33.2.31 AOAC Offi Official Method 972.16 Fat, Lactose, Pro Protein, tein, and Solids in Milk Mid-Infrared Spectro Spectroscopic scopic Method First Action Action 1972 A. Prin Princi ci ple ple
Analysis of milk by IR is based on absorp Analy ab sorption tion of IR energy energy at specific spe cific wave num bers by CH groups in fatty acid chains of fat moleecules (3.48 m), by car bonyl mol bonyl groups in es ester ter link ages o f fat moleecules (5.723 m), by peptide mol peptide link ages between between amino acids acids of protein pro tein mole molecules (6.465 m), and by OH groups in lactose moleecules (9.610 m). Total mol Total solids solids (TS) or solids-not-fat sol ids-not-fat (SNF) are com pute d by as assign signing ing ex per periimen mentally tally deter de termined mined factor factor to percent per centage of all other solid milk com po ponents, nents, and by adding adding this amount to appro pri priate ate % fat, protein, protein, and lactose, or by direct multiple regres re gression sion calcu calcula lations tions using using instru instrument ment signals signals at com binations of above-mentioned wavelengths. wavelengths. Latter Latter method has been shown to be more accu accurate rate method of de ter termin mining ing milk solids. solids. Analy Anal ysis by IR is de pend pendent ent on cali cali bration bration against suitable suitable stan dard method. See Definitions of Terms and Explanatory Notes for calcu cal cula lation tion of regres regression sion lines. The above wave num bers can be gen gener erated ated by means of (1) series series of filters, fil ters, each absorb absorbing ing all wave num bers of of electro electromag magnetic netic spectrum spectrum but one, or (2) a Michelson inter interfer ferom omeeter, each wave num ber of inter in terest est selected selected from the full spectrum spectrum by mathe mathemat matiical means. B. Perfor Performance mance Speci Specifi fica cations tions
Num ber of firms man manu ufac facture ture vari various model instru instruments ments based upon princi princi ple, ple, A. It is im per peraative that indi individ vidual ual instru instrument ment utilized utilized meet follow following ing perfor performance mance speci specifi fica cations, tions, based upon analy analysis of 8 test sam ples: Standard devi Standard deviaation of differ difference ence between between du pli plicate cate instru instrument ment esti es timates: mates: Fat, prot protein, and lactose . . . . . . . . . . . . . . . . . . 0.02% Total To tal soli solids . . . . . . . . . . . . . . . . . . . . . . . . 0.04% Mean differ difference ence between between du pli plicate cate instru instrument ment esti estimates: mates: 0.02% Fat, prot protein, and lactose . . . . . . . . . . . . . . Total To tal soli solids . . . . . . . . . . . . . . . . . . . . . . . . 0.03%
Standard Stan dard devi deviaation of differ difference ence between between instru instrument ment esti estimates mates and values values by refer reference ence methods: methods: Fat [989.05 [989.05 ( ( see [991.20 ( ( see see 33.2. 26)], pro tein [991.20 see 33.2.11)], and lac lactose tose [896.01 [896.01 ( ( see see 33.2.21) or 930.28 or 930.28 ( see 33 3.2.22)]. . . . . . . . . . . . . . . . . 0.06% see 3 Total To tal solids solids [990.20 ( see 33 3.2.43)] . . . . . 0.12% see 33.2.44) and 990.19 ( see see 3 Mean differ difference ence between between instru instrument ment esti estimates mates and values values by reference refer ence methods: methods: Fat, protein, protein, and lactose 0.05% Total To tal solids 0.09%
Calculate Calcu late standard standard devi deviaation of differ difference ence as in 969.16 969.16D D ( see see 33.2.29), where S D = alge alge braic differ difference ence either either between between du pli plicate cate instru in strument ment esti estimates mates or between between instru instrument ment esti estimates mates and values values by refer ref erence ence methods. methods. C. Precau Precautions tions
