Female householding in late eighteenth-century America and the problem of poverty.

Smith, Daniel Scott

Families in early modem England were more than twice as likely to be headed by woman as were white households in late eighteenth-century America. On average, in a preindustrial English community, women headed between one-sixth and one-fifth of all households. In the United States in 1790, by contrast, less than one-thirteenth had female household heads or householders.(1) The incidenc of female householding provides an important window on preindustrial American society. Its magnitude and ramifications deserve analysis and discussion for at least four reasons.

First, householder was a significant status, particularly because households were very important units of social and economic life in the preindustrial era. Householders served as the directors of the predominant institution of economic production. While not a primary category in thinking about the family, house-holding nevertheless was a consequence of other statuses that mattered more centrally, at least for men. Since newlywed couples established households becoming a head (and wives were considered co-heads, especially with respect to matters that were internal to the household) conveyed independence from the control of parents. Indeed, the contrast between autonomy and dependence was at the heart of legal definitions and other perceptions of social roles in early modern Anglo-America. The category of female household heads thus is crucial fo assessing one critical arena of social autonomy for women.(2)

Second, like their current counterparts, women who were particularly likely to head households in preindustrial England and America also were disproportionately prone to be impoverished. Consequently they were subjects fo support from and intervention by agents beyond the household: informal aid from kin, charity from private individuals or voluntary groups, and poor relief from public authorities. As the wages and assets of these female household heads wer lower than those enjoyed by male family heads, on average they were closer to poverty. Additionally, as women householders frequently had responsibility for the support of minor children, they shared their meager incomes with others, increasing the numbers facing poverty.

A heterogeneous group, women who headed households had some stability of residence and thus status in the community. Some were even "widows of means," who had the inherited resources and practical skills to support themselves and their families comfortably. But for many women familially unattached to the households of husbands or fathers, private troubles became, in many instances, public problems. The study of the incidence of female-headed households provide an indirect entry into the analysis of one large segment of the population of the poor of preindustrial society, a portion that was relatively larger in England than in America.

Third, historians need to provide a temporal perspective on the rapid growth in the number and percentage of female householders in both countries in recent decades, a phenomenon that is closely associated with the "feminization of poverty." The number of female householders, women enumerated first in both family and non-family living arrangements, increased by 98% in the United State between 1970 and 1990, and the share of such units expanded from 21% to 28% of all American households over that interval. In 1940, by contrast, only 15% of householders were women. During the last two decades, as Figure 1 shows, the absolute percentage increase was roughly that of the preceding three decades an also that of the entire century and a half between 1790 and 1940. A family form that was once unusual is no longer so, and historians cannot ignore major differences between past and present.(3)

Finally, as is the case for many areas of social history, the historiography of female-headed families critically needs comparative and analytical perspectives While demographic, urban, labor, welfare, and women's historians have contributed to the study of this subject, they have characteristically focussed on a single locality. At most, the geographic scope of these researches has extended to a handful of places within a region. Cities have been the focus of historians of poverty in preindustrial America, even though nearly 95% of the population lived outside of the handful of American urban places. While the local history approach has value, and is nearly irreplaceable for the reconstruction of the biographies of ordinary individuals, the typicality of findings unearthed by these local inquiries remains problematic.(4)

Even more important than the absence of a comprehensive geographic picture has been the neglect of a larger theoretical perspective that links familial and demographic outcomes to economic structure. Not only does social history need t establish variations, it also requires a framework that organizes a pattern among the differences.

I

An analytical extension of the demographic and economic framework used by late eighteenth century American and English writers provides the basis for understanding the Anglo-American difference in the incidence of female-headed households. Contemporaries in what may be called the Malthusian-frontier school were correctly impressed by the extremely high rate of population growth in eighteenth-century America, and they fitted this observation into a larger conceptual framework.(5)

The rate of population growth, they theorized, was positively related to the availability of land and consequently inversely related to population density. Compared to Europe, the components of more rapid American population growth included both differentials in natural increase, driven by the mechanisms of early marriage made possible by the availability of land that facilitated famil formation in the New World environment, and migration, as people moved from mor to less densely settled areas--both from Europe to America and toward the frontier within North America.

The framework of these commentators was highly parsimonious. Nothing else--whether technological change or productivity growth, the preventive check that limited the number of births by cheating nature within marriage, or the incidence of disease independent of economic conditions--determined the rate of population growth. Neither did more complex cultural, institutional, or historical conditions matter unless they could be tied to the sinews of their economic-demographic framework. Eighteenth-century writers on population were certainly not pure theorists, as they tended to focus on selected issues that were frequently related to immediate political problems.(6)

Following from the analysis of the sources of population growth, eighteenth-century writers also argued that economic and social organization were tied to density. Higher population densities lowered wages which in turn led to the development of manufacturing. Urbanization too was spurred by the demographic expansion that caused the countryside to become over-populated and pressured individuals to seek opportunities elsewhere. Poverty--in both the sense of the absence of access to the means of production by individuals and th sense of a supply of labor that exceeded demand in the aggregate--also was strongly related to density.(7) Consequently, poverty was less common in Americ than in England or western Europe, but this favorable situation was, many writers thought, necessarily transitory. Geographic expansion of the territory of the United States could, however, delay that development, which had ominous political ramifications for republican institutions, into the future.(8)

The optimism of political economists of the early Republican era concerning the short-run prospects for avoiding poverty persisted over the longer run in the United States. As one comparative historian has put it, the distinctive America assumption about poverty among the able bodied has been that there is none. American wages were too high and opportunities for employment too great for poverty to be a serious problem. Before American reformers could get a hearing for schemes to deal with poverty, they first had to demonstrate that it existed at all, a task that was unnecessary for their European counterparts.(9)

Commentators on poverty, both in the eighteenth century and later, neglected th large numbers of "life cycle" poor--orphans, widows, and the aged, who were grouped with the sick and physically disabled into the normative category of "worthy." Women at some risk to be heads of families were a only a potential segment of this non-controversial group of the poor. Despite their numbers, the were not seen within the socially constructed "problem" of poverty that was articulated by social and moral commentators. They were not blameworthy, as the proximate origin of the poverty of widows was, of course, the deaths of their husbands. Older widows were typically classified with the dependent rather than with the laboring poor: that is, as unable to work rather than unwilling to work, which meant, by an alternative interpretation, that they were unable to find work.(10) Deserted wives and especially mothers of illegitimate children were not, of course, viewed so sympathetically. By contrast, the able-bodied or labor-market poor faced two difficulties in garnering public support or charitable sympathy. As individuals, they mainly had only themselves to blame for their unfortunate situations. Collectively, it was feared, the numbers of labor-market poor would expand precipitately if relief were offered.

