The Concept of sādhana in Chinese Buddhist Logic

I would like to express my gratitude to my colleague, Dr. Liqing Qian, with whom I discussed the main idea of this paper, and who kindly helped me correct my English, and also to Prof. Birgit Kellner, who carefully read my paper and made many expert suggestions of various kinds towards its improvement. Needless to say, I am responsible for all remaining mistakes.

1. Introduction

In his article “More on parārthānumāna, theses and syllogisms” ([1991] 1999), Prof. Tom J. F. Tillemans gave us a brief account of the interpretation of sādhana (means of proof) in Dharmakīrti’s works as well as in the tradition following him. In this article, he showed us, on one hand, the development concerning this concept in Dignāga’s thought from the NMu to the PS, and on the other hand, insightfully explained the theoretical significance of this development through a comparison with the Aristotelian syllogism. In short, in Vasubandhu’s logical works as well as in Dignāga’s NMu, the sādhana was identified with the linguistic expression of the three members, i.e., the thesis (pakṣa), the reason (hetu) and the example (dṛṣṭānta). In Dignāga’s PS as well as in Dharmakīrti’s tradition, it was identified only with the reason and the example. This exclusion of the thesis-statement from sādhana highlights a particular Buddhist conception of what is decisive for the acceptability of an argument, one which contrasts with the principles governing the Aristotelian syllogism. The decisive or probative factor in an argument, according to this new conception of sādhana, is the truth of the premises, and not merely the logical form of an inference.

The present paper, as an extended observation based on Tillemans’ abovementioned article as well as on an earlier seminal article of Prof. Masahiro Inami,^[1] will show that in the Chinese tradition of Buddhist logic, the concept of sādhana (neng li 能立) was consistently interpreted as the reason-statement together with the statements of the positive and negative example, or directly as the trairūpya, the triple characterization of a correct reason. This interpretation of sādhana was explicitly ascribed to Dignāga himself as one significant innovation over earlier masters. Although the Chinese tradition appears to have proclaimed its theoretical exploration as exclusively based on the NP and the NMu, its novel interpretation of sādhana in fact only finds support in the PS. Like the tradition following Dharmakīrti, Chinese interpreters following Dignāga also took to various hermeneutic strategies in order to harmonize their new interpretation with the old one from the NP and the NMu, which is also found in various pre-Dignāgan Buddhist texts on logic.

In Chinese sources one also comes across the view that Indian Buddhist logicians after Dignāga held the new interpretation instead of the old one. Moreover, these logicians were reported in Chinese sources to have interpreted the “incompleteness” (nyūnatā) of an argument as referring to the incompleteness of the three characteristics, and not as indicating the incompleteness of the three statements that constitute a proof. This new interpretation of nyūnatā together with the above new interpretation of sādhana was inherited by Chinese logicians. In light of this new interpretation of nyūnatā, the present paper tries to demonstrate once more, “from a slightly different angle” than Tillemans, that it does not just reveal a terminological difference, but that it points to more fundamental issues at stake and indicates “how logic works”^[2] for Buddhist thinkers.

2. The twofold meaning of sādhana in the NP and the NMu

The literal meaning of sādhana is “means of proof.” As we know, sādhana is one of the eight topics in the basic framework of the NP. The eight topics are: (1) demonstration (sādhana), (2) refutation (dūṣaṇa, neng po 能破), (3) pseudo-demonstration (sādhanābhāsa, si neng li 似能立), (4) pseudo-refutation (dūṣaṇābhāsa, si neng po 似能破), (5) perception (pratyakṣa, xian liang 現量), (6) inference (anumāna, bi liang 比量), (7) pseudo-perception (pratyakṣābhāsa, si xian liang 似現量) and (8) pseudo-inference (anumānābhāsa, si bi liang 似比量):

NP 1:

sādhanaṃ dūṣaṇaṃ caiva sābhāsaṃ parasaṃvide /
pratyakṣam anu-mānaṃ ca sābhāsaṃ tv ātmasaṃvide //

NPCh 11a28–29:

能立與能破 及似唯悟他，現量與比量 及似唯自悟。

Demonstration (sādhana), refutation (dūṣaṇa) and their pseudo-forms (ā-bhāsa) are for the understanding of others. Perception (pratyakṣa), inference (anumāna) and their pseudo-forms are for the understanding of oneself.^[3]

The sādhana is the foremost among the eight topics. The sections on sādhana and sā-dhanābhāsa are the most extensive ones in the whole text of the NP. In this context, the term sādhana refers to a three-membered argument and stands in contrast with dūṣaṇa (lit. “means of refutation”). While the former is aimed at proving a view, the latter serves the purpose of refutation. Therefore, we could translate the term sādhana when used in this sense as “demonstration” or rather “argument,” i.e., the linguistic expression of a proof.^[4]

The three members or statements making up a sādhana are the thesis (pakṣa, zong 宗), the reason (hetu, yin 因) and the example (dṛṣṭānta, yu 喻). The last one normally consists of two parts, the positive example (sādharmyadṛṣṭānta, lit. “example by similarity,” tong fa yu 同法喻) and the negative example (vaidharmyadṛṣṭānta, lit. “example by dissimilarity,” yi fa yu 異法喻):

NP 2:

tatra pakṣādivacanāni sādhanam / pakṣahetudṛṣṭāntavacanair hi prā-śnikānām apratīto’rthaḥ pratipādyata iti //

NPCh 11b1–3:

此中宗等多言名為能立，由宗、因、喻多言開示諸有問者未了義故。

Here [among the eight topics,] the sādhana is the [three] statements consisting of the thesis and the other [two factors, i.e., the reason and the example], because the object [yet] unknown to the questioners is made known by these [three] statements consisting of the thesis, the reason and the example.^[5]

NP 2.4:

eṣāṃ vacanāni parapratyāyanakāle sādhanam / tadyathā / anityaḥ śabda iti pakṣavacanam / kṛtakatvād iti pakṣadharmavacanam / yat kṛtakaṃ tad anityaṃ dṛṣṭaṃ yathā ghaṭādir iti sapakṣānugamavacanam / yan nityaṃ tad akṛtakaṃ dṛṣṭaṃ yathākāśam iti vyatirekavacanam // etāny eva trayo’vayavā ity ucyante //

NPCh 11b19–23:

如是多言開悟他時，說名能立。如說聲無常者，是立宗言；所作性故者，是宗法言；若是所作，見彼無常，如瓶等者，是隨同品言；若是其常，見非所作，如虚空者，是遠離言。唯此三分，說名能立。

The statements having these [factors, i.e., the thesis, reason and example, made] on the occasion of convincing others, are the demonstration. For instance, “sound is impermanent” is the statement of the thesis. “Because of being produced” is the statement of the property of the subject [i.e., the statement of the reason]. “Whatever is produced is observed to be impermanent, like a pot, etc.” is the statement of the positive concomitance with the sapakṣa [, i.e., the statement of the positive example]. “Whatever is permanent is observed to be not produced, like ether, etc.” is the statement of the negative concomitance (vyatireka) [i.e., the statement of the negative example]. Only these three members are stated [to be the demonstration].^[6]

A three-membered sādhana can consequently be written in its full form as follows:

Proof (1)
Thesis: Sound is impermanent,
Reason: for sound is produced.
Positive example: Whatever is produced is observed to be imper-manent, like a pot;
Negative example: Whatever is permanent is observed to be not produced, like ether (ākāśa).

In the NMu, the three-membered sādhana together with its various pseudo-forms is also the foremost topic.

The same idea of sādhana as in the NP is found in the exposition of the basic theoretical framework in NMu 1, together with the autocommentary:

NMu k. 1a:

pakṣādivacanāni sādhanam

宗等多言說能立。

The sādhana is the [three] statements consisting of the thesis and the other [two factors, i.e., the reason and the example].

NMu 1.1:

由宗、因、喻多言，辯說他未了義故，此多言於《論式》等說名能立。又以一言說能立者，為顯總成一能立性 (sādhanam iti caikavacananirdeśaḥ samastasādhanatvakhyāpanārthaḥ),^[7] 由此應知隨有所闕名能立過。

Since the object [yet] unknown to someone else is made evident [to that person] by these [three] statements consisting of the thesis, the reason and the example, these [three] statements are said to be the sādhana in the Vādavidhāna and other [logical works of Vasubandhu]. Now, the expression “sādhanam” [here in k. 1a] is in singular form so as to show that the sādhana is a united [whole, even though it is comprised of three statements]. Thus it should be understood that the lack [of any of these statements] is called a fault of the sādhana.^[8]

In all the passages cited above, the grammatical phenomenon that the word vacana (state-ment, yan 言) appears in plural form (vacanāni/vacanair, duo yan 多言) betrays the view of the NP and the NMu that a sādhana has more than two members, that is, specifically three.

