¹ Crane, Stephen, War Is Kind and Other Lines, in The Works of Stephen Crane, ed. Follett, Wilson, vol. 6 (New York, 1926), p. 122Google Scholar.

² Hill, Christopher, “Pottage for Freeborn Englishmen: Attitudes to Wage Labour in the Sixteenth and Seventeenth Centuries,” in Socialism, Capitalism, and Economic Growth: Essays Presented Maurice Dobb, ed. Feinstein, Charles H. (Cambridge, 1967), p. 347Google Scholar.

³ Soltow, Lee, “Long-Run Changes in British Income Inequality,” Economic History Review, 2nd ser, 21 (April 1968), 17–29Google Scholar.

⁴ Deane, Phyllis, “The Implications of Early National Income Estimates for the Measurement of Long-term Economic Growth in the United Kingdom,” Economic Development and Cultural Change, 4 (1955), 3–38Google Scholar; Deane, Phyllis and Cole, W.A., British Economic Growth, 1688–1959, 2nd ed. (Cambridge, 1969), p. 2Google Scholar.

⁵ King, Gregory, Natural and Political Observations and Conclusions upon the State and Condition England, in Two Tracts by Gregory King, ed. Barnett, George E. (Baltimore, 1936)Google Scholar

⁶ Probate valuations are plentiful from 1660 to the early eighteenth century and then dwindle away across the middle of the eighteenth century. They become suddenly abundant again in 1796, when a new light probate duty was imposed. Given this timing, the related study of probated wealth chose to sample at these benchmark dates: 1669–1670; 1699–1700; 1731 and 1739–1741; 1810; and 1875. A main purpose for the present occupational study was to give occupational distributions that could be used to adjust for some of the social bias in the probate data. Hence the choice of occupational samples centering on 1670, 1700, 1740, and the wartime period including the year 1810.

⁷ A small but significant minority of children were adults, according to the many census and burial records giving both ages and occupations in the early nineteenth century. I have counted adult “son” and “daughter” entries as having no occupation whenever none was mentioned. Such entries are counted in the totals for children, as though they were minors, and not in the totals for men and women.

⁸ In addition to the infinite literary evidence on work by women (and children), we have some quantitative glimpses for the late eighteenth century. See the lists of inhabitants of Corfe Castle, Dorset in 1790 (Dorset Record Office and Cambridge Group) and the same for Cardington, Bedfordshire in January 1782 (Bedfordshire Record Office and Cambridge Group). For their work and earnings in poor families, see Rev.Davies, David, The Case of Labourers in Husbandry (Bath, 1795)Google Scholar, and SirEden, Frederick Morton, The State of the Poor, original ed. (London, 1797), vols. 2 and 3Google Scholar.

⁹ Here I imply that the recorded labels understate the true share in poverty. But what is meant by a “true” share here? The implicit norm is the share of the population that would have received this label their local parish clerks were systematically queried about their status, as in a Parliamentary survey or a census.

¹⁰ Tawney, A.J. and Tawney, R.H., “An Occupational Census of the Seventeenth Century,” Economic History Review, 5 (Oct. 1934), 98–103Google Scholar.

¹¹ Galenson, David W., “The Indenture System and the Colonial Labor Market: An Economic History of White Servitude in British America” (Ph.D. diss., Harvard Univ., 1979), chaps. 4 and 5Google Scholar.

¹² The variety of burial labels extended beyond occupations, social class, marital status, age, and place of origin. A few clerks volunteered moral commentary on the deceased—e.g., “an infamous slut,” “a decayed gent,” and “a sober young man of 34.” Others helpfully told us the cause of death, as in the entry “This man's body was found in the broads. The Coroner sat on it and ruled it was an Act of God.”

¹³ A “cowper” may or may not have been a cooper. And did a “pig tobber” do it with pigs or with pig iron? Other cases yielded to reflection: we concluded that a London “French master of the Matthew Matticks” was not a shipmaster after all.

The socio-occupational categories used below are denned and illustrated in WP Appendix Tables Al and A2.

¹⁴ See WP, Appendix Table A3.

¹⁵ Barnett, Two Tracts by Gregory King, p. 23. This figure may be a bit high. Certainly King's assumed mean age of 12 years for all children seems high, given the low share of children over 20.

¹⁶ See WP, Appendix Table A3. The same tendency for laborers to be older than craftsmen or the average occupied male shows up in 1871 census totals for England and Wales: the average age of agricultural and general laborers (among males over 15) was 39.2 years, while the overall occupied-male average was 36.9 and the average for tailors, blacksmiths, carpenters, joiners, cotton manufacturers, shoemakers, and bootmakers was 36.6.

¹⁷ See WP, Appendix Table A4, and the author's “An Algorithm for Probate Sampling,” Journal of Interdisciplinary History, in press.

¹⁸ Johnson's definitions defy social ranking: Farmer—(1.) One who cultivates hired ground; (2.) one procewho cultivates ground, whether his or another's. Yeoman—(1.) a man of small estate in land; a farmer; a gentleman farmer; (2) anciently a title given to soldiers; (3.) It was probably a freeholder not advanced to the rank of gentleman; (4.) It seems to have been likewise the notion of a gentleman servant. Husbandman—One who works in tillage. These near-quotations are from his Dictionary of the English Language (London: Thomas Tegg, 1831 ed.), vol. I, pp. 693 and 925; and vol. II, p. 1079.

¹⁹ See WP, Table 3 and Appendix Table A4.

²⁰ See WP, pages 26–42 and Appendix Tables A5 through A7.

