. 6
( 14)


Thompson, Appeal of One Half of the Human Race, p. x.
Barbara Bergmann, “The Effect on White Incomes of Discrimination on Employment,”
Journal of Political Economy 79 (1971), pp. 294“313, and “Occupational Segregation,
Wages and Pro¬ts when Employers Discriminate by Race or Sex,” Eastern Economic
Journal 1 (1974), pp. 103“10.

Explaining occupational sorting 137

Bergmann™s crowding model assumes that workers are paid wages
equal to their marginal product, and thus is consistent with the results
of the last chapter.
This chapter will present alternative models of occupational segre-
gation based on comparative advantage. In the ¬rst section I assume
that, owing to their lesser strength, women were less productive than
men in some occupations, and then ask what division of labor a com-
petitive market would produce. These models, like Bergmann™s
crowding model, predict occupational segregation by gender. Unlike
Bergmann™s model, however, the occupational segregation results in a
smaller wage gap than would have been the case without such segre-
gation. In Bergmann™s model, occupational segregation is harmful to
women, reducing their wages below what they would have been without
segregation. In the models presented here, occupational sorting is
bene¬cial to women, saving them from feeling the full force of their
lower productivity. Women™s role in child-bearing may also lead them to
choose different occupations than men. The second section of this
chapter will examine the implications of women™s role in child-rearing
and suggest that women were more likely to choose occupations in
cottage industry because these occupations were more easily combined
with their child-care responsibilities.
The explanations for occupational segregation provided in this
chapter are not meant to apply to all occupations, only to occupations
without barriers to entry that blocked competition. Later chapters
(5 and 6) will address less competitive occupations with barriers to entry
and will explain why circumstances were different in those occupations.
Although such a split at ¬rst seems unproductive and even misleading in
a study discussing occupational segregation, I believe that in this case
such a distinction allows a better description of the complex Industrial
Revolution labor market. In more competitive parts of the labor market,
where jobs were relatively easily entered, the gender division of labor was
one that promoted ef¬ciency. In other areas, where it was possible to
monopolize occupations, men found it to their advantage to create
arti¬cial barriers to women™s employment, and both women and the
ef¬ciency of the economy were reduced. One explanation for occupa-
tional segregation does not suf¬ce for the whole labor market “ in certain
areas women were excluded from male occupations, but in other areas
they competed with men. The claim that uni¬es these two strains is that
women bene¬ted from competition. Where the labor market was com-
petitive, occupational sorting reduced the wage gap. Women only suf-
fered from gender discrimination in those parts of the labor market that
were not suf¬ciently competitive.
138 Gender, Work and Wages in Industrial Revolution Britain

The gender division of labor among workers in openly competitive
industries seems to have been a response to two basic gender differences:
differences in strength, and differences in child-rearing roles. Both of
these differences had profound implications for labor, and both have at
least some biological basis. Social institutions were important for
determining child-rearing roles, but at least two parts of child-rearing,
pregnancy and breast-feeding, are biologically female. Differences in
strength and child-bearing are the only two biological differences
between the sexes that I assume; nowhere in this analysis do I assume
that women are any different from men in either intellectual ability or
personality. The remainder of this chapter discusses how the market™s
response to these two biological differences created the gender division
of labor that we observe in much of the labor market.

I. Occupational sorting based on strength
In the last chapter we saw that sex differences in strength caused sub-
stantial differences in productivity. Having recognized these differences
in productivity, we can move on to discuss whether the observed division
of labor between the sexes can be explained by gender differences in
strength. This section will show that the simple fact that men are gen-
erally stronger than women is suf¬cient to explain much of the observed
division of labor between the sexes. It will also show that this division of
labor was bene¬cial to women in the sense that it allowed them to earn
higher wages than they would have earned without sorting. I will ¬rst
present models of sorting, and then argue that these models are good
descriptions of the competitive portions of the labor market.

A. Models of sorting
In his classic paper, Roy claimed that workers are sorted into occupa-
tions according to comparative advantage. Workers differ by ability, and
those with more ability are sorted into those occupations where prod-
uctivity is most responsive to ability. He offered this model as an
argument against the view that “the distribution of incomes is an arbi-
trary one that has developed by the process of historical accident.”3
Applying this model to the division of labor between the sexes suggests
that women should be sorted into occupations requiring less strength
and men into occupations requiring more. This allocation of labor is

A. D. Roy, “Some Thoughts on the Distribution of Earnings,” Oxford Economic Papers 3
(1951), pp. 135“46.
Explaining occupational sorting 139

ef¬cient, rather than arbitrary. I will ¬rst present models of sorting for
piece-rate and time-rate work, and then examine whether the evidence is
consistent with these models.
A model is, by de¬nition, a simpli¬cation of reality. The models
presented here make a number of simplifying assumptions. They
abstract from reality by assuming that individuals differ only by strength.
Gender is relevant only because it is strongly correlated with strength. In
reality workers differ in many other ways, including cognitive ability and
skill. However, since I do not believe cognitive ability is correlated with
sex, it cannot explain gender differences. Skill is correlated with gender,
but it is endogenous. I assume that individuals of either sex could
acquire equal amounts of skill if they had the incentive to do so. In these
models I assume that skill does not determine the division of labor
among the sexes because individuals of either sex could obtain the skills
necessary for any occupation. If there are barriers that prevent women
from acquiring skill in a certain occupation, then the market is not
competitive. Chapters 5 and 6 discuss what happens when such barriers
exist. In the models presented below, strength imposes the only exogen-
ous constraint on individuals. Strength is to some extent endogenous, but
as we saw in Chapter 2 changes in ¬tness levels do not erase the male
strength advantage, so I will assume that the male strength advantage is
exogenous, a biological given.

Model A: Job choice with piece-rate wages4
In some cases the employer could observe output and thus could pay
each worker a wage equal to his or her true marginal product. This
description applies to work done on piece-rate wages, where the worker
was paid for his or her output rather than for time input. In these cases
employers were generally indifferent to who their workers were. In
handloom weaving, for example, the employers paid for output and had
relatively little contact with their workers. 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 even by a journeyman
renting a loom. 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

For similar models see Roy, “The Distribution of Earnings,” and Michael Sattinger,
“Assignment Models of the Distribution of Earnings,” Journal of Economic Literature 31
(1993), pp. 831“80.
140 Gender, Work and Wages in Industrial Revolution Britain

