T H E I N T E N S I O N A L C H A R A C T E R O F LAKS. AN.

A A N D

SA~IKARA

IN NAVYA-NYAYA* by

B. K. MOTILAL

Cambridge, Mass.
The term lak.san,a i n Sanskrit has been used i n m a n y senses, 1 b u t i n logical treatises it m a i n l y stands for "definition". The general p r o b l e m of " d e f i n i t i o n " as it is f o u n d i n Sanskrit philosophical texts has been studied by m a n y scholars. 2 Mlle. Biardeau has skillfully h a n d l e d the question emphasizing the N y g y a a n d the Vedfintic traditionsa; a n d quite recently J. F. Staal has attempted a " f o r m a l i s a t i o n of the theory of definition" i n N a v y a - N y f i y a4 with the help o f the n o t a t i o n s of Boolean algebra. 5 I n his c o n c l u d i n g remark, Prof. Staal notes that u n l i k e a large n u m b e r of cases i n I n d i a n logic the doctrines of lak.satza a n d sarhkara " s h o w a m a r k e d extensional character". 6 The p u r p o s e of the present paper is to examine the p r o b l e m s o f lak.san,a a n d sathkara in some detail i n order to * I wish to express my gratitude to Professor Daniel H. H. Ingalls for the many remarks and suggestions which have been instrumental in shaping this paper. I also thank my friend J. Masson for his helpful remarks on points of style. 1 In the grammatical tradition, for instance, lak.sa.na stands for sf~tra or grammatical rules, and lak~ya stands for words. (Cf. Mahabha.sya, Paspag~ttmika, Varttika 14). J. F. Staal has discussed this in his paper "The Theory of Definition in Indian Logic," JAOS, 33 (1961), pp. 122-6. One remark may be added to his discussion. The scheme for a samj~a satra of Pa.nini roughly corresponds to the notion of "nominal or syntactical definition" of the modern formal logicians. Such definitions are explained as "conventions which provide that certain symbols or expressions shah stand (as substitutes or abbreviations) for particular formulas of the system". They "may be theoretically dispensed with and all formulas written in full." See Alonzo Church, Dictionary of Philosophy, (D. D. Runes) (New York), p. 74-5. See A. B. Keith, lndian Logic and Atomism, pp. 153-4, S. Radhakrishnan, Indian Philosophy, II, p. 47, A. Foucher, La Compendium des topiques, pp. 7-14, & D. H. H. Ingalls, Materials for the study of Navya-Nyftya Logic (to be abbreviated as MN), t951, pp. 80-1. 8 See her "La d6finition dans la pens6e indienne", JA, 1957, pp. 371-84. 4 Navya-Nyfiya is usually considered to have begun with Gafige~a, whose date is now usually placed in the 14th century A.D. See D. C. Bhattacharyya, Bange NavyaNyaya Carca (Calcutta, 1952). 5 J.F. Staal, op. cit., p. 126. 6 He also refers to I. M. Bochefiski, who holds that Indian logic is intensional and thus differs from Western logic which is more or less extensional in character.

