INTRODUCTION

THE Mantrapāṭha, of which this is the first critical edition, contains the Mantras or Prayers to be recited in connection with the domestic rites and ceremonies as taught in the Apastambiya Gṛhyasūtra¹. It is included in the great corpus of Sutras, known as the Apastambiya Kalpasūtra of the Black Yajurveda (in 30 Praśnas), and forms Praśnas XXV and XXVI of this corpus, while the Gṛhyasutra itself forms the twenty-seventh Praśna. This is stated by Caundappa, who wrote a commentary on the Apastambīya Sutras in the fourteenth century A.D., in the following lines:

पञ्चविंशेऽथ षड्विंशे
गृह्यमन्त्राः प्रपञ्चिताः । प्रश्नेऽथ सप्तविंशे
स्याद् गृह्यतन्त्रविधिक्रमः ॥ *

(* See A. C. Burnell, Sanskrit MSS. at Tanjore (1880), p. 16 b, also in the Indian Antiquary, vol. i, p. 5 seq.. The Paribhashās form part of Praśna XXIV, not Praśna XXV (as stated by Dr. Būhler in S. B. E., vol. ii, p. xii, and repeated by Prof. Oldenberg, S. B. E., vol. xxx, p. xxix). See Prof. Max Mūller in S. B. E., vol. XXX, p. 312 seq.)

TITLE.

Tradition does not seem to have fixed upon any definite title for these Mantras. They are referred to simply as ’the Mantras’ (mantrāḥ), or as List of Mantras’ (mantrasamāmnāya, commentaries on Ap. Grhy. 4, 2), or as ’the chapter, or the two chapters of Mantras’ (mantrapraśnaḥ, or mantrapraśnau, or mantrapraśnadvayam, or as ’the text of the Mantras for recitation’ (mantrapāṭha). I have chosen *MANTRApāṭha,’ which is warranted by some of the MSS., as being the most convenient title, though the majority of the MSS. are in favour of Mantrapraśna[^2-1]. The Telugu edition has the title Ekagnikāṇḍamantrapraśnadvayam, and Haradatta’s commentary on the Mantras goes by the title Ekāgnikāṇḍavyākhyā, or Ekāgnikāṇḍamantrāṇāṇ vyākhyā, sometimes also Mantrapraśnabhāshya. In his commentary on Ap. Dharmas. II, 2, 3, 16 Haradatta uses the word. ‘Mantrapāṭha’ for the Mantras occurring in the Taitt. ār. X, 67 (see also Mahānārāyaṇa-Upanishad 19, 2, ed. Colonel G. A. Jacob) to be recited at the Vaiśvadeva rite, and referred to by Ap. Dharmas. II, 2, 3, 13-II, 2, 4, 8. In this case ‘Mantrapāṭha’ seems to be used in a general sense for ’text of the Mantras.’ In the same way Mantrapāṭha, as well as Mantrasamhita and Mantrabrāhmana, is found as the title of similar Collections of Prayers belonging to the different Vedic schools[^2-3].

‘Aufrecht’s Catalogus Catalogorum (1891), p. 430, gives both 13 and The title T is given by Rice, MSS. in Mysore and Coorg (Bangalore, 1884), p. 46; Benares Catalogue in the ‘Pandit,’ vol. iii, p. vii; G. Oppert, Lists of Sanskrit MSS. in Southern India, II, 2505. Būhler, Catalogue of Sanskrit MSS. in Gujarat etc., 1, p. 4, has 487913: (but is not this MS. identical with our MS. E.?) and आपस्तम्बमन्त्रसंहिता Our MS. E. has मन्त्रपाढा ; so also Kielhorn, Lists of Sanskrit MSS. purchased for Government during the years 1877-81 (Poona, 1881), p. 21. Oppert, I, 2943, gives (sic).

Our MSS. B. Bu. W. give the title ; so also Benares Catalogue in the Pandit,’ vol. iii, p. vii; P. Peterson, Second Report in Search of Sanskrit MSS. (1884), p. 178.

is given by Burnell, Tanjore MSS., p. 16b; also in the Catalogue of Sanskrit MSS. in the Sanskrit College Library, Benares (Allahabad Government Press, n. d.), p. 99. Our MS. Wh. has title Ekāgnikānda, see below, p. xxxviii.

षङ्गिराद्यैः अप्रये स्वाहा
सोमाय स्वाहा
विश्वेभ्यो देवेभ्यः
ध्रुवाय भूमाय
ध्रुवचितये अच्युतचितय
câa: gà fe amauf: etc. See Būhler, Apastamba, 2nd ed.,

II, p. 64.

A Mantrapāṭha of the White Yajurveda is quoted by P. Peterson, Second Report in Search of Sanskrit MSS. (1884), p. 173. Several MSS. of a Mantrasamhita of the White Yajurveda are given in the Catalogue of Sanskrit MSS. in the Sanskrit College Library, Benares, p. 40. A Madhyandinīya-Mantrasamhita, lithographed, appeared at Bombay, 1890. It is a modern compilation, beginning with, and containing such prayers as तर्पणमन्त्राः, पञ्चगव्यमन्त्राः, अर्धमन्त्रः स्नानमन्त्रः etc. A Samavedamantrasamhita, lithographed, was published by Pranaśamkara and Dayā- samkara (Jirṇadurge, 1888). This is quite different from the well-known Mantrabrāhmaṇa belonging to the Gobhila-Grhyasutra, though it is also intended as a prayer

THE MATERIALS FOR THE EDITION OF THE TEXT.

This edition of the Mantrapaṭha is based on the following MSS. :–

(a) MSS. of the Text, in Devanagari characters, with accents.

1. W. MS. Wilson 468 of the Bodleian Library, Oxford, containing the first Praśna only. Paper, 15 leaves (ff. 37-51 of the volume), 7 lines on a page. It begins: श्रीगणेशाय नमः ॥ प्र॒ सु ग्मंत etc. Colophon : इति and ait : 9 Probably eighteenth century. Correct.
2. B. A modern copy, procured by Dr. G. Būhler and presented by him to the Vienna University Library, where it is entered under ‘Manuscripta I, 299 (I, 83695).’ Paper, 22 leaves, 13 lines on a page. It begins: श्रीगणेशाय नमः ॥ अथ मंत्रप्रनो लिख्यते ॥ हरिः ॐ ॥ प्र सु ग्मंतां etc. The title is given on the first page as follows: आपस्तंवगृह्ये मंत्र प्रश्नमारं- भोयम् ॥ The colophon : इति मंत्रप्रश्नः समाप्तः ॥ Very good copy.
3. E. MS. No. 17 belonging to the Elphinstone College, Bombay. Paper, 33 leaves, 11 lines on a page. The title-page has the following: ( नंबर १७ ) ( सने १८७०७१ ) | ( आपस्तंबमन्त्रपाढा पत्राणि १७ - ९ - १८२६). The MS. begins: श्रीगणेशाय नमः ॥ हरिः ॐ प्र सु ग्मन्त etc. The colophon runs as follows: श्रीरस्तु ॥ इदं पुस्तकं आपस्तम्बीयमन्त्रपाढानाम्नोः लिखितं ॥ संवत् १८२६ विरोधीसंवत्सरे मार्गशिर्षवद्य ६ सोमवारे लिखितं ॥ स्वार्थ परार्थं च ॥ छ ॥ २ ॥ ५ ॥ पुस्तकमिदं गीरिधरदशपुत्रोपनामकस्य ॥ २ ॥ Not very correct.

(b) MSS. of the Text, in Grantha characters, unaccentuated.

4. Bu. Palm leaf MS., No. 50 of Burnell’s Collection in the India Office Library. See A. C. Burnell, Catalogue of a Collection of Sanskrit MSS. Part I. Vedic MSS. (London, 1870), p. 16 (lxiv). Why Burnell should have described the ‘Mantrapraśnadvaya’ as being ‘Taittiriya Aranyaka V and VI,’ I cannot understand. The MS. is fairly correct book to be used for domestic rites. The above-quoted Benares Catalogue, p. 26, mentions a Mantrapadha of the Samaveda. There are several MSS. and lithographed editions of a Mantrasamhita of the Rigveda (which does not, however, belong to the Asvalayana-Grhyasūtra, but is of a more modern character). Sayana (Rgveda- Bhashya I, I, I, Max Mūller’s 2nd ed., vol. i, p. 24, 1. 12) quotes a Mantrabrāhmaṇa of the Rigveda, by which he certainly does not mean the Aitareya-Brāhmaṇa.

5. Wh. Palm leaf MS., No. 25 of the Whish Collection in the library of the Royal Asiatic Society, 37 leaves, 6 lines on a page. The first leaf contains the following entry: The Mantra-praśnam, or Mantras for different Brahmanical ceremonies, in two Chapters of 17 and 22 Lectures-C M Whish 1822. The MS. was probably written towards the end of the eighteenth century. Very correct.

(c) MSS. of Haradatta’s Commentary on the Mantrapāṭha.

6. Hw. Palm leaf MS., No. 26 of the Whish Collection in the library of the Royal Asiatic Society, 135 leaves, 5 lines on a page, Grantha characters. Entry on the first leaf: This volume contains the Mantraprasna-bhāshyam: compleat. by Haridattacharyyah-C M Whish Calicut 1824. The MS. looks a little older than No. 25 of the same collection (Wh.). Correct.
7. Hbg. A modern copy, procured by Dr. G. Būhler and presented by him to the Vienna University Library, where it is entered as MS. 312 =I, 134855. It is written on paper in Grantha characters. Good copy.
8. Hbd. A Devanagari transcript (on paper), made for Dr. G. Būhler from a Grantha MS. of the library of the Mahārāja of Mysore, and presented by Dr. Būhler to the Vienna University Library, where it is entered as MS. 296 I, 134819. Very incorrect.
9. HHg. Palm leaf MS., from Tiruvidaimarudur, procured for me by Dr. Eugen Hultzsch. Very small Grantha characters, and difficult to read. 42 leaves, about 13 lines on a page. Size 17 x 2 in. The MS. is correct, but the writing is perfect poison to the eyes, and the MS. has therefore only been consulted for the most important passages.

(d) Other Materials.

10. P. A MS. of the Apastambīya-Samskara-Prayoga, in the Bodleian Library, Oxford (MS. Sansk. d. 1). This Prayoga gives the Mantras in extenso, but not in the order in which they occur in the Mantrapaṭha, as the ceremonies are arranged differently from the Apastambiya Grhyasutra. The MS. is imperfect, ff. 2 and 9 being missing. Consequently not all the Mantras are found in this MS. I thought it of some importance to collate the Mantras in this MS. for the sake of those grammatical peculiarities and irregularities which will have to be discussed below. It is interesting to see that even in this Prayoga MS.-which is quite independent from all the MSS. described above-it did not occur to the author or scribe to alter such strange readings as खेऽनसः खे रथः etc. It is a paper MS., in Devanagari character, probably written about 100 years ago, and was bought in 1886 by Dr. G. Thibaut, then at Benares, for the Bodleian Library.
11. T. An edition, in Telugu characters, of Haradatta’s commentary on the Mantrapāṭha, together with the texts of the Apastambiya Gṛhyasūtra, and of the Mantrapātha, printed at the Adisarasvati Press, Chennipur, 1890. The complete title is as follows:-

    Śrī Apastambamaharshaye namaḥ. Sasūtraikāgnikāṇḍa-mantra-bhashyam vivāhopanayanādi-sakala-smārta-karmopayukta-mantrārtha-jijñāsūnām atyamtopakārāya Apastambiya-gṛhyasūtram, ekāgnikāṇḍa-mantra-prasnadvayam śrī-Haradattācārya-viracita mantrabhāshyam ca sukhabodham sampuṭīkrtya, aneka-prācīna-tāļapatra-likhitagranthāvalokana-pūrvakam lekhaka-pramādādi-doṣān yathāmati pariśodhya Vāvilla, Rāmasvāmiśāstriṇā ādisarasvatī-nilayākhya-svīya-mudrā-kṣara-śālāyām mudrāpya prakaṭīkṛtam. vijayatetarām cennipuri. 1890 samvatsaram.