men, and their families, and apprentices I believe.”5 In a case like this,
where the employer did not even know who produced the cloth, workers
could choose the occupation best suited to their strength endowment.
This section models the individual™s choice of occupation in a market
where each worker is paid a set price times the number of units he or she
produces, and can freely choose his or her occupation. Given gender
differences in strength, and variations in the return to strength across
industries, the model predicts gender differences in wages and occupa-
tions that match observed outcomes. The model can also be used to
predict gender patterns of wages and occupations, and to examine how
changes in the occupational distribution and the wage gap result from
changes in demand or technology.
Models are simpli¬cations of reality that allow us to understand reality
better, so I begin my discussion of this model by describing the
assumptions on which it is based. First I assume that workers have a
range of different levels of strength. We can imagine giving each worker
a percentile score “S” indicating where he or she falls relative to others
in terms of strength (i.e., the person with a score of 20 is stronger than
20 percent of the workforce and weaker than 80 percent of the work-
force). If we compared strength scores across the sexes, women would
have lower scores than men. For the purposes of this model I will assume
that the male and female strength distributions do not overlap, and that
all individuals below the median are female, and all individuals above the
median are male. In reality there would be some overlap, though
measures of strength suggest the amount of overlap would be quite
small. Table 3.1 shows, for men and women age 30 to 34, three
measures of strength at the 90th, 50th, and 10th percentiles. In every
case the 90th percentile for females is well below the 10th percentile for
males. Thus, assuming that all females have strength scores lower than
all males is not too far from the truth. Those concerned about this
assumption can turn to the appendix to this chapter (pp. 336“7) for a
more general model which allows the male and female distributions to
I also assume that an individual may choose any occupation, and will
earn in that occupation an income equal to the number of units he or she
produces (qi) times the piece-rate in that occupation. An individual™s
output in a particular job is a function of the worker™s strength score.
Since jobs differ in their use of strength, the relationship between output
(qi) and strength (S) will be different for different occupations. In some
jobs strength is no advantage; in others it is very important. We can

BPP 1816 (397) III, p. 167.
Explaining occupational sorting 141
Table 3.1. Male and female strength distributions

Grip strength (kg) Arm strength (kg) Strength index

90th percentile 124 114 234
50th percentile 102 94 198
10th percentile 82 74 161
90th percentile 65 60 123
50th percentile 49 49 102
10th percentile 36 36 77

Strength scores for individuals aged 30“34.
Source: Henry Montoye and Donald Lamphiear, “Grip and Arm Strength in
Males and Females, Age 10 to 69”, Research Quarterly 48 (1977), pp. 109“20.

Units of

Strength Score (S)
Figure 3.1 A general example of a strength“productivity relationship

imagine graphing the relationship between strength and output in a
particular job; Figure 3.1 shows an example of such a relationship. The
relationship between strength and productivity is increasing in this
example, but it need not be. If strength does not affect output, then the
strength“productivity relationship will be a ¬‚at line. It is theoretically
possible that greater strength might actually reduce output, though if
this did occur it would be rare. In general the relationship need not be
linear, but in this section I will use linear functions for simplicity.
The worker takes the piece-rate as given and chooses the occupation
that gives him or her the highest income. To take a speci¬c numerical
example, we can imagine two occupations, A and B, whose products are
good A and good B. Strength is not valuable in occupation A, and any
142 Gender, Work and Wages in Industrial Revolution Britain


Value of Output ($)



0 25 50 75 100
Strength Score
Figure 3.2 A speci¬c example: productivity as a function of strength in
occupations A and B

individual can produce twenty units of good A in a week. In occupation B,
however, strength is advantageous, and those with more strength can
produce more output; in this case the number of units an individual can
produce is equal to twice that individual™s strength score (qi ¼ 2Si). We
can graph both functions on the same graph if we value the output in
money. If the piece-rate paid in occupation A is $10 per unit of good A,
then any individual can earn $200 per week in occupation A. If the
piece-rate paid in occupation B is $4 per unit of good B, then the income
an individual can earn is a function of S, and is equal to eight times
the strength score ($8Si). The relationships between strength and income
in each occupation are graphed in Figure 3.2.
If individuals are free to choose their occupations, then the com-
petitive market will produce the outcome that is optimal both in terms
of maximizing each individual™s income, and in terms of maximizing
society™s total output. If each individual chooses the occupation that gives
him or her the highest income, then in the case depicted in Figure 3.2
anyone below the 25th percentile in strength would choose occupation A,
and anyone above the 25th percentile would choose occupation B.
This outcome is optimal for individuals; any individual who changes
Explaining occupational sorting 143


Value of Output ($)



0 25 50 75 100
Strength Score
Figure 3.3 The ef¬ciency costs of moving workers

occupations would earn less. The outcome is also optimal in the sense
that the total value of the output is maximized; if any individual changes
jobs it will reduce the total value of the goods produced in the economy.
Suppose that we tried to make the gender division of labor more equal
and moved women with strength scores between 10 and 20 from
occupation A to occupation B, while at the same time moving men with
strength scores between 45 and 55 from occupation B to occupation A.
The women, who had been producing goods worth $200 each, would
now produce goods worth between $80 and $160 each. The men, who
had been producing between $360 and $440 worth of output each,
would now produce only $200 worth of output each. Both the men and
the women would be worse off, and the total value of society™s output
would also be lower, so the outcome would be inef¬cient. In Figure 3.3,
area X is the amount of value lost by moving the women to occupation B,
and area Y is the amount of value lost by moving the men to occu-
pation A. There are no compensating gains. Both men and women
earn less, and the total value of output produced has declined by the
sum of these two areas.
A change in demand for the goods would alter the piece-rate prices,
and change the incentives to workers, thus changing the output mix. If
consumers decide they don™t want as many units of good B, then there
144 Gender, Work and Wages in Industrial Revolution Britain


Value of Output ($)



0 25 50 75 100
Strength Score
Figure 3.4 A decline in the price of good B

will be an excess supply of good B in the market, and the price of good B
will fall. Suppose, for example, the price of good B falls to $2.50 (see
Figure 3.4). Individuals who work in occupation B will now earn their
strength score times ¬ve (5Si), and individuals between the 25th and
40th percentile will ¬nd it to their advantage to switch from occupation
B to occupation A. As a result, society produces more of good A and less
of good B.
This model allows us to examine the implications of strength for
gender differences in occupations and wages, and to examine how those
differences would respond to changes in demand or technology. If we
assume that individuals below the 50th percentile are female, and
individuals above the 50th percentile are male, then, in the case in
Figure 3.2, workers in occupation A are all female, and workers in
occupation B are one-third female. The women in occupation A all earn
$200 per week. The women in occupation B earn different incomes, but
average $300.6 The men in occupation B earn on average $600 per
week.7 Within occupation B the female“male wage ratio is 0.5. Aver-
aging women in both occupations, the female wage is $250 per week, so

The average income is calculated by taking the midpoint on the line segment between
S ¼ 25, where income would be 200, and S ¼ 50, where income would be 400.
600 is the midpoint between 400 and 800.
Explaining occupational sorting 145

the female“male wage ratio for the workforce as a whole is 0.42 (250/
600). Note that, in spite of the fact that all individuals in occupation B
earn more than all individuals in occupation A, the wage gap is not
caused by barriers preventing women from entering occupation B.
If occupational segregation is the result of sorting by comparative
advantage, it reduces the size of the wage gap rather than increasing it.
If, in Figure 3.2, we moved all the women below the 25th percentile
from occupation A into occupation B, their earnings would fall from
$200 to an average of $100, and the female“male wage ratio would fall
to 0.33 (200/600). Moving to the male-dominated occupation would
make the women worse off, not better off. Division of labor according to
comparative advantage improves women™s relative productivity because
it allows them to work where the penalty to their lesser strength is
smallest. Specialization improves the productivity of women relative to
men. Toman has found evidence of this phenomenon among slaves in
the US South. Plantations large enough to use the gang system could
increase output per worker because they could divide the labor
according to comparative advantage, giving the most strenuous tasks to
the strongest workers, and using the weaker workers for less strenuous
tasks. Under the tasks system, which did not allow for such sorting,
female slaves were only 40 percent as productive as male slaves. Under
the gang system, which allowed for sorting according to comparative
advantage, the marginal productivity of a female slave was 60 percent as
much as a male slave.8 The model presented here implies that occu-
pational sorting raised women™s productivity, and thus their earnings, by
sorting them into occupations that were to their comparative advantage.
This is the opposite of the conclusion implied by Bergmann™s crowding
model, where reductions in occupational segregation would decrease the
wage gap.
This model can also explain how the gender division of labor or the
size of the wage gap can change in response to changes in the demand
for goods, or the introduction of new technology. Changes in the gender
division of labor are sometimes interpreted as evidence that custom
determined this allocation. For example, Laurel Thatcher Ulrich,
observing the movement of women into weaving in New England,
concludes that: “The shift in the division of labor described here . . .
demonstrates an essential feminist argument “ that gender is socially
rather than biologically determined.”9 However, this occupational shift

