Pramāṇa · Search · About · API

The Religion and Philosophy of the Veda and Upanishads v2.c40.s2: § 2. The Extant Upaniṣads

The Religion and Philosophy of the Veda and Upanishads v2.c40.s2

Language
English

§ 2. The Extant Upaniṣads

v2.c40.s2.1

The texts which have come down to us are certainly not the beginning of the intellectual activity of the time: they bear every trace in their manner and matter of representing the result of many discussions, and deviations of opinion among the priestly schools. The doctrines have been handled and rehandled, and the names to which they are ascribed cannot be taken *au pied de la lettre*: as in the case of Yājñavalkya, we cannot assume that they really prove or indicate activity in philosophical research on the part of the men who are mentioned. It is also important to note that the texts, such as they are, are, in the case of the more important at any rate, not the productions of a single hand: they are redactions, perhaps made at more than one time,¹ of varying philosophic arguments, and they therefore contain very various doctrines, which are not often consistent.

v2.c40.s2.2

¹ BAU. consists of three parts (i–ii, iii–iv, v–vi); ii. 4 is repeated in a secondary form in iv. 5. The 8 books of the Chāndogya seem each distinct. AA. consists of three parts, ii. 1–3; ii. 4–6; iii. 1–2. See especially Deussen's analyses in *Sechzig Upanishad’s des Veda*. Full bibliographies, translations, and a valuable intr. are found in R. E. Hume's *Thirteen Principal Upanishads* (1921).

v2.c40.s2.3

It is perfectly possible to arrive at a rough estimate of the relative periods of the Upaniṣads. The first place must probably be accorded to the Aitareya Āraṇyaka² in its philosophic portion, that is the first three sections of the second book, and probably the Aitareya Upaniṣad, which fills the remaining three sections of the second book, is not to be dated later than any of the other Upaniṣads: the only evidence of later date alleged by Deussen³ in the case of the Upaniṣad cannot be taken as sufficient to prove his view. After these works must certainly come the Bṛhadāraṇyaka Upaniṣad in its main portion, books i–iv, and the Chāndogya Upaniṣad. The Chāndogya is in almost every case, as far as one can judge, secondary in its versions of matter which it shares with the Bṛhadāraṇyaka Upaniṣad. But at any rate it is certain that the Taittirīya Upaniṣad, which forms books vii–ix of the Taittirīya Āraṇyaka, is later than the two last-mentioned: it exhibits the doctrine of five⁴ as opposed to three elements,⁵ and that in a way which renders the simple view that the distinction is unimportant most improbable. This is a matter of real consequence in the progress of thought,⁶ and the argument is of quite different weight than that derived by Deussen from the fact that the Aitareya Upaniṣad⁷ knows of four classes of living beings to three of the Chāndogya:⁸ the classification is merely mentioned *en passant*, and is of no moment in other Upaniṣads. The Kauṣītaki is much later than the others, except the Taittirīya, but is probably earlier than it. Its version⁹ of transmigration is a new edition of that in the two great Upaniṣads, the Bṛhadāraṇyaka¹⁰ and the Chāndogya,¹¹ and its account¹² of Gārgya is a more developed one than that in the Bṛhadāraṇyaka,¹ while in another section² it is clearly later than the Aitareya.³ Finally the Kena, which is in part metrical and is in itself a part of much longer Upaniṣad, the Jaiminīya Upaniṣad Brāhmaṇa, is no doubt to be ranked as the last of the older Upaniṣads. The Jaiminīya Upaniṣad Brāhmaṇa itself is of the nature of an Āraṇyaka rather than a mere Upaniṣad: that it does not bear the title Āraṇyaka is perhaps an idiosyncrasy of the Sāman schools, none of which actually call any text of theirs an Āraṇyaka. As an Āraṇyaka it has no claim to any very early date: it seems in fact to be certainly later than the Aitareya. The Taittirīya Āraṇyaka,⁴ on the other hand, apart from the Upaniṣad section (vii–ix and x), is of most miscellaneous content and various date. The Bṛhadāraṇyaka Upaniṣad forms part of the last book of the Çatapatha Brāhmaṇa,⁵ and the title shows the close relation of Āraṇyaka and Upaniṣad. In the case of the Chāndogya, the first two sections of the work are of the Āraṇyaka type, but as with texts attached to the Sāmaveda generally do not bear that name. The Kauṣītaki, like the Aitareya, is part of a longer Āraṇyaka.⁶

v2.c40.s2.4

² Keith, *Aitareya Āraṇyaka*, pp. 43 ff.; Winternitz, *Gesch. d. ind. Lit.* iii. 616 f.

