This appendix describes the method used to maximise the period of time over which age at marriage was calculated in chapter 5. The problem arises because the parishes making up a particular group may not all begin to produce reliable data from the same date. To wait until all were yielding reliable data would mean a late starting date, but to make no correction for ‘missing’ parishes in the early decades of a tabulation would carry the risk that apparent changes were due simply to the entry of a new parish into the group at a particular point in time. It is necessary to produce a series in which a ‘missing’ parish contributes in a rational, if arbitrary, fashion to the mean for the group as a whole even before it has come ‘on stream’.

Consider the case of Birstall and group 2 using the data contained in table 5.2. The table suggests that a correction for Birstall is only necessary for 1630–9 and earlier decades, since the parish made a contribution to the group total in 1640–9 at much the same level as in the next decade. Appearances are deceptive, however. Reconstitution began in Birstall in 1595, so that, if truncation bias is to be avoided, marriage age information can only be used from 1645 onwards, and the proportional contribution of Birstall in the 1640s shown in table 5.2, therefore, is based on a total of marriages for 1645–9 rather than for the whole decade.