Toman, “The Gang System.”
Laurel Thatcher Ulrich, “Wheels, Looms, and the Gender Division of Labor in
Eighteenth-Century New England,” William and Mary Quarterly 55 (1998), p. 34.
146 Gender, Work and Wages in Industrial Revolution Britain

need not be the result of shifts in social expectations. The movement of
women into handloom weaving can be explained by the combination of
biological difference and technological change. In the model just
described we would expect changes in technology or demand to alter
both the division of labor between the sexes and the wage ratio. Suppose
that, as in Figure 3.4, the demand for good B falls. Women between the
25th and 40th percentile would move into occupation B. Occupational
segregation increases as the representation of women in occupation B
falls to one-sixth, but the wage gap also falls. In the new case, women in
occupation B earn only $225 on average, but the average earnings of
the men falls even more, to $375. Women in occupation B now earn
60 percent as much as men. The improvement in their relative wage is
the result of the movement of the weakest women into occupation A,
and of the decrease in the demand for the strength-intensive product
which reduced the market return to strength.
Taking a case more closely related to the Industrial Revolution,
suppose that a new strength-intensive technology is introduced for
producing good A. The new occupation, occupation C, produces the
same good as occupation A, but has a steeper strength“productivity
line. An individual working in occupation C can produce (2Si“60) units
of output per week. The ease with which high-strength individuals can
produce good A will increase the supply of this good and decrease its
price. Suppose the price of good A falls so that the piece-rate for
producing a unit of good A is only $5. If the price of good B is $2.50 as
in Figure 3.4, individuals in that occupation earn 5Si, and women with
strength scores between the 20th and 50th percentiles will choose
occupation B. Occupation C produces the same output as occupation A,
so the price of good C is also $5. Individuals in that occupation
will earn 5(2Si À 60) ¼ 10Si À 300. Individuals with a strength score
above the 60th percentile will choose occupation C (see Figure 3.5). In
this case, occupation A contains only women, occupation B contains
both women and men, and occupation C contains only men. The
overall wage gap increases, and women on average earn only 32 percent
as much as men.10
While this is a hypothetical example, it does bear some resemblance
to what happened in textile manufacturing during the Industrial Revo-
lution. Hand spinning, like occupation A, did not require strength.

Two-¬fths of women earn $100 in occupation A, and the three-¬fths of women in
occupation B earn on average $175. Overall women earn on average $145. The one-
¬fth of men in occupation B earn on average $275, while the rest earn on average $500
in occupation C. Overall, men earn on average $455.
Explaining occupational sorting 147



Value of Output ($)



0 25 50 75 100
Strength Score
Figure 3.5 Entry of a new occupation

During the Industrial Revolution, however, this old technology was
replaced by a new technology that did require strength. The earliest
spinning machines, the jenny and the throstle, did not require much
strength and were worked by women. However, the spinning mule,
which did require a great deal of strength, could produce a better quality
of yarn and became the dominant machine. Women were initially used
for mule-spinning, but as the mules grew larger men came to dominate
that occupation.11 Men became mule spinners, and women moved
out of hand spinning as the wage in that occupation dropped. Young
women and boys were used as assistants to mule spinners, but overall the
demand for female workers in spinning fell substantially, partly because
the total number of workers required to spin yarn fell. Many of the
women who were no longer employed in hand spinning moved into
handloom weaving, whose strength requirements were greater than in
hand spinning but less than in mule spinning. The introduction of a new
strength-intensive technology, which replaced an old technology
requiring no strength, made women worse off and increased the overall
wage gap in the economy.

Men continued to dominate mule spinning even after the self-actor reduced the need for
strength because the mule spinners™ union was able to exclude women. See Chapter 6,
and Freifeld, “Technological Change and the ˜Self-Acting™ Mule.”
148 Gender, Work and Wages in Industrial Revolution Britain

Model B: Time-rate wages with sex as a signal of strength
In many cases the employer could not readily observe individual output
and thus could not pay piece-rate wages. For the second model I make
an extreme informational assumption “ that employers observe only the
worker™s sex and know nothing else that would help them measure the
worker™s productivity. Where output of an individual worker is not
observed, the employer must pay a time-rate wage. The employer can
observe sex, which is a signal of strength, and in a competitive market
must pay each worker his or her expected productivity given sex. With
insuf¬cient information on individual productivity, each woman is
assumed to have the average female strength endowment, and each man
the average male strength endowment. The result is statistical discrim-
ination; workers of each sex will be treated as a homogenous quantity,
and all persons of the same sex will receive the same wage, regardless of
their individual productivity.
Firms produce a product, and the amount that an individual can
produce in a day is a function of strength, though not necessarily pro-
portional to strength. If strength increases productivity, then the employer
will expect that men on average will produce one level of output, and
women another. The ratio of average female output to average male
output will be different at different ¬rms, and I will call it ri, where the
subscript i indicates that each ¬rm has its own female“male productivity
ratio. We can imagine arranging ¬rms by ri, starting with the ¬rm with the
lowest r and progressing to those ¬rms with higher levels of r. We can then
construct an “r-pro¬le,” which is a step function, as in Figure 3.6. The
length of the x-axis (the origin to point L) is the total amount of labor
employed in the economy, and the length of each step on the r-pro¬le is
proportional to the number of workers hired by each ¬rm. The market
wage ratio will be the wage ratio that clears the market, that is, where
¬rms™ demand for female workers equals the supply of females, and ¬rms™
demand for males equals the supply of males. Point X divides the
workforce into males and females. The distance between the origin and
point X is the number of males in the workforce, and the distance
between point X and point L is the number of females in the workforce.
Each ¬rm will compare its productivity ratio to the market wage ratio.
If ri < wf /wm the ¬rm will hire males, since females are too expensive
relative to their productivity. If ri > wf /wm the ¬rm will hire females,
since their relative productivity is greater than their relative wage. A ¬rm
is willing to hire both males and females only if ri ¼ wf /wm, so that the
¬rm is indifferent between male and female workers. Figure 3.6 shows
how the market wage ratio is determined. The vertical line through point
X intersects the r-pro¬le at the seventh ¬rm, which has an r of 0.55. This
Explaining occupational sorting 149