86

B.K. MOTILAL

determine to what extent the above statement can be maintained. F r o m a very early period the N y ~ y a school considered three elements to be the main concern of a philosophical treatise: uddega (enumeration o f the philosophical concepts), lak.san, a (definition), and par~k.sa (examination of those concepts). The technique o f definition, however, in which the Navya-Naiy~yikas evince so deep an interest, was not clearly developed in the early school. V~tsygyana (3rd century A.D.?) stated that the purpose of definition (lak~an.a) was to differentiate an entity from that which does not possess the nature or essence 7 (tattva) o f that entity. Thus, insofar as the term " n a t u r e " or "essence" remained vague in meaning, the notion o f lak.san, a also remained vague. In the new school, an attempt was made to avoid this vagueness by specifying that the purpose of definition is to distinguish the definiendum (lak.sya) f r o m all entities that are different from it (itara-vyavartakatvam). Or, to put the pragmatic value o f definition more clearly, it was said that its purpose was to facilitate the inference that the definiendum is distinct from any other entity (itara-bheddnurndpakatvam). A Ny~ya definitiod s (lak.san. avdkya) usually consists o f two parts~ a lak.sya (definiendum) and a lak.san, a (definiens). I n each case o f a true definitiod, it ,will be possible to formulate a sort o f miniature syllogistic inference, o f the f o r m : " A (is) B, because C", 1~ where the definiendum will occupy the subject position (pak.sa), "distinct f r o m others ''11 will be the sddhya, and the definiens will be the hetu. The content o f the above inference evidently points to a construction o f the definiendum and the definiens as co-extensive terms a n d "co-extenv Two different readings of the Nygtya-Bha~ya are available: uddi~t.asya tattvavyavacchedako dharmo laks.a.nam, and uddist.asya atattva-vyavacchedako dharmo lak~anarn. I agree with Miss Biardeau in favouring the latter as stating the general principle of definition less vaguely than the other reading, but I think that the first reading may be more in keeping with Vatsy~tyana's style. s By definitiod is meant a sentence which expresses a definition. u Good Sanskrit never employs a copula in the present tense; it distinguishes attribution from predication by word order or by choice of structure. lo The use of such types of sentences in representing the inferrential process in Indian logic clearly indicates that its language is somewhat intensional in character. Cf. R. Carnap's The Logical Syntax of Language (1959), Part IV, p. 246, where the author, while enumerating some of the important examples of intensional sentences, mentions also "Because A, therefore B". xl It may be noted here that this form of inference apparently fails when the definiendum is a keval~nvayin (unnegatable) term like vacyatva ("being nameable") etc. To avoid unnecessary complications I wish to leave aside such exceptional cases in the present discussion.

lak.sa.na AND sarhkara IN NAVYA-NYAYA

87

sive" m a y seem to have the same m e a n i n g as the t e r m "samaniyata" in Sanskrit. H e r e one m a y note t h a t the N y ~ y a t h e o r y o f definition does n o t generally rest on s y n o n y m i t y , b u t on sameness o f reference. Prof. W . V. O. Quine in his Two Dogmas of Empiricism has s h o w n t h a t definition, in some o f the b e s t - k n o w n senses o f the term, rests on " t h e t h e o r y o f m e a n i n g " , i. e. o n s y n o n y m i t y , 1~ b u t t h a t in a n o t h e r sense, definition m a y rest on sameness o f reference o r extension. 18 This latter sense is f o u n d frequently in m a t h e m a t i c a l literature. I n s o f a r as the N a i y ~ y i k a s claim t h a t the definiendum m u s t be co-extensive with the definiens, the N y h y a t h e o r y o f definition m a y be described as extensional a n d Professor Staal deserves credit for having noticed this a n o m a l y . But here some c a u t i o n m u s t be excercised. I n t r a n s l a t i n g "samaniyata" as "co-extensive" we are liable to miss one i m p o r t a n t point. The w o r d samaniyata contains the n o t i o n o f niyama w h i c h is usually explained as a vydptir e l a t i o n (cf. niyama~ cdtra vydpakat~). Thus, samaniyatatvam has been analysed b y the N a i y ~ y i k a s as follows: A. x is samaniyata with y if and only if x is pervaded by y and also the pervader of y. (tatsamaniyatatvam tad-vyapyatve sati tad-vyapakatvam). T h e following p r e l i m i n a r y r e m a r k is necessary to explain this condition. T h e N a i y ~ y i k a s in their logical analysis use a l a n g u a g e structure which is carefully f r a m e d so as to a v o i d explicit m e n t i o n o f quantification, class, a n d class m e m b e r s h i p . C o n s e q u e n t l y their l a n g u a g e structure shows a m a r k e d difference f r o m t h a t o f the m o d e r n western logicians. I n class logic we use the n o t a t i o n " e " for class m e m b e r s h i p a n d thus "xea" is r e a d as " x is a m e m b e r o f a " o r " x belongs to the class ct". T h u s we can express the fact t h a t two general terms F a n d G have the same extension b y using quantification a n d class a b s t r a c t i o n as follows: (x) (xea if and only if xel~). 14 13 Cf. From a Logical Point of View (Cambridge, Mass., 1961), Essay II, pp. 24-7. 18 Op. cit., Essay VIII, p. 312. "A general term t is said to be definable in any portion of language which includes a sentence S such that S has the variable x in it and is fulfilled by all and only those values of x of which t is true. Definability so construed rests only on sameness of reference - sameness of extension on the part of t and S". It is evident that S here is what is called an "open sentence" of the form "x is F". Thus schematically, the definitiod may be represented as "t ~ Fx". Open sentences, like terms, have extensions, and the extension of open sentences is the class of all objects of which the open sentence is true. (See W. V. O. Quine, Methods of Logic, New York, 1961, pp. 89-104). Accordingly t and S have been said to have the same extension here. 14 Here a and 1~stand for the class-abstractions corresponding to the terms F and G. That is, a designates the class of all objects of which the term F is true. This is but another way of saying that the extension of F is the class ct. Similarly with 13.

