Bostock - Kripke on Identity and Necessity

Kripke on Identity and Necessity Author(s): David Bostock Source: The Philosophical Quarterly, Vol. 27, No. 109 (Oct., 1977), pp. 313-324 Published by: Blackwell Publishing for The Philosophical Quarterly Stable URL: http://www.jstor.org/stable/2218951 . Accessed: 27/02/2011 19:01 Your use of the JSTOR archive indicates your acceptance of JSTOR's Terms and Conditions of Use, available at . http://www.jstor.org/page/info/about/policies/terms.jsp. JSTOR's Terms and Conditions of Use provides, in part, that unless you have obtained prior permission, you may not download an entire issue of a journal or multiple copies of articles, and you may use content in the JSTOR archive only for your personal, non-commercial use. Please contact the publisher regarding any further use of this work. Publisher contact information may be obtained at . http://www.jstor.org/action/showPublisher?publisherCode=black. . Each copy of any part of a JSTOR transmission must contain the same copyright notice that appears on the screen or printed page of such transmission. JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship. For more information about JSTOR, please contact [email protected].

Blackwell Publishing and The Philosophical Quarterly are collaborating with JSTOR to digitize, preserve and extend access to The Philosophical Quarterly.

http://www.jstor.org

313

KRIPKE

ON IDENTITY

AND NECESSITY

BY DAVID BOSTOCK

I Kripke- holds both that proper names are rigid designators, and that identity-statementsin which both terms are rigid designators are necessarily true if they are true at all. My purpose in this paper is to examine his arguments for these theses, particularly the second of them, and to show that they are inconclusive. Kripke's main argument for the second thesis is most clearly expounded in his article "Identity and Necessity", and is most easily stated in the idiom of possible worlds. In this terminology a rigid designator is defined as one that designates the same thing in all possible worlds in which it designates. Suppose, then, that 'a' and 'b' are two rigid designators, and that 'a-b' is true. Then 'a' and 'b' each designate the same thing in one possible world, viz., the actual one. So, since they are rigid, they each designate the same thing in all possible worlds in which they both designate. That is to say that 'a=b' is true in all possible worlds in which 'a' and 'b' both designate, and hence that 'a exists & b exists -- a=b' is true in all possible worlds whatever, and is thereforea necessary truth. This, says Kripke, is all he really means when he claims, speaking loosely, that 'a=b' is a necessary truth. First, it is worth noting that this formal argument contains a flaw. For surely when Kripke claims (speaking loosely) that 'a=b' is a necessary truth, we are to understand him as intendingto rule out the possibility of a existing while b does not (and vice versa). But this does not follow simply from the stated definition of a rigid designator. That definition does not by itself rule out the possibility that 'a' and 'b' are both rigid designators, that 'a=b' is true (in the actual world), and that there are possible worlds in which 'a' designates while 'b' does not. Maybe something does prevent this situation arising, but if so it is not just the definitionof a rigid designator. What else might it be? To answer that we shall need a more informativeaccount of rigid designators than is given by the definitionin termsof possible worlds. We shall need some way of telling whether an expression is a rigid designator, some way of telling what an expression does designate in various possible worlds. Now Kripke does give us an account of what he means by a possible world, namely that it is what we may call a "counterfactual situation", 1See "Identity and Necessity", in Identity and Individuation, ed. Munitz (1971), pp. 135-64 (hereafter IN); also "Naming and Necessity", in Semantics of Natural Language, edd. Davidson and Harman (1972), pp. 253-355 (hereafter NN).