Figure 3.6 Determination of the wage ratio in Model B

means that the market wage ratio must be 0.55 to clear the market. The
six ¬rms with r™s lower than 0.55 will hire only males, and the six ¬rms
with r™s higher than 0.55 will hire only females. The seventh ¬rm is
indifferent to hiring males and females, and will hire both. Note that this
model predicts nearly complete occupational segregation. Only ¬rms
whose ri is exactly equal to the market wage ratio will hire both men and
women. Thus even high levels of occupational segregation are consistent
with ef¬cient sorting in a competitive market. While only 9 percent of
occupations in the 1904 report of the US Commission of Labor were
integrated by gender, this fact is not proof of discriminatory constraints
on occupational choice.12
The entry and exit of ¬rms, or expansion and contraction of
employment at existing ¬rms, will cause the r-pro¬le to shift and may
change the market wage ratio. Suppose that a low-r ¬rm replaces a high-r
¬rm, as when mule-spinning replaces hand spinning. If we remove the
¬rm whose r is 1.0, and replace it with another ¬rm whose r is 0.3, then
the new r-pro¬le shifts down and to the right, as depicted in Figure 3.7.
At the old wage ratio of 0.55, there would be women unemployed and
an excess demand for men. To clear the market the wage ratio will fall.
The vertical line originating at point X intersects the new r-pro¬le at a
¬rm whose r is 0.5, so the new market wage ratio is 0.5. Thus we ¬nd
that, as in Model A, the replacement of a low-strength technology with a

This statistic comes from Goldin, Understanding the Gender Gap, p. 77.
150 Gender, Work and Wages in Industrial Revolution Britain

Figure 3.7 A change in technology

higher-strength technology (as happened in spinning) would cause the
relative wages of women to fall.
The two models presented here depict two extremes. In the ¬rst
model, the output of each worker is perfectly observable, while in the
second model the employer cannot observe individual output at all, and
knows only the gender of the worker. Even when time-rate wages were
paid, we might think that the employer received some information about
the individual productivity of the workers. Output, though not perfectly
observable, may be observable with some error. If the employer initially
knows only the gender of the worker, but over time learns more about
the individual productivity of that worker, then the employer will be able
to tailor the wage more closely to the individual™s ability. A third model
presented in the appendix (Model C) describes this case in greater
detail. An important prediction of this third model is that casual workers
are more likely to have uniform wages than regular workers. This
matches the pattern of wages we observe in agriculture. Female day-
laborers were usually casual workers hired for only a few weeks, so
employers did not have much information about individual productivity,
and their wages were uniform. Often all women received the same wage.
Male laborers were more likely to work regularly, and we do observe
more individual variation in wages for male workers. Table 3.2 shows
the wages paid to all workers on July 15, 1836 at the Apley Park farm in
Shropshire. All the women were paid the same wages, but only eight of
Explaining occupational sorting 151
Table 3.2. Wages paid to laborers at the Apley Park farm,
July 15, 1836

Name Daily wage (d.)

Sam Partridge 36
Sam Pace 24
Richard Piver 22
John Hadley 20
Samuel Crump 20
John Rhods 20
William Pace 20
John Blackshaw 20
Henry Barret 20
Thomas Littleport 20
Jim Brown 20
Moses Davis 18
William Childs 12
Edward Hollans 8
Sarah Hadley 10
Eliz. Crump 10
Sarah Brown 10
Mary Barret 10
Jane Pace 10
Sarah Harris 10
Jane Parton 10

Source: Shropshire Record Of¬ce 5586/5/17/28.

the fourteen men earned the modal wage. The uniformity of female
wages has sometimes been interpreted as evidence of the importance of
custom in setting wages, but it may be that farmers paid all women the
same wage because they were casual workers and thus the employer had
little information about individual productivity.
Before looking for evidence that these models describe the labor
market during the Industrial Revolution, I will point out a few things
that are not implied by the models of occupational sorting presented
here. First, it is not necessary for the strength requirements of jobs to be
so great that women cannot do these jobs at all. Women were capable of
plowing, but that doesn™t mean that a woman would do the job as well or
as quickly as a man. In both of the models presented above, I assume
that women can do all of the jobs, but sometimes have lower product-
ivity than men (an exception is occupation C in Figure 3.5, where
women below the 30th percentile have zero productivity). Nor is it
necessary that the differences in productivity be large. Gullickson argues
152 Gender, Work and Wages in Industrial Revolution Britain

that since the difference in the strength needed for weaving different
types of cloth was small, the occupational sorting by type of fabric could
not have been based on this difference. She claims, “The difference in
strength requirements, as well as the actual strength of men and women,
was slight, however, and the assertion of male superiority was far more
important in this new sexual division of labor than was physical
strength.”13 However, the size of the productivity difference need not be
large. A small difference in productivity can lead to complete occupa-
tional segregation as easily as a large difference. Even if the differences in
productivity were relatively small, maximum ef¬ciency could only be
achieved if each person was assigned to the task where he or she would
be most productive. Differences in strength can lead to complete
occupational segregation even if the differences in productivity are
relatively small, and women simply work more slowly than men.
Nor does the fact that women can be observed doing heavy work
disprove the strength hypothesis. The fact that women sometimes did
strength-intensive tasks does not prove that they were as productive as
men in those tasks. Women were capable of plowing, but rarely did so.
Gielgud concludes that, “Most women were too sensible to take on a job
they knew to be beyond their strength unless particular circumstances
required them to do it . . . and it seems that men recognized it as work
which was too physically dif¬cult for most women and did not expect it
of them.”14 When women did plow, it was usually on small family farms
that did not hire laborers.15 Deborah Simonton claims that, “The
continued involvement of women across a broad spectrum of ¬eld tasks,
including the most arduous, undermines the notion that purely bio-
logical characteristics or maternal functions determined female agricul-
tural roles.”16 However, if the demand for strength is high in the
economy, or if institutional restrictions prevent women from entering
more suitable occupations, women will be assigned to strenuous tasks
even if they are less productive than men in those tasks.17

Gay Gullickson, Spinners and Weavers of Auffay: Rural Industry and the Sexual Division of
Labor in a French Village, 1750“1850 (Cambridge: Cambridge University Press, 1986),
p. 109.
Gielgud, “Nineteenth Century Farm Women,” p. 110.
Pinchbeck, Women Workers, p. 9. If her husband was busy with other work a woman
might end up plowing because hiring a laborer rather than employing a family member
created principal-agent problems.
Simonton, European Women™s Work, p. 130. See also Hudson and Lee, “Women™s
Work and the Family Economy,” p. 9.
In Germany, where institutional restrictions were tighter and markets were less
competitive, women seem to have done more plowing than in Britain. Ogilvie, A Bitter
Living, pp. 119, 151, 200.
Explaining occupational sorting 153

