. 4
( 14)



Female/Male Wage Ratio







0 20 40 60 80
Figure 2.2 The female“male wage ratio by age in agriculture
Source: Joyce Burnette, “How Skilled Were English Agricultural Laborers in the
Early Nineteenth Century?” Economic History Review 59 (2006), p. 714.

competitive market employees are price-takers, but will adjust their
hiring so that the marginal product of labor equals the wage. The actions
of employers determine the demand for labor, and the wage is deter-
mined by the interaction of this demand for labor and the supply of
labor. An alternative to the competitive model is the monopsony model.
A monopsony occurs when there is only one employer who can hire the
worker, and in this situation the employer can use his market power to
pay wages lower than the marginal product of labor. However, econo-
mists generally do not believe the monopsony model has a wide appli-
cation.7 If there is evidence that workers have a choice of possible
employers, economists generally assume that markets are competitive
and wages are equal to the marginal product of labor.
Economists who believe that markets are competitive use the gender
wage gap as evidence of productivity differences between men and

For an attempt to apply the monopsony model more broadly, see David Card and Alan
Krueger, Myth and Measurement: The New Economics of the Minimum Wage (Princeton:
Princeton University Press, 1995).
Women™s wages 81

women. While it is possible to measure productivity using production
functions, it is much easier to use observed wage differences to measure
productivity differences. Goldin and Sokoloff, for example, assume that
the relatively low female wage in the US North indicates that female
workers had a relatively low marginal product, and they use this
observation to explain the rise of textile manufacturing in the North.8
When aggregating the amount of labor used by manufacturing ¬rms,
Sokoloff counts an adult woman as the equivalent of half an adult man
because women™s wages were about half of men™s wages.
Females and boys have been treated as equal, in terms of their labor input, to one-
half of an adult male employee, with these weights having been drawn from
evidence on the relative wages of the groups prevailing near the end of the period.9

In a comment on this article, Jeffrey Williamson questions whether
assuming a constant productivity ratio over time is valid, but does not
question the assumption that the wage ratio is an accurate measure of
the productivity ratio.10 Similarly, Atack, Bateman, and Margo assume,
based on the wage ratio, that an adult female worker is equal to 60
percent of an adult male worker.11 Doraszelski also assumes that relative
wages measure relative productivity when he uses wage rates to calculate
the contributions of female and child workers to aggregate labor in his
study of French industry.12
Not everyone agrees that the wage gap is evidence of productivity
differences. On the other side of the debate are those who believe that
women™s lower wages were not justi¬ed by productivity differences, but
were set by custom. For example, Pamela Sharpe claims that the wages
of female servants were “a matter of custom bearing little relationship to
economic determinants.”13 Sonya Rose claims that “Women could be

Claudia Goldin and Kenneth Sokoloff, “The Relative Productivity Hypothesis of
Industrialization: The American Case, 1820 to 1850,” Quarterly Journal of Economics 99
(1984), pp. 461“87.
Kenneth Sokoloff, “Productivity Growth in Manufacturing during Early
Industrialization: Evidence from the American Northeast, 1820-1860,” in Stanley
Engerman and Robert Gallman, eds., Long-Term Factors in American Economic Growth
(Chicago: University of Chicago Press, 1986), pp. 702“3.
Jeffrey Williamson, “Comment,” in Stanley Engerman and Robert Gallman, eds.,
Long-Term Factors in American Economic Growth (Chicago: University of Chicago Press,
1986), pp.729“33.
Jeremy Atack, Fred Bateman, and Robert Margo, “Productivity in Manufacturing and
the Length of the Working Day: Evidence from the 1880 Census of Manufactures,”
Explorations in Economic History 40 (2003), pp. 170“94.
Ulrich Doraszelski, “Measuring Returns to Scale in Nineteenth-Century French
Industry,” Explorations in Economic History 41 (2004), pp. 256“81.
Sharpe, Adapting to Capitalism, p. 114.
82 Gender, Work and Wages in Industrial Revolution Britain

paid low wages because they were women. They earned a customary
wage, not one which was generated out of open competition in a sexually
neutral labor market.”14 Similarly, Deborah Simonton claims that, for
nineteenth-century women, “Wages, like the gendered character of many
jobs, rested on custom rather than real labor value.”15 While Woodward
¬nds early modern male wages responding to supply and demand, he
thinks female wages were different:
The low rates of pay given to most women were rooted in convictions about their
physical, economic and social, intellectual, and political inferiority which char-
acterized English society into the present century, and which were underscored
by biblical authority. Their rates of pay were not simply re¬‚ections of the supply
of and demand for female labor.16

This group of historians explains women™s wages not in terms of their
productivity, but in terms of social expectations of women™s inferiority.
There are three main ideological assumptions about women that are
usually blamed for keeping women in low-paid work: women were
assumed to be weak, unskilled, and dependent on men. Some historians
suggest that assumptions about women™s physical weakness justi¬ed low
wages and kept them con¬ned in certain occupations. Michael Roberts
suggests two reasons why seventeenth-century farmers saw women as
“the weaker vessel” and thus allocated them the lightest tasks. One was
the physical weakness caused by pregnancy, and the other was the moral
weakness evident in the biblical story of the fall, where Eve gave in ¬rst
to temptation.17 An important point is that the assumption of female
weakness is not tied to physical reality; Deborah Simonton claims that
“The association of women with weakness was not necessarily a bio-
logical notion, since the association of woman as the ˜weaker vessel™ was
as much an ideological construction as it was physical.”18
Women might also be assigned to low-paying jobs because they were
assumed to be unskilled workers. Sonya Rose suggests that women
factory workers were paid less than men and were not given jobs with the
potential for advancement because it was assumed that they could not
acquire the necessary technical skills.
Employers considered mechanical aptitude to be a purely masculine trait. They
talked about men™s “natural” technical ability and women™s mechanical

Rose, “Gender Antagonism,” p. 208. 15 Simonton, European Women™s Work, p. 170.
Donald Woodward, “The Determinants of Wage Rates in the Early Modern North of
England,” Economic History Review 47 (1994), p. 37.
Michael Roberts, “Sickles and Scythes: Women™s Work and Men™s Work at Harvest
Time,” History Workshop 7 (1979), p. 11.
Simonton, European Women™s Work, p. 34.
Women™s wages 83

ineptitude as though this was a gender difference everyone recognized; it was
common sense. The belief that women naturally lacked facility with machinery
was in fact widely held, and employers used it to justify paying women less than
men for the same jobs.19

The assumption that women were unskilled was strong enough to lead
to the relabeling of work done by women. Bridget Hill found that census
of¬cials were unwilling to categorize occupations hiring women and
children as skilled.
Albe Edwards, the man responsible for the reclassi¬cation, met with a problem
when he found certain occupations which technically were classi¬ed as “skilled”
had to be down-graded to “semi-skilled,” “because the enumerators returned so
many children, young persons, and women as pursuing these occupations.”
Edwards did not hesitate to lower the status of certain occupations when he
found women and young people worked in them in large numbers.20

In this case the categorization of occupations as skilled or semi-skilled
re¬‚ects ideology rather than characteristics of the job.
Women™s low wages are often said to result from the fact that, being
dependent on men, they “needed” less income. Sonya Rose, for
example, emphasizes the expectation that women were secondary
earners, who did not need to support a family, and whose wages were
only supplementary to the wages of the men on whom they were
dependent: “Women were workers who could be paid low wages
because of an ideology which portrayed them as supplementary wage
earners dependent on men for subsistence.”21 Deborah Valenze also
claims that “the level of a woman™s earnings was determined by an
assumption that her wage was a supplement to some other (most likely a
breadwinner™s) wage.”22
However, while it is true that contemporaries did hold these beliefs
about women™s inferiority and dependence, it does not necessarily follow
that these beliefs were the cause of women™s low wages. While I do not
question the claim that contemporaries believed women to be weak,
unskilled, and dependent on men, I do question whether these ideolo-
gies were the real motivations behind the actions of employers, or
whether they were simply the justi¬cations given by employers to dis-
guise their true motivations. Unfortunately it is dif¬cult to determine
whether women™s wages were set by markets or by custom because both
theories suggest that women would earn lower wages than men. The

Rose, Limited Livelihoods, p. 27. 20 Hill, “Women, Work and the Census,” p. 90.
Rose, “Gender at Work,” p. 117.
Valenze, The First Industrial Woman, p. 108; see also p. 89.
84 Gender, Work and Wages in Industrial Revolution Britain

remainder of this chapter examines the wage data carefully and argues
that in competitive portions of the labor market women™s lower wages
were the result of their lower productivity, and were set by the market
rather than custom.