3 4

DAVID

BOSTOCK

conceivedas a situationof this,the actual, world(IN, p. 148). To use his oun example,whenwe say 'If Nixon had bribedsuch and such a Senator, Nixon would have got Carswellthrough',we are speakingof Nixon and Carswell(who are, of course,inhabitantsof the actual world),and saying what wouldhave happenedin this,the actual, worldif a certain"counterfactualsituation"of the worldhad been actual. To say, then,that a designatordesignatesthisor that in a certainpossibleworldis to say that it has that designationin a certaincounterfactual situation,and I thinkthis in be must to it turn taken mean that has that designationwhenit occursin the sentence(or sententialthat-clause)whichwe use to specifythe situation. For the point whichKripke mainlyinsistsupon is that when we say 'If Nixon had . . .' it is Nixon wmeare talking of, and not (e.g.) some possible

counterpartof Nixon. That is, the word'Nixon' is used to designateNixon whienit occursin counterfactual sentences;and this is Kripke's main way of illustrating whathe meansby the claimthat 'Nixon' is a rigiddesignator. So this yieldsthe followingcriterion:an expressionis a rigiddesignatorif and onlyifit is used to designatethesame thingbothin ordinary("factual") assertionsand in counterfactual assertions.Moreexplicitly,it has to designate the same thing when it occurs in any counterfactual,2 i.e., in any of a it counterfactual what there and situation, specification designateshas to be the same thingas it ordinarilydesignatesin ordinary,factual,assertions. Two commentson this criterionare worthmakingat once. First,on thisaccountof the matterthereis no need forany extracaveat about possibleworldsin whichthe designatordoes not designateanything, whichindeed is a whollymisleadingway of talking. For the designatoris not itselfthoughtof as occurringin this or that possibleworld;what is in questionthroughoutis our actual uses of the designator,in thisworld,the worldbeingas it is. Kripkehimselfis at pains to pointout that he is not consideringhow the word would be used by people in a counterfactual situation(a different possibleworld);he is concernedwithhow we actually do use the word to describethat counterfactualsituation (that possible world)(IN, p. 145). To illustrate,if I specifya situationas one in which Nixon neverexisted,thenI specifya situationsuch that,if it wereactual, the word 'Nixon' (used as we use it) would designatenothing. But in fact the situationis not actual, so in factthe worddoes designate,and indeedit of the situationas designatesNixon even as it occursin the specification one in whichthereis no Nixon. For it is theactual Nixon (and not,ofcourse, ofhim)thatwe are excludingfromthe situationin question. any counterpart In a word,'Nixon' designatesNixon bothin the factualassertion'Nixon is still alive' and in the counterfactualassertion'if Nixon had never lived, then . . .' (cf. NN, p. 290).

20f course there will be counter-examplesthat are not allowed to count, forexample when the designator occurs in quotation marks or in reported speech. Besides, I may name my dog 'Nixon'. For simplicityI foregoany more exact statementof the criterion.

KRIPKE

ON IDENTITY

AND NECESSITY

315

Second, it should be noticed that the present criterionfor being a rigid designator does not anywhere introduce the notion of possibility. Now we might take it to be an essential feature of Kripke's definition of a rigid designator that it confines attention to what the designator designates in possible worlds. Perhaps Kripke should be interpretedas allowing that even a rigid designator may designate differentthings in impossibleworlds. If so, when we rewrite Kripke's definitionin terms of the behaviour of the designator in counterfactuals,it will be important to add a restrictionto counterfactuals with possible antecedents. I do not myselfthink that this would be true to Kripke's intention,but anyway it is easily seen that if the restriction is essential then we cannot hope to finda non-circularargumentfromrigidity of designation to the necessity of identity. For in order to determinewhether 'a' and 'b' are both rigid designators we should then have to determine first which counterfactuals containing them had possible antecedents, and this would involve (inter alia) determining whether the antecedent of 'if a had not been b, then .. .' was a possible one. But a verdict on that point was precisely what the argument was hoping to establish. Clearly the project must collapse into circularityif we have to reach the correct verdict before the argument can get started. Hence, if the argumentis to achieve anything, it must at least be possible to carry it through by starting with the supposition that the antecedent is possible. (We should then conclude, ifKripke's argument works, that the initial supposition was mistaken and that the antecedent is after all impossible. But wleshould now reach the result as a consequence of argument,and not as a preconditionforgettingthe argument started.) But this is to grant that Kripke's argument must, if it is to achieve anything, be capable of being conducted without the restrictionto possible antecedents in the definitionof 'rigid designator'. Hence I deliberately do not include that restriction. Now let us try out this criterion on a test case. Let us take over the usual (inaccurate) story about 'Everest' and 'Gaurisanker', viz., that Everest was named "Everest" when seen from India and "Gaurisanker" when seen from Tibet. And for vividness let us add to the story that there was once a border dispute between India and Tibet, each claiming possession of the mountain. Now, presupposing this background, consider the counterfactual If Everest and Gaurisanker had turned out to be differentmountains, there would have been no border dispute between India and Tibet. I take it to be uncontroversial that this counterfactual is intelligible, and for the reasons given above I take it that nothing, in this initial stage of the argument, can be made to depend upon the question whether the situation envisaged in the antecedent is a possible one. (Notice, incidentally, that we can and do quite intelligiblyuse counterfactualswith impossible antecedents, for example in reductioad absurdum arguments.) No doubt in Kripke's view the antecedent is not possible, for if it is possible that Everest and Gaurisanker should have turned out to be different,then it is also possible that

