The Structure of the Sarvadarśanasaṃgraha

This article shows in detail that the widely held view according to which the Sarvadarśanasaṃgraha has a hierarchical structure is mistaken. It further argues that at least some parts of the texts were independent essays before being incorporated into the Sarvadarśanasaṃgraha.

"The Advaitin Mādhava's … highly influential Sarvadarśanasaṃgraha … presented a hierarchical schema of sixteen philosophical schools beginning with the materialist philosophy of Cārvāka … and culminating, predictably, in Advaita Vedānta." Other authors who made similar claims are Erich Frauwallner, 4 Gerald Larson (2008, p. 282: "Altogether [Mādhavācārya] treats some sixteen systems in what appears to be a hierarchical manner"), Alberto Pelissero (2020) and Akihiko Akamatsu (Akamatsu 2017, p. 163: "Mādhava … places his own school [Advaita] at one extreme and ranks other schools' teachings by their degree of conflict with his own system.") We may here also quote Robert Goodding's doctoral dissertation, which has not been published but is available on the internet (2002, p. 8): "His Sarvadarśanasaṃgraha is an arrangement of the various positions in Indian philosophy that Mādhava knew starting from the materialist Cārvākas and Buddhists that he thought had the least validity, up to the Pātañjalīya Yoga system and Śaṅkara's Advaita that is the highest expression of the truth." An inspection of the text reveals that this understanding of the structure of the text is not supported by any evidence. Given that the highest truth finds expression in a form of Vedānta (Śaṅkara's Advaita Vedānta to be precise), we would expect that the philosophies of Rāmānuja and of Madhva, which are also forms of Vedānta, find a place near the end of the book. In reality they are discussed in chapters 4 and 5, far from the end. And the claim that Sāṃkhya and Yoga-discussed in chapters 14 and 15 respectively-are close to Advaita Vedānta is not based on any statements in the Sarvadarśanasaṃgraha. On the contrary, these two philosophies are presented as the main representatives of a vision of the world, pariṇāmavāda, that is the very opposite of the vivartavāda that Ś aṅkara adheres to. 5 I'll return to this below. There is no trace of a hierarchy in the remaining chapters. 6 How then could the idea of the Sarvadarśanasaṃgraha as a hierarchically structured text arise? I think that the way the text presents the first four philosophies -Materialism, Buddhism, Jainism and the school of Rāmānuja-goes some way to explain this misunderstanding. The philosophy of chapter 1 (Cārvāka) is refuted at the beginning of chapter 2 (Buddhism); the philosophy of chapter 2 is refuted at the beginning of chapter 3 (Jainism); the philosophy of chapter 3 is refuted at the beginning of chapter 4 (Rāmānuja). But there it stops. 7 Chapter 5 (Madhva) does not refute chapter 4, chapter 6 (Pāśupata) does not refute chapter 5, etc.
