l'Li+�
v
lLi
there are winning chances. The first of these is a particularly fine piece of analysis.
l'Lid5+ 15 'it>c6 ! l'Lif6 16 lt:Jg7 'it>g5 17 l'Lie8 ! lt:Jg4 1 8 'iti>d5 ! and wins. 2) 10 'iti>g4 1 1 l'Lid4 ! 'iti>g3 1 2 lt:Je6 ! lt:Ja6 ( 1 2 . . .lt:Jd3 1 3 'ili>f6) 1 3 'it>d7 ( 1 3 d7 ? lt:Jb8 ! ) lt:Jb4 1 4 'ili>c7 lt:Jd5+ 15 'it>c6 ! lt:Jf6 1 6 lt:Jg7 ! lt:Jg4 ( 1 6 . . . 'iti>g4 17 lt:Je8 ! ) 17 lt:Jh5+ 'it>h4 18 'iii>d5 ! 'it>xh5 1 9 'iii>e6 ! and the pawn promotes. The second position is a simple but neat idea. .••
+I=
105
Halberstadt, 1961 (105) : White's first move is forced, because if Black's knight es capes the draw is certain: 1 l'Lie5 ! 'it>h3 (after l . . .l'Lig2+ 2 'iii>g4 ! 'iti>gl 3 'iii> g 3 ! we have a position of recip rocal zugzwang; Black to play loses after 3 . . . 'iti>h l 4 l'Lid3 'iti>gl 5 l'Lib4 ! l'Liel 6 d4 ! l'Lig2 7 l'Lic2 !) 2 d4 ! l'Lig2+ 3 'ili>f3 ! (not 3 'iti>e4? 'iti>h4 ! 4 l'Lif3+ 'it>h5 drawing) l'Liel + (3 ...lt:Jh4+ 4 'it>e4 ! ) 4 'it>e4 ! l'Lic2 5 d5 ! l'Lib4 6 d6 ! lt:Ja6 7 'it>d5 (7 lt:Jc6 lt:Jc5+ 8 'iii>d5 ! lt:Jd7 9 'it>e6 ! is also possible, trans posing into the main line) lt:Jb8 8 lt:Jc6 ! lt:Jd7 9 'it>e6 ! lt:Jc5+ 10 'iii>e7 ! and now: 1 ) 10 'it>g3 (10 ... 'it>h4 1 1 lt:Jd4 �g4 transposes to the main line) 1 1 lt:Jd4 'iti>g4 ( l l . . .'iii>h4 12 lt:Jb3 lt:Jxb3 13 d7 ! lt:Ja5 14 'iii>d6 lt:Jb7+ 15 'iti>c7 ! wins because the pawn promotes with check) 12 lt:Je6 ! l'Lia6 1 3 �d7 lt:Jb4 ( 1 3 . . . 'iii>f5 14 'iti>c8 ! ) 14 'it>c7 ! ...
106
+I=
Halberstadt, 1961 Shakhmaty v SSSR (106): White wins by 1 lt:Jh5 ! 'it>f7 2 e4 ! (not 2 e3? �g6 ! 3 'iti>xe8 �xh5 !) 'iii>f8 (after 2 ... 'it>g6 White can now play 3 'it>xe8) 3 e5 ! �f7 4 e6+ ! �f8 5 e7+ 'iti>f7 6 lZJg3 lZJf6+ 7 'it>d8 lZJe8 8 l'Lif5 and wins. There are 38 reciprocal zug zwangs with lZJ+�d3 v lZJ and 44 with lZJ+�d2 v lZJ, none of them very exciting. We end with the longest win in the ending of lZJ+d� v lZJ (34 moves). Readers should not be sur prised to learn that reciprocal zug zwangs crop up during the solution.
lb+l!.
+I
107
108
v
lb 83
+/-
Original
Original
(107): 1 'it>b7 (the transposition 1 �c7 lbb5+ 2 'ifr>c6! is equally good) �b5 2 'ifr>c6 ! lbd4+ 3 'it>c5 ! lbb3+ (3 ... lDf3 4 d6 ! lbe5 5 'it>d5 ! trans poses to the main line) 4 'ifr>b5 lbd4+ 5 �c4 lbf3 6 d6! lbe5+ 7 'ifr>d5 ! lbd7 8 �e6 ! lbb8 (8 ... lbb6 loses more quickly after 9 lbf5 'ifr>a2 10 lbe7 ! �a4 1 1 �d7 lbb2 1 2 'it>c7 lDd3 1 3 �g6 lbc5 1 4 'ifr>c6) 9 'iPe7 ! 'it>b2 (or 9 ... 'ifr>a2 1 0 �d8 lbc6+ 1 1 'itr>d7 ! �5+ 12 �e6! lbd3 1 3 'ifr>f6 lbc5 14 �e7 ! lbd3 1 5 lbe6 lbe5 1 6 lbd4 ! �a3 17 �e6 ! and now 17 ... lbd3 1 8 �d5 and 1 7 . . .lbg6 1 8 lbc6 'itr>a4 1 9 � f7 both win for White) 10 'itr>d8 ! �6+ ( 1 0. . . lba6 1 1 lbe6) 1 1 �d7 ! �5+ ( 1 1 .. .lbd4 1 2 �c8) 12 �e6 ! �3 ( 1 2 . . . lbg6 1 3 lDf5 'itr>c3 14 �h4! lbf4+ 15 'ifr>f5 !) 1 3 'itr>d5 ! (and not 1 3 �f6? lbc5 ! 14 'itr>e7 lbd3 ! 15 �6 lbe5 ! 16 lbd4 �c3 ! drawing, in contrast to the note to Black's 9th move) lbb4+ 14 �e4 lba6 15 lbe6! �b8 16 lbc5 ! and after this lengthy introduction we have arrived at the critical position:
Readers will notice a striking re semblance to the analysis of diagram 100, and those who worked their way through the analysis of that po sition will not find this one too diffi cult. Basically, the positions with 'itr>e5 v 'ifr>c2 and 'itr>d5 v 'itr>c3 are both reciprocal zugzwang. 'itr>d5 v �c3 is zugzwang because Black to play must move his king to the second rank (as . . .'itr>b4 loses lba6+). Then White can win by lba6 followed by 'ifr>e6 because Black's king is simply too far away. The position with 'itr>e5 v 'iii>c2 is slightly more subtle. White is not threatening lba6 because Black can reply ...lbxa6 and meet d7 by ...lbb8. If Black plays . . .'itr>d2, however, then the pawn promotes with check and so lba6 wins. If Black plays . . .'itr>c3, White of course replies 'itr>d5 . There fore Black has nothing better than ... 'itr>b2, but then, as we shall see be low, Black loses because he can no longer meet �e6 by . . . 'ifr>d2. The main line continues:
84 lb+� v lb
'it>c2 16 After 1 6 . . . 'it>a3 1 7 'it>d5 B lack has to go back with his king ( 17 ... 'it>b4 1 8 llla6+), when 1 7 . . . 'it>b2 1 8 c,i;>e6 leads into the main line. Of course, 1 6...'it>c3 17 'it>d5 ! at once puts Black in zugzwang. 17 'it>eS! Reciprocal zugzwang 'it>b2 17 1 7 . . . 'it>d2 (as usual, 17 ... 'it>c3 is met by 18 'it>d5 ! ) fails to 1 8 llla6 ! llld7+ 1 9 'it>e6 ! lllb6 20 lllb4 'it>c 1 2 1 llld5 llla4 22 'it>e7. 18 c,i;>e6 Black's king is not on the d-file, so the trick 1 8 llla6? lllxa6 ! 19 d7 lllb8 ! doesn't work. 'it>c3 18 18 ... 'it>c2 loses to 19 llld7 llla6 20 'it>d5 ! 'iti>d2 2 1 c,i;>c6 lllb4+ 22 'ifi>b5 llld 3 23 'it>c4 c,i;>e3 24 lllc5 lbe5+ 25 c,i;>d5 ! . Note that if Black could play 1 8 . . . 'it>d2 here, then he would draw after 1 9 llld7 llla6 20 'it>d5 c,i;>e3 ! 2 1 'it>c6 c,i;>d4 ! approaching the d-pawn. This explains why 'it>e5 v c,i;>c2, for example, is a reciprocal zugzwang rather than a simple zugzwang. With
White to play, 'it>e6 is met by 'it>d2!. Black to play must make a conces sion: either to move his king away from d2, put his king on the tacti cally exposed d-file, or fall into a further reciprocal zugzwang after ... 'it>c3. 19 'it>dS! Not l 9 lbd7? llla6 20 'it>d5 lllb4+ ! drawing. Now Black is again to move in a reciprocal zugzwang and the rest is easy: 1 9 ...'it>d2 20 llla6 ! llld7 2 1 c,i;>e6 ! lllb6 22 lllc7 llla4 (22 ...'it>c2 23 llla 8) 23 'it>e7 lllb6 24 'it>e8 'it>e3 25 llla8 ! lllc4 26 d7 ! llld6+ 27 'it>e7 lllb7 28 lllb6 c,i;>d4 29 lllc8 'it>c5 30 lbd6 lba5 3 1 'it>e8 lbc6 32 lllf5 'it>b6 33 llle7 with capture of the knight or pawn promotion next move.
...
•••
.••
a
..
.,
. ·· .
It is remarkable that the formation with lllc5 and �d6 v lllb 8 gives rise to so many reciprocal zugzwangs. some with the kings on the kingside and some on the queenside. There are a total of seven, given here to gether with the length of the Black to play winning lines (we need only
b
' · .
c
d
7
214
279
23 1
185
6
230
492
432
347
5
257
279
203
131
4
185
91
57
62
3
92
51
24
38
l
83
85
36
44
lb+� v lb 85
give the king positions): �d5 v �c3 ( 1 5 moves), �e5 v �c2 ( 1 7 moves), �f5 v �g3 ( 1 5 moves), 'iti>g5 v �h3 (17 moves), �f4 v �h4 ( 1 8 moves), �b5 v �a7 ( 1 4 moves) and �e8 v �c8 (4 moves). We saw the first two in diagram 107 just above, while the third and fourth were analysed in diagram 100. The fifth leads into the fourth after one move. We have not analysed the last two, but they are
relatively simple and may be safely left as an exercise for the reader. The table on the previous page is a breakdown of the 4 1 28 reciprocal zugzwangs in the ending of lb+� v lb, according to the square occupied by White's pawn. Congratulations to anybody who can discern a logical pattern in this table!
White's winning chances are far fewer than in Chapter 1 . The bishop can hold up the advance of the pawn from long range, and the only hope of advancing the pawn lies in block ing the bishop's influence with the knight. This requires the co-opera tion of both king and knight, so the first condition for a win is that White's king and knight both have to be well placed. However, even this is often not enough for success. Take, for exam ple, the following position:
play 2 .!Lib5 (after 2 .!Lic4, Black just makes a pass move with his king) and then �b6, but even this doesn't help; Black passes, and then meets 3 'it>b6 by 3 ... i.c8 4 .!Lid4 i.d7 5 lLic6 i.h3, when 6 b5 is met by 6 . . . i.fl . White cannot make progress. The key point is that the bishop is always occupying a diagonal with at least five squares. The king and knight be tween them can only cover four, so the bishop cannot be driven off the diagonal by direct attack. The only method is to use the knight as a blockader (as with 2 .!Lib5 above) and then attack the bishop with the king. However, in this case the knight ob structs the pawn and the bishop has time to switch to another diagonal. Let us shift diagram 1 09 up one square:
I
109
= =
Original (109): The position of Black's king is irrelevant, since he can draw without ever using his king. After 1 .!Lid6, White threatens to play 2 .!Lic4 followed by b5, but Black replies l . . .i.a6. The only way to force the bishop off the a6-fl diagonal is to
l lO
+I-
Original
tD+L!i v .t 87
(1 10): In this case Black cannot maintain his blockade because there isn't enough space for the bishop. After I lDd7 ! i.a7 2 tDb6, there is al ready a threat of 3 '1tr>b7, so 2 ...i.b8 is forced, but then 3 tDc8 ensures the pawn's advance. White wins after 3 ...�g3 4 b6 �f4 5 lDd6 �e5 6 tDe8 �d4 7 tDc7 �c4 8 �b7 �c5 9 tDa6+. If Black's bishop starts on a5, then he still loses: I tiJd7 �g3 2 tDc5 �g4 3 tiJb7 (the a5-d8 diagonal is loo short and White can force the bishop away) i.el 4 b6 i.f2 5 tDc5 ! i.g3 6 b7 i.b8 7 tDa6 i.a7 8 �c7 �f5 9 �c8 �e5 10 tDc7 and 1 1 tDb5 winning. In fact both these positions are won even if Black moves first, but the analysis is not relevant to our dis cussion. Of the four endings with a minor piece and a pawn against a minor piece, the ending with tD+L!i v i. is the only one in which a draw is possible without the aid of the de fender's king (except for the trivial case of opposite-coloured bishops). A wide range of tD+L!i v i. positions may be evaluated as drawn without any need to consider whether the de fender's king is close enough to the pawn. Averbakh gave a pair of dia grams which conveniently show the 'drawing zone' for the white pawn. (111): This diagram applies if Black has a light-squared bishop. If the pawn is on any of the marked squares, and the bishop is control ling the light square immediately in front of the pawn, then the position
111 Averbakh, 1958 is a draw. It follows that the best winning chances are with an edge pawn, because in this case the bishop can never mount a successful de fence on its own. If Black possesses a dark-squared bishop, then the diagram is just a mirror image of the above:
Averbakh, 1958 As in the first chapter, we will divide up the material into four sec tions according to the file occupied
88 lO+� v .t
by the pawn. Of the four cases, the first is the most interesting. 2. 1 : 2.2: 2.3: 2.4:
lO+� v .t lO+b� v .t lO+c� v .t lb+cill v .t
2.1:
88 1 06 1 24 138
ltJ+lill v i.
We will start with the white pawn on a7. This ending occurs fairly often in practice, and is almost equally often mishandled. The position of most importance forover-the-board play ers was discovered by Horwitz in 1 885.
1 13 +IHorwitz, 1885 (version)
(113): After 1 l0c5, threatening 2 l0b7, Black must play 1 . . ..ia8. It might seem that White can continue with 2 �b8, forcing the bishop to move, and then 3 l0b7 promoting the pawn. However, 2 �b8? is a blunder throwing away the win, because Black replies 2 . . . �d6 and suddenly
White cannot win. The reason is that after 3 �xa8 �c7 the parity is wrong (see diagram 1), while 3 l0b7+ �d7! reaches a reciprocal zugzwang with White to play; the knight must move, but then Black's king returns to d6. Horwitz's win runs: 1 l0c5 .ia8 2 l0b7 (the key to the solution is to play the knight to a5) �e6 3 l0a5 �e7 (3 ...�d5 4 �b8 �d6 5 l0c4+ is hopeless because 5 . . . �d7 loses to 6 l0b6+ and 7 l0xa8, while 6 . . .�c6 6 �xa8 �c7 7 l0d6 puts Black in zug zwang) 4 �c8 (the second key idea; e7 is the best square for Black's king, so White passes) �e8 (4 . . .�d6 5 �b8 �d7 6 l0b7 puts Black in recip rocal zugzwang, while 4 . . .�e6 5 �b8 �d6 6 l0c4+ transposes into the note to Black's third move) 5 l0c4 (now that the king is on e8, Black cannot move his bishop be cause of l0d6+ and l0b7) �e7 6 �b8 (moving in for the kill) �d8 (or 6 ...�d7 7 l0b6+ and 8 l0xa8) 7 lOaS �d7 8 l0b7 �c6 9 �xa8 �c7 10 l0d6 and wins. In fact, Horwitz's position was slightly more subtle than this simpli fied version. He started with W�c7, l0b8, �a7 v B�b4, .ta8. White is aiming for the previous diagram, but he must also prevent ... �b5-a6, which would guarantee the draw. The solution runs 1 �b6 ! (keeping the black king at a safe distance) �c4 2 l0a6 �d4 (2 . . . .ih 1 3 l0c5 .ia8 4 l0b7 �d5 5 �c7 ! �e6 6 l0a5 transposes to the previous diagram) 3 l0c5 �d5 4 l0b7 �e5 5 �c7 �e6 6 l0a5 and we are in the analysis of diagram 1 1 3 . We will refer to the
lb+L!.
winning method of diagram 1 1 3 as the 'Horwitz win'. The key point about diagram 1 1 3 is that a position with �c7 v �e7 is winning for White, but a position with �b8 v �d7 is winning only un der favourable circumstances, so White must manoeuvre carefully before playing his king to b8 . Once you have seen the winning idea, it no longer appears especially difficult, but a number of grandmasters have failed to spot it.
• • • • • • • • • • • • • • • • • • • • • • • • · -• • @ • • m • • • -/
1 14
Stein-Dorfman USSR 1971 (114): This is actually a double example, because an identical posi tion arose in the game Sakaev-Sunye Neto, Sao Paulo 199 1 . Neither of the grandmasters playing Black found Horwitz's winning line. Stein-Dorf man continued 1 �el �bl ? (as we know from diagram 1 1 3, the Hor witz win continues with 1 . ..ll:ia4) 2 �d2! �xal 3 �cl ! lLic4 4 �c2! 'h-lh. Sakaev-Sunye Neto, on the other hand, went 1 �e3 �bl ? ( 1 .. .ll:ia4) 2
v
.t 89
�d2! lLic4+ 3 �dl ll:ia5 4 .ig7 ll:ib3 5 .if6 lLic5 6 .ig7 lLJci3 7 .ial ! ll:ib4 8 .ig7 lLic2 9 �d2 ll:ia3 10 �dl lLic4 1 1 .ial ! lbe3+ 12 �d2 ! 1'2-1/2. Study composers (and judges) have sometimes been equally igno rant:
-I+
1 15
Atanasov, 1975 Comm., '64 ' (115): We have taken the position after two introductory moves. The composer believed that White could draw, but in fact the position is an easy win for Black, even when White moves first: 1 'it>g6 ll:ib6 2 �f5 ll:ic4 (the simplest win is by 2 . . . ll:ia4 3 .ial �c2 4 'it>e4 'it>bl 5 �d3 ll:ic5+ 6 �c3 �xal 9 'iPc2 ll:id3) 3 .ial �c2 4 �e4 and now the composer gave 4 . . . �b l ? 5 'iPd3 �xal 6 �c2 with a draw. However, Black can play 4 ...ll:ib2 5 'iPe3 ll:ia4 6 �e2 �c l with the Horwitz win. A second example is a 'White to play and draw' study by Grushko, awarded a Commendation in the 1 979 Chervoni Girnik tourney. This
90 lLJ+8 v i.
reached the position W�c5, i.al v B �b3, lLJb2, 8a2, when the com poser continued 1 �d4 �c2 2 �e3 �bl ? (2...lLJa4 wins as usual) 7 �d2 �xal 8 �c l with a draw. I have found other such cases, which only goes to show how depressingly igno rant of elementary endgame theory some composers and judges are. The following group of studies are all based on the Horwitz win (in deed, it is hard to imagine an inter esting position with lLJ+8a7 v i. which does not depend on the Hor witz win).
cannot keep White's king penned in, since 2 ...�e7 loses to 3 lLJds+ ! ; after 2 . . . �d8 3 lLJe6+ �e7 4 lLJg7 White wins at once) 3 lLJds ! (gaining a vital tempo) i.al 4 �f7 ! and now that his king is on the f-file, White wins by the Horwitz method.
=I 1 17 Nestorescu, 1977 (end of study) 5th HM, Bulletin Problemistic
1 16
+I+ Schostak, 1930 Shakhmatny Listok
(116): Black to play mates in two, so we may assume that White moves first. At first sight White cannot win, because the Horwitz method de pends on White's king occupying the f-file, but it turns out that Black's pieces are on particularly unfavour able squares: 1 �g8 ! (not 1 �g7? .i.c3+ 2 �g8 �e7 with an easy draw for Black) i.c3 2 lLJc7+ ! �d7 (Black
(117): In this example White must play accurately to deny Black a Hor witz win. The solution runs 1 i.c3 ! lLJe2 and now: 1 ) 2 i.b2? (this move fails be cause Black gains a vital tempo by attacking the bishop later) �gl ! 3 �g4 �f2 4 �f5 (White must head towards a3, otherwise Black has a straightforward Horwitz win) �e3 5 �e5 �d3 6 �d5 lLJf4+ 7 �c5 (or 7 �c6 �c2 8 i.g7 lLJd3) �c2! 8 .i.al �bl , followed by 9 . . .lLJd3+ and 1 0...lLJb2. 2) 2 i.eS? (or anywhere else fur ther down the diagonal) �gl ! 3 �g4 �f2 4 �f5 �e3 and White loses be cause he cannot play his king to e5
lLJ+� v .t 91
or e6 (the latter because of ...lbci4+), so he will be too slow. 3) 2 .tal l (the only square to draw) �gl 3 ..t>g4! �f2 4 �f5 ! ..t>e3 S �e5 ! �d3 6 �d5 ! lDc l 7 �c5 ! �b3+ (or 7 . . . ..t>c2 8 �b4! �bl 9 i.f6 lDd3+ 10 ..t>a3) 8 �b4 (White is just in time) lDxal 9 �a3 drawing.
draws by 4 �h5 �f5 ! or 4 lbci7 .th8 ! 5 �g5 �d5 ! 6 �g6 �e6) �e2 (Black is to move in this reciprocal zugzwang; moving the bishop only makes White's task easier, for exam ple l . .. .th8 2 �g2 �e2 3 �g3 �e3 4 �g4 �e4 5 �g5 ! �d5 6 �g6 !) 2 �g2 ! �e3 (or 2 ....te5 3 ..t>h3 ! �f3 4 �h4 ! �e4 5 ..t>g5 ! and the king penetrates) 3 �g3 ! �e4 4 �g4 ! �d5 (4 . . .�e5 5 liJd7+) 5 �f5 ! .th8 6 ..t>g6 ! �e5 7 �f7 ! �d6 8 lDg6 with a Horwitz win. In fact, Pankov's study was one of a pair of positions. The other one was W�a8, lLJh6, �h7 v B�a5 .tf6, but this effectively just a reflection of the above diagram; the solution runs 1 ..t>a7 ! �b5 2 �b7 ! �c5 3 �c7 ! ..t>d5 4 ..t>d7 ! �e4 5 �e6! .th8 6 �f7 ! .tal 7 �g6! (this move is the only real difference; White prepare liJf5) ..t>d5 8 liJf5 ! .th8 9 �f7 and again While reaches the Horwitz win. The following study is probably the most elegant pre-computer com position. (119): As before, White's target is to get his king to c7. White can try: 1 ) 1 lDb6? .tb7 ! puts White in reciprocal zugzwang, a situation un changed by the further move 2 ..t>d6 �d8 ! . If Black were to play in this latter position, then he would be un able to move his bishop, because . . ..thl would be met by lbci5, but playing . . . �e8 allows �c7. How ever, White to play cannot maintain his bind and so the position is a draw. 2) 1 �eS! (after this White can choose between �d6 or �e6, so he can always gain the opposition) .ta8 •
+I= Pankov, 1993 Sp. Pr., Dolgov-70 Tny.
1 18
(1 18): We know from diagram 1 1 3 that White will win if his king can reach f7. The effect of this is to immobilise the knight and bishop. If the bishop moves, then Black loses control of g5 and this gives White's king an easy route to f7 via g5 and g6; on the other hand, if the knight moves then Black's king gains ac cess to e6 and White's king can no longer occupy f7. Thus most of the action is restricted to king moves, and so this position depends on the opposition. White wins by 1 �gl ! (not 1 �g2? �e2 ! and White falls victim to a reciprocal zugzwang; af ler 2 �g3 �e3 ! 3 �g4 �e4 ! B lack
92 lll +� v i.
• ••• • �J • • • B B B B • • • • tt:JB B.tB B • • • • • • • • • • • • +I= Prokop 1 930 lst Pr., Ceskoslovensky Sach
1 19
+I=
1 20
Original
,
(the lines 1 . . . i.g2 2 lllb6 ! i.b7 3 �e6 ! 'iti>d8 4 �d6! and l . . .i.b7 2 'iti>d6 ! 'iti>d8 3 lllb6 ! leave Black in reciprocal zugzwang, while 1 . . .i.c6 2 'iti>d6! i.b7 3 'iti>c7 leads to the Hor witz win straight away) 2 lllb 6! i.b7 3 �e6 ! (Black's . . . i.e4-a8-b7 has changed the parity, so White can win with this switchback) �d8 4 �d6 ! 'iti>e8 5 �c7 with the Horwitz win. The above compositions repre sent the best creative achievements of pre-database days. However, these studies were fairly basic and com posers don't seem to have made the most of the possibilities afforded by this ending. There are 66 reciprocal zugzwangs in the ending of lll+�a7 v i., and I would like to present a few original positions based on some of these. (120): Suppose firstly that White is to play. The key to the position is that the situation with �f4 v 'ifi>f2 is reciprocal zugzwang, as the follow ing analysis proves:
1 ) 1 'ifi>f4? 'ifi>f2 ! (now White must allow Black's king to either e2 or e3) 2 'iti>eS (or 2 lllb 3 'iti>e2! 3 lllcS �d2 ! 4 �e5 �c3 ! 5 llla4+ �b4 6 lllb6 �bS) �e3 ! 3 �d6 �d4 ! 4 �c7 �c5 5 tlld3+ �b5 drawing. 2) 1 �f5! �f2 (or l ...�g2 2 �e6 �f3 3 �d7 �e4 4 �c7, and White wins after 4...�e5 5 lllb3 ! or 4 ...�d5 5 lllb 3 ! �e6 6 llla 5; l ...�fl 2 �e6 is even worse) 2 �f4 ! �el (2 . . .i.hl 3 �e5 �e3 4 lllb 3 ! �d3 5 llla5 and now Black will lose a tempo when White threatens t'Llc6, e.g. 5 . . . �c3 6 �d6 ! i.a8 7 �c7 �b4 8 �b8 or 5 ...i.a8 6 'iti>d6! 'iti>d4 7 �c7) 3 �e3! (another tricky move, but the idea is to keep Black's king away from a6 and b6; not 3 �e5? �d2! 4 lllb3+ �c3 5 tlld4 �c4, nor 3 lllb 3? �e2 4 lllc5 'iti>d2 ! 5 'iti>e5 �c3 !) and now: 2a) 3 � 4 lllb 3 �g2 5 �d4 'iti>f3 6 �c5 'iti>e4 7 �d6 �f5 (or 7 . . . �d3 8 'iti>c7) 8 �c7 �e6 9 lllaS with a Horwitz win. 2b) 3 i.b7 4 �d4 �d2 5 lllb3+ �c2 6 llla5 (gaining time) i.a8 7 ...
...
lb+� v .t 93
Wc5 �c3 8 �b6 �d4 9 �c7 is the same. 2c) 3..ihl 4 �d4 �d2 5 lDb3+ Wc2 6 lDas, and White's subsequent Wc5 will gain time by forcing the bishop back to a8. 2d) 3... �dl 4 lbd3 �c2 (4 ....tb7 S �d4 �d2 6 lDc5 .ta8 7 lDb3+ �c2 8 i0a5 wins, as in line 2b) 5 �d4 ! �b3 6 �c5 ! �a4 7 �b6 ! Uust in time) �b3 8 lLJe5 �b4 9 lDc6+ �c4 lO lDas+ �d5 1 1 �c7 with the Hor witz win. Black to play draws by 1 ...�g2 2 �4 (or 2 �f5 �f3 and Black's king is free to head for a6) �f2 !, and the reciprocal zugzwang arrives with White to play.
White's �c5 will threaten lDc6, and then Black will have to lose time playing his bishop back to a8. On the other hand, if White's knight moves then he not only loses time, but also gives B lack's king access to e6 and thereby d7. White can play: 1) 1 �c2? �e2 ! (now White is to play in the reciprocal zugzwang) 2 �c3 �e3 ! 3 �c4 �e4 ! 4 �c5 (4 lDf7 �f5 ! 5 �c5 �e6 also defends) �e5 ! 5 �b6 (5 lDc6+ �e6 !) �d6 ! and White is kept out of c7. 2) 1 �cl! and now: 2a) 1 ...�e2 2 �c2 ! and play pro ceeds according to the direct opposi tion. 2b) 1 .. ihl 2 �c2 �e2 3 �c3 �e3 4 �c4 �f4 5 �c5 (now Black must lose time) .ia8 6 �d6 reaching c7. 2c) 1 . ..td5 2 �c2 �e2 3 �c3 �e3 4 �b4 �d4 5 �b5 ! (this is similar) .ia8 6 �b6! and wins. 2d) 1.. 'iPC2 (this is the most resil ient defence, because 2 �d2? �f3 ! 3 �d3 �f4 ! 4 �d4 �f5 ! only draws; White's king cannot move to d5, while 5 �c5 �e5 ! regains the oppo sition) 2 �b2! (this is the key; taking the direct opposition doesn't win, so he seizes the distant opposition in stead) �e2 3 �c2! gives rise to a re ciprocal zugzwang. It is important to realise that only the positions with Kcx v Kex are reciprocal zugzwang. The positions with Kbx v Kfx are winning for White whoever moves first, because when the kings march down the files, White's king will end up on b6 and Black's on f6, but then .
.
+I=
121
Original (121): Black to play draws com fortably after l .. .�d2, heading for d6, so we may assume that White is lO move. The play resembles Pan kov's diagram 1 1 8, but here there is an extra subtlety. As usual, White's aim is to reach c7 with his king. If Black ever moves his bishop, then
94 llJ+l!. v i. �c7 wins for White. Thus 3 �f2 is met by 4 �b3 �f3 5 �b4 ! �f4 6 �b5 ! �f5 7 �b6 ! winning, while after 3 �f3 4 �b3 ! �e3 5 �c3 ! (the pattern repeats itself) f4 6 �b4 ! �e4 7 �c4 ! �f5 8 �b5 ! �e5 9 �c5 ! �f6 10 �d6, followed by �c7 we have finally reached a nor mal Horwitz win. The zigzag manoeuvres by the two kings are striking, and bring to mind a famous Grigoriev pawn end ing (see diagram 1 2 1 in my book Tactical Chess Endings). The following position is perhaps the most interesting reciprocal zug zwang with 0.+£!.a7 v i.. ...
•••
• • • • D • • • • • • • • • • • • • • • . . . ... . • • ••• • • =lb· =I-
122
Original (122): First of all suppose that Black is to play: 1 ...i.a8 (or 1 . ..�gl 2 0.e3 �h2 3 �d2 �g3 4 �d3 �f4 5 �d4 ! and Black cannot play 5 ...i.b7 because of 6 0.c4 preventing ...�f5, while 5 ...i.a8 6 0.d5+! �f5 7 0.c7 ! i.b7 8 �c5 �e5 9 �b6 i.hl 10 liJb5 �e6 1 1 �c7 ! is a Horwitz win) 2 �e2 ! i.f3+ (2 ...i.e4 3 �e3 i.a8 is
the same) 3 �e3 ! i.a8 (Black has no really good square for his bishop; 3 ... i.b7 4 0.d2 �g3 5 0.c4 �g4 6 'iti>d4 �g5 7 �c5 �f6 8 0.a5 i.a8 9 d6 and 3 ...i.c6 4 0.d2 �g3 5 0.c4 i.h l 6 d4 f4 7 �c5 �f5 8 �d6 both gain time by attacking the bishop, with a Horwitz win in both cases) 4 0.d2 ! �g3 5 0.c4 ! (the di rect 5 d4? �f4 ! 6 c5 e5 ! al lows Black's king back) �g4 6 0.b6! (White repeatedly gains time by at tacking the bishop; 6 0.d6? �g5 ! 7 �d4 �f6 ! 8 �c5 �e6 and 6 �d4? f5 ! 7 �c5 �e6 ! only lead to a draw) i.b7 7 0.c8 ! (White's knight manoeuvres are hypnotic; now he prevents ...�f5) with the lines: 1) 7 i.hl 8 0.d6 (White gains time by threatening 0.e4) i.a8 9 �d4 'ifi>f4 10 �c5 �e5 1 1 0.b5 i.h 1 12 'iti>b6 followed by 'iti>c7 . 2) 7 �g5 8 �d4 ! �f6 9 �c5 ! �e6 10 'iti>b6 ! and again White's king reaches c7 . 3) 7 i.a8 8 0.e7 ! (not 8 0.d6? 'iti>g5 ! 9 'iti>d4 �f6 ! and draws) �g5 9 �d4 ! 'iti>f6 10 0.d5+! (the crucial point; White has totally blocked the approach of Black's king since e7 is covered, while ... �e6 and ...�e8 allow 0.c7+) 'iti>f5 ( 10 ...�f7 1 1 �c5 ! i.b7 12 0.b4 �e6 1 3 �b6 ! and 1 1 'iti>c7) 1 1 0.c7 ! (once again harassing Black's poor bishop, in order to cover e6 with gain of tempo; 1 1 �c5? i s bad after 1 l ...�e5 ! 1 2 0.c7 i.h l 1 3 0.b5 �e6 !) i.b7 ( l l .. .i.c6 12 �c5 ! i.b7 1 3 �d6 fends off Black's king) 12 �c5 ! �e5 1 3 �b6 i.h l 14 0.b5, followed by �c7, and White wins.
...
...
.•.
tEi+� v .t 95
Now we move on the case of ��a6 v .t. Here there is a distinc tion according to whether or not the bishop has the same colour as the promotion square. Starting with the case of a pawn on a6 against a dark squared bishop, the result depends on whether White's king can reach b7 without allowing Black's king to occupy a5 or b5 . In this case l£ib6 will ensure the pawn's promotion, because with the pawn on a6 B lack bas no defence corresponding to ....ta8 in the Horwitz position. Most 1uch positions are just a question of counting, so we content ourselves with a single example:
• • • • . . ·� ij . • • • • • • • • •ttJ• • • • • • • • ••• • • • • ;
+I=
123
Original (123): Black to play draws easily �er 1 . . .�f3, so assume that White moves first. He wins by using the di rect and distant opposition, and the resulting play is remarkably similar lO that in diagram 1 2 1 . 1 �g6 ! �f3 ( l ...�fl 2 �f5 �e2 3 �e6 ! trans poses) 2 �f5 ! (this position is recip rocal zugzwang, as are the positions
with �e5 v �e3 and �d5 v �d3) �e2 (if Black moves his bishop, White just plays 3 �e5 heading for b7) 3 �e6! (3 �f4? �d3 4 �e5 �c4 defends) �e3 4 �e5 ! (the zigzag manoeuvres can only arise when there is some reason why White can not take the direct opposition; in this case it is because the squares e4 and d4 are out of bounds to White's king) �e2 5 �d6 ! �d3 6 �d5 ! �c2 7 �c6 �b3 8 �b5 ! (a final finesse; White keeps the black king away from a5 and b5, and uses the time to improve the position of his knight) �a3 9 l£id6 �b3 l 0 l£ic8 .tb8 1 1 �b6 �b4 12 �b7 �b5 1 3 l£ie7 �a5 14 l£ic6+ and wins. The following position is worth knowing, as it contains a common tactical device.
• • • • . . . m • • • . . - . • • • • • • • •• • • • • • •@• •
=I+ Holm, 1910 Sydsvenska Dagbladet Snallposten 124
(124): White draws by sacrificing his bishop: 1 .lgl ! l£if3 2 .th2! (but not 2 �e2? l£ixg 1 + ! 3 �t2 l£ie2 ! and king cannot approach the pawn)
96 l'Ll+� v i.
l'Llxh2 3 �e2 ! l'Llg4 (in order to save the pawn, Black is obliged to ad vance it to h2, but then we reach the draw of diagram 2) 4 �f3 ! �g6 5 �g3 ! h2 6 �g2 ! �g5 7 �h l ! and Black cannot make progress. The tactical point in the following study, while attractive, has less prac tical importance.
-
.
. . . - . - - . . . . . . - - . � ·� · �· � � � � � � � � . . . • • • m• =/+ 1 25 Pogosiants, 1976 (end of study) lst Pr., Bulletin Problemistic
(125): The finale of this long and complex study is noteworthy in it self. White draws by 1 �f2 ! l'Llxgl ( 1 ...l'Llg5 2
1 26
=I+ S.Loyd, 1860 The Chess Monthly
(126): Black to play wins by 1 . . .l'Llf3+ 2 �e2 �g3, since 3 i.c6 is met by 3 . . . l'Lld4+. White to play draws by 1 i.d7 ! (not the immedi ate 1 i.c6+? �gl !, because Black threatens 2... l'Llg2 and there is no de fence when the pawn is on h3, for ex ample 2 i.hl �xhl 3 �fl �h2 4 �f2 l'Llf3 5 �fl �g3 wins for Black) h2 ( 1 . . .l'Llf3+ 2 �e2! l'Lld4+ 3 �e3 h2 4 �xd4 is also a draw) 2 i.c6+! �gl (or 2...l'Llf3+ 3 �e2! h l' • 4 .txf3+!) 3 .t h l (Black's king is on g l rather than f2, so there is no Hor witz win) �xh l (3 . . . l'Llg2+ 4 �e2! l'Llf4+ 5 �el ) 4 �f2 ! drawing. (127): After 1 . ..l'Lld l , White must take care to choose a good square for his bishop: 1) 2 .tf6? �b5 ! 3 �f3 (3 .te7 a2 ! 4 .tf6 �c4 5 .tat �d3 and the king reaches c2) �c4 ! 4 �e2 lLlc3+! and White loses because 5 �d2 is met by 5 ...l'Lle4+. 2) 2 i.e5? �b5 3 � �c4 4 �e2 l'Llb2! 5 �d2 �b3 ! 6 �cl l'Lld3+!
lll +l!. v i. 97
(128): If only Black could get his bishop to the long diagonal then he would draw, but thanks to the un fortunate position of his pieces this proves impossible: 1 lllf4 ! i.g4 (or l . ..i.f5 2 llld5 ! i.e4 3 lbf6+! ) 2 llld 3 ! i.h3 (or 2 ..�c8 3 lbe5 ! i.h3 4 a7 i.g2 5 lllc6 !) 3 llle5+ ! �d6 4 lllc4+ ! �d7 5 llle 3 ! and Black has no moves. There are a total of 46 reciprocal zugzwangs in the ending lll+l!.a6 v i.. The following position depends on two of these (readers should also refer back to diagram 1 20, which is crucial for understanding this posi tion). .
=I= 1 27 Tjavlovski, 1963 (end of study) Szachy and again the bishop turns out to be on a bad square. 3) 2 i.g7 (2 i.h8 is also good) �b5 3 � ! �c4 4 �e2! lllb2 5
+I=
129 Original
+I=
128 Original
(129): Black to play draws easily by l .. .�f2 followed by 2 . . .�e3, so we may assume that White is to play. The crucial reciprocal zugzwangs are the positions with �g3 v �fl (Zl ) and �f4 v �f2 (Z2), with the other pieces as in the diagram. We also note that the position with �f4 and l!.a7 v �f2 and i.a8 is reciprocal
98 lLJ+� v J.
zugzwang (Z3) - see diagram 120 for the analysis. White may try: 1) 1 �g3? �fl ! ( l .. .J.a8? 2 �f4 ! �f2 3 a7 ! is Z3) is Zl with White to play. After 2 a7 .i.a8 ! 3 �f4 �f2 ! White finds himself in Z3, while 2 �g4 is met by 2 ... �el ! fol lowed by . . . �d2 and Black's king is heading for b6. 2) 1 ltJd3? J.a8 2 �g3 �fl ! 3 �f4 �e2 ! 4 ltJc5 �d2 ! 5 �e5 �c3 ! 6 �d6 �b4 draws. 3) 1 �gS? �fl ! (not l ...�f2? 2 �f4 with Z2) 2 �f4 (or 2 �f5 �el !, and Black draws after 3 lLxl3+ �d2! 4 ltJf4 J.a8 or 3 �e5 �d2 ! 4 l0b3+ �c3 5 ltJc5 .i.a8) �f2 ! is Z2 with White to play. Black draws by 3 �e5 (3 a7 .i.a8 ! or 3 l0d3+ �e2! 4 ltJc5 �d2 ! ) �e3 ! 4 l0b3 (4 �d6 �d4 !) �d3 ! 5 ltJc5+ �c4 6 lLJe4 �d3 ! and White cannot make progress. 4) 1 �g4! and now: 4a) 1 ...�f2 ( 1 ....i.a8 2 �f4 �f2 3 a7 ! is Z3 with Black to play) 2 �f4 ! (Z2) and now Black has no waiting move. In particular, every move by the bishop weakens Black's posi tion: 4al ) 2...J.f3 3 ltJd3+ �e2 4 lLJc5 (threatening ltJe4) .i.a8 (Black can not play ...�d2 as in line 3 above) 5 �e5 ! �e3 6 �d6 �d4 7 l0b7 ! �c4 8 �c6 ! �b4 9 a7 �c4 10 �c7 and wins. 4a2) 2 J.c6 3 �e5 �e3 4 �d6 (gaining a tempo) J.a8 5 ltJb3 �d3 6 lLJa5 �d4 7 l0b7 wins as in line 4al . 4a3) 2... J.hl 3 �e5 �e3 4 l0b3 ! �d3 5 ltJc5+ ! �c4 6 ltJe4 ! and, thanks to the unfortunate position of
the bishop on h l , Black cannot play ...�d3. 4a4) 2 ... �el 3 �e3 ! and Blact has nothing better than to continue 3 ....i.a8, when 4 ltJd3 gives White aa improved version of line 4b. 4a5) 2...J.a8 3 a7 ! is Z3. 4b) 1 ...�n 2 �g3 ! (Zl) J.a8 (ex 2 ....i.hl 3 �f4 ! �f2 with line 4a3) 3 �f4 ! �el (3 ...�f2 4 a7 ! is Z3) 4 �e3 ! �d 1 5 ltJd3 �c2 6 �d4 ! �b3 7 �c5 ! �a4 (7 ... �c3 8 l0b4 �b3 9 l0c6 �a4 10 �b6 also wins) 8 �b6! �b3 9 l0c5+ �b4 10 l0b7 �c4 1 1 �c6 �b4 ( 1 l . ..�d4 1 2 �c7) 12 �c7 winning. If we move the pawn back to aS. then we are already in relatively un known territory. First of all, here are two sound studies. In both cases White must sacrifice his bishop to reach a draw, but the ideas are merely slightextensions of diagrams 124 and 125 respectively.
•••
=I+
1 30 Gulaev, 1952 Drosha
�+.0, v i. 99
(130): 1 i.c5 ! (White must not waste time; I i.g7? �e6 ! 2 i.d4 h3 3 i.gl llle l + ! 4 �d l lllf3 ! wins for Black after 5 �e2 lll x gl+! 6 �f2 �2 ! or 5 i.h2 lllxh2 6 �e2 �f5 7 �f2 �g4) h3 ( 1 ...llle l+ 2 �dl ! draws, but not 2 �d2? h3 ! 3 i.gl 00+!) 2 i.gl ! llle l+ 3 �dl ! (not 3 �d2? �f3+!; the bishop must be sacrificed on h2, not gl) �f3 4 i.h2 (this is the manoeuvre of diagram 1 24) lllxh2 5 �e2 ! �g4 (or 5 ... �e6 6 �f2 ! lllg4+ 7 �g3 h2 8 �g2) 6 �f3 ! �e6 7 �g3 ! h2 8 �g2! and draws.
(after 3 ...llle4+ 4 �f3 llld2+ 5 �f2 Black is not making progress) 4 'ifi>f3 ! (an attractive move) lllx g 1 + 5 �g4! h3 6 �g3 ! (this reciprocal zug zwang draws, just as in diagram 125) h2 7 �f2 ! lllf3 8 �fl ! and draws. However, the study composers (and judges) don' t always get it right.
• • ••• • • • • �,� . . . • • • • •• • • • ••• • • • m • • • • • • -/+ Kazantsev, 1985 Comm., 1idskriftfor Schack
1 32
=I+
131 Falk, 1991 Springaren
(131): There are two possible moves: 1) 1 i.f2? h3 ! 2 i.g3 (2 �xd4 �g2! wins) �g2! (but not 2...lllf5+? 3 �f2) 3 i.b8 lllf5+ 4 �e2 lllg 3+ 5 J.xg3 �xg3 ! 6 �fl h2 ! wins. 2) 1 �f2! (this looks horrible, but 1n fact it is a draw whoever moves first) lllf5 ( l . . .h3 2 �fl) 2 �fl ! �g3+ (or 2 ...h3 3 i.b6) 3 �f2! �e2 !
(132): 1 �c5 (or 1 i.cl lllf6! and Black wins after 2 �c5 llle4+ 3 �d4 lllf2 or 2 �a5 llle4 3 �b5 lllf2 4 .i.b2 �d3 5 i.g7 a3) a3 2 �d4 lllf6 3 �d3 and now: 1 ) 3 llle4? (the only move ana lysed by the composer!) 4 i.e3 ! (not 4 i.h6? �c5+ 5 �e2 llle6, nor 4 i.f4? lllf2+ 5 �d2 �g4) �c3 5 i.d4 ! llla4 6 �d2 ! lllb2 7 �cl ! drawing. 2) 3 �g4 (3 . . .llld7 also wins, albeit more slowly) 4 .i.c3 (4 �d4 �f2 5 i.c3 transposes) 4 ...�f2+ 5 �d4 llld l 6 i.al �a2 and the bishop is trapped. ...
...
JOO
lb+� v .t
Averbakh considered the case with �b4, lba3 and �a5 against .ta6 and attempted to delineate the zone in which Black's king must stand in order that the position is drawn with White to play. However, Averbakh's zone contained some errors. The fol lowing diagram shows the correct zone.
Averbakh, 1958 (133): Averbakh's erroneous dia gram included the squares d3, e3, f3 and g4. We will consider the follow ing squares for the black king: d3, f4, g4, g5 and h5. These demonstrate the key features which determine the shape of the drawing zone. B�d3: White wins by 1 �c5 �e4 2 �b6 .tc8 (or 2 ....tfl 3 lbb5 �d5 4 a6 ! �e6 5 a7 .tg2 6 �c7 ! with a Horwitz win) 3 lbb5 �e5 (3 ...�d5 4 lbc3+ ! is the same) 4 .!Dc3 ! (this was the move Averbakh overlooked; 4 lba7 .tg4 5 a6 �d6 ! 6 .!Des+ �d7 ! is only a draw) and now: 1 ) 4 .th3 5 a6 ! .tg2 6 �c7 is a Horwitz win. •••
2) 4 �d6 5 lbe4+ ! �e7 (after 5 ...�d5 White wins by 6 lbc5 �d6 7 lbb7+ followed by a6) 6 lbc5 ! .th3 7 a6 ! followed by lbb7 and a7. 3) 4 �d4 5 lba4 ! �c4 6 .!Des �b4 7 lbd3+ �a4 8 lbe5 �b4 9 ten .th3 (or else lbd6) 10 a6! .tg2 1 1 a7 with the usual win. Readers should note that the posi tion after 4 lbc3 ! (W�b6, lbc3, �aS v B�e5, .tc8) is vital to the whole ending of lb+� v .t and we will have cause to refer to it several times in the future. B�4: 1 �c5 �e5 ! (this move draws, unlike . . . �e4) 2 �b6 .tfl (Black has an alternative draw by 2 . . ..tc8 3 lbb5 �e6! 4 lbc3 �d7 S lbe4 �d8 ! 6 lbc5 .th3 7 a6 �c8 8 lbb7 .tg2 9 lbd6+ �d7) 3 lbbS �e6 ! 4 a6 �d7 ! (Black is just in time to keep White's king away from c7) 5 a7 .tg2 drawing. B�g4: 1 �c5 �f5 2 lbb5 ! �eS (or 2...�f6 3 �b6 ! .tc8 4 lbd6 .th3 5 a6) 3 �b6 ! .tc8 4 lbc3 ! transpos ing into the B�d3 analysis. B�gS: 1 �c5 �f6 2 �b6 (2 lbbS �e7 3 �b6 .txb5) .tn 3 lbb5 �e6 transposes to the B�f4 analysis. B�hS: 1 lbb5 �g6 2 �c5 ! �f6 3 �b6 ! .tc8 4 lbd6 .th3 5 a6 and White simply promotes the pawn. Many positions eventually re duce to the B�d3 win given above. In the following position, for exam ple, Black cannot avoid defeat de spite the relative proximity of his king and the fact that it is his turn to move. (134): We may as well assume that Black is to play: •••
•••
.!LJ+� v .i.
+I-
1 34 Original
1 ) 1 .i.c4 2 .!LJf5+ d 3 3 .!LJd6 ! .i.a6 4 'iii>d5 ! 'it>e3 5 'it>c6 �d4 6 'iii>b6 ! .i.fl 7 .!LJb5+! also wins) 3 .!LJd6 .i.a6 4 'it>d5 ! 'iii>f4 5 �c6 ! 'it>e5 6 .!LJb5 .i.c8 7 'it>c7 .i.a6 8 �b6 ! .i.c8 9 .!LJc3 ! transposing to diagram 133. 2) 1 .i.n 2 'it>d5 ! .i.a6 3 'it>c5 .lc8 4 .!LJc2+ c3 6 00+ 'it>b3 7 'it>b6, followed by tDe7, and White eventually reaches the Horwitz win. 3) 1 .i.c8 2 'it>d5 'iii>f4 3 .!LJe6+ �e3 4 .!LJc5 .i.h3 5 a6 .i.fl 6 a7 ! .lg2+ 7 'it>e5 (but not 7 d4 !, beading for a6) .i.a8 8 .!LJd7 'iii>d3 9 �b6 .i.b7 10 'it>d6 and wins. 4) 1 'iii>d3 2 'it>d5 ! 'it>e3 trans poses to line 2. 5) 1 .i.b7 (the most interesting move) and now: 5a) 2 tDb3? .i.g2! (surprisingly, this is the only move to draw; after 2 . ..i.f3? 3 'it>d6 ! .i.e2 4 'it>c5 'it>e4 5 �4 ! White gains a crucial tempo, winning after 5 ... .i.fl 6 tDb5 ! or S .i.a6 6 'it>b6 .i.fl 7 .!LJb5 ! 'it>d5 8 •••
JOI
a6 ! ) 3 'iii>d6 (3 a6 'it>d3 ! ) .i.fl ! 4 'it>c5 'iii>f4 ! 5 .!LJd4 'it>e5 6 .!LJb5 'it>e6 7 a6 'it>d7 and draws. 5b) 2 .!LJe6! .i.a6 (other lines are similar, for example 2 . . ..i.f3 3 'it>d6 .i.e2 4 'it>c6 'it>e4 5 tDc7 ! 'it>d4 6 'it>b6 d2 4 .!LJf4 .i.a8 5 'iii>d4 'iii>c2 6 'it>c4 'it>b2 7 .!LJd5 'it>a3 8 .!LJb6 .i.c6 9 'it>c5 or 2...'iii>d2 3 'it>d6 .i.a6 4 .!LJc7 .i.fl 5 'it>c5 'iii>e 3 6 tDb5) 3 'it>d5 ! .i.b7+ 4 c6 'iii>d4 7 .!LJb5+ ! 'it>c4 8 a6 .i.g2+ 9 'it>b6 ! 'it>d5 1 0 a7 winning. This section would not be com plete without a couple of reciprocal zugzwangs (selected from the 36 available with .!LJ+�a5 v .i.). The fol lowing position depends on two of them.
•••
•••
•••
...
.
...
+I=
1 35 Original
(135): Black to play draws with 1 . . . 'it>d2, so we may assume that White is to move. The positions with 'it>e4 v 'iii>e2 and 'it>d4 v 'iii>d2 are both reciprocal zugzwang. The analysis runs:
102 ©+£!. v .i.
1 ) 1 �e4 �e2! 2 �d4 (or 2 lLJd5 �d2 ! 3 �d4 �c2 ! 4 �c5 �b3 ! 5 �b6 .i.fl 6 lDc7 �b4 and Black is just in time to prevent lDb5) �f3 (Black can also draw with 2 ... �d2, but this is simpler) 3 lLJd5 �g4 4 lLJb4 i.b5 ! 5 �c5 i.fl 6 lLJd5 �g5 7 lDc3 �f6 8 lDb5 �e7 and Black's king arrives in time. 2) 1 �eS! �e2 2 �e4! (the first reciprocal zugzwang) and now: 2a) 2 .i.d3+ 3 �d4 ! .i.a6 4 lLJd5 �f3 transposes to the main line. 2b) 2 i.b7+ 3 lDd5 ! (and not 3 'itd4? �f3 ! 4 lLJd5 �g4 !, drawing after 5 �c5 �f5 ! 6 lDb4 �e6 7 �b6 .i.g2 8 a6 �d7 or 5 �e5 .i.a6 ! 6 lLJc7 i.fl 7 �d6 �f5 8 �c6 �f6 9 �b6 �e7) i.a6 (3 . . .�f2 4 �d4 �f3 5 lDb4) 4 lDc7 .i.b7+ 5 �d4 �f3 6 'itc5 �f4 7 �b6 .i.c8 8 lDd5+ �e5 9 lDc3 transposes to diagram 1 33. 2c) 2 �el 3 �d5 ! (not 3 �d4? �d2 ! 4 lLJd5 �c2 ! as in line 1) �d2 4 ..tid4 ! (the second reciprocal zug zwang) �e2 (bishop moves are no better, e.g. 4 ....i.e2 5 lLJc4+!, 4 ....i.b5 5 lDc4+ �dl 6 �c5 .i.a6 7 lLJd6, or 4 ... .i.b7 5 lLJc4+ �d i 6 �c5 .i.a6 7 lLJd6) 5 lLJd5 ! �f3 6 lDc7 ! .i.c8 7 lLJb5 'itf4 8 �d5 ! .i.a6 9 �c6 �e5 1 0 'itb6 ! .i.c8 1 1 lLJc3 ! and again we reach diagram 133. •••
...
...
With the pawn on a4, Averbakh found the winning zone analogous to diagram 133. (136): The three marked squares are the only positions for the black king which enable White to win if he has the move, i.e. Black's king has to be in a truly awful position if White
• • • • • • • • • • • • - . . . ts . • • • •• • • lb. • • • • • •*B* 1 36
Averbakh, 1958 is to win. We will consider three squares for the black king: el, f1 and h2. B'itel: Black draws after 1 �c4 �d2 2 �b5 .i.d8 ! (not 2 ... i.c7? 3 lDb4 ! when 3 ...�c3 is impossible, so White wins by 3 ... .i.d8 4 lDd5 �c2 5 lDb6 !) 3 lDb4 'itc3 ! 4 lLJd5+ 'itb3 ! and White cannot play lDb6. B� : White wins by 1 �c4 �e2 2 �b5 .i.d8 3 lDb4 �d2 (after 3 ...�e3 4 lLJd5+! �d4 5 lDb6 ! �e5 6 a5 �d6 7 a6 ! White wins by one tempo) 4 lDd5 'itc2 (again Black is one tempo too slow) 5 lDb6! �b3 6 a5 ! .i.c7 7 a6 .i.b8 8 �c6. B�h2: 1 �c4 �g3 ! 2 �b5 .i.d8 ! 3 lDb4 �f4 ! 4 lDd5+ �e5 ! (it takes Black four moves to reach d6 from h2, so he is a tempo up over the B�fl analysis) 5 lDb6 �d6! 6 a5 �c7 7 a6 �b8! drawing. In the following practical exam ple, the defender's king was on just such a bad square. (137) : White to play loses since his king is simply too far away: 1
lb+� v i. 103
-/+ Vescovi-Playa World Junior Ch 1992
1 37
�b7 tiJe6 (not 1 . ..�f5? 2 �c6! tiJe6 3 i.e7 ! with a draw after 3 ... �g4 4 �d5 ! lbg5 5 �d4 ! h4 6 �e3 ! or 3 ... lbg5 4 �d5 h4 5 �d4 ! h3 6 1.d6 !) 2 .te7 (2 i.h4 �f4 3 �c6 �g4 ! 4 i.f2 lbf4 ! 5 �d6 lbe2 6 �e5 �g3 ! 7 �d4 �f3 8 i.e 1 h4 is also lost) �f4 ! 3 �c6 lbg5 ! 4 �d5 h4 ! 5 �d4 h3 0- 1 in view of 6 i.d6+ �f3 ! 7 �d3 lLJe4 8 i.h2 lbg3 followed by ... �g2. White would also have lost if bis king had started on a7 or b8, but on any other square the position would have been drawn. The following example was less accurately conducted by the players concerned. (138) : The proximity of White's king guarantees a draw although, thanks to the poor position of his bishop, accurate defence is neces sary: 1 . . .lbg4+ 2 �d4 ! (the only move; 2 �d5? �h4 ! ensnares the bishop, while 2 �d6? h4 3 �d5 �f4 ! 4 �d4 �g3 5 i.fl lbf6 is the win of diagram 1 33) �f4 3 i.g2
I=
1 38 Brito-Kurajica Las Palmas 1994
(White makes life a little difficult for himself; the simplest draw was by 3 i.fl, and then 3 ...lbh2 4 i.a6 h4 5 �d3 h3 6 �e2 �g3 7 .ib7 or 3 . . . h4 4 �d3 �g3 5 �e2) h4 4 �d3 �g3 5 i.b?? (this move should have cost half a point; White could still have defended by 5 i.fl !, intending 5 . . . lbh2 6 �e2 ! or 5 . . .lbf6 6 �e2 lbh5 7 �el ! lbf4 8 i.a6 h3 9 i.b7 lbg2+ 10 �fl ! tiJe3+ 1 1 �e2) �f2 ! 6 �e4 (6 i.c8 h3! wins) h3 ! 7 �f4 lbe3 ! 8 .thl h2 ! (Black's accurate play has given him a winning posi tion, but now he is afflicted by that traditional weakness - an inability to find the Horwitz win) 9 �e4 lbg4 (9 . . . lbg2 was more direct, heading for h4) 1 0 �d3 �gl ? (now the win disappears entirely; it was time for reverse gear with 10 ...lbe3) 1 1 �e2 ! lbf2 1 2 i.b7 lbh3 1 3 i.hl lbf4+ 14 �el �xhl 15 �f2! lbg6 1h- 1h. There are 1 3 reciprocal zug zwangs with lb+�a4 v i.. This is one of the most interesting:
104 �+8 v �
• • • • • • • • -*- • • • • • • • • 8. • • • . . ·�· . . � . • • ••• =I-
1 39
Original (139): Suppose that Black moves first. The curious feature of this posi tion is that White displays an amaz ing reluctance to play a5. At first sight it cannot do any harm to ad vance the a-pawn, but it turns out that White's plan involves dominat ing Black's bishop on the a6-fl di agonal, and for this it is essential that the b5-square is covered by the pawn. The analysis runs: 1) 1 �c4 (this allows White to gain a tempo by �d4) 2 �4 ! �el (2 . . . �gl 3 a5 transposes) 3 �e3 ! �dl 4 �d4 ! �a6 5 �c5 �e2 6 a5 �e3 7 �b6 �fl S �d6 �d4 9 �b5+ ! �d5 10 a6 ! �e6 1 1 a7 and Black is one tempo too slow to pre vent White's king arriving at c7. 2) 1 �e2+ 2 �e3 ! �a6 3 �d3 �g2 4 �c5 �c4 5 �d4 �fl 6 �3 �f3 7 a5 ! �g4 S a6 �f5 9 a7 ! �g2 10 �c5 ! �as 1 1 �b6 ! �e4 12 �5+ �d5 1 3 �b7 followed by the Hor witz win. 3) 1 �el 2 �e3 ! (now Black cannot play ...�g2) �b7 (2 . . . �fl is ...
...
...
line 2) 3 a5 �a6 4 �d3+! �dl 5 �b4 ! �b5 6 �d4 ! �cl 7 �c5 �fl S �d5 �b2 9 �3 �a6 10 �b4! �cl 1 1 �4 �c2 12 �6 �e2 (12 ...�d3 13 �c5) 13 �b5 �d3 14 a6 �e3 15 a7 �f3 16 �c5 and wins. 4) 1 �gl 2 �e4 ! �h2 (2 ...�fl 3 a5 �h2 4 �e3 ! �h3 5 �f2+ ! �h4 6 �d3 ! �g5 7 a6 ! �f5 S �d4 ! transposes to the main line) 3 �c5 ! �fl 4 �f2 ! (White must first force the bishop to c4, so that a later �d4 gains a tempo; the immediate 4 �e4? is too slow after 4 ... �g3 5 �d3 �g2+ ! 6 �e5 �c6 7 a5 �b5 ! S �d4 �g4) �c4 (B lack's bishop is running out of squares on the fl -a6 diagonal) 5 �e3 ! �g3 6 �d4! and now Black has two possible bishop moves, but they both run into tactical problems: 4a) 6 �e2 7 �d3 ! �g4 S a5 ! �f5 9 a6 ! �f3 (the bad position of Black's bishop prevents ...�e6) 10 �c5 ! �e4 ( 1 0... �e6 1 1 �b6 ! �e7 1 2 �5 �g2 1 3 �c7 ! �aS 14 a7) 1 1 �f2! �g2 1 2 �d6 ! �fl 1 3 a7 ! �g2 14 �3 leading to the Horwitz win. 4b) 6 �n 1 �d3 ! �g4 s a5 ! �f5 9 a6 ! (now the problem is that Black cannot play both . . . �g2 and ... �e6) �f6 10 a7 ! �g2 (accurate play is still required) 1 1 �f4 ! �as 1 2 �d5+ ! �f7 1 3 �c5 ! �b7 14 �b4 �e6 15 �b6 ! �aS 1 6 �c7 ! and the Horwitz win arises yet again. The situation is much simpler when White is to play. After 1 �e3, Black continues l ...�g2 2 � �h3 3 �5 �fl ! 4 �3 �g4 ! 5 a5 �f5 6 a6 �g2! 7 �d4 �e6, making it back in time, while 1 �e4 gives Black's ...
...
...
.!LJ+� v i. 105
king access to d 1 , so allows a draw by 1 . . . �el ! 2 �e3 �d l ! 3 �d4 (3 i&:5 i.fl 4 lLJ
(140): Black to play draws easily by l . ..�d2, so assume that White moves first. His first task is to drive the black bishop off the d 1-a4 diago nal: 1 llX:3 ! (not 1 �e3? i.a4 ! 2 llX:3 i.e8, and Black draws comfortably once he has improved the position of his bishop) i.c2 ( l ...i.g4 loses more quickly after 2 a4 �f2 3 a5 i.c8 4 .!LJe4+ �f3 5 lLJc5 �f4 6 �d5 !) 2 �e3 ! (this reciprocal zugzwang is the only way to drive off Black's bishop) �fl (or 2 ...i.f5 3 a4! i.c8 4 �d4 �d2 5 .!LJe4+ �c2 6 �c4 !, and White wins after 6 . . .i.a6+ 7 �b4 ! i.fl 8 a5 ! i.e2 9 lLJ
+I=
140 Original
106 lLJ+.0.
v
.i
• • • • • • • • . -�· . • • • • 8. • • • • • • • . . .... � • • • •• +I=
141
Original 2) 1 .ih3 (holding White up for several moves) 2 �f5 ! .ifl (after 2 ... �g2 3 �f4 ! �gl 4 a5 .ifl 5 �e5 .ie2 6 l0e3 �f2 7 l0d5 �f3 8 �d6 �g4 9 �c6 �f5 10 l0c3 ! .ia6 1 1 l0b5 ! �e5 1 2 �b6! .ic8 1 3 l0c3 ! we have reached diagram 133) 3 �f4 ! (White is aiming to confine the enemy king; Black cannot prevent this as . . . �g2 is impossible) .ie2 (the line 3 ....ia6 4 �g3 ! �gl 5 l0e3 ! .ic8 6 l0d5 ! .ia6 7 a5 .ib7 8 l0b4 �fl 9 �f4 ! �e2 10 �e5 ! �e3 1 1 �d6 ! �d4 1 2 l0c6+! �e4 1 3 l0a7 .ia6 14 �c6 ! �e5 1 5 l0b5 ! .ic8 16 �c7 .ia6 17 �b6! .ic8 18 l0c3 ! is another transposition into dia gram 1 33) 4 �g3 ! �gl 5 l0e3 ! .ia6 (5 ....id3 6 a5 .ia6 7 tLJd5 transposes to the previous note) 6 l0d5 ! .ib7 (after 6 . . ..ic4 7 l0c3 �fl 8 �f3 ! �e 1 9 �e3 ! .ifl 10 a5 ! .ia6 1 1 tLJd5 �dl 12 l0b4 ! we have transposed to the main line) 7 l0c3 ! (not 7 l0b4? �fl ! 8 �f4 �e2 ! 9 �e5 �d2; it is essential to cover e2 in order to prevent Black's king escaping) �fl ...
8 �f4 ! �f2 9 a5 ! �el 10 �e3 ! .ia6 l l l0d5 �d l 12 l0b4 ! (now we are in line 3 of diagram 139; for the sake of completeness, we give the rest of the main line without commentary) .ib5 13 �d4! �c l 14 �c5 .ifl 15 lLJd5 �b2 1 6 l0e3 .ia6 17 �b4! �c 1 18 l0c4 �c2 19 l0d6 .ie2 20 l0b5 �d3 21 a6 �e3 22 a7 .if3 23 �c5 �f4 24 �d6 .ih 1 25 �c7 �e5 26 l0d6 .ia8 27 l0c4+ �e6 28 �b8 .ig2 29 l0a5 �d7 30 l0b7 ! with either promotion or capture of the bishop next move.
2.2:
ltJ+b� v i.
Of the four sections in chapter 2, the b-pawn has received the least atten tion from composers and analysts, although this is probably a result of chance more than anything else, be cause some of the positions are very interesting. We already know from Averbakh' s diagrams 1 1 1 and 1 1 2 that this ending is only interesting when the pawn is on at least the fifth rank (or can move there immedi ately). There are winning positions with the pawn further back, but these depend on the defender's bishop being badly placed. With the pawn on b7, the basic principles are not at all difficult. White only has winning chances if his king is supporting the pawn. We will start off by supposing that Black's bishop is on a7 . This is clearly not a very good square for the bishop, because White can finish the game by playing l0c6. However,
&o+l!. v .t 107
even if Black's bishop starts on the h2-b8 diagonal, it very often ends up on a7 later, so we need to know when White wins. Averbakh proved that White wins in almost all cases:
142 Averbakh, 1958 (142): With White to play, Black only draws if his king starts on one of the two marked squares. We will consider the following squares for Black's king: b3, e6 and a4. B�b3: 1 &oc7 �b4 ( l . . .�c4 2 �b5 .tb8 3 �b6 �d5 4 &oc7+! �c4 5 &oa6 .tg3 6 !Oc7) 2 &ob5 .tb8 3 �b6 (the key move; White keeps Black's king out while he plays the knight round to a6) �c4 4 &oc7 fol lowed by 5 !Oa6 expelling the bishop and then 6 &oc7 shutting it off. This winning method also works when Black's king starts on c4, d4 or e5. B�e6: Now the simplest win is by 1 �c7 �d5 2 &ob4+ �e6 3 &oc6 i.gl 4 �c8 .th2 5 &ob4 ! , followed by !Oa6 and !Oc7 . When Black's king starts on d8, White plays 1 �d6 ! �e8 2 �c7 with the same win.
B�a4: The draw is obvious after 1 �c7 (or 1 &oc7 �a5 ! 2 &ob5 .tb8 ! and White cannot play �b6) �b5 ! 2 &ob8 .tgl 3 &oc6 .th2+. Suppose now that White's king is on c8, his pawn on b7, and Black's bishop is on h2. Then White wins if he can block the diagonal with &oc7, while Black draws if he can play his king to c6 and cover the c7-square. This situation is therefore just a mat ter of counting. White is better off if his king is on d7. If Black's king starts on f7, then the route round to b6 is very long (via f5 and e4, but even then White can play �c6 causing more trouble for Black), whereas White only needs to play his knight to a6, preventing . . ..tb8, and then &oc7 forces the pawn home. Indeed, with the black king unfavourably placed on f7 White will almost always win. The closer Black's king is to the queenside, the better are his drawing char.ces. Here is a marginal case, which also happens to be a recipro cal zugzwang: (143): White's king is keeping his counterpart at a distance, but Black has two possible drawing ideas. The first is to reach a6 with his king, for example if Black's king were on c4 in the diagram, then Black to play could draw by l . . . .ta7 followed by . . .�b4-a5, but not 1 . ..�b4? 2 �b6 ! and White wins. The second idea is the obvious plan of transferring the king to e7 and ultimately d8. White cannot win if there is no possibility of cutting off Black's bishop by !Oa6-c7. The tension between these
108 lLJ+�
v
.i.