Differences in fat readings for homogenized and unho un homog mogeenized te st p or ortions tions of same milk should be <0.05% to assure as sure accu accurate rate results results at high fat levels. levels. If larger differ dif ferences ences occur oc cur and servic ser vicing ing ho mog mogeenizer does not correct correct fault, consult con sult manu man u fac facturer. turer. Changes in moisture mois ture va po r co n tent within instru in strument ment console console will cause changes in op ti tical cal 0 and shift in calii bration cal bra tion level. Re place de sicca nt frequently, fre quently, prefer preferaa bly at end of each day, as 3–4 h are required required to re store equi equilib lib rium condi con ditions. tions. For best ac cu curacy, racy, cal i brate brate with typ e of milk to be anaalyzed (herd, indi an individ vidual ual cow, homog homogeenized, unho unhomog mogeenized, market, mar ket, etc.). Do not use mixtures mixtures of cream and milk for calii bration. cal bra tion. Avoid abnor abnormal mal (low lactose) lac tose) milks for cali cal i bration. bra tion. Single Sin gle pumping pumping of milk throug h in stru strument ment sam ple cell shou ld purge 99% of p revi vious ous test portion. portion. To test purging purging effi efficiency, ciency, perform per form fat deter de termi mina nations tions on H2O and single single pasteur pasteurized, ized, homog ho mogeenized wh ole mil k i n sequence: se quence: H 2O, H2O, milk, milk, H2O, etc., until until total total of 2 0 de ter termi mina nations tions have been obtained. obtained. Calcu Cal culate late Purging effi efficciency = ( M1 W 2 )
100 M 2 W 2
where M 1 and M 2 are first and second sec ond values values for milk and W 2 sec second ond value for H2O. Instru In strument ment must be well maintained maintained and function functioning cor correctly. rectly. Malfunc Mal functions tions that influ influence ence cali cali bration bration can cause large errors. er rors. I. Fat D. Cali Calibra bration tion
Before Be fore first cali cal i bration, bration, check linear linearity ity of out put put sig signals. nals. Mix accurately accu rately measured measured volumes volumes of H2O and homog homogeenized cream to pre pare ca 8 mix mixtures tures of known rela relative fat contents contents cover ering ing required range. Pre pare test portions as in 925.21 925.21 ( ( see see 33.2.02), and pump each mixture mixture into instru strument ment twice, us ing sec ond read readings ings to pre pare plot against rel relaative concen concentra trations. tions. If plot is no t lin ear, ad just as indi indicated cated in oper operat ating ing manual. manual. Re peat measure measurements ments and ad just justments ments until until plot is linear. linear. If ana analyz lyzing ing unho unhomog mogeenized milks, check linear linearity ity again with ca 4 dilu dilutions tions of unho unhomog mogeenized cream or high fat milk. If these mixtures mix tures devi deviate ate signif signifiicantly from linear linearity, ity, see C and check differ dif ferences ences between between readings readings for homog homogeenized and unho unhomog mogeenized milk test portions. Determine Deter mine % fat for series se ries of 8 preanalyzed [989.05 [989.05 ( see see 33.2.26)] milk test portions of type to be ana analyzed ( see see C). Pre pare test portions as in 925.21 ( see analyze in tripli triplicate. cate. see 33.2.02) and ana Com pare av aver erages ages of second second and third in stru strument ment readings with standard stan dard values values and follow follow instruc instructions tions in oper operat ating ing manual manual for required re quired changes to cali cali bration. bration. Alter Al terna natively, tively, for better accu ac curacy, racy, use sim ple lin linear ear regres regression sion equation equa tion that relates relates instru instrument ment esti estimates mates and standard standard refer reference ence values, val ues, using using latter latter as de pend pendent ent variable, variable, to correct correct these esti estimates. mates. In earlier in struments, struments, which do not have this ca pa bil bility ity in instru instrument ment software, soft ware, this calcu cula lation must m ust be per performed formed manu manually. To perform perform
INTERNA NATIONAL TIONAL 2005 AOAC INTER
this calcu calcula lation tion on cali cali bration bration data that are collected collected on succes successive sive days or weeks, it is neces necessary sary to record record instru instrument ment esti estimates mates as well as final fi nal results. results. Pur pose pose of this al alter terna native, tive, preferred preferred proce procedure dure is to avoid ad just justing ing instru instrument ment slope controls. controls. E. Deter Determi mina nation tion
Pre par e tes t por tio n as in 925.21 ( see 33.2.02). Oper Operate instrument instru ment in accor accordance dance with manu manufac facturer’s turer’s instruc instructions. tions. Use A, B, or A + B filters filters for measure measurements ments when using using filter filter spec trom tromeeter; use ester ester carbonyl, am amide ide II, and hydroxyl spectral infor informa mation tion when using using interferometric spectrom spectromeeter.
