one does, and the utmost caution is therefore to be observed in approaching it. An identification has been suggested between the roots of the Old Jav. words kělěm, “to sink”, and surup, “to become submerged”, so that rup and lěm would be variants of one another. The present writer formerly agreed with this view, but has since had doubts on the subject. For there is also a Karo word kěneṅ, “to sink”; and why should we not be allowed to identify this nĕṅ also with lěm ? That, however, inevitably leads to the identification of něṅ and rwp ;and then there would be an end to all serious research.
51. In raising the question of the universal validity of phonetic law in IN, one must not use the phenomena of root-variation as evidence against such validity. There is a phonetic law of interchange of vowels as between Karo and Toba (§ 19), whereby every Karo ě is represented in Toba by o, and thus Karo ěněm, “six”, is onom in Toba. Now the pendant to the Karo ikěl, “to laugh” , is not eṅkol in Toba but eṅkel. But we have no right whatever to assert that this is an exception to the law; on the contrary, the matter stands thus: there was originally a root for to laugh with two variants, kel and kd, whereof the first has been preserved in Karo, the other in Toba.[1]
Determination.
52. In § 41 we were introduced to roots of two sounds, such as ju, “to aim, to have a certain direction”, li, “to buy”. Now alongside of these roots of two sounds there always run roots of three, which have therefore another consonant after the vowel. Beside the above-mentioned root li Karo has a root lih, “to appropriate to oneself” ; beside ju Sund. has jul in tujul, “to point, to direct a letter to someone” ; beside jd in Old Jav. ipi, “to sleep”, there is pit in ipit, “to talk in one's sleep”. Through many languages there runs a root kas alongside of ka, “to open, to release” , and the like:
- ↑ [See also Essay IV, § 348.]