Suppose for a moment (contrary to the evidence) that each succeeding chapter does indeed refute the preceding one. Would this allow us to conclude that the structure of the Sarvadarśanasaṃgraha is hierarchical in the sense indicated above? Clearly, this conclusion would only be justified if this hierarchy were transitive. There is a priori no reason to think it is. The logic of the pecking order among chickens confirms this. This order, we learn, can be circular. That is to say, if chicken A is higher on the pecking order than chicken B, and chicken B higher than chicken C, this does not exclude that chicken C may be higher on the pecking order than chicken A. 8 The study of potential circularity in pecking orders goes back to H. G. Landau (1953). It is related to the fact that the relation of dominance is not transitive: A is dominant over B and B is dominant over C does not necessarily entail that A is dominant over C. The arguments used in chapter 2 to refute the philosophy of chapter 1 are not necessarily valid from the point of view of the philosophy of chapter 3; and the same might apply to all later chapters. Indeed, the arguments used in chapter 2 to refute the philosophy of chapter 1 are not valid from the point of view of Advaita Vedānta. To quote Nicholson (2015, p. 165): "This process is sometimes problematic, since the grounds by which the Buddhist disproves some Cārvāka theories are not acceptable to the Advaitin. A refutation of Cārvāka hedonism on the basis of the Buddhist doctrine of momentariness, for instance, is unacceptable, since Advaita does not accept that all entities in the world only exist for a moment before they pass out of existence." There are other factors that weaken the thesis that hierarchy determines the sequence of chapters of the Sarvadarśanasaṃgraha. We have seen that this text contains a rejection of the doctrine of modification (pariṇāmavāda) of Sāṃkhya and Yoga in favor of the doctrine of illusory manifestation (vivartavāda). Vivartavāda is a feature of Ś aṅkara's philosophy but also of Bhartṛhari's philosophy as presented in chapter 13 ("Pāṅini's philosophy"). Since the author of the Sarvadarśanasaṃgraha sides with Ś aṅkara, he must have preferred Bhartṛhari's thought to that of Sāṃkhya. He should therefore have placed the chapter on Sāṃkhya before the chapter on Pāṅini's philosophy. In fact he put it after that chapter. This is one more reason to abandon the idea that the chapters of the Sarvadarśanasaṃgraha are ordered in a hierarchical manner.
More than 60 years ago Minoru Hara (1958, p. 11) made the following observation: "It seems to me likely that what can be shown of the Nakulīśapāśupata-darśana will be found true of the other chapters of Sarvadarśanasaṃgraha also. That is, it was Mādhava's practice to take two or three basic texts of the system, quote a number of their basic definitions word for word, summarize a number of secondary doctrines in a manner as close to the original as brevity would permit and then embox the whole between an introductory paragraph and a concluding summary." Hara's observation suggests that the different chapters of the Sarvadarśanasaṃgraha are more or less independent of each other, that they contain little essays dealing with the philosophies concerned. In fact, his observation is valid for a number of chapters, but not for others. It primarily fits the chapter that Hara translated (chapter 6), but also the immediately following chapters 7, 8 and 9, which, like chapter 6, deal with the views of various followers of Ś iva. These chapters are very little "philosophical" in the sense that they almost completely lack justifications and avoid debates. They look like independent essays that could have been produced, and published, on their own. They raise the question whether all chapters must be looked upon as potentially independent treatises, so that the Sarvadarśanasaṃgraha is really a collection of independent essays, some of them minimally adjusted so as to fit into the whole.
With this in mind, let us look at the chapter on Sāṃkhya (Chap. 14). This chapter begins with a question: "How can we respect the doctrine of the illusory manifestation (vivartavāda) in view of the doctrine of modification (pariṇāmavāda) propounded by the Sāṃkhyas, which is its watchful enemy (paripanthin)?" 9 This is an unusual beginning. It implicitly criticizes a central feature of Sāṃkhya philosophy-pariṇāmavāda-even before its presentation. This shows that at least this chapter is more than an independent essay on the philosophy concerned. A closer look at the text reveals that the chapter on Sāṃkhya is not independent at all.
