M ost histories of interwar unemployment have assumed a homogeneity INwithin the labour markets of developed industrial capitalism. Such studies claim that, while the Edwardian economy was still marked by preindustrial residues of casualism and underemployment, this was swept away by the I914-8 conflict. Interwar labour markets can thus be analysed by assuming a clear distinction between the condition of the unemployed and the employed: a distinction that was neither appropriate nor possible in the Edwardian period. The statistics of the unemployment insurance scheme have been interpreted in this manner. While historians have accepted that these provide 'a highly imperfect measurement of the phenomenon', this has commonly been seen as a problem of inaccuracy of collection methods, to be remedied by more sophisticated adjustments or disaggregation.1 This uncritical acceptance of analytical categories derived from contemporary administrative practices has been compounded by models of the labour market developed after the Second World War. The labour markets of the I920s and the i98os have been seen as broadly similar, distinguished only in that the earlier period was more manual in orientation and enticingly free of those structural constraints imposed by the growth of state regulation since the I940s. If the interwar labour market is recognized as 'modern', then the key break with the past occurred during the First World War. There is, therefore, a general assumption that this war somehow swept away the archaic employment practices of the Edwardian period, which had been widely condemned by labour market reformers before I914. This paper questions this assumption. As a result of chronic labour shortages, the First World War certainly reduced the incidence of underemployment-but its effects were largely temporary. As unemployment mounted in the winter of I920-I, traditional forms of labour management reasserted themselves. Various forms of underemployment re-emerged in those sectors of the labour market where they had long been practised. Gross irregularity of employment, a major cause of poverty and pauperism, was endemic in many trades and was present on the fringes of nearly all the remainder. An appreciation of this fact causes us firstly to qualify the 'optimist' view of the slump years, which argues that the social impact of the recession was limited and which claims that living standards rose among most of the working class in this period.2 Secondly, it renders conventional ways of defining and