1 Day, A., ‘Kinship networks and informal support in the later years’, in Grebenik, E., Höohn, C. and Mackensen, R. eds., Later phases of the family cycle: demographic aspects (Oxford, 1989), 183–207.Google Scholar

2 Ruggles, S., ‘Family demography and family history: problems and prospects’, Historical Methods 23 (1990), 22–31.CrossRefGoogle Scholar

3 Wrigley, E. A. and Schofield, R. S., The population history of England, 1541–1871: a reconstruction (London, 1981).Google Scholar

4 For further discussion on the mortality and fertility changes of the late nineteenth and early twentieeth centuries, see Glass, D., Population: policies and movements in Europe (London, 1940)Google Scholar; Woods, R. I., ‘Approaches to the fertility transition in Victorian England’, Population Studies 41 (1987), 283–311CrossRefGoogle ScholarPubMed; and ‘Mortality patterns in the nineteenth century’, in Woods, R. I. and Woodward, J. eds., Urban disease and mortality in nineteenth-century England (London, 1984), 37–64Google Scholar; Coleman, D. and Salt, J., The British population: patterns, trends and processes (Oxford, 1992).Google Scholar

5 The results published by Wrigley and Schofield suggest that the decline in the crude death rate might have started before 1840, although the fall was only marginal; see Wrigley, and Schofield, , The population history of England, 534–5Google Scholar. The family reconstitution project currently under way at the ESRC Cambridge Group for the History of Population and Social Structure will provide further information about the mortality changes of this peirod; see Wrigley, E. A. et al. , English population history from family reconstitution (Cambridge, forthcoming).Google Scholar

6 The indices are calculated by the following procedures. First, based on the annual age-specific mortality data published in Mitchell's, B. R.Abstract of British historical statistics (Cambridge, 1962), pp. 38–44Google Scholar, the average age-specific mortality rate was computed for every ten-year period. Second, the average age-specific mortality rate of 1841–1850 was chosen as a baseline (that is, the index of this peirod is set to 100). By comparing this baseline with the average age-specific mortality rates recorded in later decades, indices of mortality change were calculated for other periods. When indices of this kind are used, choosing any particular value as the baseline may have some effect on the level of indices calculated for other periods and this may, in turn, influence the general conclusion. However, since the baseline used here is itself a ten-year average, it is likely to represent a less arbitrary baseline than that which would be obtained from a single year. Moreover, in the process of data analysis, for this study, data from other periods were also used as the baseline in order to examine the impact of baseline selection. The results confirmed the general trend, which will be discussed in the following sections.

7 These statistics, and the parity progression ratio which will be discussed below, are calculated from the 1911 census and the 1961 fertility survey data published by the General Register Office; see Census of England and Wales 1911, Vol. XIII, Part II:Google Scholar ‘Fertility of marriage’ (HMSO, 1927), Table 19: ‘Families classified by age of wife at marriage, duration of marriage and size of family (number of children born), and Census 1961 England and Wales, ‘Fertility tables’ (HMSO, 1966)Google Scholar, Table 1: ‘Married women by age at census and size of family’. All the women included were ever-married women who completed their families. These women were grouped by birth cohort. Each birth cohort included women born within a period of ten years, except those of the 1866 and 1881 cohorts. These two cohorts included only women born between 1866 and 1870 and those between 1881 and 1885, because the data of those born between 1871 and 1880 are not available.

8 Figure 4 shows that among women born between 1866 and 1870, the parity progression ratios at high parities were slightly lower than those of the later cohorts. This might be explained by the difference in data selection. For all other birth cohorts, the women included were all aged 45 and over, while the 1866–1870 birth cohort included only women aged 40 to 44 due to the fact that the information of those aged 45 and over is not available. Since women, especially women with a large family size, may still have children after age 44, excluding those aged 45 and over is likely to cause a reduction in parity progression ratios of high parities.