The first Praśna ends

प्र सु ग्मंताष्टादश । हिरण्यवर्णाः पंचदश । सोम- ष्षोडश । सोमायाष्टादश । आतिष्ठेमं त्रयोविंशतिः । सत्येन ता मंदसाना एकान्नविंश- तिरेकान्नविंशतिः । शर्मेकविंशतिः । इह गाव एकान्नविंशतिः । उदीर्ष्वीतो दश । अपश्यं त्वा द्वादश । आरोहों नव । विष्णुर्योनिं करोमि ते प्राजापत्यमष्टादशाष्टादश । प्रातरमिं चतुर्दश । इमां खनाम्युदसौ द्वादश द्वादश । अथीभ्यां विंशतिरष्टादश ॥ अचीभ्यां ते नासिकाभ्यां । उदसी सूर्यो अगात् । etc…. हिरण्यवर्णीयुचयः पावकाः । प्र सु ग्मंता धियसानस्य सचणि ॥ १८ ॥ प्र सु ग्मता तानि ब्रह्मोत शँसति ॥ हरिः ओं श्रीकृष्णार्पणमस्तु ॥ प्रथमप्रश्नस्समाप्तः ॥

At the end of II, 14, T has:

इत्येकाग्रिकांडे द्वितीयप्रश्ने चतुर्दशः खंडः द्वितीय पूर्वभागः समाप्तः ॥ It ends: द्वितीय: प्रश्नस्समाप्तः समाप्तञ्चायं प्रश्नव- यात्मक एकामिकाण्डमंचः ॥ ॐ तत्सत् ॥

These materials proved quite sufficient for the critical restoration of the text of the Mantrapāṭha, and it is not likely that more MSS., which might have been procured from India, would have been of much avail.

Commentaries

Commentaries on the Mantrapaṭha by Sāyaṇa (Vidyaraṇya), and by Sudarsanācārya, are mentioned by Dr. G. Oppert (Lists of Sanskrit Manuscripts in Southern India, vol. ii, 2083, 6790, 10089; 7263). But Dr. Hultzsch, who was kind enough to make enquiries about these MSS., could find no trace of them. ‘No Pandit,’ he writes, ‘knows anything about the existence of these commentaries.’ One very obliging Pandit, Triambaka Sastriar of Kumbakonam, indeed offered to ‘make’ a commentary of Sayana. He writes to one of Dr. Hultzsch’s correspondents:

‘But if you want “Sayaṇacharya’s bhâshya,” as almost all these mantras occur in the Rigveda and Yajurveda, I think we can write it out of these vêda bhashyas.’+++(5)+++

Let us hope that we shall never find a Bhashya of Sāyaṇa made up in this fashion, though an authentic commentary of Sāyaṇa, and also of Sudarśanarya, if they really exist, would be most valuable. As it is, we must be satisfied with Haradatta’s commentary.

There can be no doubt that Haradatta, the author of the commentary on the Mantrapāṭha, is identical with the Haradatta who wrote the Anākulā Vṛtti on the Apastambiya Grhyasūtra, and the Ujjvalā Vṛtti on the Apastambīya Dharmasūtra. Now it has been proved by Dr. Būhler that this Haradatta cannot have written later than between A.D. 1450-1500, and (if he was also the author of the Padamañjarī, a commentary on the Kāsikā, which is not at all impossible)1 may be much older.

(1 See Dr. Būhler’s second edition of the Apastambiya Dharmasutra (Bombay Sanskrit Series, No. XLIV), parti, p. viii seq. )

As none of our MSS. can claim so early a date as the fifteenth century, it follows that Haradatta’s commentary must be considered as the highest authority for the reconstruction of the text of the Mantrapāṭha. Whenever, therefore, a reading was supported by Haradatta (H.), it has invariably been adopted.

The Mitakshara-Vṛtti on the Dharmasutra of Gautama is probably the work of the same Haradatta. See Dr. Būhler in S. B. E., vol. ii (2nd ed.), p. xlvii n. For other works ascribed to Haradatta, and possibly a biography of this learned author (Haradattiya? Haradattacaritra ?), see Th. Aufrecht, Catalogus Catalogorum (Leipzig, 1891), p. 755.

• We have, of course, always to distinguish between a reading supported by a special remark of Haradatta-only these have been pointed out in the notes by ‘H.’-and readings found in any one or some of our MSS. of Haradatta. A reading found in Hw., Hbg., Hbd., or HHg. has no more, or even less, weight than that of a single text MS., as the copyists of the commentary are not so careful as those of the text.

Grantha vs Devanagari

As to the MSS. of the text, the Grantha MSS. had to be considered as of higher authority than the Devanagari MSS., for the orthographical peculiarities of the latter leave no doubt that they go back to Grantha MSS. Besides, there is every reason to assume that the Southern MSS. represent more accurately a text belonging to the school of the Apastambins. For, as Dr. Būhler has shown1, both Brahmanical tradition, and the information to be derived from inscriptions, as well as the actual state of things in modern India, prove that the Apastambins belong to Southern India. The Taittiriya literature, generally, must have flourished in the Dekhan at a very early period, and for the best Taittiriya MSS. we have always to turn to the South, where, indeed, most of these MSS. are found.

The Grantha MS. Wh. is, as a rule, more correct than Bu., the other Grantha MS., which stands nearer to B. W. E. than Wh. The MSS. E. and W. stand much nearer to each other than they stand to B.

The Telugu edition (T.) has only been collated after the text was already in type. Its value is that of a good MS., and it agrees on the whole with the text commented on by Haradatta. I have quoted it in the critical notes in all important cases. Its importance, like that of P., lies in the fact that it also confirms the authenticity of all those strange and ungrammatical readings which an editor, at first sight, feels inclined to remove by conjectural emendations, and which, nevertheless, have to be retained, as they are confirmed by Haradatta’s commentary, and by the best MSS.

• 1 See S. B. E., vol. ii (2nd ed.), pp. xxxiii-xxxvii.

• See some excellent remarks by Dr. H. Lūders, Die Vyasa-Śiksha (Kiel, 1895), P. 51.

GRAMMATICAL IRREGULARITIES.

There are numerous cases in these Mantras where every editor would be tempted to have recourse to conjectural emendations. But on closer examination he will remember that he has to edit, and not to correct his text, and that even a grammatically impossible reading has to be retained, if it is warranted by the best authority.

There canis absolute is gram- पूर्वी. ???

Take for instance I, 1, 9. Nothing would be easier than to correct this verse and restore the readings found in Rv. VIII, 91, 7. not, of course, be the least doubt that: for nonsense, that the vocative with the third person matically impossible, that yet is the correct reading, and not yet.

Yet the evidence of the MSS. and the explanations of Haradatta leave no doubt that the faulty readings are those known among the Apastambins. खेनसः खे रथः is the reading of all the MSS., and is explained by Haradatta.

The reading of W.: (sic) is clearly a weak attempt of some copyist to justify the third person by changing the vocative into a nominative. gef, again, is supported by the readings of all but one of the MSS., and by Haradatta.

As I had to edit, not to correct the prayer book of the Apastambins, I could only give the text as it is warranted by the best MSS and by the Commentator, and as we must assume that the Apastambins repeated it on the occasion of the bride’s bath, without exactly knowing themselves what they were repeating.

I, 1, 10. Here again there is not the least doubt that ??? is the correct reading, but the explicit statement of Haradatta, that d शुत्वापः is the reading known among the Apastambins (प्रसिद्धस्तु पाठः समु ufafa), forbids our adopting the correct reading given by the MSS. W. Bu., and found Av. XIV, 1, 40. We find for again I, II, II, and in I, 11, 6 Haradatta supports the reading of the best MSS. ???? for विश्रयाते by saying शकारस्य सकारः. It might be said that we could not have both and in the same verse. But the same promiscuous use of the two sibilants occurs also in some of the Aśoka inscriptions, and, as Dr. Būhler has pointed out, is due to a laxness of pronunciation to be observed in most parts of India’.

(‘The vacillation is just the same as when the inhabitants of Gujarat say in one sentence é fum kahê chhe (“what does he say?”), and in the next tamé sum kahyum (“what did you say?”). Dr. Būhler in Epigraphia Indica, vol. iii, p. 136. See also M. Bloomfield, the Kauśika-sutra, p. lx; J. Amer. Or. Soc., vol. xiii, p. cix seq.; and S. B. E., vol. xlii, pp. 255, 331; J. Wackernagel, Altindische Grammatik, p. 226.)

I, 3, 14. Following the authority of the best MSS. and of Haradatta. I had to adopt the reading सप्तपदा बभूव, though सप्तपदावभूव, the reading found in Dr. Būhler’s MS. of Bharadvāja-Grhy. I, 16, and in Hir. I, 21, 2 (ed. Kirste), is far more plausible. But of two Hiranyakeśin MSS. which I have seen myself, one reads सप्तपदा बभूव, the other पदा बभूम.

I, 8, 2. The reading i at gayamura is certainly ungrammatical, and it would be easy enough to propose emendations. We might correct उद्यमाणास् to उचमाणाँस, and all would be well. Or, we might read इरां वहन्तो घृतमुचमाणास्, taking these words as a parenthetical sentence. Compare Hir. I, 29, 2; Aśv. II, 9, 5; Śankh. III, 5 3. But the varietas lectionis and Haradatta’s commentary leave no doubt that the ungrammatical reading is the traditional reading of the Apastambins. Baudh. I, 8 has the same reading.

I, 8, 7. Here, again, the reading fat of Av. VI, 78, 2 might easily be substituted for fat, which gives hardly any sense. The latter, however, is the reading of all the MSS., and is explained by Haradatta as meaning: यथा तयोरन्यचापेचा न भवति तथा, or परमान्त- रात्मानी.

• 1 Here also the Mantra is evidently corrupt.
• • It is hardly necessary to state that if I quote in the notes a reading as occurring, e.g. in ‘B. W. Wh.,’ or in E. Bu.,’ etc., it always means that it occurs in B. W. E. with the accents, and in Wh. Bu. without the accents. The MSS. give frequently the verb with the Anudatta in a dependent clause; as will be seen from the critical notes, I have corrected the accentuation in these cases. In the accentuation of vocatives, also, the MSS. are frequently wrong, and I thought it better to follow the general rules of Vedic accentuation. Especially frequent are such cases of wrong accentuation towards the end of the book. See I, 13, 3a; II, 2, 3; 6, 14a; 12, 10a; 13, 5; 10; 14, 2d; 15, 12; 17, 1; 11; 12; 18, 35; 36: 39; 44; 20, 1; 21-23; 33; 35; 21, 19; 32b; 33; 22, 5a; Ioa.

I, 9, 1. One might think that the accents which are only found in the Devanagari MSS. have little authority, and as a matter of fact, the MSS. are not always reliable in this respect. In E., especially, the accents are often placed quite arbitrarily. Yet, although I have frequently allowed myself to deviate from the MSS. with regard to the accentuation, I have never done so without giving the readings of the MSS. in the critical notes. For we find in Haradatta’s commentary references to the accents which prove that they also were fixed by tradition. Such a case occurs I, 9, 1, where Haradatta confirms the reading of our accentuated MSS : इ॒ह गावः॒ प्र जा॑यध्वमि॒हाश्वा॑ इ॒ह पूर्वषाः । by saying सर्वचामन्त्रितनिघात छान्दसत्वान्न भवति. This verse occurs in Av. XX, 127, 12 in the following form:

इ॒ह मा॑व॒ प्र जा॑यध्वमि॒हावा॑ इ॒ह प्र॑रुषाः ।
इ॒हो म॒हस्र॑चि॒णोऽप॑ पू॒षा नि षीदति ॥

Now we find in Par. I, 8, 10, Bhār. I, 17, and Hir. I, 22, 9 the following version of the first line:

इह गावो नि षीदन्त्विहाश्वा इह पूरुषाः ।

That is to say, in order to make the first line harmonize better with the second line, the vocatives were changed into nominatives, and the second person प्र जायध्वम् was changed into a third person नि षीदन्तु In our Mantrapāṭha, however, the vocatives were changed into nominatives, while the second person was retained. The Mantra occurs also Baudh. I, 8 (agreeing with Mantrapātha ), and MBr. I, 3, 13 (agreeing with Av.), but unfortunately we do not know how the corresponding line in these texts was accentuated. As regards MBr. I, 3, 13, Sayana’s commentary he says that the second person stands for the third. person 1 - may possibly refer to गावः, अश्वाः, and पूरुषाः being accentuated as nominatives.

