In the Family Expenditure Survey (a continuing form of survey undertaken from the beginning of 1957 onwards by the Social Survey, Central Office of Information, on behalf of the Ministry of Labour), members of private households at a two-stage stratified sample of addresses in the United Kingdom are visited by interviewers and invited to keep fully itemised records of all their expenditure for a period of fourteen days. In each household in which every member who is working full-time or is 16 years of age or over agrees to keep records, a separate record book is provided for each spender for each of two consecutive seven-day periods. All spenders in a co-operating household keep records for the same period, but the survey is so conducted that, throughout the year, fresh households are visited each week and different households begin recording their expenditure on different days, normally days other than Saturday or Sunday. Information is sought during interviews about some types of expenditure, e.g. insurance, railway season tickets, electricity supply, over a period longer than two weeks. Information about income is also obtained. For convenience, the seven-day period covered by the first set of record books of a household is referred to as the first week of the survey of the household's expenditure and the following seven-day period as the second week, irrespective of the day on which the records begin. Analysis of the results of the survey during 1957 and 1958 showed that, for various groups of items, the average per household of the expenditure recorded by households during the first week was significantly higher than the average of that recorded during the second week.1 Such inter-week variations were found in earlier surveys; for example, a series of small-scale experimental surveys in 1950 and 1951,2 and the large-scale Enquiry into Household Expenditure in 1953-4.3, 4 Variations of this kind have also been noted in surveys in other countries; for example, in Israel.5 136