In both the NP and the NMu, the term sādhana is also used in a more limited sense for the reason(-property), i.e., the predicate of the reason-statement, in the above proof “producedness” or “being produced” (kṛtakatva). Now, the sādhana is in contrast with sādhya (suo li 所立), the property to be proved or the inferable property, i.e., “impermanent” in the above proof. The former property possesses the force of proving and the latter property is to be proved by it to be present in the subject, i.e., “sound.” In this sense, the term sādhana, when used as a substantive, can be translated as “means of proof;” when used as an adjective, it can be rendered as “proving.” The translation of sādhana as probans and sādhya as probandum, which is frequently encountered, is also suitable for the present context.

This meaning of sādhana can be found in the NP’s classification of four types of contradictory reason (viruddha, xiang wei 相違), and as well as in that of ten types of pseudo-example (dṛṣṭāntābhāsa, si yu 似喻):

NP 3.2.3:

viruddhaś catuḥprakāraḥ / tadyathā / (1) dharmasvarūpaviparīta-sādhanaḥ / (2) dharmaviśeṣaviparītasādhanaḥ / (3) dharmisvarūpaviparīta-sādhanaḥ / (4) dharmiviśeṣaviparītasādhanaś ceti //

NPCh 12a15–16:

相違有四，謂法自相相違因，法差別相違因，有法自相相違因，有法差別相違因等。

The contradictory [reasons] are of four kinds, as follows: (1) the [reason] proving the opposite of the own form of the [inferable] property, (2) the [reason] proving the opposite of [some] specific attribute of the [inferable] property, (3) the [reason] proving the opposite of the own form of the property-possessor, and (4) the [reason] proving the opposite of [some] specific attribute of the property-possessor.^[9]

NP 3.3–3.3.2:

dṛṣṭāntābhāso dvividhaḥ / sādharmyeṇa vaidharmyeṇa ca // ta-tra sādharmyeṇa tāvad dṛṣṭāntābhāsaḥ pañcaprakāraḥ / tadyathā / (1) sādha-nadharmāsiddhaḥ / (2) sādhyadharmāsiddhaḥ / (3) ubhayadharmāsiddhaḥ / (4) ananvayaḥ / (5) viparītānvayaś ceti // … vaidharmyeṇāpi dṛṣṭāntābhāsaḥ pañcaprakāraḥ / tadyathā / (1) sādhyāvyāvṛttaḥ / (2) sādhanāvyāvṛttaḥ / (3) ubhayāvyāvṛttaḥ / (4) avyatirekaḥ / (5) viparītavyatirekaś ceti //

NPCh 12b1–4:

似同法喻有其五種：一、能立法不成，二、所立法不成，三、俱不成，四、無合，五、倒合。似異法喻亦有五種：一、所立不遣，二、能立不遣，三、俱不遣，四、不離，五、倒離。

The pseudo-example is of two kinds, by similarity or by dissimilarity. Of these, first, the pseudo-example by similarity is of five kinds, as follows: (1) [an example where] the proving property (sādhanadharma) is not established, (2) [an example where] the inferable property (sādhyadharma) is not established, (3) [an example where] both [properties] are not established, (4) [an example] without [the statement of] positive concomitance and (5) [an example where] the positive concomitance is reversed. … Second, the pseudo-example by dissimilarity is of five kinds, as follows: (1) [an example where] the inferable property is not excluded, (2) [an example where] the proving property is not excluded, (3) [an example where] both [properties] are not excluded, (4) [an example] without [the statement of] negative concomitance and (5) [an example where] the negative concomitance is reversed.^[10]

In the Chinese translation of the name for each contradictory reason, the word sādhana is consistently rendered as “reason” (yin 因). The Indian commentator Haribhadra also follows the same technique in glossing this word as hetu (reason).

When commenting on the first kind of contradictory reason, i.e., the dharmasvarūpaviparītasādhana, he says:

NPṬ 39,4–5:

atra dharmasvarūpaṃ nityatvam / ayaṃ ca hetus tadviparītam anityatvaṃ sādhayati tenaivāvinābhūtatvāt /

Here the own form of the [inferable] property is permanence. Now, this rea-son (hetuḥ) proves (sādhayati) the opposite (viparīta) of that [own form of the inferable property (dharmasvarūpa)], i.e., impermanence, because [it] is invariably concomitant only with that [opposite property].

When commenting on the word sādhanadharmāsiddha as the name leading the NP list of pseudo-examples, Haribhadra says:

NPṬ 44,5–11:

sādhanadharmo hetur asiddho nāstīti bhaṇyate / tataś ca sā-dhanadharmo’siddho’smin so’yaṃ sādhanadharmāsiddhaḥ / … evaṃ sā-dhyobhayadharmāsiddhayor api bhāvanīyam /

That is to say, the proving property, the reason, is not established, i.e., does not exist. Hence, this sādhanadharmāsiddha is that in which the proving property is not established. … With regard to the sādhyadharmāsiddha and ubhayadharmāsiddha, it should also be thought in this manner.

Haribhadra analyzes the term sādhanadharmāsiddha as a bahuvrīhi compound, and iden-tifies the sādhanadharma (proving property) with hetu (reason).^[11] On the term sādha-nadharma, the NPVP explains further that: “This is both sādhana and property. Thus sādhanadharma. What does it mean? The reason.”^[12] Here, the sādhanadharma is ana-lyzed as a karmadhāraya compound. It refers to the property which is resorted to as the means of proof (sādhana) in an argument and therefore possesses the force of proving.

When commenting on NP 3.3.1.(4) on ananvaya, Haribhadra directly glosses sādhana as hetu:

NPṬ 46,7–9:

vinānvayena vinā vyāptidarśanena sādhyasādhanayoḥ sādhya-hetvor ity arthaḥ sahabhāva ekatravṛttimātram / pradarśyate kathyate ākhyā-yate / na vīpsayā sādhyānugato hetur iti /

The meaning is: without [the statement of] positive concomitance, i.e., with-out the presentation of the pervasion, [merely] the co-occurrence, i.e., merely the occurrence in one place, of the inferable and the sādhana, i.e., of the inferable and the reason (hetu), is indicated, i.e., is stated or announced, [but] not the reason as followed by the inferable in accordance with the requirement of pervading.^[13]

In the NMu classification of the pseudo-example, the name sādhanadharmāsiddha is replaced by sādhanāsiddha, hence sādhanadharma by sādhana.

Here, the sādhana is also used in the sense of reason(-property):

NMu 5.3:

「餘此相似」(k. 11d) 是似喻義。何謂此餘?謂於是處所立、能立及不同品，雖有合、離而顛倒說。或於是處不作合、離，唯現所立、能立俱有，異品俱無。如是二法或有隨一不成、不遣，或有二俱不成、不遣。

That “all other [kinds of example] different from them are pseudo-[examples]” means the pseudo-example. Which are those other [kinds] different from them? They are [examples] where there is [the statement of] the positive concomitance or [of] the negative concomitance with regard to sādhya, sādhana or asapakṣa (i.e., an individual used as negative example),^[14] nevertheless, it is stated in reversed manner; or [examples] where only the co-occurrence of sādhya and sādhana or [only] the co-absence [of sādhya and sādhana] from vipakṣa is indicated, [but] without the statement of the positive concomitance or of the negative concomitance. [Pseudo-examples also include such cases where] with regard to these two properties [i.e., the sādhya and the sādhana], either (anyatara) [of them] is not established (asiddha) or not excluded (avyāvṛtta), or both (ubhaya) are not established or not excluded.^[15]

The word sādhana (sgrub pa/sgrub par byed pa/sgrub byed) does not occur in the corre-sponding PS IV kk. 13–14, nor in the PSV on these verses:^[16]

The pseudo-form of that [i.e., example] is [an example where] the reason (gtan tshigs), the sādhya or both are not established in or not excluded from the asapakṣa (mi mthun phyogs), or [where] the concomitance is reversed in two ways [i.e., in either positive or negative fashion], or [where] the concomitance is absent. (k. 13)

[An example where] the [inferential] sign (rtags) is not found and so on, or [where] the positive concomitance or the other [i.e., the negative concomi-tance] is reversed, is not a [correct] example. The [mere] aggregation [of two properties in one place] is not a [logical] connection, because the [logical] connection is [yet] not explicated. (k. 14)^[17]

The term sādhana has here been completely replaced by the words gtan tshigs (hetu) and rtags (liṅga = hetu); the above cited passage NPṬ 46,7–9 also demonstrates that these terms are synonymous. Therefore, we can see that the terms sādhanadharma, sādhana and hetu are interchangeable as referring to the reason-property.

In the NMu, the term *sādhanahetu (neng li yin 能立因) must be considered as another relevant term relating to sādhana:

NMu 8:

「餘所說因生」(k. 15b)^[18] 者，謂智是前智餘。從如所說能立因生，是緣彼義。

The sentence “[the inference (anumāna)] which is different [from perception] is derived from the reason as presented [above in the discussion of sādhana]” means that the [inferential] cognition is different from the above [perceptual] cognition. It is derived from the *sādhanahetu as presented above. That is to say, it is based on that [*sādhanahetu].^[19]

Although I have found no Sanskrit material to confirm a karmadhāraya interpretation of this term, it is highly probable that it, if it were in Sanskrit, must be construed in the same way as the term sādhanadharma: the former refers to a reason which possesses the force of proving, while the latter refers to a property which possesses the same force. Both terms refer to the reason(-property).