v2.c40.s2.5

³ *Phil. of Up.*, pp. 23, 24.

v2.c40.s2.6

⁴ ii. 1.

v2.c40.s2.7

⁵ CU. vi. 2. So TU. ii. 8 is obviously later than BAU. iv. 3. 33.

v2.c40.s2.8

⁶ An advanced and profound doctrine may be early in appearance, as Yājñavalkya's view in BAU., for philosophy does not present any orderly advance of ideas, and Yājñavalkya was evidently too subtle for his age, which, however, was strongly influenced by views which it could not wholly adopt.

v2.c40.s2.9

⁷ iii. 3.

v2.c40.s2.10

⁸ vi. 3. 1.

v2.c40.s2.11

⁹ i.

v2.c40.s2.12

¹⁰ vi. 2.

v2.c40.s2.13

¹¹ v. 3–10.

v2.c40.s2.14

¹² iv.

v2.c40.s2.15

¹ ii. 1.

v2.c40.s2.16

² iii.

v2.c40.s2.17

³ iii. 3.

v2.c40.s2.18

⁴ See Keith, *Taittirīya Saṁhitā*, i. pp. lxxviii ff.

v2.c40.s2.19

⁵ It is preserved in two recensions; the later probably is the Mādhyandina, the earlier the Kāṇva, which places ÇB. x. 6. 4, 5, before the rest of the text. ÇB. xiv. 1–3 ranks as the Āraṇyaka, 4–9 as the Upaniṣad in the Mādhyandina recension. Cf. Liebich, *Pāṇini*, pp. 62 ff.

v2.c40.s2.20

⁶ Keith, *JRAS.* 1908, pp. 363 ff.; *Çāṅkhāyana Āraṇyaka*, pp. v ff.

v2.c40.s2.21

The next group of Upaniṣads is marked by the fact that it is composed of texts in metre, with very slight exceptions. It contains the Kaṭha, the Īçā, the Çvetāçvatara, Muṇḍaka and Mahānārāyaṇa,⁷ that is the last book (x) of the Taittirīya Āraṇyaka. The texts are distinguished from the previous group, not merely by their form, but by their contents. The old discussions with their tentative and confused efforts to reach definite results are replaced by a definite phraseology, in which the results of speculation have been summed up in brief dogmas, expressed often with an obvious desire for the paradoxical and the bizarre. The allegories, not rarely found in the Upaniṣads of the older class, which share them with the Āraṇyakas, disappear altogether. More marked still is the relation of the texts to the other collections: the Kaṭha bears the name of a famous school of the Black Yajurveda whose Saṁhitā is still preserved for us, but its nature is that it is really a rewriting, from a philosophical as opposed to a ritual point of view, of the story, found in the Taittirīya Brāhmaṇa, of Naciketas and the winning of boons from Death by him. The Mahānārāyaṇa is reckoned in some authorities as the tenth book of the Āraṇyaka of the Taittirīya school, but the Āraṇyaka is a body of very various and confused contents, and between it and the Upaniṣad there is no inherent connexion at all, such as always exists in the previous group. Similarly the Īçā Upaniṣad has succeeded in obtaining entry as a book (xl) of the Vājasaneyi Saṁhitā, with which it has nothing really to do, and which has been much added to in the course of time, parts of its contents in books xxxi, xxxii, and xxxiv¹ belonging in conception to the period of the Brāhmaṇas. The Çvetāçvatara and the Muṇḍaka have not even a nominal effective connexion with any extant text, beyond the fact that the Muṇḍaka uses the Atharvaveda (x. 7), and we need not suppose that they ever had any such connexion.

v2.c40.s2.22

⁷ Cf. R. Zimmermann, *Quellen d. Mahānārāyaṇa-Upaniṣad* (Leipzig, 1913); *IA.* xliv. 130 ff., 177 ff.

v2.c40.s2.23

¹ For its several parts see Scheftelowitz, *ZDMG.* lxxv. 201–12.