have this ca pa bil bility ity in instru instrument ment software, software, this calcu cal cula lation tion must be performed manu performed manually. To perform perform this calcu calcula lation tion on cali cali bration bration data that are collected collected on succes successive sive days or weeks, it is neces necessary sary to record re cord instru instrument ment esti estimates mates as well as final fi nal results. results. Pur pose pose of this alter al terna native, tive, preferred preferred proce procedure dure is to avoid ad just justing ing instru instrument ment slope controls. controls. I. Deter Determi mina nation tion
Pre par e tes t por tio n as in 925.21 ( see 33.2.02). Op er ate instrument instru ment accord according ing to manu manufac facturer’s turer’s instruc instructions. tions. When using using interferometric spectrom spec tromeeter, use ester ester car bonyl, am amide ide II, and hydroxyl spectral spectral infor informa mation. tion.
II. Protein Protein IV. Solids (To (Total) tal) F. Cali Calibra bration tion
Proceed as in D, first and third paragraph, Proceed paragraph, substi substitut tuting ing calcium pro pi pio onate solu solution tion (dissolve (dissolve 15.0 g pure calcium calcium pro pi pio onateH2O in H2O and dilute dilute to 1 L with H2O at 20°C) for homog ho mogeenized cream to pre pare mixtures mixtures of known rel ative concen con centrates trates from 0 to re quired maxii mum for pro max protein. tein. De ter termine mine % protein pro tein for se ries of 8 preanalyzed [991.20 [991.20 ( see having see 33.2.11)] milk test portions having range of pro tein con tent ap proxi proximately that of pop ula ulation tion of milks to be ana an alyzed. Pre pare test portions as in 925.21 ( see see 33.2.02) and anaalyze in tripli an triplicate. cate. Com pare av aver erages ages of second second and third instru in strument ment readings readings with standard standard values values and follow follow direc directions tions in oper op erat ating ing manual manual for making making required required ad just justments ments to cali cali bration. bration. Alter Al terna natively, tively, for better accu ac curacy, racy, use sim ple lin linear ear regres regression sion equation equa tion that relates relates instru instrument ment esti estimates mates and standard standard refer reference ence values, val ues, using using latter latter as de pend pendent ent variable, variable, to correct correct these esti es timates. mates. In earlier ear lier in stru struments, ments, which do not have this ca pa bil bility ity in instru in strument ment software, software, this calcu calcula lation tion must be performed performed manu manually. To perform perform this calcu calcula lation tion on cali cali bration bration data that are collected collected on succes suc cessive sive days or weeks, it is neces necessary sary to record record instru instrument ment esti es timates mates as well as final fi nal results. results. Pur pose pose of this al alter terna native, tive, preferred pre ferred pro cedure ce dure is to avoid ad just justing ing in strument strument slope controls. controls. G. Deter Determi mina nation tion
Pre par e tes t por tio n as in 925.21 ( see 33.2.02). Oper Operate instrument instru ment accord according ing to manu manufac facturer’s turer’s instruc instructions. tions. When using using interferometric spectrometer, use ester es ter car bonyl, am amide ide II, and hydroxyl spectral spectral infor informa mation. tion. III. Lactose Lactose
J. Cali Calibra bration tion
Use standard standard method 990.19 ( see 990.20 ( ( see see 33.2.43) or 990.20 see 33.2.44) to deter determine mine % total total solids solids for series series of 8 milks. Deter Determine mine % fat, % protein, protein, and % lactose. Calcu Calculate late mean differ difference ence for total total solids solids (TS ) F P L = a, where F , P , and L are esti es timates mates of fat, protein, pro tein, and lactose, lactose, respectiv ely. For routine routine control control of cali cali bration, bration, ana analyze addi ad ditional tional se ries of milks and ad just value for mean dif differ ference ence in accor ac cordance dance with accu accumu mulated lated data. Calculate Calcu late % total total solids solids = a + F + P + L Alter Al terna natively, tively, cali cali brate brate by mul multi ti ple ple re regres gression sion to calcu calculate late equa tion for es esti tima mation tion of either either TS or SNF as function