The PS has a corresponding definition of inference for oneself (svārthānumāna):

PS II k. 1a–b:

svārthaṃ trirūpāl liṅgato’rthadṛk /^[20]

[Inference] for oneself consists in observing an object through a triply charac-terized sign.^[21]

Here, the term *sādhanahetu has been replaced by liṅga,^[22] and liṅga is just another name of hetu. Now, we have a series of synonyms: sādhana, sādhanadharma, *sādhanahetu, hetu and liṅga. All of them refer to the reason(-property) in this connection.

With this survey, we have exhausted almost all the occurrences of sādhana in the NP and the NMu. In both texts, the term sādhana sometimes refers to a “demonstration,” i.e., a three-membered argument, and at other times the reason(-property). There is no third option.

3. The new interpretation following the Pramāṇasamuccaya

Therefore, it seems surprising or even strange to some critical thinkers^[23] that Chinese classical commentators consistently proclaim the sādhana to be the reason-statement together with the positive and negative example-statements, or directly to be the trairūpya (yin san xiang 因三相), the triple characterization of a correct reason. Since the statements are three, and the characteristics of a correct reason are also three, the nature of sādhana as being three-membered is still perfectly preserved in this interpretation.

Moreover, this interpretation of sādhana is ascribed to Dignāga himself as one significant innovation over earlier masters. Kuiji 窺基 (632–682 CE) says:

YMDS 37–38 / 93a29–b2:

陳那能立，唯取因、喻，古兼宗等。…宗由言顯，故名能立。

The sādhana of Dignāga only includes the reason and the example, while in early times the thesis and others are also included. … The thesis is elucidated through the statements [of the reason and two examples]. Therefore, [the reason-statement and two example-statements] are named sādhana.

YMDS 50 / 93c28–94a3:

古師又有說四能立，謂宗及因、同喻、異喻。世親菩薩《論軌》等說能立有三：一宗、二因、三喻。以能立者，必是多言。多言顯彼所立便足，故但說三。

The early masters also talk about four [members of] sādhana. They are the thesis, the reason, the positive example and the negative example. The Bodhi-sattva Vasubandhu in the Vādavidhi and other treatises says that there are three [members of] sādhana, i.e., (1) the thesis, (2) the reason and (3) the example. This is because the sādhana is necessarily comprised of more than two statements, and [sādhana of] more than two statements is already adequate for elucidating that which is to be proved (sādhya).^[24] Therefore, only three [members] are asserted [by him].^[25]

YMDS 52 / 94a14–17:

今者陳那因、喻為能立，宗為所立。自性、差別二並極成，但是宗依，未成所諍。合以成宗，不相離性，方為所諍，何成能立？故能立中，定除其宗。

Now, Dignāga [asserts that] the reason and the example are means of proof (sādhana), and the thesis is what is to be proved (sādhya). Both the subject (*svabhāva, zi xing 自性) and the property (viśeṣa, cha bie 差別) [i.e., the qualificand and the qualifier in the thesis-statement,] have been well established (prasiddha, ji cheng 極成) [i.e., accepted by both the proponent and the opponent]. They are merely two substrata of the thesis-statement (*pakṣāśraya, zong yi 宗依), but not [by themselves] the point under disputation. Only when [they are] combined together so as to produce a thesis-statement, the invariable concomitance (avinābhāva, bu xiang li xing 不相離性) [of the subject with the property as expressed in the whole thesis-statement] then constitutes the point under disputation. So, how can these [two substrata] be the sādhana? Therefore, the thesis shall certainly be excluded from the sādhana.

Here, the reason-statement together with the example-statement is identified as sādhana. This time, the sādhana is in contrast with sādhya, the thesis-statement, in that the reason and the example are means of proof and the thesis is merely what is to be proved. Although the sādhana here is also in contrast with sādhya, the sādhana and the sādhya here are different from the interpretation of the NP and the NMu where they are taken only as the reason-property and the inferable predicate. The hetu in Indian logic can mean either the whole reason-statement or only the reason-predicate in that statement. Hence, the exegetical movement from the reason-predicate to the whole reason-statement is not prima facie breaking news. Nevertheless, the implied significance of this movement is very important. It concerns not a mere terminological shift, but a shift of perspective in the basic consideration of what makes a good argument good. When used in this new sense, the term sādhana may be translated as “probative factor.”

In order to harmonize this new interpretation with the NP and NMu passages where the sādhana is clearly said to have more than two statements–i.e., three members^[26] –the example is carefully counted as two members, i.e., the positive example and the negative example. The reason together with these two example-statements can then easily be interpreted as the three members of the sādhana.

Kuiji continues:

YMDS 53 / 94a17–21: 問：然依聲明，一言云「婆達喃」，二言云「婆達泥」，多言云「婆達」。27今此能立，「婆達」聲說。既並多言，云何但說因、喻二法以為能立？答：陳那釋云：因有三相，一因、二喻，豈非多言？非要三體。由是定說宗是所立。

Question: However, according to Sanskrit grammar (śabdavidyā, sheng ming聲明), a single statement is called vacanam, a pair of statements is called vacane, [and] more than two statements are called vacanāni. Here, the sādhana is mentioned in the form vacanāni. Since it is of more than two statements, why do you only assert the reason-statement and the example-statement, these two, as sādhana? Reply: Dignāga explains that the reason has three characteristics, i.e., the reason and the two examples. Aren’t they three statements [in all]? It is not required that there shall be three separate substrata (san ti 三體). Hence, the thesis shall be definitely asserted to be [merely] what is to be proved (sādhya).

Furthermore, when commenting on the last sentence in the NP 2.4 passage cited above (唯此三分，說名能立), Kuiji says:

YMDS 304; 113b25–29:

《理門論》云：「又比量中， 唯見此理： 若所比處， 此相審定(遍是宗法性也)； 於餘同類， 念此定有(同品定有性也)； 於彼無處， 念此遍無(異品遍無性也)。是故由此生決定解。」(NMu 5.5) 即是此中唯舉三能立。

The NMu says:

“and in an inference, only the following rule is to be ob-served: when the [inferential] sign (liṅga, xiang 相 = hetu) is ascertained in the subject of inference (anumeya, suo bi 所比), that is, the reason is pervasively a property of the subject (pakṣadharmatva, bian shi zong fa xing 遍是宗法性), and in cases other than [the subject], we remember its being [certainly] present in cases similar to that [subject in possessing the inferable property], that is, the reason is certainly present in similar in-stances (sapakṣe sattvam, tong pin ding you xing 同品定有性), and its being [pervasively] absent where that [inferable property] is absent, that is, the reason is pervasively absent from dissimilar instances (vipakṣe’sattvam, yi pin bian wu xing 異品遍無性),^[27] then knowledge of this [subject] is gener-ated.”^[28] This means the same as [when it] is [claimed] here that only three [members of] a sādhana are presented.

Here, the three members of sādhana are further identified with the three characteristics of a correct reason, the alleged basic criteria for a good argument in Buddhist logic. The presupposition made here is that the reason-statement and especially the positive and negative example-statements are nothing but the expression of the three characteristics, in the sense that these three statements are true if and only if the three characteristics are fulfilled.

This kind of interpretation of sādhana, though without being supported in the NP and the NMu, can indeed be supported from the PS, Dignāga’s magnum opus and his final work. Recent studies by Tom J. F. Tillemans have already showed that although in the NMu, Dignāga did consider the thesis-statement to be a member of sādhana, “in PS Dignāga did not consider the thesis-statement as being a sādhana, but nevertheless he most likely allowed its presence in a parārthānumāna.”^[29]

As pointed out by Tillemans, one passage from PS fits quite well with the intention to exclude the thesis-statement from sādhana while nonetheless letting it remain in the arrangement of a proof:

PSV ad PS III k. 1cd:

tatrānumeyanirdeśo hetvarthaviṣayo mataḥ // (k. 1cd)

yan lag rnams la rjes su dpag par bya ba bstan pa gaṅ yin pa de ni kho bo cag gi sgrub byed ñid du bstan pa ni ma yin te de ñid las the tsom skye ba’i phyir ro // ’on te gtan tshigs kyi yul gyi don yin pa’i phyir de ni de ma sgrub par byed do (de ma sgrub par byed do K: des bsgrub par bya’o V) // (K 124b6–7, Kitagawa 1965: 471,5–8)

In this regard, the presentation of what is to be inferred (anumeya) is held to concern the goal of the reason. (k. 1cd)

Among the members, the presentation of what is to be inferred is not presented by us to be the sādhana, because doubt will arise from it. However, because it concerns the goal of the reason, it [i.e., the thesis,] is to be established by that [i.e., the reason].^[30]

Besides the exclusion of the thesis-statement from sādhana, the equivalence of the sādhana with the expression of three characteristics can also be found in the PS:

PSV ad PS III k. 1:

trirūpaliṅgākhyānaṃ parārthānumānam.^[31]

Inference for others (parārthānumāna) is the communication of a triply char-acterized sign (liṅga).

