we might try to remove the difficulty by omitting vaso at the end of the refrain, but this would be against the general character of these hymns. We want the last word vaso, if possible, at the end of both lines. But, if so, we must admit two cases of synizesis, or, if this seems too clumsy, we must omit tvâ.

I shall now proceed to give a number of other examples in which the same consonantal synizesis seems necessary in order to make the rhythm of the verses perceptible to our ears as it was to the ears of the ancient Rishis.

The preposition anu takes synizesis in

I, 127, 1. ghrĭtāsyă vībhrâ̄shtĭm ă͡nŭ* vāshtĭ sōkĭshâ̄. Cf. X, 14, 1.

The preposition abhi:

I, 91, 23. râ̄yō bhâ̄gām săhăsâ̄vānn ă͡bhĭ* yūdhyā.

Here Professor Kuhn changes sahasâvan into sahasvah, which, no doubt, is a very simple and very plausible emendation. But in altering the text of the Veda many things have to be considered, and in our case it might be objected that sahasvah never occurs again as an epithet of Soma.

[paragraph continues] As an invocation sahasvah refers to no deity but Agni, and even in its other cases it is applied to Agni and Indra only. However, I do not by any means maintain that sahasvah could not be applied to Soma, for nearly the same arguments could be used against sahasâvan, if conjecturally put in the place of sahasvah; I only wish to point out how everything ought to be tried first, before we resort in the Veda to conjectural emendations. Therefore, if in our passage there should be any objection to admitting the synizesis in abhi, I should much rather propose synizesis of sahasâvan, than change it into sahasvah. There is synizesis in maha, eg. I, 133, 6. ăvār mă͡hă* īndră dâ̄drĭhĭ srŭdhî̄ nāh. Although this verse is quoted by the Prâtisâkhya, Sara 522, as one in which the lengthened syllable dhî of srudhî does not occupy the tenth place, and which therefore required special mention, the original poet evidently thought otherwise, and lengthened the syllable, being a syllable liable to be lengthened, because it really occupied the tenth place, and therefore received a peculiar stress.

The preposition pari:

VI, 52, 14.

mâ̄ vō văkâ̄msĭ pă͡rĭ*kākshyâ̄nĭ vōkām,
sūmnēshv̆ īd vō̆ āntămâ̄ mădēmā.

[paragraph continues] Here Professor Kuhn (Beiträge, vol. iv, p. 197) begins the last pada with vokam, but this is impossible, unless we change the accent of vokam, though even then the separation of the verb from mâ and the accumulation of two verbs in the last line would be objectionable.

Hărĭ is pronounced as hă͡rĭ*:

VII, 32, 12. yă īndrō hă͡rĭ*vâ̄n nă dăbhāntĭ tām rĭpāh.

II, 18, 5. â̄ kātvâ̄rīmsătâ̄ hă͡rĭ*bhīr yŭgâ̄nāh.

Hence I propose to scan the difficult verse I, 167, 1, as follows: