^{page 86 note 1} See, amongst others: Felber, E., Die indische Musik der vedischen und der klassischen Zeit, Wien, 1912Google Scholar; Haug, M., über das Wesen und den Wert des vedischen Akzent, München, 1873Google Scholar; Bake, A. A., ‘The practice of Samaveda’, Proceedings and transactions of the seventh All-India Oriental Conference, 1933, 143. The last article contains some remarks on Rgvedic recitation. (The Sāmaveda, studied chiefly by E. Simon, A. C. Burnell, and M. śeṣagiri śāstri, stands completely apart from Rg- and Yajur-vedic recitation.)Google Scholar

^{page 86 note 2} Professor W. S. Allen recorded some dozen verses of Rgvedic recitation from a pandit referred to as ‘A‘. The writer is indebted to Professor Allen for the iue of this recording.

^{page 86 note 3} Some eight hymns, totalling over 100 verses, were recorded by the writer in Bombay. The reciter (referred to as ‘B’) was a very old man, but the consistency and accuracy of his recitation (not to mention its vigour) from memory makes his testimony most valuable.

^{page 86 note 4} Some six hymns, totalling over 80 verses, were recorded by the writer from a pandit (referred to as ‘C’ ), a man in his thirties, who was also remarkably consistent and accurate in his performance.

^{page 86 note 5} The writer is indebted to Dr. Bake, however, for some further information, (i) An impression of Rgvedic chant from a Nepalese VājasaneyĪ (recorded in 1956), who, presumably, had heard these recitations from eastern Rgvedls. (ii) An impression of Rgvedic chant from a Brahman in Belgaum (recorded in 1938). This testimony is not so valuable as that of (i) since the reciter had gathered his impressions probably in Mahārāstra. On the other hand, to judge from his recitation, he may have been a Rgvedi from Madras. Illustrations are given at the end of the article.

^{page 87 note 1} In C's recitation the low-tone was nearly always a minor third below mid-tone while the hightone was frequently a very full third above. The correspondence of the incidence of low-tone between B and C showed almost negligible variation. In the case of A, however, the passages recited were of insufficient length to draw definite conclusions with regard to low-tone.

^{page 87 note 2} The term ‘anudātta’ is used in this article to denote those syllables which are marked with the anudatta sign (a horizontal stroke below the line) in the Rgvedic text. An exception is made (for descriptive convenience only) where a succession of two or more anudātta signs are found at the beginning of a line, in which case only the last syllable so marked is termed anudatta. The preceding syllable(s) in such a series, as well as all unmarked syllables following upon svarita, are termed ‘pracaya’. The term ‘anudātta’ corresponds to the sannatara (‘more constricted’.) anudātta of Pāṇini (Astādhyāyī, I, ii, 40), who uses the term anudātta for all syllables not udātta or svarita.

^{page 87 note 3} Svaritas (marked with a vertical stroke above the line) on short vowels in closed syllables are relatively rare.

^{page 87 note 4} See p. 88, n. 2.

^{page 88 note 1} The combination of short vowel plus homorganic nasal has the effect of counting the vowel long. At the end of a line, a short vowel plus -m or -ḥ has the same effect.

^{page 88 note 2} Pracaya means ‘accumulation’, i.e. an accumulation of (in Pānini's sense) anudāttas, i.e. unimportant syllables. The Rkprātiśākhya, III, 11, states: svaritād anudāttānām paresām pracayab svarab /udāttasrutitām yānty ekam dve vā bahūni vā/ ‘The anudāttas following a svarita have the accent (known as) pracaya. One, two or more acquire the sound of the udātta’. The tones at the beginning of a verse cannot always be determined accurately. This happens for two reasons: (i) a hymn is generally recited with a mounting tension and excitement, accompanied by a very gradual rise in pitch throughout (i.e. the mid-tone of the last verse of a hymn of some 20 verses may be a third above the initial mid-tone); (ii) hymns are often recited by Brahmans in a group, each individual taking a verse in turn, and the habit of rather slurred entry which thus develops may be carried over by a reciter into individual recitations.

^{page 88 note 3} See p. 87, n. 3.

^{page 88 note 4} In A's recitation, all svaritas on long vowels before pracaya showed a fall of a semitone. The lower tone in this case lasted for nearly the whole length of the second half of the vowel (as opposed to the swift return to mid-tone heard in the recitation of B and C). C sometimes recited this type of svarita on a long vowel before anudātta. It is possible that A's recitation showed a Yajurvedic influence, since some of the passages recited by him are given by Rgvedic Brahmans on certain occasions in Yajurvedic fashion. The anudātta, however, was almost invariably recited on the high-to-mid(-low)-tone glide by A, showing an identical pattern with that of B and C.

^{page 88 note 5} The remarks concerning the tones of the svarita apply also to the svaritas known as jātya and ksaipra. Where these svaritas are immediately followed by udātta, however, the recitation becomes rather artificial. As well as the rise and fall of the svarita being maintained, a further high-tone is recited, presumably to indicate the anudātta required to introduce the following udātta. According to the context (i.e. whether the svarita vowel is short or long, and whether it is preceded or not by an udātta syllable in the case of monosyllables), a waver or Kampa may be heard one, two, or three times.