• • • • • LS . • • •<&t>• • • • • • • . - . . • • • • • • •ttJ• • • • • =I-
143
Original two ideas, which tug at Black's king in different directions, is the key to the reciprocal zugzwang. Firstly suppose that Black moves first: B 1) 1 .i.g3 (now Black cannot meet ll:ih4 by ... �e5) 2 ll:ih4 ! (but not 2 ll:ie l ? �c3 ! when Black gains time and draws by 3 ll:if3 �b4 ! 4 �b6 .i.t2+ 5 �a6 .i.g3 !) and now: B la) 2 �c4 3 lLJf5 ! .i.b8 4 lLie3+ �b4 (or 4 ...�d4 5 ll:id5 �c4 6 ll:ic7 .i.a7 7 ll:ib5 .i.b8 8 �b6 �d5 9 ll:ic7+! �e5 10 ll:ia6 .i.d6 1 1 l:i:Y::.7) 5 �b6! and again White wins by play ing his knight to a6. B lb) 2 .i.b8 3 ll:if5+ �e5 (or 3 ...�c4 4 ll:ie3+) 4 lLie3 �e6 5 lLid5 ! .i.a7 (5 ...�e5 6 l:i:Y::.7 �d4 7 �b6 and 8 ll:ia6) 6 ll:ib4 .i.b8 7 ll:ia6 .i.a7 (Black's bishop has been driven to c7 so White wins as in the previous diagram) 8 �c7 �d5 9 ll:ib4+ �e6 10 ll:ic6 .i.gl 1 1 �c8 .i.h2 12 ll:ib4 ! winning. B2) 1 �e4 2 lLiel ! (heading for a6) �d4 3 ll:ic2+ �c3 (Black must ...
•••
•••
•••
return, as 3 ...�e5 4 ll:ib4 �e6 5 ll:ia6 is hopeless) 4 ll:ie3 ! �d4 (4 ... �b4 5 �b6!) 5 lLid5 winning as usual. B3) 1. �d3 (losing vital time) 2 ll:ih4 ! �d4 (2 . . .�e4 3 ll:ig6! �f5 4 ll:ie7+ �e6 5 l0d5 ! ) 3 lLif5+ �e5 4 l0e7 �e6 5 l0d5 ! winning as in line B l b. B4) 1 �c4 2 lLie3+! �b4 3 �b6! �a4 4 ll:id5 �b3 5 ll:ic7 and 6 ll:ia6 wins . B5) 1 �eS 2 lLie3 ! �e6 3 ll:id5 ! with line B 1 b again. With White to play the position is drawn, although very accurate de fence is necessary: 1 ll:ie l ( 1 �b6 �d5 is useless, while after 1 ll:ih4 �e5 ! 2 ll:if3+ �e6 ! 3 ll:id4+ �e7 ! 4 ll:ib5 �d8 Black reaches one of his target draws) �c3 ! (l...�c4? 2 l:i:Y::.2 ! wins as in line B2 above) 2 �b6 (2 ll:if3 �b4 ! 3 �b6 �c4 ! 4 lLid2+ �d5 ! 5 ll:ib3 �d6 6 ll:ic5 .i.c7+ ! 7
...
•.•
•••
•.•
.
ttJ+l!. v .t 109
12 tiJb5 i.b8 1 3 �b6 �d5 14 tiJc7+! �c4 1 5 llla6 wins. W4) 2 �d4? 3 tiJc2+ ! �c4 4 ;;,c6 wins as in line B2. W5) 2 i.g3! (here the bishop is relatively immune to knight attacks) 3 tiJf3 (3 tiJg2 �d4 !) �c4 ! 4 tiJd2+ (4 lllg 5 �d5 ! and 4 �c6 �b4 ! are also drawn) �d5 ! 5 lllb3 �e6 6 �c5+ �e7 ! 7 llla6 �d7 and Black has saved himself. The following position is very at tractive, but it has one curious fea ture. The composer intended it to be a 'White to play and draw' position, but by reversing the colours it would make a nice 'White to play and win' study, the solution being the line Amirjan missed. ...
...
-/
144 Amirian, 1974 Shakhmaty Moskva
(144): Black's king is not very well placed, and he must play with great accuracy in order to win: 1 i.b6 �g3 ! (threat 2 . . . tiJf2, so White must transfer his bishop to h2) 2 i.c7+ �f3 ! 3 i.h2 (White is on the
verge of drawing by �h4, so Black must prepare to counter this move) �f2 ! 4 �f5 (now 4 �g4 and 4 �h4 both lose to 4 . . .tiJe5+ ! 5 �h3 tiJf3 ! 6 i.c7 t£ig5+ ! 7 �h4 �fl 9 i.b6 t£ie4 ! , followed by . . .tiJf2) tiJel ! 5 �e4 (5 i.c7 loses to 5 ... �fl 6 i.b6 llld 3!) .!llf3 ! 6 i.f4 (or 6 i.b8 lllg 5+! 7 �f4 ttJe6+ 9 �e5 �f3 9 i.a7 lllf4 10 i.gl �g3 1 1 �e4 lllh3 and Black wins) �e2 ! 7 i.e3 tiJh2 (the com poser only took into account the line 7 ... tiJg5+? 8 �f4 ! tiJh3+ 9 �g3 ! and White draws; Black can also win by 7 . . .llle 5) 9 �f4 lllg4 ! 9 i.gl �fl ! and the pawn promotes. Of the pre-database efforts, only Halberstadt's analysis is worth quot ing:
B B B<&t>• B B B D • • ••• • • • • • • • • . � . . • • • • • • • • 145
+I= Halberstadt, 1933 'Regards '
(145): Although the win is quite short, White must avoid a devilish stalemate trap which would be quite easy to overlook during a game: 1 t£ie4 ! (and not 1 tiJd5? i.g5 ! 2 �h8 �h6! 3 g8'ii' i.f6+ ! 4 tiJxf6 with
110 tLJ+� v .t
stalemate; of course, promoting to a rook makes no difference) and now we have a reciprocal zugzwang. B lack to move loses after l . . ..te7 ( 1 . . ..id8 is the same) 2 �h8 ! �h6 3 lLif2 ! (White cannot promote im mediately, but he can improve the position of his knight without allow ing . . . i.f6) i.h4 4 lLig4+ and White can promote in safety next move. Amazingly, an inferior version of this study (by Galushka) was given a commendation in the 1984 Chervoni Girnik tourney, but I will spare read ers the details.
146
l ) 2 .id6 (here White gains a tempo by attacking the bishop) 3 �f5 ! �g7 4 �e6 ! .ih2 5 �d7 �f6 6 lDe6 .ib8 7 �c8 .ia7 8 lLid8 and 9 lLic6. 2) 2 i.h2 (and here Black is still not threatening . . .�f8 because of lLie6+ followed by lLif4) 3 �f5 i.g3 4 �e6 �g7 5 �d7 wins the same way. 3) 2 i.b8 (in this line B lack falls victim to a reciprocal zug zwang) 3 �f5 ! �f8 4 lLic6 ! (but not 4 �e6? �e8 ! 5 lLic6 i.c7 ! and it is White who is in zugzwang) i.c7 (4 ...i.d6 5 �e6 ! i.c7 6 �d7 wins as in line l ) 5 �f6 ! (the first reciprocal zugzwang) �e8 6 �e6 ! (and the sec ond; Black cannot prevent White's king penetrating) �f8 7 �d7 and again White wins as in line l . The following position makes use of some of the earlier analysis. ...
...
...
+I= Halberstadt, 1930 Ceskoslovensky Sach
(146): Black to move draws easily by I ... �f7, so assume that White moves first. The only knight move is 1 lLlc6! ( 1 lLia6? i.a7 ! 2 lLib4 i.b8 ! 3 lLic6 .td6! 4 lLid8 �f8 ! is a draw), and White wins after l....ic7 2 lLid8 ! (White gains a vital tempo; Black cannot play . . . �f8 or . . . �g7, so he has nothing better than to move the bishop) and now:
+I=
147 Original
(147): Black's main hope is to reach a6 with his king, but by accu rate play White can prevent this: l
lLJ+,0, v .i. 1 1 1
�e2 ! (not 1 �f2? �a2 ! 2 ..t>f3 .i.b8 3 �e4 �b3 ! 4 �d5 ..t>b4 5 tal6 ..t>a5 6 �c6 ..t>a6) �a2 (after 1 . . ..i.b8 2 �d3 �a2 3 �c3 White imprisons Black's king, and gains time to advance his own king to b6, for example 3 ....i.h2 4 �b4 .i.g3 5 �c5 .i.b8 6 �b6 �b3 7 lDe3) 2 �d3 ! �b3 3 �d4 ! (threat lDe5) .i.c7 (3 . . . .i.b8 4 ..t>c5 ! �a4 5 �b6 ! is the same) 4 �c5 ! ..t>a4 5 �c6 ! (gaining access to b6) .i.b8 6 �b6 ! .i.f4 7 iDb2+ �b3 8 lDd3 .i.h2 9 tDc5+ �c4 10 tDa6 .i.gl + 1 1 ..t>c6 i.a7 12 tDc7 ! �b4 1 3 lDb5 winning as in diagram 142. There are 54 reciprocal zug zwangs with tD+.0ib7 v .i.. The fol lowing position is based on one of these.
+I=
148 Original
(148): It is certainly not obvious that 1 lDd4 ! results in a position of reciprocal zugzwang. Black may re ply: 1) 1 �g3 (if Black's king strays more than three squares away from d7, White wins with iDb5 followed ...
by 'ita7) 2 lDb5 �f4 3 �a7 ! ..t>f5 4 �b6 ! .i.b8 5 tDc7 followed by tDa6, winning. 2) 1 .i.h2 2 �a7 ! (the defect of a bishop move is that Black can no longer take on d4 with check) .i.g3 3 lDb5 �f5 4 �b6! winning as in vari ation 1 . 3 ) 1 �f4 (the most resilient de fence) 2 lDc6! (now 2 iDb5? only draws after 2 . . . ..t>f5 ! 3 �a7 �e6 ! 4 ..t>b6 �d7) .i.d6 (Black cannot avoid playing his bishop to a square where it will exposed to attack by a later lDb5; 2 ... .i.c7 loses after 3 ..t>a7 ..t>f5 4 lDd4+ ! ..t>f6 5 tDb5 ! .i.h2 6 �b6 ! ' intending lDc7) 3 �a7 ! ..t>e4 (3 ...�f5 4 lDd4+ ! heads for b5 while at the same time preventing . . . ..t>e6) 4 �b6 �d5 5 tDb4+! (Black's bishop blocks d6) �e6 6 �c6! .i.b8 7 tDa6 .i.a7 8 �c7 winning as in diagram 142. It is worth explaining why the po sition after 1 lDd4 ! is a draw with White to play. White would have to move his knight, but 1 iDb5 allows 1 . . . ..t>f5 ! while 1 tDc6 .i.h2 (but not 1 . . ..i.f4? 2 tDe7 ! �g5 3 �a7 and Black cannot play ... �f6) 2 tDe7 (or else . . . �f5) �g5 ! 3 �a7 �f6 ! 4 lDd5+ �e6 ! allows Black's king to approach. ...
...
Now suppose that White's pawn starts on b6. If Black has a dark squared bishop, then White will soon play b7, leading to a situation we have already analysed, so we will concentrate on the case in which Black has a light-squared bishop. The most important analysis of this
112 �+� v �
ending was undertaken by Averbakh, who produced four completely accu rate diagrams showing the drawing zones for Black's king.
number two) 3 �d7 �a6 ! (number three) and White cannot make pro gress. The same drawing line works if Black's king starts on c5 or b4. Bc,tbs: After 1 �c6 ! ( l ...�c8 2 �a7+) Black is in the reciprocal zugzwang; White wins with L..c,tc5 2 �b8 ! �c8 3 �7+ �xd7 4 b7 ! .
149 Averbakh, 1958 (149): This is one of the most fa vourable arrangements of pieces for White. There are threats of �b3-c5, �c6-b4 or �c6-b8 . With White to play, Black can only draw if his king starts on one of the four marked squares. We will consider the follow ing squares for Black's king: a3, d5, a4, b5. Bc,taJ: White wins easily after 1 �c6 ! c,ta4 2 �b8 �fl 3 b7. Bc,tdS: Once again White is suc cessful with 1 �c6 ! c,tc5 2 �b8 ! �fl 3 b7. Bc,ta4: After 1 �c6 c,tb5 ! the po sition is reciprocal zugzwang. Black to play would lose (see Bc,tb5 be low). However, this position is gov erned by parity; White cannot move his king, and so there is no way to lose a move. Black draws after 2 �b8 �c8 ! (reciprocal zugzwang
1 50 Averbakh, 1958 (150): White's king is less favour ably placed on a7, and the (White to play) drawing zone is correspond ingly larger. White threatens �c6-e7 or �c4-d6, as well as c,tb8-c7, trans posing to the previous diagram. We will consider the following black king squares: f7, e7, e5, d5, b4. B
lt:J+.0, v i. 113
B�b4: White wins by 1 lt:Jc6+ ! �b5 2 lt:Je7 i.a6 3 lt:Jf5 �a5 4 lt:Jd6 and the bishop is dominated. Black has much better drawing chances if his bishop occupies the long diagonal, rather than the re stricted diagonal a6-c8. In the fol lowing position, the (White to play) drawing zone for the black king has expanded to 12 squares.
152 Averbakh, 1958
15 1 Averbakh, 1958 (151): White's only threat is to play lt:Jb7 followed by �b8. Black draws if his king is within two squares of attacking the b6-pawn, otherwise he loses. The square e3 is not in the zone because c5 is inac cessible after 1 lt:Jb7 �d4 2 �b8. Returning to the situation with the bishop on a6, if White's king is on c5 then his winning chances are worse than in diagram 150 because he has to play the king to c7 in order to cre ate a serious threat: (152): We will consider the fol lowing squares for Black's king: e8, e6, e3, c2, e2.
Be8: White wins after 1 'iti>c6 'iti>d8 2 lt:Jb7+! �e7 (2 ...c8 3 lDc5 !) 3 �c7 i.fl 4 lt:Ja5 i.a6 5 lt:Jb3, fol lowed by lt:Jc5 . The same method works with the black king on e7. B�e6: 1 �c6 �e5 2 Wc7 �d5 3 lt:Jc6 ! c5 4 lt:Jb8 ! i.c8 5 lt:Jd7 + is the winning line. Be3: When Black's king is ad jacent to d4 he draws by 1 c6 'ifi>d4 ! 2 lt:Jb3+ (2 �c7 �c5 ! comes to the same thing) e5 ! 3 'iti>c7 'iti>d5 ! 4 lt:Ja5 �c5 ! 5 lt:Jc6 �b5 with the re ciprocal zugzwang of diagram 149. B�c2: If Black's king is within three squares of b5, then he can draw by a different method: 1 c6 �c3 ! 2 c7 �b4 ! 3 lt:Jc6+ �b5 !, with the reciprocal zugzwang of diagram 149 again. B�e2: In this case Black loses, since after 1 �c6 �d3 2 �c7 ! the square c4 is out of bounds, and so Black's king does not arrive on b5 in time. Positions with the pawn on b6 were thoroughly explored by Kosek
114 t0+.0. v .t
in the 1920s and 1 930s. Here are three of his elegant compositions.
154
+I=
Kosek, 1932 Ceskoslovensky Sach Kosek, 1923 La Strategie (153): Black's king is certainly close to the pawn; too close, in fact, since his bishop is seriously im peded. White wins by 1 tLlf5 ! (White must prevent ... .ic8, but the alterna tive 1 tLlg6? .ia8 ! only draws after 2 tLle5+ 'it>b5 ! and Black's bishop set tles on the long diagonal) .ia8 2 tLld4+ ! 'it>c5 3 tLle6+ ! �c6 (3 ... �b5 4 tLlc7+) 4 tLlc7 ! .ib7 (White has im proved his knight position with gain of tempo, and now administers the killing blow) 5 tLld5 ! .ic8 6 tLle7+ winning. (154): White can try: 1) 1 tLlc5? .ih3 and Black trans fers his bishop to the long diagonal with an easy draw. 2) 1 tLlb4? �d7 ! is reciprocal zugzwang. A king move is met by 2 ... 'iti>d6, while 2 tLld3 gives Black's king access to c6, with a draw after 2 ... 'it>c6 3 'it>a7 �b5 4 tLle5 'iti>c5 !.
3) 1 tLle5! .ia6 ( l . . ..ih3 loses to 2 tLlc6+ ! �d7 3 b7) 2 'it>a7 ! (in order to win, White must transfer the move to Black by triangulating) .ic8 3 'it>a8 ! .ia6 (3 ... �e8 4 �b8 ! 'it>d8 5 tLlf7+ �d7 6 lLld6 .ia6 7 �a7 ! wins) 4 'it>b8 .ic8 (Black has no good move; 4 ... �e7 5 �c7 'it>e6 6 tLlc6 is also lost) 5 lLlf7+ �d7 6 tLld6 .ia6 7 'it>a7 ! and the bishop is driven away. (155): White to play must prevent Black transferring his king to c5, with an easy draw (for this reason Black to move draws by l . ..�b5): 1 tLld4 ! (this is reciprocal zugzwang) and now: 1) 1....ic8 2 tLlc6+ ! 'iti>a6 3 tLld8 ! �b5 4 tLlb7 .ih3 5 tLld6+ ! wins for White. 2) 1...�b4 2 �b8 ! �a4 3 ltic6! �b5 (or 3 ... .ib5 4 �c7 .ia6 5 tLlb8 .ifl 6 b7) 4 �c7 with the reciprocal zugzwang of diagram 149. 3) 1...�a4 2 �b8 ! �b4 (2 ....ifl 3 tLlc6 ! .ia6 4 �a8 ! �b5 5 'it>a7! .ic8 6 tLle7) 3 �a8 ! (White must
lLJ+.0. v J. 115
• • • • • = • • .t.B • • • 0 "� • • • • • • • • • • • • .lt:J. • • • • • 155
+I=
1 56
=I+
Kosek, 1934 Ceskoslovensky Sach
Sindler, 1955 Ceskoslovensky Sach
take extreme care; 3 �a7? 'iii>a5 ! leaves White to play in a reciprocal zugzwang and Black draws by 4 �6+ �b5 ! 5 0ie7 'iii>a5 ! 6 0if5 �b5) �a4 (other moves are no bet ter; 3 . . . 'iii>c5 1 oses direct!y to 4
pawn promotion) b3 2 J.a3 ! �d3 3 J.b2 ! (not 3 �f6? �c2 4 'iii>e5 0ic3) �c2 4 J.al ! (Black's knight is a long way from b2 and White's king ar rives just in time) 0ic3 5 'iii>f6 ! 0ia4 6 'iii>e5 ! 0ib2 7 d2 allows 0ie4+ followed by 0id6, with an
116 lLJ+L>, v .t
2 ...f3 3 'iti>c6 both lose) 3 'iti>c4 ! (the second reciprocal zugzwang) f4 (3 ...d2 4 lLie4+ ! e3 5 lLid6 .ta6+ 6 'iti>b4 'iti>d4 7 a5 and 3 ...f3 4 �d5 win for White) 4 'iti>d4 ! (the third reciprocal zugzwang) f3 (4 ... 'iti>f5 5 'iti>d5 ! f6 6 d6 wins) 5 'iti>d5 ! e3 6 e5 ! (the fourth and fi nal reciprocal zugzwang) 'iti>d2 7 'iti>d6! c3 8 c7 ! and wins. The following position makes use of some of the above analysis. +I=
157 Original
instant win for White. White's main plan is to reach c7 with his king, and Black's defence is to be ready to meet 'iti>d6 by a move attacking the knight on c5. White may try: 1 ) 1 'iti>bS? e2 ! puts White on the wrong end of a reciprocal zug zwang. Black draws after 2 c6 .th3 or 2 'iti>c4 e3 ! , when 3 'iti>d5 is again met by 3 . . . .th3 . White can only occupy the long diagonal when ... .th3 doesn't work, for example when Black's king is on f3 or f5 . 2) 1 b4! (the threat is 2 c4 'iti>e2 3 'iti>d5, when 3 . . . .th3 doesn't work because White can interpose on e4) e2 ( 1 . . .f2 2 lLie4+ ! e3 3 lLid6 ! wins, as does I . . .g2 2 b5 when Black has no good move as 2 . . . f2/g3 fails to 3 lLie4+, while 2 . . . f3 allows 3 c6) 2 b5 ! (the first reciprocal zugzwang; the super subtle 2 'iti>c3? fails to 2 . . . d l ! and now both 3 d3 c 1 ! and 3 b2 'iti>e2 4 c2 e3 5 c3 e2 ! lead to a draw) 'iti>e3 (2 . . . d2 3 lLie4+ and
-!+
158
Cheron, 1952 Journal de Geneve (158): Cberon intended this to be a 'White to play and draw' study, but actually Black can win whoever is to move first. Both Cberon's intended line and the missed win are very in teresting. After 1 .td7 g3 ! 2 .th3 we have: 1) 2 lLih4? 3 c5 ! (3 c4? tips the balance back in Black's favour after 3 ...e4 ! 4 c5 f3 5 d4 f2 ! 6 e4 lLif3 ! 7 f4 lLig 1 ! 8 .tn lLie2+) e4 4 d6! (intending to approach the pawn from behind) •••
.!iJ+.0. v � 1 1 7
�f3 5 �e5 ! tLlg6+ 6 �d4! (not 6 �5? .!iJf4! 7 �fl tLlg2 8 �e5 �f2 9 �e4 .!iJh4 10 c;j;>f4 tLlgl ! 1 1 �h3 �gl ! and Black wins) 'jj;>f2 7 �e4 ! �h4 8 �f4 ! .!iJf3 9 �fl tLlgl 10 �g4 ! liJe2 11 �h3 ! drawing. 2) 2 lDel ! 3 �c3 (the key point is that 3 �c5 loses after 3 ... .!iJd3+! 4 �c6 tLlf4 ! 5 �fl �e4 6 �d6 �d4, leading to one of the above recipro cal zugzwangs) �e4 ! 4 c;j;>d2 .!iJf3+ ! 5 �c3 (or 5 �dl c;j;>e3 6 �g2 'jj;>f2 7 i.h l .!iJh4) tLlh4 ! 6 �d2 �f3 ! 7 �el �g2+! 8 �d2 (8 �fl .!iJf4 !) 'jj;>f2 9 i.c8 lbh4 1 0 �h3 .!iJf3+ 1 1 c;j;>d3 �g5 and Black wins. The following over-the-board example features a missed opportu nity by Black. •••
=I=
159 Handoko Berg Gausdal 1992 -
(159): White was to move, and the game continued 1 g6 �c5 ! (not l ...�c4? 2 tLlc6+ �c5 3 liJe5 ! �a2 4 �g5 ! �d6 5 �f6 ! �bl 6 g7 �h7 7 �g4 �g8 8 .!iJh6 �h7 9 �f7 and White wins) 2 �e5 �h5? 3 g7 ! �f7
4 �f6 �a2 5 .!iJf5 �g8 6 'jj;>g6 i.a2 7 .!iJh6 �bl+ 8 c;j;>f7 �h7 9 �f8 c;j;>d5 10 tLif7 1-0. By playing his bishop to the a2-g8 diagonal via h5 and f7, Black gave White the chance to gain a vital tempo with 'jj;> f6. He could have drawn by 2 ... �c4! 3 g7 (3 c;j;>f6 c;j;>d6 ! ) �a2 ! (the bishop must move to a square which does not allow a fork after a later .!iJd4+, for example 3 . . .�b3? loses to 4 .!iJf5 and now 4 ...c;j;>c6 is impossible, while 4 ... �g8 5 .!iJh6 �h7 6 c;j;>e6 c;j;>d4 7 .!iJg4 �g8+ 8 �e7 c;j;>c5 9 'jj;>f8 is hopeless) 4 lbf5 �c6 ! 5 c;j;>f6 c;j;>d? ! 6 .!iJh6 c;j;>e8 ! and Black is just in time. Finally, it is worth noting that stalemate can sometimes arise in this ending. It is not worth giving a dia gram, but the position Wc;j;>el, �a2 v Bc;j;>e3, tLlc3, .0.g3 is an example. White can draw by means of 1 �c4 ! g2 2 �fl !. With the pawn on b5, Averbakh again described drawing zones for the black king. However, two of his three diagrams were incorrect. (160): The first diagram is rela tively unfavourable for Black. His bishop is on the shorter of the two diagonals, and White's king is ac tively placed. The immediate threat is tLic5-b7, driving the bishop off the diagonal, and then b6. Averbakh's drawing zone included the squares e4, f4 and g4, but it turns out that White to play can win if Black's king starts on one of these three squares. The principle involved is quite sim ple; Black draws if his king starts within three squares of d8, or within
1 18 ll.J+� v .i
161 Averbakh, 1958
Averbakh, 1958
two squares of the pawn, but other wise he loses. We will consider the following squares for Black's king: g5, e4, d3. B�g5: After 1 ll.Jc5 �f6 ! 2 ll.Jb7 .iel 3 b6 .if2 4 ll.Jc5 �e7 ! 5 b7 .ig3 ! 6 ll.Ja6 �d8 ! Black's king ar rives just in time. B�e4: White wins by 1 ll.Jc5+ ! �f5 2 ll.Jb7 ! (but not 2 ll.Jb3? .id8 ! 3 ll.Jc5, when Black need not repeat moves, since 3 ... �e5 ! leaves White on the wrong end of a reciprocal zug zwang) .ic3 (Black has no good square for his bishop; 2....id2 loses as in the main line, while 2 ... .iel 3 b6 �e6 4 ll.Jc5+! �e7 5 ll.Je4 ! domi nates the bishop) 3 b6 ! �e6 4 ll.Jc5+ ! �e7 5 ll.Jd3 ! (now the bishop cannot reach the b8-h2 diagonal) .id4 6 b7 ! .ia7 7 �c7 �f6 8 ll.Jb4 �e6 9 ll.Jc6 .igl 10 �c8 .ih2 1 1 ll.Jb4 ! and wins. B�d3: Black draws by 1 ll.Jc5+ �c4 ! 2 ll.Jb7 .ic7 ! . (161) : White's king i s not s o well placed on a6. The knight must travel
to e6 in order to drive the bishop off the a5-d8 diagonal, but even this is not enough to win in many cases. Averbakh included the squares a2, c2 and d2 in the drawing zone, but in fact these squares are wins with White to play. As before, the princi ple is simple; in order to draw, Black's king must be within four squares of d8 or two squares of at tacking the pawn. We will consider the following squares for the black king: h5, g4, e3, d3, d2. B�h5: 1 ll.Jc5 �g6 ! (but not 1 . ..�g4? 2 ll.Je6 ! , when White wins after 2....if6 3 b6! �f5 4 b7 .ie5 5 ll.Jc5 ! or 2....ih4 3 b6 �f5 4 b7 .ig3 5 ll.Jc5 ! .ic7 6 �b5 .ib8 7 �b6 .ih2 8 ll.Ja6 .igl + 9 �c7 �e6 10 �c8 .ia7 1 1 ll.Jc7+) 2 ll.Je6 .ih4 3 b6 �f7 (3 ... �f6? 4 ll.Jf4!) 4 ll.Jc7 (4 b7 .ig3) .ig3 ! 5 �b5 �e7 ! 6 �c6 �d8 ! drawing. In this line Black saved himself because his king could not only reach d8 in four moves, but could do so in such a way as to avoid
ti:)+� v i. 119
impeding the bishop transfer ...i.h4g3. If the Black king starts on h4 then he loses, because after 1 ti:)c5 Black has nothing better than 1 ...�g4. B�g4: If Black's king is within one square of f5 then he has an easy draw by 1 lLJc5 �f5 ! . B�e3: White wins by 1 ti:)c5 ! i.h4 (or 1 . . .i.c7 2 ti:)e6 i.h2 3 b6 ! �e4 4 ti:)c7 ! �d4 5 �b5 ! �e4 6 �c6) 2 b6 �d4 3 b7 i.g3 4 �b6 �d5 5 ti:)a6 i.f2+ 6 �c7 ! i.gl 7 �b4+ �c5 8 ti:)d3+ �d4 (8 ... �d5 9 �f4+) 9 �c6 i.h2 10 ti:)c5 i.b8 1 1 �b6. B�d3: Black draws comfortably by 1 ti:)c5+ �c4 ! 2 ti:)e6 i.h4 3 b6 i.g3 ! 4 b7 �d5 !. B�a3: In this case the method is slightly different. Now 1 ti:)c5 �b4 2 �e6 i.a5 3 ti:)f4 �a4 4 ti:)d5 i.d8 ! leads to a draw. B�d2: After 1 lLJc5 ! �c3 2 ti:)e6 ! i.h4 3 b6 ! �b4 4 b7 ! i.g3 5 �b6 ! i.f2+ 6 �c6 ! (we have transposed to diagram 144, but we repeat the main line here) i.a7 7 �c7 ! (Averbakh only mentioned 7 ti:)c7) �c4 8 ti:)d8 ! �d5 9 lLJc6 ! i.gl 10 ti:)b4+! �c5 1 1 �d3+ �d4 1 2 �c6 .i.h2 1 3 ti:)c5 with an easy win. (162): In general, Black is better off with his bishop on the longer di agonal. Even with the white king on the relatively active square c6 the drawing zone extends as far as White's second rank. In the above diagram, we first of all dispense with the exceptional squares b8, f6 and f8 . I n these case White wins b y 1 ti:)b6 ! and Black, for one reason or another, is unable to reply ... i.b8.
Averbakh, 1958 In the more general situation, White's main idea is to play 1 ti:)b6, threatening �b7. If Black replies 1 . . .i.b8, then 2 ti:)c8 followed by 3 b6 advances the pawn. We will consider the following squares for Black's king: a2, bl, f3, e3, f4, f5, g8, g7. B'iii>a2: Here 1 ti:)b6 'iii>b 3 2 'iii>b7 'iii> b4 3 �a6 �c5 4 ti:)c8 i.b8 ! 5 b6 �c6 is a simple draw. This draw operates whenever Black's king is within three squares of the pawn, with the sole exception of e2 (when Black cannot play ...�c4 after 1 ti:)b6 �d3 2 �b7). B�bl: White wins by 1 ti:)b6 ! i.b8 2 ti:)c8 �c2 3 b6! �b3 4 ti:)d6 �b4 5 ti:)b5 �a5 6 ti:)c7 'iii>a4 7 ti:)a6 i.h2 8 b7 . The same method works when Black's king starts on e2. B�f3: This is not substantially different from the last line: 1 ti:)b6 i.b8 2 ti:)c8 �e4 3 b6 �f5 4 ti:)d6+ �e6 5 ti:)b5 i.h2 6 �7+ is winning. B�e3: After 1 ti:)b6 �d4 ! Black's king is close enough to the pawn.
120 lb+l!. v i.
B�4: This is slightly different. White wins because he can gain a tempo by checking Black's king: 1 lbb6 i.b8 2 lDd5+ �e4 3 b6! i.h2 4 b7 i.b8 5 lbc7
.•
+I-
163 Suetin-Korchnoi Leningrad 1951
lDd5+!
lLJ+� v i. 121
.t.• • • • • • • • • •<&t>• • . L\ . • • • • • • • • • • .lb. • • • • • •• +I=
164
Original Wc7 and the advance of the b-pawn. This is a fairly long-winded plan, but luckily Black's king starts out on the most distant square possible. In deed, at first sight there is little de fence to this plan, for example 1 �d6 Wg2 2 �c7 �f2 3 �b8 i.h 1 4 lLJd4 We3 5 lLJc6 �d3 6 �c7 leads to the pawn running through. However, Black has a remarkably subtle de fence to White's plan. The idea is to meet 1 �d6 �g2 2 �c7 �f2 3 Wb8 by 3 ...i.g2! 4 lLJd4 i.n ! 5 b6 i.g2 ! 6 �c7 i.a8 ! , an incredible round trip by the bishop to end up back at a8. Now the position is drawn, because the plan of lLJc6 and Wb8 doesn't work when the pawn is oo b6, while playing for lLJb7 and Wb8 is too slow after 7 lLJc6 �e3 8 � �d4 9 lLJb7 �d5 10 �b8 �c6. Amazingly, Black's 3 ...i.g2! is the only move to draw, because after other bishop moves White can pre vent Black moving his bishop on to the fl-a6 diagonal, for example 3 ..i.d5 4 lLJb4 ! i.g2 5 �c7, 3 ...i.e4 .
4 lLJb4 ! �e3 5 lLJc6! �d3 6 �c7 ! or 3 ... i.f3 4 lLJd4 ! i.g2 5 lLJc6 ! . White only wins if Black is unable to exe cute the switchback with his bishop, so this defence puts the ball back in White's court. Now that the general ideas are clearer, we can give the de tailed analysis. White may try: 1 ) 1 lLJd4? (too direct; the black king is in time to attack the pawn) �g2 2 �d7 �f2 ! 3 �c7 �e3 ! 4 lLJc6 �d3! 5 �b8 �c4 ! and Black draws. 2) 1 �d6? �g2 ! 2 �c7 �f2 ! is a position of reciprocal zugzwang with White to move. If White moves his knight then Black draws by . . . �e3, for example 3 lLJb4 �e3 ! 4 lLJc6 �d3 !, while 3 �c8 is met by 3 ...i.g2! 4 lLJd4 i.fl ! 5 b6 i.g2 ! 7 �c7 i.a8 ! as before, since White's king has moved away from the de fence of c6. Finally, 3 �b8 was al ready analysed in the introduction to this position. 3) 1 �d7! �g2 ( 1 .. .�gl 2 �c8 amounts to the same thing, as Black has nothing better than 2 . . . �f2) 2 �c8 ! (threatening 3 lLJb4/d4 i.f3 4 �c7 �g3 5 lLJc6 i.e2 6 a6 i.a6 7 lLJb4; of course 2 �c7? �f2 ! is wrong) �f2 (after 2...�f3 Black's bishop can no longer reach g2 and White wins by 3 �b8 ! i.d5 4 lLJ
122 lb+� v .i.
c6, so Black can only move his king, but he can't move away from the pawn (or else .!Db4 wins). After 3 �e2 (3 ... �f3 4 �b8 ! wins as be fore) the fl -a6 diagonal has been blocked and White wins by 4 �b8 ! .i.g2 (4 ....i.e4 5 .!Db4 ! stops 5 ....i.d3 and 4 ... .i.d5 5 .!Db4 ! .i.c4 6 b6 ! pre vents 6 ....i.d5) 5 .!Db4 ! �d2 (Black has no chance if he cannot play ....i.fl ) 6 .!Dc6 ! �c3 7 �c7 ! .i.fl 8 b6 ! .i.a6 9 .!Db8 ! .i.fl 10 b7 and wins. •••
= • • • • • • • • • • • - � ·l'D· • • • • • • • • • • • • • . . . +I=
165
Original (165): Black to play draws by l .. .�f2 2 �a7 �f3, so assume that White is to move. He may try: 1) 1 �b7? �fl ! is, amazingly, a position of reciprocal zugzwang. White has various winning attempts, but all of them fail because there are no tactical tricks working in his fa vour, for example: la) 2 �a6 .i.d8 ! 3 .!Df4 (intend ing .!De6) .i.g5 ! and Black either transfers his bishop to e3 or, after 4 .!Dd5 .i.d8 !, repeats the position.
lb) 2 �c6 (intending .!Db6-c4, forcing ....i.d8, and then �d7) �e2! (but not 2 ... �f2?, when 3 .!Db6 fol lowed by .!Dc4 wins) 3 .!Db6 �d3 ! and .!Dc4 is prevented. le) 2 .!Dr4 (hoping for �a6 fol lowed by lbe6, but without allowing the defence of line la) .i.d2! (unfor tunately for White, Black has an analogous defence) 3 .!De6 .i.e3 ! 4 �c6 .i.a7 ! 5 lLJc5 �e2 6 .!Da4 �d3 7 .!Db6 �d4 drawing. 2) 1 �a7! (triangulation pro vides the key) with four possibili ties: 2a) 1 ...�h2 (now Black's king is too far away) 2 �a6 .i.d8 3 �b7 .i.h4 (3 . . . �h3 4 �c6 .i.a5 5 .!De3 �g3 6 lbc4 .i.d8 7 �d7 and 3 ... .i.a5 4 �c6 �h3 5 .!De3 followed by .!Dc4 are also winning for White) 4 �c6 .i.f2 5 .!Db6 �g3 6 .!Dd7 .i.a7 7 .!Db6 .i.b8 8 lbc8 �f4 9 b6! �f5 10 .!Dd6+ �g6 1 1 .!Db5 .i.h2 1 2 .!Dc7 .i.gl 1 3 b7 .i.a7 1 4 .!Db5 .i.b8 1 5 �d7 and wins. 2b) 1 ...�g2 (now White gains a tempo with a knight check) 2 �a6 .i.d8 3 .!Df4+ �f3 4 .!De6 .i.f6 5 b6 .i.e5 6 b7 �e4 7 �b6 .i.b8 8 .!De? winning. 2c) 1 ....i.d8 2 �b7 (threatening 3 �c6, when Black must play 3 ....i.a5 to prevent .!De?, but then 4 .!Db6 and 5 .!Dc4 wins) .i.h4 (2 ... .i.a5 3 �a6 .i.d8 4 .!Df4 wins because 4 ....i.g5 is impossible) 3 �c6 .i.f2 4 .!bb6 �g2 5 .!Dd7 .i.a7 6 .!Db6 and wins. 2d) 1 �n 2 �b7 ! (not 2 �a6? .i.d8 ! 3 �b7 .i.a5 ! 4 �c6 �e2! 5 .!Db6 �d3 ! and Black stops .!Dc4) puts Black on the wrong end of the •.•
lll +� v .t 123
reciprocal zugzwang. The various king moves all have a tactical flaw: 2d l ) 2 �e2 (2 ... �g2 loses in the same way) 3 �a6 .td8 4 lllf4+ ! (gaining a tempo) �e3 5 llle6 ! .tf6 6 b6 wins. 2d2) 2 'iPfl 3 �a6 .td8 4 lllf4 and . . ..tg5 is impossible because of �3+. 2d3) 2 �el (again the problem is a knight check) 3 lllf4 ! .td2 4 �3+ ! �e2 5 lllc5 ! .ta5 6 �c6 �f3 7 lllb7 .tc3 8 b6 .td4 9 lllc5 ! win ning. 2d4) 2 .td8 3 �c8 .ta5 4 llle3+ �e2 5 lllc4 and the white pawn ad vances. .••
...
...
...
The Averbakh diagrams 1 1 1 and 1 1 2 show that almost all positions with the pawn further back are drawn. Indeed, White can only win either by some immediate tactical point or by exploiting a bad enemy bishop position to push the pawn and reach one of the endings considered above. There are 9 reciprocal zug zwang with lll+�b4 v .t, 5 with ��b3 v .t and 5 with lll+�b2 v .t. None of them are very interesting, and I will just give one (which is fairly visual) without a diagram: W�b7, llle5, �b4 v B�d6, .tc7. The longest win in the ending �+b� v .t is the following position (26 moves). (166): 1 b6 ! .tg5 2 �e5 ! .td8 3 b7 ! .tc7+ 4 �d4 ! (but not 4 �d5? �d3 ! 5 �c6 .tf4 followed by ...�e2 trapping the knight) �b3 (trying to reach a6) 5 �c5 ! (White must take time out to head off Black's king)
+I=
1 66
Original �a4 6 �c6! (gaining access to b6 in order to stop ...�a5) .tf4 7 �b6 ! (not 7 llle 3? �a5 ! 8 llld5 .tb8 ! 9 lllc7 .ta? ! 10 lllb5 .tb8 ! drawing) �b4 8 llld2 ! (the knight escapes to the only possible square) .te3+ 9 �a6 ! (after 9 �c6? .ta7 ! Black's king does reach a6) .tf4 10 llle4 ! (now White must prevent ... �c5) .tc7 ( 1 0 ... �c4 1 1 �b6 ! �d5 1 2 lllc5 .tg3 1 3 llla6 ! .tf2+ 14 �c7 ! .tgl 15 lllb 4+! transposes to the main line, so Black tries to hold up �b6) 1 1 lllg5 ! (again preventing .. .�c5) .th2 1 2 llle6 ! (now �b6 cannot be delayed) .tg3 1 3 �b6 ! .tf2+ (the threat of lllc7 forces Black to transfer his bishop to the less effective square a7) 14 �c6 ! .ta? 15 �c7 ! (we have seen this win before, for example in diagram 1 6 1 with B�e3) �c4 1 6 llld8 ! �d5 17 lllc6 ! .tgl 1 8 lllb4+! �c5 1 9 llld3+ �d4 20 �c6 .th2 21 lllc5 .tb8 22 �b6 .th2 23 llla6 �c4 24 lllc7 .tgl + 25 �a6 with promotion or bishop capture next move.
124 lLJ+� v .t
2.3
ltJ+� v .lt
This ending has, for some reason, re ceived much more attention from analysts than the corresponding case with a b-pawn. We start with the pawn on c7. For the most part, the pre-existing analy sis has been accurate, so we begin with a survey of established mater ial. Although there are some simi larities in the general principles in volved, the precise details show many differences from the previous section 2.2. The main reason is that the diagonal a6-c8 is significantly longer than a7-b8, so being forced onto the shorter diagonal is not so disadvantageous as with the b-pawn. Averbakh constructed one of the most important drawing zones.
Averbakh, 1958 (167): The corresponding draw ing zone with the pawn on b7 con sisted of two squares (see diagram
142), so we can see how Black's drawing chances have improved. We will consider the following squares for Black's king: a7, e8, f7, f6, f5, a3. The general principle is that White threatens either to play his knight to c5, followed by �e7-d8, or the other way round, i.e. first �d7 and then lLJa4-c5. In order for Black to draw he must be in time to disrupt these manoeuvres. B�a7: After l lLJf6 2 lLJa4 �e5 3 lLJc5 ! .tg2 4 �d8 .th3 5 lLJd7+) 2 lLJa4 .tn 3 �d8 .th3 4 lLJc5, followed by lLJd7. B'iii>f6: After l �d7 (or l lLJd7+ �f7 ! 2 lLJc5 .tcs !) �e5 ! 2 lLJa4 �d5 Black prevents lLJc5 and draws. B�fS: Now Black is too far away from f7 and White wins by l lLJd7 �g6 ( 1 . . . �f4 2 lLJc5 .i.c8 3 �e7 �e5 4 �d8 !) 2 lLJc5 .i.c8 3 �e7, fol lowed by �d8. B�a3: If Black is within two squares of c5 then he draws by l lLJd7 �b4 2 lLJc5 .tc8 !, but if his king starts on a2, then White would have time to win by 3 �e7 and 4 �d8. Two positions by Ch6ron illus trate some of the subtleties which can arise with this type of ending.
lLJ+l>, v i. 125
168
+I= Cheron, 1960 Journal de Geneve
(168): White wins as follows: 1 lLJc4! The key to this position is that White should not commit his king, because in some lines it goes to b6, in others to c6 and in yet others d6. It is tempting to play 1 �c6?, threat ening 2 lllb 7, but Black draws by l ...i.c8 ! 2 lllc4 �f6 ! 3 �d6 i.a6! (but not 3 ... i.b7? 4 �d7 ! i.a6 5 �6 !, threatening �c6-b6, and win ning after 5 ...i.fl 6 �d8 i.h3 7 �4 ! i.g4 8 lllb6 and llld 7) 4 lllb6 (4 �d7 i.xc4) i.b7 ! (this is drawn by the previous diagram) 5 'itr>d7 �e5 ! 6 llla4 �d5 stopping lllc5 . i.c8 1 The only defence to the threat of 2 �6 followed by �c6, for example l...�f6 2 llld6 ! �e7 3 �c6 ! and 4 �b7. 2 lllb6! Not 2 llld6? (2 �d6? �f6 ! 3 lllb6 i.b7 4 llld5+ �f7 is also bad) i.g4 ! (after 2 ...i.h3? 3 lbe4+ ! Black can not continue 3 ...�g6 4 �d6 ! �f7 ...
because of 5 lllg 5+!, and the alterna tive defence 3 ... 'ifr>f4 loses to 4 �d5 ! i.c8 5 lllc5 �f5 6 �d6 'ifr>f6 7 llld7+ �f7 8 lllb 6 ! i.b7 9 �d7 ! i.g2 10 'itr>d8) 3 llle4+ �g6! 4 �d6 �f7 ! (now this square is available to the king, and Black draws) 5 lllg 5+ 'ifr>f6! 6 lllf3 (threatening to win by 7 llle5 i.c8 8 llld7+ followed by 9 lllb6 and Black's king will not be in the Averbakh drawing zone) i.c8 ! 7 llle5 i.a6 8 llld7+ �f7 ! 9 lllc5 i.c8 ! and Black holds on. 2 i.b7 3 lbdS! Once again White must not com mit his king. After 3 �d6? �f6 ! Black enters the drawing zone. The move played threatens llle7 fol lowed by �b6, and so forces the bishop to move again. i.cS 3 Now White wins by 4 'ifr>c6 i.h3 (Black's king cannot move to f5 or g6, and after 4 ...'ifr>h6 5 lllb6 i.a6 6 'itr>d7 Black is much too far away) 5 lllb6 i.g2+ 6 �d7 i.b7 7 'itr>d8 �f6 8 llld7+ �f5 9 lllc5 and the pawn promotes. (169): The solution runs: 1 lLJc4! Not 1 lllb 7? �e3 ! 2 llld6 i.g4 3 lllc4+ �d4 ! 4 lllb6 i.f3+ ! 5 'itr>d6 i.b7 ! and Black draws. 'ifr>f4 1 Black would prefer to keep his bishop on c8, where it is ready to meet lllb6 by ...i.a6 and llld6 by ... i.h3, so he will move his king whenever possible. The squares e3 and e4 are out of bounds, while after l ...�e2 ( l ...i.f5 2 llle5+ and 3 llld7) ...
...
...
126 lLJ+� v .t
.... . . . • D • • ·�· . . � . . . • • • • • • ••• • • • • • • • • 1 69
+I=
Chiron, 1955 Journal de Geneve
2 liJd6 i.h3 3 ltJe4 'it>e3 4 lLic5 i.c8 5 �d5 we have transposed into the main line. If Black were now to move, he would lose quickly. The king cannot go to e4 or f5, so he must either play the king away from the pawn or make an unfavourable move with the bishop. White can transfer the move to Black by triangulating. 2 �dS! White must go round the triangle clockwise. 2 'it>c5? is a mistake, al lowing Black to run back with his king: 2 ...�f5 ! 3 liJd6+ 'it>e6 ! 4 ltJxc8 �d7 ! 5 b6 'it>xc8 ! and White can not win. 2 i.b7+ There is nothing better, for exam ple: 1) 2 i.a6 3 liJd6 'it>g5 4 'it>c6 and 5 liJb7. 2) 2 i.h3 3 'it>d6 i.c8 (or else liJb6-d7 wins; White lost one tempo with 'it>c6-d5-d6, but Black has lost two with his bishop) 4 liJb6 i.b7 5
liJa4 i.c8 6 ltJc5 �f5 7 e6 e7 ! 'it>e5 7 ltJc5 ! f6 e7 i.h3 7 liJb7, followed by ltJc5) �e2 4 liJd6 i.d7 5 ltJe4 d5 ! transposes to the main line . 3 e3 (otherwise �d6-e7-d8 wins) 8 �d5 ! (but not 8 �d6? �d4 !) �d2 (Black's long-suf fering king has to move away again) 9 �d6! e7 ! �d4 1 1 'it>d8 ! i.h3 12 liJd7 ! cutting off the black bishop. Chekhover composed a study based on a similar theme. ...
...
...
...
...
170
+I=
Chekhover, 1960
lb+� v i. 127
(170): In this position White re peatedly improves the position of his knight with gain of tempo. Only when White's knight is optimally placed does the king advance. The solution runs l lbd6 ! (not 1 lbc5? �c2 2 �d5 �c3 ! 3 �d6 �c4) i.d7 (or 1 . . .i.a6 2 �c5 and 3 �b6) 2 �4 ! (threat 3 lbb6, followed by �c5-d6 and lbd7) i.c8 3 lbb6 ! i.b7 4 lbd5 ! (another similar manoeuvre; the threat is 5 lbe7, followed by �c5-b6) i.c8 5 lbe7 i.d7 (5 ...i.h3 6 �e5 intending lDf5) 6 �d5 'iti>d2 7 �d6 ! i.h3 8 lbd5 (planning lbb6 and lbd7) i.c8 9 lbb6! i.b7 10 lba4 �e3 1 1 lbc5 i.c8 1 2 �e7, followed by �d8, and White wins. If White starts with his king on b4, then the position is drawn. After 1 �6 i.d7 2 lbc4 i.c8 ! 3 lbb6 i.b7 4 �5 i.c8 ! 5 lbe7 Black can defend by 5 ... i.h3, because White cannot easily threaten lDf5 .
m.t.m m m m W'� m m - - - m mtt:Jm m - - - - -·- - - - - - - =I=
171
Chekhover, 1960 (171): This position is drawn, but Black to move must play accurately:
1) 1...�e2? 2 �d4! is reciprocal zugzwang. White wins after 2 ...'iiif3 (2 ... 'iti>dl 3 lbe7 wins by the previous diagram) 3 lbb6 i.a6 4 �e5 �e3 (4 . . . i.b7 5 lba4 �e3 6 lbc5 ! i.c8 7 'iti>d5 ! ) 5 lbc4+ �d3 6 lbd6! 'iti>c3 7 'iiid5, followed by �c6 and lbb7. 2) 1...�e4? 2 lbb6 i.a6 3 lbc4 'iti>f4 (Black cannot play ...i.c8) 4 lbd6 'iti>e5 5 �c6 !, again followed by lbb7. 3) 1 ...�c2! 2 'iti>d4 (2 'iiib4 �d3) 'iti>b3 ! 3 lbb6 i.b7 4 �c5 i.a6 draws. The following reciprocal zug zwang is closely related to the above position:
I
172
= -
Original (172): First of all suppose that Black is to play. He may try: 1 ) 1 .i.h3 2 �e5 (gaining a vital tempo by threatening lDf5) i.d7 3 �d6 i.h3 4 lbd5 i.c8 5 lbb6 i.b7 6 lba4 �e3 7 lbc5 i.c8 8 �e7 �d4 9 'iiid8 and wins. 2) 1...�f3 (now Black cannot meet �e5 by ... �d3) 2 �e5 �e2 3 �d6 winning as in line 1 .
..
128 lLJ+� v .t
3) 1 'iii>d2 (allowing a deadly knight check at move 4) 2 ltid5 .ic8 3 ltib6 .ib7 4 ltic4+ 'iii>e 2 5 lLid6 .ia6 6 'iii>c5 and wins. 4) 1 'iii>d l (this is simply too far away) 2 d5 'iii>d2 3 'iii>d6 ! .ih3 4 lLid5 .ic8 5 ltib6 .ib7 6 ltia4 win ning as before. With White to play, 1 'iii>d5 'iii>d3 ! leaves White a tempo down over the Black to play variations; Black draws after 2 'iii>d 6 .ih3 3 lLid5 .ic8 4 ltib6 .ib7 5 ltia4 'iii>c4 6 lLic5 .ic8 !. White can also try 1 lLid5, but then 1 .. ..ic8 ! gives rise to a second recip rocal zugzwang with White to play (the Black to play analysis appeared in the previous diagram). After 2 ltib6 (or 2 q.,e5 d3 !) .ib7 ! (but not 2 ... .ia6? 3 'iii>e 4! .ib7+ 4 ltid5 ! .ic8 5 'iii> d4 ! and White has turned the tables) 3 ltic4 .ic8 ! 4 lLid6 .id? ! White is not making progress. •••
...
• • • • . -�· . • • • • • • • • . - . . • • • • ••m • • • • • • =I+
173
Borge-V. Fedorov Tastrup 1992 (173): Black to play would win by 1 . . .ltib3 2 .ic3+ 'iii>d5 3 .ib2
lLic5+ 4 'iii>c7 ltid3 5 .ia3 'iii>c4, but in the game White was to move. The position closely resembles the pre vious two, but the defender's king is in a slightly better position than in diagram 170, and so the result should be a draw. The game contin ued: 1 'iii>c7? The correct defence was 1 .icl ! ltib3 ( 1 . . .ltie2 2 .id2 ! and 1 . ..'iii>e4 2 'iii>d 6! are also drawn) 2 .ib2+! (gaining a vital tempo to remove the white king from d7, where it is vul nerable to ... ltic5+ followed by ...lLid5 3 'iii>c7 lLic5 4 .icl ! 00 5 .id2, and White draws. ltib3? 1 Missing a win by 1 .. .'iii>e4 ! 2 .ic 1 ltie2 ! 3 .id2 (3 .ib2 'iii>d3 4 'iii>b6 'iii>c4 !) 'iii>d 3 4 .ih6 ltid4 5 .ic l ltib3 6 .ib2 ltia5 7 .icl ltic4, followed by playing the king to d l . 2 .ic3+! Perhaps Black had overlooked that the bishop can be transferred to the other side of the pawn. 'iii>dS 2 3 .ib2! lLic5 Now White has set up the correct defensive formation, and although Black carried on for another 26 moves, he never again had a winning position: 4 .ic 1 ! ltib3 5 .ib2 'iii>c4 6 'iii>b 6 lLic 3 8 'iii>b5 ltib3 9 .ia3 ltid4+ 1 0 'iii>c5 ltie2 1 1 .ib4+ 'iii>d3 12 .ia3 lLid4 13 .ib2 lLie6+ 14 'iii>b4 'iii>d2 15 'iii>c4 ltif4 16 .ia3 lLid3 17 'iii>d4 lLif4 1 8 .ib2 ltie6+ 19 'iii>c4 ltig5 20 'iii>d4 ltif3+ 2 1 'iii>c4 ltie l 22 'iii>d4 lLid3 23 .ia3 ltif2 24 .ib2 lLic4 ltia4 2 7 'iii>b4 ••.
•••
lb+� v i. 129
�c3 28 'iti>c4 lLibl 29 i.b2 'iti>e2 30 �d4 lf2_1h. The following study deals with a quite different type of position.
+I=
174
Kalinin, 1974 Rus (174): White can try: 1 ) 1 liJd5? i.b2 ! (surprisingly, this is Black's only drawing move; l . . .i.d4? loses to 2 'iti>e7 ! 'iti>g6 3 �f4+ ! rJ;h7 4 ltJe6 ! i.c3 5 'iti>e8 i.b4 6 llld8 rJ;g7 7 liJc6 ! i.f8 8 llle7 !) 2 �e7 'iti>g6 ! 3 liJf4+ 'it>h7 ! 4 rJ;es (or 4 �g2 i.a3+! 5 rJ;e8 rJ;g7 6 lllf4 ..t>f6 ! 7 liJd5+ rJ;e6 !) i.g7 ! with a draw. 2) 1 liJh5+! rJ;f8 2 lLif4 ! c:llg7 (or 2 ...i.b4 3 rJ;f6 ! i.c3+ 4 rJ;g6! rJ;e7 5 �6) 3 rJ;e7 ! rJ;h7 (3 ...i.b4+ 4 rJ;e8 ! i.a3 5 ltJd5 ! i.f8 6 ltJe7 !) 4 lLig2 ! (White wants to play his knight to f5, but normally Black can prevent this with ... 'iti>g6 at an appropriate moment; the correct route is there fore via h4, covering g6) rJ;g6 (or 4 . . . i.b2 5 rJ;es ! and White wins af ter 5 ...i.g7 6 liJh4 ! i.h6 7 lLif5 or 5 ... i.a3 6 liJf4 rJ;g7 7 llld5 ! i.f8 8
llle7 !) 5 ltJh4+ ! 'iti>g5 6 lLif5 ! i.b4+ 7 liJd6 ! and wins. We end with one of the 60 posi tions of reciprocal zugzwang in the ending of lll+�c7 v i..
=I-
175
Original (175): Suppose firstly that Black is to move. If he moves his king to anywhere except for g6 or f5, then White wins by 2 lLif6, followed by 3 ..t>d8 i.b7 4 liJd7 and 5 lllc5 . Other moves: 1) 1 ...rJ;g6 ( l . . .rJ;f5 2 'iti>b8) 2 rJ;d7 ! (threatening 3 rJ;e7 i.h3 4 liJf6 i.c8 5 llld 7, followed by rJ;d8 and ltJc5) c:llf7 (2 . . . i.b7 3 lLie7+! c:ll f7 4 ltJc6 ..t>f6 5 llla5 i.a6 6 liJb3 i.c4 7 rJ;d8 i.e6 8 lllc 5) 3 lLih6+ ..t>g6 4 ..t>e7 i.h3 (or 4 ...i.b7 5 ltJf7 ! i.a6 6 ltJe5+! rJ;f5 7 liJd3) 5 liJf7 'iti>g7 6 ltJe5 i.c8 7 rJ;d8 i.a6 8 liJd7 and wins. 2) 1...i.e4 ( l . . .i.d5 loses in the same way) 2 llle7 (threatening rJ;b8, and so gaining a vital tempo) i.f3 3 rJ;d8 i.b7 4 ltJc6, with ltJa5-b3-c5 to come.
130 lLJ+� v i.
3) 1 i.f3 2 'iii>d 7! 'iii>f4 (by far the best defence) and now: 3a) 3 lLJf6? i.b7 ! 4 lLJe8 'iii>e5 ! 5 lLJd6 i.g2 ! drawing, but not 5 ...i.f3? (nor 5 . . .i.a6? 6 'iii>c 6 ! ) 6 'iii>d8 i.g4 7 lLJc4+ and now ... 'iii>d5 is impossi ble, so Black cannot stop lLJb6-d7. 3b) 3 lLJe7? i.g4+ 4 'iii>d8 'iii>e 5 draws easily. 3c) 3 lLJh6! i.b7 (or 3 ... i.g2 4 li)f7, when White wins after 4 ...i.fl 5 'iii>e6 i.g2 6 'iii>f6 i.h3 7 lLJd6 or 4 ...i.h3+5 'iii>e7 i.g4 6 'iii>f6) 4 lLJf7 ! (threatening 5 lLJd8 i.a6 6 lLJe6+ and 7 lLJc5) i.g2 (4 ...i.a6 5 lLJd6 i.fl 6 lLJc4 i.g2 7 'iii>d8 i.b7 8 lLJas loses) 5 'iii>e7 i.h3 (5 ...i.b7 6 lLJd8 i.c8 7 lLJe6+ 'iii>e5 8 lLJc5 ! ) 6 'iii>f6 'iii>e 3 7 lLJd6 ! 'iii> f4 8 lLJc4 i.c8 9 lLJb6 i.b7 10 lLJd7 'iii>e 3 1 1 lLJc5 i.c8 12 'iii>e7 and wins. With White to play, after 1 'iii>d7 ( 1 'iii>d8 is met by l . . .i.b7 !, while 1 lLJe7 'iii>f6 ! is a draw after 2 'iii>d8 i.b7 ! or 2 'iii> d7 'iii>e 5) 'iii>f4 ! 2 lLJf6 (2 lLJe7 'iii>e5, 2 'iii>e7 i.b7! and 2 lLJh6 i.h3+ 3 'iii>d6 'iii>e4 are no better; in the lat ter line, Black is able to occupy the h3-c8 diagonal without losing a tempo) i.b7 ! 3 'iii>d6 'iii>e3 4 lLJd7 'it>d4 Black's king is sufficiently ac tive to draw. •.•
Now we move on to lLJ+�c6 v i.. Averbakh gave very detailed draw ing zones for this situation; the com puter has verified all his analysis. In each of the four cases, the general principles governing the position are straightforward. (176): This is the most favourable case for White, since Black's bishop
Averbakh, 1958 is on the shorter diagonal and White's pieces are well-placed. Even so, the drawing zone is relatively large. The basic principle is that White threatens lLJci6-c4, driving the bishop away. Black can prevent this if his king is within two squares of c4, which explains most of the zone. We therefore need only consider two special cases: B'iii>a8: White wins by 1 'iii>c8 ! i.c5 (or l . ..i.a5 2 lLJd6 'iii>a7 3 lLJc4 i.el 4 'iii>d7 i.g3 5 lLJci6 !) 2 'iii>c7 i.gl 3 lLJd6 i.a7 (3 ... i.f2 4 'iii>d7 i.b6 5 lLJc4) 4 'iii>d7 i.b8 5 lLJb5 i.h2 6 lLJd6, followed by c7. B'iii>b8: In this case White cannot drive the bishop off the a5-d8 diago nal, for example 1 lLJci6 i.a5 ! 2 lLJc4 i.c7. (177): The drawing zone expands even further when we move White's king to the less active square d5 ; in deed Black's king can be anywhere except for the right-hand edge of the board. In order to win, White needs time to play 'iii>d6 and lLJa3-c4. This
ll:i+.0. v .t 131
178
Averbakh, 1958
Averbakh, 1958
will force ... i.d8, whereupon �d7 drives the bishop off the diagonal. Normally, Black can draw if his king IS within three squares on c4, be cause this manoeuvre is prevented. This deals with the entire bottom half of the board. Black can also draw if his king is within four squares of c6. We consider the fol lowing squares: g2, g l , g6, g5. B�g2: Black draws by 1 �d6 �f3 ! 2 ll:ia3 �e4 (2 ... .tgl 3 �d7 .lh2 4 ll:ib5 i.b8 also draws) 3 ll:ic4 .ld8 ! 4 �d7 �d5 !. B�gl: With his king on g l , nei ther drawing condition holds and Black loses after 1 �d6 ! �t2 2 ll:ia3 ! .le3 3 ll:ic4 ! i.f4+ 4 lbe5 !. B�g6: White wins by 1 �e6 ! �g5 2 �d7 !, followed by ll:id6-c4. B�gS: Black's king can squeeze around and draw by 1 �e6 �f4 ! 2 &3 �e4 !. (178): White is slightly better off with his king on b7 . There is the fairly direct threat of ll:id4-e6 fol lowed by �a6, but in addition White
has a second plan involving ll:id6, followed by �c8-d7 and ll:ic4. We will consider the following squares for Black's king: g6, f5, f4, e2, d2. B�g6: White wins by 1 �c8 ! (but not 1 ll:id6? �f6 ! 2 �c8 �e7 ! 3 ll:if5+ �e8 !) i.a5 2 �d7 �f6 3 ll:ia3, followed by ll:ic4. B�fS: This is a draw because Black is in time to prevent ll:ic4 after 1 �c8 i.a5 2 �d7 (2 ll:id6+ �e6 !) �e5 3 ll:id6 �d5 . B�f4: Now Black is too far away from e6, so White wins by 1 �c8 .ta5 2 ll:id6 i.b6 (2 . . .�e5 3 ll:ic4+) 3 �d7, again with ll:ic4 to come. White wins the same way if Black's king starts on e3. B�e2: Now White can win with his other plan: 1 ll:id4+! �d3 2 ll:ie6 !, followed by �a6. B�d2: In this case Black can de fend the bishop by playing his king to b4: l ll:id4 �c3 2 lbe6 i.a5 3 �a6 �b4. with a draw. (179): In the last of the four Aver bakh positions, B lack's bishop is
132 lb+L!. v i.
Averbakh, 1958 on the longer diagonal. This is a fa vourable situation for Black, and again White (to move) can only win if Black's king lies near the right hand edge of the board, or if it starts on the very unfavourable square a8. White's plan is 1 lbc7 followed by 2 'it>c8 i.a7 3 lbb5 i.b6 and then either ..ti>d7 and lba3-c4, or b7 and lbd4-e6. This plan is so slow that Black's king has to be truly badly placed for it to succeed. We will con sider the following squares for the black king: a8, f7, g6, g3, f2, fl. B�a8: White wins by 1 �e7, put ting Black in zugzwang, and the pawn promotes after 1 ... i.h2 2 lbd6 'itib8 3 'itid7 !. B'itir7: White can exploit Black's bad king position by 1 lbc7 ! i.a7 ( 1 . . .'itig8 2 �c8 i.a7 3 lbb5 i.b6 4 'itid7 'itif7 5 lbd6+ 'itif6 6 lbc4 and 1 . .. 'itif6 2 'itic8 ! i.a7 3 lbd5+ 'itie6 4 c7 are also lost for Black) 2 lba6 i.b6 3 lbb4 i.a5 4 lbd5 winning. B'itig6: It is quite surprising that Black's king is better placed on g6
than f7, but in this case 1 lbc7 ..ti>fS '. (Black is able to head for the centre without playing his king to f6. where it is exposed to a knigh1 check) 2 'itic8 i.a7 ! 3 lbb5 i.b6! 4 'itid7 (4 'itib7 i.a5) �e5 5 lbd6 'itid.S is a draw because Black is in time to prevent lbc4. B..ti>g3: 1 lbc7 �f3 2 �c8 ! i.a7 3 lbb5 i.b6 4 lbd6 'itie2 (or 4 .....ti>f4 S �d7) 5 lbc4 i.gl 6 'itid7 and the pawn promotes. B�f2: 1 lbc7 i.a7 ! (not L...ti>e3? 2 �c8 ! i.a7 3 lbd5+ ..ti>d4 4 c7, nor 1 . ..�f3? 2 �c8 ! i.a7 3 lbb5 i.b6 4 lbd6 winning as with B..ti>g3) and now: 1) 2 lbdS and 2 lbe8 are both met by 2 ...i.b8 !. 2) 2 �c8 i.d4! (the bishop mus1 be ready to transfer to the b8-h2 di agonal, but 2 ... i.c5? 3 lbb5 i.b6 4 lbd6 followed by lbc4 wins, as does 2 ... i.e3? 3 lbe6 ! i.b6 4 'itib7 i.a5 S ..ti>a6) 3 lbb5 i.e5 ! 4 ..ti>d7 i.b8 ! and White is going round in circles. 3) 2 lba6 i.b6 ! 3 lbc5 �e3 ! (the black king started one square closer to c4) 4 lba4 i.a5 ! 5 lbb2 'itid4 ! and Black is just in time. B'itifl: Now White wins because after 1 lbc7 ! i.a7 2 lba6 i.b6 3 lbc5 ! 'itie2 4 lba4 ! i.a5 5 lbb2 ! Black's king cannot move to d3. Now that we have covered the fundamentals, we can move on to some correct and not so correc1 compositions. (180): Cberon's study is quite tricky, but we have already done most of the work in the previous diagram: 1 i.c5+! (not 1 ..ti>c8? lbe4
lLi+� v i. 133
=I+
1 80
Cheron, 1952 La Fin de Partie 2 i.d8 lL!d6+ ! 3 b7? lL!d l ! 3 i.gl lLif2 and 2 �c8? lL!d l ! 3 i.gl lL!f2 lose be cause White's king is outside the drawing zone. It follows that White bas to play 2 i.gl ! first, and after 2 ... lLid l 3
+I=
181
Kosek lLid7 i.d8 5 �c8 i.a5 6 lLie5+
..
...
134 lLJ+� v �
+I= Prokop, 1952 1 st Pr., Bulletin Ouvrier des Echecs
=I= 1 83 Kling and Horwitz, 1851 Chess Studies and Endgames
�d4 6 lLJc4+ �b4 7 �b7 is similar) 4 lLJd3 ! �gl (forced) 5 lDe5 �d4 6 lLJc4 (the third domination by the knight) �f2 7 �d7 �g3 8 lDd6! winning. 2) 1 ...�d4 2 lDe4 ! �b3 3 lDd6! �c5 (3 ... �c3 4 �b6 �a4 5 lLJc4 !) 4 lDc8 ! (a truly amazing move; not be ing content with blocking the pawn by �c7 , the knight now creates a further obstruction, and yet it is pre cisely this move which enforces the rapid advance of the pawn) �f2 (the threat was 5 �b7 �gl 6 c7 �h2 7 lLJd6, and 4 ... �b4 loses to 5 �b7 �el 6 c7 ! �g3 7 lDd6 !) 5 �d7 ! �el 6 c7 ! �a5 7 lDb6! and wins. The following position is of a completely different type. (183): Kling and Horwitz be lieved that White to play could win this position, giving the analysis 1 �e7 �h7 ( 1 . ..�a3+? really does lose after 2 �e8 ! �f8 3 f7+ �g7 4 lLJg6 �d6 5 lDe7 !) 2 f7 �a3+? 3 �e8! �g7 4 lLJc4 ! �b4 (or 4 ...�c5 5
lDb6 �g6 6 lLJd5 �f5 7 lDe7+!) 5 lDe3 �a3 6 lLJd5 �f8 7 lDe7 and wins. However, Bianchetti showed that Black could draw by transfer ring his bishop to the f8-h6 diagonal: 2 ...�c l ! 3 �e8 �h6! 4 lLJc6 �g6! 5 lDe7+ �g5 ! 6 lDg8 �g7 ! with a clear draw. It is worth mentioning one inter esting tactical possibility. In the fol lowing position White wins by an underpromotion. (184): Here White wins by 1 f7 ! �g8 2 f8� ! �c4 3 lDf6, followed by �g7 mate. However, this only works because White's knight is favour ably placed. If the knight starts OD d8, for example (Sackmann, 1 9 1 2), then the position is a draw after 1 f7 �g8 ! . The following reciprocal zug zwang, chosen from the 57 with lLJ+�c6 v �. is moderately baffling. (185): There are two important points to bear in mind. First of all, the play depends fundamentally OD
1 82
liJ+� v .i 135
=I= 1 84 Gurvich, 1929 (end of study) 3rd HM, Shakhmaty the reciprocal zugzwang in diagram 172. The second is that it may not be good to play c6-c7, because White intends to dominate Black's bishop on the c8-h3 diagonal, and for this White needs d7 to be covered by the pawn. Suppose that.Black moves first: 1 ) 1...�d2 (now Black's king is exposed to a later knight check) 2 c7 .ic8 3 liJe7 ! .id7 4 liJd5 ! (not 4 �d5 ? �d3 ! 5 �d6 .ih3 6 liJd5 i.c8 ! 7 liJb6 .ib7 8 liJa4 �c4 9 liJc5 .ic8 !) .ic8 5 liJb6 .ib7 6 liJc4+ (and here it is - White gains a tempo) �c2 7 liJd6 ! .ia6 8 �c5 and wins. 2) 1. .ibS 2 c7 ! .id7 3 liJe7 ! is the reciprocal zugzwang of diagram 172. 3) 1...�f3 2 c7 .ic8 (2 ... �f4 3 liJe7 ! �g5 4 �c5 �f6 5 �b6 ! wins) 3 liJe7 .id7 (3 ... .ih3 4 �e5 threat ens liJf5 and wins after 4 ....id7 5 �d6) 4 �e5 �e2 5 �d6 .ih3 6 liJd5 i.c8 7 liJb6 .ib7 8 liJa4 �e3 9 liJc5 i.c8 10 �e7 wins for White. ..