316

DAVID

BOSTOCK

they should have been different(an inferenceKripke himselfendorses, NN, p. 332). I lave chosen to write "turnedout to be different"only because this is the more normal way of putting the counterfactual: it is the supposed discoveryof the differencethat would have prevented the border dispute. But any reader who supposes that something might hang on this point may substitute a counterfactual which begins 'if Everest and Gaurisanker had been differentmountains'-say one that continues '. . . that fact would have been discovered in the course of the border dispute between India and Tibet'. Anyway, our question is: do the words 'Everest' and 'Gaurisanker', as they occur in this counterfactual,designate the same thing as they ordinarily designate, viz., Mount Everest? It is, I hope, fairly clear that neither of them is being used to designate any othermountain, so the alternative to consider is that one or both of them is not being used to designate at all. And in view of the symmetryin their occurrence, it seems rather odd to have one of them designating and the other not, so the relevant alternative seems to be that neither of them designates. But how could that be? How, one might ask, are we supposed to know what mountain or mountains are being talked of-what situation the counterfactual situation is-if neither of those expressions designates a mountain? Well, the obvious suggestion is that the proper names are in this context standing in fordefinitedescriptions,since definitedescriptionsare well knonil to be capable of introducing a topic of discourse without thereby referring to anything. Furthermore,in the set-up I have described (which is presumed knowxnto the speaker) it is evidently quite in order to say that the speaker is bearing in mind how Everest came to be named "Everest" and how it came to be named "Gaurisanker", and that he would not have expressed himselfas he did if he had not had these facts in mind. So I have no strong objection to the view that he is here using 'Everest' more or less as shorthand for the definitedescription 'the mountain named "Everest" when seen from India', and similarly for 'Gaurisanker'. The counterfactual could thus be re-expressed as If the mountain called "Everest" when seen from India and the mountain called "Gaurisanker" when seen from Tibet had turned out to be differentmountains, there would have been no border dispute between India and Tibet. But unfortunatelyit is not at all clear that the availability of this paraphrase yields the solution to our problem. That problem still remains as the problem whether, in the paraphrase, the two definitedescriptions do each designate a mountain. And how could we settle this issue? The only really strong ground that I know of for saying that a nonempty definite description does not designate (in a certain occurrence) is when we cannot plausibly take that description as having major scope. This may be tested for by applying the Russellian expansion. Of course it is