v2.c40.s2.24

In this group it is quite clear that the Kaṭha takes the first rank: its style is suggested already by the metrical part of the Kena and by verses found in the Bṛhadāraṇyaka in a later addition.² The Īçā is clearly dependent on the Kaṭha,³ and the Çvetāçvatara is certainly⁴ well aware of the Kaṭha, and is probably a good deal later even than the Īçā, which is free from the sectarian spirit of the Çvetāçvatara in which the philosophy leads up to the glorification of Rudra as the god *par excellence*. The Muṇḍaka uses apparently the Çvetāçvatara,⁵ and it again is probably presumed by the Mahānārāyaṇa.⁶

v2.c40.s2.25

² iv. 4. 8–21.

v2.c40.s2.26

³ Īçā 8 compared with KU. v. 13. With Īçā 9–11, cf. Kena 3.

v2.c40.s2.27

⁴ Deussen, *Sechzig Upanishad’s*², p. 289. The irregular form *brahmam* (i. 9, 12) suggests that its author was far from a Çiṣṭa, but there is nothing to show that this group of Upaniṣads comes from another *milieu* than the older texts as Jacobi (*Die Entwicklung der Gottesidee*, p. 20, n. 1) suggests.

v2.c40.s2.28

⁵ Cf. iv. 6 f. with iii. 1. 1 f. That Īçā answers the Muṇḍaka is implausible, despite Barua, *Prebuddh. Ind. Phil.*, pp. 259 f., 424.

v2.c40.s2.29

⁶ Cf. x. 2 f. with ii. 1. 8 f.

v2.c40.s2.30

The last group of important Upaniṣads consists of the Praçna, the Maitrāyaṇīya, and the Māṇḍūkya. They are in prose, but they differ essentially in every respect from the older prose Upaniṣads: the diction is in style much more elaborate than the early prose, and in the case of the Maitrāyaṇīya, which Max Müller wrongly believed early in date, the language is obviously closely allied to classical Sanskrit, which it follows in the introduction of greater development and complexity of style. The order of the Upaniṣads in this group is certain as regards the Praçna and Maitrāyaṇīya,⁷ and probable as regards the Māṇḍūkya.⁸ In philosophy they show their wide knowledge of the earlier texts, which they freely use. The Praçna is proved to be later than the Muṇḍaka by the fact that it presupposes it and quotes it.⁹

v2.c40.s2.31

⁷ vi. 5 quotes PU. v. 2.

v2.c40.s2.32

⁸ Cf. Māṇḍ. 3 with Maitr. vi. 4.

v2.c40.s2.33

⁹ iii. 5. Both texts show the use of *tha* in the 2nd plural Imperative as in Pāli and Buddhist Sanskrit, but we cannot say if this is more than an error of tradition, and it would be rash to accept on the strength of such evidence Buddhist influence, via Çaiva sects, as suggested by Hume (*Upanishads*, p. 7). For other errors in the Muṇḍaka, see below, p. 504, n. 2.

v2.c40.s2.34

With these the list of important Upaniṣads closes: but the manufacture of these works went on indefinitely down even to modern times. The later Upaniṣads claimed connexion with the Atharvaveda, and it is probable that the Muṇḍaka and the Praçna were really so connected: the others had no real bond of union, but were pleased to assume one for their own purposes. These Upaniṣads fall into four classes, according as they (1) merely develop doctrines already found in the older Upaniṣads; or (2) devote themselves to the glorification of the Yoga practices, by which religious ecstasy was produced; or (3) deal with the condition of the Sannyāsin; or (4) finally give themselves up to the glorification of Rudra-Çiva.

v2.c40.s2.35

The history of the text of the Upaniṣads is of interest, but not of importance for philosophy: the Atharvan Upaniṣads seem the first to have been collected into one set: we know of collections¹ of 34 or 52,² in which case others from the older type were added. A later collection, which arose in South India, has 108 Upaniṣads, including all the great Upaniṣads of the older type. A collection of 50 Upaniṣads was made at some unknown date and was translated into Persian in 1656 for Dara Shukoh: a literal Latin translation from the Persian made by Anquetil du Perron in 1801–2 was the means of bringing the Upaniṣads to the notice of Schopenhauer,³ and winning for them that place in the history of philosophy which they have never lost. The collection includes in 12 Upaniṣads the older texts, 26 Atharvan Upaniṣads and 8 others, and makes up the total by adding four chapters of the Vājasaneyi Saṁhitā.⁴ Of the dates of the making of these collections we have no proof: it is, however, possible that by the time of Çaṅkara (9th cent. A.D.) the Praçna and Muṇḍaka were already studied together. It is certain that he made use of all the Upaniṣads included in the three great groups, save the Maitrāyaṇīya and the Māṇḍūkya, which he perhaps did not regard as an Upaniṣad at all. To him also are ascribed, though with doubtful justice in all cases, commentaries on these Upaniṣads, except the Kauṣītaki and the Maitrāyaṇīya, and also a commentary on the Māṇḍūkya, which however is probably the work of a different Çaṅkara.⁵ The Upaniṣads used by Rāmānuja include all those of the great groups, save the Māṇḍūkya (though the Kārikā of Gauḍapāda is cited), the Subāla and Cūlikā, and once even the Garbha, Jābāla, and Mahā. The Jābāla and Paiṅgi as well as the Kārikā are mentioned by Çaṅkara.