The idea of assigning the reason together with the positive and negative examples to express the three characteristics can be found in the NMu as well as in the PS:

NMu 5.6:

若爾喻言應非異分，顯因義故。事雖實爾，然此因言唯為顯了是宗法性，非為顯了同品、異品有性、無性，故須別說同、異喻言。

[Objection:] If so, then the example-statement must not be a separate member [from that of the reason], because it is [designed] to express the implication of the reason. [Reply:] Although the fact is actually so, yet the statement of the reason is only meant to express [the reason’s] being a property of the subject, but not to express [the reason’s] being present in similar instances and being absent from dissimilar instances. Therefore, it is necessary to express the positive and negative examples separately [from the reason-statement].^[32]

PSV ad PS IV k. 7:

’on te de lta na dpe’i tshig kyaṅ tha dad par mi’gyur te gtan tshigs kyi don bstan pa’i phyir ro // … gtan tshigs ni mtshan ñid gsum pa can yin la / bsgrub bya’i chos ñid ni gtan tshigs kyi tshig gis bstan pa yin no // de las gtan tshigs lhag ma bstan par bya ba’i don du dpe brjod pa ni don daṅ bcas pa yin no // (K 151a2–4, Kitagawa 1965: 522,7–523,2)

[Objection:] However, if so, even the example-statement will not be separated [from the reason], because it is [designed] to express the implication of the reason. [Reply:] … Since the reason possesses three characteristics, [only the reason’s] being a property of the subject (sādhya = pakṣa) is expressed by the statement of the reason. In order to express the remaining [characteristics of the] reason other than that [first characteristic], it is meaningful to express the example.

Combining this idea with the PS’s claim that the sādhana is nothing but the expression of the triple characterization of a correct reason, it is not difficult to arrive at the conclusion that only the reason and example are sādhana, and not the thesis. To preserve the nature of the sādhana as being three-membered, counting the example as two members may suggest itself. The separation of positive and negative examples naturally results from the idea that the second and third characteristics of a correct reason are to be expressed by them respectively. At the same time, the statement of thesis is always preserved in an inference, even though it is no longer recognized as a member of sādhana. This is also Dignāga’s attitude as mentioned above.^[33]

Therefore, the Chinese conception of sādhana as the reason together with two examples or being exactly the triple characterization of a correct reason can be regarded as a natural movement from Dignāga’s late thought, and may to some extent reflects Indian views following Dignāga when Xuanzang was taught there. Although the Chinese tradition is alleged to have only the NP and the NMu as its root texts, the ideas presented in the Chinese commentarial literature on these two short treatises need not be limited to Dignāga’s early views. On certain occasions, and to a certain extent, ideas in Chinese literature are even more probably based on his later views, as well as on Indian interpretations that were produced shortly after Dignāga and were not yet influenced by the revolutionary contributions from Dharmakīrti (c. 600–660 CE). However, clues for gaining certainty on this hereto unknown historical relation might always remain ambiguous, since no special reference to the subtle differences between earlier and later stages in Dignāga’s thought could so far be found in the Chinese tradition. In this regard, recent studies on Dharmakīrti and his successors as well as on Dignāga himself will surely prove to be relevant to an improvement in our understanding of Chinese hetuvidyā.^[34]

4. The “completeness” of an argument and the identification of the pro-bative factors

As brought into light in Tillemans’ 1991 article, the above reinterpretations are not merely terminological issues, but reach deeper into the nature of “how logic works.”^[35] The theoret-ical implication of these new interpretations can be clarified if we look at the matter from a different angle, and take into account the reinterpretation of the fault called “incomplete-ness”^[36] (nyūnatā, que jian guo xing 缺減過性), which results from the new interpretation of sādhana. In NMu 1.1 Dignāga defined sādhana to be a three-membered argument, comprised of the thesis, the reason and the example. The lack of any one of these members results in the fault called “incompleteness.”^[37]

However, in the PS Dignāga says:

PSV ad PS III k. 1ab:

’dir yaṅ tshul gaṅ yaṅ ruṅ ba cig ma smras na yaṅ ma tshaṅ ba brjod par’gyur ro // (V 40b2, Kitagawa 1965: 470,7–8)

Here [in saying that inference for others is the communication of a triply characterized reason], it shall also be called incomplete when any characteristic [of the three characteristics] is unstated.^[38]

This understanding changes not only how the fault called “incompleteness” as handed down from the early phase of Indian logic is understood, but also the conception of what kind of factor contributes to the “completeness” of an argument, in the absence of which the argument becomes “incomplete” or unsound. We are now inclined to call them the “probative factors.” Indeed, a number of elements can be regarded as being capable of contributing to the “completeness” of an argument. At first, there should be certain linguistic expression with certain ideas the proponent would like to communicate to the opponent. This expression should be capable of explicating ideas in accordance with certain semantic conventions. Even the intelligence of the opponent could be presupposed as part of the necessary prerequisites for an argument to be practically adequate, since he should be intelligent enough to pick out the meaning as being the one the proponent intended to convey. In addition, one might also include the overall situation of the debate as being one where arguments from each side are to be evaluated only according to principles for thinking rationally. But not all factors which are necessary for a rational exchange to occur contribute to the soundness of an argument in the same way. Therefore, by “probative factors,” we do not mean all the necessary conditions for an argument to be “complete” in a general sense, which are nearly infinite, but only those factors which were actually selected by thinkers in the history of logic to be in the focus of their theorization of argument. The notion of a “probative factor” is therefore just a meta-logical concept, not a logical one in the usual sense. This is a concept only used to represent or recapture the main concern of a logician in his theory of argument. In fact, we are bound to select only a limited number of elements for reflecting on the soundness of a sound argument in a theoretical manner. This does not prevent us from recognizing the fact that there must be other elements remaining untheorized in our present framework, or even yet unobserved.

The mere illustration of probative factors contributes to a theory of argument just as little as a good intuition of what a sound argument may look like; it is not yet a theory of logic. The key feature of a theory is that the probative factors identified in it are at the same time considered to be the criteria for discriminating in a general way a sound argument from an unsound one. Theories of argument can be based on different approaches to the identification of different kinds of probative factor, and as a result yield different systems of criteria for sound argument. In short, different identifications of probative factors betray different conceptions of argument, which may lead to different theories of argument, or even different logical theories.

Now, let us return to the historical account as given in the Chinese literature, which is in line with PS’s new interpretation of “incompleteness.”

YMDS 57 / 94b17–21:

世親菩薩，缺減過性，宗、因、喻中，闕一有三，闕二有三，闕三有一。世親已後，皆除第七。以宗、因、喻三為能立，總闕便非。既本無體，何成能立？有何所闕而得似名？

[According to] the Bodhisattva Vasubandhu, [there are seven cases of] incom-pleteness (nyūnatā). [Of them,] three are the lack of [only] one [statement] among the thesis, the reason and the example, three are the lack of two [state-ments among them], and one is the lack of all the three [statements]. [Scholars] after Vasubandhu all exclude the seventh case. Since the thesis, the reason and the example, as three [members], form a sādhana, it is impossible for all of them to be lacking [and there is still an argument]. Since then, there would be no substratum (wu ti 無體) at all, what could be an argument (sā-dhana) and what [kind of argument] could be called faulty on account of being incomplete?

According to this view, logicians before Dignāga identified the probative factors with the statements of the thesis, the reason and the example. When one of these factors is lacking, the whole argument has the fault of incompleteness. A point that remains unclarified here is that the linguistic expression itself could contribute to the “completeness” or soundness of an argument in two ways. On one hand, the linguistic expression could be probative in representing a certain form of valid reasoning. On the other hand, it could be probative in that the reason-statement together with the example-statement is or is accepted to be true. According to contemporary logical theory, an argument is sound if and only if all its premises are true and the whole argument represents a certain form of valid reasoning. Therefore, if the identification of probative factors with the linguistic expression itself does not merely represent a good intuition of what a sound argument looks like, but is actually taken as resting upon a solid theory, it could possibly provide the Buddhist logicians with two different options in theorizing the “completeness” of an argument.

We name the first option the “formalist approach” in that the logical form itself is identified as the probative factor, forms the focus of theorization, and becomes the criterion for discriminating sound arguments from unsound ones. For the second option that takes the truth of premises to be the foremost criterion for finding sound arguments, we will speak of an “epistemic approach” or “dialectic approach,” depending on the interpretation of “truth” that is chosen. If we interpret a true statement as being approved by ascertained evidences (*niścayaprasiddha), we have an “epistemic approach.” If interpreting a statement’s being true as merely being accepted to be true, i.e., being equally accepted by both sides in debate (*abhyupagamaprasiddha, gong xu ji cheng 共許極成), we have a “dialectic approach.”