The chapter on Sāṃkhya does not answer the question raised at its beginning. It consists, as a matter of fact, of a brief presentation of the existents accepted in Sāṃkhya followed by a detailed discussion of the proofs that justify these. The question raised at the beginning is not answered in the then following chapter on Yoga (Chap. 15) either. It is answered at the beginning of chapter 16, on Ś aṅkara's philosophy. This chapter begins: "This doctrine of modification (pariṇāmavāda) should be censured by those who follow valid means of cognition (prāmāṇika)." 10 In fact, the first 63 lines (out of 918) of this chapter are nothing but a detailed refutation of the arguments in favor of pariṇāmavāda that had been given in the chapter on Sāṃkhya, explicitly referring back to passages of that chapter. 11 They do not refer back to the chapter on Yoga. They conclude with the observation that, for all the reasons given, pariṇāmavāda must be abandoned; instead, vivartavāda must be accepted. 12 To end it all, they quote a verse by Vācaspati that states that this manifold is an illusory manifestation (vivarta) of Brahma, who does not undergo modification (apariṇāmin). 13 It seems clear that the chapter on Sāṃkhya (Chap. 14) along with the first 63 lines of chapter 16 once constituted a unit, in which a presentation of Sāṃkhya is emboxed by passages that criticize its central doctrine: the pariṇāmavāda. This unit contained a detailed presentation and subsequent refutation of that doctrine. 14 We shall refer to it as the "Presentation and Refutation of pariṇāmavāda". To all 9 P. 311 ll. 14.1-2: atha sāṃkhyair ākhyāte pariṇāmavāde paripanthini jāgarūke kathaṃkāraṃ vivartavāda ādaraṇīyo bhavet. Unless otherwise indicated, references are to Abhyankar's edition. 10 P. 389 l. 1: so 'yaṃ pariṇāmavādaḥ prāmāṇikagarhaṇam arhati. 11 For details, see Bronkhorst (forthcoming). 12 P. 393 ll. 54-55: tasmāt pariṇāmaḥ parityaktavyaḥ/ svīkartavyaś ca vivartavādaḥ/. 13 P. 393 ll. 61-62: vivartas tu prapañco 'yaṃ brahmaṇo 'pariṇāminaḥ/ anādivāsanodbhūto na sārūpyam apekṣate//. 14 It is not impossible that the final line of chapter 14, which characterizes Sāṃkhya as nirīśvara and thus introduces the following chapter on seśvarasāṃkhya (Yoga), was added. appearances, it was once a separate text, whose purpose was to analyze and reject the pariṇāmavāda. As we have seen, it says so at the beginning and repeats it at the end.
This text, to repeat it once more, contained a presentation of the Sāṃkhya philosophy (as the embodiment of pariṇāmavāda), followed by its refutation. It appears that the author of the Sarvadarśanasaṃgraha, whoever he was, chose the Presentation for his chapter on Sāṃkhya, with no or few changes (he did not remove the initial question that allows us to reconstruct what happened). However, he separated the Presentation from the Refutation by inserting a chapter on Yoga in between, with the following result: a Presentation of Sāṃkhya (the main embodiment of pariṇāmavāda), followed by a presentation of Yoga, followed in its turn by a Refutation of pariṇāmavāda. The presentations of Sāṃkhya and Yoga became chapters of the Sarvadarśanasaṃgraha (chapters 14 and 15 respectively); the refutation/Refutation of pariṇāmavāda could not be part of these chapters, and followed, after the exposition of Yoga.
Here, then, we have a case where the author of the Sarvadarśanasaṃgraha appears to have used a pre-existing unit of text and incorporated it in his book. 15 He used most of this unit in his chapter on Sāṃkhya. What remained he put in the one but next chapter. In between these two parts, he inserted a long chapter on Yoga.
