t^^8 SCIENCE IN AFRICA 



usually been carried out in British territories at ten-year intervals, 

 and in French every five years. Secondly there are the systems 

 established in certain localities, of registering vital events such as 

 births, deaths, and marriages. Thirdly, certain special studies have 

 been made for defined areas giving much fuller data on the preva- 

 lence of diseases and general health conditions of samples of the 

 population. 



In the Union of South Africa the first simultaneous census of all 

 the provinces took place in 1904, and subsequently in 19 11, 1921, 

 and 1 93 1. On the last occasion financial stringency reduced its 

 scope to a census of Europeans only, with little more than estimates 

 for other categories of the population, but in May 1936 an addi- 

 tional complete census was made. This showed a material increase 

 for all categories of the population from 1921 to 1936, but it was 

 not considered that all the figures could be accepted with confi- 

 dence. Censuses for the South African protectorates were made in 

 1 9 II, 1 92 1, 1 93 1 5 and 1936. In Southern Rhodesia censuses, 

 started in 1901, have taken place more frequently, in 1904, 1907, 

 191 1, 1 92 1, 1926; 1 93 1, and 1936, the last mentioned being simul- 

 taneous with that in the Union. Figures for native population are 

 only estimated, being based on the number of tax-payers multi- 

 plied by a selected figure, usually 3J or 4. 



In most of the colonial territories a regular decennial census has 

 been made. In Northern Rhodesia, in 191 1, 1921, and 1931, the 

 European, Asiatic, and coloured races were actually counted, 

 while the Africans were estimated in a similar way to that used 

 in Southern Rhodesia. In both the Rhodesias special returns 

 were made from the employers of labour on mines. In Nyasaland, 

 censuses have been more comprehensive and more accurate than 

 in many other territories. In that of 191 1 all non-natives were 

 enumerated, but the estimate of native population was based on 

 the number of hut taxes paid, multiplied by 2-8, a figure reached 

 by counts in selected villages. The more recent censuses of 192 1, 

 1926, and 1 93 1 involved a considerable staff of enumerators, and 

 the records on the day of census were preceded by three weeks of 

 work by the enumerators to serve as a check on the final figures. 

 This is a system used in India and elsewhere [see P. G. Edge 1932, 

 p. 19). In 1 93 1 the population figures were supplemented by 