1 Sayaṇa on Mantra- Brahmaṇa (Royal Asiatic Society MS. Whish 85, foll. 15, 16 a ) I, 3, 13: दह गाव इति अनुष्टुवस्य च्छन्दः । गृहप्रवेशने विनियोगः । अशास्यमाना देवताः । दूह अस्मिन् गृहे गावः प्रजायचं पुत्रपौत्रादिसन्ततिद्वारेण प्रभूता भवतेति यावत् प्रार्थनायां लोट् । अब बालेयः (?) प्रथमपुरुषबहुवचनस्य व्यत्ययेन मध्यमपुरुषब- हुवचनं । अयञ्चार्त्यः । रह गृहे वधूवरसंबन्धिगावः प्रजायन्तां त उत्पाद्यन्तां तथाश्वाः तथा पुरुषाः पुत्रादयः दुर्न्निपातेपि चार्त्यः । अपि च अस्मिन् गृहे पूषापि देवता निषीदतु किंभूतः सहस्रदक्षिणः एतदुक्तं भवति येभ्यः प्रसादे गोसहस्रदक्षिणा अपि क्रतवस्संपत्स्यन्ते ॥

I, 9, 3. As we have seen that Haradatta was acquainted with a traditional accentuation of the Mantras, we may even be justified in altering the accents, against our accentuated MSS., according to Haradatta’s commentary. I have ventured to do so in this difficult Yajus, where our MSS. read प्रस्वस्थ, while the commentator explains प्र॒स्वः॑ स्थ॒ः by saying प्रखः प्रसवनशीलाः. As to स्थः for स्थ, and the puzzling form , all I can say is that Haradatta only confirms the readings of the MSS. when he says: स्थः । छान्दसो विसर्जनीयः । स्थः स्थ भवत । and प्रशोचेष्ट प्रसविषीष्ट ।

I, 10, 2. Here, again, it would have been easy enough for the commentator, as well as for the editor, to restore the readings of Rv. X, 85, 21. Haradatta knew as well as we do that faqat is the reading in the Rigveda-Samhita, yet he states that the short i of ufaafa is ‘Vedic." is the reading of all the MSS. and of Haradatta, for, the much better reading of the Rigveda. More doubtful is the reading faЯ for The MSS. both of the text and of the commentary are divided between वित्ताम् and व्यक्ताम्, but Haradatta’s explanation अर्धदत्तामसमाप्त- विवाहां is more appropriate for वित्ताम् The Devanāgarī MSS often give the reading of the Rigveda, where the Grantha MSS. have peculiar readings, and it is altogether more probable that a copyist should have substituted a reading of the Rigveda for a reading differing from it, than the other way about. For the same reasons I have adopted the reading I, 10, 1, instead of of Rv. X, 85, 22.

I, 11, 2. If there could be any doubt as to Haradatta’s thorough acquaintance with the readings of the Rigveda-Samhita, it would be removed by his note on the reading a for a. All the MSS. support the reading, and Haradatta has the following interesting remark on it The has the Anunāsika; the Bahvṛcas, however, read it as a pure vowel. In our case also, some people read an Anusvāra after the #, but pronounce the itself as a pure vowel1. Such expressions as  ’they read, they recite,’ are very significant, and go far to prove that Haradatta did not rely merely on what he saw in MSS., but based his remarks on what he heard from the mouth of Apastambins when they were reciting their prayers.

(1 ’ ऊकारस्यानुनासिकः । बहुचास्तु शुद्धमेवाधीयते । इहापि केचिदूकारात्परमनु- स्वारमधीयते । ऊकारं तु शुद्धम् ॥ The Anunasika is a nasal sound affecting or accompanying the vowel, the Anusvāra is a nasal following after the vowel. See Wackernagel, Altind. Grammatik, p. 256; Whitney, Taittiriya-Prātiśākhya, p. 8.)

I, 11, 4. Here, again, I ventured to alter the accents on the authority of Haradatta. While the accentuated MSS. have, which yields no sense at all, H. explains as being ‘chāndasa’ for . I do not believe that his explanation is right, but it shows, at any rate, that the Apastambins, heard by Haradatta, did not recite, but

This, too, is corrupt, but it is not a corruption of MSS., but one that had become stereotyped in the recitation of the Apastambins. It would not be difficult to propose emendations, but if we once begin to conjecture, there are so many possibilities that they all lose in probability. I subjoin the readings of the MSS. of H., and some possible emendations:

• भिस्सहसामइन्द्र Unaccentuated Mss.
• भिस्सह साम ईन्द्र Accentuated MSS.
• filegura de H. oftegaafas
• भि॒िस्सह॑सा म इन्द्र
• भि॑स्स॒ह सोम (hiatus) इन्द्र

Possible conjectures.

I, II, II. There is much that is ungrammatical in this Mantra. We have दशस्यन्त्वा for दशस्यन्ता, शम् for सम्, and हथः for हतः But all these readings are confirmed by Haradatta’s commentary, and are given by nearly all our MSS. Hence I should not have been justified in correcting this verse according to Rv. VIII, 31, 9, where we find the correct readings. As to for Я, see above, note to I, I, 10. H. says that for s is ‘Vedic,’ and adds that the Bahvṛcas read s.

As to दशस्वन्त्वा he says: वकार उपजनः . He explains संहथः by संहतम्, and शगुतः by कृणुतम् (लोडर्थे लट् पुरुषव्यत्यय). The verse was originally connected with some rite of the Soma sacrifice1, and when it was in the Mantrapāṭha included among the prayers to be recited at marital cohabitation after the wedding, its original meaning was no longer

The two verses Rv. VIII, 31, 8-9 would be very appropriate for an occasion where cohabitation forms part of the Soma sacrifice, as at the Mahavrata ceremony. See A. Hillebrandt, Sonnwendfeste in Alt-Indien, pp. 42-44. I agree with Pischel (Vedische Studien, I, 178 seq.) as to adhar having the meaning of yoni, but I see no reason why romaśa should not mean depa. I translate sám údho romalám hato devéshu kṛnuto dúvah, ’they join together you and sepa, and (thereby) worship the gods." understood, or only half understood. And if people repeat what they do not understand, corruptions like those occurring in our Mantra are almost inevitable.

I, 16, 2. ज्वार्चरेत् for उ॒पार्चरेत् ( Rv. X, 159, 2 ) is not only the reading of all the MSS. and T., but is also confirmed by Haradatta, who says that for is ‘Vedic,’ adding that the Bahvrcas read. It is noteworthy that the Grantha MSS. in our case write quite distinctly are, though in many cases it is impossible to distinguish between Grantha vand, the two letters being so much alike. We find also for v, as in af 1, 13, 5 for aft of Av. XIX, 8, 4. In II, 22, 4 it is almost impossible to say whether ought to be read. Following Haradatta’s note, I adopted the reading or , though the MSS. of the text are in favour of . B. supports the reading of T. प्रवणान्, which is not impossible for प्रपणान् corresponding to ч of Av. III, 15, 6. If Haradatta had read, he would have made a remark on the 2. But the most startling interchange of वकार vand occurs II, 22, 9, where the MSS. (all but one) read sza for अनु वोsहूत्, on which Haradatta remarks : युष्मदादेशस्य वसो वकारस्य च्छान्दसः पकारः ॥ Of course, when seeing पो for वो in a Grantha MS., and the regular form : immediately after, one would naturally think that this is a clerical mistake due to the similarity of the two letters vand in Grantha. But if we remember that Haradatta did not rely on MSS. alone, but heard the Apastambins recite their Mantras before he wrote his commentary, we shall have to admit that this interchange between v and is due to a dialectic pronunciation of the Apastambins, and not to a slip made by a copyist. It is highly improbable that Haradatta would have made his remarks on the authority of a single faulty MS. In fact, if he relied on MSS. at all, he must have had more than one MS., as we have seen that he mentions a various reading a for a above, p. xix. And a commentator like Haradatta knew as well as we do, that copyists are by no means infallible. If he had found for in a MS., he would have simply corrected it. His remark that is ‘Vedic for proves that he had the best authority for the reading . Moreover, we can imagine that an editor of both the text and the commentary (e.g. the editor of the Telugu edition)1 would adopt the reading ✈ on the strength of the commentary, but it is highly improbable that the writers of the MSS. E., B., and Wh. should all have faithfully and carefully followed the commentary of Haradatta in writing , which must have been as startling to any scribe as it is to us.

• 1 The Haradatta MSS. vacillate, however, in the same way as the text MSS., between प्रपणन, प्रवणन्, and प्रपणान्, but there is more authority for प्रपणन्. The readings are: प्रपणन व्यवहरन् Hw Hbg. प्रपणन् व्यवहारान् HHg. प्रवणान् व्यवहारान् T. comm. (प्रवणाँश्चरामि text of T.). प्रवणन् व्यवहरन् Hbd
• • One might even derive from +, but this is not probable.

in ver. 5 and f: in ver. 6. But the evidence of the MSS. and Haradatta’s remark ( छान्दसस्सुलोपाभावः सुमङ्गलीरिति यथा ) leave no doubt that the Apastambins-probably on account of the preceding
read

I, 16, 4. The best MS. of the text (Wh.) and the best MS. of the commentary (Hw.) read fg instead off. It is just possible that we have in this one irregularity more, but as Haradatta himself does not make any special remark about it, I thought it safer to follow the majority of MSS. which read af with Rv. X, 159, 4.

I, 16, 6. It looks strange, no doubt, that we should read

in ver. 6, while the readings of the best MSS. and Haradatta prove that we have to read in ver. 5. Compare C. R. Lanman, Noun-Inflection in the Veda, p. 376 seq. It ought II, 2, 3. I am doubtful about the reading. Both the text and the Haradatta MSS. vacillate between व्यदणन्, व्यचणन, and व्यचिणन्. Haradatta explains : विविधमस्रंसयन् ( viz. कापीसावस्थायां बीजनिर्हरणार्थम् ). The evidence of the MSS. is decidedly in favour of to be mentioned that the reading af is doubtful, as in the Grantha what I read may be meant for the Virama. And as the Virāma is, frequently omitted in Grantha MSS. even the reading a might be meant for ¶ In the same verse we ought, of course, to expect 1 T. reads not only अनु पो but also निवर्तो पो in 11, 22, 9, and सित्त्वा पो (for funt) in II, 22, 10! But the commentary in T. agrees with our MSS.