These models also do not imply that the gender division of labor will
remain ¬xed over time. Laurel Thatcher Ulrich, noting that weaving
shifted from being men™s work to women™s work in eighteenth-century
New England, claims that this shift proves “that gender is socially rather
than biologically determined.”18 However, the models presented here
demonstrate that changes over time in the gender division of labor or
wage ratio are compatible with a biologically-based explanation of the
division of labor. Even if differences in strength are biologically based,
and the wage gap is the direct result of those differences in strength, this
does not imply that the wage gap must remain ¬xed over time. Strength
was a scarce good, which received a premium in the market, but the
value of strength, and the size of that premium, varied over time. In fact,
we should expect the wage ratio to change with changes in demand and
technology. While I do assume that gender differences in strength are
biologically based, I do not expect these strength differences to produce
the same division of labor, or the same wage ratio, in all times and
places. The secret to this puzzle is to realize that the same underlying
strength ratio is converted into different productivity ratios in different
occupations. Some technologies are very strength-sensitive, so that the
underlying strength differences cause large differences in productivity.
Other technologies are less sensitive to strength, and in those cases
differences in productivity by gender are small or non-existent.
The models also make it clear that we should not expect men and
women to earn the same wage even when they do the same task.
Commenting on farm accounts from seventeenth-century Devon,
Pamela Sharpe notes that “women agricultural workers were paid less
than the rates of pay for men, even when they appear to have done
directly comparable work. For example, in May 1683, Jone Clements
and her husband were both paid for a day™s threshing. She was paid 4d.
and he was paid 5d.”19 Even if both of the Clements worked the same
hours, we should not expect men and women to earn the same wage
because we cannot expect them to have threshed the same amount of
grain. Threshing was a strenuous job, and unless Jone was very unusual
she had less strength than her husband, and would have threshed less
grain over the course of the day.20 Both of the models above predict that,
when men and women work in an occupation where strength increases

Ulrich, “Wheels, Looms and the Gender Division of Labor,” pp. 34“5.
Pamela Sharpe, “Time and Wages of West Country Workfolks in the Seventeenth and
Eighteenth Centuries,” Local Population Studies 55 (1995), pp. 67“8.
For evidence that threshing was a job requiring strength, see Stephen Duck™s poem,
“The Thresher™s Labor,” in Poems on Several Occasions (London, 1736).
154 Gender, Work and Wages in Industrial Revolution Britain

productivity, the women will produce less because they have less
strength, and will earn less than men doing the same job.

B. Anecdotal evidence of market-based sorting by strength
A closer look at the historical facts shows that the models of occupa-
tional sorting presented in the previous section are helpful for under-
standing the occupational distribution during the Industrial Revolution,
at least for the competitive sectors of the economy. The models predict
that we should observe men in strength-intensive occupations, and
women in occupations requiring less strength. Much of the occupational
pattern of the early nineteenth century is consistent with this sorting
prediction. To show that the models presented above are good
descriptions of how the gender division of labor was determined, I will
examine whether the strength-intensity of the job determined whether
men or women were employed. I look ¬rst at the overall distribution of
labor between the sexes, showing that men did the most strenuous tasks.
Then I show that some unusually strong women can be found doing
male jobs. Next I examine regional differences in the pattern of sorting,
and show that in regions where the demand for strength was lower
women earned higher relative wages. I also show that the division of
labor between the sexes changed in response to technological changes
that altered the strength requirements of certain jobs.

1. Patterns of occupational sorting
In general women tended to congregate in occupations requiring little
strength. In Chapter 1 we saw that hand spinning was one of the largest
employers of females during the pre-industrial period. Women also
monopolized such industries as lace making and straw plaiting which
required dexterity but not strength. Heavy industries, on the other hand,
employed men. To support her claim that “androcentric blindness”
prevented employers from hiring women, Jordan points to the absence
of women in certain new industries, speci¬cally iron and steel manu-
facture, engineering, railways, and mining.21 However, we would not
expect these industries to employ women in a competitive market.
They are called “heavy industries” for a reason “ the materials are heavy
and the work requires strength. Many occupations requiring heavy
work were male preserves. In the 1841 census, dock workers and ¬re-
men (those who tended furnaces) were exclusively male. Quarriers were
99.5 percent male, wood cutters 95.5 percent, sawyers 99.93 percent,

Jordan, “The Exclusion of Women,” pp. 294“5.
Explaining occupational sorting 155

and ship builders 99.95 percent.22 The lack of women workers in these
occupations is not surprising given the strength required for the work.
Both men and women worked in agriculture, but they typically did
different tasks, and the men™s tasks generally required more strength
than women™s tasks. In the eighteenth-century records of an estate in the
North Riding of Yorkshire, Gilboy found that the jobs done exclusively
by men were threshing, hedging, and plowing “ the most strenuous of
the agricultural jobs.23 Plowing is strongly associated with male workers,
to the point that the use of the plow is an important factor in deter-
mining whether agricultural work is done by men or women. Ester
Boserup notes that “the advent of the plow usually entails a radical shift
in sex roles in agriculture; men take over plowing even in regions where
the hoeing had formerly been women™s work.”24 Farm women inter-
viewed by Judy Gielgud did some plowing, but only with dif¬culty:
From the women I have interviewed who have plowed with horses . . . it is plain
that the weight of the plow, particularly when turning it at the headlands at the
end of the furrow, is what needed particular strength; the share and coulter must
be lifted clear of the furrow and the handles born down upon to swing the plow
round. Most said they simply did not weigh enough to do this with comfort.25

Women were more likely to do jobs such as spreading manure and
weeding. Henry Best, a Yorkshire farmer, summarized the harvest labor
needs of his farm in 1641. Breaking the tasks into three categories
according to the sex of workers:

Women only Both sexes Men only

Gathering after Reaping Mowing
the mower Pulling peas Binding
Raking Stooking (stacking)

All of these tasks required at least moderate amounts of strength.
Mowing (harvesting with a scythe) required a great deal of strength.

BPP 1844 (587) XXVII. On the strength required for these jobs, see Samuel,
“Workshop of the World.”
Gilboy, “Labor at Thornborough.”
Ester Boserup, Women™s Role in Economic Development (New York: St. Martin™s Press,
1970), p. 33. In Africa, where the hoe is the main tool for cultivation, agricultural work
is primarily done by women.
Gielgud, “Nineteenth Century Farm Women,” pp. 108“9. Note that the fact that
women occasionally did plow does not contradict the claim that it was not generally the
most ef¬cient use of scarce resources. Cases of women plowing usually come from
labor-scarce regions, such as the north.
156 Gender, Work and Wages in Industrial Revolution Britain