88

B.K. MOTILAL

But the Naiyfiyikas instead o f classes use p r o p e r t i e s a n d in lieu o f the relation o f m e m b e r s h i p they s p e a k in terms o f occurrence (v.rttitva) a n d its reciprocal, possession. I n other words, the simple p r e d i c a t i o n " a is a n F " (in s y m b o l s : Fa) which is interpreted in class logic as " a e a " , w o u l d be interpreted in N y g y a m e t a l a n g u a g e as ' 7 o c c u r s in a " , where the small letter '~r is a p r o p e r t y - a b s t r a c t i o n (as o p p o s e d to a class-abstraction) o f " F " . a5 M o r e o v e r , instead o f quantification, the N a i y ~ y i k a s use " d o u b l e negatives a n d a b s t r a c t substantives" to accomplish the same result, an N o t e also t h a t " n e g a t i v e s " are frequently e m b o d i e d in w h a t I shall call "'termini". 17 A n y n o u n substantive in Sanskrit can be t a k e n as w h a t I shall call a terminus inasmuch as it designates w h a t m a y be treated as a dharma (property) occurring in some locus a n d also as a dharmin (a p r o p e r t y - p o s s e s s o r ) in which some dharma or p r o p e r t y occurs. T o the Naiy~yikas, j u s t as "ghat.a.h" (a p o t ) can function as a dharma j u s t so can "ghatabhava.h" (pot-absence, i. e. l a c k - o f - p o t or otherness-than-pot). N o w , the following p r o c e d u r e is t a k e n to state the c o n d i t i o n for the coextensiveness o f the terms F a n d G. I f " F " a n d " G " are n o t a l r e a d y substantive expressions, so t h a t they c a n n o t be t a k e n as termini in the a b o v e sense, then let " a " a n d " b " be the c o r r e s p o n d i n g substantivized termini, is N o w we state, B. a is samaniyata with b iff 19 (1) a does not occur in a locus where absence-of-b does occur, and also (2) absence-of-a does not occur in a locus where b does occur. H e r e (1) gives the c o n d i t i o n for the p o r t i o n " a is p e r v a d e d (vydpya) by b " a n d (2) gives the c o n d i t i o n for the p o r t i o n " a is the p e r v a d e r (vyapaka)