• • •ttJ• • • • • .i.· � · • • • • • • . ·� � � . � � • • • • • ••• • • • • • =I-
1 85
Original 4) 1.. .id3 and now: 4a) 2 c7? .if5 ! 3 liJe7 .id7 ! with the reciprocal zugzwang of diagram 172, but this time with White to move. 4b) 2 liJe7? .ib5 ! 3 c7 .id7 ! is the same. 4c) 2 �dS? .ib5 ! 3 c7 .id7 ! 4 �d6 .ih3 5 liJf6 .ic8 ! 6 liJd5 �d3 ! 7 liJb6 .ib7 draws. 4d) 2 �cS! leads to yet another branch: 4dl ) 2. .�e3 3 liJe7 ! .ifl 4 liJf5+! �f4 5 c7 ! .ia6 6 lbd6 ! and White wins. 4d2) 2...�f3 3 c7 .if5 4 �d6 .ih3 5 �e5 ! and there is no defence to the threat 6 liJe7 i.d7 7 �d6, win ning as in line 3. 4d3) 2....ie4 3 c7 ! .if5 4 liJe7 ! .ih3 5 liJd5 ! .ic8 6 �d4 ! with recip rocal zugzwang; White wins as in line 1 of diagram 17 1 . 4d4) 2. .ifS 3 lbe7 .ih3 4 �d6 (4 c7? �d3) �f3 (the threat was �e5 and liJf5 ; because d7 is out of bounds, Black has no good square .
.
..
136 lb+� v .i.
for his bishop) 5 �e5 �g4 (prevent ing the immediate threat, but now that Black's king is poorly placed it is possible to push the pawn) 6 c7 �g5 7 lD
(186): With Black to play, White must not advance his pawn to c6 too hastily, because he needs to have his knight defended on d6. The alterna tives are: 1) 1 �h4 2 lbe3 ! (now Black cannot play ....i.g3) �g5 3 lbc4 ! (not 3 c6? .i.g3 ! 4 lbc4 .i.b8 !, and this is where White would like to play �c8 followed by liJd6, but his pawn has advanced too soon) .i.g3 4 liJd6 ! winning easily. 2) 1 �h2 2 lbe3 and again the reply ... .i.g3 is impossible . 3) 1 .i.g3 2 lbc3 ! (for 2 lbe3? .i.f2 ! see below) .i.f2 (the key line is 2 ... �g4 3 lbe4 .i.b8 4 �c8 ! .i.f4 5 liJd6 .i.g3 6 �d7 .i.f2 7 c6 ! .i.b6 8 lbc4 and wins) 3 c6! .i.g3 (Black has forced the pawn to advance, but the cost in time is too high) 4 lbb5 ! (4 lbe4? .i.b8 ! 5 �c8 .i.h2 draws) .i.b8 5 lDc7 �g4 6 �c8 ! .i.a7 7 lbb5 .i.b6 8 liJd6 �f4 9 lDc4 and wins. It is not surprising that this deli cate mechanism breaks down when White is to move. After 1 lbe3 ( 1 lDb2 .i.f2 ! 2 c6 .i.g3 ! is the same, while 1 c6 .i.g3 ! leaves White a tempo down) .i.f2 ! 2 c6 .i.g3 ! 3 lDc4 .i.b8 ! 4 lbd6 �g4 5 lbb5 �f5 6 lDc7 �e5 Black is too fast with his king. ...
...
...
I f the pawn i s further back then it is in the Averbakh drawing zone (see diagrams 1 1 1 and 1 12), so winning chances can only arise in extremely favourable initial positions. Even though there are very few examples of this situation, either practical or theoretical, there are some interest ing positions. The following position is a reciprocal zugzwang with the pawn on c5 (there are 20 altogether).
• • • • . ·�· . • • • • � . � � � � �� � • • • • • • • •• • • • • • •ttJ• • ,,; ,,V
1 86
=I-
Original
Here is one over-the-board exam ple with the pawn on the fourth rank. (187): Black's pawn is a long way back, so White shouldn't have much trouble drawing. The game contin ued 1 ... �e4 2 �g5 �d3 3 .i.a5 (3 .i.al was a safer choice, maintaining the bishop on the long diagonal) lbb3 4 .i.d8 c4 5 .i.e7? (White has
lb+� v i. 137
• • • • • • • • • • • = • • ••• • II . B • m • • • • • • • . B . =I= 0 'Cinneide-Cools European Club Cup, 1993
187
entered dangerous territory, and now be makes a fatal error; 5 i.f6 ! llld4 6 �4 c3 7 i.e7 c2 8 i.a3 ! would have drawn) c3 ! 6 �f5 c2 7 i.a3 llld2 8 .A.c 1 (or 8 �e6 lllc4 9 i.c 1 �e2) �e2 0- 1. There are 8 reciprocal zugzwangs with lll+�c4 v i., 5 with lll+�c3 v .A. and 7 with lll+�c2 v i.. Of these, only two are at all interesting. Here as one of them: (188): It is clear that White to play cannot win. After 1 'Ot>e5 (or 1 d6 i.f7) i.f7 ! 2 c5 i.e8 ! Black stops the pawn in the Aver bakh drawing zone. With Black to play, he may continue: 1) 1 i.h7 2 c5 ! i.e4 3 'Ot>e6 ! (3 �e5? i.g2 4 c6 rl;g7 ! 5 c7 i.h3 ! 6 �d6
...
=I-
188
Original llle3 (making it hard for the bishop to reach the c8-h3 diagonal; 5 c7? i.g4 ! is a draw) i.f3 (5 . . .i.e8 6 c7 ! i.d7 7 �d6 ! i.h3 8 'Ot>e7 !) 6 c7 ! i.b7 7 �e6! �f8 8 �d7 ! �f7 9 lllc4 'Ot>f6 10 llla5 i.a6 1 1 lllb3 wins. We finish with the longest win in the ending of lll+c� v i.:
.t.• • • ;}' �� • • • • .8. • • • • • • • • • • • • • • • • • • B •• Ii +I=
189
Original (189): At least the first move is easy enough. White wins by 1 c7 ! and now:
138 tll + � v �
1 ) 1 �b7 2 �e2 �g2 3 �e3 �c8 (3 . . . �g3 4 tllf7 ! �c8 5 tlld6 �d7 6 tlle4+ �g2 7 �f4 �c8 8 tllc5 �f2 9 �e5 �e3 1 0 �d5 ! �f4 1 1 �d6! heading for d8) 4 tllg6 (4 lllf7 also wins, but more slowly) �fl (4 ... 'iitg 3 5 tlle7 .id? 6 �e4 'it>f2 7 �d5 is easy) 5 tlle7 �d7 6 'it>e4 �e2 (or 6 ... 'iitf2 7 �d5 and Black's king is too far away) 7 �d4 and now we have the reciprocal zugzwang of dia gram 172. 2) 1 �f3+ (Black can put up more of a fight by transferring the bishop to the other diagonal) with two lines: 2a) 2 �d2? �g4 ! 3 �e3 (3 lllf7 �f2! is no better) �fl ! 4 tllg6 �el ! 5 tlle7 �d7 ! 6 �d4 (6 'iPe4 �d2 ! 7 'iPd5 �d3 ! 8 �d6 �h3 9 tlld5 �c8 !) �e2! and the diagram 172 reciprocal zugzwang arises with White to play. 2b) 2 'iite l! �g4 3 tllf7 ! (not 3 tllg6? �g2 ! 4 tlle5 �c8 5 �e2 �g3 6 �e3 'iPh4 ! 7 �f4 �h5 ! drawing) �g2 (or 3 ...�c8 4 �e2 �g2 5 �e3, and now both 5 ... 'iitg3 6 llld6 and 5 ... �fl 6 llld6 �d7 7 tllc4 .tc8 8 tllb6 .ib7 9 tlld5 �c8 10 tlle7 �d7 1 1 'iite4 transpose into line 1) 4 tllh6 ! �c8 (the position after 4 ... �e6 5 �e2 ! is actually easier to win when Black is to move) 5 �e2! �e6 6 tllf7 ! .ic8 7 llld6 Gust as in the note to Black's third move, White em ploys this typical manoeuvre to transfer his knight to e7 with gain of tempo) �d7 8 tllc4 (threat tllb6) �c8 9 tllb6 �b7 1 0 tlld5 (threat tlle7 and then king to b6) �c8 1 1 tlle7 �d7 1 2 �e3 �fl 1 3 �e4 �e2 14 'it>d4 (diagram 172 again; the rest ...
...
is given for completeness) �h3 15 �e5 �d7 16 �d6 �h3 17 tlld5 �c8 1 8 tllb6 �b7 1 9 tlld7 �e3 20 tllc5 �c8 2 1 �e7 'iPd4 22 �d8 ! �h3 23 tlld7 ! and wins.
2.4:
ttJ+d� v i.
White's winning chances are even more remote here than with a c pawn. Generally speaking, in order to have any winning prospects the pawn must be on the sixth or seventh ranks, but even with the pawn on the seventh there are many drawing positions. The following Averbakh drawing zone makes it clear exactly how favourable White's position must be in order to win.
Averbakh, 1958 (190): The corresponding draw ing zone for the pawn on c7 con sisted of just 1 3 squares (diagram 1 67). Here the zone has expanded to 36 squares, and includes the whole board except for the first rank and
i!D+8 v .i. 139
the square a2. Black's bishop is unfa vourably placed on the shorter of the two diagonals leading to d8, but even so the only way to drive it off is to transfer the white king to b7, hardly the quickest plan imaginable. We will consider the following squares for the black king: a2, cl, a3, b2, h2: B�a2: White to play wins by 1 �d5 �a3 (after l . .. �b3 2 i!Dd4+ ! �b4 3 i!De6 ! .i.a5 4 �c6 or 1 . .. �b2 2 �c4, White wins more easily) 2 �c4 ! (but not 2 �c5? �a4 ! , and White falls into a reciprocal zug zwang; the position is drawn after 3 i!Dd4 .i.d8 ! 4 i!De6 .i.h4) �a4 3 �c5 ! (now Black is to move in a re ciprocal zugzwang, and must allow White's king in to b5: 3 ...�a3 4 �b5 �b2 5 �a6 �c3 6 �b7 winning. B�cl : Now 1 �d5? doesn't win because of l ...�d2! 2 �c4 �e3 ! 3 �b5 �e4 ! 4 �a6 �d5 ! 5 �b7 �d6 ! and Black scrapes a draw. How ever, White can win by a different method: 1 �e7 .i.b6 (or l . .. �d2 2 eid4 �e3 3 i!De6! and wins after 3 ....i.g3 4 �e8 .i.h4 5 i!Df8 �d4 6 l0g6 ! or 3 ... .i.a5 4 �d6! .i.b4+ 5 l0c5 ! .i.a5 6 i!Db7 .i.b6 7 �c6) 2 l0d8 .i.c5+ 3 �f7 .i.f2 4 i!Db7 .i.h4 5 �e8 �b2 6 i!LJd6 followed by i!Df5e7. B�a3: Black draws after 1 �d5 .i.b6! (but not l ...�a4? 2 �c5 ! , win ning as above) 2 i!Dd4 .i.d8 ! (again not 2 ... �a4 ?, this time because of 3 l0e6 ! �b5 4 �d6 !) and White can not make progress. B�b2: This position is a draw be cause neither winning plan comes off. 1 �e7 �c3 2 i!Dd8 �c4 3 i!De6
.i.a5 ! 4 �d6 .i.b4+! 5 �c6 (White cannot play i!Dc5) .i.e7 ! leads to an easy draw, while 1 �d5 �c3 ! 2 �c5 �d3 ! 3 �b5 �e4 ! 4 �a6 �d5 ! 5 �b7 �d6! is a draw as in the B�cl analysis. B�h2: After l �e7 (or 1 �d5 �g3 2 �c5 �f4 3 �b5 �f5 and Black's king is too close) �g3 2 i!Dd4 �g4 3 i!De6 .i.a5 ! 4 �d6 �f5 ! 5 i!Dc7 .i.b4+! White's king has no ac cess to e6, so 6 ....i.e7 cannot be pre vented. Once again, Halberstadt produced the most detailed analysis of this ending and he discovered several positions of reciprocal zugzwang. The following four positions repre sent a survey of his work.
191
=IHalberstadt, 1954 Curiosites Tactiques
(191): With Black to play, White wins by l ...�a4 ( l . . . .i.h4 2 i!Dc6, l . . . .i.b6 2 i!De6 and l . ..�b2 2 i!LJc6 .i.c7 3 �c5 �c3 4 �b5, heading for b7, are simpler) 2 i!LJc6! .i.b6 (2 ....i.c7 3 �c5 ! transposes) 3 �c4!
140 lLJ+� v .i.
(this is also reciprocal zugzwang) .i.c7 (there is no other move) 4 �c5 ! (another reciprocal zugzwang, after which Black must allow White's king in to b5) �b3 5 �b5 and the king reaches b7. When White moves first, l tlJc6 ( 1 �c5 .i.h4 2 lllc6 �a4 ! 3 �d6 �b5 ! 4 llle7 .i.g3+ ! 5 �e6 .i.c7 ! is an easy draw) .i.b6 ! is the critical line, reaching yet another recipro cal zugzwang. White to play draws, for example 2 llld4 .i.d8 ! , 2 �d6 �a4 ! 3 llld4 .i.d8 ! , or 2 �c4 �a4 ! , the last being one of the reciprocal zugzwangs mentioned above, but this time with White to play. In the analysis of this position we have already come across four re ciprocal zugzwangs: Zl) �c5, lllc6 v �a4, .i.c7 Z2) 'it>c4, lllc6 v �a4, .i.b6 Z3) �d5, lLlc6 v �a3 .i.b6 Z4) 'it>d5, llld4 v �a3, .i.d8 •
Reciprocal zugzwangs can also arise when White's knight starts on e6. (192): When Black moves first, White wins after l . ..�a4 (l ...'iii>b5 2 �d5 ! is another reciprocal zug zwang; White wins immediately af ter 2 ....i.b6 3 �d6 !) 2 �c4 ! (not 2 �c5? .i.b4+ ! 3 �d5 .i.e7 ! 4 llld4 .i.h4 5 �e6 .i.d8 ! 6 lllc6 .i.c7 7 �d5 �b5 ! drawing, nor 2 �d5? �b5 ! reaching the reciprocal zugzwang with White to move and drawing af ter 3 �d6 .i.b4+ ! 4 �e5 .i.e7 !) .i.b6 3 �d5 ! (threat �c6) �b5 4 �d6 ! and there is no defence to the threat of tlJc7.
1 92
=IHalberstadt, 1954 Curiosites Tactiques
With White to play, 1 We5 (or l �d5 'iii>b5 ! with the familiar recipro cal zugzwang) �c4 ! (of course not l . ..'it>b5? 2 'iii>d5 !) 2 'it>d6 .i.b4+! draws. Now we can add two more recip rocal zugzwangs to the above list: Z5) 'iii>d4, llle6 v 'it>b4, .i.a5 Z6) Wd5, llle6 v �b5, .i.a5 Halberstadt published the fol lowing two positions as a pair, and thereby won first prize in a study competition. (193): Black to play draws by l . ..�a4 2 lllc6 .i.c7+ 3 �d5 Wb5 !, so suppose White moves first. He can try: 1 ) 1 lllc6? .i.c7+ ! 2 �d4 (2 �d5 .i. b6 ! is Z3 above) 'it>b3 ! 3 �d5 'it>c3 ! 4 'it>c5 Wd3 ! 5 'it>b5 'it>e4! 6 �a6 Wd5 ! drawing as in diagram 190. 2) 1 �d6? �a4 ! 2 lllc 6 .i.b6 ! leaves White's king too far away from d4.
/Li+l!,
v
i. 141
- - - m - LS · m • • • • . - . - � . - - - - - . • • • • • • • •
• . � � � � � B B LS B B - . . . � . . . � - -�- � . · � · � � � - . - . • • • •
3) 1 �dS! gives rise to a recipro cal zugzwang. The lines 1...�a4 2 'Lic6 ! i.b6 (2 ...i.c7 3 �c5 ! is Zl) 3 �c4 (Z2), 1 ...i.b6 2 �c4 �a4 3 'Lic6 ! (Z2), and 1 ...'iti>b3 2 ILic6 ! i.b6 3 'Lid4+ ! �b4 4 'Lie6 �b5 5 �d6 ! are all fairly straightforward. Black's best defence is 1 ...i.aS, but White wins in any case: 2 'Lib5+! (2 'Lic6? i.b6 ! is Z3) �b4 (2 . . .�a4 3 'Lid4 ! i.d8 4 /Lic6 ! transposes into 1...'ifta4 above, while 2 ...'iftb3 3 'Lid4+ ! �a4 is the main line) 3 'Lid4 ! 'iti>a4 (or 3 ...i.b6 4 'Lie6 'iti>b5 5 �d6!) 4 'Lic6 ! i.b6 5 'iftc4 ! and Black falls into Z2. (194): Black to play draws easily by l . ..�b4 2 'iftd5 (2 'Lic6+ �c5 !) �b5, so assume White moves first. There are two main possibilities: 1 ) 1 �dS? i.c7 ! 2 'iti>c5 (2 'Lic6 i.b6! is Z3 with White to play) �b3 ! 3 'iftb5 (after 3 'Lic6 �a4 ! we have Z l , but now 4 'Lic6 is a real threat) i.d8 ! 4 'Lic6 i.h4 5 �c5 �a4 ! 6 �d6 �b5 ! 7 'Lie7 i.g3+! and Black draws.
2) 1 /Lic6! i.c7 (after l . . .i.b6 2 �d5 ! we reach Z3) 2 'Lid4 and now: 2a) 2...i.d8 3 �d5 ! is Z4. 2b) 2...�b4 3 'Lie6! i.a5 (3 ... i.b6 4 �d5 �b5 5 �d6) 4 'ifi>d4 ! is Z5. 2c) 2...�a4 3 'Lie6 ! i.a5 trans poses to line 2d. 2d) 2...i.aS 3 'Lie6 'ifta4 (3 ...'iftb4 4 �d4 ! is Z5) 4 �e5 ! (not 4 �d5? �b5 !, and we have Z6 with White to play) �b5 (or 4 ... i.c3+ 5 'Lid4 i.a5 6 'iftd5 i.d8 7 'Lic6 ! with a familiar win) 5 �d5 ! (Z6) i.b6 6 �d6 ! and wins. Halberstadt's systematic investi gation of these positions is very im pressive and his first prize was entirely justified. Many other posi tions eventually reduce to one of the Halberstadt zugzwangs: (195): White to move wins by 1 'Lif6 ! (not 1 'Lid6? i.d7 2 �f2 �h2! 3 �f3 �h3 4 �f4 �h4 with an easy draw) i.g6 (after l . . .i.a4 White's king can gain a tempo by attacking the bishop, so 2 �d2 �g2 3 �c3
193
+/= Halberstadt, 1936 1 st Pr., La Strategie
� -
194
+/= Halberstadt, 1936 lst Pr., La Strategie
142 lLJ+L!. v .t
+I=
195
• = • • . ·�· . • • • • • • • • • • • • • • • • ttJ• - • • • • ••• +I=
1 96
Kosek
Original
'ii.?f3 4 'ii.?b4 .tc6 5 'ii.?c5 .ta4 6 'ii.?d6 wins easily; l . . ..tf7 2 'ii.?f2 ! 'ii.?h2 3 'ii.?f3 ! 'ii.?h3 4 'ii.?f4 ! 'ii.?h4 5 'ii.?f5 ! is Zl above) 2 'ii.? fl ! (not 2 'iPf2? 'ii.?h 2! 3 'ii.? f3 'ii.?h 3 ! 4 'ii.?f4 'ii.?h4 ! , and Z2 arises with the wrong player to move) 'ii.?h2 3 'ii.?f2 ! 'ii.?h3 4 'iPf3 ! 'ii.?h4 5 'ii.?f4 ! and Black has fallen into Z2. Black to play can draw by l ...'iitg2 2 'ii.?e2 'ii.?h3 3 'ii.?f3 'ii.?h4 4 'ii.?f4 �h5 5 'iPf5 .td7+. Even when reciprocal zugzwang isn't involved, the winning proce dure can require great accuracy. (196): 1 'ii.?e7 ! (after 1 'ii.?e8? .ta5 ! 2 lllc 1 'ii.?f2 3 lllb 3 .tb6 4 llld4 'ii.?e 3 5 llle 6 'ii.?e4 Black is a tempo up over the main line and draws easily; 1 'ii.?c 8? .tg5 ! leads nowhere) and now: 1 ) 1...'ii.?g2 2 lllc3 ! .tg5+ 3 'ii.?e 8!, followed by llld5-e7, winning. 2) 1 .tf4 2 lllc 3 .tc7 3 lllb5 .tb6 4 llld6 'ii.?e2 5 lllb7 ! (threaten ing 'ii.?d6-c6) .tf2 6 'ii.?f7 .th4 (or 6 . . . .tb6 7 'ii.?e6-d6-c6) 7 'ii.?e 8, fol lowed by playing the knight to e7.
3) 1.. .taS 2 lllc l ! (not 2 lllc 3? 'i&i>f2) and now: 3a) 2 .tb4+ 3 'i&i>e8 .ta5 4 lllb3 .tb6 5 llld4 'ii.?g2 6 'iite7 .ta5 7 'ii.?d6 .td8 8 'i&i>c5 ! (a useful idea; the im mediate 8 lllc 6 is dubious because of 8 ... .tb6; indeed, White can only win by playing llld4 again) .th4 (otherwise White plays lllc6, meet ing ....tc7 by 'ii.?b5-a6-b7 and ....th4 by 'iitd5-e6 and llle7) 9 lllc6 �h3 10 'iitd5, with 'ii.?e6 and lbe7 to come. 3b) 2 'ii.?f2 3 lllb3 ! .tb6 4 llld 4! (intending llle6, followed by 'ii.?d6 and lllc7) .tc5+ (4 ... 'iite 3 5 llle6 ! .ta5 6 �d6 !) 5 'ii.?e 8 .tb6 6 llle6 ! .ta5 7 'iite7 ! (the same plan still works) .tb4+ 8 �d8 .tc3 9 lllc7 'iitf3 10 'i&i>e8 .tf6 1 1 llld5 ! and wins. Black to play draws by l . . .�e2 2 r/;e7 .ta5 ! 3 lllc l + 'ii.?e 3 4 lllb3 .tb6 and the knight cannot move to d4. Once again, I cannot avoid quot ing one of Halberstadt's positions. (197): Black to play can draw by 1 ...'ii.?b7 2 d7 .td2, intending ... .ta5. White to play can win by 1 d7 ! (after
•••
.
•.•
•••
lb+� v i. 143
1 97
+I= Halberstadt, 1939 Chess
l ..tf6? i.e3 ! 2 lllc4 i.f4 3 d7 i.c7, Black has transferred his bishop to a good defensive position and draws easily) i.e3 ( 1 ...i.g5+ 2 ..te8 ! wins by lllg 6-e7, while after l ...i.d2 2 �e8 Black cannot play ... i.a5, and so loses the same way) 2 lllc4 ! (not 2 �c6+? ..tb7 ! 3 llld4 i.d2 ! 4 lllb3 i.g5+ 5 �e8 ..tc6 ! 6 llld2 ..td5 7 �f3 i.f6 ! , followed by ...�e6, and Black is safe) i.f4 3 llld6 ! i.d2 (3 ...i.e3 4 �e8 wins as the bishop cannot move to b6) 4 lllb 5+! �b7 (4 ... ..ti>a6 5 lllc7+ �b7 6 �e8 i.g5 7 � and 8 llle7) 5 lllc3 ! (a spectacu lar move) �c6 (5 ...i.g5+ 6 �e8 ! �c6 7 llle4 ! transposes) 6 ..te8 ! (6 �e4? i.a5) i.g5 7 llle4 ! i.h4 8 �g3 ! (now Black is too late with his king) ..td5 9 lllf5 and the knight reaches e7. (198): White to play wins by 1 �c3 ! (this threatens 2 �d7 ; l lllb6? �e8 2 llld7 i.a4 ! is only a draw) �e8 (or 1 . . .i.e6 2 lllb5 ! i.d5 3 �d4 ! ..te8 4 lllf5 !) 2 llle4 ! ..tf7
+I= 1 98 Norsia, 1923 (version) 2nd HM, L'Italia Schacchistica (2 ...i.a4 3 �e6 ! i.b3+ 4 ..tf6 ! ) 3 lllc5 ! (3 �d7? ..tg7 ! 4 llld 6 i.dl followed by ... i.h5 draws, but not 4 ...i.a4+?, which loses after 5 �d8 ! �f6 6 lllb7) i.c4 (3 ... i.d5 4 �d7 ! i.c6+ 5 �d8 ! i.e8 6 llld7 is also winning for White) 4 �d7 ! �f6 (4 ...i.b5+ 5 �d8 ! and 4 ...�g7 5 llle6+ ! ..tf6 6 �d8 ! i.b5 7 lllc5 are winning) 5 llle4+ ! �g7 6 llld6! i.b3 7 �d8 i.a4 8 lllc8 �f6 9 lllb6 !, fol lowed by llld7, and White wins. Norsia's original version had the pawn on e6 and Black's king on f8. The intended solution started 1 e7+ �f7, but White has a faster alterna tive win by 1 �d7. The following subtle position, which depends on a reciprocal zug zwang, finishes the section with a pawn on d7. (199): The knight and bishop are both reluctant to commit themselves. The reason is that White would like the knight and the bishop to be on opposite sides of the pawn; Black
144 lb+� v .t
+/-
1 99
Original would like them to be on the same side. If we ignore the precise white king position for the moment, then ... .te7 allows lbc6, followed by playing the king to e6 and lDe7, but if White plays lDc6 himself, Black replies with . . ..tc7 and now White can only win by playing his king to b7, which is usually too slow. So Black to move doesn' t want to play . . . .tc7 or ... .te7 . Playing ....tb8, . . ..tf8, ....tb4, ... .tf4 or ... .th2 al lows the pawn to promote at once, so that only leaves ....tc5, ....te5 and ....tg3. The first two are bad choices because White can gain two tempi by attacking the bishop, for example after ... .tc5 White plays �e4-d5, forcing either ... .tb6, met by lDe6 followed by �c6, or ... .te7, met by l0c6 followed by �e6 and l0e7. This leaves . . ..tg3 as Black's best bishop move. Much of the play re volves around whether Black can play ....tg3 in safety. White to play wins as follows: 1 �e3!
As explained above, moving the knight is bad, e.g. 1 lDb7/e6? .te7 !, or 1 lDc6? .tc7 ! . However, the most obvious move 1 �e4? fails for the subtle reason that 1 . ..�f2 ! leaves White to move in a reciprocal zug zwang, e.g. 2 �d5 (2 �f5 �e3 ! 3 �e6 .tg3 and 2 �d4 .tg3 ! 3 �d5 �e3 ! are no better) .ig3 ! (2 ....te7? 3 lDc6! .th4 4 �e6) 3 lDc6 (3 �e6 doesn't threaten anything serious, so Black can reply 3 . . .�e3) .ic7 ! and the black king is in the Averbakh drawing zone (diagram 1 90). �g2 1 ... After 1 . .. �fl White wins more easily: 2 �d4 �f2 (or 2 ...�e2 3 <.Pd5 ! .tg3 4 lDc6 .ic7 and now White gains a tempo with his knight, winning after 5 lDd4+ ! �d3 6 lDe6 ! .ta5 7 �c6) 3 <.Pe4 ! and the recipro cal zugzwang arrives with Black to move (see the next note for detailed analysis). 2 �d4! Certainly not 2 'otte4? �f2! . �h3 2 ... This is the most resilient defence, but it is interesting to see why the po sition after 2 ... <.Pt2 (2 ... �g3 3 �d5 ! .ie7 4 lDc6 ! wins) 3 �e4 ! is indeed zugzwang: 1 ) 3 �e2 (this allows a tempo gaining knight check later on) 4 �d5 .tg3 5 lDc6 .tc7 6 lDd4+! �d3 7 l0e6! .ta5 8
•..
tLl+8 v .t 145
3) 3 .tcS (this is still bad) 4 �d5 .ie7 5 tLlc6 !, followed by We6 and tbe7 . 4) 3 .ig3 4 'it>f5 ! (threatening �6. so Black must waste time with his bishop) .id6 5 �f6 (tbe6 is again the threat) .ic7 6 �e7 (White has gained enough time to win) 'ifi>e3 7 eie6 ! .ia5 8 'ifi>d6! .ib4+ 9 tLlc5 ! .ia5 10 tLlb7 ! with the usual domi nation of the bishop. 3 'ifi>dS! This wins now that Black's king is misplaced on h3. .ig3 3 ... .ih4 4 tLlf7! White wins easily after 4 ... .ic7 5 �c6 .ia5 6 tLlg5+ 'iti>g4 7 tLle6! Wf5 8 tLlc7 ! . S tLleS! White employs a typical ma noeuvre to gain time. The threat is eic6, followed by We6 and tLle7, so Black's reply is forced. .id8 s 6 'it>cS Now the threat is 7 tLlc6 .ic7 8 �b5, followed by �a6-b7 . 6 . .ie7+ Or 6 . . . .ic7 7 �b5 .id8 8 tLlc6 .ih4 9 �c5 !, intending �d6 and eie?. 7 �b6 .id8+ After 7 . . .�g3 8 tLlc6 .ig5 9 �c5 White again wins by blocking the h4-d8 diagonal. �g3 8 �b7 Now White wins easily: 9 tLlc6 ! .ih4 10 �c7 ! .ig5 1 1 �d6! .if4+ 12 �c5 .ie3+ 1 3 'iti>c4 .ib6 (or 13 . . . .ig5 14 'iti>d5) 14 �b5 .ic7 15 �a6, followed by �b7. .••
.••
...
Curiously, it doesn't help Black if he moves first. After 1 . ..�fl 2 �e3 ! , the only new possibility is 2 . . .'iti>e l , but then White can win by 3 �d3 ! (the point of this move is that one of White's main plans is to play tLlc6, forcing . . . .ic7, and then transfer his king to b7; this takes one move less from d3 than from e3, so the king is heading in the right direction, but at the same time Black's king is pinned down to the back rank; Black draws after 3 �d4? �d2! 4 'ifi>d5 .ig3 !) �d l (or 3 . . . �f2 4 �e4 !, as in the note to Black's second move above) 4 tLlc6 .ic7 5 �c4 and wins by run ning towards b7. Other first moves are no better, for example 1 . . .'ifi>h2 2 �e4 ! . There are 59 positions o frecipro cal zugzwang with tLl+8d7 v .i. Now we move on the situation with the pawn on d6. This is the last interesting case, because with the pawn further back White can only win in special situations.
..
200
=I-
146 tb+� v .t
(200): This is a curious position of reciprocal zugzwang. Black to move loses after l ....td8 2 tbe5 (this is a second reciprocal zugzwang) 2 . . . �h7 (or 2 . . . .tg5 3 tbg4, threat ening 4 tbf6, and transposing to the main line after 3 . . . .td8 4 tbf6 ! ) 3 tbg4 ! (again reciprocal zugzwang) �h8 4 �f6 ! and the fourth recipro cal zugzwang proves fatal. With White to play the position is drawn since 1 tbe5 (1 tbxg5 is stale mate) �h7 2 tbg4 .td8 ! 3 tDf6+ �h6 offers no winning prospects. In 1 85 1 , Kling and Horwitz pub lished a very similar position with White's knight on gl and Black's bishop on h4. Their intended solu tion was 1 �f3 .td8 and then as above. Many other authors repro duced this position, but in 1 952 Cheron pointed out the drawing line 1 �f3 .tg5 ! . Of course, by simply moving the bishop from h4 to g5, the Kling and Horwitz solution can be resurrected, but the above version emphasises the key point - that the diagram position is reciprocal zug zwang. The next two positions involve fairly straightforward promotion battles. (201): Black to move can draw easily by l ....tb5. White to play wins by 1 �c7 ! .tb7 (or l ....td3 2 �e7 ! .tf5 3 �6!) 2 �d7 ! (perhaps slightly surprising, but White must prevent the bishop occupying the a4-e8 diagonal) and now: 1) 2. .tf3 3 �e6! .te2 (3 . . ..td5 4 �d4) 4 �c6 and wins. ..
+I= Mugnos, 1 949 Revista Metropolitana
201
2) 2 ... �d2 3 tDe6! .ta6 (3 . . . .td5 4 tbd4 .tg2 5 �e7 .th3 6 tbe6 !) 4 tbd4 (the key square for the knight, dominating Black's bishop) .tfl 5 �e7 .th3 6 �e6 ! winning. 3) 2 ...te4 3 �e6! .tc2 4 �d4 ! .ta4+ 5 �c6! .tc2 6 �e6! .tb3+ 7 �e7 wins. It is worth emphasising how un usual such a win is. Even though Black's bishop is poorly placed, i.e. not controlling d7, White can only win through a combination of spe cial circumstances. If Black's king starts on e l , then he draws by 1 0£7 .tb7 ! 2 �d7 .tf3 3 �e6 .tdl ! 4 �d4 �d2 and the bishop is ready to move to either diagonal. If we return the king to d 1 , move the bishop to g6 and the knight to c8, then Black can draw this reflected position by 1 �e7 .th5 ! (making use of the extra space) 2 �f5 .te2 ! 3 �d4 .tg4 ! 4 �e7 .tc8 !. (202): Only the two kings can move, but even their movements are .
lLJ+� v .t 147
+I=
202
Szulc, 1942 Sth HM, Sach restricted. White cannot move his king to the bl-h7 diagonal, because Black extracts his bishop with check, while Black cannot allow a knight check, followed by e7. White to play wins by 1 �d2 ! (not 1 �c2? .th7+! 2 �c3 .tf5 3 e7 .td7 !, nor 1 �d l ? �b3 2 �d2 �b4 3 �e3 �c5 4 'iitf4 �d5 5 �f5 .th7+! 6 'iitf6 .tc2, with � draw in both cases) �b3 (1 ....th7 2 c7) 2 �e3 ! �c3 (2 ...�c4 3 lt:Je5+) 3 �f4 ! �d4 4 �g5 ! 'iitd5 5 �f6 ! 'iitc6 6 ltJe5+ �c7 7 e7 and wins. Black to play draws after 1 . . .'iitb 3 2 �d2 'iitb4 3 �e3 �c5 4 'iitd4 �d5, as in the analysis of 1 �dl ? above. The next position depends on re ciprocal zugzwang. (203): Black to play draws by l .. . .tc8+, when the bishop controls d7. White to play wins using a king triangulation: 1 ltJe4 ! (not 1 'iitc 7? .1c8 ! and the bishop arrives on the c8-h3 diagonal) .ta8 ( l . . . .tc6+ 2 �c7 ! makes it easy, because this is a position of reciprocal zugzwang;
+I= Dimentberg, 1949 (version by Halberstadt, 1954) Special HM, Shakhmaty v SSSR 203
l . . . .ta6 loses to 2 'iitc 7! .tb5 3 ltJc3+ !) 2 �c8! (certainly not 2 'iitc7? .tc6 and White is in zugzwang) .tc6 (or 2 . . .�c6 3 d7 ! .tb7+ 4 'iitd 8!) 3 'iitc7 ! and Black is finished. When a win exists at all with ltJ+�d6 v .t, the winning lines are normally not very complex. It is therefore surprising that the Ruma nian composer Dobrescu made an error in the following position. (204): This was intended to be 'White to play and draw', but Black is winning. I will follow the com poser's main line: 1 .ta4 lDc3 ! 2 .tb3 (or 2 .td7 d2 ! 3 .tg4 'iitd4) 'iit b4 (this wins, but it was much easier to play 2 . . . d2 3 �g4 'iitb 4! 4 .tc2 'iita3 ! 5 �f3 �b2! and White is too slow) 3 .te6 d2! 4 .tg4 �c4 (not 4 . . . 'iitc5 ? 5 �g5 ! �d4 6 'iitf4 ! lDd5+ 7 'iit g 3) 5 .te6+ (5 �g5 'iitd 3 ! and 5 'iith4 ..t>d4 6 �g3 'iite 3 ! are also lost) 'iitd 3? (the composer only analysed this move, but ironically it is the only
148 tD+.0. v .i.
204
-/+
=I+
205
Dobrescu, 1979 5th Pr., Revista Romana de Sah
Dobrescu, 1982 Sachove Umeni
legal move not to win; the simplest line is 5 . . .tt:xl5 6 .i.g4 tDf6+) 6 .i.b3 ! (now the position is drawn) lDd5 7 .i.d 1 ! (not 7 .i.a4? tDe3 ! 8 .i.b5+ tDc4 9 .i.a4 lDb2) tDe3 8 .i.f3 ! lDd5 9 .i.d 1 ! (9 'iith 4? tDc3 ! 1 0 .i.h5 tDe2 1 1 .i.g6+ 'iitc 3 !) tDc3 10 .i.b3 ! and Black cannot make progress. The following study is also un sound, but this time Dobrescu's mis take was more understandable, since the winning move is counter-intui tive. (205): The composer thought that this position is drawn, even with Black to play, but he was wrong. We will follow his main line: l . ..d3 ! 2 .i.d7 d2 ! 3 .i.c6+ (after 3 .i.a4 'it>d4, Black wins by . . . 'it>c3 and . . . tDc2) and now: 1 ) 3 'iti>d3? (3 . . . lDd5? 4 .i.a4 ! is also drawn) 4 .i.f3 ! (4 .i.b5+? tDc4 !) lDd5 and now we have the same draw as in the previous position. 2) 3 'iite5! (a very geometrical move; Black must retain the option
of going to b2 after .i.a4, but must also be able to meet .i.f3 by ...'it>f4 and . . .'it>g3) 4 .i.f3 (4 .i.a4 'iitd4! ) 'iitf4 ! 5 .i.e2 'iitg3 ! 6 'iit g5 'iti> f2 ! 7 .i.h5 'it>e 1 ! 8 'it>f4 lDd5+ ! 9 'it>e4 (otherwise . . . tDc3-e2 wins) lDf6+! winning for Black. White to play draws after 1 .i.d7 d3 2 'iti>g5 d2 3 .i.a4 'it>d3 4 'iitf4 tDc2 5 .i.b5+, followed by .i.e2.
.••
...
+I=
206
Original
lD+L!.
(206): White to play wins with an attractive manoeuvre: l lDd5 ! i.h6 2 �d7 ! (not 2 d7? i.g5 ! 3 �d6 i.dS ! 4 '0!7 �f2 5 lDc6 i.b6 ! 6 e4 6 �cS! i.a5 7 lDc6! i.b6 8 �b7 ! lose more quickly) 3
v
i. 149
+/-
207
Original (207): Black is to play. The main line continues l . . .i.c l 2 lDe4 ! i.f4 (2 . . . i.e3 3 d7 i.b6 4 lDg5 i.dS is a transposition) 3 d7 ! i.c7 4 lDg5 (threatening lDe6, and then king to b5) i.dS 5 lDe6! i.h4 (Black must move as far away as possible, since after 5 . . . i.f6 6 b5) 6 lDd4 'iti>g2 (6 . . .i.dS 7 lDc6 i.c7 S d6! i.f4+ 1 6 �c5) 1 3
150 tlJ+.0,
v
�
The following table gives the number of reciprocal zugzwangs for various pawn positions.
1 5 �d6 i.g3+ 16 cjric5 i.f2+ 1 7 cjric4 i.h4 1 8 cjrid5 cjrig4 1 9 �e6 and wins in two more moves. a
b
c
d
7
66
54
60
59
'
46
95
57
31
s
36
27
20
11
4
13
9
8
6
3
4
5
5
6
2
6
5
7
4
This material situation can be di vided into two cases: opposite-col oured bishops and same-coloured bishops. The first case is essentially trivial, because once the defender's bishop can permanently guard a square which the pawn must cross, the game is drawn. Whether or not the defender can achieve this be comes clear in three or four moves at the most. Sometimes a tactical de fence is required, as in the following example:
The only interesting idea i s that of stalemate.
=I= 209 Babitch, 1951 (end of study) 5th Pr., Shakhmaty
I
208
= =
Berger, 1922 (208): In Basic Chess Endings, Fine incorrectly gave this position as a win for White, but the position is drawn after 1 a6 .i.f5 ! 2 �e3 (or 2 �f3 .i.d3 3 a7 .i.c4 ! , but of course not 2 . . . �d5? 3 a7 ! .i.e4+ 4 �e3 ! ) .th3 ! 3 �f3 .tn ! 4 a7 .tc4 ! 5 a8'ii' .td5+ ! .
(209): White draws, even when Black moves first: 1 . . .�f5 ( 1 . . .b2 2 �xe5 ! b I 'ii' 3 .i.e4+ ! and 1 . . . �f6 2 'it>c4 b2 3 .i.e4 are no better) 2 .i.c6! b2 3 .i.a4 ! b 1 'ii' (Black has no better promotion) 4 .i.c2+ ! 'ii'xc2 is stale mate. The remainder of the chapter deals with the case of same-coloured bishops. White's winning chances are much greater than with .!Ll+L!. against .t. The main reason is that Black's bishop, by itself, can never hold up the pawn for long. Suppose, for example, that White has a pawn on b4 and there are light-coloured bishops. If White's king is on c5 and his bishop on d5, then he can always
152 .t+l!.
v
.t
drive away the defending bishop. If the bishop is covering b5 on the a6f l diagonal, then White plays .tc4, while if Black's bishop is on the a4e8 diagonal, White plays .tc6. In both cases White secures the ad vance of his pawn. So the result of al most every position depends, in part, on the proximity of the defender's king. If the defender's king establishes itself in front of the pawn, then of course the position is drawn, so we will only be examining situations in which this does not happen, either because his king is too far away, or because the attacker can keep the enemy king away from the pawn. Given the choice of approaching the pawn from the side or from behind, it is almost always better for the de fending king to approach the pawn from behind, and the bulk of the analysis will deal with this situation. We will make the usual division according to the file occupied by the pawn. 3 . 1 : .t+al!. v .t 3.2: .t+bl!. v .t 3.3: .t+cl!. v .t 3.4: .t+dl!. v .t
3.1:
1 52 166 181 195
i.+� v i.
As in the first two chapters, it is normally better for the attacker to have an edge pawn. When the pawn is on a7, with light-squared bishops, the situation is usually quite clear. If White's king
is on b6 or c7, then he can play his bishop round to b7, driving the en emy bishop off the long diagonal, and then promote the pawn. If Black can occupy c7 or b6 with his king, and position his bishop on the long diagonal, then White cannot drive the bishop away and the position is drawn. Subtlety is required only very rarely:
=I+ Koranyi, 1967 lst HM, Prokes Mem Tny.
210
(210): Black to play wins by 1 ... .th4 and 2 ... .tf6, so suppose that White is to play. He draws by 1 'iti>d6! (not 1 �d7? �f5, followed by ...�e4 and ... .td4) .th4 ( 1 . ..�f5 2 �d5 ! .th4 3 .tal ! is similar to the main line) 2 �e6 ! .tg5 (intending ....th6g7) 3 .tal ! (the only square for the bishop; by luring the enemy bishop to al, White gains sufficient time for his king to attack the pawn) .th6 4 �d5 ! .tg7 5 �c4 ! .txal 6 �b3 ! and White wins the pawn. There are no reciprocal zug zwangs with .t+l!.a7 v .t.
.i.+� v .i. 153
Now suppose that the pawn is on a6. White wins when his king is on b7 and Black's bishop is holding the pawn up on the gl-a7 diagonal:
• • • • -�· . . �- . . . ••• • • • • • • • • • • • g • • • • • • 21 1
+!-
Centurini, 1856
(211): It doesn't make any differ ence who moves first. White wins with a typical manoeuvre: 1 .i.f4 ..ld4 2 .i.b8 .i.c5 3 .i.a7 .i.d6 4 .i.gl ..lb8 5 .i.h2 and the pawn promotes. We will meet Centurini's name many times in this chapter. He con ducted the first systematic analysis of .i.+� v .i., and uncovered most of lbe basic principles. The above situation, in which the kings face each other vertically with a separation of two squares, and the pawn is one square diagonally away from both kings, is very common in this ending. We will follow Aver bakh and call it 'vertical opposition' . It is important to note exactly why lbe above line led to a win. There are two diagonals passing through the square in front of the pawn (gl -a7 and b8-a7 in the above case). Black's
bishop is best posted on the longer of the two diagonals, but he cannot maintain it there. White can always play his bishop to the square in front of the pawn, and then Black's bishop has to switch to the shorter diagonal . Whether Black can defend depends on the length of the shorter diagonal. If it is just two squares long, then White can win by a direct attack on the enemy bishop. If it is three squares long, then in almost every case White can win by zugzwang. If it is at least four squares long, then Black can draw. Using this argument, Averbakh was able to construct the drawing zone for the case of vertical opposi tion:
Averbakh, 1954 (212): This is the drawing zone for White's pawn, with dark-squared bishops (the zone for light-squared bishops is simply the mirror image). White wins if his pawn is further advanced, with two exceptions. The first arises when the pawn is on a7,
154 .t+l!. v .t
with the white king on b8 and the black king on b6. Black draws if his bishop is on the long diagonal, even though the pawn is in the winning zone. The reason is that it is impossi ble for White to play his bishop to a8 and so Black can maintain his bishop on the long diagonal. The second ex ception is diagram 257. An interesting tactical point arises in the following position.
213
=I=
Kopelovic, 1971 1 st Comm., Israel Ring Tny (213): White can draw, even when Black moves first: l . . .�e4 (White's threat was �f3 and �g4, so this move is forced) 2 �e2 ! (but not 2 �el ? �f3 ! 3 �fl .td6 4 .te3 h2 ! and wins) .tg3 (after 2 ....td6 3 �f2 ! Black is unable to make progress; White's king simply oscillates be tween e2 and f2; it is interesting to note that Black would win if he could transfer his king to g4, but this isn't possible) 2 �fl ! (now White threatens to move his bishop and then play �gl ; reaching h l draws
even if White loses his bishop in the meantime) �f3 3 .tf2 (other moves on the same diagonal also draw) h2 (3 ... .th2 4 .ta7) 4 .tgl ! h 1 'ir' (4 ...h 1 l:t is the same) with stalemate. There are six reciprocal zug zwangs with .t+l!.a6 v .t. The fol lowing position is a gentle (well. fairly gentle) introduction to the re ciprocal zugzwangs in this chapter.
+I= 214 Grigoriev, 1931 (version) Shakhmatny Listok
(214): Suppose that White moves first. The key point is that White wins if his king reaches b7; other wise Black draws. The alternatives are: 1 ) 1 �g4? ( 1 �h5? �g7 !) �g6! (now White is in a reciprocal zug zwang, which we will call Z 1 ; he has no constructive king move, because 2 �f4 is met by 2 . . ..tb8+ !) 2 .th4 (there is nothing better) .tb8 ! (sto� ping the threat of �f4, and reaching a second reciprocal zugzwang, Z2) and now:
i. +� v i. 155
l a) 3 .tgJ (White has no pass move) .ta7 ! 4 i.e5 �f7 5 �f5 �e7 and Black's king reaches d7. l b) 3 i.d8 �f7 ! 4 �f5 �e8 ! 5 J.c7 �d7 ! 6 i.xb8 �c6 ! draws. l e) 3 i.e7 �f7! 4 i.d6 i.a7 ! 5 �f5 �e8 6 �e6 �d8 ! 7 �d5 �d7 8 J.e5 .t g l and White cannot make progress. 2) 1 �g3! �g6 (after 1 .. .�g7 2 �4 ! �g6 3 i.h4 White's king has a free run to b7) 2 �g4 ! (now Black is in zugzwang) with the variations: 2a) 2 .tgl 3 �f4 i.a7 (Black no longer has a useful check on the b8-h2 diagonal, since 3 ... .th2+ is met by 4 �e4) 4 i.h4 �f7 5 �e5 �e8 6 �d6 i.gl 7 �c7 i.e3 8 �b7 followed by playing the bishop to b6. 2b) 2
..•
.••
• • • • - . . . • • • • = • •.t• � • • • • • • • • .t.• • • • • • • • 215
+I= Grigoriev, 1931 Shakhmatny Listok
(215): This study has often been reproduced, but its subtlety has never been fully explained. After the introductory moves 1 h6 ! i.g8 2 i.c8 ! (White must prevent . . .�b7; not 2 �b5? �b7 ! 3 �c5 �c7 ! 4 .te6 i.h7! 5 �d5 �d8 ! 6 �e5
156 .i. +.0. v .i.
by 4 . . .'iit b 7, and after 5 i.a4 he can play . . . 'iitc7-d6, drawing. In order to prevent Black reaching d6, White has to play 4 .i.b5 �b7 5 �b4. when 5 . . . 'iitc7 loses to 6 �c5 . Of course, Black will play 5 ...�b6 instead, and now the connection with the pre vious diagram becomes clear. If Black could simply pass after 3 i.a6 ! , then 4 i.b5 �b7 5 �b4 �b6 would be Z l with White to play, and therefore a draw. Unfortunately for Black, he has to move his bishop, and then the same line leads to a win for White because the position is identical to that arising after a black move in Z l . These general ideas are backed up by the following concrete variations: 1 ) 3 i.g8 4 i.c4 i.h7 5 �b5 �b7 6 �c5 ! �c7 7 i.b5 ! (this is line 2c of the previous diagram) i.g8 8 .i.a4 'iitc 8 9 �d6 �d8 10 i.b5 and the king penetrates. 2) 3 .i.g6 4 .i.b5 ! �b7 (4 ....th7 is line 3) 5 �b4 ! �b6 6 .i.a4 i.h7 7 �c4 heading for g7. 3) 3 .i.bl 4 i.b5 ! 'iitb7 (alternatively, 4 . . .i.h7 5 i.a4 �b7 6 �b5 'iitc7 7 �c5 ! i.g8 8 .i.b5) 5 'iit b4 ! 'iit b6 6 'iitc4 wins as in line 2a of the previous diagram. In fact the zugzwangs Z l , Z2 and Z3 are the only non-trivial reciprocal zugzwangs in the ending of i. +.0.a6 v i. (the other three involve the in stant loss of the bishop). I can only express my profound admiration for Grigoriev who, working entirely by hand, found these jewels hidden amongst the millions of possible po sitions. •••
•.•
...
Moving on to the case of the pawn on a5, the vertical opposition posi tion is still winning.
+I-
216
Centurini, 1847 (216): In this case the shorter di agonal leading to h6 is three squares long, so White can win by using zug zwang: 1 i.g7 i.g5 ( 1 .. .i.d2 2 i.h6 .i.c3 3 i.g5 i.g7 4 .i.e7 is also zug zwang; White wins after 4 ... .i.c3 5 h6 i.d4 6 i.f6) 2 i.h6 .i.e7 (2 ...i.f6 3 .i.e3 i.g7 4 i.g5 i.f8 5 i.f6 �f4 6 i.g7) 3 i.e3 i.f8 4 i.d4 �h4 5 i.e5 �g4 6 i.f6 and now Black is in zug zwang; White wins after 6 ...�f4 7 i.g7. Note that these zugzwangs are non-reciprocal; if White is to move, he can lose a tempo by 'going around again'. In the following study, White must play precisely in order to avoid the vertical opposition. (217): White to play draws by 1 �e6! i.f8 2 �d5 ! (not 2 �d7? �d2 3 �c6 �c2 ! 4 �b5 �b3 ! and White loses as in the previous diagram) �d2 3 �c4 ! �c2 4 i.g7 ! (4 �b5?
.t+�
=I+ Buzandian, 1955 1st HM, Bulletin Match Moscow-Prague
217
�b3 ! is again lost after 5 .i.c 1 .i.g7 6 �a5 .i.f6 7 �b5 .i.c3) .i.e7 5 .i.f6 ! i.d6 6 .i.e5 ! with a perpetual attack on Black's bishop.
218
+I= Umnov, 1986 Shakhmaty v SSSR
(218): The immediate advance 1 a6? doesn't win after 1 . . ..i.e3 ! 2 �e2 i.a7 ! (but not 2 . . . .tb6? 3 .tel and Black has to lose a tempo to counter
v
.t 157
the threat of .i.f2; White wins after 3 . . . .ta7 4 .tf2 ! .i.b8 5 �d3 ! �e5 6 �c4 ! winning as in the main line) 3 �d3 �e6! 4 �c4 �d7! with a draw. The correct method is 1 .i.d2 ! (also not 1 �e2? .i.f4 ! 2 �f3 .i.c7 3 a6 .i.b6 4 �e4 �e6) .th4+ ( l . ..�e6 2 a6!) 2 �e2! (2 �fl ? .i.g3 ! loses a vi tal tempo) .tg3 3 a6 ! .i.b8 4 .i.f4 ! (White's first move transferred his bishop to a better position with gain of tempo; now he reaps the benefit) .ta7 5 .i.e3 .i.b8 6 �d3 �e5 7 �c4 ! �d6 8 �b5 ! �c7 9 .i.gl �d7 10 �b6 �c8 11 �c6 ! and wins. So far we haven't said much about Black defending by approaching the pawn from the side. The reason is that it isn't a realistic possibility with the pawn on a6. For example, if White has his king on b7 and plays his bishop to b6, then it doesn't help Black to have his king on d7 . How ever, with the pawn on a5 this type of defence becomes possible. Sup pose Black's bishop is on f1 and White's king is on c7. Then Black draws if his king is on e7. White can only make progress by playing �b6 and then opposing bishops on b5 or a6, but Black can draw by meeting �b6 with . . . �d8, and then he can safely exchange bishops. In this situ ation possession of an edge pawn works against White. The following position provides a dramatic illustra tion of these ideas: (219): The obvious white moves fail: 1) 1 .i.e2? (securing the immedi ate advance of the pawn, but allow ing Black's king to become active)
158 .i.+� v .i.
+I=
219
Original 'iPg6 2 a6 .i.e4 ! 3 'iPg8 'iPf6 ! 4 'iPf8 'iPe6 ! 5
prevents ... 'iPg6) 3 .i.h7 ! (White's plan is to occupy the b l -h7 diagonal with his bishop, and then sneak his king out via g8 and f7; if Black ever plays ...'iPg5, then White replies 'iPg7) .i.a6 (3 ... 'iPg5 4 'iPg7 ! .i.c4 5 .i.bl 'iPf4 6 'iPf6! will eventually lead to the vertical opposition on the queenside) 4 .i.bl ! (the only square for the bishop, since 4 .i.c2?, for ex ample, allows 4 . . .'iPg5 ! 5 'iPg7 'iPf4! 6 'iPf6 'iPe3 ! 7 'iPe5 'iPd2 ! gaining a vital tempo, and rounding up the pawn after 8 .i.f5 'iPc3 !) .i.c4 (trying to keep White's king bottled up as long as possible; White wins more easily after 4 . . . .i.b7 5
.i.+� v .i. 159
Moving the pawn back to a4, the vertical opposition now leads to a draw, because the diagonal a5-d8 contains four squares, long enough to make it impossible for White to drive the bishop away. There are sev eral interesting positions with the pawn on a4, so we will give a slightly more detailed coverage. The composer Koranyi created two studies which eventually reduce to the following position. 220
+/-
queenside immediately. The detailed analysis runs: 4a) 9 �d6? �f6! 10 �c5 .i.fl 1 1 J.a4 �e7 and Black has his target draw. 4b) 9 .i.h7? (gives Black a vital tempo on the way to g7) �h6! 10 J. b l �g7 ! and Black can shadow the white king when it moves to the queenside. 4c) 9 .i.e4? �f4 ! 10 .i.bl �e5 ! and Black's king has slipped across die barrier erected by the white king md bishop. 4d) 9 .i.bl! (the only effective waiting move) .i.fl (the line 9 ...lti>h6 10 �d6 �g7 1 1 lti>c5 wins directly, while 9 . . .�f4 10 �d6 .i.fl 1 1 .i.a2 �e4 1 2 .i.e6 lfi>d4 13 .i.d7 .i.e2 14 �c6 �e5 1 5 �c5 .i.a6 1 6 �b6 is also winning after 17 .i.b5) 10 �e6 ! J.a6 1 1 .i.a2 (White has prepared �d5 followed by .i.c4 without al lowing Black to activate his king) �f4 1 2 �d5 �e3 13 .i.c4 .i.b7+ 1 4 �c5 .i.c8 1 5 �b6 �d4 16 .i.a6 .i.h3 17 .ib5 .i.c8 18 .ic6 �e5 19 .ib7 and wins.
• • • • m • • • • • • • • • • • � &/:! � " &/:! �, � u � • • • •• • • • • • • • • =I= 221 Koranyi, 1988 (end of study) Magyar Sakkelet
(221): The composer's main line was 1 .i.f2 .i.d2 2 �g6 lti>g4? 3 h5 ! , reaching the winning vertical oppo sition of diagram 2 1 6. However, both here and in a similar 1 97 2 study, Koranyi overlooked the sim ple drawing line 2 . . ..i.e3 ! 3 .i.el .i.d2! . The following study has a stun ning first move. (222): Black to move wins easily by 1 . . . .i.b5 2 .id5+ �c3, followed by . . . a4-a3, so we will suppose that
1 60 .i.+� v .i.
In the following study, White wins by carefully avoiding the draw ing idea from the previous position.
=I+ Sarkissian, 1976 5th Comm., Peckover Jubilee Tny. 222
White is to move. There three possi bilities: 1) 1 .td5+? loses after 1 . . .'it>c3 2 'it>e2 .tb5+ 3 'ifr>d l 'ifr>b2, and Black's pawn advances. 2) 1 'ifr> e3? (1 'it>e2? .tb5+ is also bad) .td7 2 .tf3 �c2! 3 .te4+ �c3 wins. 3) 1 'it>f2! (this is the key move; White's bishop will need access to d 1 later, so White's king heads for e l using a route which doesn't allow a black bishop check) .tb5 2 .tf3 ! 'it>c3 (White makes use of his first move in the line 2 ...�c2 3 �el ! �c3 4 .td 1 ! .tc4 5 .ta4 !) 3 .td 1 ! �d2 (after 3 . . ..tc4 4 .ta4 ! '1tb4, White draws by 5 .te8 .tb5 6 .txb5 !, but not 5 .i.d l ? .tb3 ! 6 �el �c3 !) 4 .tb3 ! .te8 5 'it>fl (the composer con tinued with 5 'it>f3 ?, but then Black can win by 5 ...�c3 !; fortunately, this doesn't affect the soundness of the study) .tb5+ (5 ... 'ifr>c3 6 .td l !) 6 '1ti>f2! .te2 7 .ta4! .td l 8 .te8 and White draws.
+I= Zur, 1984 112 HM, Int Friendship Team Tny. 223
(223): White can play: 1) 1 �c6? �f7 ! 2 .tb4 .td8! 3 'ifr>b5 .tc7 allows Black to establish his bishop on the a5-d8 diagonal. 2) 1 'ifr>c7? �f7 ! (Black draws us ing the manoeuvre from the previous position) 2 .tb4 'ifr>e8 ! and the bishop can move to d8. 3) 1 'it>c8! (avoiding a check after a later . . ..td8) .td4 ( l . ..�f7 2 .tel 'ifr>e6 3 a5 .td4 4 a6 �d5 5 �b7 �c4 6 .tg3 �b5 7 .tb8 wins) 2 �c7 ! (nOl 2 .tel ? .tb6!, but now White threat ens to move the bishop) .te5+ 3 �c6 .tf6 (the only way to meet the threal of .tel ; White has transferred his king from b7 to c6 with gain of tempo) 4 .tb4 (4 .te 1 ? .td8 ! 5 .tf2 .ta5 ! draws) .td8 5 .tc5 (the ideal square for the bishop, because it stands ready to intercept on either di agonal) .ta5 6 �b5 ! .te l 7 .tb4!
i.+� i.t"2 8 a5 �f7 9 a6 i.a7 10 i.c5 i.b8 1 1 �c6 and wins. There are 8 reciprocal zugzwangs with i.+�a4 v i.. We will take a look at two of them, not only for their intrinsic interest, but also be cause we will need the analysis later.
I
224
= -
Original (224): First suppose that Black is to move. White has two potential winning plans; the first is to play his king to c6, forcing ... i.d8, then i.cS (meeting . . . i.a5 by �b5) and the second is simply king to b5 and i.c5-b6. 1 ) 1 �g2 2 �d4 �f3 3 �c5 �e4 4 �c6 ! i.d8 5 i.d6 ! i.a5 6 �b5 ! i.el (if 6 . . .i.d8, then 7 i.c5 ! and 8 i.b6) 7 i.b4 ! i.f2 8 �c6 �d4 9 a5 �c4 10 i.d6 i.gl 1 1 a6 i.f2 1 2 i.c7 i.a7 1 3 �b7 i.f2 1 4 i.b6 and wins. 2) 1 i.d8 2 �d3 and now: 2a) 2 i.c7 3 �c4 �e2 4 �b5 �d3 5 i.c5 �e4 6 i.b6 ! i.g3 7 a5 �d5 8 a6 i.b8 9 i.f2 followed by �b6-b7 wins. •..
...
•••
v
i. 161
2b) 2 'iPl'2 3 �c4 �e3 4 �b5 �d4 5 i.c5 + ! �d5 6 i.b6 ! i.h4 7 a5 ! i.g3 8 a6 and wins as above. 3) 1 i.b6+ 2 �e4 �e2 3 �d5 ! 'it>d3 4 �c6 ! i.d8 5 i.c5 i.aS 6 �b5 i.el 7 i.b4 and wins as before. Black to move loses because if he moves his king, White advances his king to c6 and Black cannot meet �c6 by . . . �c4; if he moves his bishop to b6 White gains a tempo when his king arrives on c6, and if he plays ...i.d8, when White plays i.c5-b6 Black's bishop cannot be defended by his king. Now suppose that White is to play. As we have seen, he has two plans, but both fail, for example 1 �d4 �e2 ! 2 �c5 �d3 ! 3 �c6 �c4 ! 4 i.el �b3 5 �b5 i.d8 with a draw ing vertical opposition, or 1 �d3 �f2 ! 2 �c4 �e3 ! 3 �b5 �d4 ! (not 3 ...�e4? 4 �c6) 4 i.c5+ �d5 5 i.b6 �d6 ! 6 a5 �d7 ! 7 a6 �c8 ! and Black's king reaches the comer. The other reciprocal zugzwang is: ...
...
. �� . . • • • • • • • • • • • • �- . . . . . ·�· • • • • • • ••• I
225
= -
Original
162 .t +� v .t
(225): Suppose Black is to move: 1) 1 . .�gl 2 .t c5+ �fl (2 . . . �h2 .
3 �e4 ! �g3 4 �d5 ! �f4 5 �c6 ! .t a5 6 �b5 ! wins) 3 �e4 �e2 4 �d5 ! �d3 5 �c6 ! .ta5 6 �b5 wins. 2) 1 . .tc7 2 �e3 ! is the previous diagram. 3) 1 . ..t b6 2 �e4 ! wins as in line 3 of the last diagram. With White to play 1 �e3 .t c7 is the previous diagram, 1 �e4 �e2 ! 2 �d5 �d3 3 �c6 �c4 transposes to the White to play analysis of the pre vious diagram and 1 .t c3 .tc7 2 �e4 �e2 ! 3 �d5 �d3 ! 4 .te 1 .td8 is a clear draw. In this position, Black to play loses because he must either lose time with his king, permit a second reciprocal zugzwang or put his bishop on a square allowing White to gain a tempo by �c6. ..
.
There are three reciproc'al zug zwangs with the pawn on a3. The following position is fairly complex, but we have already done a lot of the work in the previous two diagrams. (226): Suppose firstly that Black is to play. Thanks to our earlier ef forts, we can dismiss some moves straight away. B l ) 1. .�g2 2 a4 .tc7 3 �d4 is line 1 of diagram 224. B2) 1. ...t c7 2 a4 ! is precisely diagram 224. B3) 1 ....tf2+ 2 �d3 �g2 3 a4 .t b6 4 �c4 � f3 5 �b5 .tc7 6 .tc5 and 7 .i.b6 wins because Black's king is too far away. Therefore, Black's only chance is to move his bishop on the b8-h2 .
=I-
226
Original diagonal (but not to c7). Then a4 will be impossible because the reply ... .tc7 will leave White on the wrong end of a reciprocal zugzwang. B4) 1 .. .th2. The defect with this move is that it gives up the possibil ity of . . . .te l , but in order to exploit this White has to introduce a para doxical third plan, namely 2 .t a5 . It looks very odd to block the pawn like this, but the idea is to play a4, then �d4-c4, and finally a bishop move to release the pawn. The point of the preliminary .ta5 is to advance the pawn to a4 without allowing Black to set up an immediate block ade with his own bishop. White will then have the option of freeing the pawn by moving his bishop in either direction. Black may try: B4a) 2....tg3 (returning to g3 is too slow) 3 a4 �g2 (the net result of the various manoeuvres is that both sides have lost two tempi compared to diagram 224, and we transpose into line 1 of that position) 4 �d4 �f3 5 �c5 �e4 6 .t b4 .tc7 7 �c6! .
i. +.0. v i. 1 63
i.d8 8 i. d6 i. a5 9 � b5 ! i.e l (or 9 ...i.d8 10 i.c5) 10 i. b4 ! and wins. B4b) 2 �g2 3 a4 �h3 4 i. d8 �g4 (thanks to Black's first move, he cannot play ...i.e l ) 5 �e4 (5 a5 ? �f5 ! draws) i. gl 6 a5 , followed by i.b6. After other moves on the b8-h2 diagonal, such as l . . .i. b8, White also wins with 2 i. a5, e.g. 2...�g2 3 a4 �g3 4 �e4 �g4 5 i. b6 and wins. Now suppose that White moves first. Then he may try: W l ) 1 �d3 ( 1 a4 i. c7 ! is dia gram 224) �f2 2 �c4 (2 �d4 �e2 !) �e 3 ! 3 � b5 (3 �c5 �d3 !) �d4 ! 4 a4 i. c7 ! 5 i. c5+ �d5 6 i. b6 �d6 ! draws. W2) 1 �f3 i.c7 ! 2 a4 i. d8 ! is diagram 225. W3) 1 �e4 �e2 ! doesn't help White. W4) 1 i.a5 �g2 ! (the king heads for c8) 2 a4 �h 3 ! 3 �d4 (White's bishop moves are unprofitable be cause Black's bishop can move to c7 or e l ) �g4 4 �c5 �f5 5 �c6 i. h4 ! (Black must retain the option of stopping the pawn from two differ ent diagonals) 6 i. b4 i. d8 ! 7 i.c5 i.a5 ! 8 � b5 i.el and Black's king is just in time to reach c8 after 9 i. b4 i.xb4 !. •••
Similar positions result from the analysis of the following over-the board example with the pawn on the second rank. (227): The position is drawn, but Black made a fatal error: 1 i. g4 ( 1 h4 �b6 2 i. g4 �c7 i s no better, since 3 h5 �d8 4 h6 i. c2 ! 5 �f4
227
=I= Beliavsky-Kotronias Belgrade 1993
�e7 6 i. f5 i. xf5 7 � xf5 � f7 and 3 i.xdl �d8 4 �f4 �e7 5 �g5 �f7 6 �h6 �g8 ! both lead to a draw) i. c2? (Black could have held on by l ... i.a4 2 h4 i.e8 ! 3 �f4 �b6 4 �g5 �c7 5 i. f5 �d6, and now either 6 �f6 i.h5 ! or 6 i. g6 �e7) 2 �f4 � b6 3 �g5 �c7 4 h4 ! �d8 5 �f6 ! �e8 6 �g7 ! 1-0, in view of 6 . . . �e7 (6 ... i. g6 7 �xg6 �f8 8 i.e6) 7 h5 i.d3 8 h6 winning. (228): White to play wins by 1 i. f7 ! (it is important not to allow Black's bishop to settle on the h5-e8 diagonal, since 1 h4? i. g6 ! 2 �g4 �h2 3 �g5 �g3 is a drawn vertical opposition) i.e2 2 h4 i. d 1 (for the moment Black can only wait) 3 i.g6 (White must improve his bishop po sition as much as possible before re leasing Black's king from the back rank; 3 �f4? �h2 4 i. g6 �h3 ! 5 �g5 �g3 6 i.h5 i. a4 7 i. g4 i.e8 ! is only a draw) i.e2 4 i.f5 (eventually Black's bishop must be allowed to h5, but it is important that White is
164 i. +� v i.