KRIPKE

ON IDENTITY

AND

NECESSITY

317

debatable whether we should regard the Russellian expansion as preserving the sense quite unchanged, or as providing only a rough and ready paraphrase, but I think we need not enter that issue here. It will be enough if, when we apply differentRussellian expansions corresponding to different ways of assigning the scope, one is clearly a much better paraphrase than the others. But on this test the descriptions in our example can, and indeed must, be assigned major scope. Obviously it would be quite wrong to apply the Russellian expansion within the subjunctive construction, to get If there had been just one mountain called "Everest" when seen from India, and . . . The speaker knows that there in fact is just one mountain which was called "Everest" when seen from India, and he is not taking this to be a merely counterfactual supposition. In fact he is not taking this fact to be governed by his opening 'if' at all, and the only plausible Russellian expansion is one that corresponds to major scope for the two descriptions,viz.: There is just one mountain which was called "Everest" when seen from India, and there is just one mountain which was called "Gaurisanker" when seen from Tibet, and if those mountains had turned out to be differentmountains there would have been no border dispute between India and Tibet. So far as this test is concerned, then, the definitedescriptions may perfectly well be taken as referring. Of course this does not settle the matter. I would not wish to say that a description can always be taken as referringif it has major scope. For example the description 'the sole teacher of Aristotle' obviously has major scope in the sentence 'Socrates was the sole teacher of Aristotle', but it is hardly natural to take the description as there referring(to Plato). On the other hand, I would not wish to say either that a description can never be taken as referring,and on this point at least Kripke and I seem to be in agreement. Kripke is perfectlyhappy to call a definitedescription a designator-indeed, he thinks that some definitedescriptions are rigid designators (e.g., 'the square root of 25', NN, p. 145). So the discussion so far is entirely inconclusive. No doubt there are further considerations that could be advanced on either side, but I do not see any way of settling the matter conclusively, so I now abandon the attempt. Instead, let us consider the consequences of settling the question one way or the other. On one view we say that in our original counterfactualthe names 'Everest' and 'Gaurisanker' do not designate, and we explain how this can be by saying that in this particular context they are merely abbreviating definitedescriptions which do not (here) designate. This, of course, involves abandoning the claim that those names are rigid designators according to our counterfactual criterion. This view would be consistentwith Kripke's position, if we take Kripke's position to be that rigid designators do not have to designate as usual in counterfactuals with impossible antecedents, for I have already mentioned that Kripke does think that the antecedent here is impossible.

318

DAVID BOSTOCK

B1utthen, as I also pointed out, on this view Kripke's argumentcollapses into circularity: the antecedent is impossible because the designators are rigid, and the designators are rigid, despite the fact that in this example they do not designate as usual, because in this example the antecedent is impossible. But the other view, which seems to me equally (and perhaps more) defensible, is that our names do designate as usual even in the present example. Of course they also bring to mind the associated descriptions,and that is what makes the counterfactual intelligible,but it seems perfectlyall right to ma,intain that they can do this as well as designating. We may perhaps compare Frege's doctrineabout ordinaryfactual identity-statements, which is that the two names involved do referto whatever they ordinarily refer to, but what makes the statement interesting (if it is interesting) is that they also have differentsenses. Without necessarily accepting all of Frege's views on the sense of a proper name, we call say much the same about our counterfactual: the two names do designate as usual, but what makes the counterfactual intelligible and what gives it its point is that the two names are also associated in some way with differentdefinite descriptions, and that (at any rate in this context) they do bring those descriptions to mind. In this way we keep to the letter (though possibly not the spirit) of the doctrine that names are rigid designators, according to our original and unrestricted counterfactual criterion. To prevent the collapse into circularity,let us persevere with this second view. So we now grant that 'Everest' and 'Gaurisanker' do designate as usual in our counterfactual, and (for the sake of argument) in all others; they are rigid designators. From this premise we are supposed to be able to argue that the statement 'Everest and Gaurisanker are the same (mountain)' is a necessary truth, because it is true "in" all possible worlds, i.e. all possible (counterfactual) situations. How is this argument supposed to proceed? The only route that I can see runs something like this. Our premise is (i) that the expressions 'Everest' and 'Gaurisanker' designate the same mountain when they are used in the specificationof any counterfactual situation. From this I think we are asked to infer (ii) that the mountains Everest and

Gaurisankerwill be the same mountainin any counterfactualsituation. Grantedthis, we can add (iii) that any possible situationis (or may be thosemountainswillbe viewedas) a counterfactual situation,and therefore in any possiblesituation,whichseems to be the result the same mouantain we are after. But, as our own example very clearlyshows, (ii) does not follow from (i). From the premise that the expressionsare used to designate the same mountain in our specificationof the situation, it by no means follows

thatthemountainsare the same mountainin the situation.On the contrary, in our examplethe situationis specifiedas one in whichtheyare different mountains.