1~ This can also be put as "a possessesf". The Naiyfiyikas while not objecting to the latter form, usually find it more convenient to adopt the former one. "f" actually designates the "essential" property of the entity to which F is applied. Perhaps we are here faced with some form of "essentialism". 16 Thus Prof. Ingalls has said, "The Navya-Nyfiya begins by removing all quantified statements by the use of two negatives and an abstract" in "The comparison of Indian and Western Philosophy", Journal of Oriental Research (Madras), Vol. xxii, 1954, p. 7. 17 I use the Latin word "'terminus" and its plural "termini" in place of "term" to furnish the special sense of a substantivized word which may serve as a term in a Nygya definitiod. Thus, in this special sense "gamanam" is a terminus whereas "gaechati" is not. is To substantivize a verbal expression the appropriate bhdvartha k.rt suffixes will be used, e.g. for gacchati, gamana ( = gam + lyut). To substantivize an adjective or descriptive noun (e.g. "father", "wife", etc.) one uses the taddhita suffixes -t~ or tva (-ness, -hood in English). 19 This is a common abbreviation for the expression "if and only if".

lak.sa.na AND sarhkara IN NAVYA-NYAYA

89

of b". (See definition (A) above). I n N a v y a - N y ~ y a language the conditions (1) a n d (2) are stated as follows: a is non-occurrent in the locus of the absence-of-b and is also not a counterpositive of any absence that occurs in the locus of b 2~ (Tad-abhavavad-av.rttive sati tad-van-nis..thdtyantdbh~va-pratiyogitvam). Some further remarks o n the n a t u r e of abstraction in Sanskrit are required. N a v y a - N y ~ y a language, as we know, is generally intensional. T h u s in their scheme relations such as samaniyatatva, rydpyatva, etc. are really r e l a t i o n s - i n - i n t e n s i o n as opposed to relations-in-extension. The i n t e r n a l structure of this language can better be represented with the n o t a t i o n s for intensional abstractions t h a n those of class abstraction. I propose to use the symbol " x [ ... x . . . ]" for the (intensional) propertyabstraction, where the prefix " x " , like the quantifiers, is a variableb i n d i n g operator, b u t unlike a quantifier produces a singular abstract term when attached to a sentence. T h u s " b e i n g fire" or "fireness" can be represented u n d e r this c o n v e n t i o n as " x [x is fire]". But it is i m p o r t a n t to note that when, instead of terms like fire etc., we use a " f r a g m e n t of description ''21 like lak.sya (definiendum), sadhya etc., the n o t a t i o n " x [x is lak.sya]" will behave more like what is called a n incomplete s y m b o l 22 t h a n like a singular abstract term n a m i n g an entity. I n fact a n expression 20 By using the transformation rules of class logic and truth-functionallogic one can easily show that the two schemata given below are equivalent. (a) (x) (x~a iff x ~ ) (b) - (~x) (x~a . - (xa~)) . - (3x) ( x ~ . - (x~a)) Now if we interpret "a" as "the class of all objects or loci in which a occurs" and "~" as "the class of all such objects or loci in which b occurs", then the first half of schema (b) represents condition (1), and the second half represents condition (2) of (B) above. In (b) we have eliminated the universal quantifier 'x' in favor of the negative and the existential quantifier. The manipulation is designed to show that the language of the Naiygyikas can be translated into extensional language involving quantification. The point to be noted is this: The Navya-Nyftya language may be translated into extensional language although it is not by itself extensional. This is a case in favor of the Thesis of Extensionality, which (in its strongest form) holds that all intensional languages are translatable into extensional language. For a brief exposition of this thesis see R. Carnap, The Logical Syntax of Language, 1LP, 1959, pp. 245-6. We can also eliminate "a" and "[3" and express (b) in quantificational language using only "predicates" like F, G etc. (c)-(3x)(Fx. -Gx). -(~x)(Ox. -Fx) But note that neither (b) nor (c) reveals all the essential features of Navya-Nyftya language, and that the variable x ranges over the domain of the loci. ~1 Prof. Ingalls has introduced this convenient term in his MN, p. 45. 82 By "incomplete symbol" is meant a symbol which has no meaning in isolation, but is only defined in a certain context. "They have a meaning in use, but not in isolation." Cf. A. N. Whitehead & B. Russell, Principia Mathematica, Vol. I, Ch. III, p. 66-7.