• Could it be corrupted from ? Or is it formed after the analogy of such forms as अग्मन् अनन्, etc. ? INTRODUCTION. xxiii (as in Mantra 5), but the evidence of the MSS. leaves no doubt that the Apastambins read in this verse. The text MSS. have all, so also the best Haradatta MS. (Hw.), and HHg. reads. FAR FAR, which I believe to be Haradatta’s own reading explaining the irregular by the regular form. The accents had to be corrected, as cakŕtan is impossible. II, 2, 8. The reading of Av. II, 13, 3; XIX, 24, 6 ardi afuer: is, no doubt, the correct reading of the verse, but it is not the reading of the Apastambins. It is true that in Grantha it is frequently impossible to distinguish the letters d and th. But it cannot be a mere accident that all the MSS., both Devanagari and Grantha, read distinctly , not. The Telugu edition1, too, and the Prayoga MS. read :, and, above all, Haradatta renders पर्यधिधाः by परिहितवानसि and remarks: थकारस्य धकारश्छान्दसः Hiranyakesin also reads अधिधाः (though one of Prof. Kirste’s MSS. has fT:). Perhaps, we ought to read u वासौ अधिधाः against the authority of the accentuated MSS. II, 2, 11. The form af (which is also found in Bhār., Hir., and Baudh.) is, as Prof. Kirste (note to Hir. I, 4, 6) suggests, probably a Prakritic form for, which occurs Śankh. II, 1, 30. But it should. be added that in Sankh. also a various reading af is quoted in Prof. Oldenberg’s edition (Ind. Stud. XV, p. 48). See also Prof. Oldenberg’s note on Sankh. II, 10, 4 (S. B. E., vol. xxix, p. 75 note). II, 3, 2. Here we have again a clear instance of Haradatta’s carefulness and trustworthiness in recording readings of the Apastambins which differ from those of the Rigveda. While Rv. IV, 58, 1 and Vs. XVII, 89 read itguig, the Mantrapāṭha (agreeing with TA. X, 10, 2) reads. This difference of accentuation between the Apastambins and the Bahvṛcas is pointed out by Haradatta, who separates उप and अंशुना, and says: अनुदात्तपाठस्तु व्यत्ययेन । बह्वृचास्त्वायुदात्तमेवाधीयते । I should prefer to explain in our Mantra with Sāyaṇa as meaning ‘by the muttered prayer.’ In II, 3, 24 Haradatta again refers to the accentuation, pointing out that ‘some’ (af) read asaú instead of asau. 1 However, T. has very frequently for . xxiv INTRODUCTION. II, 6, 8. It is almost too tempting to correct fag to fa, the reading of Rv. I, 23, 24, especially if we remember that in Grantha #1 might easily be misread for. But all our MSS. (also T.) have faga, and Haradatta says that the for Visarga in faga is ‘chāndasa,’ adding that the Bahvrcas read fag:. It is certainly a mistake, but as it is not merely a clerical mistake of some copyist, it has to be retained in the text as a faulty reading of the Apastambins. II, 7, 25. The reading is not only supported by all but one of our MSS., and by Haradatta’s commentary, but it is also the reading of Hir. I, 10, 6, and I, 11, 3, and (according to Prof. Kirste) also of Bhār. II, 21. The correct reading would, no doubt, beau, which we find Pār. I, 13, 1. But the correct reading was not the reading of the Apastambins, and the mistake is in all probability older than the Sutrakāras Bhāradvāja and Hiranyakeśin. II, 8, 4. Whether is a Prakritic form for or a mistake for fu (the reading of Hir. I, 10, 6) may be doubted, but there can be no doubt that we have to read in the Mantrapāṭha, as this is the reading of all MSS., and Haradatta tells us that a for T is ‘Vedic.’ II, 8, 10. is the reading of all MSS., though we should expect Я, as Kāmāyanī is the patronymic of Śraddhā. II, 9, 1. In day we have a used as a neuter in the sense of पौरुष. A similar neglect of Vrddhi occurs in राजबन्धवी: II, 17, 26, where all MSS. have the short a, and Haradatta says led zECHLI In II, 17, 8 we find an irregular Vrddhi in आन्तरिच्या for अन्तरिक्ष्या or आन्तरिचा. Haradatta says छान्दसत्वादादिवृद्धिः । उदल II, 11, 19. If Haradatta is right, we have here for, which makes good sense. Tilada, a personification of the after-birth, is addressed: ‘Thou art neither flesh, nor (part of the) abdomen.’ Compare II, 16, 7 and 8, where according to the evidence of the MSS. fga has to be read in ver. 7, and fea in ver. 8. Two different beings seem. to be meant. Grantha MSS. always double the consonant after r. For this and other peculiarities of Grantha MSS., see my edition of the Apastambīya Grhyasūtra, p. x note. 1

• Compare उपल· and उपरि, मातुल and °मातुर, Wackernagel, Altindische Gram - matik, p. 220. INTRODUCTION. XXV II, 13, 7. Haradatta is hardly right in explaining as a nominative standing for the accusative इमं’, and in deriving पर्यध्वम् from परि + अस् ’to throw 2 Perhaps we ought to read Kali, the dice.’ seems to be used (as T is sometimes) in the sense of ’to sit round. (with hostile intentions), circumsedere. II, 14, 2. कृष्णवर्त्मने is certainly a mistake for कृष्णवर्तने. But Haradatta says हे कृष्णवर्त्मने कृष्णवर्त्मन, and all the MSS. read कृष्णवर्त्मने. Compare Apast. Grhy. 5, 23, where a is used as a feminine, as if it were af. II, 15, 3. In this verse (which occurs also in a very corrupt form in Asv. II, 8, 16 and Sankh. III, 3, 1) we find again an impossible mixture of second and third persons. Nothing would be easier than to correct स्यादिरावती to स्या हू", and तिष्ठामा to तिष्ठा मा, but it is wiser to follow Haradatta, and simply state the purushavyatyaya, without tampering with the traditional text. II, 16, 2 and 5. If I had been sure that Haradatta really read and for in these verses, I should have adopted it in the text, as T. does, for Haradatta’s authority must count for more than that of the MSS. On the other hand, this would be the only case in which a reading given by Haradatta is not supported by any of the text MSS. And as he has no special gloss on, it is doubtful whether Haradatta really read it. The MSS. of the commentary have- अव चाकशत् अपश्यत् Hw Hbg. Hbd. II, 16, 2 चाकशत् पश्यन् HHg. II, 16, 5 अव चाकशत् अपश्यन् Comm. in T. द्यामिव चाकशत् आकाशमपश्यत् Hw. द्यामव चाकंशत् आकाशं पश्यन् आगच्छत Hbg. यामवचाकशत् आकाशं पश्यन् HHg. Deest in Hbd. व्यामवचाकशत् आकाशं पश्यत् Comm. in T. Hiranyakeśin II, 7, 2 has समुद्रमिव चाकशत्. All the Mantras in this chapter are beset with difficulties, and it may be doubted whether it will ever be possible to restore the text so as to bring out a satisfactory meaning. In fact, I do not believe that those 1 द्वितीयार्थे प्रथमा । इमं कलिम् । 2 ’ अस्यतेरेतद्रूपं न पुनरासेः । पर्यस्यत । d [III. 8.] xxvi INTRODUCTION. who used these charms to cure children’s diseases or to exorcise the demons harassing little children, knew the actual meaning of the words which they recited. Whether we ought to read or aft in II, 16, 3 and 6 is quite doubtful. If gafft in Śatap. Br. XI, 4, 1, 6; 14 means ’ shaggy,’ बीरिण might mean ’ hair, and सुबीरिण ‘fine-haired.’ MSS. (Hw. Wh.) support In II, 16, 7 the reading . But the best rafa, though syntactically impossible, seems to be warranted by the best MSS. and by Haradatta’s commentary. But the MSS. are corrupt. I subjoin the readings of the Haradatta MSS.: अवधावति लकारव्यत्ययः । अधोमुखा आगच्छत । Hbd. अवधावत लकारव्यत्ययः । अधोमुखा आगच्छत । HHg. अवधावत । धकारव्यत्ययः । अथोमुखा आगच्छत । Hbg. अवधावथ । थकारव्यत्ययः । अधोमुखा गच्छत । Hw. अवधावति । धावत । लकारस्य व्यत्ययः अधोमुखा गच्छत । Comm. in T. For the metre’s sake it ought to be left out altogether, but the sense seems to required, the corrected reading of Bu. Hir. II, 7, 2 also reads अवधावति. A most puzzling form is a in II, 16, 9 seq. Haradatta, indeed, has no difficulty, he quietly explains सीसरीदत by अपसर्पत, without telling us how he arrives at this meaning. There can be no doubt that the verses must have had some sense at some time or other. But, as we find them in the Mantrapaṭha (and both Haradatta and the MSS. enable us to ascertain at least the traditional readings of the Apastambins), they defy all attempts at interpretation. The words “g seem to suggest that is the name of a male demon, and his female counterpart, and it is not impossible (as Dr. Būhler has kindly suggested to me) that we have to look for something like दुतः in सीसरीदत, while सीसरम might stand for सीसर मत्. But in conjecturing anything like that, we really go beyond the time when these Mantras were fixed in the school of the Apastambins. And if we could get at the original form of the Mantras, it would only mean that they were composed in this form, but became corrupt and intelligible, before they were received into the prayer book of the Apastambins. INTRODUCTION. xxvii II, 17, 13. Here we have again a case where an entirely impossible and ungrammatical reading is warranted both by Haradatta and by the MSS. हिनसात् is explained by हिंस्युः, and as to न्येत् Haradatta says: छान्दसो यकारः ‘. II, 22, 5 seq. Here again the reading fa: is perfectly certain, though we may hesitate to adopt Haradatta’s explanation, who says: ared Ztdci afcfua: uftât ag:. For the readings of Hiranyakeśin and Pāraskara seem to point to some other corruption. In II, 22, 10 fata is certainly the correct reading. But the false reading fat at was, no doubt, the traditional reading of the Apastambins, whether they explained to themselves the as ‘Vedic,’ as Haradatta does(), or never thought of any explanation. The grammatical irregularities which have just been pointed out will, I am afraid, prove of very little value to grammarians. A few of the cases quoted may be due to dialectic pronunciation, and thus throw some light on the phonology of the Sanskrit spoken in Southern India. And I will here add a few cases of irregular contraction and euphonic combination, which may be interesting for grammarians. In I, 3, 14 we read, and the same contraction ámūhám (=ámus+ahám? or ámú+ahám?) occurs also in Baudh. I, 12, Bhār. I, 19, and Hir. I, 20, 23, which proves that the Mantra in this form was common property of the Black Yajurvedins. Another Mantra which belongs to the Taittiriyas generally is I, 4, 16 (repeated many times, see also Hir. I, 3, 6 etc.), which occurs already in TBr. II, 4, 1, 9. Both Sayana on this passage, and Haradatta on our Mantra agree in explaining fe=ayas asi, and rayas sán. Similar contractions as ayási=ayásasi have been pointed out by the late Prof. Roth (Z. D. M. G., vol. 48, p. 678). A case of elision occurs I, 13, 9 (also Hir. I, 16, 3) in af, which Haradatta rightly explains as ná asi, i. e. ná ‘si. See Wackernagel, Altindische Grammatik, p. 318. 1 The Telugu edition omits this remark (which is found in all our MSS.) in the commentary, and reads in the text.

• See my paper, ‘Das altindische Hochzeitsrituell,’ etc. (Denkschriften der k. Akademie der Wissenschaften in Wien, Phil.-Histor. Cl. Bd. XL, Wien, 1892), p. 6. d 2 xxviii INTRODUCTION. Haradatta is no doubt right in explaining fuafa II, 14, 1 as an irregular Sandhi for पितेति = pitā eti, and क्वैष्यसि II, 22, 5 for क्ष्यसि kvà eshyasi. Compare Wackernagel, 1. c., p. 321 seq. Grammatical forms which deserve to be mentioned are perhaps the difficult form I, 4, 41, for which Av. XIV, 2, 52 has the much easier reading असृचत, and अनभिशस्तीच I, 5, 18 (also Baudh. II, 10 with the long ), and a I, 11, 2, which may possibly be a contraction for at The occurrence of af with the short i inf: I, 1, 5 (also Baudh. I, 1), and in ft I, 10, 3-6 deserves also to be mentioned.

But most of the grammatical irregularities discussed above admit of no explanation from the grammarian’s point of view-as little as the gibberish of magicians. They can only be explained by assuming that these prayers were handed down by oral tradition-probably for centuries-among people who were no longer familiar with Vedic speech. What the Jaina monk Harikeśa-Bala tells the Brahmans-‘You are only the bearers of words, as it were, you do not understand their meaning, though you have learned the Vedas ‘was probably literally true at a very early period. The authors of the Brāhmaṇas could not have used the Vedic Mantras for their fantastic speculations, if these Mantras had been clearly understood. But the charms and prayers collected in the Mantrapāṭha were not even in the keeping of learned priests, like the Samhitas of the Rigveda or Yajurveda. They were used by people of all classes for their daily devotions and for all important occasions in their daily life. And folk-lorists know how folksongs and children’s rhymes become corrupted in the mouths of uneducated people, who care more for the beauty of the tunes or the fun of the rhymes, than for the actual meaning of the verses which they repeat. Just so will charms and prayers become corrupted in the mouths of worshippers who repeat them, not on account of the meaning which they carry, but on account of their intrinsic sacredness. Indeed, the less they were understood, the more sacred would these timehonoured charms and prayers become, and the stronger the belief in 1 Baudh. I, 6 has दीचामुदास्थ But see C. R. Lanman, Noun-Inflection in the Veda, p. 412. Uttaradhyayana-sutra XII, 15 (Sacred Books of the East, vol. xlv, p. 52). INTRODUCTION. xxix their efficacy, for-as an old adage says-the gods love what is mysterious, and hate what is evident 1." That the Mantras were not understood, or only half understood, may be seen from several passages where the Mantras have absolutely nothing to do with the ceremony for which they are used 2. The very first verse of the Mantrapāṭha, which is to be recited when the wooers are sent out, has really nothing to do with this act. The words varebhir varan (vara meaning also wooer’) were the only motive for using the Mantra in this connection. C Mantra I, 2, 6, beginning with the words ‘Around thee, O Indra, may our songs be,’ has certainly no connection with the ceremony of dressing the bride. But the words after around thee’ were enough to use the Mantra when putting the new dress around the bride. Mantra I, 6, 1, ‘The earth is supported by truth,’ etc., was only prescribed for the supporting of the carriage on account of the word उत्तभिता ’ supported. ’ More passages of that kind will be pointed out in my translation of the Mantrapāṭha. But how little the Mantras were understood, may be seen best from such passages as I, 8, 8, where two lines, one taken from Rv. X, 85, 42, the other from Rv. VI, 57, 6, are joined together so as to form one Mantra, though they have nothing to do 1 परोच- परोचप्रिया व हि देवाः प्रत्यचद्विषः Brhadāranyaka Upanishad IV, 2, 2, 91141 fe 291: Śatap. Br. VI, 1, 1, 2; 11; 2, 3; 7, 1, 23; VII, 4, 1, 10; 13; 16; 2, 12; 5, I, 22. A German novelist tells of an old farm-servant who knows some verses from his hymn-book by heart, but recites them quite wrongly, making sheer nonsense of them. But when he is told how he ought to recite them, he gets angry, and refuses to adopt such daring innovations. Mein alter Jochem weiss noch einige Verse aus dem Gesangbuch auswendig. Er sagt sie immer vor sich hin, aber ganz verkehrt, und es ist purer Unsinn, was er daraus gemacht hat. Ich wollte ihm nun die Verse richtig stellen. Darūber ward er sehr bös und sagte, das wāre Neues, das gelte nicht. Sein Unsinn ist ihm lieber, er hat etwas Geheimnisvolles daran, und das imponirt ihm, weil er’s nicht versteht.’ (B. Auerbach, Auf der Höhe, vol. iii.)