func tion of fat, protein, pro tein, and lactose lac tose un cor corrected rected sig nals. For recalibration, use sim pl e lin ear regres re gression sion equation equa tion which relates re lates regres re gression sion es ti mates and standard stan dard refer ref er ence values, val ues, using us ing latter lat ter as de pend pendent ent variable, variable, to correct correct these esti es timates mates and obtain obtain final final result. re sult. In ear lier instru in struments, ments, which do not have th is ca pa bil bility ity in in stru strument ment software, soft ware, this calcu cal cu la tion must be performed per formed manu man ually. To perform perform this calcu cal cula lation tion on cali cali bration bra tion data that are collected col lected on succes successive sive days or weeks, it is neces nec essary sary to record re cord regres re gression sion esti estimates mates as well as final fi nal result. result. This alter alterna native, tive, preferred pre ferred method has been demon dem onstrated strated to be more accu accurate. rate. ( Note: products that are forti tified fied or diluted, bot h o rig rigiinal Note: For products multi mul ti ple ple re regres gression sion and sim ple re regres gression sion calcu calcula lations tions must be based on regres gression sion that forces cali cali bra bration line through or oriigin. First calcu cal cula lation tion proce cedure dure in this section, section, paragraph paragraph 1, cannot cannot be used success suc cessfully fully with these products.) products.) K. Com puter-Based Cali Calibra bration tion and Control Control
H. Cali Calibra bration tion
Proceed as in D, paragraph Proceed paragraph 1, substi substitut tuting ing lactose lactose solu solu tion (dissolve (dis solve 50.00 g lactose lactoseH2O and 0.25 g HgCl2 in H2O and dilute dilute to 1 L with H2O at 20°C) for homog homogeenized cream to pre pare mixtures mixtures of known rela relative concen concentrates trates ranging from 0 to required required maxi maximum for lactose. lactose. Deter Determine mine lactose lactose for series series of 8 preanalyzed [896.01 [896.01 ( see having see 33.2.21) or 930.28 ( see see 33.2.22)] milk test portions having range of lactose lac tose content content ap proxi proximately that of population of milks to be ana analyzed. Pre pare test portions as in 925.21 925.21 ( ( see 33.2.02) and ana analyze in tripli trip licate. cate. Use aver av erages ages of second second and third values values for each test portion in es esti timat mating ing cali cali bration bration require requirements. ments. Ad just in instru strument ment controls, con trols, as di rected in oper op erat at ing manual, manual, to make L = L, where L = instru instrument ment readings readings for lactose and L = values values from refer ref erence ence method. Alter Al terna natively, tively, for better accu ac curacy, racy, multi multi ply ply instru instrument ment esti estimates mates by L/L to obtain obtain fi nal results. re sults. In earlier earlier instru instruments, ments, which do not
TERNA NATIONAL TIONAL 2005 AOAC IN TER
Program Pro gram com puter to: ( 1 ) Calcu Cal cu late S2 = S 1 + [(S 2 S 1 )/PF], where S 2 is purge-corrected signal; signal; S2 is signal signal at any instru instrument ment channel; channel; S1 is same signal signal for previ previous ous test portion; and PF is predicted predicted purge fraction frac tion = purging purging effi efficiency/100 ciency/100 ( see see C). Ap ply cor rection to all signals sig nals received received from instru instrument. ment. (2) Calcu Calculate late (CS2)/10, where CS2 is pu rge-corrected con trol milk signal. signal. Sum control milk signals at each chan nel for 10 control test sam ples, and av aver erage. age. At time of cali cali bration, bration, store and desig designate nate as required con control trol milk test sam ples. (3) Calcu Calculate late S = S2 + CSE, where S is purge- and drift-corrected signal; sig nal; CSE is control control signal signal error = (required con trol sig signal nal aver av erage) age) (deter (determined mined control control signal signal aver erage) age) for any check on control milk. Ap ply value to all ex cept control control milks. (4) Calcu Calculate late esti estimates mates from purge- and drift-corrected signals signals using us ing cali cali bration bration signals, signals, and store. Equation Equation type is Esti Estimate mate = A + B (main equation). equation). Ini tially, A = 0 and B = 1.