Our text continues:

YMDS 57–58 / 94b21–26:

陳那菩薩，因一喻二，說有六過，則因三相六過是也。闕一有三，闕二有三，無闕三者。大師至彼六十年前，施無厭寺有一論師，名為賢愛，精確慈悲，特以貫世，因明一論，時無敵者，亦除第七。自餘諸師，不肯除之。因一喻二，即因三相。

[According to] the Bodhisattva Dignāga, six cases of the fault [of incomplete-ness] are mentioned concerning the reason, the first, with the examples, the last two (yin yi yu er 因一喻二).^[39] They are the six faults concerning the three characteristics of a correct reason. [Of them,] three are the lack of [only] one [characteristic], three are the lack of two [characteristics], and there is no case where all the three [characteristics] are lacking. Sixty years before Xuanzang’s arrival at the Nālandā Temple, there was a learned master (śāstrin, lun shi 論師) Bhadraruci (xian ai 賢愛), who was famous throughout the world for his acute thinking and compassionate heart. No one could compare with him in the discipline of hetuvidyā. He also excluded this seventh case. The other masters, however, were not willing to exclude it. The reason, the first, with the examples, the last two, is exactly the three characteristics of a correct reason.^[40]

As is reported in the above passage, Buddhist logicians following Dignāga identified the factors contributing to the “completeness” of an argument directly with the three characteristics of a correct reason, the alleged basic criteria for a good argument in Buddhist logic. As a detailed account of the trairūpya formulae according to the Chinese tradition is not in place here, we can make no decision as to whether the “epistemic approach” or the “dialectic approach” was actually adopted by those Chinese logicians following Dignāga. Still, the evidence from our text clarifies that the “formalist approach” was not chosen. Towards this end, we just point out that the “incomplete” or unsound arguments which are to be ruled out by the three characteristics represent the same logical form with that of the above proof (1), which is typically a sound argument in Buddhist logic, regardless of its variety.

Indeed, it is not difficult for us to abstract the following form from the above proof (1):

Thesis: Sp
Reason: Hp
Positive example: (x) (Hx → Sx)
Negative example: (x) (¬Sx → ¬Hx)a

a p = pakṣa the subject, “sound;” S = sādhyadharma the inferable property, “being impermanent;” H = hetu the reason-property, “being produced.” Note that, the formalization here is just a provisional one adjusted to the present purpose and ignores aspects that are irrelevant.

Moreover, if we regard the negative example as merely the contraposition of its positive counterpart, we could just skip it. Then, the whole process of reasoning can be considered as beginning from the statement of the positive example and ending with the thesis-statement. If the formalization is correct, the above process clearly represents a valid reasoning.

When commenting on the NP passage on refutation (dūṣaṇa, NP 6),^[41] Huizhao^[42] has provided each one of these seven possibilities with an example.^[43]

Here, we just concern the first three where only one characteristic is lacking respectively:

YMDS 752 / 141c12–16:

闕一有三者：如數論師，對聲論立：「聲是無常，眼所見故」，聲無常宗，瓶、盆等為同品，虛空等為異品，此但闕初而有後二；聲論對薩婆多立：「聲為常宗，所聞性故」，虛空為共同品，瓶、盆等為異品，闕第二相；「所量性」因，闕第三相。

The three [kinds of incomplete argument where only] one [characteristic] is lacking [respectively] are for example: (1) When a Sāṅkhya, against a Śābdika, claims that “sound is impermanent, because of being visible (cākṣuṣatva, yan suo jian 眼所見).” The thesis (pakṣa) that sound is impermanent has a pot and a dish, etc., as similar instances (sapakṣa, tong pin 同品), and has ether, etc. as dissimilar instances (vipakṣa, yi pin 異品). The [argument] lacks only the first [characteristic] but has the last two [characteristics]. (2) When a Śābdika, against a Sarvāstivādin (sa po duo 薩婆多), claims that “sound is permanent–thesis, because of being audible (śrāvaṇatva, suo wen xing 所聞性).” [Here,] ether is the similar instance equally [accepted by both sides] (gong tong pin共同品). A pot and a dish, etc., are dissimilar instances. [The argument] lacks [only] the second characteristic. (3) The reason “being cognizable (prameyatva, suo liang xing 所量性)” [for the thesis “sound is permanent”] lacks [only] the third characteristic.

If we present all the three arguments in a “syllogistic” manner with subject-predicate statements, skip the negative example and all the individual cases cited and refer to the “positive example” just as “example” for the sake of convenience, we end up with the following three proofs:

Proof (2)
Thesis: Sound is impermanent,
Reason: for sound is visible.
Example: Whatever is visible is impermanent.

Proof (3)
Thesis: Sound is permanent,
Reason: for sound is audible.
Example: Whatever is audible is permanent.

Proof (4)
Thesis: Sound is permanent,
Reason: for sound is cognizable.
Example: Whatever is cognizable is permanent.

It is not a surprise to find out that all the unsound arguments illustrated here and the above proof (1) as a sound argument share the same logical form. The differences are only as follows:

In proof (2), the reason-statement “sound is visible” is not true, in that sound is clearly not visible. Here, the proposition Hp is false. In this case, only the first characteristic pakṣadharmatvam, i.e., the reason’s being (pervasively) a property of the subject (bian shi zong fa xing 遍是宗法性), does not obtain or is not fulfilled.^[44]

In proof (3), the example-statement “whatever is audible is permanent” is not true. It cannot be instantiated in existent individuals apart from the subject “sound,” since only sound is audible. Here, the positive example should be interpreted as a statement with existential import, like (x) ((x ≠ p & Hx) → Sx) & (∃x) (x ≠ p & (Hx & Sx)).^[45] The whole conjunction is false just because the last conjunct is false. In this case, only the second characteristic sapakṣe sattvam, i.e., the reason’s being (certainly) present in similar instances (tong pin ding you xing 同品定有性) is not fulfilled, since no similar instance or nothing permanent apart from sound itself instantiates the reason-property “being audible.”^[46]

In proof (4), the example-statement that “whatever cognizable is permanent” is also not true. There certainly are impermanent things which are not only cognizable but also different from sound, say a pot. Here, the first conjunct in the above conjunction is false. Hence, the whole conjunction is false. In this case, only the third characteristic vipakṣe’sattvam, i.e., the reason’s being (pervasively) absent from dissimilar instances (yi pin bian wu xing 異品遍無性), is not fulfilled, since the dissimilar instances, things not “being permanent” apart from sound itself, also have the reason-property “being cognizable,” like a pot.^[47]

In all the three cases, a valid form of reasoning does not perform a role in discriminating sound arguments from unsound ones. Arguments are considered as unsound only on account of the lacking of this or that characteristic. The “probative factor” is in this theory not the logical form, but the set of the three characteristics, just as the three characteristics are proclaimed by logicians following Dignāga as sādhana, means of proof. Therefore, in identifying the three characteristics as the probative factor, the “formalist approach” is not the approach actually stepped on by Buddhist logicians.

Moreover, each case above where one characteristic is lacking or is not fulfilled can be reduced to the scenario where one premise in the argument, either the reason-statement or the example-statement, is not true. In this sense, the three characteristics concern nothing formal. They are only the definition of the truths of the reason-statement and the example-statement, i.e., the definition of the truths of the premises in an argument, in the sense that all the premises are true if and only if all the three characteristics are fulfilled. Therefore, in identifying the three characteristics as probative factor, the implied intention is that the essential factors or criteria for discriminating a sound argument from unsound one should be the truths of the premises. The theory of trairūpya is only a theorization of this implied intention. It is only in this sense that the reason together with the positive and negative examples is also proclaimed to be a “probative factor,” the sādhana. Whether the emphasis is put on the three characteristics or on the reason and the examples is only a matter of the level on which this implied intention is to be presented, the level of the meta-language or that of the object-language.

However, one might argue in favor of a formalist interpretation of Buddhist logic that: Since at least the statement of the reason and that of the example are recognized as sādhana in this new interpretation following PS, there is certainly an awareness of the logical form of an argument coming to the core in Buddhist theorization of argument. In this sense, the “formalist approach” has not been totally neglected by Buddhist logicians following this new interpretation of sādhana. As a matter of fact, what is actually at stake in this new interpretation is not the form of these two members, but their truth. As we have said above, on the one hand, a good intuition of what a sound argument looks like does not by itself amount to a theory of argument, let alone to a “formalist” theory. The Buddhist view of the three-membered argument is just a representation of this good intuition. Moreover, there is only one form which is actually elaborated in this form of Buddhist theory of three-membered argument.^[48] It is nothing but a linguistic standard for all the arguments to follow. At any rate, a formal logic does not come about when there is only one form of reasoning, which is neither compared with other equally valid forms of reasoning, nor with other invalid ones. On the other hand, to adopt an approach other than a “formalist” one and to take some factors other than the logical form itself as the theoretically most significant does not necessarily imply that the other equally necessary factors, esp. the logical form, are rejected or considered to be irrelevant to the “completeness” of an argument. To adopt one particular approach just concerns the focus of theorization and relegates other possible candidates for a “probative factor” to a position at the edge of the horizon of a given framework, not outside of it.^[49]

Therefore, to interpret the Buddhist three-membered argument merely on its face value as some Indian equivalence to the Aristotelian syllogism might well be an over-interpretation.^[50] For such an interpretation, the thesis or conclusion has to be taken into account so that a form representing the complete process of reasoning can be available for further considerations concerning its being valid or not. However, this is obviously not the intention of logicians following Dignāga, in that the thesis is explicitly excluded by them from “probative factors” and from their fundamental considerations concerning an argument’s being tenable or not.

