Two
Elderly Persons and Members of Their Households in England and Wales from Preindustrial Times to the Present

Richard Wall

There is, as Peter Laslett said, a deeply entrenched belief—or rather, misbelief—that in the past all older people in England (those over the age of 65) lived in families, either their own or those of their relatives, particularly those of their married children (Laslett 1989). A considerable amount of research has now been completed which shows that this is not the case, at least in preindustrial times, even though death rates were not so high as to prevent individuals from surviving long enough to see their children marry (Laslett 1977: 184; Laslett, Wachter, and Laslett 1978; Wall 1984, 1992). Disagreement continues, however, as to whether elderly parents were disinclined to co-reside with their married children because they wanted to remain independent as long as possible or whether co-residence was ruled out by the children because of the burden that would have been imposed on their own growing families immediately or in the future. Economic hardship (potential or current) is favored by Michael Anderson to account for the reluctance of the inhabitants of rural preindustrial England as well as of some nineteenth-century towns to welcome elderly persons into their households (Anderson 1972: 229). Laslett, however, appears to believe in the persistence over centuries in England of a family system that encouraged residential independence on the part of elderly parents (Laslett 1989: 119, 121).

This chapter presents a thorough reassessment of the living arrangements of the elderly in preindustrial England. In addition, the extent to which the family and household patterns of the elderly had changed by the end of the nineteenth century also receives consideration. Anderson has argued that the urban industrial revolution of the nineteenth century was associated with a considerable increase in the frequency of co-residence between parents and married children but with evidence of variation from

place to place depending on the availability of employment for married women outside of the home, which encouraged co-residence as an elderly parent could provide child care (Anderson 1972: 223, 230). Also relevant, again according to Anderson, was the prosperity of the local community, with co-residence of parents and their married children only practical when the poverty was not too biting (Anderson 1972: 230). The expectation is, therefore, that the residence patterns of the elderly will have varied from one community to another. However, little hard evidence has yet been assembled on this point, either for nineteenth-century or preindustrial England. I hope to rectify this omission by noting the degree of local variation in the residence patterns of the elderly and relating this variation to particular features in the economy and social structure of the areas concerned, despite the fact that direct evidence on the living standards of specific populations in the past is particularly difficult to assemble.

Another factor that demands attention is the impact of the introduction of old age pensions in 1908 on the family and household patterns of the elderly. Prior to the Office of Population Censuses and Surveys making available to the ESRC Cambridge Group anonymized census data on thirteen communities from the period 1891-1921, it had been impossible to investigate this issue as public access to the records of the census (enumerators' books and householders' schedules) only becomes possible one hundred years after their compilation. Two competing hypotheses, however, already await examination (see Anderson 1972: 230-231). The first is that the award of a pension will have improved the ability of the elderly to maintain their residential independence; the second is that a co-resident elderly parent with a pension was a more attractive proposition to a married child as a potential co-resident as the income from the pension could help alleviate any temporary "life cycle" poverty in the child's family.

Finally, an attempt is made to chart the changes that have occurred in the residence patterns of the elderly since 1921, drawing on a number of local surveys and for recent times the Longitudinal Study of the Office of Population Censuses and Surveys. Explanations for such changes are then sought from a range of demographic, economic, and cultural factors. Demographic factors are important because of the potential impact on residence patterns of changes in the proportion of the population marrying, in fertility, and in the spacing of births, as well as an improvement in life expectancy, while economic factors may alter the ability and cultural factors the desire to maintain an independent household in old age.

The "Sample" Populations

It is somewhat ironic that households in preindustrial England have received more attention than have households in other parts of Europe, given

that the census material that survives for preindustrial England is so much more fragmentary and poorer in quality than that which is available for other parts of Europe. For the period before 1800 in England and Wales, there are only about five hundred censuses, each listing the inhabitants of a particular parish or township at one point in time and no more than eight recording the ages of the inhabitants while also providing adequate detail on relationships of household members to the head of the household. It is impossible, therefore, to produce a random sample of preindustrial populations or to determine how representative the communities for which information does survive might be of English society in general. All that can be claimed is that surviving censuses come from many different parts of the country and from a variety of time periods.

For the purposes of analysis, the populations have been divided into two categories: a group of smaller communities and two larger communities, Lichfield and Stoke, both located in the county of Stafford and enumerated in 1692 and 1701, respectively. Ideally, each of the smaller communities should have been analyzed separately as well, given that the family and household patterns of the elderly may well have varied across both time and space just as did other features of the social structure (see Wall 1987, and for the dates of the censuses and the number of households in each community, see the note to table 2.1, below). In practice there are so few elderly recorded in some of these censuses that any such specific influences on residence patterns would be difficult to distinguish from the effects of random variation. Consequently, the only practical option is to analyze these communities as a group. On the plus side, it is possible to examine the nature of the economy of these smaller populations in some detail as in almost all cases the occupations of the male household heads were specified. This information is set out in table 2.1 and shows that despite the small size of these communities, agriculture was by no means the dominant employment. Farmers large and small (yeoman and husbandmen) represented less than a fifth of all heads of households and even together with the laborers, many of whom would not have been exclusively or perhaps even primarily involved in farm work, constituted slightly under half of all household heads. The remainder were employed in a wide range of manufacturing and service jobs. The majority of these were intended to meet the needs of the local community (as in the case of carpenters, blacksmiths, and wheelwrights, for example), but some were producing goods for more distant markets, as in the case of the nailers of Chilvers Coton. In view of the small size of these populations, however, it is appropriate to consider them as predominantly rural in character despite the presence of some protoindustrial employment.