I. Interpreting piece-rate wages
For piece-rate wages, wage discrimination exists only if women and men
are paid different piece-rates. If women earned less than men but were
paid the same piece-rates, then the difference in earnings is clearly due
to differences in output. While there are a few examples of piece-rates
that differ by gender, such examples of wage discrimination are the
exception rather than the rule. In most cases men and women were paid
equal piece-rates when they worked at the same tasks.
Even for workers who were paid by the piece, many historical sources
report daily or weekly earnings rather than the actual piece-rates.
Section B of Table 2.1 gives examples of such reported earnings. One
reason this is true is that piece-rates could be very complex. In mule-
spinning there was a different piece-rate for every count, or ¬neness, of
yarn. In handloom weaving there was a different piece-rate for each type
of cloth. Reporting all these numbers might confuse the reader.23 Also,
contemporary readers were usually concerned about the living standards
of the workers, and thus were interested in the earnings of the workers
rather than the piece-rates per se. Daily or weekly earnings, though,
do not tell us whether there was gender discrimination. The gap
between women™s and men™s earnings on piece-rates seen in section B of
Table 2.1 may have a variety of causes, including mismeasurement,
differences in hours worked, differences in output per hour, and dif-
ferences in piece-rates paid. I will discuss these four possibilities in turn.
One possible reason for the wage gap observed in piece-rate earnings
is measurement error. Wages given for work done in the home often
con¬‚ate the earnings of different members of the family. Even if the
whole family contributed labor, the entire amount gained may be
counted as the man™s wage. In his parliamentary report on handloom
weavers, H. S. Chapman states,
On one occasion I saw a piece of shalloons woven under the following cir-
cumstances: 1. The man was in the loom weaving. 2. A boy of 10 years of age
was winding bobbins. 3. The wife was at her husband™s elbow picking the work.

For an example of a parliamentary committee being confused by wage evidence
presented by a witness, see the minutes of evidence from the 1808 Report on the Cotton
Weavers™ Petition. BPP 1808 (177) II, p. 10.
Women™s wages 85

4. His daughter was at the back of the loom taking up the broken threads of the
warp, for it was not a good warp. When the piece was taken home, the wages
would be paid as the earnings of one man, yet the piece was really the work of a

This type of mismeasurement would systematically raise estimates of
men™s earnings and lower those of women™s earnings. In the reports of
parliamentary committees we ¬nd speci¬c examples of this mismeas-
urement occurring. Witness the following exchange between a parlia-
mentary committee and a hatter from London:
How much do you earn a week now? “ Perhaps £2.8s.
Out of that what payment have you to make? “ If I had not a wife I should have
to pay 6s. out of that for picking, if ovals; if ¬‚at 9s.4d.
What do you mean by picking? “ The women are employed in picking the coarse
hairs out that are in the stuff, that my wife does.25

Note that the hatter claimed to earn 48s. per week, though by his own
admission at least 6s. of this (one-eighth) represents value of work done
by his wife.
The following example further illustrates this point. In 1824 Joseph
Sherwin, a handloom weaver from Stockport, told a parliamentary
committee that he usually earned 6s.6d. a week, and his wife 3s. by
winding bobbins for two other looms.26 However, he failed to subtract
from his wage, and add to hers, the value of her winding services for his
loom. Mrs. Sherwin received for winding 3d. out of every shilling earned
by each of two weavers who hired her services; each of these weavers,
then, earned only 9d. for every 1s. worth of cloth.27 Since she could
wind for three looms (her husband™s plus two others), Mrs. Sherwin
could earn the same amount as these two weavers, (9d. ¼ 3 · 3d.).
Joseph Sherwin admitted, “I must pay three pence out of every shilling,
if I had no wife.”28 His true wage, then, was only 4s.10½ d. (¼0.75 ·
6s.6d.), and his wife™s true wage was 4s.7½ d. (¼ 3s. þ [0.25 · 6s.6d.]).
(The difference arises only because Joseph weaves cloth worth 6s.6d. a
week, while each of the other two weavers complete only 6s. worth of
cloth in a week.) What by his statement appeared to be a wage ratio of
0.46 (3s./6s.6d.) turns out to be, in truth, approximately equal wages.

BPP 1840 (43) XXIII, p. 561. He also says, “when a manufacturer says ˜such a weaver
can earn so much,™ it may happen that the sum is really, as already explained, the wages
of two persons” (p. 559).
BPP 1824 (51) V, p. 97. 26 Ibid., pp. 419“20.
Since a shilling was worth 12 pence, 3d. out of every shilling was 25 percent.
BPP 1824 (51) V, p. 419.
86 Gender, Work and Wages in Industrial Revolution Britain

If women were credited with the work that they did, the wage ratios
would appear slightly higher, but women would still earn less than men.
A second possible reason for the wage gap in piece-rate earnings is the
difference in hours worked per day or per week. Most women had
domestic duties that reduced the amount of time available for market
work, so on average women generally worked fewer hours than men. In
1840 parliamentary investigator James Mitchell found that women hand-
loom weavers working in weaving shops earned an average of 5s.5d. a
week, while those working at home averaged only 4s. He attributed
this difference to a difference in hours worked: “The lower average of
the wages of the women working at home is, in a considerable degree,
attributable to the circumstance that many of them are married women,
and their time is partly occupied by their domestic duties.”29 When
investigators presented weekly wages, they probably did not assume that
women worked as many hours as men. Davies makes his calculations for
hand spinning explicit: “When she sits closely to her wheel the whole day,
she can spin 2 lbs. of coarse ¬‚ax for ordinary sheeting and toweling, 2½ d.
per lb.”30 Davies then assumes four days of work in a week, for earnings of
1s.8d. a week. Usually we are simply given the estimated earnings, with
no mention of time input. If we were told only that the earnings were
1s.8d. a week, we might erroneously assume this was for a full week™s
work and seriously underestimate the daily wage.31 Unfortunately we do
not know how many hours women in domestic industry spent at their
paid work. Even their employers did not know how many hours they
worked. These workers would often keep secret or even misrepresent the
number of hours they spent working, for fear that employers would
demand faster work, or to make their wages appear lower.32

BPP 1840 (43) XXIII, p. 317. The same was true in the hosiery industry; see Rose,
“Gender Segregation,” p. 167.
David Davies, The Case of Labourers in Husbandry Stated and Considered (London:
Robinson, 1795), p. 85.
Pinchbeck, Women Workers, p. 138, noted the need to correct for hours of work: “Labor
at the hand wheel was often a by-employment, and the time given to it varied according
to the housewife™s preoccupation with other duties. The small amounts earned by
laborers™ wives and others who span to supplement the family income cannot, therefore,
fairly be described as spinner™s wages, and must not be confused with the earnings “ for
the most part adequate “ of those who regarded it as a full time occupation.”
In response to a parliamentary survey, the overseer of Gestingthorpe, Essex, states, “It
is very dif¬cult to get at the amount of earnings of that part of the family which is not
employed by the farmer, as they conceal them for fear of having their allowances
diminished.” BPP 1834 (44) XXX, p. 175a. The motivation for concealing or
exaggerating hours worked might be to get poor relief from the parish, or to convince
the parliament to mandate a minimum wage, as it did for the Spital¬elds silk weavers.
For handloom weavers, investigator J. Mitchell spoke of “many of the weavers and their
ill-judging friends endeavouring to make it appear, that from dire necessity the weaver
Women™s wages 87

