108 KALIDASA AND THE GUPTAS
practice, and Kalidasa observes them carefully. Of the four Vipula forms he uses the last once only ; the figures ' for the other three out of 1410 half-stanzas in the epics are 46, 27, and 41, or 8-15 per cent, showing that the third Vipula was Kali- dasa's favourite. It is interesting to note that in the form of the syllables preceding the first Vipula Kalidasa shows special care
to select that form (^- ) which is not allowed in the second
Vipula as against that (^-^-) which is permitted in both. The Kumarasambhava has 11 cases of the first to 3 of the second form, the Raghuvat'iga 31 to 1 ; this doubtless indicates increasing care to secure elegance, and it accords with this that in the Kumarasambhava alone is the fourth Vipula found. 2
1 l'"or the Raghuvaiifa they are 32, 18, 27 out of 1096 ; Jncobi's figures (IS. xvii. 444 f.) are corrected from SIFT. /. c. The percentage in Bharavi is 0.-6 ; Magna 27.15 ; Bilhana 8-64 ; (J!r!harsa 0-53 ; Kumaradasa 2-35.
2 X agliuvahfa, mi. 71, should perhaps be read dviliyahemaprakaram. In Kumarasambhava, vii. 11 on one reading position 15 neglected as in Qifupalavadka, x. 60, both dubious (SIFI vili. ii. 7). For the schemes of the metres see chap, xx, §4-