The two larger populations fall into a quite different category. Lichfield was a city, and the cluster of settlements that would eventually become Stoke-on-Trent already had a reputation for the production of pottery. Both

the censuses unfortunately lack information on the occupations of the inhabitants, but the character of each community is relatively clear from the remarks of travelers and later historians, while references about the relative wealth of the two populations can be derived from an analysis of the hearth tax. Daniel Defoe and Celia Fiennes both commented on Lichfield. Defoe, whether from personal inspection or not, found it "a fine, neat, well-built, and indifferent large city . . . a place of good conversation and good company, above all the towns in this county or the next, I mean Warwickshire or

Derbyshire" (Defoe [1928] 1974: 80). It was the presence in the close of clergy with positions in the cathedral and other wealthy residents, not included in the census, who earned Lichfield this reputation, but there was no doubt some benefit to the rest of the city, which was by far the larger and more populous part. Fiennes was far less impressed, wondering why the bishop and other officials together with the gentry did not remove to Coventry given that Lichfield "stands so low and watrish" (Fiennes 1947: 114).

Accounts of Stoke concentrate on its role as a center of manufacture rather than on what it looked like. Nineteenth-century historians of the pottery industry tended to refer rather disparagingly to the quality and scale of seventeenth-century production. Simeon Shaw and John Ward, for example, imply that no more than between five and eight men may have been employed at a single works and that the distribution of the coarse ware that they produced may seldom have extended farther than the neighboring towns and villages (Shaw [1829] 1970: 65, 96; Ward 1843: 46). There can be no doubt, however, that there was a considerable amount of industrial activity in the area even in the late seventeenth century. Robert Plot's near-contemporary account refers to peacock coal being dug at Handley Green, which is located within the area covered by the census of Stoke, and of iron ore being worked at Longdon (also part of Stoke), while Burslem and Keele contained, respectively, the greatest pottery in the country and one of the only two centers in the country for the manufacture of frying pans (Plot 1686: 122, 126, 158, 335-336). There is ample justification, therefore, for insisting on the independent analysis of the family and household patterns of the elderly in Stoke even though some areas that were included in the census, such as Seabridge, were still predominantly agricultural even in the early nineteenth century (Shaw [1829] 1970: 60).

It is to be expected that the population of Stoke would be considerably poorer than that of Lichfield. Analysis of the hearth tax from the second half of the seventeenth century tends to confirm this. Just over a quarter of chargeable households in Lichfield were assessed on just one hearth compared to two-thirds in Stoke. Twenty-six percent of households were certified as not chargeable in Lichfield against 42 percent of the households in Stoke, although over a third of all the households in Lichfield were themselves deemed too poor to pay either the poor or church rate (William Salt Archaeological Society 1921: 153f.; 1936: 145f.). Unfortunately, how many of this last type of household there may have been in Stoke is unknown, and if they have been included in the hearth tax returns rather than simply omitted, then the differences between Stoke and Lichfield in terms of relative wealth would be considerably less than has been suggested although not totally eliminated. There is some justification, therefore, for considering Stoke the poorer of the two populations, and it will be necessary to take note of this when examining the residence patterns of its elderly inhabitants.

For the period 1891-1921, it has been possible to select the populations it is intended to study in detail rather than rely on the chance survival of a few local censuses, as was the case with the preindustrial period. The choice was dictated by a number of factors. Most of the populations were of interest because aspects of their social structure in other time periods had already received attention, for example, Bethnal Green in the 1950s from Michael Young and Peter Willmott (1957), Swansea in 1960 from Colin Rosser and Chris Harris (1965), Stoke in the mid-nineteenth century through the work of Marguerite Duprée (1989), and working-class York at its end from the celebrated study by B. Seebohm Rowntree (1901). At the same time, care was also taken to ensure that the thirteen populations, all that could be obtained given both the limited research budget and the time that would be required to clean and organize the data, represented a range of geographic and economic environments. This process eventually led to the selection of two "rural" populations, one in the northwest of the country (Morland) and one in the southeast (close to Saffron Walden). Another largely rural population was located in the southwest in the rural district of Axminster but including also the small seaside resort of Seaton. Two additional rural areas surrounded the market towns of Abergavenny in Monmouthshire and Banbury in northeast Oxfordshire, although Abergavenny also embraced the coal and iron working district of Blaenavon. Other areas were primarily or even exclusively industrial: Earsdon, Northumberland, had coal mining; Bolton, Lancashire, specialized in cotton; and Stoke, Staffordshire, as already mentioned, was famous for its pottery and by the late nineteenth century had also developed mining and metal interests. Swansea, too, had an important metal industry, although it was chosen principally for its role as a major port. York was selected as a representative of a large provincial town. Its adult male population was employed in a wide variety, of economic activities, with the food, transport, and construction sectors being the most important. Finally, three districts from within the London metropolitan area indicate the varied experiences of an inner-London population (Bethnal Green) and two suburban ones (Walthamstow and Pinner), the former very largely working class, the latter destined to be middle class but in 1891 only just beginning to experience the outward push of suburban London.

For the period after 1921, there are unfortunately only a few ad hoc surveys of a particular population at a particular date. These are also somewhat sparing in their details, indicating the numbers of elderly persons who lived alone, or alone but for the spouse, but rarely how many co-resided with their children (Wall 1992: 66-69). Such surveys suffice, however, to show the pace of change in both the household and the family patterns of the elderly before 1971 when the first of the random cross sections of the national population, taken every decade by the Office of Population Censuses and Surveys

in connection with the Longitudinal Study, can be exploited to yield a fuller breakdown of relationships within households. Even this classification, however, is less complete than those obtained from the more flexible data sets that have been produced for the historical populations. It also has to be borne in mind that the Longitudinal Study data sets are too small to permit the study of residence patterns of geographic areas smaller than that of the region, while most historical data are derived from an analysis of residence patterns in particular localities. At present there is no viable alternative strategy. No detailed data set on household patterns at the level of the individual locality is available for the present-day population, while the production of random samples of the national populations in the past are so time consuming and expensive that only one (for 1851) has so far been attempted, with that yielding little more than an overview of family and household patterns (see Anderson 1988, 1990).