The problem of differences in hours worked was most acute in
domestic industry, but even in some kinds of factory work we occa-
sionally ¬nd women working fewer hours than men. In 1816 Henry
Houldsworth, a cotton manufacturer from Glasgow, noted that the
women who picked the cotton for him could come and go as they
pleased, and “as they are a very irregular set of hands, time is not
noticed.”33 Since he paid them only for their output, and no machinery
was used in this process, he had no reason to insist on regular hours. The
result was lower average working hours for these women. Houldsworth
estimated that these women averaged seven hours a day: “Some of them,
in summer time, will work as much as eleven or twelve hours; but taking
them altogether, I do not conceive they average seven hours, summer
and winter.”34 Seven hours was a little more than half the twelve hours
¬fty minutes that Houldsworth™s regular mill-hands worked. Such a
large difference in hours, even for only a portion of the workers, could
easily bias female“male ratio of factory wages.
A third possible reason for women™s lower earnings in piece-rate work
was that women may have produced fewer units per hour. Unfortu-
nately, we do not have evidence on output per hour of piece-rate workers
because the time worked was not recorded. However, we do know that
in US manufacturing women produced fewer units per hour than men.
Claudia Goldin reports that women in manufacturing earned less than
men in spite of working the same hours and being paid the same piece-
rate, which implies lower output per hour.35
While it is dif¬cult to measure output per hour, we can observe the
actual piece-rates more readily. If the piece-rates paid to men and
women were the same, then men and women were paid the same wage
for the same output, even if they earned different amounts per week.
Wage discrimination existed only if men and women were paid different
piece-rates. The majority of the evidence suggests that, when men and
women did the same work, they were paid the same piece-rate. For
example, Table 2.2 shows the amounts paid for reaping at Gooseacre
Farm in Radley, Berkshire. This group of reapers clearly contained both
men and women, and all workers, regardless of their sex, received the

works a far greater number of hours than it would be supposed that human nature
could endure.” BPP 1840 (43) XXIII, p. 238. From the report of W. A. Miles on
handloom weavers we learn, “The weavers did not like the masters to know in what
time a chain could be woven, because they considered if the master knew that they
could earn a given sum in a short time, the price of the next chain would be lowered.”
BPP 1840 (220) XXIV, p. 383.
BPP 1816 (397) III, p. 233. 34 Ibid.
Claudia Goldin, Understanding the Gender Gap: An Economic History of American Women
(Oxford: Oxford University Press, 1990), p. 104.
88 Gender, Work and Wages in Industrial Revolution Britain
Table 2.2. Payments for reaping at Gooseacre Farm, Radley, Berkshire

Area Reaped Payment
Name of worker A-r-p Payment £ s d per acre (s.)

John Fisher 2-3-0 1-4-9 9
G. Comely 3-0-37 1-9-1 9
Ann Waters 1-3-28 0-17-4 9
J. Minns 3-0-39 1-9-2½ 9
Wm Stinson 3-2-15 1-12-4 9
Wm House 3-1-0 1-9-3 9
Mary Minns 1-0-36 0-11-1 9
T. Gunter 2-1-27 1-1-8½ 9
Mary Grimes 1-0-16 9

Source: Rural History Centre, University of Reading, BER 13/5/2.

same price per acre. The women earned notably less than the men (the
three workers who were clearly women earned on average only 12s.9d.,
which is less than half the 28s.9d. earned on average by the three
workers who were clearly men), but this is because they reaped fewer
acres. There is no way to tell from this source why the women reaped
fewer acres “ because they worked fewer hours, because they accom-
plished less per hour, or because they had fewer unnamed assistants than
the male workers. But the reason for the difference in earnings in clear:
the women earned less because they reaped fewer acres, not because
they were paid at a different rate per acre than the men.36
Gooseacre Farm was not unusual; other evidence also points to equal
piece-rates in agriculture. On the few occasions where we observe both
men and women doing the same agricultural tasks on piece-rates, they
are paid the same wages. In 1773 both Thomas Cook and Mary Dawson
were paid 1s. per load for picking stones.37 In 1778, William Thompson
of Staffordshire paid Betty Sillito, John Dunn, William Rowley, Dolly
Matthews, Betty Baker, and Nanny Greenbrough the same rate for
reaping wheat (4 1 d. per thrave).38 Helen Speechley found that women and
men were paid the same rates for piece-work on Somerset farms.39
Equal piece-rates were observed in manufacturing as well. The 1845
report on framework knitters explicitly stated that women were
employed “at the same rate of wages” as men.40 In 1840 a Welsh weaver

Rural History Centre, Reading, BER 13/5/2. The bill is undated, but is from sometime
in the 1820s or 1830s.
Records Hitchin farm of the Ratcliffe estate. Hertfordshire Record Of¬ce D/DE E110.
Rural History Centre, Reading, micro¬lm P262.
Speechley, “Female and Child Agricultural Day Labourers,” ch. 6.
BPP 1845 (609) XV, p. 101.
Women™s wages 89

told a parliamentary committee that women and girls could weave as
well as men, and received the same rate of wages.41 Even when rates
were set by law, the piece-rates applied equally to men and women.
Female silk weavers received the same piece-rates speci¬ed for
“journeymen” in the Spital¬elds Acts. In 1811 one employer threw the
trade into confusion by refusing to pay a journeywoman the rates
speci¬ed in the Spital¬elds Acts, claiming that the law applied only to
male workers. To clear up the confusion, an amendment was added to
the Act speci¬cally stating that women were to receive the same piece-
rates as men.42
If the work was done in the worker™s home rather than the factory, the
employer often did not even know who the worker was, and paid the
head of the family for the work of the whole family. The ¬nished product
was generally brought in for payment by the head of the household, and
the employer would not know if the piece was woven by the man of the
house, by his wife or one of his children, or by any other worker. When
asked whether the handloom weavers he employed were men, women,
or children, Adam Bogle of Glasgow replied, “we do not know whether
they are children, or men or women; the work is generally brought to the
works by a man; they are generally men, and their families, and
apprentices I believe.”43 If he did not know the sex of the worker, he
could not pay a price that differed by gender.
Some historians claim that female mule-spinners in Manchester were
paid lower piece-rates than male spinners. Though I have not seen any
actual piece-rates quoted (wages are always quoted as weekly earnings),
this does appear to be true because in 1829 the male spinners urged the
female spinners to form a separate union, and promised to support their
effort to obtain the same rates as the men.44 Piece-rates that differed by
gender would certainly seem like compelling evidence of discrimination,
but in this case we do have more detailed information on the industry
that suggests that the women may not have been underpaid. The male
and female mule spinners were not really doing the same job. Men hired

BPP 1840 (220) XXIV, p. 557. 42 Pinchbeck, Women Workers, pp. 177“8.
BPP 1816 (397) III, p. 167.
R. G. Kirby and A. E. Musson, The Voice of the People: John Doherty, 1798-1854, Trade
Unionist, Radical and Factory Reformer (Manchester: Manchester University Press,
1975), p. 94. Many historians have called attention to the lower wages of female mule-
spinners. See Freifeld, “Technological Change and the ˜Self-Acting™ Mule,” p. 334,
and Paul Johnson, “Age, Gender and the Wage in Britain, 1830-1930,” in Peter
Scholliers and Leonard Schwarz, eds., Experiencing Wages: Social and Cultural Aspects of
Wage Forms in Europe since 1500 (New York: Berghahn Books, 2003), p. 230. Weekly
earnings for women were about half as much as men™s earnings, but it is unclear how
much of this was due to lower output, and how much to lower piece-rates.
90 Gender, Work and Wages in Industrial Revolution Britain

and disciplined their own assistants (piecers), while the women did not.
Huberman notes that when the cotton spinning ¬rm M™Connel and
Kennedy hired women mule-spinners, responsibility for disciplining
piecers shifted to overlookers, and the ¬rm incurred additional costs in
recruiting piecers. Problems with supervision of piecers led to increases
in the amount of cotton wasted. M™Connel and Kennedy began hiring
women spinners in 1810, and between 1809 and 1817 its wastage rates
increased 63 percent. This trend was reversed in the 1820s when the
¬rm returned to hiring male spinners.45 If women mule-spinners in fact
were equally productive and were paid less, then the ¬rm™s pro¬ts should
have increased. In fact, the ¬rm found its pro¬ts falling, in spite of the
fact that it paid the female spinners lower piece-rates. So it turns out that
in this case the difference in piece-rates was justi¬ed by differences in
Instances of women being paid lower piece-rates than men did exist,
but were not the norm. A survey of wages by Sidney Webb, while not
from the Industrial Revolution period, is instructive. Webb examined a
wide variety of industries in Britain and France at the end of the nine-
teenth century, and he found twelve industries where men and women
earned equal piece-rates and only two where women earned lower piece-
rates than men.47 For the 1750“1850 period I have found only one
example of women being paid a lower piece-rate than men for doing the
same work. In the 1840 report on handloom weavers in the west we ¬nd
an employer, Mr. Peter Payne, who claims:
A woman receives 3s. a piece less on the white work than a man, and 4s. less on
coloured. This has always been the case, and the example was set by the master
weavers. Women are not so regular in their time as men, nor so able to perform
the work in the same time. Thirty men will do as much work as forty women,
and the outlay for looms, buildings, &c. is greater for a number of females than
for the male weavers, but that this outlay for looms, buildings, &c. is not con-
sidered to be equal to the difference in pay to the same extent.48