Women who headed households never slipped into membership of the "dangerous classes," whose criminal activity menaced the respectable. And as women, they were not generally considered to be potential political actors, either legally as voters or illegally as rioters or armed rebels. Central governments passed laws to compel units of local administration and close kin to do their duty in providing for these traditional needy.(11) Widowed women of all ages were particularly suited for small amounts of outdoor relief such as the provision o wood in winter. When local authorities institutionalized the poor, they attempted to compel the able-bodied of both sexes to work.

Even if legal sanctions and informal community pressure successfully privatized support for these poor, historians still need to incorporate life-course povert into their analyses of the general problem of the poor in preindustrial societies. For these traditional sources of poverty accounted for a substantial segment of the total population on public relief. These poor were the most morally entitled of all potentially dependent groups. Everywhere in early moder Europe the life-cycle poor were recognized as being worthy of support.(12)

Typically, the limited evidence suggests, about one-third of widows in England and America received such aid.(13) And they contributed a large proportion of the total welfare cases in both societies.(14) Quantitatively, then, the relative incidence of female-headed households can be regarded as an indirect measure of the economic burden of life cycle poverty on local government.

Later scholars also have neglected consideration of female-headed household and other groups in the ranks of the worthy poor. If politics as conventionally understood is the core of history, then these subjects are admittedly marginal. For example, an important, controversial interpretation of the growth and decline of the welfare rolls in American history excludes life cycle poverty an focuses solely on the relationship between work and welfare for the able-bodied poor.(15) Further, historians who were attracted by the relevance of the Malthusian-frontier perspective to American history have neglected the study of both poverty and women. Until very recently, emphasis on the low "man-land" ratio in early America, exemplified most famously by the frontier theory of Frederick Jackson Turner, coincided with an optimistic evaluation of the environmental conditions underlying that experience. The frontier, by this theory, meant that there were relatively fewer poor in America than in Europe; to dwell on the unfortunate circumstances of the small group of impoverished wa beside this central comparative point. However, the effects of the frontier on social experience do not all run in a single direction. Historians now emphasiz that the frontier was an environmental context that produced outcomes that were extreme.(16)

II

In this article I marshall the best available early American evidence on the incidence of female householding. Description and analysis focus on three large data sets: (1) for spatial variation, a sample of household heads from the 1790 U.S. census, the first enumeration of the entire population; (2) for variation by age, the extant document with the largest number of cases, the Maryland census of 1776; and (3) for an economic profile, the largest available computerized tax record, the Massachusetts valuation of 1771.

Results from a study of the 1790 printed census returns and an analysis of the variations in the incidence of female-headed families among the places in the sample provide a framework and context for local inquiries. The sex of some 27,854 white heads of families in this national sample of 79 places with surviving lists was inferred from the forename. When weighted to correspond to the number of white households in each sampled stratum, the percentage of white households headed by women fell between 6.88% and 7.64%. The higher estimate incorporates corrections for the possible misclassification of names of ambiguous gender and for the use of lists of taxables rather than of family heads for thirteen of the Southern counties in the national sample.(17)

White American female-headed households were distinctive in ways other than the gender of their head. As Table 1 shows, their size and composition reflect the likely effects of their later than average stage in the family cycle. Their households contained fewer males under 16 and a larger number of slaves compare to households headed by men. Older persons naturally were less likely to have minor children, and they had a longer period in which to accumulate slaves. Bot daughters and widows were less likely to receive inheritance bequests of land than of slaves, which were classified as personalty.(18) If a higher fraction o women family heads held slaves than did men, at the same time the shortage of family labor indicated by the smaller average numbers of whites in these households suggests a higher level of economic vulnerability. Solitary residenc was very uncommon in the United States before the second half of the twentieth century. But one-ninth of white women heading households in 1790 lived alone, compared to only three percent of households headed by men; over three-fifths o women family heads in that year had no white male over age 16 present in their households.

Table 1

Composition of white female- and male-headed households in the United States in
1790: Area with extant census listings(a)

Category Female-headed(b) Male-headed(c) Total

Total mean household size 5.09 (100) 6.64 (100) 6.54

Whites per household 3.86 (75%) 5.85 (88%) 5.70

Males aged 16 and older 0.56 (11) 1.54 (23) 1.47
Males under age 16 0.80 (16) 1.46 (22) 1.42
Total females 2.50 (49) 2.84 (43) 2.82

Nonwhites per household 1.23 (24%) 0.80 (12%) 0.83(a)

Slaves 1.17 (23) 0.77 (12) 0.80
Free blacks 0.059 (01) 0.032 ( 0) 0.034

With no other free person
present: 11.2% 3.2% 3.7%
With one or more non-whites
present: 19% 12% 12%
With no males under age 16
present: 53% -- --
With no males over age 16
present: 61% -- --

Notes:

a Included were the New England states, New York, Pennsylvania, South Carolina,
and all but three counties each in Maryland and North Carolina.

b Sample post-weighted to reflect varying distribution of female householders
relative to total white population. Data for female-headed households based on
actual count.

c Figures for households headed by males obtained by subtraction from totals
reported in W.S. Rossiter, A Century of population Growth: From the First Censu
of the United States to the Twelfth, 1790-1900 (Washington, D.C., 1909; 1966
reprint): Table 27, p. 98; Table 30, p. 100; Table 104, pp. 188-200. Table 105,
pp. 201-207; Table 114, pp. 276-291.

The purely demographic ramifications of the Malthusian-frontier theory can account for a substantial part of the smaller American incidence of female-headed households and the smaller share of its white population that was compared to England, potentially dependent. Although not all of the pieces of data are available, an accounting framework is useful for framing the problem and pointing toward the major elements of the solution. Definitionally, the female householder ratio is simply the number of female householders divided by the sum of the numbers of male and female householders. Since this ratio is a fraction, it is obvious that it may be made smaller by decreasing the number of female householders in the numerator, or by increasing the number of male householders, who appear only in the denominator. Marriage is the demographic event that does both. Almost by definition, spouse-present wives were not listable as household heads, and their measurable headship rate was zero, or nearly so. In the "western family pattern" or the "northwest European household formation system," nearly all households originated in a marriage.(19) Married men were listed as the heads of these households. Hence a high proportion of married people in the adult population, a result of early and nearly universal marriage, simultaneously reduces the numerator and increases the denominator.(20)

It is also unlikely that more than a small percentage of never-married women headed households. Widows were the only group among women who had much of a chance to be recorded as heads of households. Consequently, the share of widows in the adult female population or the number per household are very pertinent figures for understanding the overall female household headship rate and the share of female householders among all householders in a population.(21)

Compared to England, a considerably higher fraction of adult Americans had been married. Widows were a smaller percentage of the ever-married, and there were substantially fewer widows per household in America than in England. New Hampshire was the only early American locale to report its population by detailed marital status. Averaging the results from the 1767 and 1773 censuses, some 27%, 64% and 9% of females over age 15 were single, married and widowed respectively. In the English parishes with extant listings, the comparable figures are 37%, 49%, and 13%. In New Hampshire, however, female householding was low, even by American standards. Only four percent of households in the Granite State in 1790 were headed by females and only three percent of females over age 16 headed households, figures substantially lower than the national averages for 1790. Elsewhere, I have estimated that only 27% of colonial New Hampshire widows were householders, a figure radically below that of 76% for th three English parishes with available data.(22)

The only large-size extant nominal listing of an American population by age before the seventh U.S. census of 1850 covers parts of Maryland in 1776. In three Maryland counties in that year, 6.8% of females over 15 headed households a higher rate than that estimated for the white population nationally in 1790 (5.3%). In 1790, some 9.7% of white households in Maryland were female-headed, and 7.1 % of females over age 15 were household heads.