=I-
229
Cheron, 1956 Journal de Geneve ready to respond immediately with � f4-g5, gaining a tempo by attack ing the bishop; if Black's bishop is allowed to reach e8 before White plays � f4, then the position is drawn, for example 4 i.e4? i. h5 ! 5 i. f5 i.e8) i.h5 5 �f4 ! �h2 6 �g5 ! (thanks to the bishop on f5, Black cannot prevent an interception next move) i.e8 7 i. g6! i.a4 8 h5 i. b3 9 h6 i. g8 10 �f6 �g3 1 1 �g7 and wins. The following reciprocal zug zwang is the longest in the ending of i. +� v i. . Two other reciprocal zugzwangs arise in the course of the solution. (229): Suppose that Black moves first. Black will be reluctant to move his bishop and allow the pawn to a4, but playing a4 doesn't by itself guar antee victory. White must also be careful not to allow Black's king to escape from the corner, and he must also bear in mind the reciprocal zug zwang of diagram 225. After 1 ...�h l
Original (the lines l ...�fl 2 �f3 !, 1 ...i.e7 2 a4 ! �fl 3 �f3 ! and l . . .i.c 1 2 a4 �fl 3 �f3 ! i. g5 all transpose to variation 3 below) White has a diffi cult choice: 1) 2 i. b6? (this allows Black to transfer his bishop to the a5-d8 di agonal, which not only holds up the pawn but allows Black's king to slip out of the corner) i. c5 3 i.c7 i. b6 4 i. f4 �g l 5 �f3 �fl 6 i. g3 i. a5 7 a4 i. d8 8 i. f2 i. c7 9 i. h4 i. a5 ! 1 0 �e3 (White cannot make progress using only his bishop, but moving the king allows Black's king to escape) �g2 1 1 i.e7 �g3 1 2 �d4 � f4 1 3 �c5 �e5 14 �b5 i.el and Black is ready to swap bishops after i. b4. 2) 2 i.d8? i.c5 3 a4 i. b6! 4 i.h4 �g l 5 �f3 �h2 6 i.e7 �h3 7 �e4 �g4 is similar to line 1 . 3) 2 i. c7! (this is a second recip rocal zugzwang; c7 is the best square for the bishop because after 2 . . .i.c5 3 �f3 Black's king cannot escape from the corner) and now:
.t +� v .t 165
3a) 2 icS 3 � f3 .t a3 (other wise White gets in a4 for free and wins as in line 3b) 4 �e4 (White's king is now sufficiently well-posted to win by a direct run to the queen side) �g2 5 �d5 ! �f3 6 .id6 ! .tc 1 7 a4 .td2 8 �c4 ! �e4 9 .tb4 ! (Black is effectively a tempo down over line 1 and cannot exchange bishops) .te3 10 a5 and wins. 3b) 2 ..trs 3 a4 .tb4 4 �f3 (this case is completely clear; White has improved his king position and ad vanced the pawn to a4) �gl 5 �e3 �g2 6 �d4 �f3 7 �c4 ! .tel 8 .id6 .ta5 9 �b5 .id8 10 .ic5 wins. 3c) 2 �gl 3 �f3 ! (a third recip rocal zugzwang) �fl (3 ....tc5 4 a4 is line 3b) 4 .ia5 ! (this switchback is necessary to confine Black's king) with a final branch: 3cl) 4 �gl (White can again head for the queenside) 5 �e4 �f2 6 �d4 'iPf3 7 �c4 ! .ie7 (7 ... .icl 8 a4 .tg5 9 .ib6 is the same) 8 a4 .th4 9 .ib6 .te l 1 0 .tc5 �e4 1 1 .tb4 ! wins (as in line 3a). 3c2) 4 .te7 (4 . . . .tcl 5 a4 .ig5 is the same) 5 a4! .t g5 6 .ic3 ! (not 6 .ib4? .id8! and now White falls victim to the reciprocal zugzwang of diagram 225; c3 is the only other square available on the a5-el diago nal) .id8 7 .ib4 ! and now Black has been forced into the zugzwang of diagram 225. The analysis is simpler if White is to move: 1 �f3 allows Black's king to escape by l . ..�h2 !, leading to a draw after 2 .tc7+ �h3 3 .tg3 (or 3 .id8 .td6 4 a4 .ib4 ! 5 .tf6 .ta5 6 .te7 .tel 7 �e2 .ta5 ! 8 �d3 ••
••
�g4 9 �c4 �f5 10 .tb4 .ixb4) .te7 4 a4 .td8. Other lines are similar, for example 1 .tc3 .td6+ 2 �f3 �h2 3 a4 .ic7 (this is the defect of 1 .tc3; Black's bishop can immediately set tle on a diagonal which restrains the a-pawn) 4 �e4 �h3 5 �d5 �g4 6 �c6 .td8 ! 7 .id4 .ta5 ! 8 �b5 .te 1 9 .ic5 �f5 ! (now Black is just in time) 10 .tb4 .txb4 ! 1 1 �xb4 �e6 ! drawing. We conclude this section with the longest win in .t+� v .t (25 moves):
.•.
...
..•
230
+I-
Original
(230): Black is to move. White wins by 1 . . ..tc6 (after 1 . . . �f2 2 a4 �e3, White wins by 3 �c7 �d2 4 .ig4 .ig2 5 a5, leading to a winning vertical opposition, but not 3 a5? �d4 ! 4 a6 �c5 drawing) 2 �c7 ! .ie8 3 .if3 (now White improves the position of his bishop with gain of tempo; the threat is .ic6) .ta4 4 .ic6 .id l 5 .id5 (threatening .ib3) .ia4 (after 5 ...� f2 6 .ib3 .ie2 7 a4 �e3, White wins by 8 �b6 � d4 9 .ie6
166 .t +� v .t .t fl 10 .id7
3.2:
�+b� v �
We start, as usual, with the pawn on the seventh rank. In this case, the vertical opposition leads to a win for White, but the method is far from straightforward and is well worth knowing. (231): It doesn't make any differ ence who moves first, so assume that White is to play. At first sight the win is quite easy, because if White plays his bishop to b8, Black will have to transfer his own bishop to a7. Then a direct attack on the black bishop will force the pawn home. However, Black can defend against this plan by meeting .ih4-f2 with . . .
+/-
23 1
Centurini, 1856 this deadlock is quite elegant: 1 .th4 �b6 (White wins easily if Black ig nores the threat: l . . ..ie5 2 .if2 .if4 3 .ia7 .i g3 4 .ib8 .if2 5 .ih2 .ia7 6 .i g l ) 2 .if2+
i. +8 v .t 167
a a a.ta . - . . �-� . d . . � d - d - - - . - . . a • m m - - - •.t• - • ,,
+/-
232
Centurini, 1856 �a7 (or 3 . . . i. a7 4 i.c7) 4 i.c7, but he has far more trouble when Black moves first. Nevertheless, Black cannot escape his eventual fate after l . . .�a7 2 i. g5 (the first step is to threaten i.c7, which forces Black's king to b8) i. g3 3 i.d8 �b8 4 �b6 (now White can transfer his king to a6) i. h2 5 �a6 i. gl (this only de lays the end by one move, because the bishop has to return immedi ately) 6 i. h4 i. h2 7 i. f2 (the threat of i. a7+ forces Black to move his king, but then White's own king penetrates to a8) �c7 8 �a7 �c6 9 �a8 �b5 10 i.a7 �a6 (the horizon tal opposition has been transformed into a vertical opposition; this case is particularly easy) 1 1 i. b8 i. gl 1 2 i.g3 i. a7 1 3 i.h2 and wins. White also wins if the king posi tions are reversed. (233): Black to play can simply win the pawn, so suppose that White moves first. He wins by 1 �f4 ! �e7 2 � g5 ! � f7 3 �h6! �f6 4 i. c2 (a waiting move forcing Black to give
233
�
+/=
Pierce, 1899 (end of study) British Chess Magazine White's king access to either g6 or h7) �e7 (after 4 . . .i.c4 5 �h7 the king reaches h8, and White wins by playing his bishop to g8 as in the previous diagram) 5 �g6 �e8 (or 5 . . . .t f7+ 6 �h7 i.e6 7 �h8 �f6 8 i. h7 i. f7 9 i. g8) 6 �f6 �d7 (or 6 . ..i. a2 7 .l g6+ and 8 i. f7) 7 i. g6 �d8 8 i.f7 i.h7 9 �g5 �e7 10 �h6 and the bishop is trapped. There is just one reciprocal zug zwang with i.+8b7 v i. , but it is completely trivial: W�c8, i. a8, 8b7 v B �d6, i.c7. With the pawn on b6, both the vertical and horizontal oppositions are winning for White. In the case �c7 v �c5, White wins by playing his bishop to b7, forcing the enemy bishop onto the short a6-c8 diago nal, when zugzwang is decisive. The horizontal opposition is worth a dia gram: (234): White to play wins by 1 .l b5 (putting Black in immediate
1 68 .i. +� v .i.
• • • • B.t.Bi.B B B B B B . . - �� • • • • • • • • • • • • • • • • 234
+/-
Centurini, 1856 zugzwang) .i.c8 2
235
+I=
Obukhov-Mishuchkov Moscow 1991 (a second zugzwang forces Black to cede control of b7) .i. h3 8
. . � . � •-*-• • • • " � �.t � • u • • • • • • • • • • • • • • • • • • • • 236
=I-
Chiron (1927) and Dehler (1922) (236): This reciprocal zugzwang is a combined effort. Dehler ana lysed it with White to play, while Cheron contributed the Black to move analysis. It is closely related to the previous position. If Black is to
i. +� v .t 1 69
play, then White wins by 1 .. ..i.a6 ( l ....t f3 2 .td5 ! .i.xd5 3 'itxd5 ! 'itd7 4 'itc5 ! 'itd8 5 'itd6 'itc8 6 'itc6 ! ) 2 *c6 ! .i.e2 (or 2 . . ..t c8 3 .i.c4 .i.f5 4 *b7 ! .te4+ 5 'ita7 as in the main line) 3 'itb7 ! .i.f3+ 4 'ita7 .i.e4 5 .tc4 (White wins as in the previous diagram) .t f3 6 .i.a6 .i.e4 7 .i.b7 .if5 8 .t f3 .tc8 9 .t g4. It is perhaps more surprising that White to play cannot lose a tempo: 1 'itc5 ( 1 .i.d5 'itc8 ! is an immediate draw, while after 1 .i.d7 .i.f3 2 i.f5 i. g2, Black simply avoids putting his bishop on b7 until White has played i.e6) i. f3 (but not 1 ...'ite7? 2 i.d5 ! .t a6 3 'itc6 'itd8 4 i.e6 ! and wins as in the previous diagram) 2 .id5 (2 'itb5 .i. b7 draws) .i.g4, fol lowed by ...'itc8, and Black draws. In the following example Portisch draws by accurate defence based on the Cberon/Dehler reciprocal zug zwang and the ideas contained in diagram 235.
(237): White was to play. The game continued 1 i. d5 ! 'itc7 2 'itb2 'itd6 3 i. g2! (not 3 i. b7?, when Black gains a vital tempo and wins by 3 ....t a4 4 .i. g2 i. c6 ! 5 i. h3 'ite5 6 'itc3 'itf4 ! 7 'itd2 'itf3 !) .t a4 4 'itc3 i.c6 5 i. h3 ! (not 5 .t n ? 'ite5 ! 6 'itd2 'itf4 ! 7 'ite l 'itf3 and now White's king cannot move to fl , so Black wins after 8 .th3 i. b5 !) 'ite5 6 'itd2 ! (6 'itd3? loses to 6 ... 'itf4 ! 7 'ite2 i. b5+ ! 8 'itd2 'itf3 ! 9 'ite l i. a6) 'itf4 7 'itel ! i.b5 8 i.c8 'ite3 9 i. b7 ! i.d3 10 i. g2 ! (this is the re ciprocal zugzwang in the previous diagram) i.e2 1 1 i.c6 i. c4 1 2 i. b7 'itf4 1 3 i.c6 i. d3 14 i. b7 'it g4 1 5 .t g2 i.a6 1 6 'itd2 'itf4 1 7 'ite 1 112-112. The following two positions de pend on various tactical points:
=I=
238
Cheron, 1952
=I+
237
Portisch-Tal Bled Ct (1) 1965
(238): White can only draw be cause of a tactical point which is worth knowing as it can arise in vari ous .t +� v .t positions. After 1 b7 ( 1 i.d7 i. g2! transposes to the pre vious diagram), Black plays the
1 70 .i. +£!. v .i.
suiprising l .. .i.c8 ! and none of the available promotions helps White. 2 b8'ii' and b8l:I are stalemate, while 2 b8i. i.h3 (but not 2 . . .i. a6? 3 i.d7 and mate next move) and 2 b8l'Ll i. h3 (not 2 ... i. b7? 3 i. d7 ! i. g2 4 l'Lla6 ! i. f3 5 l'Llc5 ! either winning the bishop or forcing mate) lead to noth ing. White wins if his bishop starts on b5 . The simplest line is 1 b7 ( 1 i. fl also wins) i.c8 2 b8i. ! i.h3 3 i.c7+ and mate next move. If we return the bishop to a4 and move Black's bishop from h3 to g4, then White again wins, this time by 1 i.d l ! i. c8 2 �c6 i. a6 3 i.g4 !. Moving White's bishop to c6 re sults in a draw, regardless of whether Black's bishop starts on h3 or g4.
the extra tempo if he moves first, but in fact he can draw with a suiprising tactical resource: 1 i.d6 ! ( 1 i.b6? g2 loses as before) g2 2 i. g3 ! i. g5 3 i. f4 ! i.f6 4 i.e5 ! i.e7 5 i. d6 ! i.d8 6 i. c7! and Black cannot escape from the perpetual attack. We have already seen a reciprocal zugzwang in this section (diagram 236); there are five in total with i.+l!.b6 v i. . Now suppose that the pawn is on b5. Averbakh's drawing zone (dia gram 2 1 2) implies that the vertical opposition leads to a draw. The hori zontal opposition is no different:
=I=
240
Berger 239
=I+
Silaev, 1978 (end of study) 1 st HM, Shakhmaty v SSSR (239): If it is Black to play, then after 1 . . .g2 2 i. b6 i.el 3 i.c5 �g3, for example, he wins exactly as in diagram 232 (with Black to play). At first sight White cannot make use of
(240): Black draws after 1 i.c5 i. b8 ! 2 b6 'iti>a5 ! (but not 2 ... i. g3? 3 b7 �a5 4 �d5 ! �a6 5 �c6 !) 3 �d5 b5 ! 4 b7 i. c7 ! 5 i.d6 (there is no way to make progress) �b6 and White loses his pawn. The following selection of three composed positions is based largely on entertainment value. They could
.i. +� v .i. 1 71
just as well have appeared in the ear lier sections, since in all three cases the pawn quickly steps forwards.
• • • • • • • • • • • • ·�· . . . . . � • • • •• • • • m . . . ·� 241
line, while 6 ... .i.a7+ 7 �fl �g4 8 �e2 wins as after 4 . . . .i.c7 above) 6 �gl �g4 7 �g2! .i. b8 8 .i.f2 ! .i.c7 (8 . . .�f5 9 .i. g3 ! .i.a7 10 �f3 trans poses, but not 9 �f3? �e6 1 0 .i. g3 �d5 heading for b7) 9 .i. gl ! (now that White has a bit more space he can challenge the h2-b8 diagonal) .i.b8 1 0 .i.h2 ! .i.a7 1 1 .i. g3! (B lack is in zugzwang and must give way) �g5 1 2 �f3 �f5 1 3 .i.f4 (the pattern is repeated) �f6 14 �e4 �e6 1 5 .i.e5 �e7 1 6 �d5 �d7 1 7 .i.f4 �e8 18 �c6 �d8 1 9 �b5, followed by �a6, and White wins.
+I=
Grigoriev, 1931 2nd HM, Shakhmatny Listok (241): Black to play draws easily by 1 . . . .i.d8 ! (not 1 . . ..i.f2? 2 .i.gl !), so assume that White moves first. The first few moves are easy enough: l b6 ! .i.f2 2 b7 ! .i.a7 3 .i.gl ! (not 3 J..c7? � g4 ! 4 � g2 � f5 ! 5 �f3 �e6 ! and Black will gain a vital tempo when he plays ...�d7) .i.b8 4 i.f2 ! (after 4 .i. b6? �g3 Black sim ply heads for the white pawn; the move played creates an escape route for White's king while at the same restricting the enemy king) .i.h2 (4 ... � g4 5 � g2 ! transposes into the main line, while 4 ... i.c7 5 �gl .i.b8 6 �fl .i.c7 7 �e2 �g4 8 �d3 �f5 9 �c4 �e6 1 0 �b5 �d7 1 1 �a6 .i.b8 12 � b6 wins as in diagram 233) 5 .i.el (a waiting move designed to release the white king) .i.c7 (or 5 ....i.b8 6 �g l and now 6 ... �g4 7 �g2 ! .i.a7 8 i. g3 leads to the main
242
+I=
Silaev, 1986 4th HM, Golden Fleece Tny (242): Black to play draws easily by L.�g3 2 g6 .i.f8 !, so suppose that White is to move. The winning line runs 1 .i.a5 (not 1 g6? .i.f8 ! 2 �d3 �g3 3 �e4 �g4 ! 4 .i.d8 .i.g7 ! 5 .i.e7 .i.b2 drawing; first of all White has to improve the position of his bishop) .i. a3 ( 1 . . ..i.f8 2 �d3 ! � g3 3 �e4! �g4 4 .i.d8 ! �h5 5 �f5 ! followed by g6 wins as in the
1 72 .t +� v .t main line) 2 g6! .if8 3 .ib4 ! (thanks to White's first move, this manoeu vre is much more effective; if White had challenged the bishop on f8 with .ic5, Black would have been able to play . . . .i g7, when no further chal lenges would have been possible, whereas now .ic3 is possible) .ih6 4 .id2! (not 4 �d3? �g3 5 �e4 �g4! 6 .i d2 .i g7 ! ; White must achieve a favourable set-up before he advances his king, which in this case means having his bishop on c3, or reach ing .id2 v .if8) .i g7 (after 4 ....if8, Black's bishop doesn' t control e5 and White can win by 5 �d3 � g3 6 �e4 �g4 7 �e5 ! .i g7+ 8 �e6 !) 5 .i c3 ! .ih6 (after 5 ....if8 6 �d3 ! �g3 7 �e4! �g4 8 .if6! White wins as in the main line) 6 �d3 ! � g3 7 � e4 ! � g4 8 .if6 ! .i g7 (this tricky move is the best defence; after 8 ....i f8 9 �e5 ! h5 1 0 �f5 ! �h6 1 1 .ic3 .ig7 1 2 .id2+ �h5 13 .ig5 White wins as in diagram 234) 9 �e5 ! h5 10 �f5 !. The composer's solution stopped here, but the maga zine EG continued with 1 0. .. �h6 1 1 .ie7?, allowing Black to draw by 1 1 . . ..ib2. However, White can win by l 0. . . �h6 1 1 .i g5+ ! �h5 (now White has to lose a tempo in order to put Black in zugzwang) 1 2 .ie7 �h6 1 3 .if6! .if8 14 .ic3 .ig7 1 5 .id2+ � h5 1 6 .ig5, transposing to dia gram 234. (243): This is an attractive idea, even though properly speaking the study is unsound. White wins by 1 �e6! ( 1 g6+? �e8 ! 2 �e6 fails to 2 . . . �f8) .ie7 (a neat stalemate de fence; after 1 . . ..i b4 2 �f7 White
243
+/=
Taniev, 1986 2nd Comm., Revista Romana de Sah wins easily) 2 g6 ! �e8 3 .ig5 ! (not 3 .ie l ? �f8 ! , nor 3 .if6? .i xf6 4 � xf6 �f8 ! ) .ia3 (after 3 ... .if8 4 .ie7 .ih6 5 .ia3 .ig7 6 .id6! Black has fallen into the reciprocal zug zwang of diagram 236) 4 g7 (this is the flaw; White can also win by 4 .ic 1 .i f8 5 �f6 .ib4 6 �g7 ! , pene trating to h7) .if8 5 g8.i ! .ia3 6 .if7+ ! �f8 7 .ih6#. The following practical example uses some of the ideas discussed ear lier. (244) : White was to play in the game, which continued 1 �h6. ECE claimed that 1 �f6 leads to a draw, but this is not so. The critical posi tion arises after 1 f6 (while this re tains the win, it makes it much more complex) �e3 2 �g7 (White can only win by backtracking; after 2 .if5? .ie8 ! 3 .ie6 �f4 ! Black has achieved a drawn vertical opposi tion) �f4 3 �h6 ! .ie8 and now ECE stops with an '=' sign. The winning line is instructive: 4 .i g6 ! (and not
i. +� v i. 1 73
+/-
244
Milev-Ermenkov Bulgaria 1977 4 i.c2? � g3 ! , followed by . . . �h4. again with a drawn vertical opposi tion; White has to find a way to transfer his bishop to h5 without al lowing Black's king to reach h4) i. a4 5 i.d3 �e3 (after 5 ...i.e8 6 i.e2 the bishop reaches h5) 6 i. bl i.e8 7 i.c2 � f4 (or 7 . . .�d2 8 i. f5, followed by i. g4) 8 i.d l i.f7 9 i.h5 (target achieved!) i. g8 1 0 g6 �f5 1 1 g7 winning as in diagram 233. The game continuation was much simpler. After 1 �h6 i.e8 2 i.f5 �e3 (or 2 ... i. f7 3 i. g4 �e5 4 i.h5 i. b3 5 g6 �f6 6 g7 winning as in diagram 233) 3 i. g4 i. f7 4 i.h5 i. g8 (not the best defence, but it no longer matters) 5 i.e8 (now White doesn't even have to find the dia gram 233 win, because there is no defence to g6 followed by i.f7) �f4 6 g6 1-0. There are 16 reciprocal zug zwangs with i. +�b5 v i. . The fol lowing position offers six for the price of one.
- - - - - - Jt m • • • ·�· . . - . - . . - . . - . - . • • • •• =I-
245
Original (245): Suppose that Black moves first. It is easy to imagine that White doesn't want to move his bishop, be cause this would give Black's bishop a chance to settle on a diagonal con trolling the b-pawn. It is harder to see why Black's bishop is also im mobilised. If the bishop moves on the a l -h8 diagonal, then it must go to a square allowing i.c5 to be met by a move onto the a5-d8 diagonal. This means e5 or f6, but both these squares allow White to gain a tempo by attacking the bishop with his king, for example 1 . . . i.e5 2 �g6 �g2 3 �f5 i. b8 4 i.c5 i.c7 5 �e6 �f3 6 �d7 i. a5 7 �c6 �e4 8 i.d6 and 9 i.c7, with an easy win. Mov ing the bishop on the a5-e l diago nal has the defect that it gives up control of e5, and so White's king can cross to the queenside more eas ily. It turns out that B lack is sooner or later forced to make this conces sion in any case, so yielding imme diately simply transposes to the main line after, for example, 1 . . .i.d2
1 74 .t +,0, v .t 2 �g6 �g2 3 �f5 �f3 4 .td8 .te3 5 �e5 ! �e2 6 �e4! .tf2 7 .tc7. It follows that Black's best chance is to avoid moving his bishop for as long as possible, but White has the opposition, and Black has to give way at some stage. The main line is 1 ...�h2 ( l ...�g2 2 � g6 ! leads to a second reciprocal zugzwang; after 2 . . . �g3 3 �g5 ! we have transposed back into the main line) 2 �h6 ! (again reciprocal zug zwang) �h3 (at any stage Black may give in and move his bishop, but we do not consider this at every move because the best Black can hope for is transposition to the main line) 3 � h5 ! (reciprocal zugzwang num ber four) � g3 (after 3 ....te l 4 .t d8 .tf2 5 .tc7 .te3 6 .td6! .t b6 7 �g5 White wins more easily) 4 �g5 ! (number five) �f3 (4 ....td2+ 5 �f5 ! �f3 6 .td8 .te3 7 �e5 ! �e2 8 �e4 ! .t f2 transposes to the main line) 5 �f5 ! (number six) and now: 1 ) s .td2 6 �e5 �e2 7 �e4 .t g5 (7 ....t e l 8 .td8 transposes) 8 .t c7 .t d2 (8 . . ..te3 9 �d5 �d3 1 0 .t d6! .t b6 1 1 �c6! i s easy) 9 �d4 .tel 10 �c4 .td2 1 1 �c5 transposes to line 3. 2) s �e2 6 �e4 .td2 (6... .te l 7 .t d8 .t f2 transposes) 7 .tc7 .tel 8 �d4 also transposes to the line 3. 3) s .tel 6 �e5 ! �e2 7 �e4 ! (not 7 �d5? �d3 !, and White cannot avoid a drawn vertical opposition; White must keep the enemy king at arm's length while he improves the position of his bishop) .t g3 (7 ....th4 8 .t c7 .t e l is the same) 8 .t d8 .t f2 9 .t c7 (not 9 �d5? �d3 ! 1 0 .te7 ...
...
...
.t b6! 1 1 �c6 .t a5 !, but now l O �d.S is a threat, because at the end of this line White has .tc7) .t e l (9 . . . .t gl 1 0 �d5 �d3 1 1 .td6! .t b6 12 �c6!) 1 0 �d4 (once again not 10 �d5? �d3!) .td2 ( 10....tf2+ 1 1 �d5 �d3 12 .t d6 ! is a simple win) 1 1 �c5 �d3 and now play is governed by the analysis of the previous diagram_ White has to play his king to a6. drive Black's bishop onto the a5-d8 diagonal, and then manoeuvre his bishop to a5, all without allowing Black to set up a drawn vertical op position with . . . �a4. There are two possibilities: 3a) 12 �c6? (the natural move. but it only draws) .te3 ! 1 3 .t b6 .t g5 ! ( 1 3 . . ..t d2? 14 .t f2 .t a5 15 � b7 .td8 16 �a6 �c4 17 .te l ) 14 .t f2 .td8 ! 1 5 � b7 �c4 1 6 �a6 � b3! ( 1 6 . . . �b4? 1 7 .t b6 ! .th4 1 8 .t a5+) 17 .te l �a4 ! with the stand ard draw. 3b) 12 �b6! (it appears odd to block the pawn, but this is the short est route to a6) .te3+ (or 12 ... .te l 1 3 �a6 .t f2) 1 3 �a6 ! .t f2 (B lack keeps his king on d3 to try to ob struct the transfer to a5, but White wins using the manoeuvre analysed in the previous diagram) 14 .t b6 .t g3 15 .t g l .t c7 (Black is on the short diagonal, so White only needs to reach a5 with his bishop) 1 6 .t f2 �e2 17 .tc5 �d3 1 8 .t b4 �c4 19 .ta5 .tb8 20 b6 �c5 2 1 b7 �c6 and wins as in diagram 233. With White to play matters are much simpler. After 1 �h6 �h2 ! 2 �h5 �h3 ! 3 �g5 �g3! 4 �f5 �f3 ! Black retains the opposition, and
.t +L!i
White has to make a serious conces sion: either to allow Black's king to e4, or to permit Black's bishop to occupy a diagonal restraining the pawn. After, for example, 5 .tc5 .t a5 6 �e5 �e2 7 'iti>d5 �d3 8 .td6 'ifi>c3 9 �c6 �c4! Black draws easily. Now readers may like to examine diagram 255, which includes a more detailed explanation of the logic be hind this type of position. Now suppose that the pawn is on b4. We start with one of the most fa mous unjustified resignations in the history of chess.
. �� . . • • • • • • • • • • ••• �f�@• • • • m • • • • • • • • • • =I= Capablanca-Janowski New York 1916
246
(246): After 1 �d5 Janowski could see no defence to the plan of b5, .te5-d6, �c6 and .t c7, and so resigned. Had he known about the vertical opposition then he would probably have found the drawing idea - Black must play his king to c4. Even though this appears to be a lengthy detour, it turns out that the
v
.t 1 75
king arrives in time. The drawing line is l . ..�f4 ! 2 .td4 (trying to ob struct Black's king; 2 .te5+ �e3 ! 3 b5 'iti>d3 ! 4 �c6 'ifi>c4 ! is also drawn) �f3 ! 3 b5 (or 3 .tc5 �e2! 4 �c6 �d3 ! 5 �d7 .t g5 6 b5 �c4 !) �e2! 4 �c6 �d3 ! 5 .tb6 .t g5 6 �b7 (6 .tc7 i.e3 ! 7 .td6 �c4 !) �c4 7 �a6 (now Black must take into the ac count the analysis given in the pre vious two diagrams) �b3 ! (delay would be fatal; 7 ... .th4? 8 .te3 �d3 9 .t g l .td8 10 .t f2 loses as in line 3b from the previous diagram) 8 .tf2 .td8 ! 9 .tel �a4 ! with a drawn ver tical opposition. A similar position arose in Taima nov-Fischer, Buenos Aires 1960, but Fischer found (or already knew) the drawing idea. If White's bishop starts on d2, then he wins after 1 'iti>d5 !, for exam ple l . .. �g4 2 b5 'ifi>f3 3 �c6! �e4 (Black cannot gain a tempo by at tacking White's bishop, as 3 . . .�e2 loses to 4 .t f4 �d3 5 .tc7 !) 4 .tel �d3 (otherwise .t g3-c7 wins) 5 �b7 ! .tg5 6 .tf2 'ifi>e2 7 .t gl .td8 8 �a6 �d3 9 .t f2 winning as in the analysis of diagram 244. The following practical exam ples were also inaccurately played. (247): This differs from the pre vious position in that the bishops are light-squared, so if the pawn crosses b5 Black will only be able to draw if his king can move in front of the pawn. White may continue: 1) 1 .tdS? (the move played in the game) �f6 ! (Black heads for d8; l ...'ifi>f5? loses after 2 .tc6 .tc8 3 b5 �e6 4 .tg2 �e7 5 �b6 �d6 6 �a7)
1 76 .i. +� v .i.
+I=
247
• • • • . . . • • • • �� d .a. � • d • d D • • • • •-*-• • • • • • • • • • I=
248
Janosevit-lvkov Yugoslav Ch 1972
Dahl-Borisova Pohja 1 985
2 .i. c6 (2 'ili>d6 .i.b5 ! 3 .i.c6 .i.a6) .i. g4 3 'ili>d6 (3 b5 'ili>e7 ! 4 b6 'ili>d8 !) .i.e2 ! 4 .i.d5 .i.fl 5 'ili>c5 'ili>e7 6 .i.c4 .i.h3 7 .i.a6 'ili>d8 8 'ili>b6 .i.g4 9 'ili>b7 .i.c8+ 10 'ili>a7 .i.d7 ! 1 1 'ili>b8 .i.c6 1 2 .i.c8 .i.b5 1 3 .i.g4 .i.d3 14 .i.c8 .i.c4 1 5 .i.b7 .i.b5 1 6 'ili>a7 112.112. 2) 1 'ili>d6 (this would have won) .i.e8 2 .i.d5 ! .i.b5 (or else .i.c6) 3 'ili>c5 ! (forcing an interception) .i.fl 4 .i.c4 .i. g2 (Black's king is too far away) 5 'ili>d6 'ili>f4 6 b5 'ili>e3 7 b6 .i.b7 8 .i.b5 'ili>d4 9 .i.c6 .i.a6 1 0 'ili>c7 'ili>c4 1 1 .i.b7 .i.b5 1 2 .i.c8 and the pawn promotes. (248): This is similar to the pre vious position, but Black's bishop is better placed, which should have en abled her to achieve a draw. We fol low the game continuation: 1 .. ..i.c2 (1 ... .i.g6 also draws, but not 1 ... .i.f5? 2 'ili> d6 .i.c2 3 .i.c4 ! .i.a4 4 .i.d5 ! .i. b5 5 'ili>c5 !, winning as in line 2 of the previous diagram) 2 .i.c4 (2 'ili>b6 'ili>f6 3 'ili>a7 'ili>e7 4 .i.c6 'ili>d6 5 b5 .i.a4 ! and 2 .i.c6 'ili>f6 3 b5 'ili>e7 ! 4 b6
'ili>d8 ! 5 .i.b7 .i.d3 are also dead drawn) .i.a4 3 .i.d5 'ili>f6 4 'ili>d6 (4 'ili>b6 'ili>e7 5 'ili>c7 .i.b5 !) .i.b5 ! (Black is a tempo up over the analysis of 1 .. . .i.f5?) 5 'ili>c5 .i.f l ? (a very odd decision; Black could have drawn by simply repeating moves with 5 . . ..i. a4) 6 .i.c4 ! (now White is win ning) .i.g2 7 b5 (7 'ili>d6 was simpler) 'ili>e5 8 b6 (not 8 'ili>b6? 'ili>d4 ! 9 .i.e2 'ili>e3 ! with a perpetual attack on the bishop) .i.b7 9 .i.b5 (the simplest method was 9 .i.b3, putting Black in zugzwang, with an easy victory after 9 . . . .i.a8 1 0 .i.a4 .i.b7 1 1 .i.d7 .i.a6 1 2 'ili>c6) 'ili>e6 1 0 .i.c6 .i.c8 1 1 .i.f3 'ili>e7 1 2 'ili>c6 'ili>d8 1 3 .i.e2 (White gets there in the end) .i.f5 14 'ili>b7 ! .i.e4+ 1 5 'ili>a7 'ili>e7 1 6 .i.a6 'ili>d8 1 7 .i.b7 .i.f5 18 .i.f3 .i.c8 1 9 .i.g4 1-0. There are seven reciprocal zug zwangs with the pawn on b4, but we only have space to mention one: (249): Suppose that White is to move. 1 b5 .i.a4 ! is hopeless, so he may try:
i. +£!. v i. 1 77
moves away from c3 then c5 wins, for example 1 �a3 2 c5 ! i. d3 3 i. d5 a4 4 i. c4 i.e4 5 b5 a5 6 b6 winning as in diagram 234. Thus Black is reduced to 1 i.dl (after I . .. i. b3+ 2 �c5 ! Black cannot stop b5) 2 d4 ! , but this second recipro cal zugzwang is fatal. After 2 ...i.c2 3 i. d7 Black cannot move onto the fl -a6 diagonal, while 2 ...'iii>a3 3 �c5 ! i.e2 4 i.d5 and 2 ...i.e2 3 i.d5 ! �a3 4 �c5 ! �a4 5 i.c4 i.f3 6 b5 a5 7 b6 winning as in diagram 234. ...
...
249
=I-
Original
I ) 1 i. d7 ( I i. f3 i. a4 !) i.d l ! 2 �d4 (2 �d3 �b3 3 b5 b4 4 b6 i. f3 ! is a draw, while 2 �c5 i. e2 ! is similar to line 2) i.e2! 3 i.e6 i.b5 ! 4 �c5 i. e8 (now we can see the defect of White's first move; if the bishop were still on the long di agonal, White would win here with i.c6) 5 i.d5 �c3 with a drawn verti cal opposition position. 2) 1 �cS i. d3 ! 2 i. d5 (2 i.b5 i.g6 draws) c3 ! is a draw. 3) 1 �d4 (this threatens 2 i.d7, followed by b5, and White can meet ...i. a4 by b6) i.d l ! gives rise to a second reciprocal zugzwang. Black is ready to meet 2 i. d7 by 2 ...i.e2! , so White has nothing better than to transpose into the above lines, for example by 2 �c5 i.e2 ! . The main reason White to move cannot win lies in line 2 above. So long as Black is ready to meet c5 by ...i. d3 followed by ...c;l;ic3, White cannot make progress. When B lack moves first he can not avoid a concession. If his king
Now we move on to the case of the pawn on b3. This situation has given rise to several instructive prac tical examples, which we can ana lyse using the theory developed earlier in this section.
+I=
250
Penrose-Franklin British Ch 1961 (250): White was to play in the game, and he won as follows: I i.d l i. d3 (after I . ..i. c8 2 i.c2 b6 3 i. f5 ! i.a6 4 g4 ! �c7 5 g5 d6 6 g6 !
1 78 .t +,0, v .t �e7 7 �g5 .t c4 8 .t c2 .t d5 9 g7 .t c4 1 0 �h6 .t g8 1 1 �g6 White wins as in diagram 233) 2 g4 �b8 3 �e5 (3 g5 �c7 4 .t g4 ! �d6 5 .t f5 ! .t c4 6 g6! �e7 7 �g5, transposing to diagram 233, would have been considerably simpler) �c7 4 �f6 � d8 5 .t a4 .te4 6 g5 (this position is winning for White, but the deci sive plan is not entirely straightfor ward: first White plays his king to f8, so that the black king is kept out while White transfers his bishop to h5 ; finally � f7-f6 and .t g6 drives the black bishop away) .i.d3 7 �f7 .t c4+ 8 � f8 .td3 9 .i.e8 .te4 1 0 .t h5 .td3 1 1 �f7 .te4 1 2 �f6 .td3 1 3 .t g6 .te2 (or 13 ....t c4 14 .t c2 �e8 15 �g7 .te2 16 �h7 .tc4 1 7 g6) 14 .t c2 .t h5 15 .t a4 �c7 1 6 .t b3 1-0. Black to play draws easily by 1 . . . 'it>b6 2 .id 1 �c7 3 g4 �d6 4 g5 �e7. Black lost this position partly be cause his bishop started on an unfor tunate square which made it difficult to restrain the pawn by covering g6. In the following example Black's bishop is in a better position. (251): By a curious coincidence this drawn position arose in two games played just a year apart: Kveinys-Vetemaa, USSR 1986 and Bernard-Gizynski, Polish Ch 1987, with White to move in both cases. One game did indeed end in a draw, but Black lost the other one. White may try: 1) 1 .tf5 (this was the continu ation in Kveinys-Vetemaa) .td 1 2 .te4 and now:
25 1
I
= =
l a) 2 ... �b4 (the game continu ation; Black heads for the vertical opposition) 3 .tf3 .t a4 (3 . . ..tc2 also draws, but not 3 . . ..t b3? 4 �e5 �c3 5 g4 ! 'iti>d2 6 g5 �e3 7 .td5 ! .t c2 8 .te6 ! and White wins) 4 g4 �c3 S �e5 �d3 6 .t d5 .te8 (or 6 ... �e3 7 .t f7 .tc2) 7 g5 �e3 ! 8 �f6 �f4 ! 9 .te6 .t h5 ! 10 .td7 .tf7 11z_112. lb) 2 ...�b6 (here Black has an alternative draw which depends on a sharp tactical resource) 3 .tf3 .tc2 4 g4 �c7 ! 5 .te4 .t b3 ! 6 �e5 �d8 7 �f6 �e8 ! 8 g5 .tf7 ! (not 8 ... �f8? 9 g6 ! � g8 1 0 g7 winning as in dia gram 232 with Black to play) 9 .tg6 �f8 ! and the stalemate trick saves the day. 2) 1 .t f3 (the continuation of Bernard-Gizynski is more testing. because Black is forced to go for the vertical opposition) .td7 2 .te4 � b6? (Black tries to defend as in line 1 b, but it doesn't work here; the correct line was 2 ... �b4 ! 3 .tf5 .ta4 4 g4 �c3! 5 g5 .te8 ! 6 �e5 �d2! 7 �f6 �e3 ! 8 .te6 �f4 !) 3 .t f5 ! .ta4 4 g4 ! �c7 5 �e5 (missing a simpler
i. +LS v i. 1 79
win by 5 g5 �d6 6 g6! �e7 7 �g5 ! i. b3 8 i. bl �f8 9 �h6 !) �d8 6 �f6 ! i.d l (after 6... �e8 7 g5 ! �f8 8 g6! i. b3 9 g7+ �g8 1 0 i.d7 White wins as in diagram 232) 7 g5 ! i.h5 8 i.e4 i.e8 9 i. f3 (9 . . . �d7 1 0 i. d5) 1-0. If Black's king starts on a7 instead of a5 then he loses because after 1 i. f3 he is no longer able to use the defence of line 2 - it is simply too far to go round to the rear of White's pawn. Yet another drawn position was thrown away in the following exam ple.
• • • • • • • • • • ••• • • • • . . . � - m • • •-* • . . . • . i. . • I=
252
Ravn-Vaitonis Munich OL 1958 (252): White's bishop is already covering the important g4 square, so he should have no trouble holding the game. To begin with White did defend well: 1 . . . �e3 2 �b2 i. g2 3 i.g4 i.e4 4 �c3 �f4 5 i. d7 g5 6 �d2 i. f5 7 i. c6 (the first signs of faltering, since White makes prob lems for himself with this move; the
simplest line was 7 i.a4 �g3 8 �e3 and Black cannot advance his pawn) g4 and now: 1 ) 8 i. a4 (it is essential to pre vent . . . �g3-h2, hence this move, which arranges to meet . . . �g3 by i. d l ) g3 (8 . . . i.e4 9 �el) 9 �el ! (now that Black's king is blocked out of g3, White can approach with his king; however, when this position arose in the game Sydor-Pokojow czyk, Polish Ch 1976, White played 9 �e2? and lost after 9 . . .i. d3+ ! 1 0 �el �e3 1 1 i.c6 i.e4 ! 1 2 i. xe4 � xe4 ! 1 3 �e2 �f4 ! 14 �el �e3 0- 1 ) i. d3 10 i.c6! �e3 1 1 i. g2 ! with the reciprocal zugzwang of dia gram 236 . 2) 8 �e2? (the game continu ation) �g3 ! (of course; Black simply plays his king to h2, and then there is no stopping the pawn) 9 �e3 �h2 10 i. f3 g3 ! 0- 1 . There are three reciprocal zug zwangs with i. +LSb3 v i. , but we will skip these for reasons of space. Now we move on to the case of the pawn on b2. The following two studies are both unsound, but the Grigoriev is notable, nevertheless. (253): After 1 b4 Black may play: 1) 1...�g5 ? 2 b5 ! i.e2 3 b6! i. a6 4 �b4 �f6 5 �a5 i.c8 6 i. b5 i.b7 7 i. a6 ! i. g2 8 i.c8 �e7 9 �a6! �d8 10 i. b7 i. fl + 1 1 �a7 ! and wins. 2) 1 .. i.e2? (this was Grigoriev's main line; now White can win by very exact play) 2 i.d5 ! (not 2 �d4? �g5 ! 3 i.d5 �f6 ! 4 i.c4 i. g4 5 �c5 �e7 6 b5 �d8) �g5 3 i. c4 ! i. g4 (3 . . .i. f3 4 b5 ! �f6 loses to 5 �b4 ! .
180 i. +L!. v i.
I
253
= =
=I=
254
Grigoriev, 1931 Shakhmatny Listok
Gurgenidze, 1977 2nd HM, Kommunist
q;e7 6 q;a5 ! �d6 7 q;a6 �c5 8 i.n , but not 5 b6? q;e7 6 �d4 �d6) 4 b5 ! � f6 5 b6 ! (now it is the other way round, since 5 �b4? only draws after 5 . . . �e7 ! 6 �a5 i.c8! 7 b6 i. b7 ! 8 i. fl q;d7 9 i.h3+ �d6 10 �b5 �e7 ! 1 1 q;c5 i. f3) i.c8 (or 5 ... i.f3 6 �d4 ! i. b7 7 i.d5 i.a6 8 �c5 �e7 9 q;c6 �d8 10 i.e6 ! i.d3 1 1 �b7) 6 �d4 �e7 7 q;c5 ! �d7 8 i.b5+ ! �d8 9 �c6 ! i.d7+ 10 �d6 ! i.c8 1 1 i.c4 i. b7 1 2 i. e6 ! and White wins by the reciprocal zugzwang of diagram 236. 3) 1 i. dl! 2 �d4 (2 b5 i. a4) � g5 ! 3 �c5 �f6 4 b5 �e7 ! 5 b6 �d8 ! and B lack's king is too fast. (254): The composer's intention was the neat variation 1 b4 �b3 ! 2 b5 �c4 ! 3 b6 q;b5? 4 b7 ! i. f4 5 �xf4 ! �a6 6 b8J:r. ! and wins, but the study is unsound. B lack can draw by 3 . . . i. f8 ! 4 i.c7 (4 �e5 �c5 !) i.c5 ! 5 b7 i.a7 ! 6 i.b6 i.b8 ! 7 i.e3 q;b5 8 q;d5 i. g3 9 i.c5 i.b8 10 i.d4 i.c7 ! and White cannot make progress.
There are three reciprocal zug zwangs in the ending i. +L!.b2 v i., but we will not discuss these in de tail. We end with the longest win in the ending of i.+bl!. v i. (35 moves).
...
/-
255
Original (255): Black may play: 1) 1 i.g7+ 2 q;d3 i.e5 (2 ...i.f6 i. 3 b6! is essentially the same) 3 ...
i. b6 ! i.d6 4 b5 ! (Black loses more
.t +� v .t 181
quickly in this line because his king is badly confined) .t g3 5 .td8 .tf2 6 .t a5 �c l (or 6 . . . .t g l 7 �c4 �c2 8 .t c7 ! .t f2 9 .t d6 .t b6 10 �d5 !) 7 .td2+ ! �b2 8 .te3 ! .t g3 9 b6! �b3 10 .t d2 ! .t f2 1 1 b7 ! .t a7 12 .te l .tb8 1 3 �d4 .th2 14 �c5 and wins. 2) 1 .th6 2 b5 ! .t d2+ 3 �b3 ! .t a5 (the toughest defence) 4 �a4 ! (the first step is to dislodge Black's bishop from the a5-d8 diagonal; White can achieve this with gain of tempo) .td8 5 .t b4 ! .t b6 (the only defence to the threat of .t a5) 6 .ta5 ! .t f2 7 �b3 ! (now White must ex tract his king from a4, but he must be careful to confine Black's king, for example 7 �b4? �c2 8 �c4 .te3 ! draws after 9 .t c7 .td2! or 9 .tb4 .t b6! 10 �d5 �b3 1 1 �c6 .td 8 ! 12 .t d6 �c4) .te3 (7 ... �e2 8 �c4 ! .te3 9 .t d8 transposes to the main line) reaching the key position W�b3, .t a5, �b5 v B �d l , .te3. The basic principle governing this position is that when White's bishop is on c7 he is threatening to advance his king, and Black has to reply by playing his bishop to the e l -a5 di agonal. This prevents b6 on account of ....ta5, and it nullifies the king ad vance, because at a later stage Black can meet .t d6 by . . . .t a5 (and not ... .t b6, which would allow White to gain a tempo with �c6). On the other hand, when White's bishop is on d8, Black's bishop must be on the g l -a7 diagonal (with the bishop on d2, White would win by simply pushing the b-pawn). The detailed analysis runs 8 .tc7 ! (not 8 .td8? �d2!, with a draw after 9 .tc7 �d3 ! ...
or 9 �c4 �c2 ! 10 �d5 �b3 l 1 .te7 .t b6 ! 1 2 �c6 .t a5 !) .t d2 (8 . . . 'ot.>d2 loses to 9 �c4! �c2 10 �d5 �b3 1 1 .td6! .t b6 1 2 �c6! .t a5 1 3 .tc7 !) 9 .t d8 ! (not 9 �c4? �c2 ! 10 �d5 �b3 1 1 .td6 .ta5 ! and White cannot gain time by hitting the bishop with �c6; after the move played White threatens b6) .te3 10 �c3 ! ( 10 �c4? �c2 ! is still bad) �e2 ( 10 ....t f2 1 1 �d3 �c l 1 2 .t g5+ ! wins as in line 1 ) 1 1 �c4 .tf2 ( 1 1 . . ..t g l 12 .tc7 .te3 1 3 .td6 .tb6 14 �d5 !) 1 2 .tc7 .te l 1 3 �d4 (we have transposed to line 3 of diagram 245; the rest is just for the sake of completeness) .t d2 14 'ot.>c5 �d3 15 �b6! .te3+ 1 6 �a6 ! .t f2 17 .tb6 .t g3 1 8 .tg l .tc7 19 .t f2 �e2 20 .tc5 �d3 2 1 .tb4 �c4 22 .t a5 .t b8 23 b6 �c5 24 b7 �c6 25 .t b6 �d7 26 .t f2 �c6 27 .t g l �d7 28 �b6 .t c7+ 29 �a7 �c6 30 �a8 �b5 3 1 .t a7 �a6 32 .tb8 .tb6 33 .tf4 .t a7 34 .tc7 with capture of the bishop or pawn pro motion next move.
3.3:
i.+� v i.
As usual, we start with the pawn on the seventh rank. The first situation to consider is the vertical opposition . According to the Averbakh drawing zone, the result should be a win, but this is not the whole story. In fact, the game is decided by the position of White's king; if it is on b8 then he wins, but if his king is on d8 then the position is a draw. (256): In this case the win pro ceeds as expected: 1 .t f3 .t f5 2 .t b7
182 .t +.0. v .t
+/-
256
Centurini, 1856 .lg4 3 .lc8 .le2 4 .lh3 .t a6 5 .lg2 �c5 6 .lb7 and the pawn promotes. It makes no difference if Black's king starts on d6: White wins by 1 .l a6 .lg4 2 .lc8 .le2 3 .lh3 .la6 4 .t n .
for example 1 .la6 .lg4 2 .lc8 .le2 3 .lh3 .t a6 4 .lg2 �e6 5 .t n .lb7! 6 .le2 �d6! 7 .lc4 �c6 and White is not making progress. Black's king can start on b6 without changing the result, since White cannot prevent Black transferring his king to d6, for example 1 .ld 1 (threat .la4-d7) �c6 2 i. a4+ �d6 ! with the same draw as before. Even if Black's bishop is on the shorter diagonal, the result is still a draw, for example if we have B �b6 and B i. b7 in the diagram, I i. g4 i.a6 2 i.c8 i.e2 3 i. b7 i. g4 ! 4 .l f3 i. f5 5 i. h5 �c6 6 i.e8+ �d6! holds on. The horizontal opposition also depends on the location of White's king. If the situation is �g6 v �e6, then White wins, but reversing the kings results in a draw.
Centurini, 1856 (257): Although it is easy to drive Black's bishop onto the short di agonal, Black cannot be forced into zugzwang because he cannot be de nied a waiting move with his king,
(258): It makes no difference who moves first. The simplest win is to play a waiting move; any reply al lows White's king to reach g8, when he has a winning vertical opposition.
.t +� v .t 183
One possible line is 1 .t b2 .tb4 (or l . . . �d5 2 �h7) 2 �g7 �f5 3 �g8 �g6 4 .t g7 .t c5 5 .tf8 .t e3 6 .t b4 .th6 7 .tc3 winning. If the kings are reversed, White can force his king to e8, but this doesn't help because here the vertical opposition is drawn, for example 1 .t f6 .th6 2 �e7 �f5 3 .tc3 �g6 ! 4 �e8 �f5 and White cannot make progress. In the following position, White cannot head directly for d8 with his king, because the vertical opposi tion is a draw. Greater subtlety is re quired.
259
+/-
Dobrescu, 1980 (end ofstudy) 3rd Comm., Book Jubilee (259): The winning technique is well worth knowing. White is aim ing to establish his king on c5 . This keeps Black's king at arm's length while White prepares the winning manoeuvre .te8, �c6 and .t d7, forcing . . ..t a6, whereupon �b6 traps the bishop. The danger is that Black's king will sneak around the back and set up a drawn horizontal
opposition with �d6 v �b6. We suppose that Black moves first: 1 ...�f3 2 .t f5 and now: 1) 2 .t b7 3 �f6 ! �f4 4 .t d7 �e3 5 �e5 �d3 6 .tc6 (not 6 �d6? �c4 ! 7 .t c6 .tc8 ! 8 .t g2 �b5 ! drawing) .tc8 7 �d5 �c3 8 �c5 transposes into line 2 . 2) 2 .ta6 3 �f6 ! �f4 4 .t g6 (threatening �e6) .t c8 (4 . . . .tb7 5 �e6 �e3 6 �e5 �d2 7 �d6 �c3 8 �c5 .tc8 9 .te8 loses more quickly) 5 .t d3 (now Black is in zugzwang) .t h3 (the other lines are 5 . . ..t b7 6 �e6 �e3 7 .tb5 �d4 8 �d6 ! �c3 9 �c5, with a second zugzwang, and 5 . . . �e3 6 .t f5 ! .t a6 7 �e5) 6 .t b5 .t c8 (6 ... �e4 loses at once because of a tactical point: 7 �e7 .t c8 8 �d8 .t b7 9 .t c6+) 7 .tc6 (once again Black is in zugzwang; he must either allow White's king onto the e-file, or play 7 ... .th3, when 8 �e7 gains a tempo by threatening .td7 ; in either case White's king reaches c5 ) �e3 8 �e5 ! �d3 9 �d5 (not 9 �d6? �c4! 1 0 .t g2 �b5 !) �c3 10 �c5 �d3 1 1 .te8 .t g4 1 2 �d6 .tc8 1 3 .td7 .tb7 14 �c5, followed by �b6, and White wins. (260): In this position White's king can occupy an active position of the c-file. Eventually he can pene trate to b8 and achieve a winning vertical opposition. We assume that Black moves first: l . . .�a5 2 �c5 .t f5 3 .tc4 (White wants to play �c6 without allowing an annoying check on the long diagonal; this move prepares an interposition by .td5) .td7 (or 3 ... .th3 4 �c6 .tc8 5 .tn , as in the main line) 4 �d6 (now .••
...
184 .t +�
260
v
.t
+I-
Herberg, 1943 (end of study) Deutsche Schachzeitung Black has to allow the king in to c6) .t c8 5 �c6 �b4 (5 . . . .th3 6 �b7 �b4 1 .t n .t g4 s �b8) 6 .tn �a5 7 .t e2 (Black is in zugzwang and must allow the king to reach b6 or b7) �b4 8 b6 .th3 9 �b7 .t f5 10 �b8 �a5 1 1 .t f3 �b6 12 .tb7 with the usual vertical opposition win. There are 1 1 reciprocal zug zwangs with .t +�c7 v .t . The fol lowing position is based on one of them. (261): How should White defend his bishop? The alternatives are: 1 ) 1 �e2? .tg4+ ( 1 ....ta6+ 2 �e l .t c8 ! leads to the same position) 2 �e l .t c8 ! (White has fallen into a reciprocal zugzwang) 3 .te2 (White has no king moves, and 3 .ta4 �b2 4 �d2 �a3 ! 5 .td 1 �b4 ! allows Black to gain a tempo with his king, and thereby round up the pawn) �c2 ! 4 �f2 d2 ! (a second reciprocal zug zwang) 5 .t f l (once again, moving the other way gives Black a free tempo: 5 .t b5 �c3 ! 6 �e3 �b4 !)
+I=
261
Original .t b7 ! (reciprocal zugzwang number three) and now: l a) 6 .tg2 .t c8 ! 7 �f3 �d3 8 �f4 �d4 9 .t f3 �c5 1 0 �e5 .th3 and White cannot make progress. l b) 6 .t c4 �c3 ! 7 .t d5 .tc8 8 �e3 �b4 9 �d4 �b5 is also drawn. l e) 6 .t bS �c3 ! 7 �e3 �b4 ! 8 .t c6 .tc8 9 �d4 �a5 1 0 �c5 �a6! prevents the white king penetrating to b8, and so leads to a drawn hori zontal opposition. 2) 1 �el! (now Black is in the re ciprocal zugzwang) and now: 2a) 1 .td7 (Black gives up the possibility of checking on b7) 2 .te2 ! �c2 (2 ....tc8 3 .t n �c2 4 �e2 �c3 5 �e3 ! wins as in line 2b} 3 � f2 ! �d2 4 �f3 ! (this is the key difference; a check on b7 would force the king to return to f2, but White can meet a check on c6 by �f4) .te6 5 .tn and White's king is free to advance. 2b) 1. .t b7 2 .te2 �c2 (2 ....tc8 .t 3 n �c2 4 �e2 and 5 �e3, win ning as in the main line) 3 �f2! �d2 ...
••
.i. +�
(trying to prevent �e3) 4 .i.f l ! (re ciprocal zugzwang; the bishop can not move in the other direction for the usual reason that Black would gain a tempo with his king) �c2 (Black has to allow White's king to reach e3 or f3) 5 �e3 �c3 6 .i.e2 (by repeating the manoeuvre, White's king eventually forces its way to c5) �b3 7 �d4 �b4 8 .i.d3 .i.c8 9 �d5 .i.b7+ 10 �d6 and now Black has to allow the king to reach c5 or c6, when White wins as in the previous diagram.
v
.i. 185
followed by .i.b6. However, this win was only possible because Black's bishop started on a very bad square. If Black's king is on d5 and his bishop on f4, then the position is a draw after 1 .i.a5 �c4 ! 2 .i.c7 .i.d2 3 .i.g3 .i.aS ! 4 .i.e l .i.d8 ! 5 .i.f2 �b5 ! and Black reaches safety. With the kings on d7 and d5, the position is completely drawn. How ever, White can win with �d7 v � b5, provided B lack's bishop is already on the short diagonal:
Now suppose that White's pawn is on c6. The results with the vertical opposition are significantly different from the case of the c7-pawn.
263
262
=I=
(262): This is diagram 256 shifted down a rank. It is drawn because the shorter diagonal (a5-d8) contains four squares and so B lack cannot be put into zugzwang. But if we move Black's king from b5 to d5, then White to play wins by 1 .i.aS ! .i.g3 2 i.c7 ! .i.e 1 3 i.h2 .i.aS 4 .i.g l ,
+I=
(263): This position is drawn if it is shifted up a rank, but here White wins by 1 .i.c7 ! .i.d2 2 .i.d8 (making use of the extra space) .i.f4 3 .i.e7, followed by .i.d6. If Black's bishop starts on the long diagonal, then the position is generally drawn, for ex ample with White's bishop on e l and Black's bishop on h2, White to play cannot prevent ...�c4 and ...�d5. The conclusion is that the verti cal opposition is a draw with the pawn on c6, but White can win cer tain favourable cases of the diagonal
186 .t+� v .t
opposition. Not surprisingly, the horizontal opposition is completely drawn; it doesn't make any differ ence whether White has �d5 v �b5 or �b5 v �d5.
265
=I=
264
Chiron, 1952 (264): The CMron and Dehler re ciprocal zugzwang of diagram 236 breaks down with a c-pawn. After l .. ..i.b6 2 �d6 (or 2 �d5 .tc7 3 �c5 �f7 4 .th4 �e6 5 �b5 �d5 with the horizontal opposition) .td8 3 .td4 Black has an extra square and so can not be forced into zugzwang. The position is a draw after 3 ....ta5 ! 4 .tf6 .td8. The underpromotion ideas of dia gram 238 can also occur with a c pawn. (265): Here White to play wins by 1 c7 ! .td8 and now either 2 c8.t, with the lines 2 . . . .th4 3 .td7+! and 2 ....tb6 3 .te7 !, or 2 c8tlJ .tc7 (after 2 ... .th4 3 tLJ
+I=
then the analysis is much the same; the only extra possibility is 1 c7 ! .td8 2 c8tlJ .ta5, but then 3 tLJ
.t+� v .t 187
White continues 3 .i.f4? �a7 ! 4 .i.d6. Then if Black plays 4 .th4? he loses to 5 �c5 �a6 (or 5 ... .td8 6 .tg3 �a6 7 �d6! �b5 8 �d7 !) 6 .i.f4 .i.e7+ 7 �d5 ! .i.d8 8 �d6 �b5 9 �d7 ! , and we arrive at the key po sition W�d7, .i.f4, �c6 v B�b5, .i.d8 for the whole study. If Black plays 9 . . . �c5, White wins by 1 0 .i.c7 ! , while 9. . ..ta5 loses to 1 0 .i.c7 ! .i.el 1 1 .i.d8 .i.g3 12 .i.e7, fol lowed by .i.d6. Black's problem in this line was the position of his bishop on d8, which prevented him improving the position of his king by 9 . . . �c4, because White could simply take the bishop. Had Black's bishop been on a5 and not d8, then by playing . . . �c4 and . . . �d5, Black would arrive at a drawn vertical op position. This explains why Black can draw with the correct defence 4 .tel! 5 �c5 �a6 6 .i.f4 .i.a5 ! 7 �d6 �b5 ! 8 �d7 �c4 ! 9 .tc7 .i.el 10 .i.d8 .tg3 ! 1 1 .te7 �d5 ! and White has to settle for half a point. The move played threatens �c4d5, because ...�a7 is ruled out by the reply .i.c5+. 3 .tb6 A cunning defence. Black cannot play 3 . . ..tel because White simply promotes the pawn with 4 �a6, while 3 . . ..th4 4 �c5 �a7 5 .i.f4 transposes to the losing line in the previous note. 4 .tr4 White aims to transfer the move to Black, which has the curious effect of gaining a tempo ! After 4 �c4 �a7 5 �d5? (White can still win by back tracking with 5 �b5 ! , but also not •••
+I=
266
Original lose a tempo at the critical stage with a bishop 'triangulation' . The winning line runs: 1 .tc7! This is a position of reciprocal zugzwang. It is clear that it would be drawn with White to play, because the only move would be 1 .i.d8, but then l ....i.g3 and 2 ...�b8 draws eas ily. 1 .ta7+ 2 �b5 White can also play 2 �a6, but this just transposes into the main line after 2 . . . .tc5 3 �b5 ! .tf2. White's only winning chance is to play his king round to d7, but he must take care not to allow B lack's king to become active, otherwise a drawn diagonal or vertical opposition will result. 2 .tf2 3 .td6! White must not waste time. If he allows Black's king to occupy a7, then the position is drawn, but the reason why is rather subtle. Suppose ...
..•
••.
...
,
188 .t+� v .t
5 .tc5? �a6 ! ) 'iii>a6 ! 6 �e6 �b5 ! 7 �d7 'iitc4 ! Black is just in time. 4 �a7 White wins slightly more easily after 4 . . . .tf2 (4 . . . .ta7 is met by 5 .tg3 .tb6 6 .td6 transposing to the main line, but not 5 �c4? .tb8 ! drawing) 5 �c4 �a7 6 �d5 ! �a6 7 �d6 .tb6 8 �d7 ! �b5 9 .tc7 ! .tf2 10 .td8 .tg3 l l .te7 and Black is too late. 5 .tb8+! The key move, allowing White to lose a tempo. 5 �a8 6 .td6 Now Black is in zugzwang. He cannot play 6 . . . �a7 because of 7 .tcS, so he has to move the bishop. Playing it off the a5-d8 diagonal loses two tempi, because it has to re turn later, while ... .td8 (the only available square on the a5-d8 diago nal) fails because White's later �d7 will attack the bishop. 6 .td8 Or 6 . . ..tf2 7 'iitc4 .th4 8 �d5 1;a7 9 'iii>e6. 7 �cS �a7 Now White wins by 8 .tg3 .te7+ (8 . . . 'iii>a6 9 'iitd 6! �b6 1 0 �d7 ! �b5 wins as in the main line) 9 �d5 .td8 1 0 'iii>d6 �a6 1 1 �d7 �b5 12 .td6 (the fatal moment; Black cannot play . . . �c4) .ta5 1 3 .tc7 ! .tel 14 .td8 .tg3 15 .te7 followed by .td6. ••.
situation is quite interesting, so we will spend some time on it. First of all, virtually all 'opposition' posi tions, horizontal, vertical or diago nal, are drawn. The following case is an exception:
•••
.••
There are 1 7 reciprocal zug zwangs with .t+�c6 v .t, but we have already seen one in the previous diagram, so I will move on directly to the case of the pawn on c5. This
+I=
267
Original (267): White to play wins by I .tc6! .th5 2 .tb7 .te8 3 .ta6 (Black has no time to transfer his king to b4) .td7 4 .tbS .tc8 5 c6 �d5 6 c7 �d6, with a winning horizontal opposi tion. Black to play draws by 1 ...�c3! 2 .tc6 .th5 3 .tb7 .te8 ! 4 .ta6 �b4!. The following position has con siderable practical importance. (268): In the game, B lack failed to appreciate the danger and played 1 . . ..td4?, when White won by 2 .th6 �e7 (Black actually resigned at this point) 3 .tg7 .tc5 4 f6+ �e6 5 f7, when White has a winning hori zontal opposition. The correct de fence was 1 .. .�f8 ! 2 .tg5 �g8 3 .tf6 .te 1 4 .td4 .th4 ! 5 .te3 �f8 and White cannot make progress.
.t+l!. v .t 189
268
+I=
Kurajica-Markland Hastings 1971 This position again emphasises the point that defending with the king to the side of the pawn offers relatively few chances of saving the game. If Black's king were on h3 instead of e8, the position would be drawn whoever moves first.