He suggests that part of the difference is justi¬ed by the cost to the
manufacturer of slower work, but that part of it is not justi¬ed.
Of course a cloth ¬nished in three weeks is a different good from a
cloth ¬nished in four weeks, and these goods may reasonably have

Michael Huberman, Escape from the Market: Negotiating Work in Lancashire (Cambridge:
Cambridge University Press, 1996), p. 39.
Ibid., pp. 25“9.
Sidney Webb, “The Alleged Differences in the Wages Paid to Men and to Women for
Similar Work,” Economic Journal 1 (1891), 635“62. In one of the later industries, com-
positing, women were disadvantaged by legal restrictions on the hours they could work.
BPP 1840 (220) XXIV, p. 401.
Women™s wages 91

different prices. Mr. Payne was right to be concerned about the number
of weeks a worker keeps the cloth, because if the worker ties up the
material for an extra week, the employer loses one week™s worth of
interest on the value of the raw material. The interest rate faced by these
employers would have been fairly high, given the imperfections in the
capital market, and the high cost of the yarn.49 The input costs of one
piece of cloth is given in the 1840 report:
Value of the wool £17.10s.4d.
Labor £5.12s.
Materials £2.16s.10d.
Rent, wear and tear £0.18s.
The cost of the raw wool alone accounts for two-thirds of the total cost
of the cloth. The yarn given to the weaver would also include the value of
some of the labor, materials, and rent. The value of the yarn, then, was
many times greater than the weaver™s wage. Mr. Payne™s description
suggests that in this case the employer was providing buildings and
looms as well, so lower wages for women might represent implicit rent
for the capital equipment. The cost of delayed weaving was high, and
thus we should expect the employer to pay a lower price for cloth kept
out longer. A woman generally took longer than a man to weave a given
cloth, both because of her lower productivity and because she was likely
to devote less time to the work. Some idea of this time difference is given
in another report on handloom weaving, which claims that a man would
take three weeks to weave a piece of 46 yards, while his wife would take a
month. On cloth that took a man a fortnight to weave, his wife would
take two days longer.51
Completion time, however, is clearly observable and there is no reason
for the manufacturer to use sex as a signal for this quality. If the
employer was really concerned with completion time rather than the
gender of the worker, it is not clear why he would pay according to sex
and not speci¬cally according to weaving time. In fact, employers else-
where speci¬ed in their wage contracts that they would reduce the price
of a piece that was out longer than a speci¬ed time period. A “ticket”
specifying the wage contract, given out by a Carlisle manufacturer in 1838
states, “6d. per day deducted off work kept out longer than 28 days.”52

See Joel Mokyr, “Editor™s Introduction: The New Economic History and the Industrial
Revolution,” in J. Mokyr, ed., The British Industrial Revolution: An Economic Perspective
(Boulder: Westview Press, 1993), p. 109.
BPP 1840 (220) XXIV, p. 374. 51 BPP 1840 (43) XXIII, pp. 435“6.
BPP 1840 (220) XXIV, p. 597. Another example of such a ticket from Londonderry
speci¬es a 3d. per day deduction for work kept over twenty-one days.
92 Gender, Work and Wages in Industrial Revolution Britain

Employers in the west of England had no valid reason to use sex as a
signal of completion time, because time was perfectly observable; they
seem to have chosen a noisy signal (gender) over the actual variable of
interest (time). The story of paying women less because they take longer
does not hold up. We can only conclude that this practice was an example
of wage discrimination against women.
If wage discrimination exists, economic theory tells us that two forces
will be at work to eliminate it: the substitution of women for men, and the
failure of ¬rms who continue to employ men at higher rates. If women do
the same work cheaper than men, the economic incentive to employ them
in preference to men is great. Only an employer with a “taste for dis-
crimination” will not do so. Discriminating employers, however, will be
vulnerable to bankruptcy if the market is competitive. Discrimination
may exist in disequilibrium, but it will not be a dominant characteristic of
the labor market. In this case competitive forces do seem to have been
eroding the wage difference by removing men from the occupation of
handloom weaving. Anthony Austin, reporting on handloom weaving in
Somerset, Wiltshire, Devon, and Dorset, says of serge weaving:
women are employed, who will readily undertake it, at a lower price than men
receive . . . indeed, it appears to be a custom in every trade to pay women at a
lower rate than men for the same article. I have found it in the broad-cloth trade,
in the blanket trade, and in the silk-velvet trade. By this process (unless the men
consent to take the lower rate of wages) the whole of the weaving is gradually put
into the hands of women . . . and the men are compelled to seek other work.53

As economic theory tells us, and as Austin suggests, such a difference in
piece-rate wages is not an equilibrium. If women are paid less than men for
the same output, women workers will be substituted for men. Thus, wage
discrimination will lead to no men being employed in that occupation.
Male workers were aware of these economic forces, and they some-
times demanded equal wages for this reason. A ban on women workers
was preferable, but if women were to work, they must earn the same
wage as men, since this was the only way to maintain male employment.
In 1833 David Sloan, manager of the Bridgetown Mills in Calton,
Glasgow, reported:
that the women originally agreed to spin for wages at a rate one thirteenth lower
than the males, but the association having heard of their being engaged to work
had emissaries on the way on the very ¬rst day . . . that a deputation of the
association waited upon him the same day, to tell him that if they could help it,
they would not allow the women to be employed at all as spinners, but that in all

BPP 1840 (43) XXIII, p. 442.
Women™s wages 93

events they would fall on means to prevent their being employed for lower rates
than those which they had ¬xed.54

The male workers of Bridgetown Mills rioted, and the female spinners
had to be escorted to and from work. The riots ceased when the ¬rm
announced that male and female spinners would be paid at the same
rate, and the female spinners continued to work in peace. A male spinner
commented that “the chief reason was to prevent the lowering of wages,
in which the association in the end succeeded.”55 The male workers
were well aware that their jobs could not be maintained if women were
allowed to work at lower wages. The position of the Bridgetown union
was not unique; Sonya Rose has found a number of similar examples in
the second half of the nineteenth century. In hosiery factories, the men
voted for an equal piece-rate because when the rate was not equal, men
were replaced by women.56 During a strike over women carpet workers
in Kidderminster, the union said “If the looms are supposedly within the
compass of a women, let her do it and be paid like a man.”57 Since
unequal piece-rates reduced male employment, men actively opposed
such differences.
While I have found piece-rate wage discrimination in handloom
weaving in the south-west, the most common practice was to pay men
and women the same piece-rate. In all the instances where men and
women were paid the same piece-rate, we can con¬dently say that there
was no wage discrimination. Any differences in weekly earnings arose
because of differences in time devoted to work, or from differences in
productivity. Unfortunately, it is more dif¬cult to identify wage dis-
crimination in time-rate wages, but even there differences in product-
ivity seem to have been the main cause of the wage gap.

