Contrary to the "marriage squeeze" hypothesis, brides and grooms appear to adapt to rather than to be constrained by the age distribution of partners available, even in extreme conditions. This is evident from the substantial variability through time in the proportions who marry a partner at each single year age difference. Variations in the series are systematically associated with variation in the corresponding cohort sizes. These features suggest that age preferences are not rigid, as the "marriage squeeze" hypothesis assumes, but flexible. "Marriage squeeze" is found to be virtually absent in England and Wales during the period covered, despite large fluctuations in births that might have been expected to result in "squeeze". Two formulations of age preference are discussed. Direct evidence on preferences suggests that they are flexible, in a specific sense, and that the marriage market could as a result operate in a flexible way. A new view of marriage market dynamics is developed, arising out of these empirical findings, encompassing preferences, marriage candidacy and related aspects. The paper concludes by proposing a new approach to modelling the marriage market.