(5) Calcu Calculate late cali cali bration bration equations equations and store. (6 ) Store random random test portion signals, signals, esti estimates, mates, and standard method refer reference ence results. results. (7 ) Calcu Calculate late and store mean differ dif ference, ence, standard standard devi deviaation of differ dif ference, ence, limits limits for mean differ dif ference, ence, and popu popula lation tion standard standard devi de viaation of differ difference ence for statis sta tisti tical cal quality quality control. control. ( 8) Store results re sults for checks on purging purging effi ef fi ciency tests, homog ho mogeeni niza zation tion effi efficiency ciency tests, linear linearity ity tests, and re peat peatabil ability ity tests. L. Instru Instrument ment Controls Controls
For instru instrument ment with second sec ondary ary slope controls, controls, oper operate ate in un corrected uncor rected mode. For instru instruments ments without without second secondary ary slope controls, con trols, reduce reduce inter interfer ference ence correc correction tion coef coeffi ficients cients to 0. M. Control Control Milk
Pre pare ade adequate sup ply of pas pasteur teurized, ized, homog homogeenized, preserved preserved milks with % fat about that of popu popula lation tion aver aver age. Transfer Transfer amounts for single single tests to test sam ple con contain tainers ers and keep refrig re friger erated. ated. Zero-adjust instru instrument ment channels. channels. Heat 11 test sam ples to 40°C, and pump each through in instru strument, ment, collect collecting ing signals. signals. Aver Average age last 10 results results for each signal to cal calcu culate late required required signal. signal. Record required re quired control milk sig signals nals to 4 deci decimals to prevent round rounding ing errors er rors in contin continu uous control control situ situation. N. Setting Pri mary Slope Controls
For filter filter IR spectrom spectromeeters, deter termine mine % fat ( F ), ), protein protein ( P ), ), and lactose ( L lactose series of 8 preanalyzed [e.g., 989.05 989.05 ( ( see L) on series see 33.2.26), 991.20 ( 991.20 ( see 896.01 ( ( see 984.15 ( ( see see 33.2.11), 896.01 see 33.2.21), or 984.15 see 33.2.24)] milks of type to be an alyzed by instru strument. ment. Warm test sam ples to 40°C, mix thoroughly, thoroughly, and pump through instru instrument, ment, col lect lecting ing and correct cor recting ing signals signals for purge and drift. For each signal sig nal calcu calculate late regres re gression sion equation equation of type S = a F + b P + c L, where S is sig nal nal and F , , and L are results results for standard. standard. At fat channel, channel, adjust slope P am pli plifi fica cation tion by 1/a. At protein protein and lactose lactose chan nels, adjust by 1/b and 1/c, respec respectively. tively. Check success success of attempt attempt to obtain obtain coef coeffi ficient cient of 1 for main com po ponents nents by repeating test. Re Read ad just slope am pli plifi fica cations tions if neces essary. sary. Lock slope controls controls and do not change settings set tings thereaf thereafter. ter. Warm 11 test sam ples of con control trol milk and rede re deter termine mine required required control control milk values values as indi indicated cated above. For interferometric spec trom tromee ters, re-zero by collect col lectin in g background back ground spectrum o f thermostated dis distilled tilled water. water. Ratio test sam ple emit emittance tance spectrum spectrum against water water background background emittance emittance spectrum. spec trum. O. First Cal Caliibra bration tion