269
-I+
Sokolsky-Lipnitsky Moscow 1 950 (269): Black was to play in the game, and he won using the idea of
the previous diagram: l . ..�e4 2 �g2 (in this position, White is too slow to play his king to the rear of Black's pawn, for example 2 �g4 'iti>d3 3 �f5 .te3 4 �e5 .td2 ! ) �d3 3 �fl .tc3 (the first step is to drive White's bishop onto the long diagonal) 4 .td8 .tb4 5 .tf6 �c2 6 �e2 �b3 (now White cannot prevent Black winning just as in the previous dia gram) 7 .te5 (or 7 �dl .ta3) .ta3 8 .tf6 .tb2 9 .tg5 c3 10 .th6 c2 1 1 �d3 .ta3 12 .tg5 �b2 1 3 �c4 �bl 14 �b3 .tc 1 and wins. Curiously enough, it doesn't help White if he moves first. After 1 �g4 'iti>e4 White cannot make use of his extra tempo to move his king nearer the queenside, while after 1 'iti>g2 �e4 ! 2 �fl �d3 ! White cannot play 3 �el because of 3 . . . .tc3+, so again the extra move doesn't help. White to play would draw if his king had started on h5, because then he would have time to reach the rear of Black's pawn: 1 'iti>g6! �e4 2 'iti>f7 ! �d3 3 �e6 !, and so on. The following position is also im portant. (270): This position could have arisen from the Sokolsky-Lipnitsky game given above after inferior play by Black. In order to draw, White to play must prevent Black achieving the winning formation of king on b3 and bishop on a3. This is possible provided he plays accurately: 1 .tf6! .ta3 (the threat is ....tb2, forcing the bishop off the long diagonal, and then . . ..td4 followed by . . . 'iti>b2) 2 .tc3 .tb2 and now Black must find the right move:
190 �+8 v �
• • • • • • • • • • • • • • • m -·· . . • • • • • • • • •••@• • 270
=I+
1 ) 3 �aS? �f6 4 �d2 �b2 ! 5 �b4 �d4 6 �a5 �b3 7 �c 1 (or else Black plays ...�b2-a3, winning as in the previous diagram) �c3 ! 8 �c7 �b2+ 9 �dl c3 10 �f4 c2+ 1 1 �e2 �a2 with a standard win. 2) 3 �d2! �c l + 4 �d l ! �a3 5 �d2 ! �b2 (Black attempts to put White in zugzwang) 6 �b4 �f6 (threatening . . . �b2) 7 �c3 ! �g5+ 8 �d 1 ! �c 1 (Black makes another zugzwang attempt) 9 �b4 �b2 (Black finally gets his king to b2, but only at the cost of a miserable bishop position) 10 �e7 �f4 (threatening to win with 1 1 ...�e5) 1 1 �f6+ ! �b3 (Black's king is now on the required square, but his bishop is on the wrong diagonal: c l -h6 rather than c l -a3) 12 �g7 �d6 (the final at tempt; Black threatens ... �a3) 1 3 �c 1 ! and draws as in diagram 268. (271): This position arises near the end of a 'White to play and draw' study. Black is to move, and Tim man's analysis continued 1 ...�d3? 2 �g6 �e2 3 �h4 ! f3 4 �f5 ! �f2 5
271
=I+
Tzmman, 1990 (end of study) Schaakwerk I �e7 �el 6 �c5 ! �d2 7 �e4 ! with a standard draw. However, Black can win by means of the unlikely-look ing move 1 . . .�f5 ! . The point is that White's king is held back while Black improves his position by ad vancing his pawn to f3 and moving his bishop to h2. Then Black wins by playing his king to g2. The analysis runs 2 �h6 (2 �g7 f3 3 �f7 �gl 4 �e7 �g4 5 �e6 �h2 wins the same way) �g4 ! 3 �g7 f3 4 �g6 (4 �f6 �g5+) �gl 5 �h6 �h2 6 �f2 �h3 ! 7 �g5 �g2 !, followed by either . . . �g3 or . . .� g l , and Black wins. It follows that the study is unsound. It's time for the reciprocal zug zwangs to put in an appearance. There are 1 3 altogether in �+8c5 v �; the following analysis contains four of them. (272): Some introduction is nec essary, otherwise the play probably won't make much sense. Leaving the queenside pieces fixed, there are five reciprocal zugzwangs based on
j,+,0, v j, 191
+I=
272
Original different king positions. They are �h4 v �h2, �g4 v �g2, �g3 v �gl , �f4 v �f2 and �h3 v �h i . The last o f these plays no part in the analysis, but the other four are all important and the most crucial of all is �f4 v �f2. because virtually all lines lead to it. Let's consider the �f4 v �f2 po sition more carefully. It is easy to un derstand that White has a problem when he moves first; a king move gives Black access to e3, while a bishop move along the a4-e8 diago nal allows ... �e2. A bishop move on the other diagonal allows . . .j,c6, when Black has the added flexibility ofrestraining the pawn along either of the two diagonals passing through c6. It is harder to grasp why Black to play should not simply move his bishop on the long diagonal. The an swer lies in the manoeuvre j,d7-e6d5. Suppose, starting from �f4 v �f2, the moves 1 �e5 �e3 2 j,d7 �d3 3 j,e6 �c3 4 j,d5 occur. If Black's bishop is on b7, then he can
move to a6, retaining control of c8. If the bishop is on, say, g2, it can only cover c8 by moving to f1 and a6 ( . . .j,h3 can be met by a later j,e6). This costs Black an extra tempo, which decides the game. The analysis is very complex, so I will limit myself to the critical lines. 1 �h4! Not 1 �g4? �g2 ! (now White is on the wrong side of the reciprocal zugzwang chain) 2 �f4 �f2 ! 3 �e5 (or 3 j,d3 j,c6 ! 4 j,e4 j,a4 5 j,d5 �e2) �e3 ! 4 j,d7 �d3 5 j,e6 �c3 ! 6 j,d5 �b4 7 �d6 j,a6! (Black gains a tempo because he already controls c8 and so his king can head back without loss of time) 8 c6 �b5 ! 9 c7 �b6 with a drawn horizontal opposition. �g2 1 Most other lines eventually lead into the �f4 v �f2 reciprocal zug zwang: l ) 1 j,f3 2 j,a4 (but not imme diately 2 �g5? �g3 ! 3 �f5 �f2 ! 4 �f4 j,b7 ! and White is in zug zwang; before White plays �g5, he has to make a preparatory anti-zug zwang move) �g2 3 �g5 ! �g3 4 �f5 ! �f2 5 �f4 ! j,b7 6 j,b5 ! and Black has fallen into the �f4 v �f2 zugzwang. 2) 1 �hl 2 �g5 (actually 2 �h3 wins more quickly, but for our purposes it is enough to prove that Black cannot avoid reaching �f4 v �f2) �g l (2 . . . �g2 3 �g4 ! is an other transposition) 3 �f5 ! �f2 4 �f4 ! reaching the main line. 3) 1 j,c8 (this actually holds out longest, but it allows White to ...
...
...
...
192 i.+� v i.
win in a more technical manner) 2 i.c6 ! (this is yet another reciprocal zugzwang, this time with a slightly different arrangement on the queen side) �gl (2 . . . i.a6 3 �g4 �gl 4 �f4 �f2 5 i.a4 i.b7 6 i.b5 ! is the standard reciprocal zugzwang) 3 �g3 ! �fl 4 'ii?f3 ! �el 5 �e3 ! i.h3 (5 . . .i.a6 6 i.a4 i.b7 7 �d3 i.g2 8 i.d7 i.fl + 9 'ii?c 3 i.g2 10 �d4 transposes) 6 'ii?d3 i.fl + 7 �c3 i.h3 8 i.a4 i.g2 9 i.d7 i.f3 10 �d4 �d2 1 1 i.e6 ! i.c6 1 2 �e5 ! �c3 1 3 �d6! i.b7 1 4 i.d5 i.a6 1 5 c6 ! wins as in the main line. 2 �g4! � 2 ...i.f3+ 3 �f4 ! 'it>f2 4 i.d7 i.b7 5 �e5 �e3 6 i.e6 ! transposes to the main line. 3 �f4! i.g2 Black has to make a concession. The alternatives are: 1 ) 3 'ii?e l 4 i.d3 i.c6 5 �e5 ! 'ii?d2 6 'it>d6 ! i.b7 7 i.c4 ! 'ii?c 3 8 i.d5 ! transposes to the main line. 2) 3 i.a8 4 �e5 �e3 5 i.d7 ! (but not 5 i.e8? i.b7 6 i.f7 i.c6 ! 7 i.e6 i.a4 8 �d6 �d4 ! and Black draws) i.b7 6 i.e6! �d3 (6 ... i.c6 7 �d6 !) 7 i.d5 ! reaches the main line yet again. 4 �eS 'ii?e3 ••.
•.•
5 i.d7! 6 i.e6!
�d3
�c3 Now White wins by 7 i.d5 ! 'ii?b4 8 c6 i.fl 9 �d6! i.a6 10 i.g2 �a5 1 1 'ii?c5 ! i.c8 1 2 i.fl ! i.g4 1 3 c7 i.f5 14 i.c4 and we have transposed to diagram 260. The remaining c-pawn positions are largely based on practical ex amples. The one point that emerges
clearly is how often drawn positions are thrown away by players repeat ing the 'Janowski error' , i.e. not playing the king to the rear of the enemy pawn.
I
273
= =
Smirin-Vaiser Tel Aviv 1992 (273): Black's pawn is not far ad vanced, and he has yet to activate his king, so the position should be a clear draw. Nevertheless, White lost it in short order: 1 ...�g5 2 �b4 �h4 3 �c3 (not actually throwing the draw away, but it is an alarming sign that White shows every intention of heading for el with his king) �g3 4 i.d5 f4 5 'ii?d 2? (this is one square too far; it was White's last chance to change direction and draw by 5 �d4 ! i.f5 6 �e5 i.g4 7 �f6 ! i.f3 8 i.b3 i.c6 9 i.dl ! i.d7 10 �g5 !) i.f5 (suddenly White is lost; after 6 �el i.h3 Black wins by diagram 268) 6 �c3 i.h3 7 �d4 i.g2 8 i.c4 and White resigned, since 8 . . . f3 leads to a decisive horizontal opposi tion.
i.+l!i
The following game is another example of the 'Janowski error' .
• • • • • • • • • • • • m �� n n . . . • �+��- • �t; � . w ,?0; w�� R • • • • • • • • 274
=/= Koen-Botsari Manila worn OL 1992
(274) : This position is drawn, but with White to play it requires accu rate defence by Black. 1 i.e5 i.h4 ! (not 1 . ..i.e l ? 2 f4 ! �b6 3 f5 i.h4 4 �d5 ! 'it>b7 5 'it>e6 and White wins as in diagram 269) 2 f4 and now there are a number of possibilities: 1) 2 'it>b5? 3 'iitd5 ! 'it>b4 4 f5, followed by 'it>e6 and i.d6-e7, win ning for White. 2) 2 'it>a4? 3 'it>f3 ! (a surprising winning move; after 3 'it>d5 'it>b3 ! 4 f5 'it>c 2 ! 5 'it>e6 'it>d3 ! Black reaches the typical drawing position) 'it>b3 4 �g4! i.d8 5 'it>h5 'it>c4 6 'it>g6! 'it>d5 7 i.f6! i.a5 8 f5 'it>e4 9 i.g7 i.d8 10 i.h6 and White is just in time; Black cannot prevent i.g5, followed by the pawn's advance. 3) 2 'it>b6? (played in the game) 3 'it>d5 ! 'it>b7 4 'it>e6 �c6 (4 ... 'it>c8 was a tougher defence, but White would win as in diagram 269) 5 i.g7
v
i. 193
(5 f5 was considerably faster) �c7 6 i.f6 i.g3 (loses at once, but even 6 ...i.f2 7 f5 'it>c8 8 i.g5 i.d4 9 'it>f7 would lose by diagram 269) 7 f5 ! 'it>c8 8 i.e7 i.f4 9 f6 i.g5 10 f7 i.h6 1 1 'it>f6 'it'd? 12 i.a3 1-0 . 4) 2 'it>b4! (the unique drawing move) 3 'it>d5 (after 3 'it>f3 'it>c4 4 �g4 i.d8 5 �f5 'it>d5 Black's king is in time) 'it>b3 ! 4 f5 'it>c2! 5 �e6 'it>d3 ! 6 i.f6 i.e l 7 i.e7 i.c3 ! 8 'it>f7 �e4 9 'it>g6 �f3 ! 10 i.f8 'it>g4 with a drawn vertical opposition . •••
There are six reciprocal zug zwangs with i.+l!ic4 v i. and seven with i.+L'!ic3 v i., but none at all with i.+L'!ic2 v i.. We will not be ex ploring these. This section ends with the longest win in the ending of i.+cl!i v i., in deed the longest win in the whole ending of i.+l!i v i. (40 moves). The analysis of this win is fairly compli cated, so we will start with two pre liminary positions.
...
•.•
.•.
+/-
275
Original
1 94 .t +�
v
.t
(275): Firstly suppose that Black moves first. All the available moves weaken Black's position: 1 ) 1 ..'iii>bl (this allows White to gain a tempo with a bishop check) 2 .ib7 (threatening to win by 'ifi>b4) .ie6 (2 . . . .ie8 3 'iii>b4 'iii>c2 4 'iii>a5 'iti>c3 5 .i a6 ! 'iii>d4 6 'iii>b6 !, followed by .ib5, wins for White) 3 .ie4+ (3 c6? .id5 !) c;i;>a2 4 c6 'iii>a3 5 .ic2 .id5 6 c7 ! .ie6 7 .id l .ic8 8 'iii>c4 .ih3 9 �b5 .id7+ 1 0 'itr>b6 .ic8 1 1 c;i;>a7 and the king penetrates to b8. 2) 1....ie8 (if Black is unable to move his king, he can only move the bishop along the a4-e8 diagonal; playing to a4 or b5 gives White a free tempo after c;i;>b4, so this is the tough est defence) 2 i. b7 (threatening to win with 'itr>b4) and now: 2a) 2 c;i;>a2 3 'ifi>b4 ! 'itr>b2 4 'it>a5 �c3 5 .ia6! winning as before. 2b) 2 .tf7 3 .ia6 .id5 (3 ... .ie8 4 'it>b4 .ic6 5 ct>a5) 4 'iii>d4 .ic6 5 'iti>e5 ! c;i;>b2 6 c;i;>d6! .i f3 7 .ic4 !, fol lowed by .id5. 2c) 2 .ih5 3 .ic8 (a surprising move) .if3 (after 3 ....ie8, White can still win by 4 'itr>b4 c;i;>b2 5 c;i;>a5 c;i;>c3 6 .i a6 !) 4 c;i;>d4 'ifi>b2 5 .ie6! .ic6 (5 ... c;i;>a3 6 .id5 ! .ih5 7 c6! c;i;>b4 8 c7 .i g4 9 .if3 .ic8 10 .ie2 ! .ih3 1 1 c;i;>d5 c;i;>a5 12 c;i;>c5 .if5 13 .ic4 wins as in diagram 260) 6 'itr>e5 ! c;i;>c3 7 c;i;>d6! .ib7 8 .id5 ! transposing to the main line of diagram 272. If White is to move, then he has to transfer the move to Black by 1 .ib7 (now White threatens to win by 'iii>b4, as in line 2 above) .ig4 2 .ie4 (threat c6) .id7 3 .id5, and then White wins as above. .
••.
•..
•••
+/-
276
Original (276): Suppose that White is to play; his winning plan is not par ticularly lengthy, but it is very hard to find. The decisive manoeuvre in volves playing .ia6, but first he must force Black's king to the inferior square d l . The main line runs 1 .id3 (now Black is in zugzwang; the im mediate 1 .ia6 .ig2 2 .ib5 'itr>bl is heading in the wrong direction White can still win by backtracking with 3 .ic4, but if he goes any fur ther by 3 .id7?, then 3 ...'itr>a2 ! draws) and now: 1 ) 1 .ig2 ( l . . . .ib7 2 c;i;>d4 is the same) 2 �d4 .ic6 3 c;i;>e5 c;i;>d2 4 c;i;>d6 ! i.b7 5 .ic4 ! 'iti>c3 6 .id5 ! transposes to the main line of dia gram 272. 2) 1 .id7 2 'iti>b4 .ic6 3 'itr>a5 'iitd2 4 c;i;>b6 .if3 5 .ia6! wins easily. 3) 1...'ifi>dl 2 .ia6! (now White is threatening to win by 'itr>b4, gaining a tempo when the king arrives on b6) .ig2 3 .ib5 (thanks to the White's first move, Black's king remains bot tled up) .i f3 4 .id7 �e2 5 'iti>d4! .••
•••
i.+l!. v i. 195
�d2 6 i.e6 ! i.c6 7 �e5 ! �c3 8 �d6 ! i. b7 9 i.d5 ! again transposing to diagram 272. Suppose now that B lack moves first. The only new line is l . . .�bl ( 1 ... �d 1 2 i.a6 ! transposes into the White to play analysis), but then 2 i.e6 i.e8 3 i.d5 �al (3 . . . 'ili>c l 4 i.b7 i.h5 5 i.c8 i.f3 6 �d4 wins as in line 2c of diagram 275) 4 i.b7 transposes to diagram 275, line 2. Now we can deal with the longest win.
+I=
277
Original (277): The winning line runs 1 �b4! (first of all, White must im prison Black's king) i.c6 ( l ...i.b3 2 c5 ! i.d5 3 �a3 ! i.c6 4 i.g6 trans poses to line 1 below) 2 c,t>a3 ! and now: 1 ) 2 i.d5 3 c5 ! i.c6 4 i.g6 (now Black is in zugzwang) i.e4 (4 . . . i.d5 5 i.d3 i.c6 6 �b3 i.f3 7 i.c4 i.e4 8 �c3 i.c6 9 i.e6 i.e8 10 i.d5 is diagram 275) 5 i.f7 ! (but not 5 i.e8? �bl ! 6 �b3 i.d5+ 7 ••.
�c3 'it.>a2 ! 8 �d4 i.b7 ! 9 i.f7+ c,t>a3 ! 10 i.d5 i.a6 ! 1 1 c6 �b4 ! 12 c7 c,t>b5 ! 13 �e5 �c5, with a draw) i.g6 (5 . . . �bl 6 �b3 ! �c l 7 �c3 i.c6 8 i.c4 is diagram 276) 6 i.c4 i.e4 7 c,t>b3 i.f3 8 i.e6 i.c6 9 �c3 i.e8 10 i.d5 leads to diagram 275. 2) 2 i.e4 (the toughest defence) 3 i.g8 ! i.h7 4 i.d5 i.d3 5 c5 ! i.e4 6 i.c4 i.c6 7 i.e6 �bl 8 �b3 ! i.e8 9 �c3 �al (9 . . . �c l 10 i.c4 i.c6 is diagram 276) 10 i.d5 i.d7 (this is essentially diagram 275; the rest of given for the sake of completeness) 1 1 i.b7 i.g4 1 2 i.e4 i.d7 1 3 i.d5 i.e8 14 i.b7 i.h5 15 i.c8 i.f3 1 6 'it.>d4 �b2 1 7 i.e6! i.c6 18 �e5 ! �c3 19 �d6! i.b7 20 i.d5 ! i.a6 21 c6 ! �b4 22 i.g2 �a5 23 �c5 ! i.c8 24 i.fl ! i.g4 25 c7 i.f5 26 i.c4 i.d7 27 �d6 i.c8 28 �c6 c,t>b4 29 i.fl �a5 30 i.e2 �b4 3 1 �b6 i.d7 32 c,t>b7 i.e6 33 �b8 �a5 34 i.f3 �b6 35 i.b7 i.f5 36 i.c8 i.d3 37 i.h3 i.a6 38 i.g2 �a5 39 i.b7 and wins. ...
3.4:
ii.+d� v ii.
Of all the files the pawn can stand on, the centre files offer the least win ning chances. We start with the pawn on d7. In this case it doesn't matter whether the kings stand on c8 and c6, or e8 and e6; the vertical opposi tion is always drawn. This is not at all surprising, since both diagonals leading to the promotion square con tain at least four squares, so Black can never be put in zugzwang. Like wise, the horizontal opposition (with
196 .t+� v .t
kings on c6 and e6, or vice versa) al ways leads to a draw. There are some marginal cases involving the diago nal opposition, so we concentrate on these.
279
=I= Kling and Horwitz, 1851
278
(278): White cannot win this po sition, for example 1 .ie8 .idl ! (not l . . . .ic2? 2 .ih5 .ia4 3 .ig4, fol lowed by .id7) 2 .ig6 (there is no good square for the bishop on the e8h5 diagonal) .ia4 ! 3 .ih5 'it>e5 ! (not 3 . . . �g5 ? 4 .idl .ib5 5 .ie2 .ia4 6 .ia6, followed by .ic8 and .id7, nor 3 ....ib5? 4 .ig4) 4 .ig4 'ittd 6! with a vertical opposition draw. However, if White's bishop starts on h5 then he wins immediately by 1 .ig4 and 2 .id7. If we change the diagram by mov ing White's king to f8 and Black's king to d6, then White to play wins by 1 .ie8 ! .idl 2 .ib5 (making use of the extra square on the queenside) .ih5 3 .ic4 and 4 .tf7. The next position is based on the above ideas.
=I= Herberg, 1959 Tijdschrift v. d. KNSB
(279): White wins by 1 e7 ! .ib5+ ( 1 . . .'it>f7 2 'it>d6 .ib5 3 .id7) 2 'it>c7 ! .ie8 (2 ... 'it>f7 loses to 3 'it>d8 and 4 .id7) 3 'it>d8 ! .ih5 4 .ig4 ! (4 .id7? is a mistake, allowing Black to draw by 4 . . . 'it>f6 ! 5 .ie8 .ie2 6 .ig6 .ib5 ! 7 .ih5 'it>e5 8 .ig4 'it>d6!) .ig6 5 .id7 ! 'it>f6 6 .ie8 ! .id3 7 .ih5 ! (thanks to White's fourth move, this square is available for White's bishop) .ib5 8 .ig4 �e5 9 .id7 ! and wins . The following study is one of the most attractive compositions in the whole ending of i.+� v .i. (280): White's bishop can clear the pawn's path by moving along either diagonal, so Black's bishop must always retain the dual option of occupying either the a5-d8 or the h4d8 diagonal. In order to understand the solution, we have to see why the immediate advance of White's king is unsuccessful. If White plays 1 �e6? 'it>d3 2 'it>f7 �e4 3 'itte 8 'ittd5, then Black's king is close enough
.t+.0i v .t 197
m • • • . LS . • • • • • · � 1% . � � � � 1% � � � • • • • • • • • • ••• • . . - . •
280
+/=
Halberstadt, 1939 2nd Comm., Tijdschrift v. d. KNSB to set up the vertical opposition after 4 .tb6 .th4 ! 5 .ta5 d3 2 �g5 .td2+ 3 e4 and Black draws just as after 1 'i>e6?. There is a similar possibility on the queenside, whereby White plays �b6, forcing ... .tf2+ and also gain ing two tempi. Once again, this fails if played immediately: 1 'i>d4? ci>f3 ! 2 ci>c5 �e4 ! 3 'i>b6 .tf2+! 4 c8 �d5 ! 6 .tg5 i.a5 ! 7
.tf4 'i>c6! and Black sets up the ver tical opposition. This is where the second key idea comes in. White has to put Black into zugzwang and force a conces sion. The winning move is 1 e4 ! is reciprocal zugzwang; White to play can only hope to win with a king move, but we have already ana lysed all the possible moves and de termined that they don't work. After 1 �e4! Black may play: 1) 1 d5 (heading for b6) �e2 3 �c6 ci>d3 4 'i>b6 (White claims his two free tempi) .tf2+ 5 'i>b7 (according to Cheron, 5 �c7 �c4 6 .tg5 'i>b5 leads to a draw, but actually White can still win by 7 �d8 ! .tb6+ 8 �e8 ! .tc7 9 .td8 .tg3 10 .ta5 .th4 1 1 .te l .tg5 1 2 .td2 ! .th4 1 3 .th6, followed by .tf8-e7; however, the move played is simpler) .tel 6 d5? c5 'i>d3 4 c4 ! .tel (3 ... .td6 4 b6, promoting the pawn, and win ning after 4 . . . .tg3 5 'i>b6) 4 'i>b5 ! .tg3 5 'i>b6 .tf2+ 6 e4 8 .te7 .ta5 9 .td6 and .tc7. .•.
...
198 .i+.0. v .i
3) 1 .ic3 2 �f5 ! .ie l 3 �g5 as in line 4. 4) 1 �dl 2 �f5 (heading for g5) �e2 3 'iii>g5 .id2+ 4 �g6 (not 4 �f6? �f3 ! 5 .ic7 �g4 ! , which re ally does draw) .ie l 5 'iii>f7 c.itd3 6 �e8 'iii>e4 7 .ic7 .ih4 8 .id6 and wins. 5) 1 .ig3 2 'iii>d5 ! �d3 3 �c6! (first free tempo) .ie l 4 �b6 (sec ond free tempo) .it2+ 5 'iii>b7 .iel 6 �c8 c.ltc4 7 .ie7 .ia5 8 .id6 and 9 .ic7. There are 14 reciprocal zug zwangs with i.+.0.d7 v .i. ...
...
...
that having the pawn further back fa vours White, but the extra square on the d8-h4 diagonal makes all the dif ference. White to play wins by I .ie7 ! .iel 2 .id8 (making use of the extra space) .ib4 3 .ic7 and 4 .id6. Of course, Black to play can simply take the bishop, but even if White's bishop were on h4, then Black could draw by 1 . . .'iii>e4 ! heading for d5 and the vertical opposition. If the dia gram is modified by moving White's king to f7 and Black's king to d5, then White to play wins by 1 .ie7 ! .iel 2 .if8 .ih4 3 .ig7 and 4 .if6.
Now we will suppose that the pawn starts on d6. As one might ex pect, all positions involving the ver tical or horizontal opposition are drawn. On the other hand, the di agonal opposition results in a win, since the exceptional situation of diagram 278 (lack of space for the bishop) can't occur.
• • • . -�- . . -�- . • • ••• - . . . • • • • • • • • • • • • •
28 1
+I=
(281): This is basically diagram 278 shifted down a rank. It is curious
=I=
282
Averbakh, 1954 (282): A position analogous to diagrams 236 and 264 is drawn with a d-pawn Gust as it was with a c pawn). After l .ih5 .ih3 2 'iii>e5 (or 2 .ig6 .id7 ! 3 .if5 We8) White threat ens to play his king round to c7, which wins if Black simply plays passively. The correct antidote is for Black to activate his own king, and aim for a vertical opposition: 2 ...'iii>g7 (Black can afford to wait for
i. +£!, v i. 199
one more move before activating his king, but 2 ...i.d7 3 �d5 i.a4? is one passive move too many and then White wins by 4 'iitc5 ! rt;g7 5 'iit b6 i.d7 6 'iitc7 i. h3 7 i.e8 'iit f6 8 i. d7 i.fl 9 i.c8 i. b5 10 i. b7) 3 �d5 � f6 4 �c6 'iit e5 5 'iit c7 �d5 6 i.e8 �c5 with the standard draw. The following position illustrates a typical defensive idea:
I
284
= =
Centurini, 1856
I
283
= =
Averbakh, 1954 (283): After 1 �e6, Black defends by l . . .i. a5 ! (not l . . .i. g5? 2 d7 'iitc7 3 i.c5 ! , followed by i. e7) 2 i. f6+ �c8 ! (2 . . . �e8? 3 i. e7 ! and mate next move) 3 d7+ 'iit b7 4 'iite7 �c6 ! 5 'iite8 �d5 ! and Black defends by 'running round the back' . In a simi lar way, the symmetrical line 1 'iitc6 i.g5 ! enables Black to hold on. If Black lacks one of these lines then he loses, for example if his bishop starts on c l then White wins by 1 �e6 ! . The underpromotion ideas of dia grams 238 and 265 can also occur with a d-pawn.
(284): After l d7 ( I i. f7 i. g4 ! 2 i. h5 i. d7 is drawn by the previous diagram) i.e8 ! 2 d8.!D (2 d8i. i.h5 ! also leads to a draw) i.a4! (2 ...i.d7? 3 i. f7 ! i. g4 4 .!Dc6 ! i. f5 5 .!Des ! leads to mate next move, while after 2 . . .i. b5? or 2... i.h5? White wins the bishop by 3 .!De6+ !) 3 i. f7 i.d7 ! (now it is clear that this position is reciprocal zugzwang) 4 .!Db7 i.e8 and Black draws. If White's bishop starts on b3, then 1 d7 ! i.e8 wins after either 2 d8i. or 2 d8.!D, the former because of the mate on a4 and the latter because Black cannot play his bishop to a4. With the bishop on c4, White wins by l d7 ! i.e8 2 d8i. ! (not 2 d8.!D? i. a4 ! 3 .!De6+ �e8! 4 i.e2 i.c6 !). When the bishop is on d5, 1 d7 ! i.e8 again wins after 2 d8 i. or 2 d8 .!D, the latter because of the line 2 ... i. a4 3 .!De6+! �e8 4 i. f3, when Black is either mated or loses his bishop to a knight fork. Moving the bishop to e6 gives a draw, while if the bishop starts on h7 White wins by 1 i. g6 !
200 i.+!'!, v i.
( 1 d7? i.e8 ! is a draw) i.g4 2 i.f5 ! i.xf5 3 'it>xf5 ! cj;f7 4 'if;>e5 !.
I
285
= =
Forintos-Liberzan Moscow 1971 (285): Black's king is well placed to create a vertical opposition, so he should be able to draw. In the game Black lost rather quickly, and ECE (not to mention 600 Endings by Por tisch and Sarkozy) reprinted the end ing claiming that it is winning for White. The game continued: 1 'it>e5 The immediate king advance al lows Black an easy draw after 1 'it>f6 'it>f4 2 'it>f7 'it>f5 ! . i.b4 1 The position after 1 �e5 is a draw even with White to move, so Black's waiting policy is fully adequate. He could also have drawn by 1 . . . i.e7 2 i.f6 i.b4 3 i.d8 (now, however, 'it>d5 is a threat, so Black has to take action) 'it>f3 ! and now: 1) 4 i.c7 'it>g4 ! (4 . . . i.c3+? loses to 5 'it>d6! 'it>e4 6 'if;>d7 ! and 4 ... i.e7? 5 i.d6 ! i.h4 6 �f5 ! i.d8 7 i.f8 i.h4 ...
8 i.g7 i.e7 9 'it>g6 ! 'it>e4 10 'it>f7 wins for White) 5 i.d6 i.c3+ ! 6 'it>d5 i.f6 ! 7 'it>c6 'it>f5 8 'it>d7 'it>e4 ! 9 i.e7 i.c3 10 i.d8 i.b4 ! 1 1 i.c7 'it>d5 ! with a clear draw. 2) 4 'it>d5 'it>g4 ! (4 ... 'it>f4? fails to 5 i.c7+ ! 'it>f5 6 i.d6!) 5 i.c7 i.e7 ! 6 'it>c6 'it>f5 7 'it>d7 i.h4 8 i.d6 'it>e4 ! 9 i.e7 i.el 10 i.d8 i.b4 ! 1 1 i.c7 'it>d5 ! and Black once again reaches a standard draw. 2 'it>d5 After 2 i.h6+ 'it>d3 3 i.g5 White threatens 4 'it>d5, but Black can hang on by 3 . . . 'it>c4 ! 4 i.d8 i.c5 5 i.c7 i.e7 6 i.d6 i.h4 7 i.f8 i.g5 8 'it>d6 'it>d4! 9 i.e7 i.c l 10 i.f6+ 'if;>e4 ! . White's most dangerous attacking formation is to have his king on d5 and bishop on f6. Black must be pre pared to meet this by . . .'it>f3, so as to meet 'it>c6 by . . . 'it>e4 and i.d8 by ... 'it>g4. Likewise, if White plays his bishop to f6 and king to f5 , then . . . 'it>d3 is the requisite response. Black has two possible defensive plans; either to play his bishop to e7 whenever possible, making it hard for White set up either of these for mations, or to move his king in the correct direction to be ready with the appropriate counters. 2 'it>d3? Only this move loses. Black could have defended with the first plan by 2 ...i.e7 3 i.e5 (or 3 'it>c6 'if;>e4 4 'it>d7 i.b4 5 i.f6 'it>d5) 'it>d3 4 i.d6 i.h4 5 i.c7 i.g5 6 'it>d6 'it>e4 7 'it>d7 'if;>d5 with the vertical opposition. It was also possible to adopt the second plan by 2 ...'it>f3 3 Ji.f6 'it>g4, heading for the horizontal opposition. •.•
.t+.0. v .t 201
.ta3 3 .tf6! and White won after 4 .id8 .ib4 5 .tc7 .te7 6 'iti>c6! 'iti>e4 7 d7 ! 1-0.
=I+
286
Prokes, 1946 (286): The composer's intended solution was 1 .ie4 d2 2 .ic2 .tg4
(2 . . ..txc2 is stalemate) 3 'iti>bl ! .if5 4 'iti>al ! drawing, e.g. 4 ... 'iti>b4 5 'iti>b2 or 4 . . . .te6 5 'iti>bl .ib3 6 'iti>al ! . He believed that 1 'iti>bl d2 2 .ie4 ! c3 4 'iti>a2! .ih5 (or 4 ... 'iti>c2 5 'iti>a3 ! 'iti>c l 6 'iti>b4 ! .ih5 7 .ib3 ! .tg6 8 'iti>c3 ! and White is just in time with the verti cal opposition) 5 .ib3 ! .tf7 6 'iti>a3 !, and White is saved by stalemate. Fortunately, this unintended solution can be removed by starting White's bishop on g8. Then 1 'iti>bl ? d2! 2 .th7 'iti>b3 ! 3 .tg8+ 'iti>c3 4 'iti>a2 fails to 4 ... .tc2, so the unique solution is 1 .th7 ! d2 2 .ic2! and so on. There are two reciprocal zug zwangs in the ending .t+.0.d6 v .t.
The following position is based on one of them.
+I=
287
Original (287): The solution is short but not at all simple. The immediate pawn advance only draws, for exam ple 1 d7 .th4 ! (but not the other way; White wins after l .. ..ia5? 2 'iti>c6 ! .id8 3 'iti>b7 'iti>b3 4 c4 6 .id8 .iel 7 i.e7 .ta5 8 .id6) 2 b3 ! 3 .ic5 .id8 ! 4 .ie7 .ta5 5 'iti>f7 'iti>c4 6 'iti>e8 b3 1 ... If Black plays 1....tb4, White executes his main threat of 2 d7 ! .ta5 3 �c6 'iti>b3 (the cunning try 3 . . . 'iti>a3 doesn't help; after 4 'iti>b5 .ic7 White simply waits by 5 .if6, when Black has nothing better than 5 . . .'iti>b3 6 .te7 transposing) 4 �b5 ! (note this position; it arises several
202 .t+� v .t
times during the solution) i.c7 5 i.e7 (now Black is in zugzwang) 'iti>c2 6 'iti>c6 i.a5 7 'iti>b7, followed by 'iti>c8 and i.d6-c7. 1.. 'iti>b2 prevents 2 d7? because of 2 . . . .ta5 ! 3 'iti>c5 'iti>c3 ! (now Black can activate his king) 4 'iti>b5 i.c7 ! 5 'iti>c6 (or 5 i.e7 'iti>d3 ! 6 'iti>c6 i.a5 ! 7 'iti>b7 �c4) i.a5 ! 6 'iti>b7 'iti>c4 7 'iti>c8 'iti>d5 and the ver tical opposition is not far off. How ever, I . .. 'iii>b2 loses to a second white threat: 2 'iti>c6 (now Black cannot re ply ...'iti>c3, because this would block his bishop) �b3 3 d7 i.a5 4 'iti>b5 ! winning as before. Finally, 1 �a3 loses to 2 d7 i.a5 3 'iti>c6 'iti>a4 4 i.h4 (zugzwang) 'iti>b3 5 'iti>b5 i.c7 6 i.e7 with the usual win. The move played meets the direct threats since 2 d7? i.a5 ! transposes to a line given above, while 2 'iti>c6? 'iti>c4 ! and 2 i.e7? i.b4 ! offer no win ning chances. 2 i.f6! By covering c3, White renews his threat of d7, for example 2 ... i.d2 loses to 3 d7 ! i.a5 4 'iti>c5 ! 'iti>a4 5 i.h4 i.c7 6 'iti>c6 i.a5 7 i.g5, win ning as after 1 . . .'iti>a3 above. 2 .taS 2 ...'iti>a4 3 d7 i.a5 loses to 4 'iti>c5 !, so this is the best move, anticipating White's d7. Now 3 d7? 'iti>b4 4 'iti>c6 'iti>c4! is a draw. 3 i.h4! This surprising move creates a re ciprocal zugzwang. 3 i.b4 Black has to make a serious con cession. 3 ... 'iti>b4 and 3 . . .'iti>c3 are ruled out by 4 i.el+!, while 3 ... 'iti>a4 4 'iti>c6! is lost as above. Other king .
...
••.
•••
moves take the king too far away, while 3 ... i.b6 fails to 4 �c6 i.a5 5 'iti>b5 . The move played costs two tempi. .taS 4 d7! 5 'iti>cS! Black has lost too much time and loses as in the note to White's second move. The are very few interesting ex amples, either theoretical or practi cal, with the pawn further back. However, the following study is at tractive:
288
+I=
Bron, 1974 2nd HM, Bulletin Problemistic (288): White wins by 1 'iti>g6 ! (not 1 i.b2? 'iti>g8 2 'iti>g6 i.f8 3 e6 i.b4 with a comfortable draw, for exam ple 4 i.g7 i.a3 5 i.h6 i.e7 6 'iti>f5 i.a3 7 'iti>e5 'iti>h7 8 i.e3 'iti>g6) i.g7 ( l . ..i.f8 2 'iti>f7 ! i.a3 3 e6+ 'iti>h7 4 i.g7 i.b4 5 i.f8) 2 e6 ! 'iti>g8 3 e7 ! i.f8 4 e8i. ! (not 4 e8lt'l? i.a3 5 i.g7 i.e7 ! drawing, but Black should avoid 4 ... .te7? losing to 5 i.g7 !
.t+l!. v .t 203
.td8 6 lt.Jd6 ! .tg5 7 lt.Jf5 !) .ta3 5 .tg7 ! and mates. There are 12 reciprocal zug zwangs with .t+l!.d5 v .t. The next position provides a sample.
.tc4 4 d6 .tb5 ! 5 .tc8 �f3 ! is a draw) .ta4 ! (reciprocal zugzwang Z3) 4 .tb7 .tb3 5 d6 .ta4 ! 6 'itd5 'iii>f4 ! 7 .tc6 .tb3+! (this is the key move and it explains why Black loses in the analogous line with Black's bishop on b5) 8 'iii>c5 .te6 ! 9 �b6 �e5 10 �c7 'iii>d4 ! and Black manages to draw. 2) 1 .tf7+! 'ith4 ( l .. .'iii>h6 2 �f6 ! .ta4 3 d6 .tb5 4 �e7 wins easily) 2 .te6 ! and now we have Zl with Black to move: 2a) 2 �g3 (2....ta4 3 .tc8 ! 'iii>g3 amounts to the same thing) 3 .tc8 ! (Z2) and now: 2al ) 3 �f3 (sets Black up for a check on the long diagonal) 4 .tb7 (now the defence with ....tc4 and . . . .tb5 doesn't work because d6 is check) .td3+ (4 ... �e3 5 �e5 ! 'it>d3 6 .tc6! .ta6 7 d6 .tc8 8 .te4+ and 9 .tf5) 5 �e5 ! 'iii>g4 6 .tc8+ �g5 7 d6 .tb5 8 'iii>e6 .ta4 9 �e7 �f4 10 .td7 ! .td 1 1 1 .te8 .tg4 12 �f6 �e4 1 3 .tf7 ! and wins. 2a2) 3 .ta4 (3 . . . .td3+ 4 �e5 ! .tb5 5 .tb7 transposes) 4 �e5 ! (Z3) .tb5 5 .tb7 .tc4 6 d6! .tb5 7 'it>d5 and now the defence with . . . �f4 and . . . .tc4+ doesn't work, because Black's bishop is too close to the white king. 2b) 2 �h3 (a cunning defence) 3 �f4+! (not 3 .tc8? �g3 ! and we have Z2 with White to play) 'iii> g2 4 .tc8 ! (now White wins because he has driven Black's king further away) .tc4 (or 4 . . . �f2 5 �e5 �e3 6 .tb7 ! .tc4 7 d6! .tb5 8 �d5 ! .td7 9 .tc6! .th3 10 �e5 ! .tg4 1 1 .te8 .th3 1 2 'iii>f6 �d4 1 3 .tf7 ! �c5 14 'it>e7 !, ...
...
+/=
289
Original (289): This position depends on a set of three reciprocal zugzwangs. The most fundamental one is �e5 and .tc8 against �g3 and .ta4. We shall analyse the various reciprocal zugzwangs as they arise during the course of the solution. White may play: 1 ) 1 .te6? �h4 ! and now White is on the wrong end of a reciprocal zugzwang (which we will call Z l ) : l a) 2 �f6 �g3 ! 3 �e5 (3 d6 �f4 ! 4 �e7 �e5 ! is the vertical op position) .te8 ! 4 .tc8 .ta4 ! is the re ciprocal zugzwang of line le. lb) 2 d6 �g3 ! 3 .tcs �f3 ! 4 �e6 �e4 ! 5 .td7 .tn with the verti cal opposition. l e) 2 .tcS �g3 ! (reciprocal zug zwang Z2) 3 �e5 (White's main idea is to play .tb7-c6, but here 3 .tb7
.••
••.
204 .t+.0. v .t
winning for White) 5 d6! i.b5 6 i.b7+ �h3 (6 ... �f2 7 �e5 'it>e3 8 �d5 ! i.d7 9 i.c6 ! i.h3 10 'it>e5 ! i.g4) 7 'it>e5 �g4 (after 7 . . .�g3 White wins more easily, since Black cannot prevent 'iii>f6-e7 at move 10) 8 �d5 i.d7 9 i.c6! .tf5 (9 . . ..tc8 1 0 �c5 �f5 1 1 �b6! 'it>f6 12 �c7 'it>e6 1 3 i.b7 i.d7 14 i.d5+) 10 'it>e5 ! 'it>g5 (in this position Black can stop �f6, but it only prolongs the game without changing the result) 1 1 i.d5 i.d7 1 2 i.e6 ! i.a4 1 3 i.c8 �g6 ( 1 3 ...i.b5 14 �e6 i.a4 15 'it>e7 �f4 1 6 i.d7 !) 14 'it>e6! 'it>g5 1 5 �e7 'it>f4 1 6 i.d7 ! i.d 1 1 7 i.e8 i.g4 1 8 'it>f6 'it>e4 1 9 1J.. f7 ! i.d7 20 �e7 ! and wins. There are two reciprocal zug zwangs with .t+.0.d4 v .t, four with .t+.0.d3 v .t and three with .t+.0.d2 v .t. None is especially interesting, so we will skip them and look at the longest win in the ending of .t+d.0. v .t (25 moves).
(290): White can try: 1) 1 d6? i.c6 2 i.e4 i.d7 ! 3 �f4
(3 'it>d4 'it>g3 !) i.a4 4 i.d5 (or 4 i.f5 'it>g2 5 i.c8 �f2! 6 �e5 �e3) �h3 5 .tf7 i.b5 6 'it>e5 �g4 7 �f6 'it>f4 ! with the vertical opposition. 2) 1 .te4! .tn ( l . . . .th3 2 'iii>f4 ! .tfl transposes to line 2b) 2 i.d3 ! i.h3 (2 . . . i.g2 3 d6 ! wins now that White's bishop is more actively posted, for example 3 . . . i.c6 4 �d4 ! 'it>g3 5 'it>c5 ! i.d7 6 i.b5 i.h3 7 �b6 �f4 8 'it>c7 ! 'it>e5 9 i.d7 ! .tn 10 i.c8 i.b5 1 1 i.b7 or 3 ... i.h3 4 i.b5 ! 'it>g3 5 'it>e4 ! 'it>h4 6 'it>d5 �g5 7 'it>c6 'it>f6 8 �c7 !) 3 'it>f4 ! (confining Black's king is more important than pushing the pawn; 3 d6? 'it>g3 is a draw) and now: 2a) 3 i.d7 4 i.f5 i.b5 5 i.c8 i.a4 6 'it>f5 ! (and not 6 'it>e5? 'it>g3 !, when we have Z3 from the previous diagram, with White to play) �g3 (or 6 ...'it>h3 7 d6 'it>h4 8 'it>e6 'it>g5 9 'it>e7 'it>f4 10 i.d7 ! with an easy win) 7 'it>e5 ! and now Black is on the wrong end of Z3. 2b) 3 .tg2 4 i.e4 ! (still not 4 d6? i.c6 ! 5 'it>e5 'it>g3 ! 6 .tf5 'it>f2 7 i.c8 'it>e3 !) .tfl (or 4 . . . .th3 5 'it>e5 'it>g3 6 i.f5 ! i.g2 7 d6 ! i.c6 8 i.c8 ! i.e8 9 'it>e6 'it>f4 10 i.d7 ! i.h5 1 1 'it>f6 i.g6 1 2 i.c6 .tf5 1 3 i.e8 i.g4 14 i.f7) 5 i.f5 ! i.c4 (5 ... .tb5 6 i.c8 i.a4 7 'it>f5 ! transposes to line 2a) 6 d6 ! i.b5 7 i.c8 �g2 8 i.b7+ (we have transposed into line 2b of the previous diagram; the rest is only given for completeness) �h3 9 'it>e5 �g4 1 0 'it>d5 i.d7 1 1 i.c6 ! i.f5 1 2 'it>e5 ! �g5 1 3 i.d5 i.d7 14 i.e6 ! i.a4 1 5 i.c8 'it>g6 1 6 �e6 ! �g5 17 ...
...
+I=
290
Original
i.+l!i
�e7 �f4 1 8 i.d7 ! i.d l 1 9 i.e8 i.g4 20 �f6 �e4 21 i.f7 ! i.d7 22 �e7 ! i.a4 23 i.e8 ! and promotes or cap tures the bishop in two more moves.
v
.t 205
We end with a table summarising the 1 60 reciprocal zugzwangs in this ending, divided up according to pawn position.
The first table lists the reciprocal zugzwangs with opposite-coloured bishops, and the second gives the toals for same-coloured bishops.
· ·· 1 .. ··
c
d
0
1
3
4
2
0
4
1
0
0
0
6
1
0
1
1
0
0
4
0
0
0
0
2
•
b
c
d
0
0
8
10
4
5
13
1
5
10
16
13
6
..
7
7
5
1
3
3
3
3
4
2
7
3
0
1
7
3
7 6
' ··.·
·
The battle of knight against bishop is a central theme in chess. One of the spe cial features of chess, which makes the game unusually interesting, is the re markable fact that, despite moving in completely different ways, knight and bishop have more or less equal value (on an 8x8 board !). If all the chess pieces had significantly different values, then the question as to whether to exchange one piece for another of a different type would arise very rarely, because the outcome would clearly benefit one side. With knight and bishop, however, one must be constantly assessing whether a particular exchange favours White or Black. Many popular opening systems (Nimzo-Indian, Exchange Spanish, 3 tDc3 ..tb4 in the French) are based on the exchange of bishop for knight, and in all these openings it is not clear, even after decades of investiga tion, whether the knight or the bishop is more valuable. When the position is simplified, and pawn exchanges lead to more and more open lines, the bishop gradually becomes more valuable relative to a knight. Of course, such a rule has many exceptions, for example if the bishop is impeded by its own pawns then it can be almost worthless, but nevertheless it is true far more often than not. It follows that in a very simplified positions such as ..t+L'!. v tD, the bishop will be very effective. Despite this, the bishop has limitations. The main one is that the bishop is monochromatic; if you have a light-squared bishop and the defender can set up a firm blockade on a dark square, then that is usually the end of the matter - the game is simply drawn. The upshot of this is that ..t+L'!. v tD offers the superior side better winning chances than either ..t+L'!. v ..t or tD+L'!. v ..t, but probably not as many as tD+l!. v tD. In the battle of bishop against knight, one weapon is of exceptional value - zugzwang. The bishop can usually lose a tempo fairly easily, while the knight cannot. This is reflected in the relatively high number of reciprocal zugzwangs in ..t+L'!. v tD, a total of 21 12. As usual, the material is divided up according to the file occupied by the pawn, but here there is an extra section because there are a number of situations in which White is obliged to under promote. These positions tend to depend not so much on the file of the pawn, but on the piece White promotes to. 4. 1 : ..t+al!. v tD 207 4.2: ..t+bl'!. v tD 223 236 4.3: ..t+cl!. v tD 4.4: ..t+dl!. v tD 248 4.5 : Underpromotion 255
i.+l!, v lll 207
4.1
i.+aLS v lb
The a-pawn situation is influenced by the colour of the bishop. If White has a dark-squared bishop, Black can draw by moving his king to a8, even if this involves losing the knight. However, this feature has less impor tance than one might imagine, be cause if Black can get his king to a8 in some way, it is normally not nec essary to give up the knight in the process. We start with the pawn on a7. This situation is normally resolved fairly quickly. If the white king is actively placed, supporting the pawn, then Black normally loses unless he has an immediate draw by perpetual check. The following positions illus trate this point:
moves first, because Black is re stricted to moving his knight back and forth between c7 and a8. White wins by l i.b4 (the immediate 1 i.d6 may be met by 1 ...lllc7, and White has no instant win because of the stalemate) lllc7 2 i.el llla8 3 i.a5 (the best square for the bishop, covering both b4 and c7) lllc7 4 'it>b6 llla8+ (after 4 ... llld5+ 5 'it>a6 Black has no knight check) 5 'it>a6 (now there is no stalemate, so Black has to move his king) 'it'd? 6 'iti>b7 and wins.
Kovalenko, 1976 Schakend Nederland
+I-
29 1
Averbakh, 1955 (291): This example is fairly typi cal. White has the wrong bishop for his a-pawn, but this doesn't help Black. It makes no difference who
(292): In this example, White just manages to avoid Black's fate in the previous diagram: 1 lllf2 ! 'it>h4 (after 1 ... i.g3 2 'it>fl ! i.xf2 3 �g2 ! White reaches h 1 ) 2 'iii>f l ! (2 lllh 1 ? 'iii>h 3 3 �f2 i.g5 4 �f3 i.h4 ! loses straight away, while after 2 'it>e2? 'it>g3 ! 3 'it>fl �f3 Black wins as in the previous diagram) 'it>g3 3 llle 4+! (not 3 lllh l +? �f3 !) 'it>h3 4 lllf2+ ! (Black's bishop is on an inferior square and this allows White to save
208 j_+� v ttJ
himselt) �g3 5 tlJe4+ ! �f3 6 tiJd2+ ! j_xd2 stalemate.
+I= Galushko, 1987 Special Pr., Akerblom Mem. Tny. 294
+I=
293
Averbakh, 1955 (293): The situation is similar when the kings are the other way round. White's advantage is so large that Black can only hope to save himself by giving immediate perpet ual check. White to play wins by 1 j_e6 ! �c5 2 'it1b7 �b5 3 j_f7 'it1c5 (3 . . .�a5 4 j_e8 is the same) 4 j_e8 and Black is in zugzwang. Black to play draws by playing l . ..tiJd7+! 2 'it>c8 (White should avoid 2 �a8? �c7 ! and mate next move) tiJb6+ ! and White can only avoid the checks at the cost of losing his pawn. The following position is more complex. White has to play precisely in order to avoid Black's stalemate defence. (294): White wins by 1 j_c2 ! ( 1 j_d 1 ? effectively loses a tempo, be cause White has to play j_c2 later in order to clear the path for his king; Black draws after 1 .. .�f2 ! 2 �d2 'it1g3 ! 3 ..t>e3 'it1h4 ! 4 j_c2 tiJh8 ! 5
..tf4 'it1h5 ! 6 'it>fS 'it1h6 7 ..tf6 tlJg6! and Black is saved by stalemate) tlJh8 2 j_d 1 ! tlJg6 (White also wins after 2 . . .�f2 3 ..t>d2 ! tlJf7 4 �d3 �g2 s ..td4 'it1g3 6 j_b3 ttJhs 7 'it1e5 ! 'it1g4 8 �f6 ! ) 3 �c2 ! (White has gained the useful tempo �c2 and now his king is in time to attack the knight on g6) �f2 4 'it1d3 ! 'it1g3 5 �e4 ! 'it1h4 6 Wf5 ! winning. Black to play draws by 1 .. .'it>e2. There are 12 positions of recipro cal zugzwang with j_+�a7 v ttJ, but most are uninteresting. The follow ing example, which bears a certain resemblance to the previous posi tion, is based on the most attractive. (295): White can try: 1) 1 'it1g2 ? (this try fails because Black can gain a tempo by attacking the bishop with his king) 'it1d2 ! 2 'it>f3 �c3 ! 3 'it1e4 'it1b4 ! 4 j_fl (or 4 j_e8 'it1a5) �c5 ! and White's king cannot make further progress. 2) 1 j_c6? (1 j_d3? 'it1d2 ! is simi lar to line 1) �e2 2 'it1g2 �e3 3 �g3
.t+L!. v lb 209
. . - B • • B - - . B..t• • B - - - - . . - . - '0�% • � • d d %.. � +I=
295
Original �d4 4 f4 'iltc5 is an easy draw for Black. 3) 1 .tfl? (this avoids the prob lem of in line 1 , but now White falls victim to a reciprocal zugzwang) �a8 ! (now White's problem is that any bishop move on the f1 -a6 diago nal will give the black king an extra tempo in the race for the a-pawn) 2 �g2 (2 .tg2 0ic7 3 .tf3 'it>d2 ! is an easy draw, so this is the only other possibility) 0ic7 ! (a second recip rocal zugzwang) 3 i.a6 (White has no useful king move) 'iltd2 ! 4 'iltf3 �c3 ! 5 'ilte4 'iltb4 ! 6 'iltd4 'ilta5 ! , fol lowed by ...b6, and Black wins the pawn. 4) 1 .ta6! (this nullifies Black's threat of playing his king towards the a-pawn, because ... 'ilta5 isn't espe cially effective when the knight is blocking b6) 0ia8 (or 1 . . .'iltd2 2 'iltf2 ! �c3 3 e3 ! 'iltb4 4 �d4 ! 'ilta5 5 .tn uid Black cannot play ...'iltb6, so White wins by 5 ...'iltb4 6 .te2 'ilta4 7 �c5 'ilta5 8 .tn) 2 .tn ! (by clearing b6, Black threatened 2 ...'iltd2, so the
bishop has to move, and this is the only safe square) 0ic7 (or 2 . . .0ib6 3 'iltg2 ! 'iltd2 4 �f3 'iltc3 5 'it>e4 'iltb4 6 'it>d4) 3 'iltg2 ! 0ia8 (the line 3 ... 'iltd2 4 'it>f3 'it>c3 5 'it>e4 'it>b4 6 'iltd4 ! is clearly winning for White) 4 f3 ! (the only way to win; thanks to the reciprocal zugzwang, White's king has moved to g2 with gain of tempo, and now he can sacrifice his bishop) 'it>xfl 5 'it>e4 ! 0ic7 6 'it>d3 ! 'iltf2 (or 6 ...0ia8 7 'it>d4) 7 'ii?c4 ! 'it>e3 8 'iltc5 ! 0ia8 9 'iltc6! 'it>e4 10 'iltb7 ! and wins. The 10 consecutive unique moves and loss of tempo by the white bishop make this an appealing study. Black to play draws by l . . .'it>d2 ! , gaining a tempo when the king ar rives on b4. Now we move on to the pawn on a6. The defender's chances are much better with the pawn one square fur ther back, and indeed some positions are drawn without an immediate per petual check.
2 JO i.+£>, v lt:)
(296): Black to play wins after l . . . i.b6 2 �el i.a5+ 3 ..ti>e2 i.d2, so suppose that White moves first. He can draw by transferring his knight to f3 : 1 lDh2 ! i.c5 (the study composer Soukup-Bardon once gave l . . .i.h4 2 lDf3 ! i.g3 as a win, but White can draw by playing 3 �e3 !) 2 lDf3 ! (not 2 lDfl ? i.d4 3 ..ti>el i.c3+ 4 �e2 i.d2) i.b6 (or 2 ... i.f8 3 lDM+ ..tg3 4 lDf3 i.h6 s ..tin ! , and now the wrong bishop element becomes important) 3 lDh4+ (or 3 lDel + �g3 4 lDf3 ! transposing) �g3 4 lDf3 ! (not 4 lDf5+? �f4) i.d4 5 lDd2 ! �g2 6 lDf3 ! and Black cannot make progress. It might seem that Black should be able to run White out of moves, but this is never possible; when Black's king is on g3, White can play lDd2, while if the king is on g2, then White can check on el or h4 (except if the bishop is on f2, when lDh2 is possible). The previous position was drawn because the defender's king was well placed. In the following position White to play cannot prevent Black setting up this drawn position. (297): After 1 i.e5+ Black has to choose carefully where to move his king: 1) 1 �d7! 2 i.c7 lDa7 ! draws by the previous diagram. 2) 1 �cS? 2 i.c7 ! (this position is reciprocal zugzwang) ..ti>d5 (after 2 ...�b5, 3 i.b6! puts Black in imme diate zugzwang, which in fact is also reciprocal) 3 �b6 lDe7 4 i.f4 lDc6 (or 4 ... lDc8+ 5 �b7 'it>c5 6 i.g3 �b5 7 i.c7) and now: ...
...
I
297
= =
Averbakh, 1955 2a) 5 i.cl? (it is important not to let Black's king reach d7) �d6! 6 i.f4+ (or 6 i.a3+ �d7 ! 7 i.f8 ..ti>c8 ' and Black draws) �d7 ! 7 i.g3 lDe7! drawing as in the previous diagram. 2b) 5 i.h2 lDe7 6 �c7 lDc6 7 i.gl and the pawn advances. 3) 1 �d5? 2 ..ti'b6 wins as in line 2, but not 2 i.c7? ..ti>c5 ! drawing because White is on the wrong end of the reciprocal zugzwang (3 i.b6+ may be met by 3 .....ti>b5 or 3 ...�d6). (298): Here White wins by 1 ..ti'b8 ! (but not 1 a7? lDb6+ ! 2 �b8 lDd7+ ! 3 �c8 lDb6+ ! 4 �d8 ..ti'b7) lDb6 2 i.e6 ! (it is essential to prevent the check on d7) ..ti>b5 (or 2 . . . lDa4 3 a7 lDb6 4 i.g4 ..ti>c5 5 �b7 �b5 6 i.f3 ..ti>c5 7 i.e2) 3 �b7 ..ti>a5 4 a7 �b5 5 i.g4 c5 6 i.e2 and Black is in zugzwang. Black to play draws by 1 . . .lDb6+ !. There are 6 1 positions of recipro cal zugzwang with i.+£>,a6 v lD, but we have already seen two in diagram 297, so we will move on to the case of the pawn on a5. ...
.t+L>, v ttJ 211
• • • •
• %¥J
¥J ¥J � ¥J'
r¥J' l :;
a -
. � . � . :;:,� � • R B �� ,WJij i.. n
�
� ...
-
.
� � .
�'
w�
•
r:��
• • • • .
+I=
298
-
.
• • • • 299
.
I
= =
Averbakh, 1955
Original
Here Black's drawing chances are even better. Diagram 297 shifted down a rank is, of course, drawn. If Black's knight is firmly established on c5 then whether White can win depends on the position of Black's king. If it is on d6, then the position is drawn (by a method analogous to diagrams 296 and 297), but White has winning chances if the king is on b4, c4 or d4. The following diagram illustrates the possibilities. (299) : Black can draw this posi tion, even with White to move, but it requires very accurate defence: 1) 1 i.f3 tbd7 + ! 2 �c7 tbc5 ! 3 �c6 �c4 ! followed by . . .�b4 and Black draws. 2) l i.f7 tbd7+! 2 �c6 (or 2 �c7 �c5 3 �c6 tba6! 4 �b6 tbb8 ! 5 .te6 �c3 !, followed by . . . �b4, and Black draws) tbb8+! 3 �b7 �c5 ! (but not 3 ... tbd7 ? 4 i.e6! tbc5+ 5 �b6 tba4+ 6 �b5 tbc5 7 i.c4 tbe4 8 a6 tbd6+ 9 �c6! and White is win ning) 4 i.e8 'it>b4 ! 5 �b6 �c4 and Black draws.
3) 1 i.e8 �d5 ! followed by . . . �d6 draws as in diagram 296, for example 2 i.b5 �d6! 3 .i.c6 tba6 ! . Averbakh claimed that l . ..�c4? also leads to a draw, but White wins by 2 i.c6 ! (this is reciprocal zugzwang, just as it was a rank further up - see diagram 297) �d4 (or 2 . . .�b4 3 i.b5 ! tbe4 4 a6 tbd6 5 i.d7) 3 �b5 ! tbe6 4 i.g2 tbc5 5 i.f3 tbe6 6 �c6 (Averbakh didn't mention this move) tbd8+ (6. . .tbc5 7 i.e2) 7 �b6 tbf7 8 a6 tbd6 9 i.b7 tbc4+ 10 �c7 and the pawn queens. If White's bishop starts on g4 then he wins by I i.e6! tba4+ 2 �b5 tbc5 3 i.c4, while with a bishop on g2, 1 �b5 ! tbe6 2 'it>c6 is decisive. The following position is cer tainly worth knowing . (300) : The defender must avoid this position at all costs. Black to play is in immediate zugzwang, but even with White to move the posi tion is lost: 1 i.e8 ! (not 1 i.c6? �c4 ! and White lands in a reciprocal zugzwang; the position is only a
212 i. +�
v
tD
. - . - - - � - . B�l1 B B . . . � - - - - - . - - - 300
+/-
Original
draw after 2 i.e8 �d5, followed by . . . ..ti>d6) ..ti>c4 2 i.c6 ! ..ti>d4 (after 2 . . . ..ti>b4 3 i.b5 White has achieved his ambition) 3 ..ti>b5 ! tDe6 4 i.g2, and White wins as in the analysis of 1 . . . ..ti>c4? in line 3 of the previous diagram. The following example shows that these ideas can be important in practice.
301
=I+
Nikolaidis-Botsaris Athens Open 1992
(301): In the game White was to play. The position is drawn because White can reach the important recip rocal zugzwang given in the pre vious diagram, with Black to move. 1 ..ti>d5 (there are several ways to reach the critical position; 1 tDc4 a4 2 tDd2+ 'iti>c2 3 tDc4 ! i.c3 4 ..ti>b5 ..ti>b3 5 'iti>c5 ! is another possibility) a4 2 tDc4 ! (not 2 tDb5 ? Wb4 3 c6 i.g7 4 b6 i.h6 5 Wc6 i.e3 and Black wins) i.c3 3 c5 ! i.b4+ (the key position) 4 �b5 ? (we know that d3 is the best place for the king to support a knight on c4, and b5 is the worst; 4 �d4 ! and 5 d3 was an easy draw) i.el ! 5 c5 i.c3 ! (now Black has transferred the move to White) 6 Wd5 b4 ! 7 tDe3 i.d2 (Black shows the first signs of falter ing; he could have won by 7 ... i.g7 8 tDc4 i.f6 9 tDe3 'itiic3 10 tDc4 i.e7, just as in line 3 of diagram 299) 8 tDc4 i.e l ? (throwing away the win completely; it was necessary to go back with 8 . . .i.c3 !) 9 'iti>d4 ! (White doesn't miss the second chance to transfer his king to d3) i.f2+ 10 d3 ! i. g l 1 1 tDe5 a3 1 2 ..ti>c2! �a4 13 tDc4 a2 1 4 �b2 112-1/2. If Black were to play in the dia gram, then he could win by 1 . . .a4 2 tDc4 (or 2 tDb5 i.c3 3 'iti>d5 'iti>b4 4 ..ti>c6 i.g7 5 �b6 i.h6 6 c6 i.e3) i.c3 ! reaching the position at move five in the game. Generally speaking, Black is also safe with the formation of king on d6 and knight on c7. However, there is one trick which he must avoid. Black can always dodge this trick if he starts from a typical position, but
i.+� v ttJ 213
in the following example it is un avoidable.
+/=
302
Original (302): White to play wins by 1 �c5 ! (and not 1 'iii>b 5? 'iii>d 6! 2 'it>b6 llid7+) 'iii>d8 ( l . . .'it>d7 2 'iii>b6 ! and l . . . llid7+ 2 �c6! lose more quickly) 2 'it>b6 ! llid7+ 3 �b7 ! (not 3 'it>c6? llie5 + ! 4 'it>b7 llic4 ! 5 a6 llia5+! 6 �b6 ltJc4+ ! 7 �c5 'it>c7 ! drawing) llic5+ (3 . . .'it>e7 4 i.f5 llic5+ 5 'it>c6 llia6 6 'it>b6 llib4 7 i.e4 'it>d8 8 'it>b5) 4 �c6! llia6 (4 . . .llid7 5 i.f5 llie5+ 6 �b7 is easier) 5 'it>b6 llic7 (5 . . . llib4 6 i.e4 �c8 7 'iii>b5 llia2 8 a6 llic3+ 9 �b6 ! llia4+ 10 'iii>a5 wins) 6 'it>b7 �d7 (now White can execute the trick mentioned above) 7 i.f5+ ! (if White loses time then Black can de fend, for example 7 i.e4 'it>d6 ! 8 i.f5 �c5 ! ) �d6 8 i.c8 ! (not 8 i.h3? �c5 ! ; the key difference with the bishop on c8 is that 8 ... 'it>c5 9 �xc7 ! �b5 is met by 10 a6! ) llib5 (Black is in zugzwang and must allow the pawn forwards) 9 a6 ! �e7 10 i.h3 �d6 1 1 i.fl ltJc7 12 a7 ! winning.
Black must watch out for this trick when he is defending with his king on d6 and knight on c7, but nor mally White cannot force it. Sup pose White's king is on b7 and he puts his bishop on the h3-c8 diago nal; then Black can reply . . . 'it>c5. On the other hand if his bishop is on f1 to prevent this, Black can play ... ltJe6 instead, transferring his knight to the safer square c5 .
• • • • . . . � • •
r:fJf! i" r:�
• •
:• .':!
" ?� • • • '@ �
i'Q:,;
� �
"· "
wr: rrJ
�� �J • ��?:;?cc?? - � �£�••· 8 ?::�
•
•
303
�;:;::c;
•
rJ%
•
• •
•
=I=
Sevitov, 1937 (303): The following study was considered to be a win by both Cheron and Averbakh. Only ECE correctly gives the position as a draw. After 1 i.e5 (1 a6 and 1 i.xh6 are both met by l . . . 'it>c7 with a clear draw) Black may play: 1 ) 1 llig4? 2 i.g3 ! (the only safe square on the h2-b8 diagonal; 2 i.f4? llif6 ! 3 a6 llid5+ ! 4 'it>b5 'iii>c8 5 i.e5 llic7+ and 2 i.d6? �d7 ! 3 'iii>c5 llif2 4 a6 llie4+ ! 5 'iii>b4 'it>c6 ! both lead to a draw) llif6 3 a6 ! �c8 (or 3 ...llid5+ 4 'it>c5 ! ltJc7 5 a7 ! 'it>c8 6 �c6 ! ) 4 �c5 ! llid7+ (or 4 . . . llie4+ ...
214 �+� v �
5 �c6 �xg3 6 a7 !) 5 �b5 �f6 6 �c6 .!Des 7 a7 and White wins. 2) 1 ttJt'S? 2 a6 .!De7 3 cjr>c5 �c8 4 �c6 �a7+ 5 cjr>b7 �c8 6 �f6+ �d7 7 �e7 wins for White. 3) 1 �f7! (Averbakh failed to mention this move at all) 2 �g3 (2 a6 �xe5 3 a7 �c6+) �d7 ! 3 cjr>b5 �d8 ! (3 ...�d6+? loses after 4 �xd6 cjr>xd6 5 cjr>b6) 4 cjr>b6 ltX6 ! 5 a6 (now Cheron stopped his analysis with the assumption that White wins, but in fact the draw follows immedi ately from diagram 296) �e7 ! 6 cjr>b7 �c6 ! 7 �c7 �a7 ! and as in dia gram 296. There are exactly 100 reciprocal zugzwangs with � +�a5 v �- One of these plays a starring role in the fol lowing position. ...
...
+I=
304
Original (304): White to move may con tinue: 1) 1 � e2? �b8 ! 2 �h5 �a6! 3 �g6 �b8 ! (now White finds him self trapped in a reciprocal zug zwang) 4 �dl (after 4 �e2 �g3 ! 5
�f5 Black gains a tempo by attack ing the bishop: 5 ... �f2 ! 6 �b5 cjr>e3 7 cjr>e5 cjr>d2 ! 8 cjr>d4 cjr>c2 9 cjr>c5 cjr>b3 with a draw) � g3 ! 5 cjr>f5 cjr>f2 ! 6 cjr>e4 cjr>e l ! (the same thing happens in this line) 7 �f3 cjr>d2 ! 8 cjr>d4 cjr>c2 9 cjr>c5 �b3 drawing. 2) 1 �h5! �b8 ( l . ..�c5 2 cjr>g6! �a6 transposes to the main line, while 1 . . . cjr>g3 2 �g5 ! �c7 3 cjr>f5 leads to the main line after 3 ... cjr>f2 4 �e4 �a6 5 �f3 �b8 6 �g4 or 3 . . .�a6 4 cjr>e4) 2 �g6! (this second reciprocal zugzwang is the key po sition; either Black's knight has to return to a6, when White can play �e2 with gain of tempo, or Black has to give way with his king) �a6 (2 . . . �g3 3 cjr>f5 �a6 4 cjr>e4 �f2 5 �f3 �b8 6 �g4 is again the main line) 3 �e2 ! �b8 4 �f5 ! (White edges nearer the queenside while keeping Black's king penned in) �g3 5 cjr>e4 (Black is in zugzwang; if White were to play he could win by losing a tempo) cjr>f2 6 �g4 (now Black has to make a serious conces sion) cjr>el (after 6 ... �g3 7 �c8 cjr>f2 8 cjr>d5 White can just run for the queenside) 7 cjr>e3 ! cjr>fl (7 ... �a6 8 �f3 �b8 9 �e2 is a deadly zug zwang) 8 �f3 (now Black cannOl play either ... cjr>el or . . . �a6) �gl (Black's king is too far away) 9 �e2 �g2 10 �c4 cjr>g3 1 1 �d5 �a6 12 cjr>d4 cjr>f4 13 �e4 (White operates by a succession of zugzwangs) �b8 14 �b7 cjr>g5 1 5 cjr>d5 with an easy win for White. There have been few attempts to analyse positions with the pawn
i.+� v ttJ 215
further back, so many of the posi tions in the rest of this section are from over-the-board games. To win with the pawn so far back normally requires that the defender's king is far away from the action. The fol lowing example is an extreme case.
The following example was a comedy of errors.