v2.c40.s2.36

¹ Weber, *Berlin-Handschriften*, ii. 88.

v2.c40.s2.37

² The lists of Nārāyaṇa and Colebrooke; see Deussen, *Sechzig Upanishad’s*², p. 537. For various MSS. collections see Eggeling, *I. O. Catal.* i. 104 ff.; Keith, *ibid.* ii. 205 ff.

v2.c40.s2.38

³ *Parerga*, 2, § 185 (*Werke*, vi. 427).

v2.c40.s2.39

⁴ xvi, xxxi, xxxii, xxxiv.

v2.c40.s2.40

⁵ Jacobi, JAOS. xxxiii. 52, n. 2; Walleser, *Der ältere Vedānta*, p. 55. This fact invalidates much of Deussen’s argument (*Phil. of Up.*, pp. 30 ff.). For this date of Çaṅkara cf. references in JRAS. 1916, pp. 151 ff.; Garbe, *Sāṁkhya-philosophie*², p. 157, n. 2.

v2.c40.s2.41

The question of absolute dates for the Upaniṣads is one far more difficult than that of the relative order of the texts, and admits of no decisive solution. The one argument of weight is the fact, really undeniable, that the doctrines of the Upaniṣads are presupposed by the doctrine of the Buddha, and that, accordingly, if we accept the view that the opinions of the Buddha can be gathered with approximate accuracy from the older texts of the Pāli Canon, it is probable that the older Upaniṣads are, substantially at least, older than say 500 B.C.⁶ The argument is not by any means perfect: it may be criticized on the ground that even now the date of the Buddha is only established by very conjectural means, and still more so by the fact that the date of the Buddhist Canon rests on mere tradition, which is demonstrably in many respects defective, so that the frequently repeated assertion that the Nikāyas represent approximately a period shortly after the death of the Buddha is wholly unproved and almost certainly quite untrue. A more serious argument is afforded by the really antique character of the metre of the metrical Upaniṣads,¹ such as the Kaṭha, Kena, and Īçā, but here again no absolute date is arrived at, since the only standard of comparison is either the epic, which is of very uncertain date and being not *in pari materia* not an absolutely safe criterion, or Pāli texts, which, apart from their own uncertainty of date, are written in a different speech, and, therefore, certainly not cogently to be adduced as evidence.² Perhaps more secure is comparison with the metre of the Bṛhaddevatā,³ which is with much plausibility assigned to the fourth century B.C., and this certainly suggests that the metre of these texts is older.

v2.c40.s2.42

⁶ e. g. Rhys Davids, *Dial. of the Buddha*, i. pp. ix ff.; Walleser, *Die philosophische Grundlage des älteren Buddhismus*, Part 1; *contra*, Keith, *Buddhist Philosophy*, chap. i; de la Vallée Poussin, *Muséon*, 1905, pp. 213 ff.; *Bouddhisme*, pp. 29 ff.; *ERE*. iv. 179 ff.; Lévi, *Les Saintes Écritures du Bouddhisme* (1909); Franke, JPTS. 1908, pp. 1 ff.

v2.c40.s2.43

¹ Cf. Oldenberg, GN. 1909, pp. 219 ff., and his reff.; Zimmermann, IA. xliv. 130 ff., 177 ff.; Oldenberg, GN. 1915, pp. 490 ff.

v2.c40.s2.44

² Jacobi, *Rāmāyaṇa*, p. 93.