90

B.K. MOTILAL

such as " x is lak.sya" is to be developed fully only in a descriptive phrase like "the object x such that x is the lak.sya of such-and-such a definition" (in symbols: (ix) Fx). ~3 When the Naiygyikas use such descriptive phrases for x without naming x, it is clear that they cannot talk directly about the abstract property of x by substantivizing (with suffixes "-tea", or "-t~") the term that describes x. The usual practice of the Naiygyikas under such circumstances is to use another descriptive phrase to refer to such an abstract property. Thus if x is described as the lak.~ya in a given context, the property abstraction from x will be described by the Naiygyikas as the lak.syatdvacchedaka, i. e. the limiting property of the relational abstract lak.syat~, i. e., in symbols, the limiting property of x [x is lak.sya or definiendum] in such a context. For instance, if "fire" is described as the lak.sya, we can describe the property "being fire" or "fire-ness" as the limiting property of the relational abstract lak.syat~ in such a context. By an extension of the above convention for representing intensional abstraction, I propose to use the symbol "xy [ ... x ... y ... ]" for the relation-in-intension. 24 Thus the relation of samaniyatatva etc. will be written as "xy [x is samaniyata with y]", where the order of the variables in the prefix will determine the pratiyogin and the anuyogin of the relation in question, namely, the first will be taken as the anuyogin or "subjunct" and the second as the pratiyogin or "adjunct" of the relation. ~5 Now to return to the problem of the relation between definiens and the definiendum. Let us illustrate the point with this definitiod: bh~ry~tvam vivdhita-naritvam. 26 One may fairly translate this as "wife (is) married woman", whereupon the lak.sya or the definiendum would be "wife" (bh~ryd) and the lak.sa.na or the definiens would be "married w o m a n " (vivahita-n~d). N o w "wife" and "married w o m a n " can be said to be coextensive, i. e. they have the same extension, because they can be taken as "predicators" (of degree one) having as extensions the corresponding classes which are equivalent. 27 But since "wife" and "married w o m a n "
33 Since Peano it has been customary in logic to use this iota-operator to represent the descriptive phrases like "the object x such that". 34 All the above notations for intensional abstraction have been adapted from W. V. O. Quine. See his Word and Ol~ject (Cambridge, Mass., 1960), pp. 164-6. 35 Each Ny~ya relation has a "direction", i.e. it can be conceived as proceeding from one relation to the other. In other words, a relation can always be expressed in the form "relation of y to x". In such context the Naiy~yikas describe y as the pratiyogin, and x as the anuyogin of the relation in question. On this point see Ingalls, MN, p. 40. 3e Taken from Gad~tdhara Bhatt.~ch~yya's Vivahavdd~rtha.h. 37 In thus stating the condition for co-extensiveness I have roughly followed R. Carnap. See his Meaning and Necessity (Chicago, 1956), pp. 14-25.

lak~an, a AND sarhkara IN NAVYA-NYAYA

91

c a n n o t be t a k e n as the r e l a t a o f the relation o f samaniyatatva, we ascend to the next level o f a b s t r a c t i o n , i. e. to o u r termini " w i f e h o o d " a n d " m a r r i e d - w o m a n - h o o d " . A c t u a l l y these are the two r e l a t a o f the relation o f samaniyatatva, a n d n o t " w i f e " a n d " m a r r i e d w o m a n " . I t m a y be n o t e d t h a t this p o i n t will b e missed if this relation is t r a n s l a t e d w i t h o u t reservat i o n as "co-extensiveness". N o w to m o d i f y o u r p r e v i o u s r e m a r k , a definitiod in N a v y a - N y ~ y a substitutes a lak.syatdvacchedaka for the lak.sya o f p o p u l a r language. A c c o r d i n g l y , the lak.san,a u n d e r g o e s a c o r r e s p o n d i n g change. This is clearly shown b y the definitiod given above. I n s t e a d o f vivahita-ndr? (the o r d i n a r y definiens in English), the a b s t r a c t vivahitandrftva is t a k e n as the lak~a.na. This justifies the N a i y ~ y i k a s ' claim t h a t in a correct definitiod the lak.saua s h o u l d be samaniyata with the lak.syatdvacchedaka (instead o f saying t h a t the definiens a n d the definiendum s h o u l d be co-extensive). W e can f o r m u l a t e the general definition o f the relation o f samaniyatatva in intensionat n o t a t i o n as f o l l o w s :