Perhaps it will help if I put the objection more generally in this way:

KRIPKE

ON IDENTITY

AND NECESSITY

319

the numberof entitiesreferred to whenspecifying a counterfactual situation need not be the same as the numberof entitiesin the situationso specified. One simpleillustration ofthisprincipleis where,whenspecifying a situation, I referto an entitywhichis not in the situationat all. For example,I may specifya situationas one in whichNixon does not exist,or (say) as one in whichNixon's motherhad no children.In eithercase I referto Nixonwhen the situation,but Nixon is not in the situationI specify.Another specifying a situationI referto two distinctthings, illustrationis wherein specifying but in the situationso specifiedthey are one. For example,considerthe 'If the East Indies and the West Indies had been the same counterfactual of islands (as Columbussupposed),thieworldwould have been much group smallerthan it in fact is'. Here it would be veryreasonableto say that in the situationI referto two groupsof islands,but in the situation specifying so specifiedthereis only one group. Our originalexample with 'Everest' and 'Gaurisanker'is the reverseof this,forit is a case wherein specifying the situationI refer(twice)to one mountain,and in the situationspecified thereare two. There seemsto me nothingamiss withthis suggestion,and fromrigidity ifI am rightthenit seemsquite clearthat thereis no inference to of designation(as definedby our counterfactual what can cocriterion) with be in to counterfactual this And situations. herently supposed happen statements for the of main necessity identity argument collapses. Kripke's II of this "main" argumentthat we have been in and Alongside support a also discussing,Kripke employs challengewhichis worthsome consideration. His opponentmaintainsthat the identitybetweenEverestand Gaurisanker,betweenHesperus and Phosphorus,and (perhaps) betweenCicero one. In his opinion,thingscouldhave and Tullyis in each case a contingent been otherwise.Kripke challengeshim to specifyin moredetail a situation in whichtheywouldhave been otherwise.He asks (changingthe example underwhichHesperuswouldnot have now) "Are therereallycircumstances been Phosphorus?",3and goes on to maintainthat no such situationis imaginable. Let us see whetherwe can meetthis challenge. It seems obvious that the orbitof Hesperusmighthave been different fromwhatit nowis,yetmightin facthave beensuchthattheeveningappearfromwhat ances of Hesperus(whenseen fromearth)wereindistinguishable of orbit such that the have are been now. might Similarly Phosphorus they fromwhat the morningappearancesof Phosphoruswere indistinguishable that eitherof thesesituations theyare now. I take it to be uncontroversial taken singlyis a possibleone. The next step is to claim that both together are possible. If this is admitted,then Kripke's challengeis met,because a situationin whichHesperushas one orbit and Phosphorusanotheris evidentlya situationin whichtheyare different planets. 3NN, p. 306; IN, p. 153.