• ‘It very frequently occurs in the Grhya ritual that Mantras are used at sacrifices standing in no connection with those for which they have originally been composed." Prof. Oldenberg, S. B. E., vol. xxx, p. 114 note. See also A. Hillebrandt, Ritual- Litteratur (Būhler’s Grundriss, III, 2), p. 19, and the interesting remarks by Prof. R. G. Bhandarkar, Report on the Search for Sanskrit MSS. in the Bombay Presidency during the year 1883-84 (Bombay, 1887), p. 37 seq., and by Prof. Bloomfield, S. B. E., vol. xlii, p. 480.XXX INTRODUCTION. with one another. The literal translation of the Mantra would be: ‘Stay ye here, may ye not be separated, may ye reach old age! For great welfare, O Indra.’ A grammatical construction is impossible. A similar case occurs II, 3, 2, where the first line is derived from Rv. IV, 58, 1, and the second line from Rv. I, 109, 7. A literal translation will show how absurd the combination of the two lines is: “From the ocean went forth the wave full of honey; may I obtain immortality by the muttered prayer. These, indeed, are the rays of the sun by which our Fathers went up to their common home (lit. common drinking-place ?).” It is, I believe, sufficiently clear, that the Apastambins recited these Mantras much in the same way as many a Roman Catholic will listen to Latin prayers, and many a Jew will say, and listen to, Hebrew prayers without any knowledge of the language in which they are composed. But not only that: the evidence adduced compels us also to assume that the Mantras of the Apastambins were compiled, arranged, and some of them composed by men who knew the Vedic Samhitas, but to whom the language of these Samhitas was a dead and half-forgotten language. The priests or Pandits, who finally arranged these Mantras so as to form part of the liturgical books of the Apastambins, knew, no doubt, Sanskrit as well as our commentator Haradatta, and were thus able to see the gist of many of the Mantras, and to assign to them a place in the ritual. In the case of other Mantras, which they did not understand, they were satisfied if one or two words-as a rule the first word or words of the Mantra suggested something similar to the rite for which it was used. The important thing for them was always the words, and not the meaning of the Mantras. The words were sacred, because they were either taken from their sacred literature, from the Samhitas of the Rigveda or Yajurveda, or because they were magic spells and charms hallowed by time and ancient usage. If they thought of any meaning at all, they must have felt at liberty to explain to themselves any ungrammatical forms and constructions in the same way as Sāyaṇa and Haradatta used to explain them in more recent times, as being ‘chāndasa’ or ‘Vedic.’ But all this seems to suggest that the Mantrapāṭha must have had an independent existence from the Grhyasutra of Apastamba, and that Apastamba himself was neither the author nor the compiler of the Mantrapāṭha. In order to arrive at a more definite conclusion about INTRODUCTION. xxxi this, we shall have to examine more closely into the relation of the two works.

RELATION OF THE MANTRAPAṬHA TO THE APASTAMBĪYA GRHYASŪTRA.

The relation between the Apastambīya Gṛhyasūtra and the Mantrapāṭha is very similar to that between the Gobhila-Grhyasūtra and the Mantra-Brahmana of the Samaveda. While most of the other Grhyasūtras give the prayers in full, when describing the different domestic rites, Apastamba and Gobhila merely describe the rites, and refer to the prayers as known from their respective prayer books.

Professor Knauer has shown that the Gobhila-Grhyasūtra not only presupposes, but is entirely based on the Mantra-Brāhmaṇa, that the Grhyasūtra is, in fact, nothing but a careful and systematic working up of the Mantras into a treatise on domestic rites. That certain Mantras are not given in the Mantra-Brāhmaṇa, but are quoted by Gobhila in the Gṛhyasūtra either by their Pratīkas, or in extenso, is explained by Prof. Knauer as due to the fact that the Mantra-Brāhmaṇa is strictly limited to the Grhya ritual, carefully excluding everything that pertains to the Śrauta rites1. A different theory has been proposed by Prof. Oldenberg, who believes that the two works-the Gobhila- Grhyasūtra and the Mantra-Brāhmaṇa-were ‘composed together and on one common plan.’ His chief argument is the occurrence of certain Mantras in the Gṛhyasutra, which are omitted in the Mantra-Brāhmaṇa. ‘Gobhila gave the full wording of the shorter Mantras with which the description of the ceremony could be interwoven without becoming obscure or disproportionate; the longer Mantras would have interrupted, rather tediously and inconveniently, the coherency of his ritual statements; so he separated them from the rest of his work and made a separate Samhita of them.’ He admits that there are exceptions, that short Mantras occur in the Mantra-Brāhmaṇa, and long Mantras in the Grhyasūtra, but thinks that allowance must be made for a certain inconsistency or carelessness in the distribution of the material between the two texts’-which is rather awkward, as the number of Mantras in question is not very large, so that the exceptions really make the rule illusory1.

• 1 Das Gobhilagrhyasutra herausg. und ūbersetzt von Dr. Friedrich Knauer, II (Dorpat, 1886), pp. 22-34-

In fact, Prof. Knauer has shown that of 33 short Mantras occurring in the Mantra-Brāhmaṇa no less than 28 are given in extenso in the Grhyasūtra, and only 5 in an abridged form, while, if Prof. Oldenberg’s theory were right, these 28 Mantras ought not to be included in the Mantra-Brāhmaṇa. Prof. Knauer also draws attention to the fact that tradition does not ascribe the Mantra- Brāhmaṇa to Gobhila. It would certainly be a very uncommon thing in Indian literature if a work compiled by Gobhila should have no author’s or compiler’s name assigned to it by tradition, it being far more usual that works are ascribed to a renowned author even if they do not belong to him2.

Prof. Oldenberg is inclined to extend his theory about the relation between Mantra-Brāhmaṇa and Gobhila-Grhyasūtra also to the parallel case of Mantrapāṭha and Apastambiya Grhyasūtra. No doubt, the cases are parallel, and the Sutras of Apastamba presuppose the existence of the Mantrapāṭha, but there is no reason to assume that the Mantrapaṭha presupposes the Sutras. The Mantras, certainly, presuppose a Gṛhya ritual. But as there were Vedic Samhitas and a Śrauta ritual before there existed any Śrautasūtras, so there may have been (and I believe there were) Mantra-Samhitas or prayer books for the domestic rites and a Grhya ritual before any Gṛhyasūtras existed*.

If there is any difference between Apastamba and Gobhila as regards their relation to their respective prayer books, it is this, that Apastamba is far more dependent on the Mantrapāṭha than Gobhila on the Mantra- Brāhmaṇa-so much so, that the ritual in Apastamba is hardly intelligible from the Sutras alone without referring to the Mantras.

1 H. Oldenberg in Sacred Books of the East, vol. xxx, pp. 4-8.

• See F. Knauer, Vedische Fragen, in ‘Festgruss an Rudolf von Roth’ (Stuttgart, 1893), pp. 61-64. Sacred Books of the East, vol. xxx, p. 249. Thus there existed an Atharvaveda-Samhita and Atharvan practices long before the Kausika-sutra. See the Kauśika-sutra of the Atharva-veda, ed. by M. Bloomfield, P. xxii.

• It is not right to say that the Mantras are always referred to as yajus by Apastamba (Prof. Bloomfield, S. B. E., vol. xlii, p. xliv note). He refers to prose formulas by jajus, to rcas, or verses, by using the feminine of uttara (cf.uttarām 4, 6, uttaraya 4,7 etc. etc.). The only reason why he says repeatedly attareṇa yajushā, and never uttarayarca, but always uttaraya, is that uttarena (which also means to the north," etc.) would in many cases have been ambiguous, while the meaning of uttaraya was always clear.

The Grhyasūtra not only presupposes the Mantrapaṭha, but was framed after it, To begin with, the Grhyasutra constantly refers to the Mantrapāṭha by such phrases as with the following Rg verse,’ उत्तरेण OF उत्तरेण यजुषा ‘with the following Yajus formulas But not only that; we are also told, e.g. (Ap. 4, 2), that the wooers should be sent out with the two first verses’ (viz. of the Mantrapaṭha), or (Ap. 4, 10) that the bridegroom should recite over his bride the two first verses’ (viz. of the third Anuvāka). In Ap. 8, 10 we are told that the remaining Mantras’ (viz. of the eleventh Anuvāka) should be recited at the cohabitation. This is a particularly interesting case. In most of the MSS., Mantras I, 11, 7-11 form a separate section, only in our MS. Wh. (generally the best MS.!) they form part of the eleventh Khanda. It is by no means improbable that Apastamba in the Gṛhyasūtra (8, 10) alludes to a difference of opinion with regard to this section, when he says: xaya ada ’let him mutter the rest (“the rest of the section,” says the commentary) at the cohabitation.’ Instead of saying, as he generally does, with the following (Rg verses),’ he may have chosen the phrase the rest,’ in order expressly to refute the idea of these Mantras forming a section by themselves. In Ap. 9, 9 it is said that he should worship the sun ‘with the following Anuvāka,’ i.e. Mantrapāṭha I, 16. The water for the Upanayana rite is to be mixed (according to Ap. 10, 5) with the first Yajus of the Anuvāka,’ i.e. with Mantrap. II, 1, 1.

Interesting is the manner in which Ap. 11, 1-4 refers to Mantrap. II, 3, 26-32. The Pratika is quoted of II, 3, 26, the short questions and. answers II, 3, 27-30 are referred to by the Sutra e Afaqzci The questions are for the teacher, the answers for the boy.’ Then follow the two Sūtras शेषं परो जपति and प्रत्यगाशिषं चैनं वाचयति. This can only mean, ‘The other (i.e. the teacher) mutters the rest (of the Anuvāka). But he should cause him (the pupil) to recite the (Mantra) containing a wish for himself.’

1 Sudarsanarya says: a portion of the rest of the Anuvāka.’

• About in the sense of ‘but,’ see Böhtlingk-Roth, Sanskrit-Wörterbuch, s. v. 6).[III. 8.]

I do not think that can mean anything else but the rest of the Anuvāka,’ but as the Mantra meant (II, 3, 31) is not the end of the Anuvāka, we have, I believe, to take in Ap. 11, 4 in the sense of but Mantra II, 3, 32, O Lord of the Paths, may I obtain the other end of the path," is the Mantra containing ‘a wish for himself.’ These Mantras are also interesting for another reason. If Prof. Oldenberg were right in assuming that the authors of the Gṛhyasūtras were also the compilers of the Mantra collections, surely Gobhila, as well as Apastamba, would have included such short questions and answers as को नामासि, असी नामास्मि etc. not in the collection of Mantras, but in their Gṛhyasūtras1. Yet we find them both in the Mantra-Brāhmaṇa (I, 6, 17) and in our Mantrapāṭha (II, 3, 27-30).