5. Conclusion

In the development from Vasubandhu to Dignāga and the latter’s Indian and Chinese followers, and in the new interpretation of sādhana as the triple characterization of a correct reason (trairūpya) instead of the linguistic expression of a three-membered argument, what comes to the fore is a gradually clearer conception of what is essentially decisive for an argument to be good or sound. In identifying the decisive factor as the trairūpya or the truth of premises, Dignāga and his followers lead the Buddhist theory of argument to an approach that sharply different from that of the formal logic of their European colleagues. A crucial problem that remains undecided, however, is whether the “epistemic approach” or the “dialectic approach” was adopted in further historical development after Dignāga in India and China. As we have said, we leave the answer open at the present stage. I believe that a solution will come about with a comparative study of the interpretation of trairūpya by Chinese logicians and that by Indian Buddhist logicians after Dignāga, e.g. Dharmakīrti.^[51]

References and abbreviations

Primary sources

IRMS Inmyō ronsho myōtō shō 因明論疏明燈抄 (Zenju 善珠): T68, no. 2270.

NMu Dignāga, Nyāyamukha. See Katsura [1]–[7].

NP Śaṅkarasvāmin, Nyāyapraveśa. See Tachikawa 1971: 140–144; Chinese translation, T32, no. 1630.

NPCh Chinese translation of NP: Yin ming ru zheng li lun 因明入正理論, T32, no. 1630.

NPṬ Haribhadra, Nyāyapraveśakaṭīkā. See Jambuvijaya 2009: 13–55.

NPVP Pārśvadeva, Nyāyapraveśakavṛttipañjikā. See Jambuvijaya 2009: 56–126.

PS(V) Dignāga, Pramāṇasamuccaya(vṛtti), Tibetan translation. See Kitagawa 1965.

RINM Daijō hossō kenjin shō 大乘法相研神章. Chapter 10 Ryakken inmyō nisshōri mon 略顯因明入正理門 (Gomyō 護命): T71, no. 2309, 29a5–36b24.

T Taishō Shinshū Daizōkyō 大正新脩大藏經. Tokyo 1924–1935.

YMDS Kuiji 窺基, Yin ming da shu 因明大疏. See Zheng 2010; also T44, no. 1840.

YZMS Shentai 神泰, Yin ming zheng li men lun shu ji 因明正理門論述記. Woodprint edition. Nanjing 1923.

ZYS Wengui 文軌, Yin ming ru zheng li lun shu 因明入正理論疏, abbreviated as Zhuang yan shu 莊嚴疏. Woodprint edition. Nanjing 1934.

Secondary sources

Chen [1945] 1997 Daqi Chen 陳大齊, Yinming dashu lice 因明大疏蠡測 [Observations on the Great Commentary on Hetuvidyā]. Tainan 1997.

Copi and Cohen 2005 I. M. Copi and C. Cohen, Introduction to logic. New Jersey^[52] 2005.

Frauwallner 1957 E. Frauwallner, Vasubandhu’s Vādavidhiḥ. Wiener Zeitschrift für die Kunde Süd-und Ostasiens 1 (1957) 2–44.

Hayes 1988 R. P. Hayes, Dignāga on the interpretation of signs. Dordrecht 1988.

Inami 1991 M. Inami, On pakṣābhāsa. In: Studies in the Buddhist epistemological tradi-tion. Proceedings of the second international Dharmakīrti conference, Vienna, June 11–16, 1989, ed. E. Steinkellner. Wien 1991, 69–83.

Jambuvijaya 2009 Nyāyapraveśakaśāstra of baudh ācārya Diṅnāga. With the commen-tary of ācārya Haribhadrasūri and with the subcommentary of Pārśvadevagaṇi, ed. Muni Jambuvijaya. Delhi 2009.

Katsura [1]–[7] Shōryū Katsura 桂紹隆, Inmyō-shōri-mon-ron kenkyū 因明正理門論研究[A study of the Nyāyamukha]. Hiroshima Daigaku Bungakubu Kiyō, [1] vol. 37 (1977) 106–126; [2] vol. 38 (1978) 110–130; [3] vol. 39 (1979) 63–82; [4] vol. 41 (1981) 62–82; [5] vol. 42 (1982) 82–99; [6] vol. 44 (1984) 43–74; [7] vol. 46 (1987) 46–85.

Katsura 1985 Shōryū Katsura 桂紹隆, On trairūpya formulae. In: Buddhism and its relation to other religions. Essays in honour of Dr. Shozen Kumoi on his seventieth birthday. Kyoto 1985, 161–172.

Kitagawa 1965 Hidenori Kitagawa 北川秀則, Indo koten ronrigaku no kenkyū: Jinna no taikei インド古典論理学の研究–陳那(Dignāga)の体系 [A study of Indian classical logic. Dignāga’s system]. Tokyo 1965.

Oetke 1994 C. Oetke, Studies on the doctrine of trairūpya. Wien 1994.

Tachikawa 1971 M. Tachikawa, A sixth-century manual of Indian logic. A translation of the Nyāyapraveśa. Journal of Indian Philosophy 1 (1971) 111–145.

Tang 2015 Mingjun Tang, A study of Gomyō’s “Exposition of Hetuvidyā.” Text, trans-lation and comments (1). In: Logic in Buddhist scholasticism. From philosophical, philological, historical and comparative perspectives, ed. G. Paul. Lumbini 2015, 255–350.

Tang 2020 Mingjun Tang, Yin ming 因明 in Chinese Buddhism. In: Dao companion to Chinese philosophy of logic, ed. Y. Fung. Springer 2020, 407–436.

Tillemans [1991] 1999 T. J. F. Tillemans, More on parārthānumāna, theses and syllo-gisms. In: Scripture, logic, language. Essays on Dharmakīrti and his Tibetan succes-sors. Boston 1999, 69–87.

Tucci 1930 G. Tucci, The Nyāyamukha of Dignāga. Heidelberg 1930.

Zheng 1996 Weihong Zheng 鄭偉宏, Fojia luoji tonglun 佛家邏輯通論 [A general introduction to Buddhist logic]. Shanghai 1996.

Zheng 2010 Weihong Zheng 鄭偉宏, Yinming dashu jiaoshi jinyi yanjiu 因明大疏校釋、今譯、研究 [The Great Commentary on Hetuvidyā. Critical text with notes, modern translation and investigation]. Shanghai 2010.

Footnotes and references:

[1]:

Inami 1991. Here Inami explained the status of pakṣa in an argument, and the development of relevant ideas from Dignāga to Dharmakīrti, in connection with corresponding developments in the theory of pakṣābhāsa (pseudo-thesis).

[2]:

Tillemans [1991] 1999: 78, 81.

[3]:

Cf. Tachikawa 1971: 120.

[4]:

The anumāna/sādhana distinction in the NP (cf. above NP 1) comes approximately to the inference/ argument distinction in today’s logic. I discussed this in Tang 2020: 414–416. In short, by inference we mean nowadays “a process of linking propositions by affirming one proposition on the basis of one or more other propositions.” By argument we mean “a structured group of propositions, reflecting an inference” (Copi and Cohen 2005: 7). The working of an inference does not necessitate the medium of certain linguistic expressions. An inference is only a process of pure thinking, regardless of whether or not it is expressed linguistically. However, the working of an argument necessitates the medium of certain linguistic expressions. An argument should spell out an inference in certain linguistic forms which are suitable to be understood by others. Hence, I suggest “argument” as an alternative translation of sādhana in addition to the traditional translation “demonstration.” A three-membered sādhana is called a three-membered argument throughout this paper.

[5]:

Cf. Tachikawa 1971: 120.

[6]:

Cf. Tachikawa 1971: 121–122. The phrase in the last square brackets is added in the Chinese translation.

[7]:

Inami 1991: 76, n. 33; cf. NPṬ 19,5–6.

[8]:

Cf. Tucci 1930: 5–6; Katsura [1]: 109–111; Tillemans [1991] 1999: 85, n. 14; Inami 1991: 76–77.

[9]:

Cf. Tachikawa 1971: 125.

[10]:

Cf. Tachikawa 1971: 126–127.

[11]:

See also NPṬ 47,9; 47,18: sādhanadharmo hetuḥ /

[12]:

NPVP 109,21–22: sādhanaṃ cāsau dharmaś ca sādhanadharmaḥ / ka ity āha–hetur iti /

[13]:

Cf. Tachikawa 1971: 127.

[14]:

Cf. Kitagawa 1965: 277–278, n. 615.

[15]:

Cf. Tucci 1930: 40–41; Katsura [4]: 67–68.

[16]:

See Kitagawa 1965: 527,12–529,9; 277–281.