320

DAVID BOSTOCK

Kripke is hardlylikelyto agree that this meetshis challenge. He will grant,no doubt,that a situationin whichthereare two such planetswith but synchronizedorbits,exactly reproducingbetweenthem the different appearancesof Venus as seen fromearth,is a possible one. He will also agree (cf.NN, p. 155) that ifthishad been the situationthenthe referenceof 'Hesperus' fixingprocedureswhichwe in fact used to fixthe references of and 'Phosphorus'wouldhave fixedtwodifferent planetsas the references thoseexpressions.But he willgo on to say that in factthe worldis not like that, and in factthe two expressionsreferto the same planet. We are not how those expressionswould have been used if supposedto be considering but ratherhow the wordsare in factused, the the worldhad been different, worldbeingas it is. And the questionis: could we, now,properlydescribe that possible situationas a situationin which Hesperus and Phosphorus were different planets?What,he mightsay, would makethe one planet in our imaginedsituationthe same planet as Phosphorus,and the otherthe same planetas Hesperus? But herewe can surelyreplyto Kripkein his own coin.4 Whichentities situationis not somethingthathas to be discerned figurein a counterfactual a thus and dissimilarities bringingto lightthe similarities by using telescope, betweenthat situationand the actual one, so that we can assess whether we have counterparts.Rather, we stipulatewhich entitiesfigurein the situationwhen we specifythe situationin the firstplace. So our replyis simplyto stipulatethat the situationis one in whichPhosphorushas the one orbitand Hesperusthe other. By our stipulation,it is Phosphorusand Hesperusthat we are talkingabout. And here,I thinkthe argumentreachesa stalemate. For Kripke's only replyseems to be that we cannotstipulatethat-or rather,if we do, the situationwe specifyis not afterall a possibleone. Whynot? Because, says Kripke,it is not possiblethat Hesperusand Phosphorusshouldhave been different planets. But that,of course,is preciselythe questionat issue. So, as I say, stalemate. III I do not believe that Kripke offersus any furtherargumentson this topic. The two argumentsjust discussedare fairlyclearlythe only argumentsto be foundin the article"Identityand Necessity",and these argumentsare therepresentedwithoutrelianceon any positivetheoryof how namesrefer.In the laterseries,"Namingand Necessity",Kripkedoes begin with a detailed discussionof the questionhow names refer,but again his views on the necessityof identity-statements are then independently presented (pp. 303-8),and thereis no veryclose connectionbetweenthe two. His "causal theory"of names could be taken to be an explanationof how namessucceedin beingrigiddesignators-unlike(most)definitedescriptions -but clearlyit is not the only possible explanation,and the only claim I4N, pp. 146-7; NN, pp. 266 ff.

KRIPKE

ON IDENTITY

AND

NECESSITY

321

relevantto our problemis the claim that theyare rigid. As we have seen, whenrigidityis definedby our counterfactual criterionit may be disputed whetherthis claimis correct,but even if it is the conclusionKripke desires will not follow. I shall now concludewith a briefdiscussionof a different to Kripke's,and which argumentfora conclusionwhichhas some similarity beginsfromone of the thingshe says about rigiddesignators.But I think the argumentitselfis not to be foundin Kripke'sown writings. In "Identity and Necessity" Kripke does not explicitlypresent our criterion forrigiddesignators, but insteadproposesthistest: a counterfactual designator'a' is rigidif and onlyif the sentence'a mightnot have been a' has no readingin whichit can be seen as sayingsomethingtrue (pp. 148-9). Where'a' is takenas a definitedescription it is quite clearthat thissentence has (at least) two distinctreadings,dependingon the scope assignedto the firstoccurrenceof 'a'. For example,taking'a' as 'the highestmountainin the world'we may read the firstoccurrenceof 'a' as havingmajor scope,so that the wholesentencemay be paraphrased Concerningthe highestmountainin the world: it mightnot have been the highestmountainin the world. So taken,the sentenceis presumablytrue,on the groundthat therecould have been othermountainshigherthan Everest. But alternatively we may read the sentencein a way that gives the modal operatormajor scope, so that the appropriateparaphraseis rather It mightnot have been that: the highestmountainin the worldis the highestmountainin the world. So read, the sentenceis presumablyfalse,since the highestmountainmust whichevermountainit is. (But the sentencecould certainlybe self-identical, be taken as true on this reading,on the groundthat theremightnot have been a highestmountainat all. The convention,in discussingthis matter, is to ignorethatway ofcountingit true.) Anyway,on thefirstway ofreading the sentenceit is true,and therefore'the highestmountainin the world'is not a rigiddesignator.But Kripke claims,veryplausibly,that where'a' is taken as a name thereis no truereadingof our sentence;forexample,there is no way of reading Everestmightnot have been Everest underwhichit can be seen as expressinga truth;and similarlywhere'a' is a rigiddefinitedescription. One possibleexplanationforthisdivergenceis that in the case of names thereis no distinctionof scope to be made. But the line of argumentwe requirewill not adopt thisexplanation;it will say ratherthat we can make the same scope-distinction withnames, but that withnames each reading has the same truth-value.That is, we can distinguishbetween ConcerningEverest: it mightnot have been Everest and It mightnot have been that: Everestis Everest.