In Ap. 11, 18aaaaa is, no doubt, rightly explained by Haradatta as meaning ’that group of Mantras beginning with ▼Ħ’ ( एतत् = Qalaa). The MSS. read Mantrapaṭha II, 5, 2-10, and again II, 5, 12-21 without dividing the single Mantras. I have followed the Telugu edition in numbering them. That Mantras 12-21 were not intended to be one Mantra is clear from Apastamba’s usual phrase with the following (Mantras).’ Sudarśanarya also explains ‘with the following ten Mantras.’

Quite exceptional is the way in which Ap. 8, 2 refers to Mantrap. I, 9, 8 by the words faft: ‘Sadasaspati is the second (deity to be worshipped).’ The Mantra begins and etc.

By far the most usual way in which Apastamba refers to the Mantrapāṭha is by the word ’the following.’ Whenever a number of burnt-oblations are to be offered, Apastamba simply says: ‘Let him offer the following burnt-oblations’ (ahuti), that is to say, the burntoblations which are to be offered with the Mantras following in the Mantrapāṭha. As many Mantras there are, so many oblations have to be offered. In order to know the exact number of oblations to be made, it is necessary to know the Mantrapāṭha by heart.

• Another set of very short Mantras-formulas consisting of one or two words only- Occurs in the Mantrap. II, 10, 11-18.
• *See above, p. ix.

That the Apastambins are expected to learn first ’the Mantras by heart before they proceed to the study of the Grhyasūtra, follows even from the traditional arrangement of the two Mantraprasnas before. the Grhyasūtra 2. But this method of referring to the Mantrapāṭhato chapter and verse of the ‘prayer book’-which we have just pointed out, proves that the author of the Gṛhyasutra had an actual collection. of Mantras, arranged and divided into Anuvākas (corresponding to the Khandas of our MSS.) exactly in the same manner as we find the Mantrapāṭha in our MSS., constantly before his mind when he composed his Sūtras.

Professor Knauer, when speaking of the analogous case of the Mantra- Brāhmaṇa, would go farther. He asserts that Gobhila must have had an actual written ‘book’ of Mantras ‘in his hands’ when writing his Grhyasūtra1. I do not think that it can be proved that the authors of the Gṛhyasūtras wrote or used written books. Though it is extremely probable that the art of writing was known at the time when the Sūtras were composed, I do not believe that the Sutras were intended for reading, as the very Sutra style is intended as a memoria technica for texts to be learnt by heart. The Mantras, too, were learnt by heart as prayers and hymns are learnt to the present day, not only in India and there was no occasion for writing them down. What is true even to-day, that for the Hindu ‘sacred books and all sciences exist only in the mouth of the teacher, compared with which a written text has no authority, and that they can be learnt correctly only from a teacher, not from manuscripts,’ was certainly true in the days of the Sūtrakāras.

• 1 Knauer, Das Gobhilagṛhyasutra, II, pp. 31-34, and Festgruss an Rudolf von Roth,’ p. 63.
• • See G. Būhler, Indische Palaeographie (Grundriss, I, 11), p. 17.
• • G. Būhler, 1. c. p. 4-

There are only very few Mantras which are inserted in the Apastambiya Gṛhyasūtra and are not found in the Mantrapaṭha. They are-

• (1) The Parishecana Mantras given by Ap. 2, 3 and 8: अदितेऽनुमन्यस्व and … • नुमस्थाः । There are, besides, a few Mantras which are merely indicated in the Grhyasūtra, and not given in extenso in either of the two works. They are the Jaya, Abhyātāna, and Rashtrabhrt formulas (found in Ts. III, 4, 4-7), the Prajapati verse Ts. I, 8, 14, 2, and the Vyāhrtis (Y DIET YA DIET), all referred to by Ap. 2, 7. The Rudra-Mantras (Ts. IV, 5) also, prescribed by Ap. 20, 8, do not occur in the Mantrapāṭha. They evidently formed a separate book, and had to be learnt separately. Haradatta says: अत्र रुद्राञ्जपेत् । रुद्रा अन्यत्र व्याख्याताः । from which we may conclude that Haradatta wrote a commentary on the Rudras.

अनुमते ऽनुमन्यस्व and “नुमँस्था: । सरस्वतेऽनुमन्यस्व and “नुमँस्था: । देव सवितः प्रसुव and प्रासावीः ।

The last of these Mantras (in a much longer version) is the first Mantra in the Mantra-Brāhmaṇa, while the other Mantras are given by Gobhila (I, 3, 1–3).

(2) The Mantras for the two Ajyabhāga offerings (Ap. 2, 6): अप्रये स्वाहा । सोमाय स्वाहा ।

(3) The Mantra for the Svishṭakṛt offering (Ap. 2, 7): यदस्य कर्मणोऽत्यरीरिचं यद्वा न्यूनमिहाकरम् । अनिष्टत्स्विष्टद्विद्वान्त्सर्वं विष्ट सुतं करोतु स्वाहा ॥

(4) The Mantra MÂ ÂÂ which, according to Ap. 8, 6, is to be recited optionally in lieu of the two Mantras I, 9, 9-10 of the Mantrapāṭha, prescribed by Ap. 8, 5.

(5) The Mantra यदि वारुण्यसि वरुणात्त्वा निष्क्रीणामि यदि सौम्यसि सोमात्त्वा fauf which, according to Ap. 9, 5, is to be recited at the performance of a kind of love charm. Now it is very remarkable that the very similar Mantra यद्यसि सौमी सोमाय त्वा राजे परिक्रीणामि etc., employed at the Pumsavana rite, is given in extenso by Gobhila II, 6, 7, and does not occur in the Mantra-Brāhmaṇa. This cannot be a mere accident.

One thing is certain. The occurrence of these Mantras in the Apastambīya Grhyasutra, and their omission in the Mantrapāṭha, are not due to any distinction made between long and short Mantras, as Prof. Oldenberg suggests. Numerous short Mantras occur in the Mantrapāṭha, and the Mantras occurring in the Grhyasutra are by no means all short Mantras. I can see no other reason why Apastamba should have included these Mantras in his Gṛhyasūtra than the fact that they were missing in the Mantrapāṭha which formed the basis for his treatise. If he himself had been the compiler of the Mantrapāṭha, there is no visible reason why he should have included just these few Mantras in the Gṛhyasūtra.

The agreement between Mantra-Brāhmaṇa and Mantrapāṭha with regard to the omission of the Parishecana Mantras and the Mantra afa araufa etc. seems to indicate that different Vedic schools had their prayer books for domestic rites, which were arranged according to some uniform plan. That the Mantrapāṭha is included in the corpus of Sūtras ascribed to Apastamba1 may be merely for the sake of convenience, as the two works are so closely connected with one another. But it does not follow that even the tradition of the Apastambins credited Apastamba with the authorship or compilership of the Mantrapāṭha. With the exception of the quite modern MS. E., none of the other MSS. mentions the name of Apastamba in the colophons. Nor did the commentators look upon Apastamba as the author of the Mantrapāṭha. It has struck Sudarśanarya, the commentator of Apastamba’s Gṛhyasutra, that the Sūtras Ap. 9, 2 seqq. have nothing to do with the wedding ritual in the middle of which they occur, and he explains this fact by saying that this heterogeneous matter is treated here on account of the order followed by the Mantrapātha ( मन्त्राम्नान- a). This shows that Sudarśanarya did not consider Apastamba to be the compiler of the Mantrapāṭha 3.

• 1 See above, p. ix. That Apastamba refers to the Anuvāka division of the Mantrapāṭha (see above, p. xxxiii), but not to the division into 2 Praśnas, seems to show that the division of the corpus of Apastambiya Sutras into 30 Praśnas, of which the Mantras formed 2 Praśnas, is of a more recent date.
• See my essay, ‘Das altindische Hochzeitsrituell,’ 1. c., p. 95. Similarly the author of the Khadira-Grhyasūtra looked upon the Mantra-Brahmana as an independent work, not as the work of Gobhila. This is proved by Khad. I, 3, 3 seq., where it says that the Samavartana or bath at the end of studentship comes first, and then marriage. ‘As, however, in the (collection of) Mantras marriage is treated of (first), it is explained (here) before (the bath). Prof. Oldenberg, S. B. E., vol. xxix, p. 379 seq. I doubt whether Prof. Knauer’s interpretation of this passage (Das Gobhilagṛhyasūtra, II, P. 39 seq.) is admissible.
• • Commentators are so fond of long discussions as to why a Sūtrakāra said something which he ought not to have said, that they would have certainly tried to ’explain,’ after their fashion, why certain Mantras occur in the Grhyasūtra instead of being included in the Mantrapāṭha. Their not doing so, shows also that they, at any rate, did not consider Apastamba the author or compiler of the Mantrapāṭha.

But there is also some external evidence showing that the Mantrapāṭha existed as an independent Vedic work, forming a kind of appendix to the literature of the Taittiriyas. For in the Kāṇḍānukrama of the Atreyi Śākha of the Black Yajurveda, after the 41 Kandas and the xxxviii INTRODUCTION. 3 Upanishads ascribed to Tittiri and the 8 Kāṭhakāni have been enumerated, we read: Qanfaat fafu: anus aazafafa fufa: 1 सु This Ekāgnikanda1 consists, as the commentary states, of the Parishecana Mantras fas etc., the 39 Vaiśvadeva Mantras (=TA. X, 67=Mahānārāyaṇa-Upanishad 19, 2, see Ap. Dh. II, 2, 3, 16-II, 2, 4, 8), and finally of the praśnadvayam beginning Tafa, i. e. our Mantrapāṭha”. We do not know the date of this Kāṇḍānukrama, but it seems. to be older than the final arrangement of the Taittiriya-Brāhmaṇa and Taittiriya-Aranyaka. For though it agrees with the present arrangement of the Taittiriya-Samhitā, there is, as Professor Weber has shown 3, some difference in the arrangement of the Taittiriya-Brāhmaṇa and Taittiriya-Aranyaka, as compared with the Kāṇḍānukrama. Haradatta called his commentary the Ekāgnikāṇḍavyākhyā1, because he included in it also the Parishecana Mantras and the Vaiśvadeva Mantras, occurring at the beginning of the Apastambiya Grhyasūtra, and in the Taittiriya-Aranyaka X, 67. The Mantrapāṭha seems, then, to have formed part of an Ekāgnikāṇḍa, ‘a chapter (of Mantras for the rites to be performed) with one fire, before it came to be included in the corpus of Apastambiya Sūtras. This proves, I believe, conclusively that Apastamba is neither the author nor the compiler of the Mantrapaṭha, which must have existed as an independent collection before the Gṛhyasutra was composed. It seems to me not impossible that other Gṛhyasūtras, too, may be 1 See above, p. x.