[17]:

K 152a5–6, 152b4–5: gtan tshigs bsgrub bya gñis ldan min // rjes’gro ltog pa gñis dag ste // de’i mi mthun phyogs bsal daṅ // rjes’gro med pa der snaṅ ba’o // (k. 13) rtags med sogs daṅ rjes’gro sogs // phyin ci log pa dpe ma yin // ñe bar bsdu ba ma’brel ba // ’brel pa rab tu ma bstan phyir // (k. 14); V 63a3–4, 63a7–b1: gtan tshigs bgrub bya gñis ka med // mi mthun phyogs las med ma byas // rjes’gro phyin log rnam pa gñis // ltar snaṅ rjes’gro med pa’aṅ yin // (k. 13) rtags med sogs daṅ dpe med daṅ // rjes’gro phyin ci log la sogs // ’brel par ma bstan pa yi phyir // ñer’jal’brel pa can ma yin // (k. 14) (Kitagawa 1965: 527,12–15; 529,5–8)

[18]:

Cf. Katsura [5]: 84, n. 2: anyad nirdiṣṭalakṣaṇam.

[19]:

Cf. Tucci 1930: 52; Katsura [5]: 91.

[20]:

Katsura [5]: 92.

[21]:

Cf. Hayes 1988: 231.

[22]:

See also NP 4: anumānaṃ liṅgād arthadarśanam / liṅgaṃ punas trirūpam uktam / NPCh 12b29–c2:言比量者，謂藉衆相而觀於義。相有三種，如前已說。“Inference is the observation of an object through an [inferential] sign. The sign has been said [above] to have three characteristics.” Cf. Tachikawa 1971: 128. The word *sādhanahetu is also recurrent in NMu 10.14 on prāptyaprāptisama and ahetusama. In the corresponding PSV passage, it has been replaced completely by gtan tshigs (hetu). Cf. Katsura [7]: 46, nn. 3–4.

[23]:

Cf. Chen [1945] 1997: 4–12; Zheng 1996: 29–32, 173–176.

[24]:

Note, Vasubandhu’s concept of sādhya is different from that of Dignāga in that only the inferable property but not the whole thesis-statement is said to be what is to be proved. Moreover, Vasubandhu’s concept of pakṣa is also slightly different from that of Dignāga in that only the subject is said to be pakṣa. Cf. Frauwallner 1957: 33, frg. 1–3: pakṣo vicāraṇāyām iṣṭo’rthaḥ. sādhyābhidhānaṃ pratijñeti pratijñālakṣaṇam. me daṅ sa bon daṅ mi rtag pa ñid rnams rjes su dpag par bya ba ñid du dper brjod pa’i phyir chos tsam rjes su dpag par bya ba ñid du mṅon par’dod do źes rtogs par bya’o. “The pakṣa is the object one wishes to investigate. The definition of proposition (pratijñā) is that the proposition is the expression of what is to be proved (sādhya). Examples for the definition of what is to be inferred (anumeya = sādhya) is said to be fire, seed and impermanence. Hence, it shall be understood that only the property (dharma) is intended [here] as the definition of anumeya.” Cf. Frauwallner 1957: 16.

[25]:

Cf. Frauwallner 1957: 16, n. 21.

[26]:

See the above cited passages NP 2, NMu k. 1a and NMu 1.1.

[27]:

NP 2: hetus trirūpaḥ / kiṃ punas trairūpyam / pakṣadharmatvaṃ sapakṣe sattvaṃ vipakṣe cāsattvam iti // NPCh 11b6–7: 因有三相。何等為三？謂遍是宗法性，同品定有性，異品遍無性。 For translation and discussion, see Tachikawa 1971: 121; Katsura 1985: 161–162.

[28]:

Cf. Tucci 1930: 44; Katsura [4]: 74. The adverbs “certainly” (ding 定) and “pervasively” (bian 遍) qualifying “being present” (astitva, you 有) and “being absent” (nāstitva, wu 無) respectively are probably added in the Chinese translation. Cf. the parallel passage in PSV IV (K 150b5–7): rjes su dpag pa la yaṅ tshul’di yin par mthoṅ ste / gal te rtags’di rjes su dpag par bya ba la ṅes par bzuṅ na / gźan du de daṅ rigs mthun pa la yod pa ñid daṅ / med pa la med pa ñid dran par byed pa de’i phyir’di’i ṅes pa bskyed par yin no // (Kitagawa 1965: 521,8–13)

[29]:

Tillemans [1991] 1999: 71.

[30]:

Tillemans [1991] 1999: 71, translation slightly modified.

[31]:

Kitagawa 1965: 126, n. 154. The parārthānumāna is a verbalized inference and corresponds to the sādhana in the terminology of the NP and the NMu.

[32]:

Cf. Tucci 1930: 45–46; Katsura [4]: 76–77.

[33]:

Cf. n. 30.

[34]:

Unlike Dharmakīrti (Tillemans [1991] 1999: 72–73), Chinese tradition consistently retains the thesis-statement. The idea that the thesis-statement can be known by “implication” (artha, yi 義) or by “presumption” (arthāpatti, yi zhun 義准) is absent from Chinese sources. Nevertheless, like the tradition following Dharmakīrti, the Chinese tradition also pays a lot effort and takes a roundabout hermeneutic strategy so as to explain away the word pakṣa which always takes place at the beginning of the definition of sādhana in the NP and the NMu (cf. n. 26). The gist of such a strategy is to say that this word is so placed as to indicate exactly the aim or the object of sādhana. For details, see YMDS (54–56 / 94a21–b13) ad NP 1; YMDS (86–94 / 96c11–97b7) ad NP 2; For Wengui 文軌(early 7th century)’s similar discussion, see ZYS (1.4b–5b) ad NP 2 and ZYS (2.2a–3a) ad NP 2.4; For Kuiji’s ambiguous comment on NMu k. 13cd = PS IV. k. 6cd, see YMDS (305 / 113c6–10) ad NP 2.4. Shentai 神泰(early 7th century), the author of the only extant commentary on the NMu, says nothing on relevant passages in the NMu. On one hand, he makes reference to his commentary on the NP, which however has been lost, and on the other hand, he misleadingly ascribes this new interpretation even to Vasubandhu, see YZMS (1.3b) ad NMu 1.1. Nevertheless, he has said something on nyūnatā (incompleteness), see below, n. 38. Furthermore, the author of the NP, Śaṅkarasvāmin, who was said to be a disciple of Dignāga, did know the PS. There are certain elements in the NP which can be found only in PS, e.g. NP 3.1(9): eṣāṃ vacanāni dharmasvarūpanirākaraṇamukhena NPCh 11c7–8: 如是多言，是遣諸法自相門故≈ PSV ad PS III k. 2 (K 125a5–6): ’di yaṅ chos kyi raṅ gi ṅo bo daṅ’gal bas sel ba’i sgo tsam źig bstan pa yin la / (Kitagawa 1965: 472,14–15). Moreover, the theory of four kinds of contradictory reason (viruddha, NP 3.2.3) can only be traced to the PSV ad PS III k. 26–27 (K 133b1–134a8), cf. Kitagawa 1965: 205–217. One aspect to consider in this context is the relationship between the composition of popular manuals on logic on the one hand and more theoretical works on the other hand. Elements of innovation might sometimes be diluted with more traditional viewpoint in these popular manuals composed immediately or even several centuries after the emergence of that idea. The Hetutattvopadeśa of Jitāri, which juxtaposes the NP with Dharmakīrti’s Nyāyabindu, may be thought of as one example for this tendency.

[35]:

Tillemans [1991] 1999: 81.

[36]:

For this development, see Tillemans [1991] 1999: 75.

[37]:

Cf. Tillemans [1991] 1999: 85, n. 14. When commenting on this passage, Shentai (YZMS 1.4a–b) gives three different explanations of “incompleteness,” among which the first two correspond respectively to that of Vasubandhu and that of scholars after Vasubandhu (but before Dignāga), and the last one to that of Dignāga and his followers. However, he attributes the first two explanations both ambiguously to a “cer-tain master,” cf. Tucci 1930: 6, n. 5. The definition of “incompleteness” in the NP is basically the same as that in the NMu, see NP 6: sādhanadoṣo nyūnatvam / pakṣadoṣaḥ pratyakṣādiviruddhatvam / hetudoṣo’siddhānaikāntikaviruddhatvam / dṛṣṭāntadoṣaḥ sādhanadharmādyasiddhatvam / tasyodbhāvanaṃ prāśnikapratyāyanaṃ dūṣaṇam // NPCh 12c12–15: 謂初能立缺減過性、立宗過性、不成因性、不定因性、相違因性及喻過性，顯示此言，開曉問者，故名能破。 For an English translation, see Tachikawa 1971: 129. NPVP (124,8–12) ad NPṬ (54,12–13) on this passage: sādhanadoṣo nyūnatvaṃ sāmānyeneti / nyūnatvaṃ pakṣādyavayavānāṃ yathoktalakṣaṇarahitatvaṃ pramāṇabādhitatvam iti yāvat / ayam arthaḥ–sādhanavākye’vayavāpekṣayā nyūnatāyā atiriktatāyāś ca sabhāsadaḥ purato’bhidhānaṃ yat tat sāmānyena dūṣaṇam / viśeṣatas tu pakṣadoṣodbhāvanam asiddhaviruddhānaikānti-kadoṣodbhāvanaṃ dṛṣṭāntadoṣodbhāvanaṃ vā dūṣaṇam iti / “The fault of the sādhana in general (sāmānyena) is incompleteness. That is to say, incompleteness is the fact that the members including the thesis and others lack [one or more of] the above mentioned definitions, or is contradicted by [other] means of valid cognition (pramāṇa). This means: refutation is any expression in general of the incompleteness or redundancy (atiriktatā) with regard to any member in the statement of a sādhana in front of a witness (sabhāsad); in particular (viśeṣatas), however, refutation is either to point out the fault of thesis, or to point out the fault of unestablished, contradictory or inconclusive [reason], or to point out the fault of example.”