Elderly Persons and Their Co-Residents in Preindustrial Times

There are any number of ways in which the household and family patterns of the elderly can be investigated. One of the more interesting approaches is to consider how many of the elderly were residentially isolated in that they lived either alone or with nonrelatives only. These two types of residence patterns are best considered in tandem as it is notoriously difficult to maintain a consistent definition of the household, and what one enumerator saw as a separate domestic group, another might subsume into a neighboring unit. The problem is serious enough in the nineteenth century when at least there was some attempt by a central authority, the Registrar General, to introduce and enforce a standard definition of the household (Wall 1982), but it is particularly acute with the earlier local censuses that were taken for a variety of purposes by different enumerators who, in compiling their lists of inhabitants, drew lines or left spaces between blocks of names but seldom recorded the procedures they had followed.

Another residence pattern that it is particularly important to distinguish involves those married couples who lived on their own, a very common pattern in contemporary Western societies and one that carries with it the threat of residential isolation should one of the partners die. Also of significance is the frequency of co-residence with a child, married or unmarried, with or without the presence of other people. There was, of course, no guarantee that child and parent would continue to co-reside until the latter's death, but at least there was a certain measure of security that care, economic support, and companionship would be provided if parents became dependent.

Data on these lines are set out in table 2.2 for preindustrial England, distinguishing the group of rural communities from the two urban areas of

Lichfield and Stoke. It is immediately apparent that five of the residence patterns that have been specified (living alone, with nonrelatives only, with a spouse alone, with a spouse and other persons, and with a child) capture the vast majority of the experience. For an elderly person to live with other relatives in the absence of a spouse or child was extremely uncommon in any of the communities. It is also apparent that relatively few elderly persons in preindustrial times lived totally alone: never more than 5 percent of elderly men and at most 16 percent of elderly women. As is argued above, however, those who lived only with nonrelatives should perhaps also be considered at risk of residential isolation. There were considerable numbers of such people, particularly women, in Lichfield and Stoke, reflecting perhaps the movement of women late in life into towns where there would be a broader range of cheaper accommodation. Taken together with the number of elderly who lived totally alone, it can be seen that about one in ten elderly men rising to a fifth in Stoke and around a third of elderly women rising to almost half in Lichfield were residentially isolated on the basis of the definitions set out above. These proportions are not substantially different from the proportions of persons who lived totally alone in England and Wales in 1971, as we shall see later. One important qualification has to be added. The censuses of both Lichfield and Stoke do not always specify the relationships of those persons who were not part of the nuclear family of the head of the

household, and the number of occasions on which elderly women, in particular, lived only with nonrelatives may, in consequence, be overstated. In contrast, relationships in the censuses of the rural communities are well specified, and there seems no reason to doubt that overall a third of elderly women had no relative of any sort present in their households.

However, in all the preindustrial communities, urban as well as rural, more common than either living alone or with nonrelatives was for an elderly person to co-reside with a child. Around half of elderly men shared a household with a child, married or unmarried, as did more than a third of elderly women. That fewer elderly women than elderly men lived with a child may seem surprising, as it might have been expected that when elderly women were widowed they would be taken into the household of a married child. In fact, as table 2.3 makes clear, higher proportions of elderly women than of elderly men were living with a married child, but there were many more elderly men with a co-resident never-married child. Finally, it should be noted that it was relatively rare in preindustrial times for an elderly couple to live on their own. In Stoke, in particular, there were very few such households, but even in the rural communities fewer than a fifth of elderly men and women lived alone with their spouses. This is in marked contrast to the situation of the elderly in Britain today, when more than six in every ten elderly men and a third of elderly women are living just with their spouses (Arber and Ginn 1992: 99).

A different perspective on the residence patterns of elderly persons is suggested by measuring the frequency with which they lived with a range of related and unrelated persons regardless of the number and type of other persons who might (or might not) be present. Table 2.3 contains some data

along these lines for preindustrial England. For elderly men, the most likely co-resident was a spouse, followed in the case of Stoke and Lichfield by unmarried children. Co-residence with a married child was considerably rarer. Even in the rural communities, only just over 10 percent of elderly men lived with a married child. By contrast, the presence of a nonrelative, whether servant, boarder, or lodger, was somewhat more likely, and in the rural communities where 44 percent of elderly men had at least one nonrelative in their household, they were a more frequent co-resident than unmarried children.

The position of elderly women was somewhat different. In the first place, far fewer than was the case with elderly men lived with a spouse. This is only to be expected as in preindustrial times, like today, women generally outlived their spouses and were in any case on average a few years younger than their spouses on marriage or remarriage. Even so, approximately four in ten elderly women in the rural preindustrial communities were still married. That in Lichfield this proportion fell to about a fifth indicates there may well have been considerable variation from place to place in the pattern of living arrangements, probably reflecting on this occasion not only the higher mortality of urban areas but, as was suggested above, the movement of widows into localities offering a range of cheaper accommodation. Second, elderly women were less likely than were elderly men to live with an unmarried child. In this case it could be argued that the living arrangements of elderly men and women differ, because in a significant number of cases a radical reconstruction of the parental household occurred only after the death of the father. Such a reconstruction might involve both the departure and/or marriage of any children still resident and the movement of the newly widowed mother into the household of a married child. Only repetitive high-quality censuses or population registers, neither of which are available for preindustrial England, could enable the process to be traced in detail. The third point to make is that in both Lichfield and Stoke, half of all elderly women lived with a nonrelative, whereas fewer than a third of elderly men shared a household with a nonrelative. This reverses the situation in the rural communities, where elderly men were more likely to co-reside with nonrelatives than were elderly women. As noted above, the numbers of women in Lichfield and Stoke who co-resided with nonrelatives may be somewhat inflated because not all relationships between household members were identified in the censuses, but it is unlikely that so many relationships were unstated as to reduce the proportions of elderly women living with nonrelatives to the same level as those for elderly men.

