Ever since the Registrar General began to give mortality rates for different marital condition groups, it has been observed that, in every age group and for both sexes, the widowed have higher mortality rates than the married. In his 1959 Annual Statistical Review, the Registrar General made some analysis of this differential in terms of causes of death. The excess mortality as compared with the married appeared to affect the diseases listed in Table 1. The Registrar General's commentary asked ‘Leaving aside the question of possibility of error (in statements of marital condition) is it possible that the effect of death of a spouse is such as to increase the likelihood of death in the surviving member of the partnership? Here the basis of discussion has little factual support but it seems right to suppose that in the period immediately following bereavement the general state of ‘shock’ indeed is such as to increase likelihood of death. In the majority of young and middle-age widowed persons, however, this period passes and adjustment takes place, often followed by remarriage. There will be some selection here similar in many respects to that affecting the decision whether single persons should marry. With the older person, however, adjustment is more difficult to attain, remarriage is less common and the surviving partner, particularly the man, may live under relatively unfavourable conditions.’