xy Ix is samaniyata with y] =dzsa xy [x is pervaded by y]. xy Ix is the pervader of y] =d xy [x does not occur in a locus where absence of y does occur], xy Ix is
not absent from a locus where y does occur]. This c o r r e s p o n d s to o u r c o n d i t i o n s (1) a n d (2) given above. N o w , it is easy to see t h a t the v i o l a t i o n o f c o n d i t i o n (1) gives rise to the defect o f ativyapti (too wide a d e f n i t i o n ) , a n d t h a t o f c o n d i t i o n (2) to the defect o f avyapti (too n a r r o w a definition). ~sb A third k i n d o f do.sa (defect) m e n t i o n e d b y the N a i y ~ y i k a s is asambhava which is b u t an extreme case o f avyapti. It t a k e s place iff x is a b s e n t f r o m all the loci o f y . ~ T h e 28a I shall use the ordinary sign "--d" for definition. This can be read as '~ defined
a s '~.

2sb The Sanskrit definitions for these two defects are a-lak~ya-vrttitvam and lak~yav.rttitvam respectively. We can represent them in intensional symbols as follows: (1) xy [x is ativyapta by y] -'d xy Ix does occur in a locus where absence of y does occur]. (2) xy [x is avyapta by y] =d xy Ix is absent from a locus where y does occur]. Using the convention discussed in note 20, we can express these conditions in the language of class logic: I) (3x) (xect . - (xe~)) II) (3x) ( - (xea) . xel3) Note here, that the variable x, as before, is to be taken as ranging over the domain of loci, which is not the case in the former expressions (b) & (c). 29 The Sanskrit definition is lak.sya-matrdv.rttitvam. In symbols of class logic the condition is presented (using the convention of note 20) as follows: //I) (x) ( x ~ - (x~13)) Note that the word "matra" in Sanskrit accounts for the universal quantification in symbolic expression.

92

B.K. MOTILAL

c o r r e s p o n d i n g extreme case for ativyapti t h o u g h conceivable, is n o t p h i l o s o p h i c a l l y i m p o r t a n t . But there is a n o t h e r p e c u l i a r case o f awapti n o t e d b y the N a i y g y i k a s which is p h i l o s o p h i c a l l y significant. This is technically called " aprasiddhy~vy~pti" 3 ~ I f b o t h c o n d i t i o n s (1) a n d (2) are violated, b o t h the defects avyapti a n d ativyapti t a k e place. T h e situation is as follows: x occurs in a locus where y does n o t occur a n d x is absent f r o m a locus where y does occur. 31 This brings us close to o u r second p r o b l e m , viz. sarhkara, which we shall examine presently. A t any rate, the presence o f two defects in one case is n o t h i n g to be w o n d e r e d at, for this only a m o u n t s to saying t h a t the hetu o f the inference concerned 32 is defective in two ways. A c t u a l l y , the N a i y ~ y i k a s can c o n s t r u c t an inference where the hetu is defective in all five possible w a y s : 8