Table 2 compares age-specific female householder rates for parts of three Maryland counties in that year with those for nine preindustrial English parishes, seven rural and two urban, and national figures for the United States in 1900. Although female household heads were more common in Maryland than in America generally, the headship rate of 11.1% in England was more than 60% higher than that of Maryland in 1776.

In both countries female householder rates increased sharply with age, and the climb was especially steep in England. Women under age 35 in Maryland in 1776 were actually more likely to be householders than their counterparts in preindustrial England (3% versus 2%). Middle-aged women, those aged 35 to 54, were somewhat less likely to head households in Maryland (12%) than in England (15%). Older women in Maryland, those aged 55 and older, however were substantially less likely to head households (19%) compared to older English women (32%).

TABULAR DATA OMITTED

Reflecting a lower level of fertility in the past, English women were older tha women in Maryland. Some 16% of those over age 15 in these English parishes were aged 55 and older compared to 8% of those in Maryland; likewise, fewer adult women in England (53%) were under 35 than in Maryland (66%). Standardization reveals that nearly half (45%) of the difference in the overall headship rates between Maryland and England may be attributed to the older age distribution in the latter population, while the other half is due to the higher age-specific female householder rates in England.(23)

For understanding the burden of poverty, the fraction of female-headed households is a proxy for the percentage of the potentially-dependent among all households. But demographic factors are also important in dividing the non-dependent population into those who might (or might be forced) to contribut to the dependent because of kin relationships, and those unrelated households and individuals that could only be taxed to support the poor.

The higher rates of natural population increase in America compared to England meant that old people had more children and other kin available. In Maryland in 1776, for example, about 60% of those over age 65 lived with a child. In preindustrial England, by contrast, this figure was just under 50%, a small gap that likely can be accounted for the differential in the availability of children.(24) The higher householding rate of older white women in England than Maryland is probably attributable to a lesser incidence of widows living in the households of their children than was case in America.

Furthermore, there were more kin available in the neighborhood and community in America than in England. Nearly three in ten white American households in 1790 were located within five households of another that was headed by a person with the same surname, compared to only one in twenty in three English parishes. In these English locales, some 73% of householders had no other same-surname head within a cluster of 50 households; in America in 1790, only 58% were missing such a same-surname household head within a group of 50.(25) In America the greater availability of kin living in close proximity reduced the demands place on the public welfare system.

III

Early and nearly universal marriage was the proximate demographic force driving the Malthusian-frontier system. Compared to England, it produced fewer widows relative to the numbers of married householders who could provide assistance, acting in the roles of kin, neighbors, or rate payers. The factor of density supplies the substantive mechanism in the Malthusian-frontier theory.

Table 3 confirms that areas of higher density had a higher female householder rate and ratio. The adjusted rate increased from 4.8% for places under 10 persons per square mile to 6.0% for places with a density greater than 50; the respective ratios are 6.1% and 9.0%. The female headship rate and ratio were considerably higher in areas with lower sex ratios, child-woman ratios, numbers of children per household, and average household sizes. All of these indicators capture the stage of demographic development of communities.

Table 3

Description of patterns in female householder ratio, and male and female
headship rates, U.S.A. 1790

 Female Householder Male Householder
 Ratio Rate Adjusted
Variable and Categories Raw Adjusted Adjusted Rate

Grand Mean 6.88% 7.64% 5.49% 63.32%

Main Regions

South 7.5 9.2 7.0 62.7
Middle states 5.2 5.4 4.0 63.1
New England 7.5 7.7 5.0 64.1

[Eta.sup.2] .04 .09 .12 .01
F-value 1.7 3.6 5.4 0.4
Significance level .19 .03 .01 .69

Detailed Region

Rural:

Carolinas 8.3% 8.8% 6.8% 66.8
Chesapeake 6.6 9.2 6.9 60.3
Middle states 4.5 4.6 3.7 63.8
Southern
New England 8.6 8.8 5.6 64.2
Northern
New England 3.4 3.5 2.5 64.1

Largest cities 18.4 18.9 11.7 56.4

[Eta.sup.2] .38 .39 .33 .16
F-value 8.7 9.4 7.3 2.8
Significance level .001 .001 .001 .02

White Sex Ratio,
Ages 16+

Under 100 10.7 11.3 7.4 64.1
100 to 104.9 5.2 6.2 4.7 64.1
105 to 109.9 5.2 5.9 4.6 62.4
110.0 and above 4.7 5.5 4.6 62.2

[Eta.sup.2] .39 .24 .14 .02
F-value 10.2 8.0 4.0 0.5
Significance level .001 .001 .01 .68

Density Per
Square Mile

Under 10 5.7 6.1 4.8 65.5
10 to 30 5.4 6.6 5.1 63.2
30 to 50 6.9 7.6 5.4 61.8
Over 50,
except urban 8.8 9.0 5.9 64.6
Cities 18.4 18.8 11.7 56.4

[Eta.sup.2] .29 .23 .14 .10
F-value 7.4 5.6 3.0 2.1
Significance level .001 .001 .02 .09

Notes:

For a description of the sample, see Smith, "'All in Some Degree Related'"
American Historical Review 94 (1989): 74-79. The sample was retrospectively
re-weighted so that the number of cases reflect the number of white households
in each area, rather than the white population.

The adjusted female householding ratio is an estimate of maximum fraction of al
white households headed by women; it compensates for omitting names of ambiguou
gender from recording as women and for the use of tax lists in Virginia and one
North Carolina county rather than the 1790 census listings.

The adjusted female householding rate is a maximum estimate of the fraction of
women over age 16 who were householders. The adjusted male householding ratio i
a minimum estimate of the fraction of men over age 16 who were householders.

The categorization of places by density statistically accounts for a modest percentage of the variation in the rates and ratios. Only New England seems to have been an exception to this weak pattern,(26) and the correlation of a dichotomous indicator of densities above and below 40 per square mile with the rural female headship ratio (.66) and rate (.63) was substantial only in that region. As Franklin optimistically asserted and later American commentators hoped, the dire, characteristically "Malthusian" implications of higher densities were not applicable to American conditions: there was simply too much land for expansion in the frontier economy of early America. The distribution o densities among places in rural America in 1790 was missing cases with a high enough level of density that signified a complete occupation of economic resources.