Elderly Persons and Their Co-Residents Between 1891 and 1921

The above account of the residence pattern of elderly people in preindustrial England can now be directly compared with the situation in the late

nineteenth century and early twentieth century by drawing on the anonymized data that the Office of Population Censuses and Surveys has recently made available to the Cambridge Group. Table 2.4 parallels table 2.2 by indicating how many elderly men or women were residentially isolated, or lived with their spouses, either alone or with other people, or co-resided with a child at each of the censuses of 1891, 1901, 1911, and 1921. Few differences from preindustrial times are evident. For men the most dramatic change is the rise from a very low level in the proportions who lived with other relatives in the absence of their spouse or child. There may also have been a slight increase in the proportion of elderly men living entirely on their own. In the case of elderly women, it is clear that far fewer in the late nineteenth and early twentieth century than in preindustrial times lived with nonrelatives in the absence of relatives, while more co-resided either with a child or with other relatives in the absence of a spouse or child. Yet even after a century of great social and economic change it would be difficult to substantiate the claim that the households of either elderly men or elderly women in 1921 differed in fundamental respects from those formed by elderly people in preindustrial times. The other point worth making is how little alteration there was between 1891 and 1921 in the household and family patterns of the elderly. For elderly men there is evidence of a slight rise in the percentage living on their own, and there was also a steady increase in the proportions of both elderly men and elderly women resident with a child. These increases were balanced by a decrease in the proportions of elderly people who were still married but who did not co-reside with a

child. In all other respects it is impossible to detect any consistent trends affecting the residence patterns of the elderly in this period.

An absence of change between 1891 and 1921 does not, of course, preclude significant developments between preindustrial times and the 1890s. Detailed data on family and household patterns in almost any community in the country can be abstracted from each of the decadal censuses beginning with 1851, but surprisingly few analyses have been completed given the wealth of information available, and little enough has emerged even from Anderson's 2 percent nationwide sample of enumerators' books from the censuses of 1851. Anderson, however, has now calculated the proportions of elderly men and women living alone, with nonrelatives only, with a child, or with other relatives (Anderson 1988: 436).

Of the elderly men in 1851, 5 percent lived totally alone, 16 percent with nonrelatives only, 5 percent with relatives other than a spouse and child, and 45 percent with a child. Fewer elderly men co-resided with a child in 1851 than in preindustrial times or between 1891 and 1921. However, the percentage who lived with nonrelatives in 1851 is considerably higher than it was to be in the late nineteenth or early twentieth century and higher also than what it had been in any of the preindustrial populations. The frequency with which elderly men lived with relatives other than a spouse or child is also in excess of that occurring within the preindustrial communities, although generally below the levels recorded in the later years of the nineteenth century and in the early part of the twentieth century. Finally, the proportion of elderly men living alone in 1851 appears to differ little from that between 1891 and 1921.

In a number of respects the situation with regard to the residence patterns of elderly women appears to be similar. For example, there is the same rise between 1851 and the early twentieth century in the proportions who shared a household with a child (46 percent in 1851 and between 47 percent and 52 percent during the period 1891-1921), and the proportion who lived with other relatives is, as is the case with elderly men, above that of preindustrial times but below that of the late nineteenth and early twentieth century. Elderly women, like elderly men, were also more likely in 1851 to be living only with nonrelatives than would elderly women in the late nineteenth and early twentieth century. The differences in their residence patterns as compared with those of elderly men lie in the fact that in 1851 they appear to have been less likely to have lived alone than was the case either in the late nineteenth century or in preindustrial times, while they were much less likely compared to women in the preindustrial urban populations to have lived with nonrelatives in the absence of any member of their own family.

The above account of family and household patterns of elderly people between 1891 and 1921 can also be supplemented by measuring the frequency with which the elderly co-resided with a spouse, with never-married

or ever-married children, or with nonrelatives whether or not other persons were present. Data on these lines are set out in table 2.5 and point to a marked fall, particularly between 1911 and 1921, in the frequency with which both elderly men and elderly women lived with nonrelatives. The same decade also witnessed a substantial rise in the proportions living with a married child, perhaps as a result of the housing crisis that followed the end of the First World War (for a further account, see Wall 1989). Between 1891 and 1911, a steadily increasing proportion of elderly persons lived with an unmarried child. No definite trend, however, is visible as regards the proportion of elderly living with a spouse.

These proportions can now be compared with those that pertained in preindustrial England (see table 2.3). Focusing first on the situation of elderly men, it is apparent that by the end of the nineteenth century, fewer elderly men were living with a spouse than had been the case in preindustrial times. In addition, between 1891 and 1921, many fewer elderly men lived with a never-married child than had done so in Lichfield and Stoke. However, between 1891 and 1921, many more elderly men lived with their married children than had been the case in any of the preindustrial populations. Trends in the frequency with which elderly men co-resided with nonrelatives are, however, more difficult to interpret, as in 1891 it was somewhat more likely that elderly men would share a household with a nonrelative than would elderly men from Lichfield and Stoke but somewhat less likely than would men from the rural preindustrial populations.