²¹ The equation for children's share of total population is; of course, redundant, given the equations for the shares of men and women. Yet I included it as an extra way of revealing prediction errors in my regression system: by including its coefficient in the subsequent predictions of total numbers of children in England and Wales, I have given the extrapolating predictions a chance to violate the identity that men + women + children = population. The net errors of extrapolation to the total population are displayed in Table 3 below.

The original list of dependent variables numbered 59 rather than the 37 mentioned and used here. The others referred to additional small-group breakdowns (e.g., clergy/men, grocers/men) for which the regressions showed wide margins of error, as expected. The fine breakdowns were discarded as unreliable.

²² Exceptions to the use of county-level data were the aggregation of all of Wales into a single unit, the separation of the City of London from the rest of Middlesex, and the splitting of four rising cities (Birmingham, Hull, Manchester, and Liverpool) from the rest of their counties.

²³ The occupational non-reporting rate was set at zero for all counties, as suggested above, in order to estimate what occupations would have emerged from perfect reporting. The year dummies were averaged over five-year periods including the dates of estimate: 1668–1672 for 1670, 1686–1690 for 1688, 1698–1702 for 1700, 1738–1742 for 1740, 1751–1755 for 1755, 1801–1805 for 1801–1803, and 1808–1812 for 1811.

²⁴ Wrigley, E.A. and Schofield, R.S., The Population History of England, 1541–1871: A Reconstruction (Cambridge: Cambridge University Press, forthcoming), Table A 13.5Google Scholar. For the death rates used here, see WP, appendix Tables A6 and A7.

²⁵ See WP, pp. 41, 42.

²⁶ Such standard errors may understate even the true regression errors themselves if the type of model is misspectned. Understatement could arise from autocorrelation of residuals for the same parish in different years, or from incorrect choices of functional form and independent variables.

²⁷ Deane and Cole, British Economic Growth, p. 142.

²⁸ The quotations here are, of course, from the title of Paul David's valuable exploratory article on U.S. national product before 1840 in the June 1967 issue of this Journal.

²⁹ Peter H. Linden and Jeffrey G. Williamson, “Britain's Social Tables Revised, 1688–1867,” in progress. Gregory King, himself, incidentally, has turned up in my occupational Sample 5, as an “Esquire” and a head of household, in the Marriage Duty Return for St. Bennetts Paul's Wharf, London, 1st May 1695.

³⁰ In what follows, “King” is the version in Barnett, Two Tracts, p. 31.

³¹ I am indebted to Paul David for his suggesting that I explore this possibility of a secondary-trade bias in homogeneous rural areas.

³² In 17 rural places in Sample 5 (4 in Kent; Hothorpe, Northants.; Chilvers Coton, Warks.; 10 villages in Westmoreland; and Donhead St. Mary, Wilts.) 99 males, or 6.89 percent of all labelled men (or 6.46 percent of all men), were in manufacturing trades alone, whereas only 4.31 percent of King's family heads or vagrants were engaged in all artisan and craft classes.

³³ Tawney and Tawney, “An Occupational Census,” Table X.

³⁴ Holmes, G.S., “Gregory King and the Social Structure of Pre-Industrial England,” Transactions of the Royal Historical Society, 5th ser., 27 (1977), 56, 57, and 67Google Scholar.

³⁵ Mathias, Peter, “The Social Structure in the Eighteenth Century: A Calculation by Joseph Massie,” Economic History Review, 2nd ser., 10 (Oct. 1957), 30–45Google Scholar.

³⁶ The official figures for the army, the navy, and civilian mariners tend to support the Table 3 estimates not only for 1801–1803 but for other dates as well. This can be seen by comparing Table 3 with the numbers of army personnel voted for and the number of navy personnel borne (both from House of Commons, Sessional Papers, 1868–1869, vol. XXXV, Pt. 2, pp. 693–702) and the taxed number of civilian seamen (Davis, Ralph, “Seamen's Sixpences: An Index of Commercial Activity, 1697–1828,” Economica, Nov. 1956, pp. 339–40Google Scholar). After the military numbers have been multiplied by the population shares of England and Wales in Britain for different dates, we get the following rough comparisons with the present estimates for England and Wales:

a = 1691; b = 1689–1690; c = 1712.

³⁷ “Before the 1770s, the cotton hand weavers as a body had not existed; by the late 1840s, they had effectively vanished. In three generations, the process of economic change had first created and then destroyed a new type of labour.” (Bythell, Duncan, The Handloom Weavers [Cambridge, 1969], p. 40.Google Scholar)

³⁸ In the half-century after 1755 coal output in England and Wales rose only about fourfold, according to Pollard, Sidney, “A New Estimate of British Coal Production,” Economic History Review, 2nd sen, 33 (May 1980), 229Google Scholar.

³⁹ Of the 736 enrollees into the militia at Woodbridge, Suffolk in 1814, 402 were laborers, 36 were servants, and 4 were in other menial services at the time of enrollment (WP, Appendix Table A15).

⁴⁰ I used the 1831 census to put upper and lower bounds on the shares of laborers who were in agriculture in 1811. The upper-bound estimate assumes that the agricultural share of laborers and menial service workers was as high as the highest share reported in any county in 1831—912 for Buckinghamshire. The lower-bound estimate assumes that the same share was as low as it had become for all of England and Wales by 1831, or 6002.

⁴¹ Banks, J.A., “The Social Structure of Nineteenth Century England as Seen through the Census,” in The Census and Social Structure, ed. Lawton, Richard (London, 1978), p. 190Google Scholar.