In multiple regression and multiple classification analyses of the female householder ratio and rate in 1790, variation in density among rural areas did not prove to be a statistically significant predictor. However, the interaction of density with location in New England remained statistically significant with the inclusion of a regional variables and two other demographic indicators of the stage of development (the adult sex ratio, and the mean number of children per household, an indicator that was negatively related to density) in the analysis.(27) Although a useful first approximation, density as the operative mechanism in the Malthusian-frontier theory is an imperfect indicator and not the only discernable source of variation.

Regional location and especially urban-rural residence more forcefully shaped the incidence of female householding. Cities exhibited an especially large ratio, for they were also characterized by a low male householder rate. In the seven large American cities, females headed some 18.8% of all white households, a figure more than double that for rural areas. Cities in New England, with Boston leading with 30.4%, had the highest incidence of households headed by females, while Baltimore was lowest at 13.6%. While higher urban mortality is doubtless one source of the greater incidence of female householding in cities, the more influential cause was the migration of unmarried householders and the poor toward the greater economic opportunities in cities. In 1708, for example, the mayor of Philadelphia complained that his city of brotherly love took care not only of its own poor "but almost all the poor of the province, most of them when distressed in the Countrey, repairing to the town for relief."(28)

The female householding rate and ratio were highest in the South and lowest in New York and Pennsylvania (see the bivariate results under the "none" column of Table 4). Controlling for the effects of demographic variables, however, the middle states are very close to the national average. New England had the lowes female householding rate and ratio, while the South remained the region with th highest incidence. Assuming that the adjusted results sufficiently control for demographic differences and therefore reflect variations in propensity, there i no obvious reason why the level of New England female householding should be lower than elsewhere in America.(29)

As noted previously in the discussion of Maryland patterns in 1776, slavery is the likely reason that white southern unmarried women were more able to be householders. Domestic slaves could provide the services needed for the maintenance of independent households, and field slaves, some of whom presumabl were rented out to other masters, the income.

Table 4

Multiple classification analysis of adjusted female householder ratio for place
in the United States in 1790

 Deviations from grand mean:
 Weighted Adjusted for:
Factor & Categories (N) None Factors only All variable

Density per square mile
 Under 20 (31) -2.1 -2.0 -1.1
 20 to 40 (22) -0.6 0.3 -0.1
 Over 40 (23) 1.9 0.7 0.4
 Cities (3) 11.2 13.0 9.0

Eta and beta .55 .57 .38(***

Regions
 South (30) 1.5 3.1 2.7
 Middle states (22) -2.2 -1.9 -1.2
 New England (28) 0.1 -1.9 -2.0

Eta and beta .29 .47 .42(***

Interaction Term
 Rural New England/
 over 40 p.s.m. (14) 2.9 4.0 1.5(**)
 Other (65) -0.6 -0.9 -0.3

Eta and beta .26 .36 .13(**)

Regression coefficients:
 Sex ratio over age 16 -10.5(**)
 Children per household -4.1(**)

Grand Mean=7.64% Multiple R-Square=0.63

* Sig. at .05; ** Sig. at .01; *** Sig. at .001

IV

An advantage of an accounting framework for the study of householding is that the scheme must "account" for everyone, whether or not they would qualify by current or past definitions. This framework focuses directly on different elements in the entire population and what role each played in family organization. Two groups--slave women around whom dwelling units within the enslaved population were organized and white women who were perceived to function in some respects as co-heads of their households--need to be evaluated Neither group, albeit for quite different reasons, would be listed in an enumeration as household heads. Any framework that omits the some 18% of Americans who were enslaved in 1790 is obviously incomplete. Further, slavery intertwined with welfare provisions for whites in ways beyond facilitating independent householding by white widows. In colonial Charleston, South Carolina, for example, the workhouse was filled with runaway slaves and outdoor relief was provided for white paupers.(30) In addition to slaves, wives of whit men are an ambiguous group in relationship to the status of headship. When thes groups are considered, the Malthusian-frontier model can explain why the state or collectivity in America had an even lower burden of poverty to deal with tha has already been argued in this article. This is true even though the overall fraction of female-headed households in America, as measured by the expanded perspective provided by the accounting construct, may have been quite similar t that in England.

First, a cultural norm in both societies regarded wives as deputy husbands or co-heads of their households.(31) Certainly a wife was more of a householder than was a daughter or a servant of the head. An upper limit of one-third can b placed on the extent of co-headship of wives.(32) Some 63% of adult women in Ne Hampshire around 1770 were married compared to only 49% in late eighteenth-century England.(33) Thus, compared to England, the early American female headship ratio needs to be raised by nearly five points to account for the phenomenon of co-headship by wives.

Estimates of the incidence of female householding among the African-American population must be approximate. While of interest, data for free blacks in 1790 do not provide a good guide for an estimate for the entire population. Only a little more than a tenth of blacks in areas with extant census listings were free in that year. The female householder ratio for blacks in that year was about twice as high as it was for whites in the same distribution of places where free blacks lived.

About one-fifth of free black householders were women, and about half of free blacks lived in households with no white members. Interestingly, in light of patterns since the Civil War, there was, except in the South, no marked urban-rural difference in the incidence of female-headed families in the black population.(34)

If one-half of the households of the enslaved were organized around an adult woman, then both America and England in the late eighteenth century would have nearly identical proportions of female-headed households. Yet this guess TABULA DATA OMITTED about the gender of householders among slaves is almost surely too high. Almost exactly a quarter of nearly 4,000 units in rural Louisiana between 1810 and 1860 were organized around an enslaved woman, although over half of these female-headed households contained only a solitary woman.(35)

It must be remembered that the accounting framework does not match the structur most suited to the understanding of poverty. Taking into account the families o slaves and co-headship by white wives, the former gets at the fraction of adult women who performed the householding role. Neither of these categories fails into the population that would be aided by the collectivity, either formally or informally. As co-heads, wives were in the segment of the population that serve normally as providers of welfare to the poor, either as rate-paying households or informally to neighbors. In principle, masters took care of the welfare need of slaves. In practice, slave owners sometimes shirked their duty. In 1752, the Maryland legislature complained in an act that "sundry Persons in this Province have set disabled and superannuated Slaves free who have either perished throug want or otherwise have become a Burthen to others."(36) Although the accounting framework yields a perspective that should not be overlooked, only white female householders are relevant to the poverty problem as defined by early American institutions.