• See A. Weber, Indische Studien, III, pp. 376, 387, 391; XII, p. 354. In a note to Ind. Stud. III, p. 387, Prof. Weber has already (in 1855) referred to the Bodleian MS. (W.) of the Mantrapāṭha. Indische Studien, III, p. 374. The Telugu edition gives the Parishecana and some of the Vaiśvadeva Mantras as a kind of introduction, omitting the Mantras for the Bali offerings. It begins: ॥ अथ वैश्वदेवमंचप्रारंभः ॥ अदितेनुमन्यस्व । अनुमतेनुमन्यस्व । सरस्वतेनुमन्यस्व । देव सवितः प्रसुव ॥ ओं अमये स्वाहा । सोमाय स्वाहा । विश्वेभ्यो देवेभ्यस्स्वाहा । ध्रुवाय भूमाय स्वाहा । ध्रुवचितये स्वाहा । अच्युतचितये स्वाहा । अप्रये स्विष्टकृते स्वाहा । अदितेन्वमस्थाः । अनुमतेन्वमस्थाः । सरस्वतेन्वमँस्थाः । देव सवितः प्रासावीः ॥ ॥ एका- fagicHnentia: Then follows the beginning of Haradatta’s commentary. INTRODUCTION. xxxix based on similar prayer books or Mantra collections. Such modern prayer books as the Rigveda-Mantra-Samhita1, which contain prayers for domestic rites, may possibly be developed from older books of that kind which were worked up into the Grhyasutras as we have them now. Dr. Būhler has discovered a Mantrabhashya, containing a commentary on the Mantras prescribed in the Pāraskara-Grhyasūtra*, so that there may have been a separate collection of Mantras, which formed the basis of our Pāraskara-Grhyasutra. For practical purposes such an entire separation of Mantras and Sutras, as it is carried out in the case of the Mantrapāṭha and the Apastambīya Gṛhyasutra, and (though less rigorously) in the case of the Mantra-Brāhmaṇa and Gobhila, must have been very inconvenient. That this is so even now-a-days, may be seen. from a native edition of the Apastambiya Grhyasutra in Grantha characters, which gives the Mantras in Pratika form 3. But whether they existed originally in separate collections or not, the Mantras ought, I believe, always to be regarded as belonging to an older stratum of Vedic literature than the Sutras in which they occur. This is not always borne in mind by scholars when investigating the language of the Sūtras*. 1 I possess a lithographed edition, Bombay, 1891 (Sake 1813), beginning with पुण्याहवाचनमन्त्राः, नान्दीश्राद्धमन्त्राः etc. This is different from the ‘Aévalayanasākhokta-Mantra-Samhita’ (Bodleian MS. Walker 144, cf. Max Mūller, History of Ancient Sanskrit Literature, p. 474), which begins with te, gauga, facure an etc. Neither the edition, nor the MS., contains the Mantras as found in our Aávalā- yana-Grhyasutra. See also the books quoted above, p. x, note 3. पुरुषसूक्त, विष्णुसूक्त
• Dr. Būhler’s Report on Sanskrit MSS. in Gujarat for 1872-3 (see Indian Antiquary, II, 304). • Apastambagṛhyasūtram sapratīkam (printed at Tanjore, Jyotirvilāsa Press, 1885). Instead of the Sutra af (p. 4, 2) this edition, e. g., has: raf तान् प्र सु ग्मंतति द्वाभ्यामभिमंचयेत.
• Dr. O. Franke, in his article ‘Was ist Sanskrit ? (Bezz. Beitrāge, XVII, 54 ff.), says (p. 55) that in Pap. IV, 1, 62 ‘bhasha’ is opposed to ‘Sanskrit, not to ‘chandas,” for, he says, the example given by the Kafikā (sakha saptapadi bhava) is taken from the Aévalayana-Grhyasūtra (1, 7, 19). Yet the words are not Aévalāyana’s, but, as the author of the Kāsikā knew well, the words of a Vedic Mantra (also occurring Sankh. I, 14, 6, Kaus. 76, 24, cf. Mantrap. I, 3, 14), and therefore ‘chāndasa,’ not ‘Sanskrit’ in Dr. Franke’s sense. Sakha as a feminine occurs also in Sankh. II, 2, 1, again in a Mantra,x1

RELATION OF THE MANTRApāṭha TO THE TAITTIRIYA, RIGVEDA, AND ATHARVAVEDA SAMHITĀS.

Even if we had not the testimony of the Kāṇḍānukrama, there could not be the least doubt that the Mantrapaṭha belongs to the Taittiriya- Veda. An analysis of the synoptical list of parallel passages given in Appendix B yields the following facts. There are 100 Mantras for which parallel passages are found in the Taittiriya-Samhita. Of these, 86 are identical with the corresponding Mantras of the Taittiriya-Samhita, while 14 are only similar Mantras. Most of the identical Mantras-66 in number-are not given in extenso by our MSS., but only indicated by Pratikas. It was evidently taken for granted that every Apastambin would know these Mantras by heart from his Samhita-the more so, as most of them occur several times in the Taittiriya books. In using these verses or formulas for the ritual, they had all, of course, to be recited in full. They are accordingly given in full in the Apastambiya-Samskara-Prayoga (MS. P.), and are fully commented on in Haradatta’s commentary. For the sake of convenience I have printed these Mantras in extenso1, as it would have been mere pedantry to compel the reader to refer to the Taittiriya-Samhitā every time that such a Mantra occurs. It has, however, always been stated in the critical notes when the MSS. give the Pratika only. This very practice of the scribes of our MSS. shows that a thorough familiarity with the Taittiriya-Samhita is presupposed on the part of the worshippers who had to make use of the Mantrapāṭha. Besides those 100 Mantras occurring in the Taittiriya-Samhita, many of which occur also in the Brāhmaṇa and Aranyaka, there are 20 parallel passages which are found in the Taittiriya-Aranyaka only, and 18 which are found in the Taittiriya-Brāhmaṇa only. Of these 38 Mantras, 31 are identical, and only 7 are variants of the corresponding Taittiriya Mantras. If we turn to the Rigveda-Samhita we find parallel passages for 140 The editor of T. is not consistent in this respect. Sometimes he gives the Mantras in full, sometimes their Pratikas only, and occasionally he even gives first the Pratika, with etc. after it, and then the Mantras in extenso. INTRODUCTION. xli Mantras (86 in Praśna I1 and 54 in Praśna II), many of which occur also in the Taittiriya books. But only 45 out of these 140 Mantras are identical with the corresponding verses of the Rigveda, while all the other Mantras contain either slight various readings, or are entirely different versions of the Rigveda verses. If we take into account the Mantras which occur in the Rigveda-Samhita only, and not in the Taittirīya books, there are 84 parallel passages (68 in Praśna I and 16 in Praśna II), of which only 24 are identical with the corresponding verses of the Rigveda. Still more remarkable is the result of an analysis of the parallel passages occurring in the Atharvaveda-Samhita. Of 120 Mantras occurring in this Samhita, only 9 are identical with the corresponding Atharvaveda verses, while all the rest are either different versions, or contain at least one or two various readings. Excluding the Mantras found also in the Rigveda-Samhita, and in the Taittiriya books, there are 41 Mantras found in the Atharvaveda-Samhita only, out of which only 2 are identical verses. The following table will show clearly how much closer the relation of the Mantrapaṭha to the Taittiriya books is, than its relation to the other Vedic Samhitas" Of 138 Mantras found in Ts. TBr. TĀ. 33 140 33 84 120 117 are identical, 21 have variants Rv. 33 45 J] 95 11 33 11 23 Ts. {Rv. and not in in } 24 60 33 33 33 23 Av. JJ 9 III " " 33 { Av.and notin Rv.) Ts. TBr. TĀ. 2 39 33 13 33 41 If any further proof were wanted to show that the Mantrapāṭha belongs to the Black Yajurveda, we have only to refer to those Mantras which occur in the Rigveda or Atharvaveda with variants, while the Taittirīyas have the same readings as our Mantrapāṭha. Thus Mp. I, 1 The large number of Rigveda Mantras in the first Praśna is due to the fact that this Praśna contains the Mantras for the wedding ritual, collected in the Surya hymn Rv. X, 85.

• It is hardly necessary to add anything about other Samhitas but those mentioned, as there is nothing to suggest any closer connection of the Mantrapāṭha with either the Vs., or the Maitr. S., or the Samaveda-Samhita. f [III. 8.] xlii INTRODUCTION. 7, 1 has the same readings as TA. IV, 20, 1 seq., while Rv. VIII, 1, 12 agrees with Av. XIV, 2, 47. Mp. I, 14 corresponds to Rv. I, 14; the various readings occurring in vv. 3, 5, and 7 are also found in TBr. II, 5, 5, 1; 8, 8, 9. Mp. II, 12, 6-10 agrees with TBr. II, 5, 6, 1-3, both differing from the corresponding hymn of Av. II, 10. Mp. II, 15, 2 shares with TA. X, 1, 10 the reading :, while Rv. I, 22, 15 reads सप्रथ: 1. The total number of Mantras in the Mantrapaṭha (excluding repetitions) is 590; for 264 of these Mantras I have been able to find parallel passages in the Samhitas of the Rigveda or Atharvaveda, and in the Taittiriya books. Of the remaining 326 Mantras many will be found in the other Grhyasūtras and in the Mantra-Brāhmaṇa of the Samaveda. A comparison of these Mantras would show that the Mantras of the Mantrapāṭha have much more in common with those of the Baudhāyana, Bharadvaja,and Hiranyakeśin Gṛhyasūtras-that is to say, with the Gṛhyasutras of the Black Yajurveda-than with any of the other Gṛhyasūtras. Baudhāyana, Bharadvāja, and Hiranyakeśin have frequently the same Mantras (with the same various readings where they differ from the Samhitas) as the Mantrapāṭha, though they do not prescribe them for the same occasions. Thus Baudh. I, Io agrees with Mp. I, 1, 6 in reading of Rv. X, 40, 10, although Apastamba prescribes the Mantra for the moment when the bride begins to cry at the preparatory marriage rites, while Baudhāyana includes it among the prayers to be recited at the cohabitation2. Instead of life wafa of Rv. X, 85, 28, the Mp. I, 6, 8 reads the . The same reading occurs. in Baudh. I, 8. Apastamba prescribes it for the wedding ceremony, which I believe to be identical with the European custom of barricading the bridal procession, while according to Baudhāyana it is to be recited when the bride looks at the setting sun. It has been shown that in some cases ungrammatical readings of the Mantrapaṭha occur also in the for 1 Vs. XXXV, 21 (=XXXVI, 13) also has T. See besides, p. xxiii, on II, 3, 2.
• See Das altindische Hochzeitsrituell,’ pp. 12, 42 seq.
• See ‘Das altindische Hochzeitsrituell,’ pp. 12, 67 seq., and my paper ‘On a Comparative Study of Indo-European Customs (Transactions of the International Folk- Lore Congress, London, 1891), p. 268 seq.
• See above, pp. xvii, xxiii seq., xxvii, on I, 8, 2; II, 2, 11; II, 7, 25; and I, 3, 14- 2 INTRODUCTION. xliii Grhyasutras of Baudhāyana, Bharadvaja, and Hiranyakeśin. If we had critical editions of Baudh. and Bhar., many more cases of this kind. would certainly come to light. They prove that these Mantras are derived from one source common to the schools of the Black Yajurveda. As is only to be expected, the Gṛhyasutra of the White Yajurveda, the Pāraskara-Grhyasūtra, has more Mantras in common with the Mantrapāṭha than the Mantra-Brāhmaṇa and the Grhyasutras of the Rigveda schools. The parallel passages from the Grhyasutras1 quoted in the critical notes will give some idea of the relation of the Mantrapāṭha to the Mantras of the other Grhyasūtras. A full analysis of the Mantras. occurring in the Grhyasutras has been promised by Prof. E. W. Fay, who also intends to publish a complete Index of Mantras covering all the Grhyasutras. This will, no doubt, prove extremely useful for the study of Vedic Mantras, and it is only to be hoped that it may include the Grhyasutras of Baudhayana and Bharadvaja, for which the MS. material unfortunately seems as yet to be insufficient. Of still greater importance will be such a Vedic Concordance on a large scale as announced by another American scholar, Prof. Bloomfield3. Only such a Concordance of the Pādas of Vedic Mantras will enable us to get a clearer insight into the origin of the Mantrapāṭha than it is at present possible to arrive at. With the help of such a Concordance we should, no doubt, also be able to find many more parallel passages than I have been able to point out. But there will always remain a considerable number of Mantras which were not derived from any Vedic Samhita. Such Mantras as those of II, 13, used at the ceremonies connected with the birth of a child, or the exorcisms against the demons harassing little children, found in II, 16, belong to ancient popular tradition, and are probably older than the hymns of the Vedic bards. 1 In the first Praśna I have only given a few parallel passages, as they have been given more completely in the ‘Erlāuterungen’ to my essay ‘Das altindische Hochzeitsrituell, p. 27 seqq., where I have also quoted the wedding Mantras from the Baudhayana and Bharadvaja Grhyasūtras. Editions of these two works are still among the desiderata. I regret that it was too late to make any use of Prof. Knauer’s excellent edition of the Manava-Gṛhyasutra, which has just been published.
• See John Hopkins University Circulars, Baltimore, May, 1890, vol. ix, no. 81, p. 74. See American Oriental Society’s Proceedings, April, 1892. f 2 xliv INTRODUCTION. 24 Prof. Oldenberg1, indeed, believes that, during the latter part of the Rigveda period, ceremonies such as marriage and burial began to be decked out with poetry as had long been the case with the Somal offering.’ And from metrical usage he concludes that the development of the Grhya rites in the form in which they are described to us in the Sutras, that especially their being accompanied with verses, which were to be recited at their performance, is later than the time of the oldest Vedic poetry, and coincides rather with the transition period in the development of the Anushṭubh metre, a period which lies between. the old Vedic and the later Buddhistic and epic form." I confess I am not convinced by Prof. Oldenberg’s argumentation of the comparatively late origin of the Grhya Mantras. It seems to me highly improbable that marriage rites and burial ceremonies should have been performed without any prayers, until they began to be ‘decked out with poetry’ after the analogy of the Soma sacrifice. In fact, many of the Grhya’ rites’ consist merely in invocations addressed to popular deities, or in the recitation of magic formulas, while others are no ‘rites" at all unless they are accompanied by prayers or formulas. And it has not yet been proved that the elaborate Soma sacrifice can claim a higher antiquity than the simple Grhya offerings. I am rather inclined to think that the artificial poetry of the Brahmans, as represented by the hymns of the Rigveda, and the popular poetry connected with Atharvan practices and Grhya rites, as represented by the Atharvaveda-Samhitā, and the Grhya Mantras, are two parallel streams running in different channels, approaching each other at times, and diverging widely at other times. The peculiarities of the Anushṭubh pointed out by Prof. Oldenberg may just as well be due to the fact that these Mantras were popular poetry, and as such followed different rules of metrics from those followed by the authors of the Vedic hymns. Besides, a glance at our Mantrapāṭha will show that metrical laws are disregarded in these Grhya Mantras quite as much as the rules of grammar. See e. g. I, 1, 9, where the metre is as corrupt as the text of the Mantra. Or, take the Mantras of II, 16, which offer no less difficulties to scanning than to *S. B. E., vol. xxx, p. xiv. S. B. E., vol. xxx, p. x. See the very interesting and suggestive remarks made by Prof. Knauer in the ‘Festgruss an Rudolf von Roth,’ p. 64 seqq. INTRODUCTION. xlv interpretation. Compare also such Mantras as I, 1, 3; 3,6; 4, 11; 5,7; II, 5, 22; 7, 24-26, etc., where we have either too many or too few syllables, or Mantras where we have a kind of mixed prose and verse, as in I, 3, 14; 13,7-8; II, 3, 3-12; 5, 2-9; 13, 12, etc. I doubt whether, with such a state of things, it would be safe to draw any conclusions as to chronology from the metrical usage of the Gṛhya Mantras1.