Reaping (harvesting with a sickle) required strength, but less than
mowing. The men-only tasks other than mowing required lifting the hay,
and the women-only tasks did not require lifting. Even within each task,
Best instructs his readers to take account of individual strength in
assigning workers; thus, “the strongest and ablest of your shearers
[reapers] you shoulde allways putte to the ridge, because there the corne
is rankest and strongest.”26
The sickle was used by both men and women, but the scythe was used
only by men. Michael Roberts examined 125 illustrations from Europe
in the period from the second century to the sixteenth century and found
no examples of women using scythes, though he found many examples
of women using sickles. Though Roberts was skeptical of employers™
assumptions about women™s strength, he noted that, “The strength or
stature required of the mower appears to provide the most plausible
explanation why women were so rarely employed in this way.”27 Men
mowing with a scythe remained standing and swung the scythe. They
bene¬ted from greater strength, because the scythe was heavier than the
sickle, and from greater stature, because this provided more leverage. By
contrast, shorter stature was bene¬cial to the reaper, who stooped to cut
with the sickle. While reaping required less strength than mowing, it
required more strength than other tasks done by women. Mary Haynes,
a laborer from Wiltshire, noted that “I think reaping the hardest of all
the work I have ever done.”28 She mentions having worked at stone-
picking, weeding, haymaking, and turnip-hoeing, as well as reaping, so
reaping was more demanding than all these other tasks. This suggests
that reaping required more strength than other types of farm work done
by women. Reaping required enough strength that not every woman
could do it. A farm laborer from Devon noted that his wife worked
regularly in the ¬elds, but “she can™t reap, she is not strong enough.”29
Model A predicts that, when women are observed doing relatively
strength-intensive jobs, we should also observe them earning more than
other women. This prediction holds true for reaping. Reaping was
relatively well paid work for women, and for much of English history the
female“male wage ratio was higher at harvest than at other times of year.
Among the agricultural wage ratios in Table 2.1, the wage ratio for
harvest work tends to be higher than the wage ratio for work in other

Henry Best, Rural Economy in Yorkshire in 1641, Surtees Society, vol. 33 (Durham:
George Andrews, 1857), p. 44.
Roberts, “Sickles and Scythes,” p. 12.
Evidence of Mary Haynes, widow, of Calne, Wiltshire, Women and Children in
Agriculture, p. 69.
George Moxey of Shillingford, near Exeter, Women and Children in Agriculture, p. 112.
Explaining occupational sorting 157

seasons, at least in the seventeenth and eighteenth centuries. For the
seventeenth century, the harvest wage ratios average 0.72, while the wage
ratios for hay time average 0.59. In the eighteenth century all the wage
ratios are lower, but the wage ratio is still higher at harvest (0.67) than at
other times of the year (0.48). By the nineteenth century, this effect has
disappeared, and the harvest wage ratios are no higher than wage ratios
at other times of year. This reduction in the harvest wage ratio was
probably due to the disappearance of opportunities for work reaping. In
the eighteenth century (and before) most grains were cut with the sickle.
Women could do this task fairly well, and were paid accordingly. In the
nineteenth century, however, the scythe replaced the sickle for cutting
grains, and women worked only in less well paid subsidiary activities.30
While some agricultural tasks, such as hedging and threshing, were
always assigned to men, many tasks were done by both men and women.
While we sometimes observe both men and women doing the same task
at the same farm at the same time, more often we observe a task moving
back and forth between male and female workers depending on the
circumstances. Hoeing was an example of a task that was done some-
times by women and sometimes by men. Marshall explains how the task
of weeding could vary in intensity depending on various conditions:
[T]he requisite labor varies with the state of the crop, and the nature of the soil.
A full clean crop, on a free soil, wants little labor. Nor, on such a soil, though
foul with seed weeds, is the labor dif¬cult, provided that crop has not been
suffered to run up, and hide the surface. On the contrary, a thin tall crop, foul
with couchgrass, on a stubborn soil, in a dry season, requires more labor than is
paid for. I have seen a man hoing wheat, under the last mentioned circum-
stances, at 3s. an acre. But he barely earned day wages . . . If the soil be tolerably
free, the season kind, and the crop taken in a proper state, as to growth, not-
withstanding it may be foul with seed weeds, there are women who will hoe half
an acre, a day. Such a crop is not unfrequently done, at 2s. an acre.31

Note that Marshall suggests men are used for the more laborious hoeing
tasks, and women for the less laborious hoeing tasks. Whether men or
women were used for hoeing might also depend on the local labor
demand conditions. In the discussion of Model B we saw that an
increase in demand for labor in a less strength-intensive job may change
the marginal occupation. Regions with better alternative employments
for women should be less likely to employ women for hoeing.
The agricultural task most closely associated with women was dairy-
ing, but even in the dairy the gender division of labor responded to

On the increased use of the scythe, see Chapter 7, and Collins, “Harvest Technology.”
Marshall, The Rural Economy of Gloucestershire, vol. I, pp. 123“4.
158 Gender, Work and Wages in Industrial Revolution Britain

comparative advantage. Male assistants were frequently hired to do the
most strength-intensive tasks, such as turning and washing the large
cheeses, and churning the butter.32 Milking, which required less
strength than some agricultural jobs, was commonly a female task, but
when women workers could ¬nd work in non-agricultural occupations
requiring even less strength, men did the milking. John Broad notes that
“In areas such as Buckinghamshire where women were more fully and
pro¬tably employed in lace and straw plait manufacture than the men in
agriculture, men took over the subsidiary task of milking cows, leaving
women in charge of the indoor work.”33
Turning to textile manufacture, we also ¬nd the division of labor
among tasks to be consistent with the models presented. The employ-
ment of women in the Leeds cloth industry in 1769, as described by
Arthur Young, matches the predictions of Model A. Young mentions
women employed in three different occupations: weaving broad cloths,
weaving “stuffs,” and spinning.34 The weaving of broad cloths required
the most strength. The reason for this is succinctly described in the 1841
parliamentary report on handloom weavers: “In proportion to the width
of a fabric must be the size of the loom, and therefore the strength of the
workman.”35 Weaving stuffs required little strength, and hand spinning
required no strength. A graph of this situation would look something
like Figure 3.5, with spinning as occupation A, weaving stuffs as occu-
pation B, and weaving broad cloths as occupation C. As the model
predicted, Young observed that women in Leeds were more likely to
work in spinning and in weaving the lighter cloths. However, Young did
¬nd a few women working even in the strength-intensive broad cloths,
and these women earned as much as the men. Men weaving broad cloths
could earn 10s.6d. if fully employed, but usually earned 8s. a week. The
wages of women who did this work were similar. Young noted that
“some women earn by weaving as much as the men.”36 Other women,
working in other occupations, earned much less. Women weaving stuffs
earned 3s.6d. or 4s. a week, and those spinning earned 2s.6d. or 3s. This
difference in wages is predicted by the model, and is the result of the
women™s individual abilities. According to Model A, the women who
worked as spinners could not have earned 8s. a week weaving broad
cloths because they did not have the same characteristics as the women