V

Not all white female householders were poor. But more needs to be known about the economic circumstances that allowed or forced women to become householders. In particular, data on wealth or income are needed to evaluate the relationship of female householders to the threat of poverty. Unfortunately, the available sources are not readily helpful because they incompletely cover the relevant populations. Not all currently unmarried women appear in probate records after death or on tax lists while alive; for that matter, neither do all men, althoug the coverage is substantially higher for males. For comparison we need to know the relationship of those for whom information on wealth is available to the larger population of each group.

Certainly direct comparison of those appearing in a record can be misleading, a results from the largest extant computerized tax list, that for Massachusetts i 1771, reveal.(37) For this investigation, the 1,562 women (4.01% of the total) who could be identified were compared with a random sample of men of the same size.(38) As was the case of female householding nationally in 1790, female taxpayers in Massachusetts in 1771 were more frequently found in cities: 8.5% o the total in Boston, 5.7% in Salem, and 17.2% in Charlestown.

Comparing the wealth of men and women on the list yields a temporary surprise. Women appear to be richer than men, as the former possessed an average of [pounds]72 compared to [pounds]49 for the latter. Women had even more of an advantage in categories other than real property, owning [pounds]42 compared to the [pounds]20 of men.(39) Real property was more evenly distributed, with wome holding [pounds]29 and men [pounds]31.

In 1790, however, an adjusted estimate of 8.56% of household heads enumerated i Massachusetts (including Maine) were women, more than twice the fraction of females on the 1771 tax lists. One reason for the deficiency in the numbers of women on the 1771 document is the inclusion of a poll tax, which allows more me without other property to appear on the tax valuations. By Pruitt's definition of wealth, some 19% of men named on the list had none, compared to only 6% of the women. Some 84% of women taxpayers had no rateable polls compared to only 14% of the men on the 1771 document.

Two simple, plausible assumptions produce an estimate that there were some 55.4 of female household heads had no property of any sort that was valuated in 1771.(40) Women householders were, after this adjustment, poorer than men; thei total wealth was only two-thirds of that of their male counterparts.(41) The implied estimate of the share of women's wealth in all wealth is below that found in early American probate inventories.(42)

The comparison of the unadjusted figures underscores the undeniable economic heterogeneity among female householders. Indeed, the variation in wealth among female taxpayers was greater than among male taxpayers. As the coefficients of variation (the ratio of standard deviation to the mean) show, this was primaril because women disproportionately held non-real wealth that was more dispersed i its distribution than real property.

Another distinctive feature of the assets of Massachusetts female taxpayers in 1771, like the households of female heads nationally in 1790, was that slaves were more likely to be included than was the case for men. It is possible that more of the slaves of female householders were female household servants, provided by their husbands as part of provision for old age. However, neither the 1771 tax lists nor the 1790 census indicates the sex of slaves.(43)

Table 6

Comparison of wealth profiles of male and female taxpayers on 1771 Massachusett
provincial evaluation

 "Taxed" "Heads"
 Observed Corrected
Category Males Females Females

Total Wealth 48.9 (1522) 73.7 (1522) 32.9
 Coeff. of variation 2.89 4.48 --
 Real Wealth 29.3 31.2 13.9
 Coeff. of var. 1.53 1.55 --
Other Wealth 19.6 42.4 18.9
 Coeff. of var. 6.25 7.30 --

Pct no wealth 19.0 5.9
Pct no rateable polls 13.7 84.4
Pct with slave 1.6 4.3 1.9
Pct Lenders 5.9 13.9 6.2
Pct Merchants 4.7 3.9 1.7
Pct Manufacturers 4.3 2.6 1.1
Pct with cow 58.5 53.3
 Urban 6.3 (143) 5.1 (356)
 Rural 63.9 (1379) 68.1 (1166)

Notes: Wealth is in pounds, and is the sum of [(horses + oxen) * 2], (cows *
1.5), [(goats + sheep) * 0.015], (swine * 0.4), (annual value of real wealth *
6), and the values of stock-in-trade, factorage, and money at interest. Lenders
are defined by Pruitt as those having any money at interest. Merchants are thos
with any stock-in-trade. Manufacturers are those owning any part of a works or
still. See Pruitt, "Agriculture and Society in the Towns of Massachusetts, 1771
A Statistical Analysis," (Unpublished Ph.D. thesis, Boston University, 1981),
185. Only 3 places--Boston, Salem, and Charlestown--were classed as urban.

Additionally, there is evidence, at least in relative terms, for what one historian of widowhood in England has pointed out as "the most prominent economic function of the widow in English rural society between 1500 and 1900"--money lending.(44) Some 14% of women on the list (6.2% after the adjustment) were money lenders, having money-at-interest reported, compared to 5.9% of men. However, nearly 90% of the outstanding credit provided by Massachusetts taxpayers in 1771 was supplied by men, although women participate disproportionately in this activity. As for the other, more mundane stereotype, rural female taxpayers were slightly more likely (68%) to own a cow than were male taxpayers (64%).

Despite the diversity in wealth, most American female householders were poor. I is unknown whether the fraction was higher or lower than in England. But even i the poverty rates of female householders were identical, this aspect of the social burden of life-cycle poverty was reduced in America for the demographic reasons discussed previously. The female headship ratio was less because of the higher householding rates for men, because of the younger age structure of the population, and because of the lower proportion of the widowed in the adult female population. The American inputs into the Malthusian-frontier model reduced poverty for demographic as well as economic reasons.

VI

SUMMARY AND SPECULATIONS ABOUT IMPLICATIONS

The Malthusian-frontier model can subsume under its themes all of the sources o the collective problem of poverty:

(1) the familiar "frontier model" that emphasizes the cheap land and high wages of America, and a low incidence of labor-market poverty--in short, America as "the best poor man's country."

(2) the comparative advantage of agriculture over commerce and especially manufacturing led to a very low incidence of urbanization. Both life-cycle and labor-market poor were more prevalent in cities.

(3) the purely demographic effects of early marriage and high male house-holdin rates, young age structure and low incidence of widowhood that have been documented in this paper. At any point in time, a smaller proportion of white adults in America faced the problem of life-cycle poverty.

(4) finally slavery, an institution also built on the economic facts of cheap land and high wages for free labor, also reduced the collective burden of poverty. Although it radically altered patterns of household organization, slavery simultaneously privatized the resulting poverty problem.(45)

The lesser incidence of poverty, both labor-market and life-cycle, had larger and longer lasting implications. Many eighteenth- and even nineteenth-century American communities were too small in population to justify the construction o almshouse or workhouse for their poor. Such facilities for indoor relief, which advocates thought compensated for their expense by their deterrent value, appeared first in cities, then in smaller cities and towns, and finally in consolidated poor farms in rural counties.