Sarhkara, o r rather the situation which the N a i y g y i k a s call sarhkarya, can be treated in the same m a n n e r as samaniyatatva, vyapyatva etc., viz. as a relation-in-intension between two termini (as before), belonging to the same level o f abstraction. But sarhkara b y itself is in n o sense connected with the defects o f definition (lak.san. a). The n o t i o n o f sarhkara is p h i l o s o p h i c a l l y significant only when one is trying to determine whether a p a r t i c u l a r terminus a b s t r a c t e d f r o m a n o r d i n a r y (general) t e r m can be r e g a r d e d as jati (a generic real) o r not. a4 Since o u r present concern is n o t to investigate the p r o b l e m o f j a t i or samanya p r o p e r , I shall proceed, after a b r i e f r e m a r k on the n o t i o n o f jati, to examine the two p r o l e m s
ao For further discussion regarding this subject see Ingalls, MN, pp. 80-1. 31 In notations of class logic this can be put as follows: (3x) (xect .-(xe~)). (3x) ( - (xett). xe13)J. F. Staal has explained (op. cit.) sarhkara or "overlapping" using the notations of Boolean Algebra as follows: (x. ~ @ o). (x. y =~ o). (~?. y =/: o) 32 It has been already noted that the Naiyftyikas want to reduce all definitions to an inference of the form: A differs from other entities because B. See above. 33 E.g. gaur agvo ghatatvdt (A cow (is) horse, because of potness). Here all the five kinds of hetvdbhdsa occtu'. 84 It is highly probable that the term sarhkara originally belongs to the Smr.ti literature which deals with problems of var.na-sarhkara etc. Udayana in his Kira.ndvaff on Pragastapdda-bhd.sya develops the notion of sarhkara at some length describing it as one of the five jati-badhaka's. This theory however, met trenchant criticism at the hands of ~riharsa and other philosophers. Even Raghun~ttha disputed it in his Paddrthatattvanir~pa.nam and also briefly in his Gu.na-kiran.avali-prakdha-didhiti (particularly the problem of sa~hkara with relation to bhfttatva and mftrtatva). This specific problem has been discussed by many scholars since. See Vddavdridhi (ascribed to Gadfidhara), Paddrthama.nd.anam (of Ven.idatta) ( = Prince of Wales Text Series, No. 30) etc.

lak.san,a AND sarhkara IN NAVYA-NYs

93

connected with sarhkara, e. g. how to formulate an exact definition of sarhkara, and under what condition sarhkara or sdd~karya will disqualify a terminus from being ajati. Since the second question is directly related to the problem ofjati, ! shall take it up first. The Nygya concept ofjdti, in its earlier phase, was undoubtedly related to the familiar problem of universals. The earlier Naiy~yikas betrayed a Platonic spirit of realism in upholding the doctrine of jdti. The pragmatic ground for acceptingjdti, namely the need to explain why a general name is applicable to different individuals, was however furnished almost from the beginning. It was argued that there must be some permanent, characterizing entity or essence corresponding to the general name which accounts for the name's being true of different individuals. Such a theory was however, subjected to the classic objections which have been raised against Platonism. Moreover, the idea of essence invited still more difficulties. A hierarchy ofjdti was admitted. Satt~ (existence or beingness) was accepted as the highest in the scheme. Under sattd, dravyatva, gu.natva, and karmatva (answering to the three general names, dravya or substance, gun.a, or quality, karma or movement) were accepted as the three jdti's which together are considered to be exhaustive and also mutually exclusive. Points to be stressed in this connection are these: the question of jdti cannot be compared to the medieval hierarchy of classes, where the class animal, for example, is divided into man and beast, or rather, man and not-man, nor to the modern hierarchy of classes and sub-classes of the mathematical logicians which invites the paradox of classes (Russell) and the antinomies of class and sub-classes (Cantor). a5 The Naiy~yikas thought of this jdti as something real and indestructible occuring in individuals (vyakti)26 It is as real as an ordinary particular object, say, a pot. Just as an ordinary object, a pot, is determinable by space-time co-ordinates, and any other object of a similar "type" cannot occupy the same space-time co-ordinates unless the one is totally included in and therefore a part of the other, so also only one jdti can occur in one individual and no other j~ti can occur in that india5 Theseantinomies are usually named after their discoverers. For a good exposition of them one may consult S. C. Kleene, Introduction to Metamathematies (New York, 1962), Ch. III, w167 11, 12. a6 The traditional Ny~tyaholds the curious doctrine that the jdti's are real (sat, satsvar~pa) but do not possess reality (sattd) by inherence. They were forced into this position by difficulties of hierarchy and of the relations between categories. If sattd could possess sattd there would be the faults of dtmd~raya and of infinite regress. Cf.