Pinchbeck, Women Workers, pp. 13“14; Valenze, The First Industrial Woman, p. 60.
John Broad, “Regional Perspectives and Variations in English Dairying, 1650“1850,”
in R. Hoyle, ed., People, Landscape and Alternative Agriculture (Exeter: British
Agricultural History Society, 2004), pp. 106“7.
Young, Northern Tour, vol. I, p. 89. 35 BPP 1841 (296) X, p. 3.
Young, Northern Tour, vol. I, p. 89.
Explaining occupational sorting 159

who did, and the spinners did not choose to move to broad cloth
weaving because it would not have improved their wages. Occupational
assignment, and wages, seem to have been based on individual ability
rather than strictly on gender.
In Norwich, where the cloths woven required little strength, the men
who wove them earned low wages. Young noted that “Men on average do
not exceed 5s. a week; but then many women earn as much: and boys of
15 to 16 likewise the same.”37 This suggests that women may have earned
slightly less than men, but that the earnings of men and women were
similar. The men who worked in this industry must have been unable to
earn high wages elsewhere, and must have had relatively low endowments
of strength, as the women weaving broad cloths in Leeds must have had
high strength endowments, allowing them to earn high wages.
There are numerous other examples where the allocation of women
workers within weaving re¬‚ects their comparative advantage in cloth not
requiring strength. Women were concentrated in the lighter and nar-
rower cloths, leaving men predominant in the heavier and wider cloths
that required more strength. Fewer women were engaged in ¬‚ax weaving
than in other types of weaving because it required more strength. In the
1840 Reports from Assistant Hand-Loom Weavers™ Commissioners, James
Mitchell reports, “The work is too heavy for women in general, and
hence the very small number of women in the trade. For the same cause
there are very few boys or girls.”38 Similarly, the manager of a bed-quilt
weaving shop in Bolton noted that he only employed men: “they must
be stout, able-bodied men to do the work; and ours is the best paid work
on the hand-loom weaving line.”39 Parliamentary investigator Symons
noted that:
Where wide, as in broad cloths and damask table cloths, a degree of strength is
required which no very young, very old, or weak persons can apply. Women are
debarred from entering into competition with men in these factories, not by the
material used, but the physical force required.40

Men, rather than women, worked looms for patterned ribbons, where
shuttles could weigh as much as 130 pounds.41 In Spital¬elds, women
were concentrated in plain silks, which required less strength than vel-
vets.42 The ratio of women to men weavers was 0.23 in velvets, but 0.99

Arthur Young, The Farmer™s Tour through the East of England (London: W. Straham,
1771), vol. II, p. 75.
BPP 1840 (43) XXIII, p. 351. 39 BPP 1840 (220) XXIV, p. 585. 40 Ibid., p. 615.
Simonton, European Women™s Work, p. 144.
The comparison of strength requirements is given by parliamentary investigator
J. Symons. BPP 1840 (220) XXIV, pp. 616“17.
160 Gender, Work and Wages in Industrial Revolution Britain


Number of Workers





















Figure 3.8 The distribution of male and female workers across cloths
of various piece-rates
Source: University of Leeds Archives, Clough 19.

in plain silks. While women were able to weave velvets, few did so
because of the greater strength requirement.
We can see the distribution of male and female workers over different
types of cloth by looking at the accounts of Robert Clough of Grove
Mill, Keighley.43 Clough owned Grove Mill, a spinning mill which used
water power until 1836. He also employed out-workers, male wool-
combers, and both male and female weavers. The weavers produced
many different types of cloth, and each type had its own piece-rate. If
cloths with higher piece-rates required more strength, and workers
sorted themselves according to their individual abilities, then we should
observe that women tend to work on cheaper cloth, but that a few
women were able to work on the higher-priced cloths dominated by
men. That is exactly what we observe. Figure 3.8 shows the distribution
of male and female workers across the different piece-rates in October
of 1824. Men and women can be observed over most of the range of

University of Leeds Archives, Clough 19.
Explaining occupational sorting 161

cloths. Both male and female weavers produced the cloth with the lowest
piece-rate (27s). No woman worked at a cloth with a piece-rate higher
than 48s., but only 10 percent of men did so. While different types of
cloth do not seem to have been assigned solely on the basis of gender,
there is a de¬nite gender pattern. Women tended to work on cloths with
lower piece-rates. The median piece-rate was 36s. for women and 42s.
for men.
The same pattern of a small number of women engaged in a pre-
dominately “male” occupation is found in the hosiery industry. In
general, men operated the knitting frames and women seamed the
stockings. This division of labor matched the comparative advantages of
the sexes because the knitting frame required strength. Osterud has
pointed out the “heaviness of the physical labor involved in the oper-
ation of wide frames,” noting that the strength necessary prevented
youths and older men from being employed on wide frames.44 While in
general work was allocated along gender lines, the barrier was not strict.
Although few in number compared to the men, women did work at the
knitting frames. Rose notes that women were occasionally knitters,
especially in Nottinghamshire and Leicestershire. In 1845, 7 percent of
frames were worked by females.45 When women did work as knitters,
this was most commonly on narrow frames, which required less strength.
This distribution of employment is consistent with the model™s predic-
tion that a few women with high strength endowments should be found
working with the men. No discriminatory constraints prevented women
from becoming framework knitters. Both boys and girls were apprenticed
to the trade.46 Women were hired at the same piece-rates as men, and
no combination of male workers opposed their employment. In 1819 a
union of framework knitters was formed in Nottinghamshire, and women
were included.47 Women could easily move into knitting if they found it
advantageous. William Felkin describes such a movement in 1833. When
the wages in embroidering fell because of foreign competition, many
women quit that trade and took up the stocking frame.
I believe the average wages of embroiderers in the bobbin-net trade have fallen
one half since September 1831 . . . I think runners are now earning from 9d. to

Nancy Grey Osterud, “Gender Divisions and the Organization of Work in the Leicester
Hosiery Industry,” in Angela John, ed., Unequal Opportunities (Oxford: Basil Blackwell,
1986), p. 50.
Rose, “Gender Segregation,” p. 166.
Felkin described the trade as suffering from a “constant in¬‚ux of too many apprenticed
boys and girls.” Felkin, A History of the Machine-Wrought Hosiery, p. 115.
Ibid., pp. 441“3. In the ¬rst three months of 1821, 2172 knitters received bene¬ts for
unemployment; 1559 were men, and 613 were women and boys.
162 Gender, Work and Wages in Industrial Revolution Britain

3s.6d. a week. I know that several young women are now putting into the
stocking-frame because they can earn more in it than at lace-running.48

These women had not been segregated into the embroidery trade; they
chose that occupation because it was the most pro¬table for them, and
they moved to another occupation when it ceased to be pro¬table. The
only disadvantage a woman had was her lack of strength. The 1845
parliamentary report on framework knitters emphasized this equality of
opportunity: “Vast numbers of women and children are working side by
side with men, often employed in the same description of frames,
making the same fabrics, at the same rate of wages, the only advantage
over them which the man possesses being his superior strength.”49 This
statement clearly describes a competitive market.
The one exception to the rule that men always did the jobs requiring
strength is laundry. Laundry work required heavy lifting and wringing. An
Exeter woman who had worked in both agriculture and washing thought
that “washing is harder than working in the ¬elds.”50 One laundress
remarked, “You had to be as strong as a man to lift the great wooden wash
tubs.”51 Since the work was so heavy, men sometimes helped their wives
with the heaviest parts of it. A London street-seller told Henry Mayhew
that he helped his laundress wife turn the mangle: “She earns about 1s. 3d.
a day. She takes in a little washing, and keeps a mangle. When I™m at home
I turn the mangle for her.”52 The husband of another laundress, an often-
unemployed dock worker, reported to Mayhew that “the party my wife
works for has a mangle, and I go sometimes to help; for if she has got 6d.
worth of washing to do at home, than I go to turn the mangle for an hour
instead of her “ she™s not strong enough.”53 A Manchester schoolmaster
was “liable to interruption in his academic labors, as his wife keeps a
mangle, and he is obliged to turn it for her.”54 The fact that men often
helped their wives with laundry work suggests that laundry was a strength-
intensive job that might have been better suited to men.
In spite of the strength required for laundry work, men rarely did it.55
The 1841 census lists only 583 males in laundry work, as compared to