9 Thes data are derived from Chester Beatty Research Institute: see The Chester Beatty Research Institute serial abridged life tables, England and Wales, 1841–1960 (London, 1962).Google Scholar

10 Because there is no five-year age-group fertility data available for the 1851–1855 birth cohort, the statistics calculated for the 1846–1855 cohort are used as an approximation. This may slightly overestimate the fertility level of those who were born between 1851 and 1855, if their fertility was indeed lower than those born five years earlier.

11 When the net reproduction rate is less than one, it indicates that women of a certain generation could not produce enough women to replace themselves. If mortality is constant and the fertility level is maintained, sooner or later the population will decline.

12 For further discussion of changing marriage patterns in English history, see Schofield, R. S., ‘English marriage patterns revisited’, Journal of Family History 10 (1985), 2–19CrossRefGoogle ScholarPubMed; Weir, D., ‘Rather never than late: celibacy and age at marriage in English cohort fertility 1541–1871’, Journal of Family History 9 (1984), 340–54CrossRefGoogle ScholarPubMed; Smith, R. M., ‘Fertility, economy and household formation in England over three centuries’, Population and Development Review 7 (1981), 595–622.CrossRefGoogle Scholar

13 The statistics are derived from Mitchell, B. R., British historical statistics (Cambridge, 1988), 20–2.Google Scholar

14 This suggestion is in agreement with the relevant period data. The mean age at first marriage was 25.1 for women and 26.4 for men during the 1870s. In the 1930s, they were 25.5 and 27.3 respectively. The majority in the two birth cohorts were likely to have married during these two periods.

15 For further discussion of computer simulation and differences between macro-simulations and micro-simulations, see Wachter, K., ‘Microsimulation of household cycles’, in Bongaarts, J., Burch, J. and Wachter, K. eds., Family demography: methods and their implications (Oxford, 1987), 215–27Google Scholar; Ruggles, , ‘Family demography and family history’Google Scholar; Keilman, N. and Keyfitz, N., ‘Recurrent issues in dynamic household modelling’, in Keilman, N., Kuijsten, A. and Vossen, A. eds., Modelling household formation and dissolution (Oxford, 1988), 254–85.Google Scholar

16 For detailed discussion of the CAMSIM system, see Smith, J. E., ‘The computer simulation of kin sets and kin counts’, in Bongaarts, , Burch, and Wachter, eds., Family demography, 249–66Google Scholar; Smith, J. E. and Oeppen, J. E., ‘Estimating numbers of kin in historical England using demographic microsimulation’, in Reher, D. and Schofield, R. S. eds., Old and new methods in historical demography (Oxford, 1993), 280–317.Google Scholar

17 In comparison to SOCSIM (another computer simulation system which has been developed at the University of California at Berkeley), for example, CAMSIM is less effective in dealing with changing demographic conditions in a single simulation run. But my preliminary work indicates that short-term fluctuations and their impact on kin composition could be better simulated by running several closely related simulations, if more detailed input parameters are available. See Z. Zhao, ‘A report on the OPCS project related computer simulation studies’ (unpublished research report, 1993, on file in the library of the Cambridge Group for the History of Population and Social Structure).

18 In order to avoid repetition, only kinship networks of female egos are discussed in this paper. However, kinship networks of male egos have been simulated and the results show similar patterns.

19 For details of the input demographic parameters used by CAMSIM, see Smith, ‘The computer simulation of kin sets’.

20 The mean numbers of these kin produced by the second simulation are very low. This could be explained by the following circumstances. First of all, the fertility rate used in the simulation is one of the lowest recorded in English history. Secondly, the simulation results were generated under a stable demographic regime. This means that the same demographic parameters were applied to every individual, regardless of the time when each individual was born. Finally, in this exercise illegitimate births and adoptions, which might not be negligible in the past, have not been simulated. If changing demographic rates were used and illegitimate births and adoptions were simulated, the proportion of egos having kin of particular types and the simulated kin numbers would be slightly higher.