Obtain Ob tain 20 milk test sam ples by random random selec selection from pop popu ula lation tion to be tested, and ana analyze by standard standard refer ref erence ence method for com po ponents nents to be es timated timated by IR method. Key standard standard re sults into com puter. War Warm m and pre pare 11 con control trol milk sam ples and the calii bra cal bration sam ples. Pump through in instru strument ment and collect and store signals. sig nals. Com puter ap plies purge cor correc rection, tion, calcu calculates lates control control signal sig nal correc correction tion values, values, ap plies these to calcu calculate late milk signals, signals, and stores these values values in matrix matrix with standard standard refer reference ence values. values. Select Select equation equa tion that is ap pli plica ca ble to in instru strument ment from follow following ing main equation equa tion types. For filter filter IR spectrometers: Signals Sig nals Compo Com ponents nents 1. F, P, L, TS, SNF
F A
F
L
*
*
*
FB
2. F, P, L, TS, SNF 3. F
*
4. F
*
*
*
*
*
*
[0.27 F(1)] + [0.73 F(2)] –1
–1
where FA = ester ester car bonyl (1742 cm ); F = amide amide II (1548 cm ); L –1 –1 = hydroxyl (1046 cm ); and FB = C–H bond (2854 cm ). ( Note: Except cept for TS and SNF, all main equation types force Note: Ex regression gression line through 0. With TS and SNF SNF,, ei ther may be used, but equation equa tion with inter intercept is pre preferred ferred and gives better results unless un less product prod uct being analyzed has been forti fortified fied or diluted.) For interferometric spectrom spec tromeeter: Signals Sig nals Compo Com ponents nents F, P, L, TS, SNF
F A
F
L
*
*
*
Calcu Cal culate late by multi multi ple ple lin linear ear regres regression sion main equations equations for com po ponents nents for which IR analy analysis is required. required. It is not neces essary sary to calcu cal culate late esti estimates mates for any com po ponent nent for which esti estimates mates are not required or to perform stan standard dard refer reference ence method analy analysis for that com po ponent nent except except when slopes for initial initial cali bration bration are ad justed. However, How ever, all signals signals must be controlled controlled us ing re quired quired con control trol milk signal sig nal values. values. P. Produc Production tion Analy Analysis
At begin beginning ning of each produc production tion pe riod and at defi def inite inter intervals during produc during production tion run, check control control milk signals signals and recal recalculate culate drift correc corrections tions for ap pli plica cation tion to product product test sam ples. In Inter terval val between be tween checks can be shortened shortened if drift for current current inter terval val is exces ex cessive. sive. If instru instrument ment signals signals sta bil bility ity is accept acceptable, able, 1 h inter interval val is satis sat isfac factory. tory. Signals Signals for product product test sam ples are corrected for purge and drift and then are used to calcu calculate late esti estimates. mates. Q. Quality Quality Control Control
Select Se lect produc production tion test sam ples at ran random dom at prede predeter termined mined intervals inter vals and perform perform standard standard refer reference ence method analy analyses on these. Store signals, signals, esti estimates, mates, and refer reference ence values values for these test sam ples in matrix matrix on accu cumu mula lative ba basis. sis. Check control milks and make ad just justments ments to drift correc correction tion just prior to these selec selections. tions. Inter Interval val between be tween selec selection tion (days) and num ber selected selected may be varied varied to suit labo lab ora ratory tory but 5 should be selected selected in 1 day and 20 in 1 week. Current Cur rent popu popula lation tion standard standard devi deviaations of differ difference ence between between esti es timate mate and refer ref erence ence methods methods may be calcu culated lated from last 50 differ dif ferences ences in these accu accumu mulated lated data. For quality quality control, control, calcu calculate late statis statisti tical cal limits limits for daily mean differ dif ference ence as: 3 Population standard standard devi deviaation/
N
