In the colonial anthropology of India developed in connection with the decennial censuses in the late nineteenth century, caste and religion were major topics of enquiry, although caste was particularly important. Official anthropologists, mostly members of the Indian Civil Service, reified castes and religious communities as separate ‘things’ to be counted and classified. In the 1911 and later censuses, less attention was paid to caste, but three officials – E. A. Gait, E. A. H. Blunt and L. S. S. O'Malley – made significant progress in understanding the caste system by recognising and partly overcoming the problems of reification. In this period, however, there was less progress in understanding popular religion. The Morley-Minto reforms established separate Muslim electorates in 1909; communal representation was extended in 1921 by the Montagu-Chelmsford reforms and again by the 1935 Government of India Act, which also introduced reservations for the Untouchable Scheduled Castes. Gait and Blunt were involved in the Montagu-Chelmsford debates, and Blunt in those preceding the 1935 Act. In the twentieth century, the imperial government's most serious problems were the nationalist movement, mainly supported by the middle class, and religious communalism. But there were no ethnographic data on the middle class, while the data on popular religion showed that Hindus and Muslims generally did not belong to separate communities; anthropological enquiry also failed to identify the Untouchable castes satisfactorily. Thus, official anthropology became increasingly irrelevant to policy making and could no longer strengthen the colonial state.