In addition, a number of the contrasts between the living arrangements of elderly men in preindustrial times and at the end of the nineteenth century do not apply in the case of elderly women. First, the proportions of el-

derly women still living with a spouse were some way below the proportion of elderly women in rural preindustrial England who co-resided with a spouse but were not as low as the proportions reported for the towns of Lichfield and Stoke. This contrasts with the situation of elderly men, who were less likely to be resident with a spouse than were men from any of the preindustrial populations. However, more elderly women were living with a never-married child in the nineteenth century than had lived with a never-married child in the rural communities in preindustrial times and by 1911 more even than had lived with a never-married child in Stoke in 1701. By contrast, the proportions of elderly men living with never-married children were much lower than they had been in preindustrial Stoke and Lichfield. The one major similarity in the trends over time in the residence patterns of elderly men and women was that co-residence with a married child had become much more probable by 1891.

A further perspective on the residence patterns of the elderly in the late nineteenth and early twentieth century requires that some account be taken of the degree of spatial variation. The data supplied by the Office of Population Censuses and Surveys are ideal for this purpose, because, as mentioned earlier, the thirteen populations were selected partly because of their geographic and economic diversity. To the extent, therefore, that household and family patterns are influenced by the nature of the local economy, the range of variation suggested is likely to embrace as much of the experience of the time as one could reasonably hope to cover, given that budgetary and time constraints precluded the taking of a proper national sample. Moreover, the focus on distinctive environments was enhanced due to the fact that from within each area a specific sample was drawn comprising a number of usually contiguous enumeration districts.

To illustrate the degree of variation from place to place in the residence patterns of the elderly, tables 2.6 and 2.7 set out the proportions of elderly in all thirteen communities in 1921 who were residentially isolated, or lived alone with their spouse or with their spouse and other persons but without a child, or lived with a child with or without other persons being present, or co-resided with more distant relatives. As measured by the coefficient of variation, there was least variation in the proportions of elderly men living with a child and most variation in the proportions living alone. Overall, about 6 percent of elderly men had no one else present in their household in 1921 (see table 2.4), whereas in Bethnal Green 14 percent lived alone and in Earsdon not one elderly man lived alone (table 2.6). By contrast, the percentage of elderly men co-residing with a child ranged from a high of 64 percent in Stoke to a low of 45 percent in Saffron Walden, Morland, and Bethnal Green. More variation is visible in the residence patterns of elderly women living with a child: the range extends from 63 percent in Stoke to 39 percent in Axminster.

It might be expected that many of these differences could be readily explained by relating them to specific features of the local economies, even if these are broadly defined as "agricultural," coal mining, and inner urban, but in practice such relationships are not easily detected. For example, although two of the industrial areas, Earsdon and Stoke, have a very low proportion of elderly men living on their own, other industrial areas such as Bolton and Swansea have only "average" proportions. Conversely, comparable (and low) proportions of elderly men living on their own occur in places that are known to be strikingly dissimilar (e.g., Stoke and Pinner). It may well be, of course, that a more detailed specification of the economic character of the areas in question, or possibly of subareas within them, would clarify the nature of the relationships between economic factors and household forms. This must await further analysis of the data. What can, however, be pointed out now is that it is in the rural communities around Saffron Walden that the family and household patterns of the elderly in 1921 took on their most "modern" look, with higher proportions of both elderly men and women living on their own, or with only their spouse, and lower proportions with nonrelatives only, or with a child, than was generally the case in 1921.

Residence Patterns of the Elderly After 1921

Keeping track of the residence patterns of the elderly after 1921 is not a particularly easy task because of the scarcity of detailed surveys and the degree of variability in the patterns recorded in those surveys that were taken. As early as 1929, the New Survey of London Life and Labour reported that 19 percent of elderly men and 37 percent of elderly women from the London working class were living alone, percentages that for women in particular were far in excess of those in Bethnal Green only eight years earlier (Gordon 1988: 26). Yet as the Second World War ended, in another working-class population, the mid-Rhondda, just 1 percent of men and 7 percent of women over the age of 65 lived alone (Nuffield Foundation 1947: 140-141). Such a low incidence of living alone is matched among the populations enumerated in 1921 only by Earsdon in the case of elderly men and by Pinner, Swansea, Stoke, and Earsdon again in the case of elderly women (see tables 2.6 and 2.7).

To measure the pace of change in the residence patterns of the elderly in the nation as a whole, it is necessary to continue to limit the comparison to the percentage of the population who lived alone. Comparisons can then be made with the results of a national survey taken in 1945 as well as with the far better-known investigation of Ethel Shanas, Peter Townsend, Dorothy Wedderburn, and others into the living arrangements of the elderly in three Western countries in the early 1960s (Shanas et al. 1968). In

table 2.8, the evidence of these two surveys as to the percentages of elderly men and women living alone in 1945 and 1962 is compared with the percentages in 1921 for the group of thirteen local populations (see table 2.4) and in 1971 and 1981 for England and Wales as a whole, derived from the samples taken for the purposes of the Longitudinal Study. Between 1921 and 1945, there was no change at all in the propensity of elderly men to live alone, while just 5 percent more women over the age of 65 lived alone in 1945 than had done so in 1921. Much more evidence of change is visible by 1962, although the degree of change in the residence patterns of both elderly men and elderly women that occurred between 1945 and 1962 was still somewhat less than that which was to take place between 1962 and 1981.

One additional perspective is possible, and that is a consideration of the residence patterns of the nonmarried elderly, as in table 2.9. The decision to study this particular group of elderly was largely dictated by the fact that it is the only group whose residence patterns can be examined through to 1981 using the population samples of the Longitudinal Study, as the way in which the data streams were defined in the study renders it impossible to determine whether married couples shared their household with other relatives or nonrelatives. Even so, to extend the perspective to 1981, a fairly basic classification of residence patterns is all that is possible, and table 2.9 is limited to showing how many elderly men and women lived with relatives, or only with nonrelatives, or alone. However, on the positive side, it should be emphasized that the nonmarried elderly are a very important group, not only because they constituted a large proportion of the total population over the age of 65 (e.g., more than four in ten of the men and two-thirds of the women aged 65+ in 1921) but because, lacking a spouse, they were particularly exposed to the risk of residential isolation.