BPP 1833 (450) XX, C1, pp. 50“1. Report of Mr. Drinkwater, evidence of William Felkin.
BPP 1845 (609) XV, p. 101.
Mary Puddicombe of Exeter, Women and Children in Agriculture, p. 109.
Malcolmson, English Laundresses, p. 26. 52 Mayhew, London Labor, vol. I, p. 150.
Ibid., vol. III, p. 307.
Manchester Statistical Society, 1837, quoted in Phil Gardner, The Lost Elementary Schools
of Victorian England (London: Croom Helm, 1984), p. 122.
Ruth Schwartz Cowan, More Work for Mother: The Ironies of Household Technology from
the Open Hearth to the Microwave (New York: Basic Books, 1983), p. 20, noted the fact
that women were assigned this strength-intensive task in the early US.
Explaining occupational sorting 163

50,706 females, so that only about 1 percent of laundry workers were
male. Laundry work may have appealed to women because it was often
done at home, so that it was easy to combine with domestic duties such
as child care (see Section II). In the later nineteenth century when
factory-like steam laundries replaced laundry work done in the home
there was some increased use of men in laundries, but women still did
much of the work, and did some of the heavy work. The employment of
men increased from 1 to 5 percent of the workforce, and many men were
employed in loading and unloading the washing machines, which was a
heavy task.56 Women were employed for ironing, which was a light task,
but were also employed calendering, which was a heavy task.57 While
there was some movement toward the employment of men, women still
dominated the workforce. The prominence of women in laundry work, a
task requiring strength, suggests that this may be one case where gender
roles were strong enough to overrule comparative advantage. Gender
roles associating women with housework were so strong that women did
the laundry even when the work was better suited to men™s strength.
Comparative advantage encouraged husbands to help their wives with
the most arduous tasks, but laundry as a whole remained a female
industry. It should be noted, though, that when women worked as
laundresses, they received relatively high wages. Even accounting for the
long hours, wages in laundry work were about 50 percent higher than in
agricultural work.58 When women did strength-intensive work, they
received a wage premium.
Except for laundry work, which should have employed more men, all
the patterns of occupational sorting that we have examined here comply
with the predictions of the sorting models. Workers in strength-intensive
occupations were primarily men, and workers in occupations requiring
little strength were primarily women. The gender division of labor was
one of general tendencies, not absolute barriers. A few women were
found in strength-intensive tasks such as framework knitting, suggesting
that women could do such work if they found it advantageous.

2. Unusual women
When discussing Model A above, I assumed that all women had less
strength than all men. While this is usually true, there will be cases when
it does not hold. There will be a few exceptional women who are
stronger than some men. An interesting test of the labor market, then, is
what happens to these women. If the competitive market is functioning
well, then exceptional women ought to be able to do the same work as

56 57 58
Malcolmson, English Laundresses, pp. 140“1. Ibid., pp. 143, 145. See Chapter 2.
164 Gender, Work and Wages in Industrial Revolution Britain

men and earn as much as men. We do occasionally observe unusually
strong women doing work typically done by men, and being well paid for
it. In seventeenth-century Oxfordshire, Alice George claims that “she
was able to have reaped as much in a day as any man, and had as much
wages.”59 In the 1840s Mary Haynes, a farm laborer from Wiltshire,
claimed: “I am a good reaper, as good as many men; and in harvest,
when I have worked by the job, I have earned 2s., sometimes 2s.6d.
a-day.” We don™t know how her harvest wages compared to the wages of
male reapers, but we do know that Mary earned more at reaping than her
husband™s normal summer wage of 10s. per week, or 1s.8d. per day.60
In coal-mining, the work was by piece-rate, and employers took little
interest in who their workers were. One employer told the parliamentary
investigators, “I exercise no control over them. I merely pay the men for
the coal which they bring to the bank.”61 Miners were paid by weight
for the coal they brought up, and paid their assistants themselves. This
gave the miners freedom to allocate the different mining tasks among
themselves. This suggests that Model A should describe the allocation of
labor in mining. The allocation of tasks in mining ¬ts the prediction that
men should do the more strength-intensive jobs. Hewing the coal
(wielding the pick) required more strength than pushing or drawing
containers of coal through the tunnels, so in general men hewed, and
women and children “hurried” (transported the coal).62 This division of
labor holds true in all collieries where women worked. Jane Humphries
notes that “Not only was hewing a male job, it was reserved for adult
males.”63 This also ¬ts the model because boys had less strength than
grown men, and would be expected to choose the less strength-intensive
jobs. However, this allocation of work seems to have been based on
individual characteristics rather than simply on the sex of the worker.

Journal of John Locke for March 1, 1685. Quoted in Roberts, “Sickles and Scythes,” p. 19.
When interviewed by Alfred Austin, Mary Haynes was a widow. Her husband died “not
long ago” and had earned 10s. per week in the summer and 9s. per week in the winter.
Women and Children in Agriculture, p. 69.
BPP 1842 (381) XVI, p. 243. Humphries interprets this quote as an attempt by the
employer to shift responsibility for the treatment of child workers from himself to the
adult male worker. Jane Humphries, “Protective Legislation, the Capitalist State, and
Working Class Men: The Case of the 1842 Mines Regulation Act,” Feminist Review 7
(1981), pp. 11“13. While the piece-rate payment system may not clear the employer of
all responsibility, it does suggest that work groups had some freedom in how tasks were
allocated among the workers in their group.
Women did carry coal on their backs out of the mines, which would have required
strength. I do not have measures of strength requirements for hewing as compared to
carrying coal up ladders, but would note that women have a greater disadvantage in arm
strength than in leg strength. At age 20 females have 64 percent as much leg strength as
men, but only 46 percent as much arm strength. Lynch, “Muscle Quality.”
Humphries, “Protective Legislation,” p. 10.
Explaining occupational sorting 165

Though hewing was a male job, on rare occasions women worked as
hewers, and when women did hew they earned as much as men. In 1840
one surgeon gave evidence that, “The work these women do will be
generally hurrying; but sometimes women ˜get,™ and one I have known
to do so and she earned more than her husband.”64 This woman was
obviously on the high end of the female strength distribution. Her
unusual abilities allowed her to do “men™s work” and be paid accord-
ingly. Similarly, the women who worked as coal carriers in south
Gloucestershire, carrying coal from the pits to the city, were unusually
strong women: “A few Amazons yet practice the vocation of coal car-
riers, on their own account, from the pits into the city or suburbs,
rivaling the men in strength of sinew and vigor of lungs.”65 Women were
not barred from this occupation by arti¬cial barriers, but only a few
women were strong enough to pro¬t from it. As predicted, the distri-
bution of occupations seems to have been based on individual ability
rather than simply by gender.

3. Regional patterns
Sometimes regional differences in the gender division of labor are seen
as evidence that these differences were the result of custom rather than
market forces. For example, Pamela Sharpe points out that in some
villages women would reap or hoe, while in other villages they did not.
Sharpe is not convinced that economic models can explain these dif-
ferences: “The point here is not only should we be mindful of local


. 6
( 14)