If a harsh method for dealing with the poor was delayed by their scarcity in th populations of American communities, so too was the English experiment with a more generous and extensive one, that of family allowances. Inasmuch as early modern societies first learned to deal with poverty primarily as life-cycle phenomena, societies with a larger share in the total population of those who were unable to work developed means of support that eventually could be extende to others in unfortunate circumstances for other reasons. In other words, by this admittedly speculative hypothesis, the substantial magnitude of the welfar burden of the worthy poor paved the way for entitlement of those unworthy by traditional definition.(46) However, with the important exception of veterans and their widows, first of the War for Independence and then of the Civil War, such a transformation from poor relief to more extensive entitlements did not happen in the United States until well into the twentieth century.(47)

As in England, nineteenth-century American public discourse about poverty continued to ignore the "impotent" or life-cycle poor and focused on the able-bodied poor. Despite differences in economic situation and in the analysis of the causes of poverty, American thinking about the effects of public policy on the problem followed that in England to a remarkable degree.(48) After the repeal of the Old Poor Law in England in 1834, reformers in both societies, armed with the doctrines of liberal political economy, stressed deterrence and fought against outdoor relief, even for the worthy poor.(49)

Experience, however, did not closely follow ideology. Outdoor relief remained important, especially for the life-cycle poor. In Philadelphia in 1830, for example, 498 out of 549 paupers on outdoor relief were women who were either aged or had young children; of the total, 406 were widows.(50) Indeed, even in the century of the poorhouse, more people were given outdoor relief than entere that detested institution. Outdoor relief tended to be given in small amounts, often only in kind rather than cash, for short periods and for what seemed to b special cases of need for those who were deemed to be worthy.(51)

Nineteenth-century demographic developments helped to perpetuate the distinctio between the populations of life-cycle and labor-market poor. The former tended to be old, rural, and native-born. The latter, on the other hand, were younger, urban, and increasingly over time, foreign-born.(52) This ethnic difference was probably most important in its implications for policy, as the American reform impulse, both in antebellum and Progressive eras, was distinctively located among Protestant, middle-class groups, and especially women in this milieu. Fro their perspective, the life-cycle poverty of widows could easily be identified as something that happened to "us" rather than to "them"--the urban, immigrant, Catholic working classes.(53)

Although marriage ages increased over time and a larger fraction remained unmarried relative to the incidence of celibacy in northwestern Europe, America retained its distinctive pattern of high nuptiality. The number of widows relative to all householders made the financial burden of assistance to them relatively small. Compared to England and western Europe, private charity playe a larger role in American assistance to the poor. The impotent poor who were in need for reasons beyond their control (e.g., widowhood) were not socially constructed into the "poverty problem" that was ideologically problematic.

Recent scholars of the twentieth-century welfare state have argued that America developments have been maternalist in orientation, favoring, for example, mothers' pensions over unemployment insurance. Focussed on understanding the present, social welfare historians have emphasized mothers' pensions, enacted b twenty states between 1911 and 1913 and by forty by 1920, as a turning point in the creation of the modern welfare state. State rather than local in creation and based on a status rather than a situation, mothers' pensions removed the stigma of poor relief. This program can be seen as a precursor of the kind of entitlements that became common in welfare policy during the New Deal and thereafter.(54) Similarly, feminist scholars have emphasized the inadequacy of the program by the less moralistic and more humane standards of current welfare policies.(55) In her comprehensive study of the American welfare state in comparative perspective, Theda Skocpol also has emphasized the novelty and distinctively American attributes of the mothers' pensions program. She particularly stresses the crucial role of women's organizations and publication for women in the sudden emergence of the issue and its equally swift passage by state legislatures in the 1911-13 period.(56)

To be sure, novelties and antecedents of later developments may be discerned in the maternalist orientation of early twentieth-century American welfare. This explanation is, however, incomplete. It focuses too much on the sources of reform and not enough on the size and character of those who composed the problem in need of amelioration. It also neglects the orientation of the male legislators who voted for mothers' pensions by overwhelming majorities. The distinctively maternalist orientation in the belated building of the modern American welfare state may also be seen as an ongoing legacy of the peculiar structuring of poverty in the preindustrial, Malthusian-frontier era that persisted into the nineteenth century.

First, mothers' pensions were really widows' pensions, as only three states as late as 1931 allowed never-married women to be aided. The test for eligibility for these pensions was not simply the status of being a widowed mother. Officials also assessed the moral character and behavior of the potential recipients. Second, although the program was enacted by states, it was primaril funded by cities and counties and also administered locally, by juvenile court judges or other local officials. Its administration thus did not represent a ne departure toward the creation of a centralized welfare state.

Third, the amounts appropriated not surprisingly fell far short of the total needed to meet the novel goal of the program. One of its major aims to provide funds so mothers could stay out of the labor force and their children in school.(57) In fact, in most states the stipends were, as outdoor relief had been in the preceding centuries, small supplements to income. Much of the refor legislation that expanded the role of American governments in the late nineteenth and early twentieth centuries was successful because its substance dealt with small, manageable problems. For example, variation in the timing of compulsory school attendance by states followed rather than preceded the attainment of nearly universal attendance in the relevant age groups. The law served to compel a small group of deviant children and their parents to comply.(58) Perhaps legislators originally hoped that the size of group being aided was small enough to be easily assisted. In effect, since the initial legislation was not tied to the appropriation of funds, legislators were making only a symbolic statement that reflected a perception that widowed mothers were worthy.

Mothers' pensions, then, were part of a historical context in which both labor-market poverty and life-cycle poverty were rarer than in England or western Europe. The economic problems of widows and other dependent poor remained visible to those who provided informal aid, private charity, and publi outdoor relief. Given the relatively small numbers involved and the small amounts of aid provided to each recipient, this "invisible" poverty problem was visible enough when it was perceived, as indeed it was, as close up rather than distant. A legislative innovation in the second decade of the twentieth century mothers' pensions in practice were more noteworthy for their continuity with past experience. The details of their implementation demonstrate the legacy of distinctive past on a seemingly innovative policy.

More generally, social historians need to explore further the implications of the important distinction developed in this article between the life-cycle and the labor-market poor. Both contemporary writers on poverty and later historian of the subject have neglected the former group. Their very worthiness has condemned them to obscurity. Yet, in terms of the economic burden of poverty on the non-poor and on the collectivity and, more especially, in terms of the numbers of people who were assisted, the life-cycle poor loom very large indeed

Department of History, M/C 198 851 South Morgan Street Chicago, IL 60607-6377

ENDNOTES

An earlier version of this paper was presented at a conference, "Lois Green Carr: The Chesapeake and Beyond--A Celebration," at the University of Maryland, College Park, May 22-23, 1992. The author is indebted to Mary Lynn Dietsche for research assistance with the 1790 sample and to Richard John, Steven Ruggles, Peter Stearns and Steven Wiberley for helpful suggestions for revision.