v2.c40.s2.45

³ Keith, JRAS. 1906, pp. 1 ff. Evidence of Pāli influence on the Karman of BAU. iii. 2. 13, or the BAU. iv. 3. 2 (*palyayate*), seen by Hume (*Upanishads*, p. 6), is negligible.

v2.c40.s2.46

More positive evidence cannot be found: the view that the Kaṭha Upaniṣad is older than Buddhism, because the legend resembles in some degree the legend of the Buddhist Māra,⁴ ignores the fact that the story is already found in the Taittirīya Brāhmaṇa. The suggestion of Walleser ⁵ that in the Tevijja Sutta we are to find a reference to the Aitareya, Chāndogya, and Taittirīya Upaniṣads is quite impossible,⁶ and the definite use of any particular Upaniṣad by any Buddhist Sutta has still to be proved. The similarity ⁷ of the language of the early Upaniṣads to that of Pāṇini is decidedly against a very early dating. Certainly it is wholly impossible to make out any case for dating the oldest even of the extant Upaniṣads beyond the sixth century B.C., and the acceptance of an earlier date must rest merely on individual fancy.

v2.c40.s2.47

⁴ Oldenberg, *Die Literatur des alten Indien*, p. 83.

v2.c40.s2.48

⁵ *Die philosophische Grundlage des älteren Buddhismus*, p. 67.

v2.c40.s2.49

⁶ Cf. Rhys Davids, *op. cit.*, p. 303.

v2.c40.s2.50

⁷ Cf. Liebich, *Pāṇini*, pp. 28 ff.; A. Fürst, *Der Sprachgebrauch d. älteren Upaniṣads*; W. Kirfel, *Nominalkomposition in den U.*; O. Wecker, BB. xxx. 1 ff., 177 ff.

v2.c40.s2.51

In the case of the later Upaniṣads any dating must be still more vague. Hopkins ⁸ indeed suggests the fourth century for such works as the Kaṭha, Maitrāyaṇīya, and the Çvetāçvatara, but this appears to be as regards the Maitrāyaṇīya much too early a date: style and content alike suggest that this is one of the most recent of the important Upaniṣads. In the case of the Çvetāçvatara, the date selected by Hopkins appears high in view of the fact that he is convinced that this Upaniṣad is later than the orthodox atheistic Sāṁkhya, and than the theistic Sāṁkhya-Yoga, and marks the appearance of the illusion doctrine foreign to the early Upaniṣads. But this view of his can hardly be accepted for reasons which will later be given, and there is, accordingly, no strong objection to the suggested date beyond the fact that there is no cogent evidence for it. A very different view of the date of the Çvetāçvatara, the Muṇḍaka, and the Māṇḍūkya Upaniṣads is suggested by Walleser,¹ who holds that the illusion theory is posterior to, and derived from, the nihilistic school of Buddhist thought, and that in particular the Māṇḍūkya is not necessarily anterior to the Kārikā of Gauḍapāda—taken by him as a designation of a Bengal school, not of an individual—which expounds the illusionist form of Vedānta for the first time, and which was in being by A.D. 550, but not necessarily very much earlier. This view, however, is without any sound foundation, and of these Upaniṣads no assured date can even be suggested.

v2.c40.s2.52

⁸ JAOS. xxii. 336, n. 1.

v2.c40.s2.53

¹ *Der ältere Vedānta*, pp. 5 ff. His date for the Kārikā is contrary to the tradition which makes him the spiritual grandfather of Çaṅkara; Deussen, *Sechzig Upanishad’s*², p. 574.

v2.c40.s2.54

The investigation of J. Hertel² as to the age of the Muṇḍaka Upaniṣad yields little that is positive. Its comparatively late date is assured, apart from its imitations of the Kaṭha among others, by the fact that it mentions for the first time, except perhaps for the Taittirīya Āraṇyaka,³ the seven worlds by name, that it has a list of Vedic sciences which includes the six Vedāṅgas, though that term is not used,⁴ and the Atharvaveda by that title, and uses the term Vedānta of the doctrine of the Upaniṣads. It knows in all probability the personal Brahman.⁵ In certain points Hertel indicates similarities with Jain views; thus, as the title indicates, the work seems to belong to an Atharvavedin sect which practised the habit of shaving the head or pulling out the hair,⁶ a Jain practice. The Puruṣa in the Upaniṣad is described by comparing his various members to the fire and other cosmic powers; the Jains, unlike Buddhists and Brahmans, are wont to represent the universe in human shape in their cosmographical fancies. The Upaniṣad approves asceticism, as does Jainism, and treats release as attained by one who has gained the Brahmaloka, and it is from that world that the man who is freed⁷ attains his final abode in the Jain system, and a parallel can be drawn between the terms in which the Upaniṣad and the Jain texts describe the ascent. Certain of the terms of the Upaniṣad have Jain currency (*cyu*, *nirveda*, *rāga*, *vītarāga*, *samyagjñāna*, *granthi*). It is, however, clear that in all these cases there is no reason to assume Jain priority. Nor do metrical tests do more than show what is otherwise certain, the priority of the Kaṭha.¹ Grammatical forms² of pronounced inaccuracy suggest careless composition rather than assure any definite date.