322

DAVID

BOSTOCK

But bothare false. (Indeed, we mightattemptto distinguishbetweenthem on the groundthat the second can be countedas true,but in the way we said was not allowed to count: we mightregardthe second as true on the groundthat there mightnot have been such a thingas Everest. But I confessI do not findthis altogetherconvincing.)Anyway,the firstclaim we make is that our sentencecan be read in the firstway, and so read it is false. That is, we may assert It is not the case that: concerningEverest: it mightnot have been Everest. Since Everestdoes exist,thisis equivalentto ConcerningEverest: it is not the case that it mightnot have been Everest abbreviateto whichwe may conveniently ConcerningEverest: it is a necessarytruththat it is Everest. The next claim is that whenthese sentencesare read in the firstway, with the firstoccurrenceof the name (or description)havingmajor scope, identitydoes licensesubstitutionforthat occurrence.We could argue for this firstby pointingout that thereare many other(intensional)locutions in whichthe scope of a name or descriptionseems relevant,and in all of these identitydoes licensesubstitutionwhen the name or descriptionhas major scope. For example, Ortcutt:Ralph believesthat he is a spy Concerning has the same truth-valueas the onlyman he knowswho is not a spy: Ralph believes Concerning that he is a spy providedthat Ortcuttis in fact the only man Ralph knowswho is not a spy. Next we could strengthenthe argumentby consideringinstancesof our sentence-form 'a mightnot have been a' in whichthereis a description whichhas major scope. Most sentencesof the form'Concerning the so-andso: it mightnot have been the so-and-so'are true,and they remaintrue upon substitutingfor the firstoccurrenceof the definitedescriptionany otherexpressionwith the same reference.On the otherhand, if we take one that is false,say the square root of 25: it mightnot have been the square Concerning rootof 25 thenthis remainsfalseunderany such substitution.One mightat firstbe inclinedto say that if we replacethe openingphrasewith,e.g., 'Concerning my favouriteodd number',thenthe truth-valuewillnot be preservedeven thoughthe reference may be. But thisinclinationmustbe dismissedas due to a confusionwith the otherreadingof the sentence,where indeed the substitution wouldnotpreservetruth-value.So finallywe applythisprinciple to our presentcase, and we deduce that since Everest,it is a necessarytruththat it is Everest Concerning

