________________ and Yudhisthira Mimamsaka accept that Panini composed the sivasutras, may be by divine inspiration37. . 2.3. In these rules Panini refers to a group of verbs and its subgroups38, which serve as bases for formulating verbal forms. This group is called 'Dhatupatha', and the verbal roots (dhatu) are defined by listing in the Ast. sutra 1.3.1 : "Bhuvadayo dhatavah". There are about 2000 roots which are given in an appendix called Dhatupatha, verbal root lexicon. A few more occur in the grammatical statements in the Astadhyayi and since being mentioned in the sutras these are called sautra dhatus by his commentators 39. Thus, in his grammatical treatment he does not distinguish between roots and derivatives 40, The Dhatupatha is preserved only as it was known to commentators such as Maitreya-raksita; it lists roots accompanied by meaning entries. For example, the first root entry is : bhu sattayam. Yudhisthira Mimamsaka shows that there is evidence in the Mahabhasya to show that at the time of Katyayana the root entries of the Dhatupatha were recited continuously, without the intervention of glosses. According to Nagesa, it was Bhimasena who entered the glosses in the Dhatupatha. However, Yudhisthira Mimamsaka has argued that a version of the Dhatupatha with glosses should be attributed to Panini4l. S. K. Belvalkar42 accepted that the Dhatupatha is correctly. attributed to Panini because he refers to roots and classes of roots in his rules and because the functions of markers in rules and the Dhatupatha are in accord with each other. 2.4. It is plausible to posit that lists comparable to Paninian ganas had been compiled before Panini's time, and there is some evidence in the discussions of the Mahabhasya to support this. Yudhisthira Mimamsaka43 has taken up arguments for and against attributing the Ganapatha to Panini and concluded that it should be ascribed to him. One thing is obvious about the relation between the Ganapatha and the rules of the grammar. These rules refer to groups of items, so that they presuppose these lists. For example, the rule 'Sarvadini sarvanamani' (P. Ast. 1.1:27) assigns the items 'sarva' ('whole, all') etc. to the class of pronominals (sarvanaman); the rule presupposes the set 'sarva' etc. Moreover, the list in question must have a particular order. For there are other rules which provide operations for subgroups of this set. For example, the rule 'Tyadadinam ah' (P. Ast. 7.2.102) states that the final sound of 'tyad' ('that') etc. is replaced by 'a', and there is a series of rules which apply to pronominals other than 'dvi' ('two') etc. It is clear that such rules require that lists have been set up and in particular orders. It is equally obvious, however, that these lists were set up in the first place in view of
