1920s. By then, language erosion had become markedly perceptible and was measurable by virtue of population censuses The 1891 Census was the first to inquire in Wales into the language spoken by individuals in each household-whether Welsh, English or both. As Geraint H. Jenkins, the general editor to the Social History, observes, this was by no means an uncontroversial or neutral exercise. The Registrar General and his officials got caught up in the flak of the quasi-nationalist campaigns of Lloyd George and cultural activists associated with the (first) Welsh Language Society. In their published statements, officials of the General Census office revealed a certain insensitivity to Welsh interests (not entirely abandoned, even now!) and implicitly they were keener to examine how far Wales had progressed in becoming anglicized. That this should be so was due to the fact that within Wales itself language shift was a focus of significant debate by 1891, between 'modernisers', who saw no future for the vernacular in the rapidly changing English (or imperial) world of commerce and ideas, and, essentially, 'compromisers', who felt that bilingualism was viable-by 1891 few held out prospects for Welsh mono- lingualism and the Census certainly tested this. Thus, the Census became the object of agitation on the part of these rival partisans and extended even to the local enumeration officials responsible for distributing forms in either language and for collecting the data. Consequently, the returns cannot be taken as being entirely objective or dispassionate in their details, but they are probably no more misleading than the other census criteria, as various studies have shown. Any limitations notwithstanding, since the enumerators' returns form the largest cache of information about the language in the nineteenth century, it was inevitable that the ambitious project researching the Social History of the Welsh Language should undertake an analysis of this material. The current volume presents the conclusions of the two principal investigators into the language patterns found in about twenty representative areas of Wales. In all, the sample represents approximately five per cent of the population, a reason- able proportion in the circumstances. The communities examined were not exactly randomly chosen; rather they were taken to replicate the distribution of population within Wales and, more importantly, to represent distinctive occu- pational patterns. Centres of coal mining and the metalliferous trades predominate. This is also important inasmuch as one of the aims was to test the Brinley Thomas thesis about the impact of industrialisation on population growth and migration in Wales and hence on patterns of cultural preservation. The authors' conclusions suggest that Thomas was too optimistic or sweeping in his argument. Migratory patterns were highly complicated, responding to demand for both new and replacement labour. While industry did lead to the development