II. Interpreting time-rate wages
While time-rate wages must be examined separately from piece-rate
wages, I conclude that the reasons for the wage gap in time-rate wages
were essentially the same as for the earnings gap observed where piece-
rate wages were paid. The time-rate wages in Table 2.1 suggest a wage
ratio of between one-third and three-fourths. However, a portion of this
wage gap is the result of measurement error. Women often worked fewer
hours per day than men, so the ratio of daily wages understates the ratio

BPP 1833 (450) XX, A1, p. 84. The self-actor had just been invented, but it was
probably not in use yet.
Ibid., p. 85. 56 Rose, “Gender Segregation.”
Rose, “Gender Antagonism,” p. 200.
94 Gender, Work and Wages in Industrial Revolution Britain

of hourly wages. In other cases, the failure to include in-kind payments
led to an underestimate of the wage ratio. The bondager system used in
Northumberland also led to quoted wages which understated the real
compensation to women workers. If these biases are taken into account,
the wage ratio looks closer to two-thirds than to one-third. Even if these
measurement errors are accounted for, a substantial wage gap remains.
Some historians interpret this wage gap as evidence of discrimination,
but the evidence suggests that women™s relative productivity was at least
approximately equal to their relative wage. This section will explore the
reasons for the wage gap in time-rate wages and argue that wage dis-
crimination was not an important cause of the wage gap.
In agriculture, female day-laborers frequently worked shorter days
than male wage-laborers, a fact that accounts for some (though not all)
of the wage difference.58 In 1843 a farmer from Dorset reported why
he did not hire more women: “I consider their labor dear; they want 8d.
a day, and they don™t come till nine, and are away again at ¬ve.”59 Men
generally worked 6 a.m. to 6 p.m.60 so if women worked eight hours
a day, they worked only two-thirds as many hours as men. In this case,
a daily-wage ratio of 0.5 would imply an hourly-wage ratio of 0.75. If
such a difference in hours was widespread, correcting for hours reduces
the wage gap considerably. In fact, it does appear that women commonly
worked fewer hours than men. The 1843 parliamentary report Women
and Children in Agriculture shows that it was a common practice for
women to start an hour later than men, so they could get breakfast for their
families, and to return home sooner in the evening.61 The same is found
in the “Rural Queries,” a questionnaire sent out by the Poor Law Com-
mittee in 1833. In one Cornwall parish we ¬nd that women in agriculture
worked from 8 a.m. to 6 p.m., and Leicester women are said to work only
eight or nine hours a day.62 Gilboy ¬nds the same thing in the eighteenth
century “ the women began work at 8 a.m. rather than 6 a.m.63 Differ-
ences in hours worked were common, so earnings ratios will understate
the wage ratios. The 1843 report on Women and Children in Agriculture

Gielgud, “Nineteenth Century Farm Women,” p. 102, suggested that difference in
hours worked might explain the wage gap, though she does not attempt to measure the
difference in hours worked.
BPP 1843 (510) XII, p. 88.
See, for example, Eden, State of the Poor, vol. III, p. 876; BPP 1824 (392) VI, p. 22;
BPP 1843 (510) XII, pp. 120, 169“71.
BPP 1843 (510) XII.
BPP 1834 (44) XXX, St. Anthony of the East, Cornwall, and Sheepy Magna,
Elizabeth Waterman Gilboy, “Labour at Thornborough: An Eighteenth Century
Estate,” Economic History Review 3 (1932), pp. 388“98.
Women™s wages 95

provides 109 observations on daily hours of work for women. Women™s
hours ranged from eight to twelve hours per day, and averaged 9.66
hours.64 If men worked twelve-hour days, then equal hourly wages would
imply a daily-wage ratio of 0.805, and if a woman™s daily wage was 40
percent of a man™s daily wage, then her implicit hourly wage was 50
percent of a man™s. Thus daily wages reported in Table 2.1 overstate the
size of the wage gap.
In the case of washerwomen, on the other hand, failing to control for
the number of hours worked in a day makes the female wage appear
much too large. Wages for women going out washing by the day were as
high as 2s.6d. a day in London,65 but were lower in areas farther from
London. The “Rural Queries” record of 1834 gives wages ranging from
6d. to 3s. per day, but 1s. seems to have been more typical. Some of
these wages appear to be very high relative to the wages of women in
other occupations, but most of the apparent difference is a result of the
fact that these women worked long hours, perhaps even double the
hours of other workers. A day™s work of washing might begin at 1:00 a.m.
and continue to the next evening.66 These long days of washing were
recognized as “a day and a half™s work.”67 In London in 1839 women
could earn 2s.6d. for nearly twenty hours™ work, which would put the
wage at about 1½ pence an hour.68 This hourly wage is still higher than
agricultural labor, which often paid about 1d. an hour, but the difference
is not nearly as great as it ¬rst appears, and we would expect wages to be
higher in London than in the country.
In-kind payments to servants also lead to an overstatement of the
wage gap for some workers. Many workers, particularly in domestic
service and agriculture, were employed as live-in servants. These
workers were given room and board in addition to a cash wage, and the
value of these in-kind payments was frequently greater than the cash
wage. Ignoring the in-kind portion of the wage will bias the wage ratio
down. Table 2.3 shows wages paid to agricultural servants. Two wage
ratios are presented “ the ¬rst is the ratio of money wages, and the
second is the ratio of full wages, with the value of board included.69
The value of board was a large portion of the wage, and including it has a
large impact on the wage ratio. Since women require fewer calories than

Women and Children in Agriculture. Women™s hours of work are inclusive of breaks for
meals, as is the twelve-hour day for men.
Hill, Women, Work, and Sexual Politics, p. 159. 66 Ibid., p. 158.
George, London Life in the Eighteenth Century, p. 20. 68 Ibid., p. 208.
Arthur Young™s estimate includes the value of room, but the others do not. This
omission is less serious than the omission of board because the value of the room was
small compared to the value of board.
Table 2.3. Servants™ wages (£ per year)

Male wages Female wages

Employer Servant Money Board Money Board Ratio1 Ratio2 Src

Robert Loder, Harwell, Berkshire
1613 Robert Earnold, carter 3 10.25 a
Dick, shepherd 2 10.25 a
Johan C., maid 2.38 10.25 0.79 0.95 a
Alice K., maid 1.5 10.25 0.75 0.96 a
1614 Robert Andrewes, carter 3.33 7 a
Johan Colle, maid 2.35 7 0.71 0.91 a
Mary, maid 2 7 0.60 0.87 a
1615 Ned, carter 3 11.83 a
Dick, shepherd 2.4 11.83 a
Mary 2 11.83 0.67 0.93 a
Margaret 2.25 11.83 0.94 0.99 a
1620 London Baker™s 6.5 10.4 2.17 10.4 0.33 0.74 b
journeyman & maid
1761 Bury, Lancashire, agric. 6.5 3 0.46 c
Danby More skilled 15.0 9.1 5.5 5.5 0.37 0.46 d
Kabers More skilled 9.0 9.1 3.0 6.1 0.33 0.50 d
Less skilled 5.0 9.1 2.25 6.1 0.45 0.59 d
Ormskirk, More skilled 7.0 9.0 3.0 6.0 0.43 0.56 d
Shenstone More skilled 11.0 9.0 4.0 6.0 0.36 0.50 d
Less skilled 6.5 9.0 2.5 6.0 0.38 0.55 d
Hagley More skilled 10.0 10.0 3.5 6.7 0.35 0.51 d
Less skilled 6.75 10.0 2.75 6.7 0.41 0.56 d
Bends- More skilled 10.0 12.0 4.0 8.0 0.40 0.55 d
Less skilled 8.0 12.0 2.75 8.0 0.34 0.54 d
1791 Bury, Lancashire, agric. 9.45 4.5 0.48 c
1795 Cumberland, common 9 3.25 0.36 c
1795 Northamptonshire, age 20 7.5 3 0.40 c
1821 William Stickney, Yorkshire 16.5 27 7 18 0.42 0.57 e

Ratio1 ¼ ratio of money wages.
Ratio2 ¼ ratio of full wage, including in kind payments.
a. Fussel, Robert Loder™s Farm Accounts. Where there are multiple wages for each sex,
I compare the highest-wage females to the highest-wage males.
b. Bakers of London, quoted in S. Paul Garner, Evolution of Cost Accounting (University of
Alabama Press, 1954), p. 32.
c. Eden, State of the Poor, vol. II, p. 294.
d. Young, Northern Tour. Danby gives an estimate for a maid™s board; the others are two-
thirds of the male value, which Young suggests is the correct ratio, in vol. III, p. 288. “More
skilled” servants are the headman and a dairy maid.
e. BPP 1821 (668) IX.
Women™s wages 97

men (about 73 percent of what a man requires), the cost of board for a
woman was less than for a man.70 However, since the ratio of cash wages
was well below half and thus lower than the ratio of food intake,
including board in the calculation of wages will still raise the wage ratio.
The cases presented in Table 2.3 demonstrate the error introduced by
using only cash wages. In the seventeenth century, Robert Loder of
Berkshire calculated a value for board by dividing total expenditures on
food by the number of people at his table. Including this value in the
wages of his servants raises the female“male wage ratio. This calculation
somewhat overstates the ratio, however, since the men would have eaten
more than the women. In fact, if women ate only 73 percent as much as
men, then adding board would not signi¬cantly alter the wage ratio. In
the eighteenth century, however, the ratio of cash wages was much
lower, and including the value of board increases the wage ratio even if
women received less than men. Arthur Young, on his northern tour,
asked a few farmers what value they gave to a man™s board, lodging, and
washing. The costs of lodging and washing were relatively small, so most
of this value was the cost of board. Young assumed that a woman
received two-thirds as much as a man in in-kind payments, and this ratio
is used to value these payments in Table 2.3. Even if the in-kind pay-
ments received by women were only two-thirds as much as those
received by men, including these payments increases the wage ratios
from a little more than a third to over a half.71 The same pattern is
observed for Yorkshire in 1821; here the wage ratio rises from 0.42 to
0.57 when the value of board is included. The wage gap remains sub-
stantial, but the difference is not as extreme as it ¬rst appears. If we wish
to discuss the causes of the wage gap, we must ¬rst ¬nd its correct size,
which requires including the value of in-kind payments.
A different kind of mis-measurement occurred in northern counties
that used the bondager system. Male and female laborers were hired
together, and the compensation package overstated the portion earned
by the male, and understated the portion earned by the female. In this
system the male laborer, called the “hind,” was required to provide a
woman worker, the “bondager,” to work whenever the employer desired
at a speci¬ed rate. Usually this woman was a family member, but if the