v2.c40.s2.55

² *Muṇḍaka Upaniṣad* (1924), pp. 64 ff.

v2.c40.s2.56

³ x. 27, 28; Kirfel, *Kosmographie*, p. 24*. The TA. x. is probably later than the Muṇḍaka.

v2.c40.s2.57

⁴ As it occurs in RV. Prātiçākhya (xiv. 30) and the Nirukta (i. 20), it is impossible to deduce any late date from the mere knowledge of the Vedāṅgas, but their appearance is new in the Upaniṣads as a definite group, excluding the other miscellaneous texts found in BAU. and CU.

v2.c40.s2.58

⁵ See i. 1. 1 f. and 2. 5.

v2.c40.s2.59

⁶ Çaṅkara (on iii. 2. 10) takes it as a putting of fire on one’s head, but this severity is probably not Vedic, though in keeping with Hindu and even Buddhist practices.

v2.c40.s2.60

⁷ But in Jainism freedom does not mean extinction of individual existence, as in Brahmanism; cf. von Glasenapp, *Die Lehre vom Karman*, p. 105.

v2.c40.s2.61

¹ They do not, it may be observed, really aid in dividing the text into an original and an interpolator’s hand, for the one point made by Hertel (p. 52) is clearly invalid.

v2.c40.s2.62

² e. g. *bhavate*, *parimucyanti* (pass.), *ādadāyan* (= *ādadat*), *çraddhayantaḥ*, *pravedayanti* (in sense of simple verb), *viddhi* (from vyadh), *jānatha*, *ācaratha* (imperative), *atkarvāya* (dative), *yoṣitāyām*; *vibhum* (nom. neut.) is doubtless a mere misprint; *adreçyam* is uncertain; the use of *pravadeta* for imperfect is the chief sign of later date; the irregularities have epic parallels.

v2.c40.s2.63

Another possible source of evidence is the date of Īçvarakṛṣṇa, the author of the Sāṁkhya Kārikā. It is perfectly clear that the Sāṁkhya system was substantially older than he in its substance and even in detail, so that a considerable age must be assumed for it, if Īçvarakṛṣṇa is to be assigned to the first century A.D. or even the first half of the second century A.D. Unfortunately the evidence as to Īçvarakṛṣṇa is by no means satisfactory. It seems, however, clear that he was either an older contemporary, if he is identified with Vindhyavāsa, of Vasubandhu, or still older, but this clearly, accepting the earlier date of Vasubandhu, which terminates his existence about A.D. 350, does not lead us further back than about A.D. 300, or half a century earlier.³ A further argument, however, has been adduced by S. K. Belvalkar,⁴ who holds that the commentary on the Kārikā, which was rendered into Chinese by Paramārtha about A.D. 560, was the work of Māṭhara, and is to be dated about A.D. 400. Now as the Māṭhara-Vṛtti frequently calls Īçvarakṛṣṇa Bhagavant, a term which would not readily be applied to a recent author, we may assume the date of the first century A.D. or shortly after for Īçvarakṛṣṇa.

v2.c40.s2.64

³ See Keith, *Sāṁkhya System*², pp. 79 f.

v2.c40.s2.65

⁴ *Māṭhara-Vṛtti* (*Ann. of Bhandarkar Institute*, 1924, pp. 133 ff.).