1. Unless otherwise stated, the terms female householder ratio and female headship ratio are used synonymously to refer to the ratio of female householders or heads to all householders or heads. The female householder or headship rate is the ratio of female householders to all females aged 16 and older; the denominator of the male rate is the number of males aged 16 and older. The English female householder ratio of 16.5% is the average for 70 communities enumerated between 1574 and 1821. Peter Laslett, "Mean Household Size in England since the Sixteenth Century," in Laslett, ed., Household and Family in Past Time (Cambridge, Eng., 1972), Table 4.9, p. 147. However, "Introduction: the History of the Family," Table 1.8, p. 78, yields a figure of 20.1%.

2. Mary Beth Norton, Liberty's Daughters: The Revolutionary Experience of American Women, 1750-1800 (Boston, 1980), 40-51, 125-38. Daniel Scott Smith, "The Meanings of Family and Household: Change and Continuity in the Mirror of the American Census," Population and Development Review 18 (1992): 421-456.

3. U.S. Bureau of the Census, Current Population Reports, Series P-20, No. 447 Household and Family Characteristics: March 1990 and 1989 (Washington, D.C., 1990), Appendix Table A-2, pp. 200-203.

4. Recently, for example, Waciega and Gough have disagreed about the extent of economic independence possible for widows in Philadelphia and several counties in southeastern Pennsylvania in the eighteenth and early nineteenth centuries. Lisa Wilson Waciega, "A 'Man of Business': The Widow of Means in Southeastern Pennsylvania, 1750-1850," William and Mary Quarterly, 3d Ser., 44 (1987): 60-64 Deborah Mathias Gough, "A Further Look at Widows in Early Southeastern Pennsylvania," William and Mary Quarterly, 3d Ser. (1987): 829-835; Waciega, "Reply," ibid., 835-839. A rare exception to concentration on a single locality is Lynne E. Withey, "Household Structure in Rhode Island, 1774-1800," Journal o Family History 3 (1978): 37-50. In Rhode Island, Withey notes, the percentage o female-headed households increased from 5% in subsistence agricultural towns, t 6-7% in commercial towns and 7-11% in smaller port towns, and finally to 17-20% in cities.

5. J.J. Spengler, "Malthusianism in Late Eighteenth Century America," American Economic Journal 25 (1935): 691-707; reprinted in John Cunningham Wood, ed., Thomas Robert Malthus: Critical Assessments (London, 1986), I, 72-89. Daniel Scott Smith, "A Malthusian-Frontier Interpretation of United States Demographic History before c. 1815," in Woodrow Borah, Jorge Hardoy and Gilbert A. Stelter, eds., Urbanization in the Americas: The Background in Comparative Perspective (Ottawa, Canada, 1980), 15-24.

6. Dennis Hodgson, "Benjamin Franklin on Population: From Policy to Theory," Population and Development Review 17 (1991): 639-661. Greater complexity appear in the frameworks of early nineteenth-century American writings on population. See James R. Gibson, Jr., Americans versus Malthus: The Population Debate in th Early Republic (New York, 1989). For the use of the framework as a periodizatio scheme, see Walter Nugent, Structures of American Social History (Bloomington, IN, 1981).

7. "Populous places have all times, been burthened with a larger proportion of paupers, than places where a thin or scattered population is found." J.V.N. Yates, "Report of the Secretary of State in 1824 on the Relief and Settlement o the Poor," reprinted from the Thirty-Fourth Annual Report of the New York State Board of Charities, January 28, 1901, p. 989, and in David J. Rothman, ed., Poverty, U.S.A.: The Almshouse Experience, Collected Reports (New York, 1971)

8. Drew R. McCoy, "James Madison and Visions of American Nationality in the Confederation Period: A Regional Perspective," in Richard Beeman, Stephen Botein, and Edward C. Carter, II, Beyond Confederation: Origins of the Constitution and American National Identity (Chapel Hill, 1987), 226-258.

9. Daniel Levine, Poverty and Society: The Growth of the American Welfare State in International Comparison (New Brunswick, NJ, 1988), 15.

10. For distinctions among types of poor, see Stephen Edward Wiberley, Jr., "Four Cities: Public Poor Relief in Urban America, 1700-1775," (Unpublished Ph.D. thesis, Yale University, 1975), 8-10. Conrad Edick Wright, The Transformation of Charity in Postrevolutionary New England (Boston, 1992), 28-29.

11. For a summary of early nineteenth-century laws, see Yates, "Report of the Secretary of State," 1068-1110.

12. Stuart Woolf, The Poor in Western Europe in the Eighteenth and Nineteenth Centuries (London, 1986), 32-33.

13. For example, in Aldenham, Hertfordshire, 34% received public support in the period 1641-1681. W. Newman Brown, "The receipt of poor relief and family situation: Aldenham, Hertfordshire, 1630-1690," in Richard Smith, ed., Land, Kinship, and Life-Cycle (Cambridge, 1984), Table 12.4, p. 412. About 30% of the recently widowed received outdoor relief in the period 1808-1812 in Newburyport Massachusetts. Susan Grigg, The Dependent Poor of Newburyport: Studies in Socia History, 1800-1830 (Ann Arbor, 1984), 19.

14. In nine seventeenth-century Norfolk, England, parishes, widows comprised 48 of all those receiving relief, and all categories of women supplied 61% of all pauper households. In many American communities, widows and single women living alone comprised a majority of those receiving relief. Tim Wales, "Poverty, poor relief and the life cycle: some evidence from seventeenth-century Norfolk," in Smith, ed., Land, Kinship, and Life Cycle, Table 11.3, p. 360. More generally see Richard Smith, "Some issues concerning families and their property in rural England 1250-1800," in Ibid., 72-86. In Women and the Law of Property in Early America (Chapel Hill, 1986), 184, Marylynn Salmon does not document the assertion that widows and single women comprised a majority of welfare cases.

15. Frances Fox Piven and Richard A. Cloward, Regulating the Poor: The Function of Public Welfare (New York, 1971). For a statement limiting the scope of their study to those not exempted from the necessity of working, see Piven and Cloward, "Humanitarianism in History: A Response to Critics," in Walter I. Trattner, ed., Social Welfare or Social Control?: Some Historical Reflections o Regulating the Poor (Knoxville, TN, 1983), 138.

16. Evsey D. Domar, "The Causes of Slavery or Serfdom: A Hypothesis," Journal o Economic History 30 (1970): 18-32. William H. McNeill, The Great Frontier: Freedom and Hierarchy in Modern Times (Princeton, 1983). Howard Lamar, "From Bondage to Contract: Ethnic Labor in the American West, 1600-1890," in Steven Hahn and Jonathan Prude, The Countryside in the Age of Capitalist Transformation: Essays in the Social History of Rural America (Chapel Hill, NC, 1985), 293-324.