Geert Bekaert, “Caloric Consumption in Industrializing Belgium,” Journal of Economic
History, 51 (1991), p. 638. Ogilvie ¬nds that in Germany female agricultural laborers™
meals were valued at 67 to 79 percent as much as male laborers™ meals. Sheilagh
Ogilvie, A Bitter Living: Women, Markets, and Social Capital in Early Modern Germany
(Oxford: Oxford University Press, 2003), p. 287.
I use the ratio of two-thirds because this is the ratio that Young used to calculate the
value of a woman™s board. Young, Northern Tour, vol. III, p. 288.
98 Gender, Work and Wages in Industrial Revolution Britain

hind could not provide a bondager from his own family, he had to hire
one. The bondager was paid a lower daily rate than other women
workers.72 For example, a Northumberland farmer paid Jane Thomp-
son, a bondager, 10d. per day, but paid Isabella Thompson, who was
not a bondager, 1s. per day.73 The farmer was able to pay the bondager
2d. less than the market wage because John Thompson™s contract spe-
ci¬ed that, as a condition of employment, he must provide a bondager.
Hinds complained about the bondager system because if they had no
suitable relative to provide as a bondager, hiring a bondager cost them
money. When hiring a bondager, the hind had to pay the bondager an
annual wage and provide her food for the year as well. The hind received
from the farmer the daily wage for the bondager™s work, but this was less
than he spent to hire and feed the bondager. The cash wage paid to the
bondager was nearly as much as she earned from the farmer, and the
hind had to provide her food for the year as well.74 When the bondager
was a family member the hind did not pay the bondager a wage, but it
was still true that part of the hind™s wage was compensation for work of
the bondager. This system disguised part of the female bondager™s
earnings as compensation of the male hind.
Women™s wages were lower than men™s. However, the available wage
quotes often overstate the gap. Correcting for measurement error biases
can increase our estimate of the wage ratio substantially. Initially wo-
men™s wages appear to be between one-third and one-half of men™s
wages, but correcting for measurement error suggests that women™s
wages were closer to one-half to two-thirds of male wages. The wage gap
does not disappear, and there still remains the question of why this gap
occurred. As discussed above, the wage gap has been interpreted either
as evidence of women™s lesser productivity, or as evidence that women™s
wages were set by custom rather than the market. Evidence on the size of
the wage gap is abundant, but this evidence cannot distinguish between
the two theories because both theories suggest that women™s wages
would be lower than men™s. To test the assumption that the wage gap
matches women™s lower productivity we need independent evidence on
male and female productivity. Unfortunately evidence on productivity is
much scarcer than evidence on wages. However, since we do not want to
rely on prejudice to answer this question, we must look at what limited
evidence is available. The existing evidence suggests that, at least for

Gielgud, “Nineteenth Century Farm Women,” p. 145.
The farmer is Mr. Hindmarsh, “an extensive farmer in the neighbourbood of Wooler.”
Women and Children in Agriculture, p. 297.
Gielgud, “Nineteenth Century Farm Women,” p. 330.
Women™s wages 99

manual workers, women were not as productive as men, and that the
wage gap may indeed have been justi¬ed by productivity differences.
Quotes from contemporaries have been used to evaluate relative
female productivity, but if we look at a variety of such quotes we can see
that they con¬‚ict with each other. A farm bailiff from Kent estimated
that women and boys were less productive than men in reaping: “The
boys begin to do this at about 12 years of age; a boy in this time of life
will reap about a quarter of an acre in two days, while a man would be
reaping three-quarters of an acre in the same time, and a woman half an
acre, if she worked as many hours as the man.”75 This implies that a boy
of age 12 was one-third as productive as a man, and a woman was two-
thirds as productive as a man. Another nineteenth-century author gives a
contradictory assessment. Henry Stephens claimed that a woman could
reap as much as a man: “The reapers may all be men, or all women, the
women being able to cut down as much as the men.”76 Frederick Eden
claimed that in Brom¬eld
The wages of men-servants employed in husbandry, who are hired from half-
year to half-year, are from 9 to 12 guineas a year; whilst women, who here do a
large portion of the work of the farm, with dif¬culty get half as much. It is not
easy to account for so striking an inequality; and still less easy to justify it.77

While Eden was unable to explain the wage differences he observed,
individuals interviewed by Alfred Austin and Mr. Vaughn for the 1843
report on Women and Children in Agriculture were less mysti¬ed by the
wage differences. One Dorset farmer referred to women™s shorter hours
of work (see p. 94), but others suggested that women accomplished less
work than men when working at the same tasks. When asked if women
and men worked together in the ¬elds, a Wiltshire farm manager replied,
“The women generally work together; they don™t get on so fast as the men
in their work, particularly in reaping and hoeing turnips.”78 A Surrey
landlord claimed that, in poling hops, “The value of the woman™s labor
is rather more than a third of the man™s.”79 Joseph Henley noted that
women workers in Northumberland did various tasks, including “in some
instances forking (pitching) and loading hay or corn, though when such is
the case two women are put to the work of one man.”80 Robert Loder

Women and Children in Agriculture, p. 185.
Henry Stephens, The Book of the Farm, 2nd edn (Edinburgh and London: William
Blackwood and Sons, 1845), vol. II, p. 331.
Eden, State of the Poor, vol. II, p. 47. 78 Women and Children in Agriculture, p. 62.
Ibid., p. 198.
Report of Joseph Henley, 1867, quoted by Gielgud, “Nineteenth Century Farm Women,”
p. 11.
100 Gender, Work and Wages in Industrial Revolution Britain

found women less productive than men in the early seventeenth century;
he had hired women to harvest cherries, but noted that “I think it were a
better course to hire men, for they would doe twice so much I think.”81
Given these con¬‚icting statements by contemporary observers, whom
should we trust? The best way to resolve the question of who was right is
to look for evidence that is not the expression of someone™s opinion, but
is direct evidence from output.
One of the simplest pieces of evidence for differences in productivity is
earnings differences among workers paid piece-rate wages. When
workers were paid piece-rates, their earnings depended directly on their
output, and any differences in productivity would translate directly into
differences in earnings. Claudia Goldin has noted that, in nineteenth-
century manufacturing in the US, when men and women worked together
“males earned 25% more than females, even when the work was identical,
the piece-rate was the same, and both worked for the same ¬rm.”82 If
these workers were paid the same rate per unit of output, the differences
in their earnings must have come from differences in output; Goldin™s
¬ndings suggest that women were 80 percent as productive as men.
Direct measures of cotton picked by individual slaves in the US South
suggest that a woman picked less cotton than a man in a day. Olmstead
and Rhode collected over 600,000 observations of the weight of cotton
picked in a day by individual workers from 113 plantations in the period
1801“62. A girl could pick as much as a boy until about age 15, at which
point a gender gap began to emerge. For prime-age adults, a man picked
about 18 percent more cotton per day than a woman.83 This may seem
like a relatively small difference, but cotton picking was not a particularly
strength-intensive task where we would expect to ¬nd large gender
differences in productivity. Goldin and Sokoloff suggest that manufac-
turing emerged in the US North because women were relatively less
productive in northern agriculture than in southern agriculture, and
therefore had a lower opportunity cost. The South grew cotton and
tobacco, which could make better use of female workers. They note that
a woman™s disadvantage relative to a man was comparatively small in
cotton picking: “Even though males (over age 16) had an absolute
advantage over females in cotton picking, females had a comparative
advantage and therefore picked a greater percentage of the man-days