where, N is is num ber of random random test sam ples tested that day. Action Action situ sit uations for changing changing A inter intercept cept in overall overall equations equations are: (1) Daily mean dif differ ference ence is greater than limit. (2) Six succes successive sive or 7 of 8 succes suc cessive sive or 8 of 10 succes suc cessive mean differ differences ences are on same sid e of 0. To correct correct this, ad just A in op po posite site di rec rection to mean of dif differ ferences. To correct cor rect for ( 1), ad just A in op po posite site direc direction tion by excess excess of differ difference ence over limit plus one-third of limit. Suggested Suggested addi additional tional statis statisti tical cal calcu cal cula lations tions are: (a) accu accumu mulated lated monthly mean differ difference ence standard standard devi deviaation of differ dif ference, ence, and limits limits for mean differ difference; ence; and (b) accu accumu mulated lated
INTERNA NATIONAL TIONAL 2005 AOAC INTER
annual mean differ annual difference, ence, standard standard devi deviaation of differ difference, ence, and limits lim its for mean differ difference. ence. Overall Over all ob jec jectives tives are to control control current current analy anal y ses as well as possi possi ble ble and to achieve annual annual mean differ difference ence very close to 0.
in data bank can be used to vary choices and deter determine mine net effect effect for differ dif ferent ent equations equations gener generated ated in this way. For test popu popula lation tion used first with this control control method, optimum condi con ditions tions were N = = 20 and inter interval val between between cali cali brations brations = 1 week.
R. Recalibration
S. Precau Precautions tions
Use signals signals and standard standard refer reference ence values values for last N test test sam ples in accu accumu mulated lated random random test sam ple data. Calcu Calcula lation tion may be made either ei ther to rede redeter termine co coef effi ficients cients of main equation equation or by sim ple regres re gression sion using using esti estimates mates of main equation equation as inde inde pend pendent ent variable variable and standard standard refer reference ence method values values as de pend pendent ent variable. variable. Latter Latter method would deter determine val values ues for A and B in overall over all equation. equation. Whenever When ever control control milk is changed, whether at a time of recalibration or b etween cali cal i brations, brations, run 11 test sam ples of both old and new control control milks. Use drift errors of old con control trol milk to ad just aver av erages ages for new con trol milk signals signals in calcu calculating lating required required sig nals for it. Do no t use con trol milk for lon longer ger than 1 week. With ran dom test sam ple data accu accumu mulated lated over long peri periods ods of time, num ber of test samples used for calibra bration tion and frequency frequency of calii bration cal bration can be opti optimized. mized. Signals Signals and standard standard refer reference ence values values
TERNA NATIONAL TIONAL 2005 AOAC IN TER
See C for
normal nor mal precau precautions. This con trol method uses re quired control milk signals to con control trol cali cali bration bration level in prefer preference ence to maintain main taining ing 0 settings settings at various channels channels with H2O. Advan Advantages tages are that drift errors errors identify identify signal signal sub ject to drift and that control con trol of level is in range of test analy anal y sis. Wa Water ter zeros zeros may still be checked. Drift away from 0 indi indicates cates ex tra trane neous ous absorp absorptions tions which could be due to exces cessive sive moisture moisture in console, console, buildup of scale on cell windows, win dows, or other instru instrument ment faults. Investigate contin continu uous drifting drifting in one di rec rection for 0 signal. sig nal. Refer Ref erences: ences: JAOAC 55, 488(1972); 61, 1015(1978); 62, 1202, 1211(1979); 72, 70, 184(1989). CAS-63-42-3 (lactose) (lactose) Revised: Re vised: March 1996