KRIPKE

ON IDENTITY

AND NECESSITY

323

and since (as we are supposing)Everest is in fact the same mountainas it followsthat Gaurisanker, ConcerningGaurisanker:it is a necessarytruththat it is Everest. This is the firststage of the argument,but it is onlythe firststage. For so far we have used the premisethat 'Everest' is a rigiddesignator,but not the premisethat 'Gaurisanker'is also. The unusedpremiseis Gaurisanker:it is a necessarytruththatit is Gaurisanker. Concerning this Putting togetherwithour previousresultwe can evidentlydeduce Gaurisanker:it is a necessarytruththat it is Everestand Concerning it is a necessarytruththat it is Gaurisanker. Fromthisit certainlyseemsto followthat Gaurisanker:it is a necessarytruththatit is bothEverest Concerning and Gaurisanker. But here I thinkwe stick. It does notseem to followfromthisthat is bothEverestand Gaurisanker It is a necessarytruththatsomething because it does not followthat it is a necessarytruththat somethingis Everest. Nor does it seemto followthat It is a necessarytruththat if somethingis Everest (or Gaurisanker) thensomethingis both Everestand Gaurisanker. At any rate this inferencecertainlyfails for other intensionaloperators (such as 'Ralph believesthat') in place of 'it is a necessarytruththat'. It seems that once our modal operatorhas been assignedthe narrowerscope thereis no way of inferring a conclusionin whichit has major scope. So we mustrestcontentwiththe conclusionwe have, or perhapsroundoffthe argumentby applyingexistentialgeneralizationto concludewith Thereis somethingsuch that it is a necessarytruththat it, Everest, are all identical. and Gaurisanker, Kripke mightbe contentwiththis conclusion.The premisefromwhich our argumentstartedwas that where'a' is a rigiddesignatorwe may assert a: it is a necessarytruththat it is a. Concerning Kripke mightwell claim that this premisecan properlybe expressedin the idiomof possibleworldsas Concerninga: in everypossibleworldin whichit exists,it is a. And he mightgo on to say that this was preciselythe pointhe was trying to expressin his formaldefinition, that 'a' is a rigiddesignatorif and only ifit holdsthat Concerning'a': in every possible world in which it designates,it designatesa. He mightalso go on to add that the difficulties whichwe foundearlierwhen (via our counterfactual tryingto apply his formaldefinition criterion)arose onlybecause the step of "semanticascent" whichis hereinvolvedis in this case not so harmlessas usual. Further,if we stickto the "materialmode" it becomeseasierto see how the flawwhichI began by versionthroughout, pointingout (p. 313 above) is overcome. For the argumentis simplythis.

324

DAVID

BOSTOCK

Since Everest is Gaurisanker,everypossibleworldwhichcontainsEverest containsGaurisanker,and viceversa. Further,sincein everypossibleworld whichcontainsEverest,it is both Everest and Gaurisanker,it followsthat in every possible world which containsEverest, Everest is Gaurisanker. The somewhatlongwindedreasoningwhichI have just been spellingout is perhaps a perfectlygood way of paraphrasingthis train of thoughtwhile avoidingthe idiomof possibleworlds. But thereare advantagesin the longwindedversion,forwiththatversion it is much easier to see the weak pointsof the argument.First,it clearly revealsa pointthat is noticeablyobscuredin Kripke'sown discussion,viz., that the wholeargumentreliesupon our beingable to use the usual idioms for "de re" modality,in which a modal operatoris applied to something and not counting less than a wholesentence. In Kripke'sown terminology, the phrase 'possible world', modal operatorsnearly always occur in train particular,his conclusionis stated in this ditionalde dictoconstructions; form,thoughif I am rightthereis no warrantforthis. So one lineof objection wouldbe to maintainthat thesede re locutionsare in generalmeaningthat the requiredway of readingthe crucialsentence'a less, and therefore mightnot have been a' is simplynot available.5 It is, however,rather difficult not to accept thisway of readingthe sentencewhere'a' is a definite description,and a positionwhichis closerto our ordinaryintuitionsmight but deny it fornames. Anbe to accept the de re readingfordescriptions, otherpossibilitywould be to admitthe de re readingfornames,but to say that whenso read the crucialsentenceis true. We argue,that is, that it is not a necessarytruth,concerningEverest,that it is Everest,forit might have been somethingelse.6 And a thirdline of objectionwould be that even whenan occurrenceof a name or descriptionis read withmajor scope, forthat occurrence. we cannotassumethat identitywilllicensesubstitution Certainlythe argumentwhich I sketchedfor this principlewas far from conclusive,and we have alreadyhad occasion to notice that the principle discussedon pp. 315-7 fails for counterfactuals.If in the counterfactual above we substitute'the mountaincalled "Everest" whenseen fromIndia' for'the mountaincalled "Gaurisanker"whenseen fromTibet', the resultis mustbe evidentlyto renderthe wholeunintelligible.But thesedescriptions in that counterfactual. assignedmajor scope MertonCollege,Oxford

5Doubtless Quine would take this view. 6There are hints of this view in what Dummett says of St. Anne (Frege, Philosophy of Language, pp. 111-6. But see also pp. 131-2).