v2.c40.s2.66

Unfortunately the evidence for these conclusions is peculiarly unsatisfactory. The work for which the title of Māṭhara-Vṛtti is claimed bears this title on the strength of a single colophon only in a MS. apparently of A. D. 1400, which is a very slender piece of evidence. The careful analysis of Belvalkar himself shows that it contains up to 3 per cent. of what he regards as contaminations with Gauḍapāda, and 20 per cent. of students’ additions, including a reference to the Hastāmalaka attributed to Çaṅkara, while it gives apparently an extra Kārikā, which certainly has no real claim to be original. In these circumstances, instead of claiming that we have the Māṭhara-Vṛtti established as the source of Paramārtha and Gauḍapāda, the natural conclusion seems to be that this is a compilation derived from the original commentary, now lost, to which the style of Māṭhara-Vṛtti has been given without authority in the lighthearted manner of scribes and owners of MSS. Practically fatal to the theory of Māṭhara as the author of the original comment is the fact that the translator nowhere gives or hints at this name, which is also ignored by Gauḍapāda. What we really do know of an early Māṭhara is that Jain texts¹ which refer to the Saṭṭhitanta add to it the Māḍhara, which means probably that Māṭhara wrote on the Ṣaṣṭitantra, with which agrees precisely the assertion of Guṇaratna Sūri in his commentary on the Ṣaḍdarçanasamuccaya that Māṭhara's Bhāṣya was Ṣaṣṭitantroddhāra-rūpa. Otherwise, the silence regarding Māṭhara as a commentator of the Kārikā is complete. We are, therefore, left without any evidence of the real authorship of the original commentary, which, if not by Īçvarakṛṣṇa himself, need not go back earlier than A.D. 500.

v2.c40.s2.67

¹ Nandi and Anuyogadvāra Sūtras in Weber, *Ind. Stud.* xvii. 9; *Berlin Catal.* ii. 697.

v2.c40.s2.68

Mention should perhaps be made of the argument deduced by Jacobi² from the Kauṭilīya Arthaçāstra from which he infers that the Sāṁkhya and Yoga systems were in existence as well as the Pūrva Mīmāṁsā and the materialist system of the Lokāyata by the fourth century B.C. Unhappily the evidence adduced for this is not satisfactory, as the date of the Arthaçāstra must probably³ be placed some centuries after the date of the minister of Candragupta under whose name it passes.

v2.c40.s2.69

² SBA. 1911, pp. 732 ff.; Oldenberg, *Die Kultur der Gegenwart*², I. v. 32–4.

v2.c40.s2.70

³ Keith, JRAS. 1916, pp. 130–7; 1920, p. 628; Jolly, ZDMG. lxviii. 355–9 and pref. to ed. (Lahore, 1923); Winternitz, *Gesch. d. ind. Lit.* iii. 518; O. Stein, *Megasthenes und Kauṭilya* (SWA. 1921). Jacobi's defence of his thesis (ZDMG. lxxiv. 254; *Die Entwicklung der Gottesidee bei den Indern*, p. 41, n. 1) is ineffective. Use of Yājñavalkya is shown by T. Gaṇapati Çāstrī in his ed. (Trivandrum, 1924), pp. 8 ff.

v2.c40.s2.71

From a literary point of view the Upaniṣads command interest by their introduction of new forms. The dialogue appears beside the simpler dogmatic exposition of the Brāhmaṇas, expanded and developed from the brief discussions in this form of ritual problems which the Brāhmaṇas occasionally present. The individual character of the new teaching, which was a mystery, naturally furthered this mode of treatment, and dialogues are recorded between very strange interlocutors as well as between teacher and pupil. Interesting is the fact of competitions in which the speakers contend against one another, in order to win favour and the prize offered by a rich prince. Or two Brahmans are enjoying the meal offered by admirers; one comes to them and wins food by setting a clever riddle. A son returns to his father to report that he has been questioned, and has found himself unable to return a reply; his father assures him that he has told him all he knows, and goes with him to seek instruction. The wise can be induced by strange causes to give of their wisdom. The geese sing the praises of Raikva; he is found under a cart, scratching himself; offers of cows he haughtily refuses, but, when Jānaçruti presents his daughter, he taunts him with not realizing that this was the boon with which he should have sought his favour.⁴ Janaka of Videha is the king who appears as the greatest of patrons, and Yājñavalkya the wisest of Brahmans; in one contest of wits alone he vanquishes nine opponents;

v2.c40.s2.72

⁴ CU. iv. 2.