[38]:

Tillemans [1991] 1999: 85, n. 15, translation slightly modified.

[39]:

Cf. RINM 30c29–31a2: 且「能立」者，即有二義：一一因二喻，二因一喻二。一因二喻，約因三相也；因一喻二，約因二喻也。“Here, the sādhana has two meanings: (1) one reason with two examples (yi yin er yu 一因二喻); and (2) the reason, the first, with the examples, the last two (yin yi yu er 因一喻二). The ‘one reason with two examples’ concerns the three characteristics of a correct reason. The ‘reason, the first, with the examples, the last two’ concerns the reason together with two examples.” Gomyō 護命 (750–834 CE)’s division between 一因二喻 and 因一喻二 in this passage is quite obscure. However, I have kept the room for such a subtle division in translating these two expressions differently, though Kuiji in the above passage equated the three characteristics with 因一喻二but not 一因二喻.

[40]:

In fact, Kuiji is also unwilling to exclude the seventh possibility. Just a few lines after this passage, he says: 又雖有言，三相並闕。如聲論師，對佛法者，立「聲為常，德所依故，猶如擇滅。諸非常者，皆非德依，如四大種」。此「德依」因，雖有所說，三相並闕，何得非似？由此第七亦缺減過。(YMDS 58–59 / 94b28–c3) “Again, there is also the case where the statement is complete but all the three characteristics are lacking. For example, an upholder of the permanence of sound (śābdika, sheng lun shi 聲論師), against a Buddhist, claims that ‘sound is permanent, because of being a substratum of qualities (guṇāśraya, de suo yi 德所依, cf. NP 3.2.1(4)), like cessation through analytical meditation (pratisaṃkhyānirodha, ze mie 擇滅). Whatever is impermanent is not a substratum of qualities, like the four great elements (caturmahābhūta, si da zhong 四大種).’ Although there is an argument from the reason ‘being a substratum of qualities,’ it lacks all the three characteristics. How can it not be faulty? Hence, the seventh case should also be counted as incomplete.” That is to say, there could be some linguistic expressions in which all the three characteristics are lacking, although expressions of this kind have no probative force at all. Huizhao 慧沼(650–714 CE) gives a much clearer example for the seventh possibility: 如立「聲常，眼所見故」，虛空為同，盆等為異，三相俱闕。(YMDS 753 / 141c21–22) “For example, ‘sound is permanent, because of being visible.’ [Here,] ether is the similar instance. A dish, etc., are dissimilar instances. [The argument] lacks all the three characteristics.”

[41]:

See n. 38.

[42]:

The YMDS from the commentary on NP 3.3.1(1) to the end is actually written by Huizhao after the death of his teacher Kuiji, see Zheng 2010: 605.

[43]:

See YMDS 752–753 / 141c11–22.

[44]:

Cf. NP 3.2.1(1): śabdānityatve sādhye cākṣuṣatvād ity ubhayāsiddhaḥ // NPCh 11c12–13: 如成立聲為無常等，若言是眼所見性故，兩俱不成。“When one is to prove that sound is impermanent, [the reason] ‘because of being visible’ is not established for both (ubhayāsiddha).” Cf. Tachikawa 1971: 123.

[45]:

Oetke 1994: 24, ES+eva4.

[46]:

Cf. NP 3.2.2(2): asādhāraṇaḥ śrāvaṇatvān nitya iti / tad dhi nityānityapakṣābhyāṃ vyāvṛttatvān nityānityavinirmuktasya cānyasyāsaṃbhavāt saṃśayahetuḥ / kiṃbhūtasyāsya śrāvaṇatvam iti // NPCh11c22–24: 言不共者，如說聲常，所聞性故，常、無常品皆離此因，常、無常外餘非有故是 猶豫因，此所聞性其猶何等？“An uncommon (asādhāraṇa) [reason] is: ‘[Sound is] permanent, because of being audible.’ For, since this [reason] is [certainly] excluded from both the permanent and impermanent kinds (pakṣa, pin 品) [apart from the subject ‘sound’], and since anything else which is different from permanent and impermanent is impossible, this [reason] is a cause for doubt. [The question remains:] ‘What kind of [thing] has audibility?’” For translation and discussion, see Tachikawa 1971: 124; Oetke 1994: 33–35.

[47]:

Cf. NP 3.2.2(1): sādhāraṇaḥ śabdaḥ prameyatvān nitya iti / tad dhi nityānityapakṣayoḥ sādhāraṇatvād anaikāntikam / kiṃ ghaṭavat prameyatvād anityaḥ śabda āhosvid ākāśavat prameyatvān nitya iti // NPCh 11c19–22: 共者，如言聲常，所量性故，常、無常品皆共此因，是故不定。為如瓶等，所量性故，聲是無常；為如空等，所量性故，聲是其常？“A common (sādhāraṇa) [reason] is: ‘Sound is permanent, because of being cognizable.’ For, since this [reason] is common to both the permanent and impermanent kinds [apart from the subject ‘sound’], it is inconclusive (anaikāntika). [The question remains:] ‘Is sound impermanent because of being cognizable, like a pot, or permanent because of being cognizable, like ether?’” Cf. Tachikawa 1971: 124.

[48]:

This is partly demonstrated by the constant practice of transforming a negative statement into its affirmative counterpart, cf. NP 2.3: vaidharmyeṇāpi / … tadyathā / yan nityaṃ tad akṛtakaṃ dṛṣṭaṃ yathākāśam iti / nityaśabdenātrānityatvasyābhāva ucyate / akṛtakaśabdenāpi kṛtakatvasyābhāvaḥ / yathā bhāvābhāvo’bhāva iti // NPCh 11b15–18: 異法者，…，謂若是常，見非所作，如虚空等。此中常言表非無常，非所作言表無所作，如有非有說名非有。“[The example] by dissimilarity [i.e., the negative example] is … for instance, ‘whatever is permanent is observed to be non-produced, like ether.’ Here, the negation (abhāva) of being impermanent is said by the word ‘permanent,’ and the negation of being produced is said by the word ‘non-produced,’ like non-being (abhāva) is the negation (abhāva) of being (bhāva).” Cf. Tachikawa 1971: 121.

[49]:

Indeed, there are other minor faults beyond the scope of trairūpya which concern the logical form of a statement, cf. NP 3.3.1(5): viparītānvayo yathā / yat kṛtakaṃ tad anityaṃ dṛṣṭam iti vaktavye yad anityaṃ tat kṛtakaṃ dṛṣṭam iti bravīti // NPCh12b14–15: 倒合者，謂應說言，諸所作者，皆是無 常，而倒說言，諸無常者，皆是所作。“[An example where] the positive concomitance is reversed is that: One states ‘whatever is impermanent is observed to be produced,’ when he should say ‘whatever is produced is observed to be impermanent’;” NP 3.3.2(5): viparītavyatireko yathā / yad anityaṃ tan mūrtaṃ dṛṣṭam iti vaktavye yan mūrtaṃ tad anityaṃ dṛṣṭam iti bravīti // NPCh 12b25: 倒離者，謂如說言，諸質礙者，皆是無常。“[An example where] the negative concomitance is reversed is that: one states ‘whatever is corporeal (mūrta, zhi ai 質礙) is observed to be impermanent,’ when he should say ‘whatever is impermanent is observed to be corporeal’.” Cf. Tachikawa 1971: 127, 128.

[50]:

This is exactly the conclusion of Tillemans’ 1991 article (Tillemans [1991] 1999: 78–81), where he clearly shows that the thesis or conclusion constitutes “an integral part of the syllogism,” but not of the Buddhist sādhana, and that this fact foreshadows the “fundamental incommensurability” between syllogism and sādhana. To certain extent, my treatment of the concept of sādhana in Chinese literature could be regarded as presenting the matter “from a slightly different angle.”

[51]:

I have made some preliminary attempts in this direction, esp. on the concept of sapakṣa and vipakṣa in Chinese Buddhist logic, and on a logical reading of the second characteristic under the Chinese interpretation, see Tang 2015: 289–307, 323–336. For the most extensive and profound analysis of the “epistemic operator” in Indian logic and of the trairūpya, see Oetke 1994.

[52]:

NPVP 109,21–22: sādhanaṃ cāsau dharmaś ca sādhanadharmaḥ / ka ity āha–hetur iti /