^{page 89 note 1} This feature was much more strongly marked in C's recitation than in that of A or B.

^{page 89 note 2} Svarita is a combination of udātta and anudātta (Pānini, Astādhyāyī, I, ii, 31: samāhārah svaritah)), the first part (half the length of a short vowel) being udātta (ibid., I, ii, 32: tasyādita udāttam ardhahrasvam). The Rkprātiśākhya, III, 2, also allows a vowel (long as well as short) to be divided equally: tasyodāttatarodāttād ardhamātrārdhām era vā ‘ a half mora of it (i.e. the svarita vowel) is more prominent than the udātta or a half is the same’. Both Pānini and the Rkprātiśākhya agree that the remainder of svarita is anudātta (i.e. has the quality of any unimportant syllable), while the Rkprātiśākhya adds that it is heard on the same tone as the udātta (Rkprātiśākhya, III, 3: anudāttah parah śesah sa udāttaśrutih). That these dicta on the exact nature of svarita were regarded as doubtful even in ancient times may be gathered from Patanjali's remarks under Astādhyāyī, I, ii, 32. Uvata, commenting on the above citations from Rkprātiśākhya, wisely leaves the problem open. For present purposes, it is convenient to list the components separately.

^{page 89 note 3} Apart from those cases where the jātya svarita is the principal accent or where udātta has disappeared (ksaipra) or has become svarita (praślista and abhinihita). In the last case (praślista and abhinihita svarita), did coalescence or elision not take place, there would be glottal occlusion after the udātta vowel with almost complete disappearance of the initial component of svarita. Abhinihita svarita has doubtless spread out from contexts with original -h, an influence which, historically, can be seen gathering strength in the case of praślista svarita. In addition to the above points, many words are accentless either by nature or for syntactical reasons.

^{page 90 note 1} There is occasional evidence for a certain counterbalance between anudatta and svarita, particularly in the context mentioned, but also where the svarita is introduced by an aspirate consonant. The anudātta may present a more subdued quality, almost indistinguishable from pracaya, and the svarita is recited with greater intensity. No doubt, according to strict practice, this feature is a dosa (Rkprātiśākhya, III, 18: svaritaip, nātinirhanyāt ‘one should not intensify a svarita excessively ’), but the conclusion might be that the udatta is the fulcrum between the constriction of mahāprāna and its subsequent release. Allied with this tendency are, probably, the practices mentioned in Rkprātiśākhya, III, 12: ke cit tv ekam anekarp, vā niyacchanty antato 'ksaram / ā vā śesān niyuktam, tūdāttasvaritodayam / ‘some authorities, however, constrict one or more syllables back from (the udātta), or as far back as a remaining (pracaya); it is proper though (to constrict pracaya thus only) before udātta and svarita’. The Rkprātiśākhya continues (III, 13): niyamarp, kāranād eke pracayasvaradharmavat /pracayasvara ācārah śākalyānyatareyayoh /‘constriction (over several syllables) would result, according to some authorities, (in the syllable preceding udātta or svarita) possessing the qualities of the pracaya accent; the practice (of constricting pre-udātta and -svarita pracaya) is according to (śākalya and Ānyatareya’.

^{page 90 note 2} Where anudātta occurs in place of svarita (i.e. where the context is the syllable between two udāttas), the range of the preceding alpaprāna is curtailed (since the syllable now following the first udātta does not bear this quality). The result is that the termination of such an udātta acts as if in hiatus (see 7 (c)). There is evidence here also for the occasional counterbalance which was described in the preceding note. It has, at least, the revered authority of Gārgya, Kāśyapa, and Gālava, who maintained the svarita in this context (Pānini, VIII, iv, 67). There was, it appears, even in those days an important body of scholars who were repelled by the thought of an accented syllable with mahāprāna at both ends. According to this point of view, (ii) (the quality of the initial component of svarita) is stronger than (iv) (the quality of anudātta). That alpaprana was strong in this position (i.e. over the consonant(s) immediately following the udātta vowel, with a corresponding slurring of the distinction between the preceding anudatta and the udātta) in certain areas is a contention which might be supported by the evidence of the Germanic soundshifts.

^{page 90 note 3} In these cases, the anudātta is never high-to-low-tone. In the first case (that of the anudātta final), however, the mahāprana is fully released where the following word has initial ks-, and, in this context, a high-to-low-tone anudātta may be heard (cf. Rkprātiśākhya, VI, 15). The exceptions given here to the normal incidence of low-tone on the udātta (i.e. where the svarita s long) rarely operate on the last udātta(s) to occur in a line.

^{page 91 note 1} This feature (and that of nasality, 7 (a)) was brought to the notice of the writer by Professor J. Brough, for whose interest and encouragement in the analysis of these recitations the writer is extremely grateful.

^{page 93 note 1} The sign eq1 is the ‘broken’ svarita used by southern Rgvedīs on all long vowels and by Taittirīyans on long vowels in certain contexts. Half the vowel is on one tone and the remaining half on the higher tone.