In the late nineteenth and early twentieth century, as table 2.9 makes clear, approximately two-thirds of the elderly women and just under two-

thirds of the elderly men who were not currently married nevertheless lived with at least one person to whom they were related. By 1962, this proportion had fallen, but not dramatically so. The real change had occurred with a rise in the percentage of the nonmarried elderly, both men and women, who lived alone and a marked decline in the proportion living only with non-relatives. Much the greater part of the rise between 1921 and 1962 in the proportion of elderly men living alone can therefore be explained by the decline in the percentage who were in some senses already economically independent in that they budgeted separately from the nonrelatives with whom they lived but lacked the wherewithal to establish their own households. About a third of the change in the residence patterns of elderly women can also be explained in this way. After 1962, the situation appears to change considerably with a very spectacular fall in the percentage of both elderly men and elderly women living with relatives, particularly between 1962 and 1971, matched by a continuing expansion in the proportion living on their own. The proportions living only with nonrelatives, in contrast, appear to have drifted upward between 1962 and 1971, only to fall back again by 1981.

Cultural, Economic, and Demographic Determinants of the Living Arrangements of the Elderly

At this point it is appropriate to return to the issue of whether the primary factors influencing the temporal and spatial variations in the residence patterns of the elderly were cultural, economic, or demographic. Both demographic and economic factors have clearly had a large part to play in increasing in the longer term the numbers of elderly who live on their own or

as couples on their own, by decreasing the frequency with which the elderly live with either nonrelatives or their children. The role of economic factors is evident in the long-term decline in the frequency of sharing a household with nonrelatives. Some of these nonrelatives encountered in the households of the elderly in preindustrial society would have been servants, but many others would have been lodgers or boarders or even unrelated people put together into one household by Poor Law authorities anxious to economize by arranging for the younger poor to care for the elderly poor (Erith 1978; Robin 1990: 208). It is often impossible to establish just how independent some of the lodgers and other unrelated people may have been of the household to which they were attached, as living spaces were so circumscribed. Over time, as the rise in living standards has allowed the standard of accommodation to be improved beyond measure, it has become much easier to see which individuals are living independently in the sense that they occupy separate accommodation. However, the lodgers and boarders of earlier times will have budgeted separately from the main household and on the definition of "independent living" ought perhaps to be judged as forming their own "households." From this perspective, therefore, some of the increase in the numbers of elderly living on their own is more apparent than real, and the role of economic forces in helping to bring about the new household forms rather less significant than might appear at first sight.

It is also interesting that the introduction of old age pensions in 1908 seemed to have little visible impact on the living arrangements of the elderly. The award of a pension, even if the initial payments were neither universal nor particularly generous (Thane 1990: 34-35), should have enabled a higher proportion of the elderly to maintain their own household. As mentioned above, this was Michael Anderson's expectation, and B. Seebohm Rowntree hinted at the existence of such a tendency in York when he compared the households of the poor in 1936 with those of the poor in 1899 (Anderson 1972: 230-231; Rowntree 1941: 114). There was no sign of this, however, when the censuses between 1891 and 1921 for the thirteen communities were analyzed (see table 2.4). A slight rise is visible from 1911 in the percentage of elderly men living alone but not in the percentage of elderly women, nor in the percentage of either elderly men or elderly women who lived "independently" in that they resided alone or only with nonrelatives. Nor is there any evidence to support Anderson's alternative suggestion that the introduction of the old age pension may have made it more feasible for sons or daughters with families of their own to offer to shelter their elderly parents in case of need (Anderson 1972: 231, quoting the statement of one pensioner circa 1912). Table 2.4 shows that although the percentage of elderly men and women living with a child did increase between 1901 and

1911, this merely continued a trend of the preceding decade that had therefore begun well in advance of the introduction of the old age pension.

Other changes in the living arrangements of the elderly are demographic in origin. Improvements in life expectancy over the past two centuries have obviously increased the proportion of those who survive through to old age, yet in the Britain of today the proportion of elderly women who live with a spouse is lower than the proportion who lived with a spouse in preindustrial rural England. The reason for this is that there has been a greater improvement in the life expectancy of females than of males. This has prolonged the period women spend as widows while ensuring that many more men than in the past have a spouse to provide care and companionship in old age. Another demographic change of possibly even greater significance for the living arrangements of the elderly is, paradoxically, the fall in fertility beginning in the later nineteenth century and the altered pattern of birth spacing within marriage. Prior to the demographic transition, women continued to bear children into their late 30s or early 40s. When parents reached the age of 65, it was entirely feasible, therefore, for them still to have unmarried children in their household without these children necessarily having to postpone unduly the date of their own marriage, or not marrying at all, unless of course one or both parents survived into extreme old age and still insisted on keeping the parental household intact and unchanged. In a present-day population, it is much less likely that the elderly will have a never-married child on whom they can rely, while the earlier cessation of childbearing results in children leaving the parental household much earlier in the life of the parents. The increase in the proportions who eventually marry (or cohabit with a partner) will simply intensify these trends.

It took some time, of course, before the fertility fall of the late nineteenth century and the early part of the twentieth century came to affect directly the lives of the elderly. That is why there is so little sign of change in family and household patterns of the elderly by 1921 (see table 2.5). No firm evidence can be produced to demonstrate how soon after 1921 the decline in the percentage of elderly living with an unmarried child may have established itself, but it was by no means over when Ethel Shanas and colleagues undertook their research in 1962 (Shanas et al. 1968: 186). According to their study, 22 percent of elderly men and the same percentage of elderly women in Britain in 1962 co-resided with a never-married child, percentages that were indeed lower but not substantially lower than had been the case for the elderly in 1921 when 38 percent of men and 32 percent of women over the age of 65 lived with a never-married child (see table 2.5).