G. E. Fussell, ed., Robert Loder™s Farm Accounts, 1610“1620, Camden Society, Third
Series, vol. 53 (London: Royal Historical Society, 1936), p. 148.
Goldin, Understanding the Gender Gap, p. 104.
Alan Olmstead and Paul Rhode, “ ˜Wait a Cotton Pickin™ Minute!™ A New View of
Slave Productivity,” presented at the Economic History Association Annual Meeting in
Pittsburgh, Sept. 17, 2006.
Women™s wages 101

allocated to that task.”84 If we could directly measure output in other
agricultural tasks, we would expect to ¬nd a larger gender gap.
Slave prices are consistent with Olmstead and Rhode™s measures of
productivity. Slaves were sold in a competitive market, and the price
paid for a slave should represent the current value of the slave™s future
output. If slave prices included any value put on the sexual services or
children produced by female slaves, this would increase the value of
women relative to men. In spite of their potential reproductive bene¬ts,
we ¬nd that slave women had lower prices than slave men. Girls had
higher prices than boys, but after age 16 the prices of male slaves rose
above those of female slaves. At age 32 a man cost 18 percent more than
a woman.85 Slave owners were willing to pay more for male slaves
because males were more productive.
Another way to measure women™s relative productivity is to estimate
production functions using historical data. This method consistently
¬nds that women were less productive than men in agriculture. Using
US census data, Craig and Field-Hendrey ¬nd that women were about
60 percent as productive as men in agriculture.86 Toman estimates the
marginal product of slaves and ¬nds that the marginal product of female
slaves was 40 percent of male productivity in the task system, and 60
percent of male productivity in the gang system.87 The same seems to be
true in other areas of the world as well. Benjamin and Brandt use a 1936
household survey in China to estimate the contribution of men and
women to family income in general and crop income speci¬cally; they
¬nd that women contributed 62 percent as much as men to farm pro-
duction.88 Women are also less productive than men in agriculture in
developing countries today; Jacoby found that women were 46 percent
as productive as men in Peruvian agriculture in the 1980s.89

Goldin and Sokoloff, “The Relative Productivity Hypothesis,” p. 473.
Laurence Kotlikoff, “Quantitative Description of the New Orleans Slave Market,
1804 to 1862,” in R. W. Fogel and S. L. Engerman, eds., Without Consent or Contract:
The Rise and Fall of American Slavery, Markets and Production: Technical Papers, vol. I
(New York: Norton, 1989), pp. 42“5.
Lee A. Craig and Elizabeth Field-Hendrey, “Industrialization and the Earnings Gap:
Regional and Sectoral Tests of the Goldin“Sokoloff Hypothesis,” Explorations in
Economic History 30 (1993), pp. 60“80.
J. T. Toman, “The Gang System and Comparative Advantage,” Explorations in
Economic History 42 (2005), p. 320.
Dwayne Benjamin and Loren Brandt, “Markets, Discrimination, and the Economic
Contribution of Women in China: Historical Evidence,” Economic Development and
Cultural Change 44 (1995), pp. 63“104.
Hanan Jacoby, “Productivity of Men and Women and the Sexual Division of Labour in
Peasant Agriculture of the Peruvian Sierra,” Journal of Development Economics 37 (1992),
pp. 265“87.
102 Gender, Work and Wages in Industrial Revolution Britain

Studies that estimate production functions for manufacturing also
consistently ¬nd that women are less productive than men. Craig and
Field-Hendrey estimate that women were 40 to 50 percent as productive
as men in US manufacturing in 1860.90 Cox and Nye use data on
nineteenth-century French manufacturing ¬rms to estimate the mar-
ginal product of male and female workers and ¬nd productivity ratios
ranging from 0.37 to 0.63. When they test for wage discrimination, they
¬nd no evidence of it.91 Studies of late twentieth-century manufacturing
¬nd a smaller gap, but still conclude that women were less productive
than men. Haegeland and Klette ¬nd that women were 83 percent as
productive as men in Norwegian manufacturing, while Hellerstein,
Neumark, and Troske ¬nd that women were 84 percent as productive as
men in the US.92
Evidence from production functions suggests that, in agriculture and
manufacturing, differences in productivity were large enough to explain
the portion of the wage gap that is not explained by measurement error.
Section III examines possible reasons for these productivity differences.
Before moving on, though, it is important to note that not all wage
differences could be explained by productivity differences. In less
competitive areas of the economy wage discrimination could persist. Of
all the wages in Table 2.1, the most likely candidates for wage discrim-
ination are the salaried professions. While the lower salaries of school-
mistresses may have resulted from their lower skills, those of workhouse
matrons do not seem to be justi¬ed by productivity differences.
Teachers may have experienced employer discrimination, but it is also
possible that the wage differences re¬‚ected productivity. When hired by
schools, female teachers earned lower salaries than male teachers. The
examples in Table 2.1 suggest that schoolmistresses earned one-half to
two-thirds the salaries of schoolmasters. However, some of this wage gap
re¬‚ects the fact that women teachers had fewer skills and thus taught
fewer subjects. Schoolmistresses were generally not expected to teach

Craig and Field-Hendrey, “Industrialization and the Earnings Gap.”
Donald Cox and John Vincent Nye, “Male“Female Wage Discrimination in
Nineteenth-Century France,” Journal of Economic History 49 (1989), pp. 903“20.
Torbjorn Haegeland and Tor Jakob Klette, “Do Higher Wages Re¬‚ect Higher
Productivity? Education, Gender and Experience Premiums in a Matched Plant-
Worker Data Set,” in J. Haltwanger, J. R. Lane, J. Spletzer, J. Theeuwes, and K. Troske,
eds., The Creation and Analysis of Employer“Employee Matched Data (Amsterdam: Elsevier,
1999), pp. 231“59. Judith Hellerstein, David Neumark, and Kenneth Troske, “Wages,
Productivity, and Worker Characteristics: Evidence from Plant-Level Production
Functions and Wage Equations,” Journal of Labor Economics 17 (1999), pp. 409“46.
Hellerstein, Neumark, and Troske ¬nd evidence that women were underpaid, but
Haegeland and Klette do not.
Women™s wages 103

writing or arithmetic. One set of rules for a charity school required the
master to be “One who can write a good hand, and who understands the
grounds of arithmetic,” but did not require the same of a schoolmis-
tress.93 At one Lancashire school the girls were taught by a schoolmis-
tress, except in writing and arithmetic, which the schoolmaster taught.94
Women were less likely than men to be able to write or do arithmetic and
therefore they generally did not teach these subjects. Without measures
of output it is dif¬cult to say whether the difference in skills justi¬ed the
difference in salaries, but it is at least possible.
The differences in the salaries of workhouse administrators are more
dif¬cult to explain. The position did not require extensive education. If
anything, women were better trained in the skills required to run a
household. The master and matron of the workhouse had similar
responsibilities. At Bristol, both the master and the matron had money
on hand at the end of the ¬scal year, suggesting that both had ¬nancial
responsibilities.95 Women seem to have been paid less for doing the
same job. Wage discrimination could exist for workhouse administrators
because the market was not competitive. A workhouse did not go
bankrupt if it lost money. The gender difference in salaries did not even
result in an all-female workforce because each workhouse hired exactly
one master and one matron, the former to watch over the men, and the
latter to watch over the women. Because the workhouse inmates were
strictly segregated by sex, the parish could not substitute a matron for a
master. Segregation of inmates, and the fact that there was no competi-
tion between workhouses, allowed the persistence of wage discrimination.

III. Productivity differences
The evidence presented above suggests that in the past women were less
productive than men, but the question of why they were less productive
still remains. The argument that women™s wages were commensurate
with their productivity is convincing only if there are plausible reasons
for women™s lower productivity. Two important reasons for such
productivity differences were strength and human capital. Strength was
an important component of productivity during the Industrial Revolu-
tion, and since women have on average less strength than men, they

Asher Tropp, The School Teachers: The Growth of the Teaching Profession in England and
Wales from 1800 to the Present Day (London: William Heinemann, 1957), p. 6.
John Roach, A History of Secondary Education in England, 1800“1870 (London:
Longman, 1986), p. 15.
Eden, State of the Poor, vol. II, p. 198. However, the master had more money in his
possession than the matron.
104 Gender, Work and Wages in Industrial Revolution Britain

were less productive. Women also had less human capital. Since human
capital is chosen rather than biologically determined, identifying dis-
crimination becomes more complicated in this case because human
capital differences may be both a result of and a cause of wage differ-
ences. While strength differences are largely exogenous, skill differences
are largely endogenous.