v2.c40.s2.73

the head of the last, Çākalya, splits open as the result of his failure, while robbers steal his bones.¹ Women are not excluded from contests, a maiden seized by a Gandharva or even one of Yājñavalkya's wives shows herself an adept in these questions. The gods appear; Prajāpati instructs the intelligent pupil Indra and the stupid demon Virocana.² Animals lend their quota of tuition; the pupil, who watches his master's kine increase, receives instruction from the bull of the herd, from the fire, from a gander, and from a Madgu bird.³ It would, of course, be rash to trust to the historicity of the dialogues; they are attached readily to famous names, as in the Kaṭha Naciketas and Yama carry on a dialogue in the later form of verse.

v2.c40.s2.74

¹ BAU. iii. 1; cf. ÇB. xi. 6. 3.

v2.c40.s2.75

² BAU. ii. 4; iv. 5; CU. viii. 7 ff.

v2.c40.s2.76

³ CU. iv. 4 ff.

v2.c40.s2.77

There is, of course, much that contrasts strangely with the Platonic dialogue. An element of grave risk enters into such discussions, as we have seen in the case of the rivals of Yājñavalkya; to hold a wrong view often involves risk of death in an age when the curse is believed to have power to work its own end.⁴ Again of real dialectic there is hardly any; the inquirer is ignorant and makes wrong guesses, the teacher dogmatically expounds the truth, with little or no leading up to it, and with still less reasoning for its support; curiously enough, it is in the discussions between Indra and Prajāpati⁵ that we find more of the expression of uncertainty leading to a final illumination than usual. There are, however, especially in the dialogues of Yājñavalkya, signs of planning; the outcome is attained by a process of rejection of inferior views, arranged in something like a series of lower to higher.⁶

v2.c40.s2.78

⁴ JUB. iii. 8. Cf. ÇB. xi. 4. 1; GB. i. 3. 6.

v2.c40.s2.79

⁵ CU. viii. 7 ff.

v2.c40.s2.80

⁶ BAU. iii. 1 ff.; iv. 3. 23 ff.; CU. v. 12 ff.

v2.c40.s2.81

No serious advance in logic is made over the Brāhmaṇas, though faint traces of induction by simple enumeration may be traced.⁷ On the other hand, there is a certain gain in literary form; the description of the unity of the universe produces passages of much higher feeling than was possible when dealing with the sacrifice,⁸ and distinct power is shown in the invention of such designations of the Ātman as 'Not so, not so,' which excludes all particularization, while the phrase 'That thou art' not unworthily sums up the mystery of the identity of the self and the universe. In other cases we have empty mysticism as in the unintelligible *jalān*⁹ or *tadvanam*¹⁰ as expressions of the absolute. One distinctive feature of value is the use of similes, often of complexity and interest; they are treated as conclusive arguments, though often only fanciful analogies, and though as always, if these are pressed, they tend to confuse the precise meaning of the text, and have afforded cause for doubt both to ancient and to modern interpreters.¹¹ Riddles, inherited from the sacrificial tradition, are curiously rare, though not unknown, and in the earlier Upaniṣads the employment of verse is restricted, though it occurs here and there in the Bṛhadāraṇyaka and the Chāndogya. Interesting as presaging the manner of the Pāli texts is the constant repetition of the same idea in slightly different forms; thus whatever is said of one sense is normally repeated verbatim, with only the essential changes, for another; the generic terms which would have spared the tedium have not yet been invented or even felt to be necessary and appropriate.¹

v2.c40.s2.82

⁷ CU. vi. 1. 4–6; above, Chap. 27, § 5.

v2.c40.s2.83

⁸ CU. iii. 14; ÇB. x. 6. 3.

v2.c40.s2.84

⁹ Böhtlingk, BSGW. 1896, pp. 159 f.; 1897, p. 83.

v2.c40.s2.85

¹⁰ Kena 31; Hopkins, JAOS. xxii. 362.

v2.c40.s2.86

¹¹ *e. g.* BAU. ii. 4. 7 ff.; Oldenberg, *Die Lehre der Upanishaden*, p. 183, n. 1. The idea that the late Muṇḍaka (iii. 2. 4) has Liṅga in the technical logical sense (Radhakrishnan, *Ind.* *Phil.* i. 263) is implausible. Similarly Vyaṣṭi and Samaṣṭi in BAU. iii. 3. 2 can hardly denote distinction and generality, or distributive and collective aggregates, or Siddhi, in KU. iii. 3, proof.

v2.c40.s2.87

¹ Cf. the Canonical Buddhist style; Oldenberg, GN. 1912, pp. 123 ff.

Linked texts (38)