At this point we are confronted with a difficulty. As I pointed out above, most manuscripts of the Sarvadarśanasaṃgraha have only 15 chapters. Chapter 16, on Ś aṅkara's philosophy, is missing in these manuscripts. And chapter 15, on Yoga, ends with the following sentence 16 : itaḥ paraṃ sarvadarśanaśiromaṇibhūtaṃ śāṅkaradarśanam anyatra likhitam ity atropekṣitam iti The philosophy of Ś aṅkara, which comes hereafter and is the crest-jewel of all philosophies, has been described elsewhere and is therefore here left out. This suggests that there was an independent text expounding Ś aṅkara's philosophy that subsequently came to be incorporated into the Sarvadarśanasaṃgraha as chapter 16. This, however, is impossible. Chapter 16 is not an independent text on Ś aṅkara's philosophy, because its first 63 lines are a direct response to the chapter on Sāṃkhya. If there was an independent text on Ś aṅkara's philosophy (and there is no reason to doubt this), 17 it may have been the part of chapter 16 that begins with line 64; it cannot have been the whole chapter. The first 63 lines of this chapter were not part of it. What is more, these 63 lines were inseparably connected with the preceding chapters, more precisely: with chapter 14. 15 Pre-existing does not necessarily mean that it had a different author; this question must remain open for the moment. 16 Mysteriously, Abhyankar's manuscript D-which appears to be the only one he used that contained Chap. 16 as we find it in his edition-has the sentence itaḥ paraṃ … at the end of Chap. 15. (The indications provided by the Ā nandāśrama edition are not clear enough to draw conclusions about the manuscripts it used.) This is mysterious, for what could be the point of announcing that there will be no chapter on Ś aṅkara's philosophy just before the beginning of the chapter on Ś aṅkara's philosophy? 17 Once again, independent does not necessarily mean that it had a different author; see foot note 15 above. This leaves us with a problem for which I have no unambiguous solution. It seems certain that the early textual tradition of the Sarvadarśanasaṃgraha went through an obscure period out of which two manuscript traditions arose. Most manuscripts had only 15 chapters; in the process they had removed the 63 lines that should have been inseparable from what preceded. A minority of manuscripts added the treatise on the philosophy of Ś aṅkara, but in doing so preserved those very same 63 lines that the other manuscripts had lost or discarded.
Note that these reflections are not based on a study of the surviving manuscripts of the Sarvadarśanasaṃgraha, but only on the manuscript-based editions of that text. A study of all the surviving manuscripts may throw more light on the questions here raised. Or it may not. It seems however clear that, at least in the case of his treatment of Sāṃkhya, the author of the Sarvadarśanasaṃgraha made use of an independent unit of text, which he felt free to split into two and details of which he may have modified. Also the content of chapter 16 (on Ś aṅkara's philosophy), with the exception of its first 63 lines, was once an independent unit, as is clear from the line (quoted above) that concludes the manuscripts that have only 15 chapter. These two units, as we have seen, did not coincide with chapters of the Sarvadarśanasaṃgraha: the unit on pariṇāmavāda ended up in two different chapters (Chaps. 14 and 16), and the unit on Ś aṅkara's philosophy makes up part of Chap. 16.
Can we identify more units that do not coincide with chapters? We saw that some philosophies are refuted in the chapters that follow them. The chapter on Jaina philosophy (Chap. 3), for example, is followed (not preceded!) by its refutation. This refutation is part of the following chapter, on Rāmānuja's philosophy (Chap. 4; p. 89 ll. 4.1-31), even though the refutation has little or nothing to do with Rāmānuja's philosophy (the name Rāmānuja does not occur until its very last line, l. 4.32). In other words, this refutation can be read as an appendix to the preceding chapter, just as the Refutation of pariṇāmavāda was presumably an appendix to the two preceding chapters, on Sāṃkhya and Yoga respectively. The beginning of the chapter on Jaina philosophy (Chap. 3; p. 48-55 ll. 3.1-69), similarly, contains a long refutation of the Buddhist philosophy that had been the subject matter of the preceding chapter (Chap. 2). This time the Jaina identity of this refutation is clear: it refers to Jaina doctrines. It is yet not part of the regular presentation of Jaina doctrine, which begins at line 3.69 and extends up to the end at line 3.438. The same can be said about the beginning of the chapter on Buddhist philosophy (Chap. 2; p. 16-19 ll. 2.1-40), which refutes the philosophy of Cārvāka, the subject matter of the preceding chapter (Chap. 1). This refutation is explicitly stated to come from Buddhists. I see no reason to believe that the presentation of these philosophies along with their refutations ever constituted independently existing units. The refutations in these cases, as far as we can tell, were composed by the author of the Sarvadarśanasaṃgraha while he created his work. Their presence in the text raises however another question: why are only certain philosophies refuted, while others are not?
