DISCUSSION 71 Mandals out of 57, all the Municipalities and the ZPP seat also. Of these 68 local political elite, 8 were Reddis (all won on the Congress ticket), 24 Kammas (TDP-6 and Congress-17) and 32 belonged to other categories. This actually proves the dominant position of Kammas in district politics irrespective of the ups and downs of political parties in the elections. This dominant position can be understood only in relation to their control over land and capital in the district. A comparison between Kohli's empirical data and the data collected by me reveals the following picture. Kohlfs data suggests that during Congress rule both Kammas and Reddis were represented equally among the district political elite (8 each), but the Reddis were represented as the single largest group within the Congress Party. Under TDP rule, while the Kammas improved their position considerably, the position of Reddis declined drastically. In contrast our data suggests that both under Congress and TDP rule, Kammas enjoyed a predominant position constituting the single largest group among the district political elite. Both in the Congress and the TDP, when in power, Kammas have been represented as the single largest group. The position of Reddis declined under TDP rule, but not as drastically as Kohli suggests. With regard to SCs and STs, Kohli's data suggests that their representation declined when TDP came to power. But our data suggests that the position has not changed at all. Regarding BCs our data too suggests that there is improvement in their representation under TDP, but not as significant as Kohli suggests. The other two variables which the author takes into account for the analysis of caste and party linkages are: (a) Pattern of voting behaviour of different castes (who votes for which party), and (b) the self-perceptions of the people about those who are in and out of power. The sources of data for these two variables are the author's interviews with a few members of the political elite (p. 78). But as author has himself noted, without a detailed survey, no firm conclusions can be drawn concerning these two variables on the basis of such impressionistic data (p. 78). Apart from the theoretical problems central to his approach, which have already been highlighted by Neera Chandoke in her review, we only wish to introduce a note of caution on the issue of linkages between caste and politics exclusively in terms of the caste background of the candidates of various political parties in the elections. As is shown above even the empirical data used by Kohli is faulty and hence his understanding of the linkages between caste and the political process in India is questionable. DAVULURI VENKATESWARLU Department of Political Science University of Hyderabad, Hyderabad