Kiral.l&,ali.

94

B. K. MOTILAL

vidual unless it is either included in or inclusive of the formerjdti. This principle was implicitly followed with the result that only a few termini abstracted from general names can be taken as naming jdti's. We can rephrase the condition for jdti in a specific case as follows: C. If x and y are two termini naming two general properties shared by more than one individual, and if those two properties are said to occur in one and the same individual (vyakti), then they can both be regarded as instances ofjdti, iff one of them is either included in or inclusive of the other. Violation of this rule gives rise to the situation known as sarhkara. A terminus under sarhkara is disqualified from being a jati since it violates this principleY Now to formulate the definition of sarhkara is not at all difficult, if only we make certain that the relation of "inclusion" as expressed in the above condition be reframed using intensional language. I shall conclude this discussion by examining a definition of sarhkara offered by Gad~dhara Bha.t.tgc~ryya in his Tattvacintdma.ni-dMhiti-viv.rti. The definition runs as follows :3s

Taj-jaty-avy@akatve sati taj-j~ti-vyabhicaritve sati taj-j~ti-sglmgm~dhikara~tya~ sarhkarab.

Using the notation for intensional abstraction as before, this can be written as follows: x [x possesses sarhkara] = x [x is not the pervader of a jdti a] . x Ix is not pervaded by the same jdti a]. x [x occurs in a locus where a occurs].

Example: by substituting "indriyatva" (sense-organness) for x and "p.rthivftva" (earth-ness) for a we get a true statement, which proves that the former is disqualified from being a jdti by the latter. It is however, clear that sarhkara is here defined with relation to a jdti term. But there is another kind of sarhkara which the Naiy~yikas call "paraspara sarhkara", for instance the relation between bhfttatva and mftrtatva. In order to apply the above definition of sarhkara here it is necessary to ascribe deliberately jdtitva to one of the two, so that the other can be tested for
87 The principle in Ny~ya language has been stated as follows: "sva-sdmdnddhi-

kara.nya-sv6bhdvavadv.rttitvdtitad-ubhaya-sambandhena jdti-vi~ist.a-jdtitv6vacchedena sva-samdnddhikara.ndtyantdbhava-pratiyogitvdbhava-niyama.h" Note that the expression "jdtitvdvaechedena" has the force of a universal quantification. 38 See Gddadhari (= Chowkhamba Sanskrit Series, Vol 42), p. 91.

lak.saga A~rO sarhkara IN NAVYA-NY.~YA

95

sarhkara. But such an ascription (called apadana) a9 is entirely deliberate and is made only for the purpose in question. Thus here also the whole notion of sarhkara amounts to this: it is a sort of relation-in-intension between two termini, and it determines that not both of them can be jati, i. e. at least one of them is not jati. Sarhkara, in Ny~ya theory, may disqualify a terminus from being ajati whose status asjati is yet to be proven, but it cannot do so where such status has been proven through some stronger means.
Thus from what has been said it can be seen that the doctrines of lak~aoa and sathkara can be translated into extensional language. And one will readily grant that such translation is a valuable aid to our understanding of Navya-Nyaya. But such translation does not reveal, and in fact if not used with constant caution may actually obscure, the essential features of the language-structure adopted by the Naiy~yikas. In their analysis, as we have seen, the Naiy~yikas cling to intensional abstraction at every step, and speak of their abstract properties as intensions of the predicates and sentences, rather than of classes and truth values.

~ It is usually defined as badha-kdlinecchajanya-pratiti-viyayatvam.