Detection without correction: problems in assessing the quality of English ecclesiastical and civil registration.

Reliable, good quality source material is required for any demographic study. By selecting specific examples from York during the parish register period and Sheffield during the civil registratiion period deficiencies in both ecclesiastical and civil registration are discussed with reference to how they affect infant and adult mortality calculations. In particular, the extent to which the deaths of very young infants were registered is considered in detail. Bourgeois-Pichat's biometric test, Farr's early life tables and Coale and Demeny's model life tables have all been used to correct inaccuracies within original sources. We consider the limitations of each of these methods and suggest that a reassessment of the quality of vital registration data and the methods used to make corrections is needed in order to make further advances in historical demography possible.