The philosophies that are refuted are Materialism, Buddhism, Jainism and Sāṃkhya. None of the others are. It is true that some of the other chapters contain polemical passages. The chapters on the philosophies of Rāmānuja and Madhva, for example, contain extensive criticism of Advaita Vedānta. Some other chapters criticize more or less in passing certain differing views. But only Materialism, Buddhism, Jainism and Sāṃkhya (which includes Yoga; both Sāṃkhya and Yoga are forms of pariṇāmavāda) are explicitly rejected. Why only these?
I can think of only one answer. The author of the Sarvadarśanasaṃgraha looked upon these, and only these, philosophies as in some sense "unorthodox". All the other philosophies were somehow considered "orthodox" and not therefore in need of refutation.
Perhaps this division into "orthodox" and "unorthodox" is not surprising. Materialism, Buddhism and Jainism are the eternal opponents of Brahmanical thinkers throughout much of Indian intellectual history. The critical attitude of Sāṃkhya toward the Vedic sacrifice would explain that this philosophy, too, is "unorthodox". 18 Seen this way, we must distinguish two kinds of philosophies in the Sarvadarśanasaṃgraha: some are "orthodox", others "unorthodox". And even though some of the "unorthodox" philosophies occur at the beginning of the text, the presence of the two kinds of Sāṃkhya near its end shows that the schools are not hierarchically ordered.
Certain authors believe that the pair āstika-nāstika appropriately describes how the Sarvadarśanasaṃgraha categorizes its philosophies. Andrew Nicholson (2012, pp. 107-108), for example, contrasts the use of āstika and nāstika in Haribhadra's Ṣaḍdarśanasamuccaya with that in the Sarvadarśanasaṃgraha: For Haribhadra, the six āstika schools are: Buddhism, Nyāya, Sāṃkhya, Jainism, Vaiśeṣika and Mīmāṃsā. … Haribhadra … [includes] Buddhism and Jainism. Haribhadra does this by defining the term āstika differently than it is understood by Mādhava or Madhusūdana.
Though mentioning Mādhava in the above quote, Nicholson says very little about his Sarvadarśanasaṃgraha, 20 concentrating rather on the works of Madhusūdana Sarasvatī. This leaves his claim that the Sarvadarśanasaṃgraha understands the term āstika differently than Haribhadra dangling in the air. In fact, the Sarvadarśanasaṃgraha never uses the word āstika, and it uses the word nāstika only twice. An inspection of its use of nāstika leads to a surprising result.
The first occurrence of nāstika in the Sarvadarśanasaṃgraha is on p. 2 (l. 1.14), at the very beginning of the discussion of the "philosophy of Cārvāka". Here Cārvāka is called nāstikaśiromaṇi "the crest-jewel of the nāstikas". Since the philosophy of Cārvāka is also known by the name Lokāyata (also in the Sarvadarśanasaṃgraha), 21 we may conclude that this text agreed with Haribhadra in looking upon the followers of this school as nāstikas.
The word nāstika occurs a second time on p. 255 (l. 11.204), once again as part of the compound nāstikaśiromaṇi, this time in the chapter on the philosophy of Akṣapāda (= Nyāya). The compound occurs in a debate about whether God (īśvara) created the world or not. The opponent, who is not specified, argues that it just does not do to assume that God created the world, one reason being that God could have no ground to create it for himself, and that, if He created it out of compassion for others, there would be no suffering in the world. The opponent concludes this passage with some quoted lines that he attributes to "teacher Bhaṫt˙a" (bhaṭṭācārya) and that can be found in almost identical form in Kumārila Bhat˙ṫa's Ślokavārttika. 22 Immediately following this quotation, the proponent gives an answer that begins with the vocative nāstikaśiromaṇe "Oh crest-jewel of the nāstikas!" What these two passages suggest is that the author of the Sarvadarśanasaṃgraha used the expression nāstika for all those who deny the existence of God, or at any rate His role in the creation of the world. This would then include the school of Mīmāṃsā. This would be noteworthy, since Mīmāṃsā is often thought of as a bulwark of Brahmanical orthodoxy. Remember in this context that the author of the Sarvadarśanasaṃgraha (and the author of most of chapter 16, once a separate composition), is himself a follower of a form of Mīmāṃsā,called uttaramīmāṃsā (p. 118 l. 4.283),brahmamīmāṃsā (p. 128 l. 5.5;p. 399 l. 16.85;p. 405 l. 16.156;p. 458 l. 16.776). The Sarvadarśanasaṃgraha itself calls ritual Mīmāṃsā pūrvamīmāṃsā (p. 118 l. 4.283; p. 267 l. 12.84), prācī mīmāṃsā (p. 257 l. 12.3), adhvaramīmāṃsā (p. 58 l. 3.107). At the same time, it seems unlikely that he would consider himself to be a nāstika. Did he think of the ritual Mīmāṃsakas as nāstikas?