17. For a description of the sample, see Daniel Scott Smith, "'All in Some Degree Related to Each Other': A Demographic and Comparative Resolution of the Paradox of New England Kinship." American Historical Review 94 (1989): 74-79. The sex of the head was not explicitly recorded by the census and thus had to b inferred from the forename. For two regions--the Carolinas and southern New England--the forenames of householders that contained white females and no male over age 16 were separately examined. It was assumed that females were the head of these households. In the former region, a maximum of 16 of 168 names of family heads (9.5%) that almost certainly belonged to women by the above rule were ambiguous and not easily spotted as those of females: Christian/Chritian, Desy, Frank, Frans, Jude, Love, Maple, Massey, Oma, Uley, Sillah, Sande (2), Tener, and Tamar. In southern New England, out of 365 heads, a minimum of 13 names [Alatheus, Allin (2), Damarus, Elliot, Elijah, Juda, Lewis (2), Prince, Rob't, Sam'l, and Subinet] appeared to be masculine and eight others (Asenath, Bethiah, Easter, Meriam, Noami, Robin, Toby, and Zilpha) were judged unusual enough not to be picked up as female while scanning the lists. By the lower estimate, some 3.5% of female heads would be miwsed, and by the higher, 5.2%. I order to obtain a maximum estimate, the percentage of female householders in households that contained no males over 16 was assumed to be correct. However, in households that had males over age 16, the percentage of female householders was multiplied, somewhat arbitrarily, by 8.25%.

A more serious defect arises from the loss of the original schedules of the 179 census for some states or counties. Printed versions of the original documents exist for all of New England, New York and Pennsylvania, Maryland, and the Carolinas. The compilers of the printed volumes substituted names and data from tax lists from the 1780s for Virginia and for a few counties elsewhere. In this study, data for one North Carolina county (Orange) and for all twelve counties in Virginia come from tax rather than census lists. In the southern sampling units with extant 1790 census listings, women headed over 75% more households than in the 13 counties with only tax records available.

To arrive at an upper estimate of female householders, their numbers were increased in the 13 southern places with tax list data. In making the correction, the more important assumption is that the overall averages in both tax- and census-record southern places would be identical if the areas in both categories were at the same stage of demographic development. Of lesser quantitative importance is the adjustment for the developmental stage that was assumed to be captured by the estimated sex ratio of the white population aged 16 and older. For counties where tax lists were used instead of the census records, the percent of female headed households was multiplied by 1.754.

18. Jean Buttenhoff Lee, "Land and Labor: Parental Bequest Practices in Charles County, Maryland, 1732-1783," in Lois Green Carr, Philip D. Morgan and Jean B. Russo, eds., Colonial Chesapeake Society (Chapel Hill, 1988), 330-38. Virginia and especially South Carolina limited the rights of widows to inherit slaves. Salmon, Women and the Law of Property, 149-160.

19. Peter Laslett, "Characteristics of the Western Family Considered over Time, Journal of Family History 2 (1977): 89-115. John Hajnal, "Two Kinds of Pre-Industrial Household Formation System," in Richard Wall, ed., Family Forms in Historic Europe (Cambridge, Eng., 1983), 65-104. Daniel Scott Smith, "American Family and Demographic Patterns and the Northwest European Model," Continuity and Change 8 (1993): 389-415.

20. In the public use sample from the U.S. manuscript census of 1900, nearly al (97%) of spouse-present married men were listed first in the census household. Some 42% of post-married men aged 16 and older headed households compared to 9% of never-married men.

21. In the 1900 public use sample, some 51.4% of post-married (mostly widowed) women aged 16 and older headed households compared to 3.5% of never-married women. Only 0.1% of spouse-present married women were listed first in the censu household.

22. A complete count of the 1790 listings for the entire state reveals that 3.8 of all householders were females, and that 2.6% of females over age 15 were householders. Smith, "The Demography of Widowhood in Pre-Industrial New Hampshire," in David L. Kertzer and Peter Laslett, eds., Aging in the Past: Demography, Society, and Old Age (Berkeley: University of California Press, forthcoming). The English figure is from Wall, "Women Alone," Table 4, p. 313.

23. Controlling simultaneously for both age and marital status, if such data were available, would provide a better fix on the separate roles of age and marital distribution.

24. Daniel Scott Smith, "Historical Change in the Household Structure of the Elderly in Economically Developed Societies," in Robert Fogel et al., eds., Aging: Stability and Change in the Family (New York, 1981), Table 5.1, p. 100. Also see Steven Ruggles, "The Transformation of American Family Structure," American Historical Review 99 (1994): 118-120.

25. Smith, "'All in Some Degree Related'," 44-79.

26. Density was measured for towns in New England but for counties elsewhere in the country. In general, better measurement produces higher correlations. Also see the results for the relationship of these indicators to density variations among New Hampshire towns. Smith, "Demography of Widowhood," Tables 3 and 4.

27. Together the predictors accounted for 56% of the variance in the adjusted female ratio, 51% of the adjusted female ratio but only 16% of the adjusted mal ratio.

28. Quoted by Jill S. Quadagno, "The Transformation of Old Age Security," in David Van Tassel and Peter N. Stearns, eds., Old Age in a Bureaucratic Society: The Elderly, the Experts and the State in American History (Westport, CT, 1986) 132.

29. By propensity is meant the disposition to engage in an action given the opportunity. For example, only old people who have children have a non-zero propensity to live with a child. It is possible that this assumption is wrong; the lower figure for New England might be attributed to the small proportion of non-American born in that region or some other inadequately measured demographi factor.

30. Wiberley, "Four Cities," 80-85.

31. See especially Laurel Thatcher Ulrich, Good Wives: Image and Reality in the Lives of Women in Northern New England, 1650-1750 (New York, 1982).

32. This was the fraction of married men and women who indicated their belief i joint headship in responses to a special inquiry of the Census Bureau in September 1975. Unpublished Census Bureau memo of October 17, 1975, cited in Smith, "Meanings of Family and Household," 447-448. The calculation assumes tha the subjective status of married women within the household was not lower in 1975 than it had been in 1790.

33. Smith, "Demography of Widowhood," Table 1.

34. Charleston, South Carolina, and Baltimore, Maryland, are the two southern cities above 8,000 population.

35. Ann Patton Malone, Sweet Chariot: Slave Family and Household Structure in Nineteenth-Century Louisiana (Chapel Hill, 1992), Table 1.1, p. 15. Also see, Stephen Crawford, "The Slave Family: A View from the Slave Narratives," in Claudia Goldin and Hugh Rockoff, eds, Strategic Factors in Nineteenth-Century American Economic History (Chicago, 1992), 331-350.

36. In 1755, 20% of all free blacks in Maryland were "past labour or cripples" compared to only 2% of white men. Allan Kulikoff, "The Beginnings of the Afro-American Family in Maryland," in Aubrey C. Land et al., eds., Law, Society and Politics in Early Maryland (Baltimore, 1977), 189.