Economic and demographic factors also undoubtedly help to account for much of the variation in residence patterns that emerges when the experience of one local community is compared with that of another. The pre-

cise impact of the factors, however, is difficult to measure. In part this is because the fertility and mortality rates to which the families of the inhabitants had been exposed are not easily calculated, as a large proportion of the adult population will not have been born in the area in which they were resident at the time of the census. In addition, however, there is the problem of identifying the likely intercorrelations between a range of economic and demographic factors, on the one hand, and the various components of the household, on the other. One example would be when a buoyant local market pushed some sections of the native population toward both earlier marriage and an earlier exit from the parental home while at the same time the improvement in the standard of living and consequent fall in mortality make it less likely that the parental home would be broken by out-migration of the children following an early parental death. In fact, what is particularly surprising is that the residence patterns of the elderly from the various populations are not more different given the differences in their economies and in the level of epidemic and endemic mortality to which they were exposed. It was established above that the population of Stoke in 1701 was in all probability considerably poorer than that of Lichfield in 1692. Their economies certainly differed as both did from those of the rural populations that were also undoubtedly healthier. E. A. Wrigley and R. S. Schofield, for example, suggest that mortality in the most sparsely inhabited rural parishes such as Hartland in Devon might be two or three times lower than that in a substantial town (Wrigley and Schofield 1983: 178-179), although the difference in the level of mortality between Lichfield and Stoke and the pooled data on the rural populations included above may be somewhat less than this as the latter did not include any community quite as remote or with such a scattered settlement pattern as Hartland. Nevertheless, in terms of the household patterns of the elderly, Lichfield, Stoke, and the rural populations appear remarkably similar. Strikingly, the 54 percent of men over the age of 65 in Lichfield who lived with a child is the same as in Stoke and just 5 percent more than did so in the rural populations. For elderly women, the range is from 37 percent in the rural populations to 46 percent in Stoke.

Much the same point can be made in connection with the thirteen populations enumerated in 1921. There is undoubtedly some variation in residence patterns between one population and another, usually somewhat greater in the case of elderly women than elderly men, but the variation is not that large given that the thirteen include some of the least healthy areas of the country, such as Stoke, as well as some low mortality populations such as Morland and Axminster (Annual reports of the Registrar General). There is no evidence either that any of the different local economies, whether mining, industrial, suburban, or agricultural, produced a unique family pattern. This might seem to suggest that there might be different "cultural" preferences in particular areas favoring the formation of one type of house-

hold rather than another. However, the evidence that would prove that such preferences existed and were acted upon has yet to be produced, and it seems more likely that such limited variation as there is in residence patterns does reflect the different demographic and economic circumstances of the various populations. Moreover, it is worth emphasizing that even a shared characteristic, for example, a high proportion of elderly people living alone, can occur in different demographic and economic contexts, as in the case of Saffron Walden and Bethnal Green.

Over the longer term, cultural influences may have exerted a greater impact on the sort of households that were formed. Again, however, the evidence is lacking which would establish definitely the existence of norms prescribing the residence rules for various sections of the population. By contrast with the frequent references that can be found to various forms of behavior that were deemed to be morally offensive or contrary to the natural order, such as conceiving a child out of wedlock, mésalliances, and scolding wives, little appears to have been said about when, or even whether, children should leave the parental home, or about the rights of an elderly parent to live with a child. Such norms, if they existed, therefore, must have been internalized rather than embodied in a legal code, perceived as a matter of choice by the parties immediately concerned even though their neighbors would have reacted in the same way if faced with the same situation. Laslett (1984: 364) has labeled norms of this type "noumenal normative rules," holding them applicable in particular to the process by which in England children on their marriage regularly established a household independent from that of their parents. A wider applicability, governing, for example, whether the elderly should maintain an independent household even if widowed, is also possible. Yet merely to posit the existence of such norms in one sense solves very little as Laslett himself realized. The origin of the rule system and the reason why different rules are applied in some other parts of Europe remain to be explained and could conceivably be derived from the nature of the relationship between the population and its resource base at some distant point in the past.

Also requiring explanation is why in the recent past, particularly from the 1960s, there has been the explosion in the proportion of elderly living on their own. It was suggested above that demographic change, particularly the fall in fertility earlier in the century, probably accounts for a good deal of the decline in the proportion of elderly persons residing with their children. However, it should be noted that some other investigations of the trends in family and household patterns over recent decades in a number of Western countries have concluded that not all of the trend can be explained by improvements in the standard of living and modifications to the age structure of the population (for some of the arguments, see Michael, Fuchs, and Scott 1980; Pampel 1983; Schwarz 1983). This apparent inconsistency may arise

from the fact that too much of the research has focused on possible period effects, whether the primary concern is with cultural, demographic, or economic determinants of residence patterns, when cohort effects may be of equal or even greater significance. The data that would allow such hypotheses to be investigated do not exist for England, but it seems plausible that each generation that reached old age after 1945 was wealthier than its predecessor, rendering the members of these generations increasingly reluctant to disband their households to move in with their children or other relatives even in extreme old age. In countries, such as the Netherlands, with population registers that can be linked to income tax and real property records, such hypotheses could be subjected to a thorough testing (see Bulder 1993 on the Netherlands before 1940).

References

Anderson, Michael. 1972. "Household structure and the industrial revolution: Mid-nineteenth-century Preston in comparative perspective." In Household and family in past time , ed. Peter Laslett and Richard Wall, 215-235. Cambridge: Cambridge University Press.

———. 1988. "Households, families and individuals: Some preliminary results from the national sample from the 1851 census of Great Britain." Continuity and Change 3(3): 421-438.