+I= Pogats-Papp Hungarian Team Ch 1991
306
+I=
305
Barlov-Eingorn 'Zagreb IZ 1987 (305): In the game White was to play. Barlov won by 1 i.g4 tiJd8 ( I . . . tiJd6 2 h5 is the same) 2 h5 tiJf7 3 i.e6 tbg5 (3 . . . tiJh6 4 d6
(306): White to play would win by 1 b6, but in the game Black was to move. He could have drawn by l . . .tiJe3 ! 2
216 i+L!. v ttJ
an easy win) �d8 6 ia3 (a sign of hesitation; White only needs to play his bishop to the h2-b8 diagonal, and then drive the knight away by �b6) �d7 7 ib4 �d8 8 iel tiJd6+ 9 �b8 tlJc8 1 0 ig3 �d7 1 1 �b7 ! (after 1 1 if4? Black could draw by 1 l . . .�c6 or 1 l ...tlJe7) �d8 12 ic7+ (missing an instant win by 1 2 ih4+ �d7 1 3 ie7) �d7 13 ib6? (White could still have reversed tracks, but now he permanently throws the win away) tiJd6+ ! (from now on Black defends accurately) 14 �b8 tiJb5 ! 1 5 ia5 �c6 1 6 id8 �d7 17 ih4 ..ti>c6 1 8 if2 �d7 ! 1 9 ib6 �c6 ! 20 ia5 �d7 2 1 id2 �c6 22 ie3 �d7! 23 ..ti>b7 tiJd6+! 24 �b8 tiJb5 ! 25 igl ..ti>c6 ! 26 �f2 �d7 ! 27 �b7 tiJd6+ ! 28 ..ti>b6 tlJc8+! 29 �b5 tiJd6+ 30 �c5 �c7 3 1 ig3 �b8 32 'iitb6 'iita8 112-112.
+I=
307
Original (307): In order to draw, Black only needs to have his knight on c7 and king on d6, but by accurate play White can prevent this: 1 ..tr>b4 ! (not
1 ig2? tlJe8, followed by . ..t'iJc7 and �e6 (after l .. .tiJe8 2 �c.5 White prevents ...�d6 and wins eas ily; the move played intends . . . �d7) 2 ih3+! (the only move; 2 ibS? �d5 and 2 ..tr>c5? �d7 ! 3 �b6 �c8 4 �c6 tlJe8 are comfortable draws) �d5 3 ig2+ ! ..ti>e6 4 ic6 ! (pre venting both ...tlJe8 and .....tr>d7) �e7 (after 4 . . . tlJc8 5 ..t>c5 ! Black has nothing better than . . . �e7 transpos ing) 5 �c5 ! (now White has defi nitely prevented Black from setting up the drawing formation) tlJc8 6 ib5 ! (clearing c6 for the king; not 6 a5 ? ..ti>d8 !) �d8 (6 . . . ttJa7 7 �b6 tlJc8+ 8 �c7 tiJd6 9 ic6 tlJc4 l 0 id5 tiJd6 1 1 a5) 7 �c6! tlJe7+ 8 ..ti>b7 ! tiJf5 (8 . . . tlJc8 9 in tiJd6+ transposes) 9 ifl tiJd6+ (or 9 ... ttJd4 10 'iitb6 tlJe6 1 1 a5 tlJc7 1 2 �b7 'iitd7 1 3 ih3+! �d6 14 ic8 ! win ning as in diagram 302) 10 �b8 lOe4 (10 . . . tlJe8 1 1 a5 tlJc7 12 �b7 leads to the previous note) 1 1 a5 tlJc5 1 2 ib5 (reciprocal zugzwang) �e7 1 3 'iitc7 ! �e6 1 4 �c6 tiJd7 1 5 ifl �e7 16 ih3 tlJe5+ 1 7 �c7 and the pawn cannot be stopped. There are 80 reciprocal zug zwangs with i+L!.a4 v tlJ. The most important information for the practi cal player is that the reciprocal zug zwang in line 3 of diagram 299 is preserved when it is shifted down a rank. (308): White to play wins by I ib4 ! (not 1 ..tr>b5? tlJa3+ 2 ..tr>b4 tlJc2+ !) �c2 2 �b5 ! �b3 (after this White must head for the reciprocal zugzwang with ic5 v �c3; 2 ...�d3 loses after 3 ..ti>c5 ! tlJe5 4 a5 tiJd7+ . . . �d6)
..t +.0. v lb 2 1 7
12 �b6 ! ll'ic6 1 3 i. g3 ll'ie7 14 �c7 ll'ic6 15 i. f2 and wins. Now we move the pawn back to a3.
+/=
308
Original 5 � b5) 3 i. f8 (3 ..te7 is also good, but not 3 i. c5? �c3 ! and White is on the wrong end of the reciprocal zugzwang) �c3 4 i.c5 ! �d3 (after 4 . . . � b3 5 ..t b4 ! White has passed the move to Black and wins easily, for example 5 . . . ll'ie3 6 a5 ll'id5 7 i. d6 ll'ic3+ 8 � b6 ll'id5+ 9 �c6) 5 �b4! ll'ie5 6 i. g l and now: 1 ) 6...ll'ic4 7 ..tf2 ! (curiously, this is the only square to win; after 7 i.a7? ll'id6 8 a5 �e4 ! 9 a6 �d5 ! the pawn is blocked, and Black will draw by . . . �c6) ll'ie5 (7 . . . ll'id6 8 a5 ! ll'ic8 9 a6) 8 �c5 ll'id7+ (8 ...ll'ic4 is met by 9 i.e l , so White could not have won if his bishop had been on gl) 9 �b5 ll'if6 10 a5 ll'id5 1 1 i. b6 ll'ic3+ 12
+I=
309
Chemin-Sax Subotica IZ 1987 (309): White to play would win by 1 a4 �f7 2 a5 e6 3 �b7 ll'id3 4 �b6 ll'ib4 5 i.el ll'id5+ 6 �b7, but in the game it was Black to move. Play continued l . . . �f7 2 a4 �e6 3 a5 �d7 ! 4
218 �+£!, v ll:l
move l ...'it>d4 draws because it side steps a check on the long diagonal, and B lack gets his king to h8 after 2 �c8 llif3 3 �b7 llih4! 4 �g5 �e5 !. The fundamental reciprocal zug zwang of diagrams 297, 299 and 308 finally becomes a draw when it is shifted down another rank:
3 10
+I=
Fischer-Taimanov Vancouver Ct (2) 1971 (310): In this position Taimanov (Black, to move) made one of the most famous blunders in the his tory of chess. The game continued 1 . . . �e4? 2 �c8 ! (having been given a chance, Fischer makes no mistake; 2 �e6? lllf3 ! would be a draw be cause there is no check on the long diagonal) 'it>f4 (thanks to the unfor tunate position of his king, Black cannot manoeuvre his knight round to stop the pawn, e.g. 2 ... llid3 3 �f5+! or 2 . . . llif3 3 �b7+ ! ) 3 h4 ! llif3 (3 . . . llig4+ 4 'it>g7) 4 h5 ! llig5 5 �f5 llif3 6 h6 llig5 7 �g6 llif3 8 h7 llih4+ 9 �f6 1-0. In the diagram position Black has three drawing moves. The simplest draw is by l . . .llid3 2 h4 llif4 ! 3 'it>f5 �d6, when there is nothing for White to do. B lack can also head for the draw of diagram 296 by 1 ...�d6 2 �e2 (or 2 �c8 llif3 3 �b7 llih4 4 �g5 �e7 5 �xh4 �f6) llid7+ 3 �f7 llie5+ 4 �g7 �e7 5 h4 llid7 6 h5 llif6. Finally, the odd
=I=
311
Original (311): Black can play: 1 ) 1 . .�b2? 2 �b3 ! (in diagram .
308 the analogous position was a win whoever moved first, but now it is a reciprocal zugzwang; the rea son is that if White is to play, he has to allow Black's king to reach d2, and this is a draw by line 2 below) llib5 3 a4 ! llic7 4 �c4 �c2 5 a5 'it>b2 6 'it>c5 'it>c3 7 �b5 'it>b3 8 'it>b6! llid5+ 9 �c6 llib4+ 10 'it>c5 �a3 1 1 �c4 'it>a4 1 2 'it>b6! �a3 1 3 �b5 and wins. 2) 1. .�d2! 2 �b3 llie4 ! 3 �b5 (this is the line which led to a win in the analogous positions) �e3 4 �c4 (after other moves Black draws .
i.+� v lLi 219
by . . . �d4) lLid2+! S �b4 �d4 6 a4 �dS 7 a5 �d6 ! and Black is just in time to stop the pawn. In this vari ation the pawn reaches a7, but is then brought to a halt; shifted up a rank this line would result in a pawn pro motion. We aim to continue with a posi tion based on one of the 93 recipro cal zugzwangs with i.+�a3 v i., but first, here is a useful preliminary position.
+/-
3 12
Original (312): This is won for White who ever moves first, so it is an impor tant target position. Suppose firstly that White is to play. After 1 i.c2 Black may reply: W l ) 1. lLie5 2 a4 lLid7 (the key to this position is that White must not play a5 too soon; this allows Black to set up a defence with his knight on b8 and his king attacking the pawn from the rear) 3 i.d3 ! (3 a5? lLib8 ! 4
4 i.bS traps the knight, and while the pawn is on a4 it is defended by the bishop, and so is immune to at tack by the black king; 3 . . .
.
...
220 � +� v lll
4 a5 �e4 ! 5 �b4 tlld7 ! 6 �b5 lllb8 ! 7 �b6 �d4 ! 8 �e6 �c3 !) llld7 (threatening 3 . . . lllb 6 4 �b3 �e4 5 �b4 �d4 6 �b5 �c3 ! gaining a vi tal tempo and drawing) 3 �bl (now the threat is 4 �c4, because the reply . . . �d2 no longer gains a tempo; 3 a4? is still premature, and 3 . . . lllc5 ! 4 a5 llla6 ! 5 �d3 lllb 8 ! 6 �h7 �e2 defends) lllb6 (or 3 . . .lllc5 4 �c4 llla4 5 �c2 lllb6+ 6 �b5 tlld5 7 �c5 lllc3 8 �c4 ! �d2 9 �b3 ! with a reciprocal zugzwang; Black loses after 9 . . . llle4 10 a4 llld6+ 1 1 �c5 lllb7+ 12 'it>b6 tlld6 13 a5) 4 �a2 llla4+ (after 4 ...�e4 5 �b4! 'it>d4 6 �b5 ! Black has to move his knight because the move ...�c3 is not available, whereupon 6 . . . llld7 7 a4 ! wins as in line B 1) 5 �b3 lllc5+ (5 ... lllb6 6 �b4 'it>d4 7 �b5 !) 6 'it>c4 llla4 (6 ... llld3 7 �bl llle5+ 8 �d5 tlld7 9 �f5 is line W2) 7 �b3 lllb2+ 8 �c3 ! tll d3 and now White is to play and wins by 9 �c2 as in lines W 1 and W2. Now we are ready for the follow ing position. (313): 1 �b2! (not 1 �a2? �e2 2 'it>b3 lllb6 3 �b4 'it>d2 4 �b5 'it>c3) and now: 1 ) 1. lllb6 2 �a2 �el (2 ...'it>e2 3 �c2) 3 �c2 �e2 4 �g8 (a waiting move; 4 �f7 is also possible, but not 4 �e6? 'it>e3 ! 5 �c3 �e4 ! 6 �b4 �e5 ! gaining a tempo and drawing) �e3 5 �c3 ! llla4+ (5 . . .�e4 6 �b4 ! �d4 7 �b5 ! llld7 8 a4 ! wins as in line B 1 of the previous diagram) 6 �b4 ! lllb 2 7 �b3 llld3 8 �h7 llle5 (or 8 ... lllc 5+ 9 �c4 llla4 1 0 �c2 lllb6+ 1 1 �b5 llld5 12 �c5 lllc3 1 3 ••
+I=
313
Original 'it>c4 ! �d2 1 4 �b3 ! with a recipro cal zugzwang we have seen before in line B2 of the previous diagram) 9 �c3 ! tlld7 10 �c4 (not 1 0 a4? lllc5 ! 1 1 a5 llla6 ! 1 2 �c4 �d2 13 �b5 lllb8 ! 1 4 �b6 �c3 ! and Black reaches b4 in the nick of time) lllb6+ ( 1 0. . . llle5+ 1 1 �d5 llld7 1 2 �f5 lllb6+ 13 �c5 ! is line W2 of the pre vious diagram) 1 1 �c5 llla4+ 1 2 �b4 ! lllb6 1 3 �g8 �d4 14 �b5 ! llld7 1 5 a4 ! transposes to line B 1 of the previous diagram. 2) 1 . �el 2 �c l ! (not 2 �c2? �e2 ! and it is White who falls into an unlikely-looking reciprocal zug zwang; the lines 3 a4 lllc5 4 a5 �e3 5 �c3 llla6 ! 6 �c4 �d2 7 �b5 lllb8 ! , 3 �a2 �e3 ! 4 �c3 �e4! 5 �c4 �e5 ! and 3 �c3 lllb6 4 �a2 �dl ! are all drawn) �e2 (alterna tives transpose to lines considered later, for example 2 . . . llle5 3 �c2 tlld7 4 �a2 or 2 ... lllc5 3 �c2 llla4 4 �a2 lllb6 5 �b3) 3 �c2 ! and now: 2a) 3 �el 4 �a2 �e2 5 �d5 �e3 (5 ...lDc5 6 �c3 �e3 transposes) . .
...
.t+� v .!Li 221
6 'it>c3 ! .!Lic5 7 .ta2 .!Lia4+ (7 ...'it>e4 8 'it>c4 ! .!Lia4 9 .t b3 .!Lib6+ 10 �c5 .!Lid?+ 1 1 'it>c6 wins) is line B2 of the previous diagram. 2b) 3 .!Lib6 4 .ta2 ! .!Lid? (the al ternative 4 ... 'it>el 5 .tb3 'it>e2 6 'it>c3 <,L?e3 7 .ta2 .!Lia4+ 8 'it>b3 transposes) 5 .td5 ! .!Lic5 (5 ...'it>e3 6 'it>c3 ! .!Lic5 is the same) 6 'it>c3 ! 'it>e3 7 .ta2 .!Lia4+ 8 <,L?b3 and we again reach line B2 of the previous diagram.
...
Finally, the pawn moves back to a2.
�c4 3 �e5 (3 .!Lia3+ loses to 3 ... 'it>b3 4 .!Lib5 a5 ! 5 <,L?d5 a4 6 .!Lid4+ �c3 7 .!Lib5+ 'it>b4 8 .!Lid4 .tc3 9 .!Lie2 a3) a5 4 'it>e4 .tcl 5 .!Lie 1 (5 .!Lid4 a4 6 .!Lic2 'it>b3 7 .!Lid4+ 'it>c3 8 .!Lib5+ 'it>c4 9 .!Lid6+ <,L?b4) a4 6 .!Lid3 (6 .!Lic2 is the previous note) .ta3 ! (preparing ... �c3 and then a bishop move) 7 'it>e3 'it>c3 8 <,L?e2 .te7 9 .!Lie 1 'it>c2 10 .!Lid3 a3 11 .!Lie 1 + c;t>c3 1 2 'it>d 1 a2 0-1 because of 13 .!Lic2 .tg5 14 .!Lial �b2 15 .!Lic2 .th6 and the pawn pro motes. The following position depends on one of the 69 reciprocal zug zwangs with .i+�a2 v .!Li.
• \t>. ·tb • • - ii i - . - . . . . - - . . ��; . . . . - . . �·. '$� • • • • 'Ji ,
-I+
3 14
Allan-Nunn Szirak IZ 1987 (314): Even thought Black's a pawn hasn't moved, White's king is so far way that he cannot hold the game. The game continued 1 'it>e6 .td2 (there was an alternative win by l . . ..tb6 2 .!Lifl a5 3 �d5 �b4 4 �2 .tf2 5 'it>e4 'it>c3 ! 6 .!Libl + 'it>c2 7 .!Lia3+ 'it>b3 8 .!Lib5 a4 9 'it>d3 .tc5 lO �d2 'it>c4) 2 .!Lic2 (or 2 .!Lif5 a5 3 �d5 a4 ! 4 .!Lid6+ 'it>b4 ! 5 .!Lic4 .tc3 ! 6 .!Lie3 .tg7 7 .!Lic4 .tf6 8 .!Lie3 'it>c3)
3 15
=/+
Rezvov and Chemous, 1992 (version) lst Sp. Pr., Bron Mem. Tny. (315): White can play: 1) 1 .!Lie7? .tg5 2 .!Lif5+ 'it>g6 3
.!Lig3 (or 3 .!Lid4 h5 4 .!Lif3 �f5 5 �d7 'it>f4 6 .!Lid4 h4) .tf4 4 .!Lih 1 h5 5 �d7 h4 6 �e6 .tg3 7 �d5 'it>f5 ! 8 <,L?d4 'it>f4 ! 9 'it>d3 'it>f3 and Black wins. 2) 1 �c7? h5 ! 2 .!Lie? (2 'it>d6 h4 ! 3 'it>e5 h3 ! 4 .!Lif6 h2 5 .!Lih5+
222 i.+£!.
v
ll)
'iti>g6 6 li:)g3 lPg5 !) i.c5 3 li:)f5+ 'iti>f6 4 li:)g3 (4 li:)b4 lPg5 5 ll)f3+ 'iti>f4 6 li:)b4 i.t'2 7 li:)g6+ 'iti>g5 8 tt:Je5 i.g3) h4 ! 5 ll)fl 'iti>e5 6 lPc6 i.t'2 and now Black's king marches to g2. 3) 1 lPd7! i.g5 ( l . . .h5 2 �e6 h4 3 lPf5 ! h3 4 li:)f6 ! h2 5 li:)h5+! and 6 li:)g3 draws) 2 lPd6 ! (and not 2 lPe6? �g6! when White is trapped by the reciprocal zugzwang; Black wins af ter 3 �e5 �h5 ! 4 �f5 lPh4 ! 5 �e4 lPg4) 'iti>g6 (there is nothing better since 2 ... h5 allows 3 lPe5) 3 'iti>e6! (now it is Black who is in zugzwang) 'iti>h5 4 'iti>f7 ! lPg4 5 li:)h6+! i.xh6 6 'iti>g8 !, capturing the pawn. In the original version of the study, the fact that the zugzwang was reciprocal played no part, as the po sition couldn't arise with White to play. The king triangulation is an ad ditional feature making it impossible for the solver to overlook the recip rocal nature of the zugzwang. We end with the longest win in i.+al!. v ll) (42 moves). This is also one of the longest wins in the whole ending of i.+£!. v ll), but the honour is tied with a d-pawn position (see diagram 366). (316) : The longest win is with Black to play. The main line runs l . . .�t'2 ( l . . .li:)b7 2 �b2 lPt'2 3 �c3 lPe3 transposes to the next note) 2 'iti>b2! 'iti>e3 3 �c3 ! and now: 1) 3 li:)b7 4 �c4 li:)d6+ 5 lPb4 �d4 6 i.c6 ! li:)c4 7 a4 ! li:)e3 8 lPb5 li:)c4 9 i.g2 li:)d6+ 10 lPb4 li:)e8 1 1 a5 li:)c7 1 2 i.b7 ! (this is a recipro cal zugzwang) lPe5 13 lPc5 ! lPe6 14 �c6 li:)d5 15 i.c8+ lPe5 16 i.h3 .••
• • • • • •.t• • • • • • • • • • • • • • • • • • �- . . . � � - � � � 316
+/-
Original li:)b4+ 17 'iti>b5 li:)d5 1 8 lPc5 li:)c? 19 i.c8 (and now a simple zugzwang) 'iti>e4 20 'iti>c6 winning for White. 2) 3 'iti>e4 4 �c4! 'iti>e5 5 �c5 ! li:)b7+ 6 'iti>b6 li:)d6 7 'iti>c6 li:)c4 8 'iti>c5 ! li:)b2 (or 8 ... li:)a3 9 i.c6 'iti>f4 10 i.d5 'iti>e5 1 1 i.g8, threatening 12 i.h7 and 1 3 �b4; White wins after 1 1 . . .li:)bl 1 2 a4 !) 9 i.b5 ! (the idea is i.e2 followed by 'iti>b4; this forces Black's king to move down the board) �e4 (9 ... �e6 loses to 10 i.e2 ! li:)a4+ 1 1 �b4 ! li:)b6 12 �b5 li:)d7 1 3 �c6 ! li:)e5+ 14 lPc7 ! lPd5 15 a4 ! �c5 16 a5 !) 10 lPb4 ! 'iti>e3 1 1 a3 ! (and not 1 1 'iti>b3? li:)d l ! 1 2 a4 �d4 ! 1 3 �b4 �d5 drawing) �d2 ( 1 l .. .'iti>d4 12 i.e2 !) 12 i.c6 ! (12 �b3? fails to 12 ... �cl 1 3 i.d7 ll)d3 14 i.f5 li:)c5+ ! 15 lPb4 lPb2 ! , while the other bishop moves on the b5-e8 diagonal run into knight forks, e.g. 12 i.d7? li:)d3+! 13 lPc4 li:)e5+ ! or 12 i.e8? li:)d3+! 1 3 lPc4 lPc2 ! 14 i.g6 lPb2! 15 a4 li:)e5+!) lPe3 (alter natively Black may play 12 . . . ll)d3+ 1 3 �c4 ! , and now 1 3 . . . 'iti>e3 14 �c3 •••
.i+8
transposes, while 1 3 . . . �c2 14 .ie4 ! �b2 15 a4 ! wins because the bishop is not exposed to a knight fork) 1 3 �b3 ! tlid3 14 �c3 ! tlic5 ( 1 4 . . . tlie5 15 .id5 and now 1 5 ... tlid7 1 6 .ig2 transposes, while 15 . . .tlid3 1 6 .ic4 llic5 1 7 .ia2 is line 2b of diagram 3 1 3) 1 5 .id5 with a further branch: 2a) 15...llld3 16 .ic4 tlic5 17 .ia2 is again line 2b of diagram 3 1 3 . 2b) 1 5...tlla4+ 1 6 �b4 ! tlib2 17 .ic4 ! (the net effect of the previous six moves has been to transfer the bishop from b5 to c4 with gain of tempo - a small but important im provement) cli>d2 1 8 .ie6 �e3 (af ter 1 8 . . . llid3+ 1 9 'iti>c4 Black may choose between 19 ... 'it>e3 20 i.f5 tlib2+ 2 1 �b3 tlid l 22 .ic2 ! lead ing to the main line at move 32, or 19 . . . tlib2+ 20 'iii>b 3 tlid3 21 .if5 transposing directly) 1 9 'iii>b 3 tlid3 20 .if5 ! tlie5 21 'iii>c3 ! 'iii>f4 22 .ic2 ! tlic6 23 .ia4 ! tlid8 24 'iti>d4! 'iii>f5 25 cli>d5 f6 26 cli>d6 tlib7+ 27 'iii>c6 ! tlia5+ 28 'iii>b5 ! tlib7 29 'iii> b6 tlid6 30 .ib3 ! and the pawn advances. 2b) 15...llld7 (if Black were to move here, the win would be signifi cantly easier) 16 .ig2 (a precise choice of square; 16 .ihl tlie5 1 7 a4? allows 17 . . .tlid3 ! 1 8 a5 tlic5 ! heading for the familiar defensive spot on b8) tlic5 (now 1 6 . . . tlie5 loses to 17 a4 llid3 18 a5 tlic5 1 9 .ifl ! covering a6) 1 7 .ic6 (the start of a remarkable bishop manoeuvre which ends up on b l ; the point of this move is to prevent Black simply oscillating with his knight between c5 and d7) tlid3 1 8 .ia4 (now White has a chance to transfer his bishop
v
tLi 223
to the bl -h7 diagonal) tlic5 19 .ic2 (this is the position White was aim ing for) tlid7 (now we are in line B2 of diagram 3 1 2; readers will find more detailed analysis there) 20 i.bl tlib6 21 .ia2 tlia4+ 22 'li>b3 tlic5+ 23 'it>c4 tlia4 24 .ib3 (this is why the bishop had to bury itself on b l ; had it been further down the b l h 7 diagonal, then i t would now be on the other side of the white king and this move would be impossible) tlib2+ 25 'it>c3 ! tlid3 26 .ic2 tlic5 27 'iti>c4 tlid7 28 .if5 tlib6+ 29 'iii>c5 tlia4+ 30 b4! tlib2 31 'iii>b3 llld l 32 .ic2! llif2 33 a4 'iii>d4 34 'iii>b4 ! cli>d5 35 'iii>b5 ! 'iii>d6 36 a5 ! tlig4 37 a6 llif6 38 a7 tlid5 39 'it>a5 tlic7 40 'iii>b6 tlid5+ 4 1 'li>b7 tlic7 42 i.a4 with capture or promotion next move.
4.2:
i.+b� v ttJ
The next step is to consider the b pawn. In this case the attacker need not worry about the possibility of ending up with an a-pawn and wrong bishop combination, but on the other hand the knight finds it easier to re strain a b-pawn than an a-pawn. If the defending king is in front of the pawn the result is normally a draw, but with the right bishop there are some winning positions based on mating threats: (317): In this position Black wins even if White moves first: 1 tlig4 (in the game White resigned at once) .ig5 2 tlif2 'it>f3 3 llih3 (3 tlig4 .id8 4 tlih2+ 'iii>g3 ! 5 llifl + 'iii>h 3 ! is also lost) .ih4 4 h2 .iel 5 tligl+ 'iii>f2 !
224 i.+£!. v ttJ
3 17
-/+ Rautestreich-Nemet 1964
on a6 since it cannot be attacked b\ �a7. White to play wins by preveni ing the knight check: 1 i..d6 ! �b5 2 �a7 �a5 3 i.. g3 �b5 4 i..e l and Black is in zugzwang. The formation in the top left cor ner looks like a coathook, so we will refer to this and analogous positions as 'hook' positions. There are 37 positions ofrecipro cal zugzwang with i..+l!.b7 v tt:J, some of which are featured in the following analysis.
6 tt:Jh3+ (6 tt:Je2 i..d 2 7 �h3 i.f4) �fl ! 7 tlJgl i..g 3+ winning. If the defender's king cannot reach the square in front of the pawn, then his drawing chances are poor. The following position is typical of the whole ending with i.+£!. v tt:J.
Original
Averbakh, 1955 (318): Black to play draws by 1 . ..tlJc7+ ! 2 �b8 tt:Ja6+ ! 3 'it.?c8 'it.?c6 and Black's knight is securely posted
(319): Black to plays draws by 1 . . .tt:Jc5+ !, so suppose that White moves first. If White can establish his king on b6 then he wins; other wise Black can draw. White may try: 1) 1 �d5? �d7 ! (now White has fallen into a reciprocal zugzwang) 2 i..b6 (Black has to concede his control over c5, but it doesn't help White when his bishop is blocking b6) tt:Jb8 (a second reciprocal zug zwang) 3 �c5 tt:Ja6+ ! 4 �b5 tt:Jb8 ! 5 i.. a5 (White has to waste a move
.t+� v tLJ 225
clearing b6) �d6 ! (Black vacates d7 just in time) 6 .td8 (after 6 'it>b6 tlld7+ ! White's king cannot stay on b6) �d7 7 .tb6 �d6! 8 .ta5 tlld7 ! and White cannot make progress. 2) 1 �eS? �e7 ! (reciprocal zug zwang) 2 .tb6 (or 2 �d5 'it>d7! lead ing to line 1 ) �d7 3 �d5 tllb8 also transposes to line 1 . 3 ) 1 'it>d4! and now: 3a) 1 �d7 2 'it>d5 ! is line 3c. 3b) 1 tllb8 2 �c5 �d7 3 'it>b6 ! �d6 4 .tb4+! (White mustn't waste time, for example 4 .tel ? tlld 7+! draws) �d5 (4 ... �d7 5 .tel �d8 6 i.h4+ 'it>e8 7 .tg3 tlld7+ 8 �c7 �e7 9 .th2 'it>e6 1 0 �d8) 5 .tel tlld7+ 6 �b5 ! (6 �c7? tllc5 !) �d6 7 .tg3+! �d5 8 .th2 �e6 9 �c6 and wins. 3c) 1 �e7 2 �e5 ! (reciprocal zugzwang) �d7 (2 . . . tllb8 3 c5 tllc6 5 �b6 tllb8 6 .tel wins as in line 3b, while 2 . . . tllc 5 fails to 3 .tb4!) 3 �d5 ! tllc7+ 4 e5? is most simply met by 1 .. .tllc5 since the ending of .t+tll v tD is a simple draw in this case) �e8 ( 1 . . .tllb8 2 �e5 �e8 3 �d6 is similar) 2 �e6 ! �d8 3 .ta5+ �e8 4 �d6 tllb8 5 i.d2 tll a6 6 .tf4 �d8 7 �c6, fol lowed by �b6, and White's pawn promotes. Black to play draws after l . . .�e8 ! (not l . . .�e7? 2 �e5 ! as in line 3c above) 2 �e5 (White cannot play
Moving the pawn back to d6, the result of the hook position differs from diagram 3 1 8. Now White can only win under more restricted cir cumstances.
...
...
...
+I=
320
Original (320): 1 .td5 ! (unlike the position one rank further up, this is recipro cal zugzwang) �c5 2 �a6! (not 2 .tg2? �d6 3 �a6 tllb3, followed by . . . tllc 5+, with an easy draw) �b4 3 .tg8 �a4 4 .th7 �b4 5 .tc2 (Black is in zugzwang) tllc6 6 b7 ! tll b8+ 7 �b6 �c4 8 .ta4 and wins. White also wins if he can check on d7 or e8 with his first move, for example if the bishop starts on g6, White wins by 1 .te8+! �b4 2 .td7 ! (a waiting move; the immediate 2 �a6? fails to 2 ... tllb 3 !, transferring the knight to c5) �c3 3 .te6 tllc6+ (3 . . .�b4 4 �a6) 4 �b7 tlla5+ 5 �a6 �b4 6 .tg8 transposing to the analysis just above. If the bishop cannot move to d5, d7 or e8 on the first move then the position is drawn with White to play.
226 .t+l!. v lLl
If White's king is on c7 and Black's knight is on a5 then the posi tion is drawn because the knight is immune to attack. If, however, the knight is on d8 then White usually wins.
321
Curiously enough, Krivenko was given a special 'Junior Award' for a 'White to play and draw' study in the 1 978 Roycroft Jubilee Tourney, even though the position after six moves is exactly that of Kling and Horwitz, which had been proved a win over 125 years earlier! In the following position Black's king attempts unsuccessfully to ap proach the pawn from the side.
+I-
Kling and Horwitz, 1851 (version) (321): Here it doesn't matter who moves first. White is aiming for one of the two zugzwang positions .ic4 v cjr>a5 or .ic4 v 'it>c5, but he must never allow Black to play ...ll:ie6+. He achieves this aim by 1 .ig8 (a waiting move) �a6 ( I . . .cjr>a5 and I . . .cjr>c5 both lose immediately to 2 .ic4, so this is forced) 2 .ib3 ..t>b5 (forced, but now White can transfer his bishop to d7 without allowing a check on e6) 3 .ia4+ �a5 (3 ... cjr>a6 4 .id7 cjr>a5 5 .if5 loses one move more quickly) 4 .id7 cjr>a6 5 .if5 (now 5 ...'it>b5 can be met by 6 .id3+ and 7 .ic4) �a5 6 .ig4 (another waiting move; now Black cannot avoid moving his king onto the fl -a6 diagonal) 'it>b5 7 .ie2+ �a5 8 .ic4 and wins.
322
+/-
Averbakh and Khenkin, 1958 (322): White wins, even if Black moves first: l ...�e7 (I...cjr>d7 2 .ig4+! 'it>e8 3 .ic8 'it>e7 4 cjr>d5 cjr>f6 5 cjr>d6 cjr>f7 6 �c7 cjr>e8 7 .ig4 ..t>e7 8 .id7 is considerably easier) 2 .ig4 (this allows the knight some freedom, but it is more important to keep Black's king cut oft) ll:ib7+ 3 cjr>c6 ll:ia5+ 4 ..t>b5 ll:ib7 5 .ic8 (blocking the knight's a5-b7-d8 route) ll:id6+ (or 5 . . . ll:id8 6 �c5 ! as in the note to Black's first move) 6 cjr>c5 ll:ie4+ 7 cjr>c6 ll:id6 8 .ia6 (Black's knight has been forced to a bad square; White can win by transferring his bishop
.t+� v lb 227
to d5, but he must not allow . . . lbc4 and ...lba5+) �e6 (8 ... lbt7 9 .tc4 lC!d6 10 .td5) 9 .tn �e7 10 .te2 (a waiting move; not 10 .tg2? lbc4 !) �e6 1 1 .tf3 �e7 (if now 1 1 ...lbc4, then 12 .td5+) 12 .td5 and Black is in zugzwang. An almost identical position arose in McNab-King, Brit ish Ch 1 988 and Black won using more or less the method given by Averbakh and Khenkin.
�e4 .tb6) .txgl 10 'it>g3 ! and Black loses his pawn.
+/-
324
Averbakh, 1955
323
=I+
Onate, 1969 1 st HM, Schach Echo (323): White to play can draw by precise defence: 1 'it>h3! 'it>f3 2 lbe6 ! (but not 2 lbd3 ? �e2 ! 3 lbc5 �fl ! 4 lbe4 g2 ! 5 lbd2+ 'it> g l and Black wins after 6 �g4 �f2 7 lbf3 .tc7 or 6 lbe4 .i.f4 7 lbc3 �fl 8 lC!e2 .i.e5) g2 3 lbg5+! 'ifi>f2 4 'it>h4 ! (the key move, making space for the knight; 4 lbe4+? �gl 5 lbd2 .tf4 6 lbf3+ �f2 7 �g4 .tb8 wins for Black) .ta? 5 lC!h3+! �f3 6 lbg5+! �e2 7 lC!h3 ! .tf2+ 8 �g4 ! �e3 9 lbgl ! (9 �f5? �f3 ! 10 lbg5+ �g3 1 1 lbe4+ �h2 1 2 lbg5 .te3 1 3 lbf3+ �g3 14
(324): White to play wins imme diately by 1 .td7, so suppose that Black moves first. After 1 . . .lbe8 2 .td7 ! (White must not waste time; for example 2 .i.h3? �e7 ! draws) lbg7 (or 2 ... lbd6 3 g7 lbf7+ 4 �h7 ! lbg5+ 5 �g8 �e7 6 .tf5 �e8 7 .tg6+ �e7 8 �h8 and the pawn pro motes) White can lose a tempo with the manoeuvre 3 �h7 ! lbh5 4 .tg4 ! lbg7 5 .tc8 lbh5 (5 . . . lbe8 6 .td7 ! lbg7 7 'it>h6 is the same) 6 �h6 lbg7 7 .i.d7, and now Black falls into zug zwang. The following two position are based on surprising reciprocal zug zwangs, chosen from the 78 with .t+�b6 v lb. (325): Suppose White is to play. For the moment he cannot advance the pawn because 1 b7? lbc6! per manently imprisons the white king. However, king moves are also im possible; 1 �b7? and 1 �b8? shed
228 .t+� v lLl
325
+I=
326
+I=
Original
Original
the pawn straight away, and 1
zugzwang. The logic behind it is that White would win easily, except for the unfortunate position of his bishop on a7. If he could play .tb8 and extract the bishop then Black would be dead lost. In the original position this is impossible because Black meets 1 .tb8 by 1 . . .lLlc4 ! 2 b7 lLla5 ! , but if White had a bishop check then he could promote the pawn. It follows that Black can never play his king to a dark square. White is also largely restricted to light squares for his king, because other wise he runs into a knight fork, e.g. 1
.t+.0. v lb 229
the pawn) 3 �d5 ! �e2 4 .tb8 (now that c4 is covered White can move his bishop) 'ifr>d3 5 .tg3 'ifr>d2 6 �c5 �e2 7 �b5 ll:ib7 8 .th4 ll:id6+ 9 'ifr>a6 �d3 10 .te7 winning. 2) 1 �g6 2 �e6 ! (another recip rocal zugzwang) �h5 (2 . . . ll:ic6 3 �d5 ll:id8 4 .tb8 �f7 5 .tc7 ll:ib7 6 'ifr>c6 ll:ia5+ 7 'ifr>d7 �f6 8 .td6 �f5 9 .tb4 wins, while 2 . . .�h7 3 'ifr>d5 ! �g6 transposes to the main line) 3 'ifr>d5 ! 'ifr>g6 4 .tb8 ! �f7 5 .td6 �e8 6 .tb4 ! ll:ib7 7 �c6 ! ll:id8+ 8 �c7 ll:ie6+ 9 �c8 ll:id8 10 .tc5 (zug zwang) ll:ic6 1 1 b7 and wins. ...
Now suppose that the pawn is on b5 . The results for the hook position are basically the same as with the pawn on b6, but the White to play winning process is slightly more complicated.
+I=
327
Original (327): After 1 .td4 ! we again have a reciprocal zugzwang. White wins by 1 .. .�c4 ( 1 .. .'ifr>a3 2 �a5 ! �b3 is the same) 2 �a5 ! �b3 3 .tg7
(but not 3 .te5 ? lbc5 ! 4 b6 ll:id7 5 b7 'ifr>c4 ! and White cannot prevent . . . 'ifr>c5 followed by ... 'ifr>c6) 'ifr>a3 (now 3 . . .ll:ic5 4 b6 ll:id7 5 b7 ! 'ifr>c4 loses to 6 .tf8 ! ) 4 .th6 (White must not allow . . . ll:ib2, so this manoeu vre is only possible when B lack's king is on a3) �b3 5 .t c l (unlike the position a rank up, this is a sec ond reciprocal zugzwang) ll:ic5 6 b6 ! �c4 7 .ta3 ll:ib7+ 8 �a6 ll:id8 9 .te7 lbc6 1 0 b7 ll:ib8+ 1 1 'ifr>b6 �d5 1 2 .th4 ll:id7+ 1 3 'ifr>b5 ! with a familiar win. As before, White also wins when he can give an initial check on e7 or f8, for example suppose White's bishop starts on g5. After 1 .te7+ ! 'it>c4 White can only win with a sub tle manoeuvre. He would like to play �a5. but the immediate 2 'ifr>a5? fails to 2 . . . ll:ic3 ! 3 b6 ll:id5 ! 4 b7 ll:ixe7 ! . Evidently the bishop is badly placed on e7, so first of all White must improve its position by 2 .ta3 (2 .tf8 �b3 3 .tg7? ll:ic5+ ! draws) 'ifr>b3 3 .tf8 (only this move prevents the reply . . .�c4; after 3 .td6 �c4 4 'ifr>a5 'it>b3 5 .te5? Black draws by 3 ... ll:ic5, as in the analysis of the dia gram position). Now 3 . . . �c4 4 �a5 �b3 5 .tg7 �a3 6 .th6 wins as in the diagram analysis, but even after 3 . . . �c2 Black cannot avoid transposing by 4 .tg7 ll:ic5+ (4 ... �b3 5 �a5) 5 �b6 ! ll:ia4+ (or 5 . . . ll:ib3 6 'ifr>c6 ll:ia5+ 7 �c5 ll:ib7+ 8 �b4 �d3 9 b6 �e4 10 �c4 ll:id8 1 1 'ifr>c5 win ning) 6 �a5 �b3 7 .td4, winning as above. The following two studies are very attractive; the first is based on
230 �+� v lll
reciprocal zugzwang and the second on stalemate.
328
+I=
'Zakhodiakin, 1931 lst Pr., '64 ' (328): Black's king and knight are widely separated, so one might guess that White's strategy was based on isolating and surrounding the knight, but in fact the solution depends on forcing the knight to the kingside, where it obstructs Black's king . The winning line is 1 �c5 ! (not 1 ..td5? lllb6+! 2 �c6 llla4 and the knight slips away) lllc7 2 ..td6! llle8+ and now: 1 ) 3 �d7? lllg7 ! 4 .i.g6 ..tg8 ! 5 rj;e? ..th8 ! and this position, rather surprisingly, is reciprocal zugzwang. The bishop must stay on g6 to cover h5 and f5, but this leaves Black's king free to oscillate between g8 and h8. The only way to stop this is by 6 �f7, but then Black replies 6 ... lllf5 ! and the stalemate allows the knight to escape. 2) 3 �e7! lllg7 (3 . . .lllc7 4 �f7 ! llld 5 5 g6 ! forces the pawn home) 4
�g6 ! (now the reciprocal zugzwang arises with Black to move) �g8 (Black is forced to allow a bishop check, which gives White the op portunity to rearrange his pieces) S .i.f7+ �h7 (5 . . . �h8 6 �f6 ! ..th7 1 �e5 ! �h8 8 �f4 ! transposes to the main line) 6 �f6 �h8 7 �e5 ! (not 1 �g6? llle 6! with another stalemate defence; White's aim is to reach g4 with his king when Black's king is on h7; then . . .�h8 can be met by g6 winning the knight) �h7 8 �e4 ! (the parity is wrong for the immedi ate �f4-g4, so White has to lose a tempo; his king must remain in con tact with f5, so this triangle is the only possibility to lose a tempo) �h8 9 �f4 ! ..th7 1 0 �g4 ! �h8 1 1 g6! and the knight falls.
329
I=
Brenev, 1931 (end of study) 3rd HM, '64 ' (329): White draws with a short but sharp stalemate combination: 1 .. ..i. b8+ 2 �h4 ! g3 3 lllh5 ! (but not 3 llld 5? �e4 ! 4 lllc 3+ �f3) g2 4 lllg 3+ ! �xg3+ (or 4 ...�f4 5 llle 2+!
.t+� v lb 231
�f3 6 lbgl + ! �f2 7 lbh3+! with an easy draw) 5 �h3! and promoting to queen or rook delivers stalemate. There are 1 3 3 reciprocal zug zwangs with .t+�b5 v lb, but as we have already seen some in diagrams 327 and 328 we will not explore them further. Now we will move to the case of the pawn on b4. The hook position W�a5, .tg6, �b4v B�b3, lba3 is now drawn. After 1 .td3 Black sim ply replies 1 ...�c3 2 �a4 �b2 ! , and the manoeuvre which won in dia gram 327 doesn't work here because it would involve playing the bishop to cO! The following study is unsound, but the defence overlooked by the composer is quite subtle.
�b4 5 �e4 �c4 6 .t g l traps the knight) 3 g6! lbf4 4 g7 ! llle6 5 g8l:t! (5 g8'it'? and 5 g8.t? are stalemate, while after 5 g8lb? lbc5 Black's knight arrives in time to disrupt the mating net) and wins. However, the diagram position is a draw. The ba sic idea is that Black keeps his knight on g5 for as long as possible. If White attacks it along the c l -g5 di agonal Black replies ...lbh7 (but not ... lbh3?, when .te3 traps the knight), while if White attacks it along the d8-h4 diagonal, Black replies ... lbh3 (not . . . lbh7? allowing .te7). In both cases White's king has to go all the way to the h-file to trap the knight, which gives Black's king time to ap proach the g4-pawn. The detailed analysis runs 1 . . .lbg5 ! 2 .tf6 and now: I) 2 lbe6? (this is too close to the white king) 3 �c3 lbf4 (3 . . . �a3 4 �c4 and so on) 4 .td4 (but not 4 g5? lbe6 5 g6 lbf4 ! 6 g7 lbd5+!) lbe6 5 .te3 �a3 6 �c4 �b2 7 �d5 lbc7+ 8 �d6 lbe8+ 9 cl;e7 lbc7 1 0 .td4+ �b3 1 1 .te5 lbd5+ 1 2 �e6 and wins. 2) 2 lbb3! 3 .td4 (3 �d2 lbf2 rounds up White's pawn; the move played threatens .te3) lbg5 ! 4 .te3 (4 �d3 �b3 5 �e3 �c4 and Black's king is too quick) lbh7 ! (and not 4 . . . lbf7? 5 �c3 �bl 6 .tf4 lbh8 7 g5 lbg6 8 .td6 with an easy win) 5 .tc5 (threatening .te7) lbg5 ! 6 .tb4 (6 .te7 lbh3 ! ) �al ! (Black avoids 6 . . . lbh3? 7 .td2 ! , threatening .te3, and forcing the pawn's advance) 7 .tc3+ (7 .td2 lbh7 ! 8 �b3 �bl ! 9 .tf4 �al !) �a2 ! 8 .tb2 lbh7 ! (but ...
• • • • • • • • • • • • • • • • . ··· � · • • • • •m�• • • • • • • 330
+I=
Alexandrov, 1933 (end of study) Shakhmaty (330): Black was to play in this 'White to win' study. The composer's analysis continued 1 .. .lbf2? 2 g5 ! lbh3 (2 . . . lbg4 3 .td4 �a3 4 �d3
...
232 i.+� v lLi
not 8 . . . lLih3? 9 i.c l !) and White is not making progress. If White were to play in the dia gram, he would continue 1 i.c 1 ! (preventing . . . lLig5, which would set up the defence used above) �al 2 i.e3 (covering f2 and thereby threat ening �d3) �a2 3 �d3 lLid6 4 �d4 with a simple win. The great endgame expert makes the most of his chances in the follow ing practical example:
•
• • • • • • • � . � . � � d • • • • . . . ... . • • • • • • • • • • • • 331
victim to a reciprocal zugzwang) 4 lLic6 (or 4 lLif3 g4 ! 5 lLie 1 i.e4 ! 6 �e6 �e3 7 �f6 �f2 8 �g5 g3 ! 9 �g4 i.bl 10 �h3 i.g6 1 1 �g4 i.h5+ winning as in diagram 268) g4 5 lLid4 i.c8 (5 . . . i.d3 6 lLie6+ �f3 ! 7 lLig5+ �e3 was quicker) 0- 1 in view of 6 lLic2 g3 7 lLiel i.h3 8 'it>g6 �e3 9 �g5 i.fl ! 10 �h4 �f2 1 1 �g4 i.a6 12 �h4 i.b5 1 3 �g4 i.d7+ 14 �f4 i.f5 with the usual win. There are 1 66 reciprocal zug zwangs with i.+�b4 v lLi . Now suppose that the pawn is on b3. Curiously, the hook position is now a win if White moves first, pro vided that he has an initial check on the long diagonal.
=/=
Petrovit-Averbakh Graz Open 1987 (331): The game went on l ... i.f3 2 lLid7 (White could also draw by 2 lLic4, for example 2 ...g4 3 lLie5 ! g3 4 lLig6+ �e3 5 lLih4 ! i.e4 6 �g5, 2 . . .i.d5 3 lLie5 ! i.e4 4 �e6 ! , or fi nally 2 ...i.e4 3 lLie3 ! and Black is on the wrong end of a reciprocal zug zwang) i.e4 (2... g4 3 lLie5 ! is a draw as in the previous note) 3 lLie5 ? (missing a simple draw by 3 �e7, followed by lLif6; 3 �f7 and 3 �g7 also draw, but every other move loses) i.f5 ! (suddenly Black falls
+I=
332
Original (332): White to play wins by means of 1 i.f6+ ! �c2 2 �a3 lLic l (the line 2 . . . �bl 3 i.b2 shows why shifting the position down a rank has worked against Black - he cannot play 3 . . . lLibO) 3 b4 ! lLid3 4 b5 ! lLic5
.t+�
5 b6 ! (not 5 �b4? lLJci7 !, followed by ... �d3) �d3 6 �b4 liJb7 7 �b5 'it>e4 8 �c6 llJa5+ 9 �c7 'iii>d 3 1 0 .th4 �d2 1 1 .tg3 �e2 1 2 .td6 followed by .tb4 and White wins. The following position is based on one of the 1 17 reciprocal zug zwangs with .t+�b3 v llJ.
v
lLJ 233
time White has a good reply) 3 .tf5 ! (this creates a reciprocal zugzwang; 3 .te4? lLJc5 is obviously bad, but 3 .th7? fails more subtly to 3 . . . llJel ! 4 �b4 liJf3 ! 5 �c3 lLJe5 6 b4 �a2 ! 7 b5 �a3 ! 8 b6 liJd7 9 b7 �a4 ! draw ing) and at this point there is a fur ther branch: 3a) 3 llJb2 4 b4 lLJc4+ trans poses to line 3c. 3b) 3 llJel 4 �b4 ! liJf3 5 �c3 llJe5 (5 ...llJg5 6 b4 liJf7 7 b5 liJd6 8 b6 !) 6 b4 ! �a2 7 b5 ! 'iii>a3 8 b6! lLJc6 (forced, as d7 is covered) 9 b7 win ning since 9 ...�a4 loses to 10 .td7 ! . White needs to have the option of moving his bishop to the e8-a4 di agonal, which explains why 3 .th7? was bad. 3c) 3 llJeS 4 b4 ! llJc4+ 5 �b3 ! liJd6 6 .td3 (zugzwang forces Black to give up control of b5) liJb7 7 b5 (once White's pawn starts marching, the win is straightforward) llJc5+ 8 �c4 liJb7 9 b6 �b2 1 0 .tf5 llJa5+ 1 1 �b5 liJb7 1 2 .tc8 liJd8 1 3 �c5 �c3 14 �d6 �b4 1 5 b5 18 .t n + 'it>c5 19 .tc4 and wins. ...
...
...
+I=
333
Original (333): 1 .th5 ! (White's only win ning chance is to imprison Black's knight, so he has to cover d l ; 1 b4? llJc4+ ! 2 �b3 liJd6 and 1 .tg6+? �c 1 ! 2 b4 llJc4+ ! 3 �b3 liJd6 let Black escape) and now Black has to meet the threat of .te2, totally sur rounding the knight: 1 ) 1 �al 2 .tg4 ! (avoiding the traps 2 .tg6? lLJc4+! and 2 .te2? liJd3 !, which employ stalemate to extract the knight) �bl (2 ... liJd3 3 .tf5 ! transposes to line 3) 3 .te2 �al 4 b4 liJd3 5 b5 with an easy win. 2) 1 �cl 2 .te2! �bl 3 b4 wins as in line 1 . 3) 1 lLJd3 2 .tg6 ! �al (again Black plays for stalemate, but this ...
...
...
Finally, we deal with the pawn on b2. The most interesting analysis of this ending is contained in a study by Timman. We take the position af ter the first move. (334): Timman initially published a study which depended on the dia gram being a win, but later he dis covered a draw and reversed the colours. The result is now correct, al though some of his variations con tain errors.
234 .t+L!. v lLJ
'iti>el .td3 11 ll'ie6 g4 and now Tim man gave 12 lt)d4+ ? as 'the most elegant way' to draw, although Black wins after 12 ... 'iti>e4 1 3 ll'ie6 (13 ll'ie2 .txe2 14 'iti>xe2 g3 !) 1 3 ... g3 14 lt)g5+ 'iti>e3 15 ll'ih3 g2 16 lt)gl .tb5 17 lt)e2 �f3 18 lt)gl + 'iti>g3. The correct method is 12 'ittd 2! (forcing Black to concede control of f5) .tc4 (or 1 2 . . . g3 1 3 'iti>xd 3 ! g2 14 lt)d4+ ! ) 13 lt)g7 ! .tf7 14 'iti>el !. 2b) 5 'it>e5 6 ll'ig4+! 'iti>f4 7 lt)h6! .td5 8 'iti>c3 ! .ta2 9 'it>d3 ! (and not 9 �d4? .te6! and White is in zug zwang) .te6 10 �d4 ! reaching a re ciprocal zugzwang with Black to play. There are 79 reciprocal zug zwangs with .t+.0.b2 v lt). The fol lowing position is based on one of them. •••
=I+
334
1imman, 1994
After 1 ll'ig2! Black may play: 1 ) 1. .tc2+ 2 �c3 (not 2 �xc2? g5 ! followed by ...'iti>f2 and Black wins) g5 3 �d4! g4 4 ll'ie3 ! (not 4 'iti>e5? 'it>f3 ! 5 ll'iel + 'it>f2 ! 6 'iti>f4 g3 ! 7 c;t;>g4 .td 1 + winning as in diagram 32 1 ) g3 5 ll'ig2 ! 'it>f3 6 ll'ih4+ ! 'it>f2 and now Timman gave 7 'iti>c3 .th7 8 'iti>d4 .tbl 9 'it>e5 �e3 1 0 ll'ig2+ 'iti>f3 1 1 ll'ih4+! 'it>g4 1 2 ll'ig2! .td3 13 ll'iel .tg6 14 ll'ig2 'it>f3 15 ll'ih4+ ! 'ittf2 16 ll'ixg6 g2 17 ll'if4 ! as a draw ing line, which is indeed perfectly adequate. However, 7 'itte5, which Timman gave as losing, also draws after 7 . . . 'iti>e3 8 ll'ig2+ c;t;>f3 9 ll'ih4+ ! c;t;>g4 10 ll'ig2! .ta4 l 1 ll'ie3+ 'iti>g5 12 ll'ig2 ! .tc6 1 3 ll'iel ! (and not Tim man's 1 3 ll'ie3? .tb7 14 �d4 'it>f4 15 'it>d3 'iii>f3 16 'iti>d2 .te4) .tb7 14 'it>d4 'it>f4 15 ll'id3+ ! 'it>f3 1 6 ll'ie5+ ! c;t;>n 17 ll'ig6 ! . 2 ) 1 . .t e4 2 ll'if4+ ! 'it>e3 3 ll'ih5 ! g5 4 ll'if6 ! 'it>d4 5 'it>b2 ! with a further split: 2a) 5 .tf5 6 �c 1 �e5 7 ll'ie8 ! .tg6 8 ll'ic7 ! �e4 9 �d2 'it>f3 10 •.
.•
•••
+I=
335
Original (335): The obvious 1 b4? doesn't work after l . . .lt)e4+! 2 �b3 (or 2 c;t;>d4 lt)d6 ! 3 .tg6 'it>b2) ll'id2+! 3 'it>a4 c;t;>b2 ! 4 b5 'it>c3 5 b6 lt)e4 ! 6 .tf7 ll'id6! 7 .te6 'it>d4 8 �b4 ll'ib7
.t+.0i
with a draw similar to that after aris ing after 7 �e5 in line 1 of the pre vious diagram. The solution runs 1 .tg6 ! (not 1 .tc6? lDc4 2 b4 lDd6 ! drawing after 3 �d4 �b2 or 3 .td5 c4 lDd6+ 1 1 'it>d5 lDb7 12 'itc6 lDd8+ 13 �c7 lDe6+ 14 c7 ! followed by the win of diagram 321 . ••
•.
•.
lLJ 235
• ••• • = • • •.t • • • • • • • • • • • • • • • • �� . . . . . - . +I=
336
Original
•.
We end with the longest win in the ending of .t+b.0i v lb (40 moves). (336): White wins by 1 'it>b6! (the threat is 2 'it>c6, but not 1 b4? �d2 drawing after 2 b5 lDd6 or 2 �b6 �c3 3 �c5 lDc7) and now: 1) 1. 'itd2 (allowing White to execute his threat) 2 �c6! lDg7 3 'itd7 'it>e3 4 b4 ! �d4 5 .tg8 ! lDf5 6 �c7 ! lDg7 7 �c6 lDf5 8 b5 and wins. 2) 1. lDf6 2 .tf5 ! (but not 2 .tg6? lDd7+ 3 �c7 lDe5 gaining a tempo, nor 2 .t b l ? �d2 threatening . . . �c l , with a draw in both cases)
v
lDd5+ 3 �c5 ! lDc7 4 'it>c6 lDa6 5 'it>b6 lDb4 6 'itb5 lDd5 7 .te4, trans posing to line 3. 3) 1. lLJd6 2 �c5 ! lDb7+ 3 �d4 ! (threat b4) lDd6 4 �d5 ! (not 4 b4? lDb5+! 5 �c4 lDc7 drawing) lDb5 5 'it>c5 ! lDc7 6 �c6 lDa6 7 �b6 lDb4 8 'itb5 ! lDb4 ! lDb6 17 .th7 lDc3 1 9 .tg8 ! lDe7 20 b6! lDc6+ 2 1 'itb5 ! lDd8 transposes to line 3b) 1 3 .th7 lDa4 1 4 b5 ! lDb6+ 1 5 �d4 lDa4 16 .tg8 'itc2 17 'itc4 ! lDb2+ 1 8 'itb4 lDd3+ 1 9 'ita5 lDc5 20 b6 �d3 2 1 �b5 ! lDb7 2 2 .td5 4Jd8 2 3 �c5 ! �c3 24 .tc6 �b3 25 .td7 lDb7+ 26 �b5 c3 27 .tc8 lDd8 28 �c5 ! 'itb3 29 �d6 'itb4 30 �c7 and wins. 3b) 9 lDf6 10 .tc6 lDg4 1 1 �c5 lLJe5 12 .te4 ! lD
•.•
..•
236 i.+�
v
lLl
with White to play and not Black; this actually makes the win more complicated) 1 6 i.f3 (the fact that we are following the optimal line of play makes the win appear more complex than it really is, for example if White were prepared to waste one move he could play the obvious continuation 1 6 �c4 lllb6+ 17 ..tr>c5 ll:la4+ 1 8 �d4, which loses a tempo and transposes into line 3a) ..tr>c2 ( 1 6...lllb6 17 i.d5 llla4 1 8 i.b3 wins far more easily) 17 i.h5 (Black is in zugzwang, because he cannot allow White to play i.d l ) lllb6 ( 1 7 ... �d2 1 8 i.f7 �c2 1 9 �c4 is the same) 1 8 i.f7 ll:la4 1 9 �c4 lllb2+ 20 �b4 llld3+ 21 �a5 ll:le5 (2 1 . . .lllc5 22 b6 �d3 23 ..tr>b5 ! lllb7 24 i.d5 is line 3a) 22 i.g8 �c3 23 b6 ! lllc6+ 24 �b5 ! ll:ld8 25 �c5 ! ll:lb7+ 26 �c6! ll:la5+ 27 ..tb5 ll:lb7 28 i.d5 ll:ld8 29 �c5 ! (now we are back in familiar territory) �d3 30 i.c4+ �c3 3 1 i.a6 �b3 32 i.c8 �a4 33 �d6 �a5 34 ..tr>c7 ..tr>b5 35 i.h3 �a5 36 i.g4 �b5 37 i.e2+ �c5 38 i.c4 ll:lc6 39 b7 ! with promotion or capture of the knight next move.
4.3:
on c6 facing an enemy knight on c8. Black can try 1 . . .ll:le7+, meeting 2 �d7 by 2 ...ll:ld5 and 2 �d6 by 2 . . . lllc8+ 3 �d7 ll:la7 (3 ... lllb6+ 4 �c6 lllc8 is normally inferior, be cause White has gained a tempo over the initial position). He can also try 1 . . .llla7+ with a similar idea. Both lines depend on the availability of a7; we can conclude that the de fender has better chances with a c pawn than a b-pawn. Starting with the pawn on c7, the hook position (W�b8, i.b3, �c7 v B�c6, ll:lb6) has the same evalu ation (+I=) as the corresponding po sition with a b-pawn (diagram 3 1 8). White to play can win by the same method as before, namely by 1 i.e6, but now there is an even quicker win by 1 i.a4+ �d6 2 �b7 �c5 3 i.e8. Some positions have different re sults, though:
..t+� v ttJ
In many cases the play doesn't dif fer significantly from the analogous b-pawn situation, so we will concen trate on those positions which dis play new features. Many of these differences depend on the knight having access to an extra square on the a-file. For example, suppose that White has a pawn on c7 and a king
337
=I=
Horwitz, 1884 (version) Chess Studies and Endgames (337): If this position is shifted one file to the left then White to play
i.+.0. v lLl 237
wins, but here Black can draw by making use of a stalemate defence: 1 �d7 ll:lc8 ! 2 i.e3 ll:le7 3 i.d4 ll:lc8 ! 4 i.c5 'iita8 ! 5 'iitc6 ll:lb6 ! (per haps the most surprising move in the solution) 6 i.d6 'iita7 7 i.c8 �a8 (7 . . . 'iita6 also draws) and White can not make progress.
There are 7 1 reciprocal zug zwangs with i.+.0.c7 v ll:l. The fol lowing position is based on one of them.
+/
339
-
Original 338
+/Koranyi Suomen Shakki, 1955
(338): ECE gave a rather compli cated route to victory, but White can win quite simply. After 1 i.d7 Black can only oscillate with his king, provided White is careful not to al low Black's king to escape via f6 and e5. This leaves White to free to play his king round to support the pawn: 1 . . .'iitf8 ( 1 ...�g7 2 i.e6, followed by �f6) 2 'iit g4 (White must avoid the squares f4 and e3) �f7 3 'iitf3 �g6 4 'iite4 'iit f6 5 'iitd4 'iitf7 6 �e5 'iitg7 7 i.e6 ll:lc6+ 8 'iitd6 ll:la7 9 'iitc5 and wins. Black is not helped if he moves first: l . . .�f8 2 i.h3 'iit g8 (2 . . . 'iitf7 3 i.d7 as above) 3 'iitg4 'iitf7 4 'iitf3 and White wins in the same way.
(339): White wins even if Black moves first: l ...�gl (after l ...'iith2 2 'iit b2 'iitg 3 3 'iitc3 Black can play either 3 ...r.t>f4 4 'iitd4 ll:lf5+ 5 'iitc5 ll:le7 6 i.h3 'iite5 7 i.g4 with an easy win for White, or 3 ...'iitt"2 4 i.a6 'iite3 5 i.d3 ! transposing into the main line) 2 i.d3 ! (not 2 i.a6? 'iitt"2 3 'iitb2 r.t>e3 4 'iitc3 'iite4 ! 5 'iitb4 'iitd 5 6 i.b7+ r.t>e6 drawing; Black's king must be denied the square e4) 'iitt"2 3 'iitb 2! r.t>f3 (3 . . . ll:lc8 4 �c3 �e3 5 �c4 transposes) 4 'iitb3 ! (not 4 �c3? 'iite3 ! and White has fallen into re ciprocal zugzwang; the continu ation 5 i.fl 'iite4 ! 6 'iitb4 �d5 leads to a draw) 'iite3 5 'iitc3 ! ll:lc8 (5 ...'iitf4 6 �d4) 6 'iitc4 'iitf4 7 'iitc5 ll:le7 8 i.a6 'iite5 9 i.fl ll:lc8 10 i.h3 ll:le7 1 1 i.g4 r.t>f6 12 'iitd6 'iitf7 1 3 �d7 'iitf6 14 r.t>e8 and the pawn will pro mote.
238 i.+� v lLi
Now suppose that the pawn is on c6. The hook position (W�b7, i.b2, �c6 v Bq.,c5, l'Lib5) is again +/= ; the analysis is virtually identical to that in diagram 320. The following posi tion is analogous to diagram 32 1 . Al though the analysis is rather similar, we repeat it here because of the posi tion's fundamental importance:
340
4 ...�b6 5 i.:f2+ �b5 6 i.d4 are also winning) 5 i.g5 (a waiting move; Black must play his king to the gl-a7 diagonal, and then White wins with i.e3+ followed by i.d4) �c5 6 i.e3+
+I-
Chess Player's Chronicle, 1856 (340): Here it doesn't matter who moves first. White is aiming for one of two zugzwang positions, either i.d4 v q.,b5 or i.d4 v �d5. Direct methods of heading for these posi tions result in the wrong player being to move, for example 1 i.d4+ �d5 or 1 i.e5
+I=
341
Averbakh, 1955 (341): White to play wins by con tinuing 1 i.d6 (this position is recip rocal zugzwang) �f7 (after l . . .�f5 2 i.e5 ! White's king penetrates to e6 and d7) 2 d5 l'Lib6+ 5
i. +� v t'iJ 239
rJitfl 4 rJite5 ! 'it>g8 5 rJite6 cttg7 6 rJitd7 rJitf8 7 i. h4 ctt n 8 i.e7) 2 i.h4 (2 i.e5 rJite7 !) and now: 1) 2 'it>g6? 3 'it>e5 (this is also an important winning position) t'/Jc7 (3 ...'it>g7 4 i.d8 rJitfl 5 rJitd5 ! wins as in the White to play analysis) 4 i.d8 t'/Ja6 (or 4 ...t'/Jb5 5 rJitd5 ! ..t>fl 6 'it>c5 ! t'/Ja7 7 c7 ! rJite6 8 rJitb6! t'iJc8+ 9 'it>c6 t'iJd6 10 i. h4 t'/Jc8 1 1 rJitb7 "1i>d7 1 2 i. g 3 t'/Je7 1 3 i.f2 t'/Jc8 14 i. c 5 and White wins) 5 rJitd4 ! "1i>f7 6 rJitc4 ! '1ti>e6 7 '1ti>b5 ! rJitd6 8 i.e7+ (the key move) 'it>c7 9 i.h4 "1i>d6 10 i.g3+ 'it>d5 1 1 "1i>b6 winning. 2) 2 t'/Jc7+! 3 'it>d6 t'/Ja6 ! (now we have an important drawing posi tion) 4 i.el "1i>e8 5 i.a5 rJitfl 6 rJitd5 '1ti>e7 ! (6 . . . 'it>e8? gives White a free tempo and loses after 7 'it>c4 ! 'it>e7 8 '1ti>b5 'it>d6 9 '1ti>b6 ! 'it>d5 10 i.b4) 7 '1ti>c4 '1ti>d6 ! 8 'it>b5 t'iJb8 (8 . . . t'/Jc7+ also draws) 9 c7 t'iJc6 ! 10 'it>b6 'it'd? ! 1 1 il-c3 'it>c8 and Black is safe. If both kings are moved one square to the left then the position is drawn, whoever moves first. (342): This position is drawn: 1 i.f4 'it>e7 2 'it>b6 '1ti>e6 3 i.g5 'it>d6 4 'it>b7 (White is trying to squeeze round to d7 with his king, ending up with a win as in diagram 340) 'it>d5 5 i.e7 and now: 1) 5 'it>eS? (this allows White to execute his plan) 6 'it>c8 ! t'/Jg7 7 'it>d7 t'/Je6 8 i.a3 'it>f5 9 i.d6 "1i>f6 10 ..t>e8 (White must transfer the move to Black) t'/Jg7+ ( 10... '1ti>g5 1 1 'it>e7 'it>f5 12 i.b8) 1 1 'it>d8 'it>fl ( 1 l . ..t'iJe6+ 12 'it>d7 'it>f7 13 i.a3 is the same) 12 i.a3 t'/Je6+ ( 1 2 . . .t'/Je8 13 i.c5 t'/Jg7 14 'it'd? t'/Je6 15 i.d4 t'iJf8+ 1 6 'it>d6 •••
...
...
=I=
342
Averbakh, 1955 t'/Je6 17 i.c3) 1 3 'it>d7 ! t'/Jf4 14 cttd6 t'/Jg6 15 c7 t'/Je7 16 i.b4 t'/Jc8+ 1 7 rJitd7 t'iJb6+ 18 'it>d8 and wins. 2) 5 rJite6! (the simplest anti dote; White's bishop is forced away from its optimal square) 6 i.f8 'itd5 7 i.a3 'it>e6 8 'it>c8 '1ti>d5 and now that the bishop isn't on e7 Black can meet 'it'd? by ...t'/Jf6+. However, Black can Jose if his knight is badly placed. ...
240 .t+� v lb
(343): White to play wins by sim ply penetrating with his king to b7: 1 Wb5 Wd5 2 .tf4 �e4 (2 ...Wd4 3 'iti>a6 Wc5 4 �b7 is the same) 3 �a6 �d4 4 �b7 �d5 5 .tc l �c5 6 .ta3+ 'iti>b5 7 .tb4 and Black is in zugzwang. Averbakh erroneously claimed that the diagram is a draw with Black to move, but actually White wins in this case too: 1 . ..�e7 2 .th4+ ! We6 (or 2 . . . 'iti>e8 3 �d6 !, and now White wins after 3 . . .�f8 4 .td8 �e8 5 .ta5 �f7 6 �d7 �f6 7 Wc8 �e5 8 �b7 �d6 9 .tb4+ �d5 1 0 .ta3 or 3 . . . lbb6 4 c7 lbc4+ 5 'iti>e6! lbb6 6 .tf2 lbc8 7 .tc5 lbe7 8 'iti>d6 lbc8+ 9 �c6) 3 .td8 ! (not 3 �b5? lbc7+! and Black escapes) �e5 4 .te7 (Averbakh only consid ered 4 �b5? �d5, which is indeed drawn) �f5 (4 . . . 'iti>e6 5 .td6 puts Black in zugzwang, and after 5 ... Wf6 6 'iti>b5 �e6 7 .tg3 �d5 8 .tf4 White wins as in the White to play analysis; 4 . . .�e4 5 .td6 is similar) 5 �d6 (5 .td6 is inaccurate because 5 . . . �e6 6 .tg3 We7 forces White to start the winning process again from scratch) We4 6 .td8 'iti>d4 7 .ta5 (Black can not prevent White transferring his king to b7) 'iti>c4 (7 ... �e4 8 .tb4 Wd4 9 .tc5+ �c4 1 0 .tf2 wins by the same method) 8 .tel �d3 9 .tf2 We2 (9 . . .'iti>c4 1 0 �d7 �d5 1 1 .te3 We4 1 2 Wc8 followed by �b7) 1 0 .ta7 Wd3 1 1 'iti>d7 �c4 12 �c8 'iti>d5 1 3 Wb7 �d6 14 .tf2 �d5 1 5 .th4 �c5 1 6 .te7+ 'iti>b5 17 .tb4 and White wins. In the following group of posi tions, Black defends with his king to the rear of White's pawn.
