The research design which led to the division of the 26 reconstitution parishes into four groups has been described elsewhere, as has the reason for treating group 2 as the reference group to which data drawn from the other groups should be aligned. The issue covered in this appendix is the implementation of this strategy when dealing first with infant and child mortality, and then with adult mortality.

Infant and child mortality

The four groups were chosen so that there were long overlaps between them. It will be recalled that group 1 covers the period 1580–1729 (15 parishes); group 2, 1600–1729 (20 parishes); group 3, 1680–1789 (18 parishes); group 4, 1680–1837 (8 parishes). The separate tabulations for each group showed that in overlap periods infant and child mortality rates in the groups differed, sometimes considerably, and that therefore adjustment between the rates for each pair of adjacent groups was necessary to produce a single series. The adjustment process might appear to involve reconciling two conflicting desiderata. On the one hand, it is the ratio of rates drawn from two adjacent groups over a period close to the changeover between them which is at issue. Therefore, adjustment ratios based on a brief overlap period are attractive since there is no certainty that the appropriate adjustment ratio would be constant over the whole period of a long overlap.