Recall now that the passage beginning with the vocative nāstikaśiromaṇe "Oh crest-jewel of the nāstikas!" is put in the mouth of a defender of the philosophy of Akṣapāda, i.e. in the mouth of a Naiyāyika. It does not therefore necessarily reflect the position of the author of the Sarvadarśanasaṃgraha (unlike the first occurrence of this compound, where it qualifies Cārvāka). All we can conclude from the use of the word nāstika in the Sarvadarśanasaṃgraha is that its author looked upon the followers of Cārvāka as nāstikas and that the Naiyāyikas whose views are presented in the relevant chapter also looked upon ritual Mīmāṃsakas as nāstikas. Does this make sense?
The Sarvadarśanasaṃgraha is not the first text to suggest a link between ritual Mīmāṃsā and Lokāyata. A verse at the beginning of Kumārila's Ślokavārttika does the same 23 : For the most part Mīmāṃsā has, in this world, been turned into Lokāyata. This effort of mine is made to take it to the path of the āstikas.

Pārthasārathi comments as follows 24 :
Mīmāṃsā, though not being Lokāyata, has been turned into Lokāyata by Bhartṛmitra and others by accepting the incorrect position according to which there is no fruit, desired or not desired, of obligatory and forbidden [deeds] etc.
I have studied these and other texts elsewhere, 25 and come to the conclusion that the Lokāyata here criticized is characterized by its rejection of rebirth and karmic retribution, or more simply: of "another world" (paraloka). The grammatical commentary called Kāśikā confirms this interpretation; it accounts for the words āstika and nāstika under P. 4.4.60 (astināstidiṣṭaṃ matiḥ), in the senses "he who thinks 'there is'" and "he who thinks 'there is not'" respectively, adding (Kāś I p. 448): na ca matisattāmātre pratyaya iṣyate, kiṃ tarhi, paraloko 'sti iti yasya matiḥ sa āstikaḥ / tadviparīto nāstikaḥ /. In other words, an āstika is someone who believes there is another world, whereas a nāstika does not believe that there is another world. Early ritual Mīmāṃsā fitted the bill, so that the accusation of its proximity to Lokāyata makes sense. Kumārila Bhat˙ṫa protested against this identification, and it is therefore ironic to see that the Naiyāyika in the Sarvadarśanasaṃgraha quotes from the work of this self-proclaimed āstika to support the position of the nāstikas.
Where does all this leave us? I think we can once and for all discard the notion that the sequence of chapters in the Sarvadarśanasaṃgraha reflects a hierarchical order. Furthermore, we have seen that in composing this text, use was sometimes made of more or less independently existing building blocks, which were at least to some extent adjusted to the role they had to play in the whole. This procedure can only be shown for certain portions of the text, but they raise the question whether a similar procedure was used elsewhere, too. Finally, the Sarvadarśanasaṃgraha divided its philosophies into two kinds: those that were unorthodox and needed to be refuted, and other ones that were presumably orthodox enough to be presented without judgment.
Funding Open Access funding provided by Université de Lausanne.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http:// creativecommons.org/licenses/by/4.0/.