¹ In this paper the terms ‘heavy’ and ‘light’ (syllable) will be used in the sense of guru- and laghu- as used by the Hindu metricians: viz., a ‘heavy’ syllable is one containing a long vowel or diphthong, or a short vowel followed by more than one consonant; other syllables are ‘light’.

² The 3 pl. of dhā is only dadhi-re (IE *de-dh-rai), occurring over 60 times in RV. It is historically required, with i for IE . No root in ā has any 3 pl. perfect form lacking this root-vowel i. The ending -re (IE -rai) appears as -ire (IE -rai) by our law only after a heavy syllable ending in a consonant, but as -re after a vowel and after a light syllable ending in a consonant. There is no exception in RV. Wackernagel I §25b correctly explains the exclusive appearance of ir (never ur, cf. §46 below) for IE r in this ending as due to the influence of roots in -ā and set roots (dadhi-re, etc.), where i represents IE . With roots ending in , the historic 3 pl. form in -re occurs only in cakre, dadhre, as explained above. But the analogical -rire is not much commoner; to be sure cakrire occurs 15 times, but otherwise we find only dadhrire once, and jabhrire three times. These secondary and analogical forms gave rise to the further analogical forms in -rire from roots in final consonants (cikit-rire, etc., six forms in RV., Macdonell, VGr. §§484.4a, 485 end), which are scarcely explainable except on the assumption that for a time such forms as cakre and cakrire, from roots in r, coexisted as 3 plurals. The proportion was cakre (3 pl.): cakrire = cikit-re (5 times in RV.): cikit-rire (twice).—To call the i of cakrire a ‘connecting vowel’, and regard cakr- as the root and -re as the ending, fails to explain the fact that this i does not occur before other endings beginning with consonants; e.g. the 2 sing, is cak r- se, not *cakri se. Whereas other roots which show the ‘connecting vowel’ i show it alike before the 3 pl. -re and before other consonantal endings. In passing, however, it may be noted that the analogical extension of the use of this ‘connecting vowel’ before consonantal endings in the perfect was clearly due primarily to the occurrence (under our law) of -ire in the 3 plural after a heavy syllable (but -re after a light): e.g. jajñ-ire, but dadś-re. This is proved by the fact that (surely first) -ire, and (then also) -i e,-ima, occur after roots in final consonants when the preceding syllable is heavy, but corresponding forms without i when it is light (Macdonell 484.4). Of course the historically quite distinct i from IE , in roots in -ā and se t roots, undoubtedly helped in the extension of this ‘connecting vowel’ i.