+I=
344
Averbakh, 1955 (344): White can take advantage of the momentarily poor position of the black king to force zugzwang: 1 .tel �d4 ( l . . .lbf6 2 .tc3 lbd5 3 .tg7 lbb6 4 c7 is simpler) 2 .tg3 ! (White must transfer his bishop to the a l -h8 diagonal without allowing Black to reposition his knight by ... lbc3-b5; if he misses this opportu nity Black will play . . . Wc4 and the chance will have gone for ever, for example 2 .td2? �c4! 3 .th6 lbc3 ! draws) Wc4 (or 2 . . .'iti>e4 3 �c5 lbe7 4 c7 ! �f5 5 'iti>d6 lbc8+ 6 'iti>d7 lbb6+ 7 �d8) 3 .te5 ! (putting Black in zugzwang) lbb6 4 c7 lbc8+ 5 �d7 lbb6+ 6 �d8 �d5 7 .th2 and White wins. Black to play can hold the game by accurate defence: 1 .. .�d4 2 .tel 'iti>c4 3 .td2 lbf6 ! (not 3 ... �d4? 4 .th6 and White transfers his bishop to the long diagonal) 4 .tf4 lbe8+ ! (not 4 . . . �b5? 5 .te5 ! lbe8+ 6 �d7 ! �c5 winning as in diagram 340) 5 �d7 lbf6+ ! 6 �e6 lbd5 ! 7 .td6 lbb6 8 c7 �b5 ! drawing.
i.+l!. v lLi 241
If White's bishop starts on a7, then he wins with the move by 1 i.f2 lLic3 ( 1 . . .lLif6 2 �e6 liJd5 3 i.g3 �d4 4 �d6 ! wins as before) 2 �c5 ! liJd5 3 i.g3 and the pawn advances. On the other hand, Black to play can draw by continuing 1 . . .lLic3 ! 2 i. g l �f5 3 i.e3 (or 3 i.d4 tt:'ib5+! 4 'ittc5 lLic7 !) liJb5+ 4 �c5 lLic7 ! . Owners of Averbakh' s books should note that his recommended draw ing method (which we will not de scribe here) is bad, since it leads into diagram 34 1 , which is actually lost for Black.
(4 ...ltJd6 5 c7 �d4 6 i.e7 lLic8 7 i.f8 �d5 8 �b5 ! wins) 5 i.h4 liJd5+ 6 �c5 ! lLic7 7 i.d8 lLie8 8 �d5 ! �c3 9 �e6 �c4 10 �d7 �c5 wins as in diagram 340. Black to play draws by 1 ...�b4 ! . The following position i s based on one of the 120 reciprocal zug zwangs with i.+l!.c6 v ltJ.
346
+I=
345
Averbakh, 1955 (345): Averbakh gave a compli cated winning method starting with 1 i.f2, but the quickest and simplest winning continuation runs 1 i.a5+ �d3 ( l . . .'itiib3 2 �c5 �a4 3 i.b6 wins at once) 2 i.d8 ! (not 2 i.b6? l2Jc3+ !) l2Jc3+ (or 2 . . .�e3 3 �c4 liJd6+ 4 �c5 lLie8 5 �d5 ! �d3 6 �e6 �c4 7 �d7 with diagram 340 again) 3 �c5 lLie4+ (3 . . . l2Ja4+ 4 �b4 lLic3 transposes) 4 �b4 tt:'ic3
=I=
Moravec, 1950 Ceskoslovensky Sach
(346): This study is based on the fact that the position with �e5 v i.h3 is reciprocal zugzwang. Ac cordingly, White should not con tinue 1 �eS? i.h3 ! (now White has to make a concession) 2 �d6 (2 lLig3 f2) i.g2 ! 3 lLig3 f2 with an easy win. 1 tt:'ig3? f2 ! 2 lLifl+ �f4 3 �c4 i.c8 4 �d4 i.a6 5 liJd2 i.b5 is also bad, so White must move his king, while at the same time retaining the option of meeting . . . i.h3 by �e5. This leads us to the solution 1 �d6! (a stunning first move) i.c8 ( 1 ...i.h3 2 'ifile5 transposes to the main line) 2 'ifild5 ! and now:
242 i. +.0i v llJ
1) 2...i.a6 3 �e5 i.d3 4 'it>d5 (Ftacnik analysed this to a loss in ECE, but in fact it draws; he also correctly gave 4 'it>e6 i.fl 5 lt:Jg3 i.h3+ 6 �d5 as a draw) i.fl 5 lt:Jg3 ! i.h3 6 'it>d6 ! (Ftacnik missed this move, although the logic behind it is the same as for 1 �d6 !) f2 (6 ... 'it>f4 7 lt:Jhl ! i.g2 8 lt:Jf2 ! also draws) 7 'it>e5 ! with a second reciprocal zug zwang. 2) 2 i.h3 3 lt:Jg3 i.g2 4 �e5 ! f2 5 'it>f5 ! i. b7 (5 . . . .i.h3+ 6 'it>e5 ! is the reciprocal zugzwang mentioned above) 6 'iPg4 ! (Ftacnik gave 6 'iPe5 ? as an alternative draw, but then Black wins by 6 ... 'it>f3 7 lt:Jfl i.a6 ! ) i.c8+ 7 'it>h4 ! 'iPf3 8 lt:Jfl ! i.d7 (8 . . . i.a6 9 lt:Jh2+ ! 'it;g2 1 0 lt:Jg4 ! and 8 ... i.f5 9 lt:Jg3 ! i.d7 10 lt:Jfl ! are no better) 9 lt:Jd2+! 'it>e2 10 lt:Je4 ! and White de fends. The final position with the pawn on c6 not only involves a reciprocal zugzwang, it also makes use of sev eral ideas developed in this section. •.•
(347): White may try: 1) l i.f4? (1 i.d8? lt:Je3 !) lt:Jf6! 2 'iPc7 'iPe4 ! 3 i.g3 'iPd4 (3 . . . �f5 also draws, but we will take 3 ... 'it>d4 as the main line in order to prove that the resulting position is drawn with White to move; in fact it is a recip rocal zugzwang) 4 'iPd6 (other lines are no better, for example 4 i.h4 lt:Jd5+! 5 'it>d6 'iPc4 !, 4 i.h2 'it>d5 or 4 i.f4 'iPe4 ! 5 i.g5 lt:Jd5+ ! 6 'iPd6 lt:Jc3 ! 7 'it>c5 lt:Jd5) lt:Je4+! 5 'it>e7 (5 'it>d7 lt:Jc5+ ! 6 'it>c8 'it;d5 and 5 'it>e6 lt:Jc5 + ! also lead to a draw) lt:Jc3 ! and, as we know from diagram 344, the transfer of the knight to b5 guar antees the draw. 2) 1 i.h2? lt:Jf6! 2 'iPc7 'iPd5 and White cannot play i.h4 as in line 3. 3) 1 i.g3! lt:Jf6 2 'it>c7 ! (this posi tion is reciprocal zugzwang) and now: 3a) 2 M+ 3 'it>d7 ! lt:Jf6+ 4 'it>e6 lt:Je8 (4 . . . lt:Jd5 5 'iPd6! 'it>c4 6 i.e5 ! wins by diagram 344) 5 i.d6 'it>c4 6 i.e5 ! 'it>c5 7 'it>d7 ! winning as in dia gram 340. 3b) 2 �cS 3 i.d6+! �b5 4 h5! lt:Jd5+ 5 'it>d6 ! lt:Jb6 6 i.d4 ! (but not 6 c7? 'iPa6! 7 i.d4 lt:Jc8+! 8 'it>d7 'iPb7 ! 9 i.c5 'it>a8 ! drawing as in dia gram 337) lt:Ja8 7 'it;d7 (White wins as in diagram 343) �c4 8 i.f2 'it>b5 9 i.e3 �c4 10 'it>c8, followed by 'iPb7, and White wins. 3c) 2 'iPc4 3 i.e5 ! tbct5+ 4 �d6! wins by diagram 344. 3d) 2 �e4 3 �d6 ! lt:Jd5 (or 3 ... lt:Je8+ 4 'it>d7 lt:Jf6+ 5 �e6 lt:Jd5 6 i.e5) 4 �c5 and the pawn advances. 3e) 2 �d5 (now Black's knight has no access to d5) 3 i.h4 ! lt:Je4 4 •••
•••
•••
•••
+I=
347
Original
••.
i. +£!, v lll 243
i.e7 ! lllc3 5 i. b4 lllb5+ 6 'iii>b6 'iii>c4 7 i. el and Black cannot hold up the pawn any longer. Now suppose that White's pawn starts on c5 . In contrast to the situ ation with the pawn on b5 (diagram 327), the hook position is a draw whoever moves first.
• • • • .$; r:r� • •/,:; j].. /�/' '/@ • • •••
• . //�%1 . ;�
-
�� ".U/
"':!:@
. ;fd%
. • -
� �
�· �
···��
• • • •
• •.t• • . ·� . . /-". /,•%:: ,/,i> ;:{;,,p,, ij///::;
349
+I=
Original
=I=
348
Averbakh, 1955 (348) : 1 i.e4 'itd4 2 'iii>b5 llla2 (simpler than the alternative draw 2 . . . 'iii>c3 3 i. g6 'it>b3 4 i.h5 'iii>c3 ! 5 i. dl llld5 ! 6 c6 'it>d4 7 i.b3 llle7 ! 8 c7 lllc8 9 'iii>c6 llla7+! 10 'it>b6 lllc8+ ! ) 3 i.f3 (or 3 c6 lllc3+ ! 4 'it>a6 lllxe4 !) lllc3+ 4 'it>b4 llla2+ draw ing. Note how both drawing ideas depend on the a-file. There are 1 1 3 reciprocal zug zwangs with i. +,0,c5 v lll . Three of them appear in the following analy sis. (349): The logic behind this posi tion is that Black would like to meet the advance of White's king by . . . 'iii> f2-e3-d4. White to play wins by
1 i. a6 ! (this position is reciprocal zugzwang; other moves are inferior, for example I c6? llle7 !, 1 i. c4? llle7 ! 2 'iii>e4 'iii>f2 3 'iii>e5 'it>e3 4 'iii>d6 lllc8+, 1 i. d3 ? llle7 ! 2 i.e4 'it>g4 3 'iti>d4 'iti>f4, 1 'it>e4? 'iii>f2 2 i. b5 llle7 ! 3 'it>e5 'it>e3 ! 4 'iii>d6 'it>d4 and 1 i.f3? lllf8 2 c6 llle6 ! 3 i. d5 lllc7 ! 4 'iii>e4 'iii>f2 5 'iii>e5 'iii>e3 6 'it>d6 lllb5+ 7 'iti>c5 lllc7 !) and now Black has to make a concession: 1 ) 1 llle7 2 i. b7 ! transposes to line 4. 2) 1 'it>g2, which is positionally desirable because it stays in contact with f2, fails tactically to 2 .ib7+ and 3 c6, since d5 is covered after 3 ...llle7 4 c7. 3) 1 llle5 (now White gains a tempo after 'iii>e4) 2 i.b5 ! (confining the knight; 2 'iii>e4? fails to 2 . . . llld7) 'iii>g4 3 'iti>e4 ! lllf7 4 i.e2+ 'iii>g3 5 c6 ! llld6+ 6 'it>d5 llle8 7 'it>e6 'it>f4 8 i.c4 lllc7+ 9 'it>d6 ! llla8 1 0 i. a6! 'iii>e3 1 1 i.b7 ! lllb6 12 c7 'itd4 1 3 i. a6! with a reciprocal zugzwang from diagram 346. ...
...
...
244 .t+l!. v ltJ
4) 1. 'iti>g4 (now Black cannot play . . . 'iti>f2 and approach the pawn from behind) 2 rJi>e4 ! �gS 3 c6 ! 'iti>f6 4 c7 ! ltJe7 S .tc4 ! ltJc8 6 �d5 ! ltJe7+ 7 'iti>d6 ltJc8+ 8 �cS ltJe7 9 .tn rJi>e6 1 0 .th3+ ! 'iti>f7 1 1 'iti>d6 winning for White. S ) 1 liJf4 2 .tc4 ! liJg6 3 .tdS ! (a second reciprocal zugzwang) and now: Sa) 3 ltJe5 (giving White a free tempo with his king) 4 rJi>e4 ! liJd7 S c6 ! liJf6+ 6 'iti>eS liJe8 7 .tf7 and wins. Sb) 3 rJi>h4 (losing touch with f4) 4 rJi>d4 liJe7 S rJi>eS and the king advances. Sc) 3 rJi>g4 4 rJi>d4 liJe7 (4 ... rJi>f4 S c6 !) S 'iti>eS ! 'iti>gS 6 .tb7 is similar. Sd) 3 ltJe7 4 .tb7 ! (a third re ciprocal zugzwang) liJfS+ (4 . . . 'iti>g4 S 'iti>e4 ! rJi>gS 6 'iti>eS ! ) S rJi>e4 ! ltJg7 6 .t c8 (threatening .td7; not 6 c6? liJe8 7 'iti>eS 'iti>f2 8 rJi>e6 rJi>e3 ! 9 rJi>d7 liJf6+ ! 10 'iti>d8 liJdS ! 1 1 .ta6 'iti>d4 and Black draws) liJe8 7 �dS ! ltJc7+ (7 . . . 'iti>f4 8 .td7 ! liJf6+ 9 rJi>e6 ! ltJe4 10 c6 ! liJcS+ 1 1 rJi>dS liJa6 12 'iti>d6!) 8 rJi>c6 ! liJe8 9 �d7 liJf6+ 10 'iti>e6 ! liJe8 1 1 .td7 liJc7+ 12 'iti>d6! liJa6 13 c6! 'iti>f4 14 .th3 winning. Black to plays draws by l .. .liJe7. •.
•••
...
Pogosiants, 1979 2nd Pr., '64 '
•••
•••
=I=
3SO
liJh2 ! .tf3 s ttJn ! .th 1 6 liJh2 ! .tf3 7 liJfl .tg4 8 liJg3+ ! fxg3 9 rJi>g2 and Black draws.
..•
Now suppose that the pawn starts on c4. The following study is simple but attractive: (350): Despite the miserable po sition of White's knight, he manages to draw by accurate defence. The play is self-explanatory: l . . ..t h l 2 rJi>gl ! rJi>e2 (a good winning try; 3 'iti>xh l ? loses to 3 . . . rJi>f2) 3 liJfl ! f4 4
3S l
I
= =
Beliavsky-Liberwn Baden 1980 (351): This ending was analysed by Beliavsky in lnformator and by Cebalo in ECE. Both analysts con cluded that the position is winning for White, although in fact it is a fairly comfortable draw. White was
i.+� v ltJ 245
to play in the game, which continued l 'iti>e4 and now: l ) 1 ltJg2 (the simplest move, although 1 . . .lLid l is also adequate for a draw) 2 i.d7+ b4 ! 3 f5 c5 ! (ECE only analysed 3 ...ltJh4 ?, which loses after 4 f6 ! ltJg6 5 f7 'itc5 6 i.f5 ltJf8 7 e5 ! <,i;>c6 8 cli>f6 cli>d6 9 �g7 �e7 10 �g8) 4 <,i;>e5 ltJh4 5 f6 ltJg6+ 6 e6 lLif8+ ! 7 9ile7 ltJg6+ 8 e8 ltJe5 drawing. 2) 1 ltJn (the game continu ation is also adequate for a draw, but it requires accurate play by Black) 2 i.g6 with the branch: 2a) 2 9ilb4? (as played in the game) 3 �d4 ! (Beliavsky takes his chance; this moves both threatens f5 and keeps the enemy king at arm's length) ltJg3 (or 3 ...lLid2 4 i.d3 lLib3+ 5 'iti>d5 ! ltJc5 6 i.bl lLib7 7 i.g6, sealing off the black knight and winning after 7 ...ltJc5 8 i.e8 lLid3 9 f5 ! ltJf4+ 10 d6 ltJh3 1 1 f6 ltJg5 1 2 i.g6 'iti>c4 1 3 �e5 ! lLif3+ 1 4 f4) 4 'iti>e3 ! lLi f l+ 5 9ite2 ! lLih2 6 i.h5 <,i;>c5 7 'itf2 ! 'iti>d5 8 i.e2 ! 'iti>d4 9 �g2 'ite3 (9 . . . 'iti>e4 10 'iti>g3 ! is a re ciprocal zugzwang) 1 0 f5 ! xe2 1 1 9ilxh2 ! 1-0. 2b) 2 b5 (a fairly obvious im provement; Black again heads for the pawn, but at the same time pre vents White from using his king to erect a barrier) 3 'iii>d3 (3 'itd4 'itc6 is an easy draw) �c5 4 'ite2 lLih2! and now: 2bl ) 5 i.f5 �d4 ! (not 5 ... 'iti>d5? 6 �f2 ! 'iti>d4 7 i.h3 ! ) 6 f2 lLifl ! and the knight escapes. 2b2) 5 i.h5 d5 ! (curiously, this time the roles are reversed: 5 ... �d4? ...
loses to 6 i.f3 ! 'iti>c5 7 f2 ! 'iti>d6 8 i.e2 'ite6 9 �g3) 6 i.f3+ (6 f2 'iti>e4 !) ltJxf3 7 xf3 'iti>e6! drawing. There are 85 reciprocal zug zwangs with i.+�c4 v ltJ. A number of these arise in the analysis of the following position.
...
...
•••
+I=
352
Original (352): White to play wins by 1 i.h4+ ! 'iti>g6 ( 1 . . . �f7 2 i.d8 ! and 1 . .. 'iti>f5 2 i.d8 ! transpose to later analysis, while 1 . . . g7 2 i.d8 ! 'iti>f7 3 c5 ! ltJg7 4 c6 ! ltJe6 5 i.b6 e7 6 i.c5+ ! 'iti>f6 7 d6 �f7 8 i.e3 �f6 9 i.d2 �f7 10 i.c3 is also winning for White) and now: 1 ) 2 i.d8? ltJg7 (now White falls into a reciprocal zugzwang himself; the transposition 2 . . .'iti>f5 3 c5 ltJg7 ! is also possible) 3 c5 f5 (another reciprocal zugzwang) 4 c6 lLie8 ! (the third reciprocal zugzwang) 5 i.e7 ltJc7+! 6 d6 ltJa6 and Black draws by line 2 of diagram 34 1 . 2) 2 i.e7! (now Black is on the wrong end of a reciprocal zug zwang) with various possibilities:
246 .t+£i.
v
ttJ
2a) 2 ttJg7 3 c5 tiJf5 4 .td8 ! tiJe3+ 5 �e5 �f7 6 c6 ! �e8 7 .tf6 ! ttJc4+ 8 �e6 ! tiJb6 9 c7 ttJc8 10 .td4 ltJe7 1 1 .te3 ttJc8 1 2 .tc5 ltJe7 1 3 �d6 winning. 2b) 2 �f7 3 .td8 ! tiJg7 4 c5 ! �e8 5 .tf6 ! tiJf5 6 c6 tiJe3+ 7 �e6 ltJc4 8 c7 tiJb6 9 .td4 tiJc8 10 .tc5 wins as in line 2a. 2c) 2 �fS 3 .td8 ! (another re ciprocal zugzwang) �f4 4 c5 ! �e3 5 c6 �d3 6 �e6 �c4 7 �d7 �c5 8 .th4 winning as in diagram 340. 2d) 2 �g7 3 �e6 �g6 4 .td6 tiJg7+ 5 �e5 ! �f7 6 c5 ! �e8 (or 6 . . . ttJe6 7 c6! tiJd8 8 c7 ! tiJc6+ 9 �d5 tiJa7 10 .tc5 tiJc8 1 1 �c6) 7 �f6 ! tiJh5+ 8 �g6 and the knight is trapped. 2e) 2 ttJc7+ 3 �d6! tiJe8+ 4 �d7 ! tiJg7 5 c5 �f5 6 c6! tiJe6 7 .td6 �f6 (White wins by transfer ring the move to Black, as in line 1 of diagram 342) 8 �e8 tiJg7+ 9 �d8 tiJe6+ 10 �d7 �f7 1 1 .tb4 winning. ...
...
••.
+I=
353
Original
.•.
..•
There are 4 1 positions of recipro cal zugzwang with .t+£i.c3 v ttJ and 44 with .t+£i.c2 v ttJ. The following very attractive example is based on one of the latter. (353) : This study is based on re ciprocal zugzwang and domination of the knight; the pawn on c2 plays a vital role in the latter. White can play: 1) 1 �d6? tiJg6 ! (now White has fallen into the reciprocal zugzwang) 2 c4 (as we shall see, 2 .tf3+ �a7 ! doesn't significantly alter the situ ation) tiJf4 ! 3 c5 (after 3 .tf3+ �a7 ! 4 c5 tiJd3 ! 5 c6 tiJb2 ! Black's knight
again reaches b6) tiJd3 ! 4 c6 tiJb2 5 c7 (5 .te2 ltJa4 ! 6 c7 tiJb6! is basi cally the same) ltJc4+! 6 �c6 tiJb6 ! and Black draws by diagram 337. 2) 1 �d7! tiJg6 (after 1 . ..tiJgS White traps the knight by 2 �e6 �b7 3 .th3 ! tiJh6 4 �f6 tiJg8+ 5 �f7 tiJh6+ 6 �g7) 2 �d6! (this time Black is the one in reciprocal zug zwang) and now most knight moves result in the steed being trapped. while playing the king to the b-file gives White an extra tempo when he pushes the c-pawn: 2a) 2 �b8 (2 . . . �b7 3 c4 ! is the same) 3 c4 ! �a7 (Black has no time for his knight tour to b6 because the pawn will give check when it arrives on c7) 4 c5 ! �a6 5 c6 �b6 6 c7 la!7 7 �d7 winning. 2b) 2 tiJf4 3 �e5 ! tiJg6+ 4 �f6 tiJf4 5 �g5 wins the knight. 2c) 2 ttJh4 loses to 3 c4 ! be cause the knight is so far away that White can simply push the pawn. 2d) 2 �a7 3 .tf3 ! (a second re ciprocal zugzwang) tiJf4 (moving •
.••
••.
...
••.
�+� v .!LJ 247
the king to the b-file loses as before, while 3 . . .�a6 runs into the tactical point 4 c4 ! tllf4 5 c5 ! tlld3 6 �e2) 4 �e5 ! tllh3 (after 4 . . . tllg6+ 5 �f6 .!Df4 6 �f5 tllh 3 7 �g4 tllg l 8 �dl ! White rounds up the knight with his king) 5 �f5 ! tllg l (or 5 ....!Df2 6 �f4 .!Dh3+ 7 �g4) 6 �d l (6 �g4 also wins, albeit somewhat more slowly) tllh3 7 �g4 .!Df2 8 �f3 ! �b6 9 �f4 ! tllh 3+ 10 �g4 ! tllg l 1 1 �d l ! and now Black can kiss goodbye to his knight.
tllc2 followed by ... .!Dd4, drawing) and now: l ) 2 � 3 �f4 ! with another branch: l a) 3 tlle l+ 4 �d2! .!Df3+ (or 4 . . . �f2 5 c4 ! .!Df3+ 6 �c3 transpos ing to line l b) 5 �e3 .!Del (5 ...�g2 6 c4 .!Del 7 �d2 .!Df3+ 8 �d3) 6 �d6 (not 6 c4? tllc2+ ! 7 �e4 �e2 8 c5 tllb4 !) tllc2+ (6 ... �g2 7 c4) 7 �d2 .!Del 8 c4 .!Df3+ 9 �d3 .!Del+ 10 �c3 and now the pawn can advance safely. lb) 3 �f2 4 c4 .!De l + 5 �d2 ! tllf3+ 6 �c3 .!LJM 7 c5 tllf5 (7 ...�f3 8 c6 tllf5 9 �g5 ! transposes) 8 �g5 ! �f3 9 c6 �g4 (9 ... .!Dd6 10 �d4 ! tlle 8 1 1 �e7 .!Dc7 12 �d6 tlle 8 1 3 �d5 �e3 1 4 �c5+ �d3 1 5 �b6 �c3 16 �e6 �c4 17 �d8 ! �c5 1 8 �d7 ! is a familiar win) 10 �f6! .!Dd6 1 1 �d4 ! tllb5+ 12 �c5 .!Dc7 1 3 �d8 tlle8 14 �c4 ! (a nice winning move; White gains the opposition) �f4 15 �d4 ! �f5 1 6 �d5 ! and the king penetrates to e6 and d7. 2) 2 �el (the toughest defence) 3 �f4 ! �d l (3 . . . �f2 4 c4 .!Del+ 5 �d2 is line l b) 4 c4 ! .!De l+ 5 �d4 ! tllc2+ 6 �c3 ! .!Da3 7 c5 ! tllb5+ 8 �c4 ! tlla7 9 �d5 ! �e2 (a better de fence than 9 . . . �c2 10 � b8 ! .!Dc8 1 1 �c6! tlle7+ 1 2 �d6! tllf5+ 1 3 �d7 tlld4 14 �e5 ! tllb5 15 �c6 ! .!Da3 16 �d5 ! tllb5 17 c6 ! ) 1 0 � b8 ! tllc 8 1 1 �c6! tlle7+ 12 �d6! tllf5+ 1 3 �d7 ! .!Dd4 14 �e5 ! tllc2 (Black did not have this defence in the previous note; after 14 . . . tllb5 15 �c6! White wins just as after 9 ... �c2) 15 �c3 ! (not 1 5 c6? tllb4 ! 16 c7 tlld5 !) .!Da3 16 �b2 ! ( 1 6 �b4? tllb5 ! 17 �c6 ...
...
...
We end with the longest win in the ending of �+c� v tD (35 moves):
...
+I=
354
Original (354): The winning line runs 1 �d3 ! (not 1 �d5? tlle2 2 c4 .!llc 3+! 3 �d4 tlla4) .!Df3 (White wins much more easily after l .. .�g3 2 c4 �g4 3 �e3 tllf3 4 �e4 ! tllh4 5 c5 tllf5 6 c6 ! tlld6+ 7 �e5 tlle8 8 �b6 �f3 9 �e6 �e4 1 0 �d8 �d4 1 1 �d7 �c5 12 �h4) 2 �e3+ ! (and not 2 �f4? .!De 1 + ! 3 �d4 tllc2+ 4 �d5 �e2 5 c4 �d3 6 c5 tllb4+ 7 �d6 �c4 8 �d2
248 .t+A v tLJ
tLld4+! 1 8 'iPd5 'iPd3 draws) tLlc2 ( 1 6 . . . tLlb5 17 'iPc6! and 1 6. . .tLlc4 17 c6 are easy wins) 1 7 'iPc6! (threat 'iPb5) tLlb4+ (in the line 17 ... 'iPd3 18 'iPb5 ! tLle3 1 9 c6 ! tLJd5 20 'iPc5 ! tLlc7 2 1 .te5 ! tLle8 22 'iPd5 ! White needed to find 22 consecutive 'only' moves in order to win; the continuation might be 22...'iPe3 23 .td4+ 'iPd3 24 .tb6 'iPc3 25 'iPe6 'iPc4 26 .td8 !) 1 8 'iPb5 ! tLJd5 1 9 c6 'iPe3 ( 1 9. . .tLlc7+ 20 'iPc5 tLle8 2 1 .te5 transposes) 20 'iPc5 ! tLlc7 2 1 .te5 ! tLle8 22 �d5 ! 'iPd3 23 .id6 'iPc3 24 .te7 tLlc7+ 25 'iPc5 ! �d3 26 .td8 tLle8 27 �d5 ! 'iPc3 28 'iPe6 'iPc4 29 'iPd7 �c5 30 .th4 'iPb5 31 .tg5 'iPc5 32 .te3+ 'iPd5 33 .id4 tLld6 34 c7 ! with pro motion or capture of the bishop next move.
4.4:
.tb4+! (note that, unlike the analo gous c-pawn position, 1 .tf6? fails to 1 . ..tLla7+ !) �e6 2 �c7 �d5 3 .ta3. Black to play draws by either 1 .. .tLla7+ or 1 . ..tLle7+. If we shift diagram 337 one file to the right (W�e8, .te2, Ad7 v B'iPc7, tL!f7) then the stalemate defence disap pears and White to play wins after I �e7 tLld8 2 .tf3 tL!f7 3 .tg2 tLld8 4 .id5. Black to play draws by giving immediate perpetual check.
i.+d� v llJ
There are many similarities between this section and analogous positions with a c-pawn, so our coverage need not be so detailed. Moving the pawn to the d-file increases the defender's chances still further. One reason for this is that the knight has more room to manoeuvre while it is fighting against the pawn; the second is that with the pawn in the centre of the board, it is impossible for the de fender's king to be more than four files away from the pawn. As usual, we start with the pawn on d7. The hook position (W�c8, .tc3, .0.d7 v B'iPd6, tLlc6) again has the evaluation +/=, just as with a b or c-pawn. White to play wins by 1
355
+I=
Galushko, 1990 =2nd HM, Bimov Mem. Tny. (355): Black to play draws by 1 . . .�b7 2 .td6 tLle8 !, followed by ...�c6. White to move wins by using zugzwang to penetrate with his king to either c6 or e6: 1 .i b4 ! (not 1 .id6? tLle8 ! 2 .te5 �b6) 'iPa6 (after 1 ...�b7 2 .ta5 ! tLle8 3 �b5 ! White's king advances more quickly) 2 .tel (Galushko gave a more complicated win beginning with 2 .ta5) tLle8 (2 . . . �b7 3 .ta5 as before) 3 .ia5 ! tLld6 4 �b4 ! tLle8 (4 . . . �b7 5 'iPc5! tLle8 6 �d5 tLlg7 7 i.c3, followed by
i.+�
c;t>e6) 5 i.d8 ! �b7 (5 . . . lLig7 6 c;t>c5 c;t>b7 7 i.a5 wins after 7 ... lLie8 8 c;t>d5 or 7 . . . �c8 8 �c6 !) 6 �c5 'it>c8 7 �c6! lLic7 8 �d6 ! (White's first task is to extract his bishop from d8) lLie8+ 9 c;t>e6! lLig7+ 10 c;t>f6 ! lLie8+ 1 1 'it>f7 ! c;t>d7 1 2 i.a5 (White wins easily once his bishop becomes ac tive) liJd6+ 1 3 c;t>f8 ! lLie8 1 4 i.c3 liJd6 15 i.d4 lLie8 16 i.e5 followed by the knight's capture.
356
v
lLJ 249
(the composer only gave 5 . . . i.c2, but then White can draw by 6 lLid5+ ! �f2 7 lLic3 ! and Black cannot make progress) 6 �b4 'it>d4 7 �b3 'ifi>d3! 8 'iti>b2 i.c6 9 lLidl i.a4 10 liJf2+ c;t>e3 1 1 c;t>c3 c;t>e2 and wins. There are 77 positions ofrecipro cal zugzwang with i.+�d7 v ltJ, but we will not dwell on any of these. Now suppose that the pawn starts on d6. The hook position (Wc;t>c7, i.c2, �d6 v B'it>d5, lLic5) is drawn (with a c-pawn it was +/=) because Black has an extra knight check on the a-file, for example 1 i.b3+ c;t>e5 2 i.c4 (2 �c6 lLixb3 3 d7 liJd4+ !) c;t>d4 ! 3 i.f7 lLia6+ 4 c;t>b6 lLib8 and White cannot make headway. When diagram 340 is shifted to the right the result also changes, but in this case the reason is more subtle.
-/+
Voja, 1957 Iijdschrift v. d. KNSB (356): This was intended to be a White to play and draw study, but actually Black wins. After 1 lLid5 Black may continue: 1) 1 'ifi>e2 (the simplest win) 2 liJc3+ c;t>d3 ! 3 liJd 1 i.h5 4 liJf2+ c;t>d4 ! (4 ...c;t>e2? 5 lLie4 !) 5 'ifi>g3 (5 c;t>f5 c;t>e3) �e3 6 c;t>g2 i.e2 7 c;t>g3 i.f3 with zugzwang. 2) 1 'itn (the composer's move also wins, albeit more slowly) 2 lLic3 (2 lLie3 i.h5) i.c2 ! 3 �e5 'it>e3 ! 4 'it>d5 i.e4+ (certainly not 4 . . . 'it>d3? 5 lLie4 ! ) 5 c;t>c4 (5 c;t>e5 i.f3) i.f3 ! •••
MandelOl, 1938
...
(357): White is aiming for one of two possible zugzwang positions: i.e4 v 'it>c5 or i.e4 v �e5. He could force one of these in the analogous
250 .i.+� v ll)
c-pawn position, but here it is impos sible for the unexpected reason that the h5-e8 diagonal is too short. The critical position arises after 1 �c2 rJi>c6 ! (not 1 . ..rJi>e5? 2 �e4 !) 2 �a4+ rJi>c5 ! (2 ... rJi>d5? 3 �e8! leaves Black in reciprocal zugzwang) 3 �e8 (3 �d7 lLlg6+ ! is a simple draw) rJi>d5 ! . White would like to reach a position with �h5 v rJi>c5, and Black to play; then Black would have to move his king onto the long diagonal, where upon �f3+ and �e4 would win. However, if Black is careful he can avoid this fate. He has to meet �f7 by . . .rJi>c6 (not ... rJi>c5, which would be met by �h5) and �e8 by ...'iitd5 . In other words there is a correspon dence between Black's king and White's bishop; e8-d5, f7-c6 and h5-c5 form the three pairs of corre sponding squares (all three posi tions are reciprocal zugzwang). The square g6 is out of bounds, and the three remaining squares on the h5-e8 diagonal are not enough for the bishop to break the correspondence by losing a tempo, for example after 4 �f7+ (4 �h5 rJi>c5 !) rJi>c6! 5 �h5 �c5 ! White cannot make progress. (358): White to play wins by 1 �b3 ! rJi>f5 2 �f7 rJi>g5 (2 . . . �e5 3 �e6 wins immediately) 3 �e6 rJi>g6 4 rJi>f8 (now White tries to lose a move) lLlh7+ 5 rJi>e8 lLlf6+ 6 �e7 �g7 (or 6 ...rJi>g5 7 �f7 lLle4 8 d7 ! lLld6+ 9 rJi>e7 lLlb7 1 0 �d5) 7 �c4 lLlg4 (7 . . . rJi>g6 8 �d3+ wins after 8 ... rJi>g7 9 rJi>e6 or 8 . . . rJi>g5 9 �c2) 8 �e6 lLlh6 9 d7 lLlf7 1 0 �b3 rJi>f8 1 1 �d5 lLld8+ 12 rJi>d6 rJi>g7 1 3 rJi>e7 trapping the knight.
- - - . - • = . . � � - . - . - . . - - . - m.tm m m - . - +I=
358
Averbakh, 1955 White also wins if Black's king starts on d5 : 1 �b3+! rJi>c5 2 �a2 'iitc6 (2 ... lLlg4 3 rJi>e6 lLlf6 4 �bi 'iitc6 5 �f5 'iitc5 6 �e4 leads to the same zugzwang) 3 rJi>e6 lLlh7 4 �d5+ 'iitc5 5 �e4 lLlf6 6 �bl �c6 7 �f5 �c5 8 �e4 and Black is in a fa tal zugzwang. The following position, while ap parently very similar, is rather more complex.
+I=
359
Original
.t+�
(359): White to play wins by 1 �e6, so we may assume that Black moves first. The winning line runs 1 .. .tl:ld5+ 2 �d8 ! tl:lf6 (or 2 . . .tl:lf4 3 �e8 tl:ld5 4 .ib3 tl:lf6+ 5 �e7 ! �f5 6 .ia2 winning as in the previous diagram) 3 .id3 ! (not 3 .t b 1 ? �f4 ! 4 �e7 tl:ld5+! 5 �d7 tl:lb6+ ! and Black draws after 6 �c7 tl:ld5+ ! 7 �b7 tl:lf6 ! or 6 �c6 tl:lc4! ; the latter line shows why White must cover c4) tl:lg4 (3 . . . �f4 4 �e7 ! tl:ld5+ 5 �d7! �e5 6 .ic4! tl:lf6+ 7 �e7 ! �f5 8 .ib3 again transposes to the pre vious diagram) 4 �e7 ! tl:le5 5 .tn ! (5 .ie2? �f5 ! leads to a draw since White has no check on the h3-c8 diagonal) tl:lg6+ (5 . . . tl:lc6+ 6 'it>d7 ! tl:le5+ 7 �e6! transposes to the main line; 5 . . . �f5 6 .ih3+! �f4 7 �f6 ! �e4 8 �e6 ! �f4 9 �d5 tl:lf7 10 d7 ! wins for White, but not 7 'it>e6? in this line because of 7 ... tl:ld3 !) 6 �f7 ! tl:le5+ 7 'it>e6! tl:lg6 8 .ib5 ! tl:lf4+ (White also wins after 8 . . . tl:lf8+ 9 �e7 tl:lg6+ 10 �f7 tl:le5+ 1 1 �e6! tl:lg6 12 .ic6 tl:lf4+ 13 �d7 tl:ld3 14 .ie4 tl:le5+ 15 �e6! �f4 16 �c2) 9 �f7 ! tl:lg6 10 .ia4 tl:le5+ 1 1 'it>e7 tl:lg6+ 12 �e6 tt:if4+ 1 3 'it>d7 'it>f6 14 �c8! (14 �c7? tl:ld5+ 1 5 �c6 tl:lb4+ 1 6 �c5 tl:ld3+) tl:ld5 1 5 �b3 ! �e5 16 �d7! tl:lf6+ 17 �e7! �f5 1 8 .ia2 and wins as in the previous diagram. The following study has a finish which is based on a reciprocal zug zwang. (360): The main line runs 1 �e7 ! �g7 2 .ie4! (White must be careful not to allow the black king too much freedom; 2 .id5? fails to 2 . . . �g6! 3 .ib7 �f5 4 �d8 �e6 5 �c7 �e5 ! 6
v
tLl 251
+I= Missiaen, 1955 =2nd Pr., Rinck Mem. Tny.
360
.ic8 �d5 !) �g8 3 .id5+ �g7 4 .ic4 ! (by covering the escape route via a6, White threatens �d8-c7) and now: 1) 4 lDc6+ 5 �d7 tl:le5+ 6 �e6 ! tl:lg6 (6 . . . tl:lf7 7 d7 ! �f8 8 .id5 tl:ld8+ 9 �d6) 7 .id3 ! tl:lf8+ 8 �e7 ! 'it>g8 9 .if5 �h8 (9 . . �g7 10 .th7) 10 �f7 and White wins. 2) 4 �g6 5 �d8 ! �f5 6 �c8 ! (but not 6 �c7? �e5! and White is in reciprocal zugzwang) �e5 7 �c7 ! and White's triangulation has left Black movebound. There are 1 6 1 reciprocal zug zwangs with .t+�d6 v tl:l. .••
.
...
With the pawn on d5 or further back, White's winning chances are few indeed. Positions which are won tend to fall into two categories; either White can advance the pawn more or less immediately, or White can play to trap the enemy knight. The first type rapidly transposes into the earlier sections while the second
252 .t+� v lD
can only occur in exceptional situ ations which are not very relevant for the over-the-board player. Therefore, we will finish this section with a se lection of entertaining positions rather than a systematic analysis. In the next diagram, White's army co-operates to surround and trap the knight.
�· . . . • • • • • = • • • • • .i.. •• • • • • . fj . • • • • • • • • •
2b) 4 .!Db6 (the composers be lieved that this move loses, but actu ally it is good enough for a draw) 5 .ib5 lDa8 ! (the composers only gave 5 ... .!Dc8? 6 d5 ! lDb6 7 d6! lDcs 8 d7 ! lDb6 9 d8l:t ! , when White does win) 6 d5 .!Dc7 ! 7 d6 .!De6+ ! 8 �d5 lDds with an easy draw. 3) 1 .ie8+! b7 ! .!Da4 6 .ixa4 ! �xa4 7
I"
+I=
361
Kondratiev and Kopnin, 1985 3rd Comm., teskoslovensky Sach (361): White can play: 1) 1 .t f3? lDb6 ! 2 b5 ! 3
.tc6+ c5 and now: 2a) 4 .!De7 5 .ib5 and now the composers gave the line 5 ... lDf5? 6 d5 ! lDg3 7 .td3 ! lDh5 8 d6 ! lDf6 9 a6 1 0 e6
+I=
362
Original (362): Black to play draws by means of 1 . ..lDd2+ ! and 2 . . . lDb3, when the knight escapes, so assume that White moves first. Provided that White's king is sufficiently close, Black can't move his knight, but on the other hand the bishop is immo bilised by the need to guard a3 and c3. Therefore both sides are more or less limited to king moves, and the position can be �malysed in terms of
Jl + .0.
corresponding squares. White's aim is to reach �c2 v �e3 with Black to move. Then Black would lose his knight after either l . . . liJd2 2 Jlc l ! or I . . . 'it>e2 2 Jlc l ! . On the other hand, White to play would be unable to win since his only reasonable move, 1 Jlc l +, is met by l .. .�d4 ! with another reciprocal zugzwang; in this case White could only repeat moves. Given this, it is easy to see why 'it>d l v 'it>f3 is reciprocal zug zwang. With Black to play l ...�e3 2 �c2 !, 1 . . .'it>f4 2 �c l ! �e3 3 �c2! and l .. .�f2 2 d4 ! (this variation is noteworthy, because an extra ele ment comes into play - Black's king mustn't move too far up the board or else White can just push the pawn) �e3 3 d5 ! liJd2 4 d6! all win for White, but with White to play 1 �c2 �e3 ! and 1 'it>c l �e2 ! 2 �c2 �e3 ! hold the draw. Thus we have the two pairs of corresponding squares c2-e3 and d l -f3. Working backwards, one can easily find the other four pairs e l -g4, e2-f4, e3-f5 and f2-g5. Be yond this the pattern breaks down, because if White's king is further away, the black knight can escape via d2. Using this information, the following solution becomes self explanatory: I 'it>f2 ! (not 1 'it>e2? 'it>f4 ! 2 'it>f2 'it>g5 ! 3 'it>el �g4 ! 4 �d 1 �f3 ! 5 �c l 'it>e2 ! 6 d4 liJd2 ! drawing) 'it>g4 (or l . . . 'it>f5 2 �e3 ! 'it>e6 3 'it>e2 ! 'it>f5 4 'it>d 1 ! 'it>f4 5 'it>c l ! 'it>e3 6 'it>c2 ! 'it>e2 7 Jlc l !) 2 �el ! 'it>f4 (2 ...'it>f3 3 �d i ! 'it>f2 4 d4 !) 3 'it>e2 ! 'it>f5 4 'it>d 1 ! �f4 5 'it>c 1 ! �e3 6 'it>c2 ! 'it>e2 7 Jlc l ! and White wins the knight.
v
ei 253
+I=
363
Original (363): The first move is the sur prising 1 'it>a3 ! , but the reason for this move has nothing to do with the opposition. The point is that one of Black's defensive ideas is to play . . . �e2, and after d4 simply to take the bishop. Then the question is whether or not the knight can stop the pawn. After 1 'it>b5?, for exam ple, the knight can make it back by 1 . . .�e2! 2 d4 'it>xel ! 3 d5 lLlf2 4 d6 lL!e4 ! 5 d7 liJd6+ ! 6 �c6 lL!f7 ! . It is quite easy to confirm that if White's king is on a5 , b5, a4, b4, b3, a2 or b2 then Black's knight is in time, whereas on a3 the pawn promotes. So I �a3 ! is simply the only way to defend against Black's threat of . . .'it>e2. It also follows that if White were to play in the position after I �a3 !, then the result would be a draw because White would have to play his king onto one of the forbid den squares. Thus I �a3 ! results in a position of reciprocal zugzwang, and White wins after I ... 'it>e4 (or l ...�f4 2 'it>b4 �e4 3 �c4 �e5 4 d4+ �d6
254 .t +13:. v lt)
5 d5 wins easily; once White gets his pawn moving Black has no chance) 2 �b2 ! (not 2 �b3? �d3! 3 'i1i>b2 li)f2 and the knight escapes) 'i1i>d3 3 �c l ! (defending the pawn just in time to prevent the horse bolting) �e2 (3 . . . 'itd4 4 �c2) 4 .ih4 ! �d3 (4 ... �f3 5 �c2 �g4 6 .iel !) 5 'iii>d l ! �e4 6 �c2 �d4 7 d3 �e5 8 �c3 �d5 9 d4 winning easily.
+I=
364
Original (364): After the obvious 1 .id4 ! , we once again have a situation domi nated by corresponding squares. White is aiming for a position with 'itc2 v 'itel and Black to move. He would be forced to play 1 .. .�e2, but after 2 d3! �el 3 .igl ! �e2 4 d4 Black's knight remains imprisoned and the d-pawn races away. On the other hand, White to play cannot win because after 1 d3 (1 �c 1 �e2 ! and 1 .igl li)b2 ! are no help, while after 1 .if6 �e2 ! White has to return to d4 in order to prevent 2 ...li)f2 and 2 . . . li)e3+) �e2 2 .i gl Black has the defence . . . li)b2. It follows that
'i1i>d3 v 'iti>fl is also reciprocal zug zwang; Black to plays loses after l ...'iti>el 2 'itc2 !, while with White to move 1 'itc2 �el ! just leads to the earlier reciprocal zugzwang. Hav ing determined the two pairs of cor responding squares c2-e 1 and d3-fl. one can easily work backwards to determine the other two, e4-g2 and c4-g2. Now the solution should make sense: 1 .id4! 'i1i>h2 (l ...�g2 2 �c4 !) 2 �b4 ! (White must avoid c4 until Black occupies g2; 2 �b5? is also bad because of 2 ...� g3 ! 3 �b4 'i1i>f3 ! 4 �b3 �e2 ! 5 'iti>c2 �el !) �g2 (Black must approach in order to meet the threat of � b3-c2, and 2 . . . 'it g3 loses to 3 'itb3 � f3 4 �c2 �e2 5 d3!) 3 �c4 ! �fl (3 ...�g3 4 'i1i>d3 �f3 5 'itc2! 'i1i>e2 6 d3!) 4 �d3 ! �el 5 �c2 ! 'ite2 6 d3! �el 7 .igl 'ite2 8 d4 ! and the pawn promotes. The reciprocal zugzwang statis tics for the remaining d-pawn posi tion are as follows: there are 143 reciprocal zugzwangs with .i+l3:.d5 v li), 89 with .i+l3:.d4 v li), 63 with .i+l3:.d3 v li) and 80 with .i+l3:.d2 v lt)_ We end with the longest win in the ending of .i+dl3:. v li), and the joint longest (see also diagram 3 1 6) in the whole ending of .i+l3:. v li) (42 moves). (365): In contrast to most of these 'longest win' positions, the optimal line is not especially complicated. White starts with a king manoeuvre designed to gain a tempo: l ... tba5+ 2 �b5 ! (not 2 �c5? 'i1i>b2) li)b? 3 �c6! li)a5+ 4 c;t;.>b6 ! li)c4+ 5 �c5 ! li)d2
.i. +.0. v ll:l 255
occurs in line 2 below) 19 �c5 ! ll':ld7+ 20 �c6 ll:le5+ 2 1 �d5 ll':ld7 22 .i.el �b3 23 �c6 ll:le5+ (alterna tively, 23 ...ll':lb8+ 24 �b5 ll':ld7 25 .i.g3 ll':lb8 26 .i.e5 �c2 27 �c4 �d2 28 .i.d4 �e2 29 �d5 �d3 30 .i.a7 ll':ld7 3 1 .i.g 1 �c3 32 �e6) 24 �c5 ! ll':ld7 + (24 . . . ll:lg6 25 .i.d2 ll:le5 26 .i.e3 ll:lg6 27 .i.d4 ll:lf8 28 �c6 �b4 29 .i.g7 ! winning as in the note to Black's 1 8th move) 25 �b5 and now: 1) 25 ll:lf8 26 .i.d2 (White in tends to transfer his bishop to g7) ll':ld7 27 .i.e3 ll:lf8 (27 ...�c3 28 �c6 ll:le5+ 29 �d5 ! ll':ld7 30 �e6 ll:lf8+ 3 1 �e7 ll:lg6+ 32 �f7 ll:le5+ 33 �e6 ! ) 28 .i.h6 ll':ld7 29 .i.g7 �a3 30 �c6 ll':lb8+ 3 1 �c7 ll:la6+ 32 �b6 ll':lb8 33 .i.d4 ll':ld7+ 34 �c6 ll':lf8 35 .i.g7 ll:le6 36 d7 ll':ld8+ 37 �d6 �a4 38 .i.f6 ll':lb7+ 39 �c6 ! ll:la5+ 40 �c7 and wins. 2) 25 �c2 26 �c6 ll:le5+ 27 �d5 ! ll':ld7 28 .i.f2 ll:lf8 29 .i.d4 �b3 30 �c6 �b4 3 1 .i.g7 ! ll:le6 32 d7 ll':ld8+ 33 �d6 �b5 34 .i.d4 ! (recip rocal zugzwang) ll':lc6 35 .i.f6 �b6 36 .i.h4 �b5 37 .i.g5 �b6 38 .i.e3+ �b7 39 .i.d2 ll':ld8 40 .i.a5 ll':lf7+ 4 1 �e6 �c6 42 �e7 and White will either promote or capture the knight next move. The chart on the following page breaks the 2 1 1 2 reciprocal zug zwangs down according to the loca tion of the pawn. ...
/-
365
Original (White wins much more easily after 5 ...ll:la5 6 .i.h2 ll':lb7+ 7 �b6! ll':ld8 8 d6 ! �b2 9 d7 ! �b3 10 .i.c7 ll':lf7 1 1 �c5 �a4 12 �d5 �b5 1 3 �e6! �c6 14 .i.a5) 6 �d4 ! ll':lf3+ (6 . . .ll':lb3+ 7 �c3 ll':la5 8 d6 !) 7 �e4 ! ll:lg5+ (now there is attractive echo as White re peats the manoeuvre on the other side of the pawn) 8 �f5 ll':lf7 9 �e6 ! ll:lg5+ 1 0 �f6 ll':le4+ 1 1 �e5 ! ll':ld2 1 2 .i.c5 ! (at long last White has a breathing space to improve the posi tion of his bishop) �b2 1 3 .i.b4 ll:lc4+ (alternatively, 1 3 . . .ll:lf3+ 1 4 �f6 ! �b3 15 d 6 ! ll':ld4 1 6 d7 ! ll':lc6 17 .i.e7 ! �c4 1 8 �e6 ! ll':ld4+ 1 9 �e5 ll':lc6+ 20 �d6 �b5 2 1 .i.h4 �b6 22 .i.f2+ �b7 23 .i.el ll':ld8 24 i.a5 wins) 1 4 �d4 ! ll':lb6 ( 1 4 ...�b3 1 5 �c5 ! ll:le5 16 d6! ll':ld3+ 17 �b5 ll:le5 transposes) 15 d6! �b3 1 6 �c5 ! ll':ld7+ 1 7 �b5 ll:le5 1 8 .i.a5 �c2 ( 1 8 ...ll':ld7 19 .i.b6 �c3 20 �c6 ll:le5+ 2 1 �d5 ! ll':ld7 22 .i.d4+ �b4 23 �c6 ll:lf8 24 .i.g7 ! ll':le6 25 d7 ll':ld8+ 26 �d6 �b5 27 .i.d4 ! reaches a reciprocal zugzwang which also
...
4.5:
Underpromotion
We have met underpromotion a few times in earlier chapters, but it plays
256 .t+� v lLJ
Thus some very long wins are possi ble with .t+lll v lll, the maximum being 77 moves. These 'long win' positions are characterised by the defender's king being confined to a corner, and the defender's knight being cut off from the king, but not actually trapped. However, it should be emphasised that poor king posi tion is a greater handicap to the de fender than a disconnected king and knight. If the defender's king is trapped in or near the comer, then he may lose even if the knight is re united with the king. In these cases White can use threats to the king to drive the king and knight apart. We will divide up this section ac cording to the piece White promotes to. Rook promotions arise occasion ally, but it is normally quite obvi ous when such an underpromotion is necessary. (366): This example is particu larly attractive in that it combines underpromotion with the 'excelsior' march of the pawn all the way from its starting position on the second
a much more prominent role in the ending of .t+� v lll. Of course, the material balance of :+.t v lLJ is normally an easy win, but as we shall see in the next chap ter, 2.t v lLJ is in general also a win. As this fact was only discovered in 1983, it would be reasonable to sup pose that studies composed earlier might be unsoundly based on the as sumption that 2.t v lLJ is a draw. In fact, surprisingly few studies are af fected in this way; in diagram 366, for example, White can't promote to a bishop because his other bishop is attacked by the knight. In contrast, the ending of .t+lll v lLJ is almost always drawn. Winning chances normally arise only when the defender's knight is cut off from the king. In this situation, the knight is usually either trapped very quickly or escapes to freedom and a draw. The percentage of 'marginal' posi tions is extremely low, but strangely enough when a marginal situation arises, a curious equilibrium can be set up in which the position stays in a marginal state for a very long time. '
.·
.
.
.
a
.·.
b
e
'
d
1
12
37
71
77
'
61
78
120
161
5
100
133
1 13
143
4
80
166
85
89
3
93
1 17
41
63
2
69
79
44
80
.t+.0. v ttJ 257
- - - � •@• • � � - - - - - - - - - m m m m - lS - - - - -·366
+I= Blandford, 1949 1 st Pr., Springaren
rank (the rook underpromotion by itself had been shown earlier by Kubbel). White wins by 1 .td4+! (freeing the pawn) 'it>a8 (Black's pieces are so miserably placed that he cannot prevent the straightfor ward advance of the pawn; lacking a normal defence, his only chance is to play for stalemate) 2 c4 ! tbd2 3 c5 ! tbb3 4 c6 ! tba5 5 c7 ! tbc6 (this is the cunning point of Black's de fence; he threatens to force a draw by . . . tbb8+, so White must act at once) 6 c8.l:t+ ! (the only move to win; 6 'it>xc6? and 6 c8'it'+? tbb8+ both lead to stalemate, while 6 .tc5? tbe5+ 7 'iti>d6 'iti>b7 ! and 6 .tc3 ? tbb8+ 7 'iti>d8 tbc6+ ! 8 'it>e8 'it>b7 9 'iti>d7 tbb8+ are 'normal' draws) and White wins with his extra rook. (367): White wins by 1 e6 ! tbf4 2 .td8+ ! (not 2 e7? tbd5 ! and White can only reach a drawn .t+ttJ v tiJ) 'iti>a6 (Black cannot stop the pawn by normal means so he has to try ex ceptional means) 3 e7 ! tbd5 (after
+I= 367 Troitsky, 1916 (version) Eskilstuna Kuriren 3 . . . tbe6 White can still promote to a rook, but he can also win without an underpromotion by 4 'iti>b8 tbg7 5 .tc7 'it>b5 6 .te5 tbe8 7 'it>c8) 4 e8l:t! (4 e8'ii'/.t? tbc7+! forces stale mate) with a clear extra rook. There are a few other possibili ties for a rook underpromotion. One is W'it>c2, .tf7, .0.h7 v B'iti>al, tbh2 (Herbstmann, 1927); after 1 ...tbg4 2 h7 ! tbf6, promotion to queen or bishop delivers immediate stalemate by pinning, but 3 h8l:t ! wins. Read ers should also note the rook promo tions in diagram 330 and line 2b of diagram 361 . Turning now to bishop promo tion, the following study is typical. (368): We may assume that Black moves first. The main variation runs 1 ...tbg4 2 .tg6+ ! (not 2 .tg8+? 'iti>h8 ! 3 f7 tbe5 ! with a knight fork on g6) 'it>h8 (Black is playing for stalemate) 3 f7 ! tbh6 (after 3 ... tbf6, White can win by promoting to either rook or bishop) 4 f8.t ! (and
258 i. +8
v
tt:i
It is worth mentioning that if we take the position after 3 . . . tt:lh6, and move White's bishop from g6 to fS (W
+I368 Filaretov, 1925 (end of study) Shakhmatny Listok not 4 f8'ii'+ ? tt:lg8+ forcing stale mate, nor 4 f8.l:+? �g7 ! when both 5 i.f7 ti:lf5+ ! 6 �e8 ti:ld6+ ! and 5 i.d3 tt:lg8+! 6 �e8 ti:lf6+! lead to perpetual check) and now have an ending of 2i. v tt:l. It is now known that this is a win in general, but at the time Filaretov composed his study, the general result was thought to be a draw. However, this is a very favour able case for White, with Black's king trapped in the corner and his knight cut off from the king, so Filaretov was able to prove a win. The database informs us that it takes just eight moves to mate or capture the knight, for example 4 ...tt:lg4 5 i.f5 tt:le5 (or 5 . . . tt:le3 6 i.e6 tt:lg2 7
+I= Galushko, 1985 2nd Pr., Birnov Mem. Tny.
369
(369): Black to play draws by 1 . . .tt:le8! 2 g8i. (2 g8'ii'? and 2 g8.l:? are stalemate) ti:lf6 ! 3 i.e6 ti:ld7+! 4 �a6 tt:lb8+ ! with stalemate or per petual check. White to play wins by 1
i.+!::, v lll 259
One other situation is worth men tioning, but does not deserve a dia gram. Starting with W�a5, i.d5, /::,d7 v B�a7, llla6 and Black to play (Kok, 1934), White wins by 1 ...lllb 8 2 d8i. ! (2 d8'ji'? and 2 d8l:.? are both met by 2 . . . lllc6+ 3 i.xc6 stalemate) llld7 (the situation is the similar to diagram 369; in this case it takes ten moves to round up the knight) 3 �b5 llle5 (or 3 ...lllf8 4 �c6 lllg6 5 i.f6 lllf4 6 i.c4 lllg6 7 �c7 lllf4 8 i.d4+ �a8 9 �c8) 4 i.f6 llld7 5 i.d4+ �b8 6 �c6 lllf8 7 i.f6 �c8 8 i.f7 lllh7 9 i.e7 and the trapped knight is lost. Now we come on to the most im portant case, that of a knight pro motion. The following position is a good example.
1 . . .lllg7 2 d6 �g8 3 i.e8, for exam ple 3 . . . lllxe8 4 d7 ! or 3 ... �f8 4 i.g6!) 2 d6 (2 i.f5+ is also possible, transposing to the main line) llla6 3 i.f5+ ! (not 3 i.b5? lllb 8 ! 4 �f6 �h6! 5 �e7 �g5 6 �d8 �f6 7 �c7 �e6 ! 8 i.fl �d5 ! 9 i.c4+ �c5 when White is on the wrong end of a reciprocal zugzwang) and now there are two lines: 1 ) 3 �h6 4 d7 ! lllb8 5 d8lll ! and now the second phase begins. i.+lll v lll can be very complicated, but this is an elementary case because Black's knight can only flee as far as b4, whereupon i.bl closes the net. It takes just eight moves to round up the beast: 5 ... llla6 (5 . . .�g7 6 �d6 llla6 7 lllc6) 6 �d6 �g5 (or 6 ...lllb4 7 i.bl ! �h5 8 llle6 llla6 9 lllf4+ "'g5 1 0 llld 5) 7 i.bl ! (not 7 i.d3? lllb4 ! 8 i.bl �f4 9 llle6+ �e3 ! and the extra tempo allows Black to es cape) �f4 (7 ...lllb4 8 llle6+ �g4 9 lllc 7) 8 lllc 6! �e3 9 i.f5 ! �d2 10 i.c8 ! and White is just in time. 2) 3 �g7 4 d7 ! lllb8 (4 ...lllb4 5 i.e4) 5 d8lll ! (this case takes one move longer than line 1) llla6 (or 5 . . . �f8 6 �d6 ! �e8 7 llle6 �f7 8 lllc 5) 6 llle6+! �h6 (if Black moves to the a2-g8 diagonal then he loses even more quickly, for example 6... �g8 7 �d6 lllb4 8 lllc7 �g7 9 i.bl) 7 �d6 lllb4 8 i.bl llla6 9 �c6 lllb4+ 10 �c5 llla6+ 1 1 �b6 lllb4 12 lllc7 followed by 13 �b5. Although composers have pro duced many examples in which the knight promotion wins, it has to be said that in most examples the black knight is trapped immediately after .••
...
370
+I= Missiaen, 1963 Schakend Nederland
(370): White to play wins, but the battle may be divided into two phases. In the first phase White en sures the promotion of his pawn by 1 i.d7 ! lllc7 (White wins easily after
260 J.+.0, v lLJ
the underpromotion, and is rounded up within a few moves. However, there are some compositions with additional points, and we take time out to look at a couple of these. Perhaps the most complex piece of analysis was attempted by Pogos iants. Although there were some se rious inaccuracies in the sidelines, his main ideas were absolutely cor rect. The difficulty of analysing marginal .i.+lD v lLJ positions can hardly be overstated, and I doubt if anyone else could have done better.
are characteristic of the more diffi cult winning positions, which we do not have space to examine exten sively in this book. The analysis runs 2 ...�g4 (2 ...�g3 loses more quickly after 3 'iti>e5 �f3 4 �e6 lLla7 5 �d5 ' lLlb5 6 �c4 ! lLla3+ 7 'iti>d3 lLlb5 8 .i.b6 �f4 9 .i.c5) 3 'iti>e5 and now: 1 ) 3...'iti>f3 transposes into the bracket above. 2) 3...lLla7 4 'it>d5 ! lLlb5 5 �c4 ' lLla3+ (5 ...lLla7 6 'it>c5 �f5 7 lLJd6+' 'it>e6 8 .i.b6!) 6 'it>b3 lLlb5 (6...lLlbl 7 .i.a5) 7 .i.b6 �f5 8 �b4 winning. 3) 3...'it>hS 4 �f5 'it>h6 5 .i.g5+ 'iti>h7 6 'iti>e6 with two lines: 3a) 6...lLla7 7 'ifi>f7 lLlb5 (7 . . .lLJc6 8 lLlf6+ 'it>h8 9 lLld7 lLlb4 1 0 lLlf8 lLJd5 1 1 .i.h6 and 7 ...lLJc8 8 .i.e3 �h8 9 .i.c5 'it>h7 1 0 .i.f8 are also lost) 8 .i.e3 'iti>h8 9 .i.h6 �h7 10 .i.f8 ! (n<>l 10 .i.g7? lLJd6+ !) 10 ...lLld4 1 1 lLlf6+ �h8 12 .i.g7#. 3b) 6 lLlb6 7 'it>f7 with one more branch: 3bl) 7 lLlc8 8 .i.e3 corners the knight. 3b2) 7 ..lLJd.S 8 .i.d2 �h8 (or 8 . . . lLlb6 9 lLlf6+) 9 .i.h6 ! �h7 10 .i.g7 and mate next move. 3b3) 7...'it>h8 8 .i.f4 lLld7 9 .i.d6 'it>h7 10 lLlc7 lLlb6 ( 1 0...lLlbS 1 1 .i.f4 lLJc6 12 lLle6 'it>h8 13 lLlg5 lLld8+ 14 'it>f8) 1 1 .i.f4 ! 'it>h8 1 2 lLle6 lLlc4 13 lLJg5 lLJd6+ 1 4 'it>f8 lLlc4 15 .i.g3 mating. 3b4) 7...tlxl7 8 .i.f4! lLJc5 9 lLJf6+ �h8 1 0 .i.c7 lLlb7 1 1 lLle4 'iti>h7 12 'iti>e6 'iti>g6 13 'ifi>d5 'it>f5 1 4 lLlc3 and 15 �c6. 3b5) 7...lLlc4 8 .i.f4 �h8 9 lLJc7 'iti>h7 (9 .. lLlb6 10 lLle6 lLld7 1 l lLlg5 ...
•••
/37 1 Pogosiants, 1978 (end of study) '64 '
(371): After 1 ...lLlc8 2 e8lD! we reach a position in which it is not possible to trap the knight directly (so this differs from the previous po sition). In order to win, White must combine threats to corner the knight with an attack on Black's king. This dual approach is typical of the more complex .i.+lD v lLJ wins. While this is a relatively simple example (it takes just 14 moves to win), the ideas
.
.
.t+� v lLi 261
lLif6 1 2 .ie3; 9 ... lLia5 10 lLie6 lLJc6 1 1 lLig5 lLid8+ 12 f8) 10 lLie6 h8 l l lLig5 wins as in line 3b3. It is not surprising that analytical errors are common in this type of po sition.
8 lLid6 �a8 9 lLib5 lLie6+ 10 c8 lLif4 1 1 .ib3 mating. 2b) 4 lLirs 5 lLic6+ a8 6 .if7 lLig7 (6 . . . lLie3 7 lLib4 a7 8 .ie6 �a8 9 .ic8) 7 lLid4 a7 8 d6 wins the encircled knight. Using the database, it is possible to construct phenomenally complex positions which are beyond human ability to solve or check for sound ness. The following position is espe cially curious. ••.
I372 Rossi, 1955 (end of study) 3rd Pr., L'ltalia Scacchistica
(372): 1 . . .lLif8 2 d8lD ! (2 d8'6'? lLie6+) lLig6 (2 . . .a6 3 .ie4) and now: 1 ) 3 .ib7 (the composer's line is enough to win) lLie7 4 .ie4 ! a6 (the composer's main line continued 4 ... lLig8 5 lLif7, even though 5 lLic6+ wins at once) 5 lLif7 lLig8 6 �d6 lLif6 7 .if3 ! lLie8+ 8 �d7 lLif6+ 9 e7 lLig8+ 10 'itte6 followed by .ie4-h7. 2) 3 .ibS (this alternative win is slightly faster) lLie7 4 .ic4 ! (a curi ous position; with Black to play, White wins in 8 moves, but with White to play it takes a whopping 27 moves) and now: 2a) 4 'itta8 5 .ie6 rl;a7 6 lLif7 lLig6 (6 ... lLig8 7 .ic4 lLif6 8 lLid6 a8 9 lLib5 is the same) 7 .ic4! lLif4 ...
+I=
373
Original (373) : This is probably the most complex position in the book. Due to space limitations, I will only be able to give a sketch of the winning proc ess below. 1 rs! Not 1 e5? g6 2 d5 f7 3 rl;c5 lLJc7 ! 4 .ie5 lLie8 ! 5 d5 e7 ! and Black draws. rru.1 1 Or: 1) 1 h6 2 .ie5 h7 3 e6 �g6 4 rl;d5 ! f5 (4 . . .f7 5 c5 ! lLia7 6 c7 ! e6 7 b6! lLics+ 8 b7 ! ...
...