DATE OF THE MANTRAPĀṬHA.

Needless to say that it is absolutely impossible to arrive at any safe conclusion as to the exact, or even approximate date of the Mantrapāṭha. It presupposes the Samhitas of the Rigveda, Yajurveda, and Atharvaveda, and refers to the three Vedas in II, 19, 14-16, and to the four Vedas in II, 21, 2-5. It is older than the Kāṇḍānukrama of the Atreyi school, but we know nothing about the date of this work. If I have succeeded in proving that the Apastambīya Grhyasūtra is based on the Mantrapaṭha, and that for the author of the Gṛhyasūtra the Mantrapāṭha was a sacred text with which he did not dare to tamper, but which was to him a given fact, we shall have to assume, as it must have taken some time for this sacred character to develop, that the Mantrapaṭha preceded the Grhyasūtra by something like a century. As Dr. Būhler has shown that, on linguistic grounds, Apastamba cannot be placed later than the third century B. C., we may conclude that the Mantrapāṭha cannot be much later than the fourth century B. C., though it may be much older. If Haradatta were right in explaining futu: II, 16, 14 as meaning • The Rshi of the Bodhāyana Gotra (यस्य गोत्रं बोधायनस्स बोध ऋषिः), it might prove the anteriority of the Sūtrakāra Bodhāyana or Baudhayana to the Mantrapāṭha. But this interpretation is more than doubtful. There is one other Mantra which may possibly help us some day to arrive at a more certain date of the Mantrapāṭha. This is the Gāthā to be sung by two lute players at the Simantonnayana among the Salvas. 1 On the similar state of things in the hymns of the Atharvaveda, see Prof. Bloomfield, S. B. E., vol. xlii, p. 296: ‘Atharvan metres are so generally capable of improvement, that we are in danger of singing our own, rather than Atharvan hymns, when we apply ourselves to the task of improving them.’ See also ibid., p. 584 seq. S. B. E., vol. ii, p. xlvi. xlvi 2 INTRODUCTION. Other Grhyasūtras (Hir. II, 1, 3; Aśv. I, 14, 7 ; Pār. I, 15, 8) quote a Gāthā, similar to Mp. II, 11, 13, in which king Soma is praised, and the name of the river near which the worshipper dwells, is to be mentioned1. Apastamba alone says that the first Gāthā (Mp. II, II, 12) is to be sung among the people of the Salvas, while the second one (II, 11, 13) is to be used for Brāhmaṇas generally. The Gāthā is a fragment of some Akhyāna, and can only be translated quite tentatively: “Yaugandhari only is our king,” thus spake the Salva (women ?), sitting on thy banks, O Yamuna, turning round the wheel (?). It is simply impossible to know the exact meaning of this verse, which must originally have formed. part of a longer ballad, of which only this one verse has come down to us. Yet thus much is certain that Yaugandhari is the name of a king of the Salvas, a tribe that must have been living somewhere near the Yamuna. And it seems obvious, too, that the author or compiler of the Mantrapāṭha must have had some connection with this or some other king of the Salvas. Now, the Salvas or Salvas are frequently mentioned in Sanskrit literature. They were divided into several tribes, one of which were the Yugandharas 3, as we see from the Kārikā quoted by the Kāsikā on Pāṇ. IV, 1, 173: उदुम्बरास्तिलखला मद्रकारा युगन्धराः । भुलिङ्गाः शरदण्डाश्च साल्वावयवसंज्ञिताः ॥ Śalvas and Yugandharas are mentioned together in the Mahabharata IV, 1, 12 among the ’lovely countries around the Kurus,’ and in the Mahābh. VIII, 44 (45), 14 the Salvas are mentioned immediately after the Kauravas and Pañcālas along with other nations who follow the ’eternal law’ of the Brahmanic religion. According to H. H. Wilson*, the Śalvas seem to have occupied part of Rajasthan, a Śālva Rāja being elsewhere described as engaging in hostilities with the people 1 Cf. Prof. Oldenberg, Zeitschrift der D. M. G., vol. 39, p. 88. As to Soma as the king of Brāhmapas, see Ts. I, 8, 10, 2 एषो वो भरता राजा सोमो ऽस्माकं ब्राह्मणानाँ T. Cf. Sat. Br. V, 3, 3, 12. 2 Grhyasutra 14, 4-6. According to the Kāsikā on Pap. III, 2, 46, Yugandhara is the name of a mountain. “Vishnu-Purana (London, 1840), p. 177. INTRODUCTION. 1 xlvii of Dvaraka in Guzerat.’ Ch. Lassen identifies the Salvas with the Salabastrae of Pliny, and places them between the Indus and the Aravali in Lower Rajasthan. According to our Mantra we should have to place either the Salvas, or at least the Yugandharas as forming one division of the Salvas, somewhere near the Yamuna. As the school of the Apastambins undoubtedly belongs to the South of India 2, we ought to expect the Mantrapaṭha, too, to have arisen in the South. Could there be any possible connection between the Salvas of the Yamuna, and the ‘Salvas’ or ‘Saluvas’ met with in South-Indian inscriptions of modern times 3? At any rate, this seems to be the only passage in the Mantrapāṭha which may be said to contain a historical allusion. It is to be hoped. that future researches and future discoveries of our excellent Indian epigraphists may enable us to use this allusion for chronological purposes.

ORTHOGRAPHICAL NOTES.

There can be no doubt, that if we want to know how the Apastambins actually recited their prayers, we have to take into account the phonetic peculiarities of the Grantha MSS., which-for reasons stated above (p. xv) -give the most authentic text of the Mantrapāṭha. Now, the Grantha MSS. are perfectly consistent with regard to the Sandhi of sibilants. 1 Indische Alterthumskunde, p. 613 seq. (vol. i, p. 760 seq.).

• See above, p. xv. P. 7.
• Saluvaganda II, A. D. 1428, and Saluva Narsinha, A. D. 1477, are given in a genealogical list of Vijayanagara kings, by H. H. Wilson, in the Asiatic Researches, XX, Saluva Nrsimha occurs in a Grant of Ranga II, dated in 1644-5 A. D., and Dr. Hultzsch thinks that this king, who was the protégé of Rama’s great-grandfather Bukka according to the Karnaṭā grants, may have been Sadasiva’s grandfather Nrsimha or Narasa of Vijayanagara’ (Indian Antiquary, XIII, p. 155). See also R. Sewell, Sketch of the Dynasties of Southern India (Madras, 1883), pp. 44 seq., 109, and L. Rice, Epigraphia Carnataca, Mysore, I (Bangalore, 1894), p. 215, Introd., P. 24 seq. Dr. Lūders informs me that Salva-Timma is a name of a chief minister of king Krshna Raya of Vijayanagara in an inscription from Kondavīdu, dated Saka- Samvat, 1442. Dr. E. Hultzsch (South-Indian Inscriptions, I, 1890, p. 179) gives for Sāluva, as occurring in Tamil inscriptions of these kings, the meaning, ’the hawk, a biruda? See, however, V. Visvanatha Pillai, A. Dictionary, Tamil and English (Madras, 1888), s. v. rari, ’tributary kings from Salva, regarded by the people in Southern India as intruders.” 10 xlviii INTRODUCTION. They never write a Visarga before a sibilant, but always the corresponding sibilant, and they drop the Visarga (or rather the final sibilant) before an initial sibilant with any following consonant. That is to say, they follow the practice recommended by the Vyasa-Sikshā1, and probably known already to the author of the Taittirīya-Prātiśākhya. For the teaching of the latter (IX, 1), that ‘Visarga, when followed by a spirant which has a surd letter after it, is dropped, according to Kāṇḍamayana,’ seems to imply, as the commentator explains it, that according to others the Visarga is dropped not only in this case, but also before a spirant that is followed by a sonant letter2. Devanāgarī copyists seem to have been puzzled sometimes in consequence of this way of making Sandhi. Thus, they separated : (I, 9, 3) into स्वस्थः instead of प्रस्वः स्थः, and were doubtful as to विष्णो before श्रेष्ठेन I, 12, 7, or as to ЯT before II, 4, 14. As I wished to edit the text as I believe it to have been recited by the Apastambins, I had to follow the Grantha MSS. with regard to this Sandhi of sibilants". I ought also to have written the nasals in accordance with the Grantha MSS. that is, the corresponding nasal instead of Anusvara-and to have omitted the Avagraha. But if I had done so, it would have made the separation of words almost illusory, and would have been mere pedantry. Every Sanskrit scholar knows that the Anusvāra is used as a mere. graphic sign, expressive of the nasal required by the following consonant, that, e.g. शं नो भव द्विपदे शं चतुष्पदे is meant to be read शन्नो भव द्विपदे faч nag, as the Grantha MSS. actually write. I have also followed the European practice with regard to the Avagraha, as I consider it, with Dr. von Böhtlingk, as useful for practical purposes. If we were to follow the MSS. in all respects, we should have to do away with. 1 See H. Lūders, Die Vyasa-Śiksha (Kiel, 1895), p. 57.
• See W. D. Whitney, The Taittiriya-Prâtiśakhya (Journal of the American Oriental Society, vol. ix), p. 205. Cf. J. Wackernagel, Altindische Grammatik, § 287. Dr. Oertel, in his edition of the Jaiminiya-Upanishad-Brahmana (Journal of the American Oriental Society, vol. xvi, 1, 1894), has also followed his MSS. with regard to the Sandhi of sibilants (he prints dakshinatas sa, etc.), but he leaves a final sibilant before an initial sibilant with a following sonant letter (e. g. tatassyat, etc.). The same rule is followed in the Telugu edition of our Mantrapāṭha.
• Zeitschrift der D. M. G., vol. 43, p. 602. INTRODUCTION. xlix the separation of words altogether. If we write Visarga instead of or, we give an entirely wrong impression of the actual recitation of the text. But I see no reason why we should not avail ourselves of such orthographical expedients as the Anusvāra and the Avagraha, which do not affect the oral recitation. I also adopted the usual practice of placing the figures 4 and 3 after short and long Svarita vowels, when followed by an Udatta syllable, although the MSS. never use these figures. They write, e. g. 75 arao I, 1, 3, °दु॒त्य॑स्सं वि॑° I, 4, 9, नद्यौ यान I, 7, 9, मनुष्यां वर्षन्ति I, 11, 6 etc.