———. 1990. "The social implications of demographic change." In The Cambridge social history of Britain 1750-1950 . Vol. 2. People and their environment , ed. F. M. L. Thompson, 1-70. Cambridge: Cambridge University Press.

Arber, Sara, and Jay Ginn. 1992. "In sickness and in health: Care-giving, gender and the independence of elderly people." In Families and households: Divisions and change , ed. Catherine Marsh and Sara Arber, 86-105. London: Macmillan.

Bulder, E. A. M. 1993. "The social economics of old age: Strategies to maintain income in later life in the Netherlands 1880-1940." Ph.D. dissertation, Tinbergen Institute, Erasmus University, Rotterdam.

Defoe, Daniel [1928] 1974. A tour through the whole of the island of Great Britain , ed. G. D. F. Cole and D. C. Browning. London: J. M. Dent and Sons.

Duprée, Marguerite. 1989. "The community perspective in family history: The potteries during the nineteenth century." In The first modern society: Essays in English history in honour of Laurence Stone , ed. A. L. Beier, David Cannadine, and James M. Rosenheim, 549-573. Cambridge: Cambridge University Press.

Erith, F. H. 1978. Ardleigh in 1796: Its farms, families and local governments . East Bergholt: Hugh Tempest Radfort.

Fiennes, Celia. 1947. The journal of Celia Fiennes , ed. Christopher Morris. London: Cresset Press.

Gordon, Chris. 1988. "The myth of family care: The elderly in the early 1930s." Discussion Paper 29, Centre for Economic and Related Disciplines, London School of Economics.

Laslett, Peter. 1977. Family life and illicit love in earlier generations . Cambridge: Cambridge University Press.

———. 1984. "The family as a knot of individual interests." In Households: Comparative and historical studies of the domestic group , ed. Robert McC. Netting, Richard R. Wilk, and J. Eric Arnould, 353-379. Berkeley, Los Angeles, and London: University of California Press.

——— 1989. A fresh map of life: The emergence of the Third Age . London: Weidenfeld and Nicholson.

Laslett, Peter, Kenneth W. Wachter, and Robert Laslett. 1978. "The English evidence on household structure compared with the outcome of microsimulation." In Statistical studies of historical social structure , ed. Kenneth W. Wachter, Eugene A. Hammel, and Peter Laslett, 65-87. New York: Academic Press.

Michael, Robert T., Victor R. Fuchs, and Sharon R. Scott. 1980. "Changes in the propensity to live alone 1950-1976." Demography 17(1): 39-56.

Nuffield Foundation. 1947. Old people: Report of a survey committee on the problems of ageing and the case of old people . London: Nuffield Foundation.

Pampel, Fred C. 1983. "Changes in propensity to live alone: Evidence from consecutive cross-national surveys 1960-1976." Demography 20(4): 433-447.

Plot, Robert. 1686. A natural history of Staffordshire . Oxford: Timothy Halton.

Registrar General of England and Wales. 1837-. Annual reports . London: HMSO.

Robin, Jean. 1990. "The relief of poverty in mid-nineteenth-century Colyton." Rural History 1 (2): 193-218.

Rosser, Colin, and Chris Harris. 1965. The family and social change: A study of family and kinship in a South Wales town . London: Routledge and Kegan Paul.

Rowntree, B. Seebohm. 1901. Poverty: A study of town life . 2d ed. London: Thomas Nelson.

———. 1941. Poverty and progress: A second social survey of York . London: Longmans, Green.

Schwarz, Karl. 1983. "Die Alleinlebenden." Zeitschrift für Bevolkungswissenschaft 9(2): 241-257.

Shanas, Ethel, Peter Townsend, Dorothy Wedderburn, Henning Friss, Paul Mihøj, and Jan Stehouwer, eds. 1968. Old people in three industrial societies . London: Rout-ledge and Kegan Paul.

Shaw, Simeon. [1829] 1970. A history of the Staffordshire potteries . Hanley: privately printed; reprint, Newton Abbot: David and Charles.

Thane, Pat. 1990. "Government and society in England and Wales, 1750-1914." In The Cambridge social history of Britain, 1750-1950 . Vol. 3. Social agencies and institutions , ed. F. M. L. Thompson, 1-61. Cambridge: Cambridge University Press.

Thomas, Geoffrey. 1947. "The employment of older persons." An enquiry carried out in mid-1945 for the Industrial Health Research Board for the Medical Research Council. The Social Survey , new series, 60(2).

Wall, Richard. 1982. "Regional and temporal variations in the structure of the British household since 1851." In Population and society in Britain 1850-1980 , ed. T. Barker and Michael Drake, 62-99. London: Batsford.

———. 1984. "Residential isolation of the elderly: A comparison over time." Ageing and Society 4(4): 483-503.

——— 1987. "Leaving home and the process of household formation in preindustrial England." Continuity and Change 2(1): 77-101.

——— 1989. "English and German families and the First World War." In The upheaval of war: Family, work and welfare in Europe 1914-1918 , ed. Richard Wall and Jay Winter, 43-106. Cambridge: Cambridge University Press.

——— 1992. "Relationships between the generations in British families past and present." In Families and households: Division and Change , ed. Catherine Marsh and Sara Arber, 63-85. London: Macmillan.

Ward, John. 1843. The borough of Stoke upon Trent . London: J. W. and S. Shaw.

William Salt Archaeological Society. 1921 and 1936. Collections for a history of Staffordshire . 1921, London: Harrison and Sons; 1936, Stafford: J and C Mort.

Wrigley, E. A., and R. S. Schofield. 1983. "English population history from family reconstitution: Summary results 1600-1799." Population Studies 37(2): 157-184.

Young, Michael, and Peter Willmott. 1957. Family and kinship in East London . London: Routledge and Kegan Paul.