262 .t+� v tLJ
tDe7 9 .tg3 �d7 10 .tf2 tDc8 1 1 .tc5) 5 .tb8 tDc3+ 6 �c4 ! (6 �c5? �e6 !) 6 . . .tDa4 7 .ta? �e6 8 .td4 ! �e7 9 �b5 �d6 10 .tg7 �c7 1 1 .te5+ winning the knight. 2) 1...lLJc7 2 .te5 tDa8 3 �e6 �g6 4 �d7 �f5 5 .tc7 �e4 6 �c8 �d5 7 �b7 ! �c5 8 i.d8 �b5 9 .te7 �a5 1 0 i.g5 �b5 1 1 i.e3 with zug zwang. tDbS 2 c7! �h6 3 c8tD! White wins easily after 3 . . . �h4 4 �e6 tDc7+ 5 �d6 ! tDa6 6 .ta3, while 3 ... tDc7 4 tDe7 ! transposes to the main line. lLJc7 4 �! 4 . . . �h7 5 �e7 tDc7 6 tDb6! tDa6 7 .ta3 ! tDc7 8 �d7 tDb5 9 .tb2 ! tDa7 10 tDc4 tDb5 1 1 �c6 corrals the knight. �hS 5 lLJe7! Alternatively: 1) 5...�h7 (the king goes the other way) 6 .te5 tDb5 7 .if4 tDd4 (7 ... tDc3 8 �f7 ! tDe2 transposes) 8 �f7 tDe2 9 .id2 tDd4 10 tDd5 tDf3 1 1 .tf4 tDd4 12 tDf6+ �h8 1 3 tDd7 tDf5 14 i.d2 tDd6+ 1 5 �g6 tDc8 1 6 .tc3+ �g8 1 7 .tb4 �h8 1 8 �f7 with instantly decisive threats. 2) 5 lLJa6 6 .td4 tDb4 7 �f5 �h5 8 i.b6 tLJd3 (or 8 ...�h4 9 �e4 ! �g4 10 i.a5 tDa6 1 1 �d5 ! tDb8 1 2 �d6 �f3 1 3 tLJd5 tDa6 14 �c6 �e4 1 5 i.e l ! �b8+ 16 �d6! tDa6 17 i.g3 �f3 1 8 i.h2 �g2 1 9 .te5 �f3 20 �c6 �e4 2 1 .ih2 �d4 22 i.d6 �c4 23 tDb6+) 9 tLJd5 tDb2 10 .ta5 tDc4 1 1 .ib4 ! tDb2 1 2 �e4 tDa4 1 3 lLJe3 tDb6 1 4 �f5 �h6 1 5 �e6 �g5 1 6 .td2! tDa4 17 tLJc4+ �g4 1 8 �d5 ! •••
(Black's knight is imprisoned, but it is still not easy to actually capture it) �f5 1 9 �d4 �e6 20 .ia5 ! (amaz ingly, the bishop is heading for h8; this manoeuvre is worth noting, be cause it also arises in the main line) �f5 21 i.c7 �g4 22 i.d6 �f5 23 i.e5 �e6 24 i.h8 �d7 25 �d5 ! �c7 26 .id4 ! �d7 27 .te5 �e7 28 �c6 �e6 29 i.al and the knight fi nally disappears. 6 �5 It takes seven moves longer to win this position when White is to move, so we can count ourselves lucky that Black is to move. 6 lLJa6 After 6...�h4 7 �e5 ! tDa6 8 tDc6 tDc5 9 .td4 ! tDd7+ 1 0 �e6 tDf8+ 1 1 �f5 �h5 1 2 tDe5 ! �h6 1 3 tDg4+ �h7 Black's king and knight are re united, but his precarious king posi tion tips the balance in White's favour, and White wins by 14 �f6 ! �g8 1 5 tDh6+ �h7 16 tDf5 tDg6 17 i.f2 (the critical moment; Black's king and knight cannot stay together any longer) 1 7 ...tDf4 ( 1 7 ... tDh8 1 8 tDe7 �h6 1 9 .te3+ �h7 20 .tf4 and 1 7 ...tDf8 1 8 tLJe7 �h6 1 9 �f7 ! tLJh7 20 tDg8+ ! �h5 2 1 �g7 ! tDg5 22 tDf6 !# lose more quickly) 1 8 tDe7 ! tDd3 19 .te3 �h8 20 �f5 �g7 2 1 tDc6 ! �f7 2 2 �e4 ! tDb2 2 3 .id4 tDc4 24 �d5 ! tDd2 25 tDe5+! �g8 26 i.e3 �bl 27 �c4 tDa3+ 28 �c5 ! tDc2 29 .id2 �g7 30 tDf3 �f6 3 1 �c4 ! �f5 3 2 �d3 ! �g4 3 3 tDe5+ ! �f5 3 4 tDc4 tDal 3 5 tDa5 followed by .tc3. lLJcs 7 lLJdS Or 7 ...�h6 8 .id4 �h5 9 �e6. ••.
.i.+� v lb 263
�h4 8 .i.c3 If the black king goes the other way, then White wins by 8 ...'if.?h6 9 lLie3 lLid3 10 .i.d4 lLic l 1 1 lLig4+ �h5 1 2 .i.f2! lLie2 1 3 lLif6+! �h6 14 lLie8 lLic3 1 5 .i.e3+ �h7 1 6 lLic7 �g8 1 7 �g6 lLie4 1 8 lLib5 �f8 1 9 �f5 ! lLig3+ 2 0 �f4 lbh5+ 2 1 �g5 ! lLig7 22 �g6 ! lbe6 23 �f6! lLid8 24 lLic7 �g8 25 .i.c5 lLib7 26 .i.b4 lLid8 27 lLid5 lLif7 28 lLie7+ �f8 29 lLif5+! �g8 30 .i.a3 lLih8 31 lLih6+ �h7 32 .i.c 1 lLig6 33 lLif5 lLif8 34 .i.a3 ! lLid7+ (34 ...lLig6 35 .i.d6 lLih8 36 lLie7 �h6 37 .i.h2 'if.?h7 38 .i.f4) 35 'if.?e6! lLib6 36 lLie3 ! �g6 37 .i.c5 lLia4 38 .i.d4 !. 9 .i.el+ 'if.?h3 10 lLif4+! �h2 lLia4 11 .i.b4 l l ...lLib3 12 �e4 ! 'if.?g3 1 3 .i.el + ! 'if.?g4 1 4 lLid3 �h5 1 5 �d5 lLial 1 6 �c4 lLic2 1 7 .i.f2 ! lLia3+ 1 8 � b4 ! lLibl 19 .i.el traps the knight. 12 'if.?g4 lLib2 1 2...�gl 1 3 'if.?f3 lLib2 transposes . 13 �f3 �gl 14 lbe2+ 'if.?hl 14 . . . 'if.?fl is another important variation; in this case White wins by 14 . . .'if.?fl 1 5 lLig3+ �gl 1 6 .i.c5+ �h2 17 lLie4 ! lLic4 18 lLig5 ! lLie5+ 19 �e4 ! lLig4 20 'if.?f4 ! lLif6 2 1 .i.d4 lLig8 22 �f3 lLie7 23 'if.?f2 'if.?h 1 24 .i.e5 lLic6 25 .i.d6 lLia5 26 lLie4 lLib7 27 .i.e7 lLia5 28 .i.d8 lLic4 29 .i.c7 lLid2 (an ingenious stalemate de fence) 30 lLig3+! �h2 3 1 �e3 lLib3 32 lLie2+ 'if.?g2 33 lLif4+ �fl 34 lLid3 ! �g2 35 .i.d6 lLia5 36 lLie5 ! lLib7 37 .i.e7 ! �g3 38 lLic6 (the knight is trapped) �g4 39 �e4 �g3
40 lLid4 (surprisingly, the quickest win involves playing the bishop and knight away from their present posi tions) �f2 4 1 .i.h4+ �g2 42 �d5 �h3 43 .i.e7 lLia5 44 �c5. lLid3 15 �g3 Since there are only 38 moves left to go, we can move over to a con densed format, but I will be generous and add a bonus diagram in compen sation: 16 .i.d6 lLiel 17 �f2 lLid3+ 18 �n lLib2 ( 1 8 ...lLic l 19 lLic3 lLib3 20 �f2 lLid2 2 1 .i.c5 lLic4 22 lLie4 lLia5 23 .i.b6 lLic4 24 .i.c7 lLid2 transposes into the previous note) 19 .i.f4 lLic4 20 'if.?n lLib2 21 .i.b8 lLic4 22 .i.c7 lLib2 23 lLid4 lLid3+ 24 'if.?f3 lLib2 25 lLib5
• • • • . � . . • • • • .tl). • • • • • • . . -�· - . . . • • • •• 374
+/-
25...lLlc4 (or 25 . . . lLia4 26 .i.d6 lLib6 27 lLia3 lLid5 28 lLibl lLib6 29 lLid2 lLid5 30 lLie4 �gl 3 1 lLig3 lLif6 32 .i.c5+ �h2 33 lLifl + �h3 34 lLie3 ! lLig8 35 lLig2 lLih6 36 lLif4+ �h2 37 lLid3 �h3 38 .i.e7 lLif5 39 lLif2+ �h2 40 .i.f6 lLih6 4 1 �f4 �gl 42 lbe4 'if.?g2 43 �g5 �f3 44 lLid2+ ! �e2 45 lLic4 ! lLig8 46 .i.d8 ! �d3 47
264 .i.+� v lD
tlld6 'itid4 48 �g6 �d5 49 tllf5 ! �e6 50 i.h4 �e5 5 1 i.g5 �e6 52 tllg7 +) 26 �e2 �g2 27 tllc 3! tllb2 28 tlld5 tllc4 29 �d3 tllb2+ 30 �e3 'itifl 3 1 i.g3! (3 1 �d2? tlla4 ! 3 2 i.d6 �f2 ! 33 �c2 �f3 ! 34 �b3 �e4 ! draws) 'itig2 32 i.b8 (White transfers the bishop from c7 to b8 with gain of tempo) tlla4 (now 32...�fl 33 �d2 ! tlla4 loses to 34 i.a7) 33 i.d6 tllb2 (33 . . . �fl 34 i.a3 �g2 35 �d4 �f3 36 i.c 1 ! �g4 37 tlle3+ �f3 38 tllc4 �g4 39 i.a3 and wins as in the analysis of 5 ...tlla6) 34 �e2 tlla4 35 tlle3+ �b3 36 �d3 tllb6 37 'itie4 tlla4 (37 . . .'itih4 38 i.c7 tlld7 39 �f5 ! �h5 40 i.d6 tllb6 41 i.c5 tlld7 42 i.e7 tllb8 43 tllc4 tllc6 44 i.c5 tllb 8 45 �e6 �g6 46 tlle 5+ �g5 47 i.e7+ �f4 48 tlld3+ 'itie3 49 tllb4 ! 'itid4 50 �d6 !) 38 i.b4 tllb6 39 �f5 tlla4 (after 39 . . . �g3 40 i.c5 ! tlla4 4 1 i.d4 ! 'itif3 42 tllc4 ! �e2 43 'iPe4 ! Black falls victim to a reciprocal zugzwang and White wins by 43 . . . �dl 44 �d3 ! �c l 45 i.e3+ �d l - White now passes the move to Black - 46 i.gl 'iPc l 47 i.d4 'iti>dl 48 i.e3 'iti>el 49 'itic2 'itie2 50 i.a7) 40 tllc4 �g3 41 'itie4 (4 1 'itie5 'iPg4 42 �d4 is also possible) �g4 42 'itid4 'itir5 43 i.f8 (heading for h8, just as in the note to Black's fifth move; 43 i.d6 �g4 44 i.f8
'itif5 45 i.g7 'iPe6 46 i.h8 is another way to reach the same position) �e6 44 i.g7! �f7 45 i.e5 �e6 46 i.b8 �d7 47 �d5! 'itic7 48 i.d4! �d7 49 i.e5 �e7 50 �c6 �e6 51 i.al �f5 52 �b5 'itie4 and now White can take the knight. I have mentioned in the introduc tion that I will be ignoring the effect of the 50-move rule in this book, but it is worth pointing out that by the time Black plays 52 . . . 'iPe4 in the above analysis, there have been 50 Black moves and 49 White moves without a pawn move or a capture; in other words White can take the knight exactly on the last half-move before Black could claim a draw un der the 50-move rule. Of course, White had to play extremely accu rately to win the position at all, but the accuracy required to win the po sition without at any stage losing a move is extraordinary. In the entire sequence of 53 moves, there are only two points where alternatives were possible (at moves 4 1 and 43). The remaining 5 1 moves are absolutely unique if White is constrained not to lose a tempo. Could this be the most difficult chess puzzle ever? I think this is an appropriate note on which to finish the coverage of minor piece and pawn v minor piece.
5
2.i.
v
ttJ
We are now coming to the end of a long but interesting journey. My three endgame books based on the Ken Thompson endgame databases, totalling 907 pages of detailed analy sis, are drawing to a close. In these books I have covered virtually every important 5-man ending with the ex ception of 'i'+t::, v 'i' and 2lli v /::, , The first of these proved too hard to understand; I will leave it for a future author to deal with. I decided to omit the latter after discovering that the pre-database work of Troitsky and other analysts is astonishingly ac curate. I felt that my contribution would be limited to tidying up a few loose ends, and decided that the space would be better employed covering those endings which have been more poorly served by earlier analysts. However, I can add one original fact about the ending of 2tD v /::, it contains 3 1 24 reciprocal zugzwangs. After so much analysis the readers are probably almost as tired as the author, so perhaps they will allow me a small indulgence in this final chapter. I would like to take some time to discuss the philosophy be hind these books, using the ending of 2.t v tD as an example; those who by now have become analysis junkies need not worry - they will get a final 'fix' later on in the chapter. -
First of all I would to explain a lit tle terminology, which will doubt less be familiar to many readers but will be probably be unfamiliar to others. A ply is half a move i.e. 1 e4 e5 consists of two plies. When a database is constructed, each won position is associated with a depth, which we may take to be the mini mum number of plies required to force a win against any defence. The precise definition of 'win' may affect the depth, for example two reason able definitions of win are 'give mate' , or 'either give mate or reach a winning position with fewer men on the board ' . Databases constructed according to these two definitions will assign different depths to the same position. The second definition is usually more natural; in 2.t v tD it is equivalent to 'either give mate or capture the knight without deliver ing stalemate' , and this is indeed the definition used in the Ken Thompson 2.t v tD database (the definition of 'win' becomes more tricky when there are pawns on the board, but I will skip this !). In a winning posi tion, an optimal move is one which reduces the depth by one. In the case of a white move, this corresponds to the concept of not losing time; for a black move, it means spinning out the game for as long as possible. A move which is not optimal is called
266 2.i. v lb
sub-optimal. There may be several optimal moves in a particular posi tion; if so, these are described as equi-optimal. For example, in the diagram below White to play has a unique optimal move, 1 .i.b4, but after the reply I ...t'Dd8 there are four equi-optimal moves, namely 2 .i.c3, 2 .i.d2, 2 .i.el and 2 .i.c5+. It is fitting that this last chapter deals with the ending which has felt the greatest impact from the computer - indeed, the computer changed the evaluation of the whole ending. To summarise the history of 2.i. v lb, in 1 85 1 Kling and Horwitz pub lished the following position:
+I375 Kling arul Horwitz, 1851
(375): They evaluated this as a draw on the grounds that the king and knight form a fortress which holds White at bay. For over a cen tury endgame authors repeated this view, generally giving a line such as 1 .i.e7
White is not making progress (this particular variation is from Cheron). Indeed, the argument is superficially convincing; it doesn't take long to see that direct attacking attempts come to nothing, mainly because it is impossible to check on the a5-d8 diagonal and thereby dislodge Black's king from the b6 and c7 squares. Kling and Horwitz also demonstrated that if Black is unable to set up this defensive formation, then he normally cannot resist the at tacking power of the two bishops. Nevertheless, starting from a general position with White's king far away, Black can adopt the formation of knight on b7 (or one of the other three topologically identical squares, i.e. b2, g2 and g7) and king on b6 or c7. I will refer to a position of this type as a 'KH-type position' or a ' KH set-up' . I will call the specific position of diagram 375 'the KH po sition' . Perhaps the first real note of cau tion was sounded by John Roycroft, who in 1972 (in his book Test Tube Chess) included 2.i. v lLJ in a list of endings in which, as he put it, 'The result is controversial or not known' . He added that 'only one position has been generally agreed to be drawn', presumably referring to the KH position. Roycroft correctly commented that the KH position is not really a fortress since White can dig Black's king out of the cor ner by ' 1 .i.e7
2.1 v lb 267
take up a comparable position in an other comer, though it "looks as if he can" ' . While this proved to be a pro phetic comment, the concrete line Roycroft offered is very inaccurate. After 6 �b5 White is actually 10 moves further away from a win than he was in the KH position, since Black can indeed regain his KH set up. With the aid of the database, we can identify the culprit as 6 �b5 (6 �b4 wins 16 moves more quickly). Roycroft proposed 2.1 v lb as an ideal ending for the construction of a database, and when Ken Thompson actually made the database in 1983, it overturned 132 years of endgame theory by proving that, apart from a few trivial positions in which Black can force an immediate draw, the ending is always winning for the bishops. In particular, with perfect play White can win the KH position in 45 moves. However, in one way Kling and Horwitz were vindicated by the data base. The longest win in this ending is a massive 66 moves, but the KH position plays a key role in all the very long wins with 2.1 v lb. The longest wins generally start with one bishop trapped in the corner (see dia gram 3 8 1 ), and there are some pre liminary moves during which White frees this bishop. Then the king and two bishops start to attack the black king and knight; at this stage Black's pieces occupy central positions. The attacking power of the bishops is simply too great to resist when they can attack from every side, and the black king is steadily driven towards
a corner. This phase of the play is relatively straightforward, because White's moves are based on familiar concepts: harassing Black's pieces and forcing the enemy king to retreat step by step. Finally the black king approaches a corner, but now he sets up a KH-type position. The long wins invariably pass through a KH type position (or the imminent possi bility of one). Were it not for the KH possibility, the longest win with 2.1 v lb would be far less than 66 moves. Now comes a key moment. Using a manoeuvre similar to that proposed by Roycroft, White prises Black's king out of the comer. In the KH po sition itself, Black's king is expelled from the b7-corner and promptly heads towards g7 . If Black could re store his KH set-up in the g7 corner he would save the game, but it turns out that this is impossible against ac curate play. Skilful manoeuvring by the bishops denies Black the squares he needs to manoeuvre his pieces into a new KH-type position, and he ends up with his king floating around the edge of the board (for ex ample on e7 or g5), unable to set up a KH-type position in either of the nearby corners. During this phase accurate play is especially important, because a slip up could result in Black regaining his KH set-up, which could easily set the clock back by at least 20 moves. However, assuming that White plays accurately, he gradually squeezes Black more and more, until finally the enemy king is forced back to the very edge of the board. The final
268 2J.
v
lt)
phase typically involves Black's king cowering in a corner with no pros pect of a KH-type position, while the bishops mercilessly finish him off, either by a mating attack or by sepa rating the knight from the king and trapping it. Although Kling and Horwitz did not evaluate diagram 375 correctly, they nevertheless found the de fender's basis for prolonged resis tance, and that is an achievement in itself. The KH position is so funda mental to this ending that it is no ex aggeration to say that if you can win diagram 375, then you should be able to win virtually any position with 2J. v lt). This explains why we will focus so closely on the KH posi tion. Curiously enough, since Thomp son's original 1983 discovery pro gress on understanding this ending has been rather slow. A number of authors have dutifully reproduced some of Thompson's positions, but they have usually provided just the bare main line. A single optimal line of play is not very informative; the explanation as to why certain moves are effective and others not cannot be deduced from such limited material, and this obstructs the development of true understanding. Roycroft himself has continued his investigations, publishing a se ries of articles in EG stretching from 1 9 8 3 to 1 990, but while these es tablished the general principles of the win, Roycroft concentrated al most entirely on single optimal lines of play with few, if any, variations.
However, he did identify one impor tant intermediary position (see dia gram 376 below). In general terms the winning plan is not hard to grasp, but to execute it reasonably accu rately is far from easy. Note that in the last paragraph but one I slipped in the phrase 'under standing this ending', and it is worth spending a little time to discuss ex actly what this means. For this we have to take a wider perspective. In the field of artificial intelligence, one question goes right back to the dawn of electronic computers: how can you tell if a computer is think ing? The mathematician Alan Turing put forward what is now known as the 'Turing test ' , which proposes that if you cannot distinguish the re plies of a computer from those of a human being, then the computer is thinking. Of course, this test presup poses something like a teletype link, so that the computer is not given away by the Dalek-like tones of its speech synthesiser. The advantage of the Turing test is that it takes some thing which is inherently vague and poorly-defined ('thinking') and ex presses in it a form which is actually testable. The Turing test and ver sions of it have been a hot topic ever since, but the simple fact is that at the moment no computer in the world can pass this test in an unrestricted form, or is likely to do so in the near future. In the database world what we need is a kind of Turing test for hu mans. The computer generates a da tabase for a particular ending, the
2i.
human spends some time working with the database, and then the hu man is given a test to see if he or she has 'understood' the ending. What form should this test take? Clearly, it would be pointless to expect the hu man to play optimally, as this would require a virtually complete knowl edge of the database. A more reason able test would be to pit the human, playing with the 2B, against the da tabase itself playing with the N. For example, one might set the human several positions requiring 55 moves to win. Allowing him a certain lee way, we could say that if the human can win almost every position within 75 moves, then he 'understands' 2i. v lll . In fact, some tests of this type have been performed with the end ing 2i. v lll . The value of such tests depends critically on the way the computer is programmed to use the database. Suppose, for example, that the computer plays optimally on every move or, if there is more than one equi-optimal move, it chooses between the equi-optimal moves at random. The trouble with this ap proach (which was used in the tests mentioned above) is that the com puter has such a restricted choice that the number of possible lines is very limited, and the human can work purely from memory, which is certainly no test of 'understanding' . In order to prove the truth o f this assertion, I will give a complete de scription of how to win the KH posi tion, with White to play, on the assumption that at every stage Black plays an equi-optimal move:
v
lll 269
1 i.b4 llld8 ( 1 ...llla5 2 i.c3 lllb7 3 i.d4+ c7 2 i.c3 16 llln 11 f3 20 i.c6 d4 lllg4 1 9 i.c8+ f3 22 i.a5 lllg4 23 i.c6+ c4) 40 i.g4+
...
270 2.t
v
ll)
moves at every point in the solution. Thus in the tests mentioned above, the human could succeed by simply memorising this very modest quan tity of analysis. Normal, intuitive play suffices to reach the KH posi tion, and then memory could take over. It follows that such a test is es sentially trivial from the human's point of view, and so cannot reveal anything about the human's degree of understanding. I should add that I have not em ployed any devious tricks in this analysis, for example by deliberately playing sub-optimal White moves solely because Black would then have a unique optimal reply. By us ing such tricks it is possible to sim plify the analysis even more, but the above analysis contains only optimal White moves. The apparent simplic ity of the above analysis is rather de ceptive; in fact it is the result of a considerable quantity of work. The analysis started out several times as long, but a gradual refining process of repeatedly substituting one equi optimal White move for another dis tilled it to the version given above. We can conclude that coping with optimal play presents few problems; it is the much wider range of sub-op timal defences which gives rise to the real complexity of 2.t v lt). Such a test for human 'under standing' would be more realistic if the computer were programmed to throw in the occasional sub-optimal move in order to defeat such mem ory techniques, but the best test of all would be for the subject, playing
without access to the database, to confront a strong human opponent playing with database access. The human opponent could throw in nasty tricks, such as choosing moves which require counter-intuitive re sponses in order to make progress. An ability to overcome such an op ponent would, in my view, prove a genuine understanding of 2.t v lt), Considering only optimal moves can be very misleading. As we have seen above, if Black is restricted to playing optimally, then the ending is very simple, but only considering White optimal moves has the oppo site effect - it makes an ending ap pear much more complex than it really is. In all three of my endgame books I have concluded each section with a ' longest win ' , taking the reader all the way down an optimal path. These position are often the most complex in their section, not because of their length but because of the restriction to White optimal play. Throughout all three volumes I have not felt myself restricted to giving optimal lines for the winning side (except for 'longest win' posi tions); if a simple, intuitive win was available that was slightly slower than optimal play, then I have chosen the intuitive win. Even in the 'long est win' positions I have chosen be tween White's equi-optimal moves in order to make the winning process as comprehensible as possible. How ever, while going through the 'long est win' positions it very often happened that White could win one move more slowly by a methodical,
2.i. v lLJ 2 71
intuitive method, but a tricky and highly complex method would, as if by pure chance, cut a move off the winning process. When one plays over a single optimal line of play, one is struck by how often completely baffling moves occur. However, many of the 'baffling' moves are not actually es sential parts of the winning process, but complex finesses which serve only to reduce the length of the win slightly. An important part of the hu man interpretation of endgame data bases is to distinguish between those difficult manoeuvres which are fun damental parts of the winning proc ess, and those which are mere frills. This can never be done by looking only at optimal lines of play, and I suspect that one of the reasons for the limited progress in interpreting databases is that human analysts have not broadened their scope suffi ciently when examining them. Moving on now to wider issues, one of the motivations for this series of endgame books has been to see how far one can go in understanding large endgame databases in 'human comprehensible' terms. The increas ing use of computers has left the world awash with data, which is surely accumulating in the world's data storage facilities far faster than human beings can ever examine it. Very often the raw data itself is use ful, but in most cases some general principles governing the data would be much more valuable. A good ex ample from the history of science is the solution of the problem of
planetary motion. Anyone watching the paths of the planets in the night sky can see that there is a pattern to the motion and, indeed, the ancients knew enough to predict the apparent motions of the planets correctly. Yet the real breakthrough didn't come until the late 1 6th century, when for the first time Tycho made very pre cise measurements of the planetary positions; indeed he spent most of his life compiling tables of figures for each planet. Using these tables, Kepler was able to derive his three laws of planetary motion, effectively reducing Tycho's vast mass of data into a much more easily under stood form. Yet even Kepler didn't reach the deepest level of under standing. He had found, for example, that the planets moved in ellipses, but he could provide no reason why this should be the case. It was left to Newton to penetrate one level deeper, explaining all three of Ke pler's laws in terms of a single law of gravitation. Thus in the end all of Tycho's tables could be reduced to a single sentence; the huge volume of data had been distilled into a single, powerful, universal principle. These days I suspect that scien tists are less convinced about the un derlying simplicity of nature; indeed there are certain systems which ap pear to be truly chaotic, making pre diction difficult or impossible. If nature can be chaotic, it seems even more likely that an artificial system such as chess should lack the under lying simplicity that humans are constantly searching for. Perhaps
272 2.i. v lb
one has to operate on the assumption that hidden simplicity exists, or else one has to accept that there will be some things which can never be un derstood by the organic brain. De spite this depressing thought, a glance at chess literature shows that almost all grandmasters believe there is inherent logic in chess and that there are some rules which help to find a good move. When they ana lyse or annotate a game, GMs con stantly refer to a move as 'logical' , or talk about 'the logic of the position' . Indeed some even attribute more nebulous qualities, such as 'justice' , to chess positions, for example one often reads about 'an unjustified at tack' being 'punished' . Computer databases offer a challenge and a test for this view. Some of the 'best' moves in the databases appear very hard to understand; do endings such as 'ii'+� v 'if and l:t+.i. v 2lb exceed the limit of human comprehension in chess, or is it just that we need a Kepler to uncover the underlying laws governing these endings? In one sense this question is academic, for even if these endings are even tually understood, the increasing power of computers will finally en able them to exhaustively analyse an ending so complex that no human will be able to grasp it in its entirety. In this case we will have to settle for second best; a method by which the human's performance can be im proved, even if it will never equal that of the machine. On the whole, the 1 8 months I have spent studying 5-man endgame
databases has left me feeling more optimistic about the possibilities for interpreting large endgame data bases in 'human-friendly ' terms. In the long term, the escalating com plexity of the world means that un derstanding large quantities of data will be an increasingly important matter for humanity - we should either succeed, or perhaps ultimately we will have to turn the world over to the yet-to-be-created artificial intel ligences. Some of you are probably still wondering how to win with 2.i. v lb, so here is some analysis which should help you along the way. In this ending, general concepts and plans are more important than specific variations, so we will spend some time outlining characteristic manoeuvres, using the following po sition as a concrete example.
+I-
376
Roycroft, 1986 (376): This is a typical post-KH position. Black has been forced out
2i. v lLi 273
of a corner, and now his king is on the second rank between two cor ners. White's task is to drive the en emy king onto the back rank or into a comer, while making sure that Black can never set up a KH-type position. Black's defensive formation, with the king and knight adjacent to each other on the second rank, closely resembles a true KH-position; the only difference is that Black's pieces are one square further away from the corner. The situation resembles the ending of 'ilt' v 2.t; in that case there is one true fortress, namely the Lolli position (W�f7, 'ilt'e5 v B�b7, i.b6, c6 - see diagram 420 in Secrets of Pawnless Endings for the detailed analysis). If B lack cannot achieve the Lolli position then he generally loses, but he can put up the greatest resistance by adopting what we called a pseudo-Lolli position, in which Black's pieces occupy the same relative positions as in the true Lolli position, but further away from the comer (e.g. king on c7, and bish ops on c6 and d6). By using the extra space available on the queenside, White is eventually able to break down such a position and win (see diagram 427 in Secrets ofPawnless Endings). The situation is similar with KH-type positions. The genu ine KH-position in the corner is not actually a draw, but it offers the greatest possibilities for resistance. Once Black has been forced to abandon this defensive formation, his next best chance is to set up a position such as diagram 376, which we call by analogy a pseudo-KH
position. This is easier to break down than the genuine KH position. Diagram 2 is important for the general winning process in 2i. v lLi. The white king has advanced as far as possible, but at the moment no di rect progress can be made. Black simply oscillates with his knight be tween f2 and g4, and if White meets ...lLig4 by �e4, then Black replies ...00+. If White checks on b5, then Black's king moves to f3; check again, and the king returns to e2. The difference between diagram 376 and diagram 375 is the extra space avail able on the kingside in diagram 376. White can exploit this with the typi cal manoeuvre i.e8-h5 . In diagram 375, the black knight and king com bined to cover all the squares on the a5-d8 diagonal, but here the extra square on the h-file is available for a bishop check, and this forces Black to abandon his defensive formation. We will call a manoeuvre of the type i.e8-h5 a mirror manoeuvre, because the bishop's path resembles that of a ray of light bouncing off a mirror (in this case the edge of the board). Mirror manoeuvres are fun damental to the whole ending of a v lLi. It often happens that the bish ops are operating on parallel diago nals, as in diagram 376, and then one of them undertakes a mirror ma noeuvre which results in the bishops' diagonals crossing like an X. If this manoeuvre can be carried out with gain of tempo, so much the better. Mirror manoeuvres also explain why the bishops are typically posted at a moderate distance from the enemy
274 2.i.
v
lLi
pieces. Looking at diagram 376, if the bishops were closer to the black men, then they would be within the knight's sphere of influence, and might be restricted by having vital squares guarded by the black knight. So the bishops are better posted fur ther away. But they mustn't be too far away, because then mirror ma noeuvres would not be possible. For example, if the bishop were on b7 and not c6, White certainly could not transfer it to h5 . In diagram 376 the bishops are optimally posted; both bishops have the chance to reflect off either the top or left edges of the board - flexibility is the key when it comes to bishop position. Now let's take a look at the de tailed analysis to see how similar ideas arise again and again. We are assuming that Black is to play. There are three significant lines, but in all cases White responds with the same manoeuvre; this is one reason why diagram 376 is so significant White actually has a threat (i.e. i.e8h5), and whatever Black does he cannot nullify it. lLig4 1 The other two main lines are: 1) 1. lLih3 2 i.e8 (Black cannot achieve a KH-type position in the g2-corner because the square f4 is firmly under White's control, so Black's knight is cut off from reach ing g2) lLig5 (2 ... �f3 3 i.h5+ 'it>f2 4 i.g4 lLig5 5 'it>e5 lLif3+ 6 �f4 lLih2 7 i.h5 lLifl 8 i.d6 wins as after move 6 in line 2) 3 i.h5+ �f2 4 �e5 (the plan is to play �f5-f4 and then i.b6+, constricting Black' s king even ..
more) �g3 5 �f5+ (now Black's king is driven to the edge of the board) �h4 6 i.e8 lLih3 7 i.b6 (the immediate mirror manoeuvre 7 i.a5 is equi-optimal, but this move gives Black less choice) �g3 8 i.c6 with the branch: l a) 8 lLif2 9 i.c7+ �h4 1 0 i.a5 (now White executes his mirror ma noeuvre; the threat is i.el , and Black cannot defend by ...�g3) �h3 1 1 i.d5 (this is simply a waiting move; ...�g3 and ... �h4 both lose the knight after i.el , ... �h2 has a tactical flaw, and most knight moves are met by �f4) and now: lal) 11 �h2 1 2 �f4 lLid3+ 1 3 �e3 ( ... lLie5 is impossible thanks to the position of Black's king) lLic5 14 i.b4 lLid7 15 i.e6 and quickly rounds up the knight. la2) 11 lLig4 12 �f4 lLih2 (after 1 2 ... lLif2 we transpose to line l a4) 13 i.el ein 14 �f3 lbh2+ 15 �f2 lLig4+ 16 'it>gl and wins. l a3) 11 lLidl 1 2 �f4 lLif2 1 3 �f3 is line l a4. l a4) 11 lLid3 12 i.c4 lLif2 1 3 �f4 �g2 1 4 i.d5+ �h3 1 5 �f3 (not quickest, but simplest) lLid3 1 6 i.c7 (taking the e5-square away from the knight) lLie 1 + 17 �e2 lLic2 1 8 i.e5 �g4 19 i.e4 lLib4 20 i.d6 and wins. 1 b) 8 . 'it>h4 9 i.d8+ �g3 10 i.a5 (the mirror manoeuvre strikes again) lLigl 1 1 i.el + �h3 (Black's king is forced to the edge of the board) 1 2 i.e4 lLie2 1 3 i.d3 lLig3+ ( 1 3. . .lLid4+ 14 �e4 lLie6 15 i.c4 lbf8 1 6 �f5 will quickly trap the knight) 14 �f4 lbh5+ 15 'it>f3 lbg3 (a valiant try for stalemate, but Black can't last long •••
•..
•••
..•
•.•
. .
2.t
when his king is pinned to the edge) 1 6 .i.c4 (a waiting move) lt:lht 17 .te6+ �h2 18 .tc3 �gl 1 9 .i.d4+ 'iPh2 20 .i.e3 winning. 2) 1 lLJdl 2 .i.e8 lt:le3 (this is a moment when precision is required, because Black is threatening a KH set-up with ... lt:lg2 and ...�f2) 3 �e4 (not 3 .i.h5+ �f2. setting the clock back 28 moves) lt:lfl (Black has to abandon his hopes of reaching a KH type position since 3 ...�f2 loses the knight and 3 ... lt:lg2 4 .i.b6 cuts Black's king off from the g2-comer) 4 .i.h5+ �f2 5 �f4 �el (trying to avoid being forced into the comer, but the respite is very brief) 6 .i.d6 (such a move may appear mysteri ous, but actually the reason behind it is rather simple; White threatens 7 �f3 lt:ld2+ 8 'iPg2 and Black's knight cannot move so as to cover b4) 'it>f2 7 .i.c5+ �g2 (7 ...�el 8 �f3) 8 .i.f3+ �h3 (8 ... �h2 is also met by 9 .i.d5 �h3 10 .tb4 lt:lh2 1 1 .i.el) 9 .i.d5 (once again White sets up a characteristic piece formation; his bishops stand flexibly next to each other, while his king is immune to knight checks) lt:lh2 1 0 .tb4 (g2 is covered, so there is no danger of Black's king escaping that way; this means that the other bishop is free for a mirror manoeuvre to el in order to prevent the king slipping out via h4) lt:lfl 1 1 .i.el lt:lh2. Now White must take care not to allow the black knight to escape via g4 and f6. To achieve this, White's bishop must be ready to meet ...lt:lf6 by a check on the c8-h3 diagonal. This is possible with the bishop on d5 or b7, but not •••
v
liJ 275
with the bishop on c6. Thus the right waiting move is 1 2 .tb7, winning after 12 ... lt:lfl 13 �f3 (the decisive manoeuvre) liJh2+ 14 �f2 lt:lg4+ 1 5 �gl and wins. 2 .te8 White plays his mirror manoeu vre in any case. Black cannot reply 2 ...�f3 because he would lose the knight instantly after 3 .i.h5. 2 � A concession; White can improve his king position. 3 �d3 lLlt'6 If Black simply waits, White will gradually improve the position of his bishops, for example 3 ... �fl 4 .i.c6 �f2 5 .tf4 �fl 6 .i.g3 �gl 7 �e2 and Black is boxed into the comer. 4 .tc6 White restores the initial position of his bishops, but having driven Black's king to an inferior square. Note that there is no route by which Black's knight can reach g2, since even . . .lt:lg4-e3 loses the knight to .tb6. liJh7 4 It is curious that this counter-in tuitive move lasts longest. More natural defences are refuted as fol lows: 1) 4 lt:lg4 5 .i.f4 (now Black has to make a concession, either to re treat his king to the first rank, allow ing .i.g3, or to play ...lt:lf6) lt:lf6 6 .i.g5 (in this case the short-range mirror manoeuvre comes with gain of tempo) lt:lg4 7 .th4+ (forcing the king to the edge of the board) �fl 8 .i.g3 �gl 9 �e2 (the king closes in for the kill) lt:le5 10 .i.d5 lt:lg4 1 1 •••
.••
•••
276 2j. v lLl
j.b4 lLle5 12 j.f6 lLlg4 1 3 j.d4+ 'iPh2 14 'iii>f3 and wins. 2) 4...lLlh5 5 j.e5 lLlg3 (this is forced or else the knight is trapped immediately, but now Black's king is once again driven to the edge) 6 j.d4+ 'iii>f l 7 i.f6 (another mirror manoeuvre, exploiting the fact that . . .'iii> f2 is impossible) lLlh5 8 j.b4 lbf4+ 9 'iPe4 lLlh3 (9 ... lLlg2 10 j.b5+ 'iii> g l 1 1 j.g3 is total paralysis) 10 'iii>e 3 lbgl 1 1 j.d5 (another typical move; d5 and b7 are good squares, but c6 is bad since ... lbe2-c3 covers b5) lbh3 1 2 'iii>f3 lbgl+ 1 3 'iii>g 3 lbe2+ 14 'iii>h2 and wins. 5 j.dS This move not only restricts Black's knight, it also angles for a mirror manoeuvre to h4. If this is completed, Black's king will be forced to the edge of the board and will be hopelessly cut off from the knight, so the reply is forced. 'iii>g3 5 6 'iii>e4 Threatening 'iii>f5, so the reply is again forced. 'iii>g4 6 7 j.d7+ All perfectly normal moves. If Black is to save his knight, the king has to move to the edge. 7 'iii>h5 8 'iii>f5 'iii>h6 Or 8 . . .lbf8 9 i.e8+ 'iii>h6 1 0 i.f6 'iii> h7 1 1 j.d4 (this is the quickest win, but the precise choice of d4 is not an essential part of the winning process; b2 would be just as good) 'iPg8 ( 1 1 . ..'iii>h6 12 'iii>f6 lLlh7+ 1 3 'iii> f7 lL!g5+ 1 4 'iii>g8 wins) 1 2 'iii>f6 ••.
.•.
••.
lLlh7+ 13 'iii>e7 lLlg5 14 j.c6 'iii>h7 15 'iii>f6 lLlh3 (if White had played l l j.b2, for example, it would now take him slightly longer to round up the knight, but of course the win is trivial in any case) 16 j.e3 and wins. 9 i.e7 The bishops again stand on adja cent squares; Black's knight is tem porarily movebound. 'iii>g7 9 . 10 j.e6 'iii>h6 Or 10 ... lLlf8 1 1 j.f6+ 'iii>h6 1 2 j.c4 and after any reply apart from 1 2...'iii>h5 (which allows mate in two) and 1 2 ... lbe6 (which puts the knight en prise) White will continue 1 3 j.d4 with play virtually identical to the main line. 'iPg7 11 j.c5 White wants to play his bishop to d4, but first he needs a waiting move to help force Black's king back to h6. 12 j.c4 lLllS 12 ...'iii>h6 1 3 j.d4 lbf8 transposes. 13 j.d4+ 'iii>h6 Once again the bishops stand on adjacent squares. The bishop on d4 is sealing off the black king's escape via g7, so the other bishop is free to move. 14 j.b5 This final mirror manoeuvre takes the bishop to e8, cutting off the black king's other exit at h5. 14 lLJh7 15 j.eS The position is identical to that af ter 1 1 ...lbh2 in line 2 of the note to Black's first move. lbf8 15 lLlh7+ 16 'iii>f6 ..
••.
2i. v � 277
17 �f7 18 �g8 and wins.
�g5+
Now we will return to the KH po sition for a more complete analysis.
• • m • ••• • • • • • • • • • • i. . • • • • • • • • • • • • • • • +I377 Kling and Horwitz, 1851
(377): White to play can win the knight or force mate in 45 moves. In the first phase, White undermines the KH set-up in the b7-corner and drives the black king towards g7 . There then follows an amorphous phase in which White gradually drives the black king towards the edge of the board. The key idea is that White should operate with his bishops from a 'moderate' distance. This concept arose in the previous diagram, where the motivation was to ensure the availability of mirror manoeuvres. The same idea is im portant in the following analysis, but the motivation is rather different. Let us suppose that Black's is on e6 and White's king is on c5. Then the dark squared bishop may well be best
posted on c3. The reason is that Black's king may make a run for the b7-corner or the g2-corner. In this case White may need to switch his bishop to a5 or el respectively to cut Black's king off from the corner he is aiming for. If the bishop is further away, for example on b2, then it has much less flexibility to switch onto parallel diagonals. d4 would be an equally poor square; the white king would obstruct the bishop, and it would be in range of the enemy knight. So the general rule is that the bishops should be far enough away to be out of knight range, but near enough to the centre of the board to ensure maximum flexibility. A second important idea is that of preventing tricky pseudo-KH posi tions. The one with the black king on e3 and knight on f3 (i.e. a diagonal shift of one square from the true KH position) is particularly awkward and is definitely to be avoided. �8 1 i.b4 After l ...� 2 i.c3 �b7 3 i.d4+ �c7 or l ...�c7 2 i.c3 �b6 3 i.d4+
A)
4 5 �b4
�6+ �b7
278 2-'l
v
lLJ
�b6 6 .teS+ 7 .tg3 This waiting move is a good ex ample of the 'moderate distance' principle. After 7 .ltf4 lLid8 8 .te3+ �c7 9 �b5 lLie6 there is no check on the h2-b8 diagonal (the bishop is within range of the knight), while af ter 7 .th2 lLid8 8 .tgl+ �c7 9 �b5 �d6 10 .th2+ �e7 there is no check on the h4-d8 diagonal (the bishop is restricted by the edge of the board). 7 lDd8 �c7 8 .tn+ 9 �bS transposes to line C. •••
B)
4 �d6 5 .tb6 Now there is no way back to the b7-corner, and the black king has to head for open spaces. 5 �e6 Or 5 . . . �e5 6 .tc7+ �e6 7 'itd4 lLid6 8 .tb3+ �d7 9 .ta5 lLib7 10 .ta4+ �e6 11 .tb4 lLid6 1 2 .tb3+ �d7 1 3 .td5 (this covers b7, and so prevents Black setting up another KH-type position in the b7-corner) ri;c7 14 .ta5+ �d7 15 �e5 lLib5 1 6 .tf3 (not the fastest win, but simplest given that we already know diagram 376) lLid6 17 �d5 lLie8 1 8 .tel lLic7+ 19 �e5 lLib5 20 .tb4 lLic7 2 1 .t f'2 i s diagram 376. 6 .tb3 6 �d6 Alternatively: 1) 6...'iteS 7 .tc7+ �e6 8 �b5+ �d7 9 .th2 (in this case the most distant square is best for the concrete ..•
•••
••.
• • • • ••• • • m ••• • • • • • rJ rJ • • •.t• • • • • • • • • • • 378
+!Position after 6 .tb3
reason that after 9 .tf4 lLid8 or 9 .tg3 lLid6+ White cannot play 10 �c5) lLid8 10 'itc5 lLie6+ 1 1 �d5 lLic7+ 1 2 ri;e5 �c6 13 .tgl �b5 14 .ltf7 lLia6 15 .tf2 (not the fastest, but once again White is aiming for dia gram 376) lLib4 ( 1 5 . . .lLic5 1 6 �d4 �c6 1 7 .td5+ 'itb5 1 8 .tf3 lLid7 1 9 .th4 lLib6 20 .tg3 and we have dia gram 376) 16 �d4 lLic6+ 17 'itc3 lLie7 1 8 .te8+ �a6 1 9 'itc4 �b7 20 .tg3 'itb6 2 1 .th4 lLic6 22 �d5 lLia7 23 .tf2+ ri;b7 24 ri;c5 ri;c7 25 .ta4 and the position is virtually identical to that after 5 'itf4 in line 2 of the note to Black's first move in diagram 376. 2) 6 lDd6+ 7 �c5+ 'itd7 8 'itd5 (a very favourable position) lLie8 9 .td4 lLic7+ 10 �e5 lLia6 1 1 .ta4+ 'itc7 1 2 'itd5 lLib4+ 1 3 �c5 lLia6+ 14 ri;b5 �b7 15 .tb3 lLlc7+ 16 'itc5 lLia6+ 17 �d6 lLic7 1 8 .tc4 and the enemy king is forced to the edge of the board. 7 �bS 'itd7 8 .te3 •.•
2.t v llJ 279
Here 8 .tf2 is bad because of the fork on e4 after 8 . . .l2Jd6+, while gl breaks the 'moderate distance' rule. lLJd6+ 8 9 �cS lLib7+ The main line continues 10 ..ti>b6 l2Jd6 1 1 .tf4 l2Je4 12 .th2 (a waiting move which keeps the bishop out of the black king's range) �e7 1 3 �c6 l2Jd2 14 .tc2 l2Jf3 15 .td6+! ..ti>e6 1 6 .tb3+ �f5 17 ..ti>d5 l2Jg5 1 8 .tc2+ �g4 1 9 .tc7 (the moderate distance principle in action) �f3 20 ..ti>e5 (note that 20 .te5 is very inaccurate; after 20 . . . r:j;e3 2 1 .tf6 lLif3 Black sets up a tough pseudo-KH position) ..ti>e2 (20 ... �e3 2 1 .tb6+ �d2 22 .tf5 l2Jf3+ 23 �f4 lLiel 24 .ta5+ forces the king to the edge) 2 1 .ta4 (not 2 1 .tb3 liJf3+ 22 �f4 lLid2 and there is no check on the a6-f1 diago nal) �d3 (21 ... l2Jf3+ 22 �f4 lLid2 23 .tb5+) 22 .tb5+ �e3 23 .tb6+ �f3 24 �f5 lLih3 25 .te8 (a mirror ma noeuvre trapping Black's king in the g2-corner) lLif2 26 .th5+ 'ii?g 3 27 .tc7+ ..ti>h3 28 �g5 �g2 29 �f4 lLih3+ 30 ..ti>e3 with the type of cor ner position we have seen before. •••
C)
4 lLJds 5 .tf2 Not 5 .te3 lLie6 and there is no check on f4. 5 lLib7 r:j;b6 6 .tg3+ 7 �b4 l2Jd8 <3;c7 8 .tf2+ 9 �bS A critical position. If Black tries to maintain his KH set-up then his ..•
. - . . � � � � � � � � � :$/), • • • • -�· . . i.. • • • • • • • • • m • • • • • 379
+/Position after 9 r:j;bS
king is driven to the edge of the board and he loses quickly, for ex ample 9 ... l2Jb7 10 .tg3+ ..ti>c8 1 1 'it?b6 (now we are already in the later stages of the winning process) liJd8 1 2 ..ti>c5 lLib7+ 1 3 �d5 l2Jd8 14 .tf4 (this is the most accurate square, as we shall see later, but the precise lo cation of the bishop isn't really criti cal for the winning process) ..ti>b7 1 5 ..ti>d6 ..ti>a6 1 6 .tc2 �b6 17 .ie3+ ..ti>b5 18 .td3+ �b4 19 .te4 lLif7+ 20 ..ti>e6 (now we see why f4 was the best square for the bishop on the h2b8 diagonal) l2Jd8+ 2 1 'it?e7 and wins the knight. lLie6 9 ... After 9 . . . �d6, Black's king is forced back more quickly, for ex ample 10 .tg3+ ..ti>e6 ( 1 0... ..ti>e7 1 1 .th4+ ..ti>d7 1 2 ..ti>c5+ �c8 1 3 .tg3) 1 1 �c5 lLif7 12 .tb3+ �f6 13 �d5 'it?f5 14 .tc7 l2Jh6 15 .tc2+ ..ti>g5 1 6 �e5 l2Jg4+ 1 7 �e4 lLif6+ 1 8 ..ti>f3 transposing to the position after 17 �c3 in line 1 of the note to Black's move in variation B .
280 2.i. v �
10 11 12 13
i. g3+ i.dl 'iPcS i.eS
'iPd7 llXl4+ tLJrs 'iPe6
1 ) 16 'iPe6 17 i.b7 'iPf7 (the line 17 ...'iPf5 18 'iPd4 lLig4 19 i.c8+ 'iPf4 20 i.d7 lLih2 2 1 i.d2+ 'iPf3 22 i.a5 lLig4 23 i.c6+ 'iPe2 24 i.b6 lLif2 25 i.c7 also transposes into diagram 376) 1 8 i.f3 'iPg6 19 i.e4+ lLif5 20 'iPd5 'iPg5 21 'iPe5 lbh4 22 i.c2 lLig6+ (22 ... 'iPg4 23 i.d l + 'iPg3 24 i.e 1 + 'iPh3 25 i.h5 lLig2 26 i.f2 with a quick win) 23 'iPe4 lbe7 24 i.a4 lLig6 25 i.b3 reaches diagram 376. 2) 16 ..lLin 11 'iPd5 lLig3 1 8 i.d7+ �f4 1 9 i.d2+ 'iPf3 2 0 'iPe5 lLie4 2 1 i.a5 lLif2 22 i.c6+ 'iPe2 23 'iPd4 lLig4 24 i.d8 lbf2 25 i.c7 is diagram 376 again. •••
.
380 +/Position after 13... 'iPe6
14 i.c3 An interesting example of the 'moderate distance' principle. The concrete reason why 14 i.b2 is infe rior is rather deep: after 14 ... lLie3 1 5 i.f3 'iPf5 1 6 i.c6 'iPe6 1 7 i.b7 'iPf7 1 8 i.f3 'iPg6 1 9 i.e4+ lLif5 20 'iPd5 'iPg5 2 1 'iPe5 lbh4 22 i.c2 'iPg4 23 i.d 1 + 'iPg3 there is no check on the e l-h4 diagonal. To see this in a game would be extremely difficult, but it shows how distilling general princi ples from a database may be more helpful than a thousand optimal vari ations. lbe3 14 ... 'iPfS 15 i.f3 16 i.c6 The same principle again. After this Black's king is soon forced back to the second rank and diagram 376 is not far away, for example:
Here is a 'longest win' situation in 2i. v lb (without detailed analysis):
+I-
381
Thompson, 1983 (381): One optimal line runs 1 .. .'iPf3 2 i.f8 'iPf2 3 i.c5+ ! 'iPf3 4 �b7 'iPg3 5 i.g l (now White must bring his king up to free the h l bishop) 'iPf3 6 'iPc6 'iPg3 7 'iPd5 'iPf3 8 'iPd4 'iPg3 9 'iPd3 'iPf3 10 'iPd2 'iPg3
2� v lb 281
1 1 �e2 lLif4+ 1 2 'it>n lLie6 1 3 �d5 lLif4 14 �e4 lLih5 15 �e2 �f4 16 �d3 lLif6 17 �h2+ �g5 1 8 �g2 lLih5 19 �d4 lLif6 20 �h3 lLih5 2 1 'it>e5 lLig7 (Black aims for the KH set-up) 22 �g2 �g6 23 �g3 lLih5 24 �h4 lLig7 25 �e4+ (this is the posi tion after 3 �d4+ in the previous diagram) �f7 26 'it>f4 lbe6+ 27 �g4 lLig7 28 �d5+ �g6 29 �b3 lbe8 30 �c2+ �f7 3 1 'it>g5 lLid6 32 �b3+ �e7 33 �e l lLie4+ 34 �f5 lLic5 35 �d5 'it>d6 36 �f3 lLid3 37 �c3 d6 39 �g7 c5 �a6 55 �d7 �b7 56 �d6 �a6 57 �f5 �b7 58 �f4 lLic8 59 �e4+ 'it>a6 60 �g5 lLia7 61 �d8 lLic8 62 'it>c6 lLia7+ 63 �c7 lLib5+ 64 �b8 lLid6 65 �d3+ lLib5 66 �c8 and takes the knight next move.
We end with the unique reciprocal zugzwang in the ending of 2� v lb, and (for once) a simple piece of analysis:
=I-
382
Original (382): Black to play must allow immediate mate. With White to move, 1 �g8 ( 1 �c8 lLie7+ ! 2 �c7 lLid5+) lLie7 ! 2 �c4 lLic6! either wins the b8-bishop or, after 3 �d5, gives rise to stalemate.
Index of Players and Composers The index covers all three volmes of this endgame series. Each name is followed by all the references from the first volume (Rook End· ings}, then all the references from the second volume (Pawnless Endings) am: finally those from the current volume. Each group is preceded by a code letter, as follows: R = Secrets of Rook Endings P = Secrets of Pawnless Endings M = Secrets of Minor Piece Endings The numbers are those of the diagrams. R95 Adamski P274 Agdestein R24 Ahrend P387, 399 Akerblom R104, 159 Alburt R109, M28 Alekhine M330 Alexandrov M3 14 Allan R344 Al Mokhtar R388 Alzate P l 6, 52, 406 Amelung R193, M144 Amirian P324 Archakov P308, 309 Arkell,K. M50 Aronin P245, 253 Assalini M1 15 Atanov Averbakh R23, Pl5, M l7, 25, 45, 75, 1 1 1, 1 12, 1 33, 1 36, 142, 149, 1 50, 1 5 1 , 1 52, 160, 1 6 1 , 162, 1 67, 176, 177, 178, 179, 1 90, 2 12, 282, 283, 291, 293, 297, 297, 3 1 8, 322, 324, 3 3 1 , 34 1 , 342, 343, 344, 345, 348, 358 M209 Babitch
Barbero P18 Barcza M19 Barlov R176, M305 Barry R145 Barus R388 Beliavsky M227, 285 Belli P76 Benjamin P24, 44 Benko R80, 358, P242 P32 Bent Berelovic M36 Berg R268, M159 Berger R9, P3, 64, 70, 7 1 , 92, 246, 340, M208, 240 Berner R32 Bernschutz R344 Bianchetti P322 Bisguier P88 Blandford M366 Blees R268 Boersma R364 Bogolyubov R l 42 Bolland R478 Bondarenko P13 Borge M173 Borisova M248
Index of Players and Composers 283
Botsari M274 M301 Botsaris M29 Botvinnik Brenev M329 P285, M 1 38 Brito Brodsky R47, 1 36 P109, 1 16, M288 Bron R1 15, 1 19, 492, P3 18 Bronstein P27 1 Budnikov R261 Burliaev R497 Burn M217 Buzandian M246 Capablanca P53 Carlsson P193, 240, 24 1, 248, Centurini 250, M21 1 , 216, 23 1 , 232, 234, 256, 257, 284 R 176 Chandler Chebotarev R63 Chekhover P95, 97, 98, 102, 1 30, 1 3 1 , 1 38, M l70, 1 7 1 Chernev M7 P304, 3 1 2, M309 Chemin M3 1 5 Chernous Cberon R 16, 20, 28, 30, 74, 78, 87, 90, 121, 1 22, 129, 1 30, 1 33, 147, 1 48, 1 5 1 , 1 52, 157, 163, 165, 166, 167, 168, 207, 218, 220, 230, 241 , 242, 243, 25 1 , 252, 260, 262, 263, 273, 274, 29 1, 295, 299, 301 , 3 10, 3 1 1 , 342, 347, 376, 380, 38 1, 382, 383, 384, 385, 396, 399, 400, 40 1, 4 17, 4 19, 459, 463, 464, 465, 477, 48 1, 482, 483, 485, 493, 494, 495, 496, 498, 502, 503, 504, 507, 5 13, 522, 523, 528, 529, P283, M l 8 76, 85, 1 58, 168, 169, 1 80, 228, 236, 238, 264 P22 Claesen R15 Clausen
Clement R302 Chigorin R93 P25 Comay Cools M187 Cooper, L. M84 Cvetkov R92 M248 Dahl P309 Daly Dedrle R l89, P421 , 422, 423 De Firmian R379 Dehler P36, 48, 108, M236 P75 Delaney P21 Delithanasis P77 Diaz, J. Diez del Corral P1 1 2 Dimentberg M203 P213, 344, 345, Dobrescu 346, 347, 348, 355, 356, 358, 359, 361, 363, 364, 365, 366, 370, 37 1 , 372, 373, 374, 375, 376, 389, 395, 396, 398, 400 401, 402, M99, 204, 205, 259 Dolgov P394 P285 Doncevic Donner R48 Dorfman M l 14 R89 Dvoirjs Dzindzihashvili R296 R423, P3 1 3 Ehlvest Eingorn M305 Emms Rl45 Enevoldsen, J. P50 Ermenkov M244 P273 Ernst P273 Espig Estorch M20 R50 Estrin R109, 478, P68, 73 Euwe Ml31 Falk R 1 37 Farago Fedorov M l73
284 Index ofPlayers and Composers
Fila R 1 64, 283, 3 10 M368 Filaretov R289 Filipowicz R198, 275, 284 Fine R80, 246, M310 Fischer M250 Franklin Frias R159 P320 Frie M61 Frolov R49, M285 Forintos M43 Fritz P20 Ftacnik P l 06, M294, 355, Galushko 369 R53, 54, 59, 60, 61 Ganshin R7 1 Garcia, G. Garcia Palermo R4 16 P28 Gasparian Gayson P308 Gazoni P56 R l l 1, 1 35 Gelfand Geller P19 Georgiev, Kir. P45, R363, P82 R186 Gerasimov P22 Geselschap Gheorghiu P86 Gillen M48 P23 Gipslis M63 Gligoric M98 Goldenov R210 Gorgiev Gosh P23 P133 Graetzer R l 9, 52, 83, 107, Grigoriev 1 17, 1 9 1 , 232, 236, 240, 244, 248, 250, 27 1, 277, 28 1, 282, 431, 45 1 , M214, 2 15, 24 1, 253 Gufeld R1 15 Gulaev M 1 30 Gunst R205 Gurgenidze R195, 330, 445, P32 1 , 4 15, 450, M254
M 1 84 Gurvitch Gutierrez P76 Gutman R158 Halberstadt P47, 1 9 1 , 203, 207, 2 1 8, 350, 384, M4, 65, 7 1 , 83, 90, 91, 93, 103, 105, 106, 145, 146, 191, 192, 1 93, 194, 1 97, 203, 280 R354 Halurnbirek M159 Handoko Hanneman M81 Harding M48 P263 Ha�ek R106 Healey P314 Hebden R7 1 H�bert M19 Hecht Heller P 1 20 P3 1 3 Hellers P349 Henneberger R466, M260, 279 Herberg P33 Herlin Hodgson R407, P274 Holm M 1 24 HOlzl P18 Hort M27 Horwitz R457, 458, 472, P2, 54, 55, 107, 1 87, M 1 1 3, 1 83, 278, 321 , 337, 375, 377 P89 Hurelbator Iglicky R474 R104 Inkiov R228, P40 Isenegger R379 lvanovic lvkov M247 M247 Jano§evic M246 Janowski P298 Johnstone M69 Joita P124 Joseph P103 Kabiev P326 Kalandadze
Index ofPlayers and Composers 285
M 1 74 P84 M98 Kan Karaldaic R92 Karpov P20 Karstedt R77, P41 1 Kasparian R21 1 , P27, 388 Kasparov P85 R68, 1 39 Katsnelson Kazantsev Ml32 R437 Keres Khenkin M322 R47 Khmelnitsky P443 Khortov Khuzman P46 King R l l l , 1 35 Kiselev P3 10 Kleiman R190 Kleindinst P125 Kling R457, 458, 472, P2, 54, 55, 107, 187, 278, 280, 28 1, 282, M40, 1 83, 278, 321, 375, 377 P74, 3 1 9 Knefovic Koen P302, M274 Kohlweyer P3 15 Komljenovic R158 Kondratiev R l 16, 438, P183, M36 1 Kopaev R2 1, 194, 469, 473, 476, 490 Kopelovic M21 3 Kopnin R l 16, M36 1 Koppelomiiki P189 Koranyi P357, 454, M210, 221 , 338 Korchnoi P85, M163 Korolkov P382 Korteling M5 1 P262 Kos Ko�ek M 153, 154, 155, 1 8 1 , 1 95 Kalinin Kamsky
Kotronias Kovalenko Kovalev Kozirev Kriihenbiihl Kralin Krasenkov Kremenietsky Krikheli Kruszynski Ksieski Kubbel Kuiper Kupreichik Kurajica Kuriatnikov Kurtenkov Kuzmin, G. Kuznetsov Larsen Lasker, Ed. Lazarevic Lein Lengyel Leow Lerner Levenfish Levy Liberzon Liburkin Lipnitsky Livshits Ljubojevic Lobron Lolli Loyd, S. Lurye Madi Maestre Magomedov Maizelis Makarev
M227 P42, M292 R89 P197 P299 M6 R290 M77 R298 R289, 397 R95 R440 P278, 280, 28 1, 282 P3 12 M 1 38, 268 P446 P306 R37 1 P325 R258 R339 M26 R361 P149 P75 R8 1, 406 R479 P149 M285, 35 1 P49 M269 P25 1 R423, P84 P301 P276, 294, 420 M 1 26 M73 P302 P37 P3 10 R359 P122
286 Index ofPlayers and Composers
R406 Makarychev R439 Makletsov MaksimovskikhP202 R279 Malaniuk M4 1 Mamedov M357 MandelOl R201, 226, 35 1 , Mandler 433, P9 P20 1 , 225 Maniakin P I 15, 148, 159, Mann 1 60, 190, 194, 214 P303 Manor P199, 227 Mansarliisky R144, 492 Marjanovic M268 Markland P3 1 6 Marovic R82, P3 14 Martinovsky R361 Matsukevich R4 1 6 Meister P3 1 7 Mesiarik P4 1 Mikhailov M9 Mikhalevski R358 Miles M244 Milev M30 Minero R33, 69, 328, 34 1 , Minev 435, 460, 462, 464, 465 M235 Mishuchkov P58, 59, 188, M360, Missiaen 370 M73 Mitrofanov P38 Molien R l 7, 18, P4, M346 Moravec Moreno Ramos P354, 369 M84 Morris P53, M201 Mugnos Miiller, Ka. P90 Nadareishvili R2 12, P4 1 2 R94 Najdorf PlOl Neistadt M3 1 7 Nemet P397, M66, 1 17 Nestorescu
P17 Neumann P204 Neustadtl Neverov R136 Nijboer R247 Nikolaidis M301 Nikolic, P. R105 P300 Ninov Norsia M 198 P27 1 Novik Novikov P305 Nunn P172, 173, 174, 175, 176, 177, 178, 1 79, 1 80, 1 8 1 , 1 82, 1 84, 185, 264, 33 1 , M3 14 M235 Obukhov Obukhovsky R50 M 1 87 O'Cinneide P89 O'Donnell Olafsson, F. M63 P45, 298, 3 1 1 Olafsson, H . P39, 447, M42 Olmutsky M323 Onate P 140, 343 Olympiev P3 15 Ostojic P21 Pandavos M1 18 Pankov Paoli M33 Papp M306 Penrose M250 P299 Pergericht P216, 256, M50 Petrov Petrovic M33 1 P3 1 1 Petursson P12 Pevit R374 Pfleger R468, P63, 275 Philidor M233 Pierce P300 Piket, J . P5 1 Platov, V. Playa M 1 37 M34 Podgaets M306 Pogats
Index ofPlayers and Composers 287
R352, 44 1 , P29, 205, Ruge Pogosiants 2 1 7, 239, 244, 247, 323, 368, Rumiantsev Ruszcynski 4 1 3, 448, M54, 125, 350, 37 1 Polak P25 1 Sackmann Polgar, J. Salov P305 R296 Salvio Polugaevsky R7, P93 Ponziani Salwe Popovic Rl25 Saren R 1 37, M27, 237 Sarkissian Portisch, L. Sax P425 Pospi�il R 1 77, 1 8 1, P137, Schaffgotsch Proke� Scheve M286 Pl39, 1 7 1 , 209, Schindbleck Prokop Schlechter M 1 1 9, 1 82 Proskurovsky R467 Schneider Schostak Przepiorka R70 Segal M61 , 79 Psakhis Sehner Purtov M9 Seirawan R237 Rabinovich R3 1 0 Selesniev Ratner Sellos M317 Rautestreich Selman Ravi R407 Sermek Ravn M252 Razuvaev Sevitov M77 Seyboth R364 Ree Reichhelm Shaigarovsky P105 R208, 22 1, P14, R�ti Sherwin Shirazi M 1 2, 1 3 Shmuter Rezvov M3 1 5 R204, M3 Short Richter Silaev R23 l , P26, 1 10, Rinck 1 1 1 , 1 17, 1 1 8, 126, 127, 1 32, Simagin Sindler 1 35, 1 36, 206, 215, 220, 221, Smagin 222, 223, 237, 238, 35 1 , 352, Smirin 353, 380, 38 1 , 385, 386, 39 1 , Smit 392, 407, 408, 4 16, 424, 426, Smyslov 44 1 , 442, 445 Sokolsky P342 Roche Rodriguez, A. P19, 307 Spasov Stefansson Romanovsky R35 Stein M372 Rossi P452, M376 Steinitz Roycroft
M30 P 1 2 1 , 390 P119 P60, 1 1 3 R279, P83 R 179 R93, 497 Pl34 M222 R345, M309 P96 M20 M296 R357 R278 M 1 16 R302 R278 P82 M70 P304 M35 M79 M303 R22, 29, 85 M5 R246 R82 P46 P83 M239, 242 R2 1 6, M29 M 156 P307 M273 P104 R48, 479, P3 1 8 M269 P301 P90 M 1 14 R 1 60, P17
288 Index ofPlayers and Composers
M95 Stolk R49, P74, M 1 63 Suetin Sunthornpongsathorn P303 P202 Supletsov R290, 37 1 Sveshnikov P279 Szen M94, 202 Szulc R258, M3 10 Taimanov R288, M34, 237 Tai R247 Tangborn M243 Taniev P45 1 Tarasiuk R2 16 Tarason R105 Tarjan R245, 259, 357, 454 Tarrasch P339, M381 Thompson R144, P377, M27 1 , Timman 334 M68, 127 Tjavlovski P199 Tkachenko P29, M87 Tolstoi P243 Topcheev R94 Trifunovic R84, P1 28, 129, Troitsky 1 4 1 , 142, 224, 409, M367 M80 Truliko Tseitlin, M.S. P61 R345, P44, 306 Tseshkovsky P1 83, M2 18 Umnov P3 19 Vaganian M273 Vaiser P3 1 7 Vaisman M252 Vaitonis R l 1 , 26, 34, 5 1 , 335 Vancura Van den Bosch M28
Van den Ende M60 Van der Weide P3 16 P99, 100, 1 14, 123, Vandiest 143, 144, 145, 146, 147, 170, 198, 208, 210, 2 1 2 Van Riemsdijk R363 Vescovi M 1 37 Vidmar R 142 Vistaneckis R261 Vlasenko P438, M87 Voja M66, 356 Von der Lasa P277, 436, 437 Vreeken M26 Vyzhmanavin R8 1 Wahls P296 Walter P30 P24 Welin Westerinen R374 Wiemer R397 R32 Wolk R192 Wotawa Yakimchik P200, 325 Yandemirov M36 Yusupov M78 R204, 288, P61 Zaitsev, I. Zaitsev, N. P2 1 1 P226 Zaitsev, V. M328 Zakhodiakin P77 Zapata Zinchuk P393 Zinn R1 19 Zuckertort R160 Zilger P296 Zur M223 Zytogorsky P292
Followi n g up h i s successful volumes on rook e n d i n g s and pawn less endings John N u n n and his computer database turn the i r attention to the intricacies of endings with just knig hts, bishops a n d pawns - tricky endings of great practical i m portance. How many g a m es have been lost because a pl ayer d i d n ' t find the right way to use his knight to halt a passed pawn? Th e book a l so provides definite answers to some of the puzzles that have taxed the greatest chess m i n d s for centuries: precisely when do two kn ights win a g a i n st a pawn, and how do two bishops overcome a knig ht? •
Comprehensive covera ge of many types of minor-piece endings
•
Full of instructive ideas and i m portant new insig hts
•
Every move checked by computer database
•
Written by one of the wo r l d ' s foremost endgame theoreticians
G randm aster J o h n N u n n is a key m e m be r of the three-times si lver medal w i n n i n g E n g l ish Olympiad tea m , for whom h e has also won fo u r individual gold meda ls. H e regards his third place in the World C u p as h i s g reatest i n d ividual achi evement. He is one of Batsford ' s m ost respected auth ors, a nd was awarded the British Chess Federation Book of the Year awa rd (with Pete G riffiths) for Secrets of Grandmaster Play. Th e pub lication of Secrets of Rook Endings was reg a rded by m a n y reviewers as an event of great i m porta nce: " A book for anyone fascinated by the su btlety and precision of practical endgames. "
Bill Hartston, The I n dependent
" An extraord i n a ry u ndertaking which wi l l become the defi n itive work for centuries to come . "
John Walker, Th e Oxford Times
Also by John Nunn New Ideas in the Pirc Defence New Ideas in the Four Knights The Complete Pirc Beating the Sicilian
3
John Nunn's Best Games Secrets of Pawnless Endings
For a co m p lete l ist of other Batsford chess books please write to:
B.T. Batsford Ltd 4 Fitzhard inge Street, London W1 H OAH
£1 7.99 ISBN 0 - 7 1 3 4 - 7 7 27 - X
1 1 11
9 780 7 1 3 4 7 7 2 76