different from the final e of sarve or âste, or the final o of sarvo for sarvas or mano for manas *. It should be remarked that in Greek, too, the final diphthongs corresponding to the e of sarve and âste are treated as short, as far as the accent is concerned. Hence ἄποικοι, τύπτεται, and even γνώμαι, nom. plur. In Latin, too, the old terminations of the nom. sing. o and u, instead of the later us, are short. (Neue, Formenlehre, § 23 seq.)

Here the o of gopâ is treated as short, in order to get

instead of

1121

which is perfectly

legitimate at the end of an Ushnih.

2. The long ↑ and û are treated as short, not only before vowels, which is legitimate, but also before consonants:

vii. 62, 4. dyāvābhumi adite trâsîthâm nah (see Trisht. § 5).

The forms isiya and râsîya in vii. 32, 18, occur at the end of octosyllabic or Gâyatra pâdas, and are therefore perfectly legitimate, yet Professor Kuhn would change them too, into īsiya and râsîya. In vii. 28, 4, even mâyî is treated as mâyĬ (see Trisht. § 5); and in vii. 68, 1, vītām as vitam. If, in explanation

* A very strong divergence of opinion is expressed on this point by Professor Bollensen. He says: O und E erst später in die Schrifttafel aufgenommen, bewahren ihre Länge durch das ganze indische Schriftenthum bis ins Apabhramsa hinab. Selbstverständlich kann kurz o und e im Veda erst recht nicht zugelassen werden.' Zeitschrift der D. M. G., vol. xxii. p. 574.

of this shortening of vîtam, vîhi is quoted, which is identified with vìhi, this can hardly be considered as an argument, for vihi occurs where no short syllable is required, iv. 48, 1; ii. 26, 2; and where, therefore, the shortening of the vowel cannot be attributed to metrical reasons.

3. Final m followed by an initial consonant is allowed to make no position, and even in the middle of a word a nasal followed by a liquid is supposed to make positio debilis. Several of the instances, however, given in support, are from Gâyatra pâdas, where Professor Kuhn, in some of his later articles, has himself allowed greater latitude; others admit of different scanning, as for instance,

i. 117, 8. mahah kshonasya asvina kanvāya.

Here, even if we considered the dispondeus as illegitimate, we might scan kanvaya, for this scanning occurs in other places, while to treat the first a as short before no seems tantamount to surrendering all rules of prosody.

4. Final n before semivowels, mutes, and double n before vowels make no position *. Ex. iii. 49, 1. yasmin visva (Trisht. § 5); i. 174, 5. yasmiñ kākan; i. 186, 4. sasmin(n) undhan †.

*Professor Kuhn has afterwards (Beiträge, vol. iv. p. 207) modified this view, and instead of allowing a final nasal followed by a mute to make positio debilis, he thinks that the nasal should in most cases be omitted altogether.

+ Here a distinction should be made, I think, between an n before a consonant, and a final n following a short vowel, which,

5. Final Visarga before sibilants makes no position *. Ex. iv. 21, 10. satyah samrat (Trisht. § 5). Even in i. 63, 4. kodih sakha (probably a Gâgata), and v. 82, 4. savih saubhagam (a Gây. § 7), the long î is treated as short, and the short a of sakhâ is lengthened, because an aspirate follows.

6. S before mutes makes no position. Ex. vi. 66, 11. ugrâ aspridhran (Trisht. § 3).

7. S before k makes no position. Ex. visvaskandrah, &c.

8. Mutes before s make no position. Ex. rakshas, according to Professor Kuhn, in the seventh Mandala only, but see i. 12, 5; kutsa, &c.

9. Mutes before r or v make no position. Ex. susipra, dirghasrut.

10. Sibilants before y make no position. Ex. dasyûn.

པ

11. R followed by mutes or sibilants makes no position. Ex. ayur givase, khardih, varshishtham.

12. Words like smaddishtîm &c. retain their vowel short before two following consonants.

We now proceed to consider a number of pro

according to the rules of Sandhi, is doubled, if a vowel follows. In the latter case, the vowel before the n remains, no doubt, short in many cases, or, more correctly, the doubling of the n does not take place, e. g. i. 63, 4; 186, 4. In other places, the doubling seems preferable, e. g. i. 33, 11, though Professor Kuhn would remove it altogether. Kuhn, Beiträge, vol. iii. p. 125.

* Here, too, according to later researches, Professor Kuhn would rather omit the final sibilant altogether, loc. cit. vol. iv. p. 207.

sodial rules which Professor Kuhn proposes to repeal in order to have a long syllable where the MSS. supply a short:

1. The vowel ri is to be pronounced as long, or rather as ar. Ex. i. 12, 9. tasmai pâvakă mrilāya is to be read mardaya; v. 33, 10. samvaranasya risheh is to be read arsheh. But why not samvaranasya risheh (i. e. siarsheh) ?

2. The a privativum may be lengthened. Ex. agarah, amritah.

こ

こ

^

3. Short vowels before liquids may be long. Ex.

こ

こ

こ

こ

こ

narah, tarutâ, tarati, marutâm, harivah, arushi, dadhur

Ex. ganân, sanitar, tanûh, upa nah.

5. Short vowels before the ma of the superlative

may be lengthened. Ex. nritama.

6. The short a in the roots sam and yam, and in am (the termination of the accusative) may be lengthened.

7. The group ava is to be pronounced aua. Ex.

avase becomes auase; savita becomes sauita; năvă becomes naua.

8. The group aya is to be changed into aiă or ea. Ex. nayasi becomes naiasi.

31

9. The group vă is to be changed into ua, and this ua to be treated as a kind of diphthong and therefore long. Ex. kanvătămah becomes kānuatamah; varunah becomes uarunah.

10. The short vowel in the reduplicated syllable of perfects is to be lengthened. Ex. tatanah, dadhire.

11. Short vowels before all aspirates may be lengthened. Ex. rathâh becomes rathâh; sakhâ becomes sakhâ.

12. Short vowels before h and all sibilants may be lengthened. Ex. mahini becomes mahini; usigâm becomes usigâm; rishate becomes rishate; dăsat becomes dasat.

13. The short vowel before t may be lengthened. Ex. vâgavatah becomes vâgavatah; atithih becomes atithih.

14. The short vowel before d may be lengthened. Ex. udaram becomes udaram; ud ava becomes ud ava.

15. The short vowel before p may be lengthened. Ex. apâm becomes apâm; tapushim becomes tapushim; grihapatim becomes grihapatim.

16. The short vowel before g and g may be lengthened. Ex. sânushag asat becomes sânushag asat; yunagan becomes yunagan.

Let us now turn back for one moment to look at the slaughter which has been committed! Is there one single rule that has been spared? Is there one single short syllable that must always remain short, or a long syllable that must always remain long? If all restrictions of prosody are thus removed, our metres, no doubt, become perfectly regular. But it should be remembered that these metrical rules, for which all this carnage has been committed, are not founded upon any à priori principles, but deduced by ancient or modern metricians from those very hymns which seem