A. Strength
The impact of strength on sex differences in employment has been noted
by many historians. Merry Wiesner claims that the gender division of
labor in agriculture in the early modern period was partly due to dif-
ferences in physical strength, “with men generally doing tasks that
required a great deal of upper-body strength, such as cutting grain with a
scythe.”96 Judy Gielgud notes that “Individual women could and did use
a scythe, but it was too heavy an implement for them to use all day
keeping level with the other mowers, as was essential at harvest” and
that, more generally, “a man™s strength might enable him to accomplish
more of a given task than could a woman in the same time, where both
were working at full stretch.”97 Edward Shorter notes that “spading and
ploughing the ¬elds was too much for women to manage as a rule “
great strength being necessary to maneuver a Norfolk plow behind a
team of percherons.”98 Joan Lane suggests that strength requirements
in¬‚uenced occupational sorting: “Older boys were a minority in textile
factories because they could work in trades requiring physical strength.”99
More generally, Brian Harrison suggests that the gender division of labor:
is older by far than the “capitalism” to which it is sometimes ascribed. It occurs
wherever manual labor is at a premium, and re¬‚ects the fact that on average men
surpass women in sheer muscle-power. Whenever heavy labor was involved, the
sexes had rarely worked together in the past.100

Sometimes strength is included as one factor among many. Mary
Freifeld™s story about the male domination of mule-spinning faults the
male spinners™ union for excluding women following the adoption of the

Wiesner, Women and Gender in Early Modern Europe, p. 106.
Gielgud, “Nineteenth Century Farm Women,” pp. 67“8, 85.
Edward Shorter, “Women™s Work: What Difference Did Capitalism Make?” Theory
and Society 3 (1976), p. 517.
Joan Lane, Apprenticeship in England, 1600“1914 (Boulder: Westview Press, 1996),
p. 15.
Brian Harrison, “Class and Gender in Modern British Labor History,” Past and Present
124 (1989), p. 125.
Women™s wages 105

self-actor in the 1840s, but blames the strength requirements of the
machines for pushing out women in the 1830s.101 Pamela Sharpe
blames both strength and guild restrictions for keeping women out of
wool-combing.102 Sheilagh Ogilvie notes that in Germany gender dif-
ferences in strength were “one factor in¬‚uencing women™s choice of
work,” but that this in¬‚uence was marginal because of “countervailing
institutional in¬‚uences, such as guild rules excluding females from
sedentary industrial pursuits, thereby pushing them into farmwork and
However, while strength is sometimes offered as a possible explan-
ation of the division of labor, often it is either ignored, or discussed but
ultimately rejected. Elizabeth Roberts notes that strength was sometimes
used as justi¬cation for men™s higher wages, but she rejects this
explanation, arguing that women sometimes did heavy labor, and that
“In some cases these assumptions appear to have been based on gender
stereotyping rather than on reality.”104 Deborah Simonton discusses the
role of strength in determining the gender division of agricultural work,
but argues that strength is not suf¬cient to explain the results observed,
since “the persistence of woman as ˜the weaker vessel™ was as much an
ideological construction as it was physical.” Simonton concludes that
custom and gender roles were “instrumental” in determining who did
what, not biological strength.105 While Sandy Bardsley suggests that
older men may have earned lower wages because they were “less capable
of hard physical labour,” she rejects strength as an explanation of gender
differences and suggests that “social conventions” rather than physical
strength prevented women from mowing with the scythe. Bardsley even
warns the reader against the “dangers in assuming that physical strength,
rather than gender, determined division of labor in the late medieval
economy.”106 Gay Gullickson also doubts that differences in strength
between the genders were large or important.
Whether women were signi¬cantly weaker than men in the early nineteenth
century is debatable, however. Studies of women™s farm labor demonstrate that
rural women were as accustomed to strenuous farm work as men were, and the
size and strength differences between the sexes were probably not large. In fact,
it seems more likely that the paeans to male strength (and intelligence) which
began to appear in the nineteenth century are more a re¬‚ection of male

Freifeld, “Technological Change and the ˜Self-Acting™ Mule.”
Sharpe, Adapting to Capitalism, p. 24. 103 Ogilvie, A Bitter Living, pp. 286, 326.
Roberts, Women™s Work, p. 14. 105 Simonton, European Women™s Work, pp. 31“4.
Sandy Bardsley, “Women”s Work Reconsidered: Gender and Wage Differentiation in
Late Medieval England,” Past and Present 165 (1999), p. 21 and footnote 20 on p. 11.
106 Gender, Work and Wages in Industrial Revolution Britain

psychological distress over the entry of women into weaving and knitting then
they are evidence of women™s inferiority.107

By referring to claims that have been made in the past about women™s
lesser intelligence, Gullickson hopes to convince us that differences in
strength were equally ¬ctional. However, strength and intelligence are
very different things, and there is a great deal of evidence that gender
differences in strength are very real and very large.
Because the miracles of modern technology have made our own lives
so easy, we can underestimate the importance of strength in determining
wages in the past. For most of history, manual occupations dominated,
and in manual occupations biological differences in strength matter.
Strength was a scarce factor of production, and was rewarded in the
market. Women could do, and did do, physically demanding jobs, but as
long as men could produce more output per day, men would earn more.
The ubiquity of the wage gap across time and place makes sense if it
is not an arbitrary difference created by society, but a re¬‚ection of
women™s lower productivity in manual labor. This section will present
evidence establishing that the difference in strength between the sexes is
large, and will argue that this difference in strength led to differences in
Numerous physiological studies have measured various kinds of
strength, including arm, leg, and hand grip strength. Adult women
clearly have less strength than adult men. Figure 2.3 shows relative
female strength from two studies of adults. Strength is originally meas-
ured as torque or force exerted, but the results are presented here as
ratios of female to male strength. The study by Lindle et al. examines the
leg strength of 654 individuals, and the study by Lynch et al. examines
both leg and arm strength of 703 individuals.108 Women have 46 percent
as much arm strength as men at age 20, and their relative position
increases with age as male arm strength deteriorates. Women™s relative leg
strength is fairly constant between ages 20 and 75, at about 60 percent of
male leg strength. This strength gap between the sexes appears during
adolescence. Figure 2.4 shows relative strength in a number of activities
for teenage girls and boys. At ages 8“12 girls are only slightly behind

Gay Gullickson, “Love and Power in the Proto-Industrial Family,” in Maxine Berg,
ed., Markets and Manufacture in Early Industrial Europe (London: Routledge, 1991),
pp. 218“19.
R. S. Lindle et al., “Age and Gender Comparisons of Muscle Strength in 654 Women
and Men aged 20“93,” Journal of Applied Physiology 83 (1997), pp. 1581“7; N. A.
Lynch et al., “Muscle Quality. I. Age-associated Differences between Arm and Leg
Muscle Groups,” Journal of Applied Physiology 86 (1999), pp. 188“94.
Women™s wages 107
0.9 Lynch, leg
Lynch, arm
Female/Male Strength Ratio








20 30 40 50 60 70 80

Figure 2.3 Female“male strength ratios: adults
Sources: Lindle, “Age and gender comparisons,” Lynch, “Muscle quality.”

Female/Male Strength Ratio






6 8 10 12 14 16 18 20
Elbow flexion Elbow extension Knee extension
Leg lift Handgrip force
Figure 2.4 Female“male strength ratios: teens
Source: Roy Shephard, Physical Activity and Growth (Chicago: Year Book
Medical Publishers, 1982), Table 5“11.
108 Gender, Work and Wages in Industrial Revolution Britain

boys; female strength by various measures ranges from 86 to 99 percent
of male strength. After age 12, though, boys rapidly pull ahead and the
ratio drops rapidly to about 60 percent.
Part of the male advantage is due to the fact that male bodies are
larger. Men are taller than women and weigh more. Throughout the
world, men are about 7 percent taller than women.109 Longer bones give
men better leverage, so that the same muscle movement will do more


. 4
( 14)