A factory is a strange thing to invent, because for most trades in most centuries it was cheaper to leave the worker at home and collect the product.
The previous lessons have built the machines, the fuel and the transport. What they have not explained is the building. It is tempting to treat the factory as an automatic consequence of machinery, since big machines need big sheds, but some of the earliest factories held no powered machinery at all and one of the earliest powered ones produced no imitators for fifty years. This lesson asks why work moved out of the cottage, tests three answers against each other, and ends with the awkward fact that in 1851 most British workers had still never been inside one.
Work before the factory
The system the factory displaced is called putting out, and it was the ordinary way of organising manufacturing across Europe.
A merchant bought raw wool, cotton or yarn and distributed it to households over a wide district. The household worked it on its own equipment, in its own time, and returned the product to be paid for by the piece. The merchant owned the material and not the tools; the worker owned the tools and the hours and not the material. Nobody supervised anybody. In the West Riding, the Lancashire cotton districts and the Nottinghamshire hosiery villages this supported hundreds of thousands of people and had done so for generations.
Its virtues are worth stating before its faults. It needed almost no fixed capital, so a merchant met a slump by putting out less work rather than by idling a mill he had paid for. It drew on labour that could not have gone anywhere else, since a woman with small children could spin between other tasks and could not have left the house to do it. And it fitted the agricultural year, letting the same family harvest in August and weave in January.
What was wrong with putting out
The complaints of the merchants are documented in detail, because they took them to Parliament repeatedly, and they are not mainly about speed.
The first is embezzlement. When material worth more than a week's wage is in the hands of a worker the merchant cannot see, some of it stays there. Weft was retained, wool was substituted, yarn came back short weight and cloth came back wetted to make up the weight. Parliament legislated against it repeatedly, and in 1777 the Worsted Act created something genuinely unusual: a salaried inspectorate, funded by a levy on the manufacturers themselves, with power to search premises across the worsted districts.
The second is quality. A merchant collecting cloth from four hundred households receives four hundred slightly different cloths, and a piece that fails at finishing has already had everyone's labour spent on it. The third is time. Work went at the household's pace, so it stopped for harvests, fairs, funerals and Saint Monday, the widely observed custom of not working on the first day of the week after a Sunday spent drinking. A merchant with an export order could not promise a date he did not control.
Example. A worsted manufacturer puts out £10,000 of wool a year and earns a gross margin of 10 per cent on the trade. Contemporaries claimed that something like 5 per cent of material issued was retained by the workers. What does that do to his business, and why did competing firms agree to pay for a shared inspectorate rather than hire their own?
His margin is £1,000 and his losses are , so embezzlement takes half his profit, which is enough to explain the vehemence of the petitions. The reason he will not simply hire his own inspector is that the deterrent is a public good among the firms. An outworker frightened out of retaining wool stops retaining everybody's wool, and since workers in a district spun for several merchants in turn, a firm paying for enforcement is buying protection for its rivals. Left to themselves each firm underspends on monitoring and all of them lose, which is exactly the situation a compulsory levy fixes. Hold on to the logic, because it points somewhere else too: if monitoring at a distance is expensive and hard to share, then bringing the work to a place where monitoring is nearly free solves the same problem by a different route.
Now you. Why could the merchant not solve embezzlement by simply paying a higher piece rate?
Answer
Because the temptation does not depend much on how well the work is paid. Retaining four ounces of wool is worth the same to a worker on a high rate as on a low one, and the chance of being caught is unchanged, so a higher rate raises the merchant's costs without altering the worker's calculation. A high rate helps only by making the job worth keeping, and it can do that only if the merchant can identify who cheated and stop giving them work, which is the thing he cannot do when the yarn comes back through intermediaries from four hundred houses. Reputation did work in tight, stable communities, and putting out survived longest in exactly those. What defeats the system is scale.
Cromford
Richard Arkwright's mill at Cromford in Derbyshire, opened in 1771, is where the institution begins, and it begins for a reason the previous section does not supply.
Arkwright's water frame, described in an earlier lesson, could not be worked by hand. It needed continuous rotary power at a steady speed, and in 1771 that meant a water wheel. A wheel is one indivisible lump of capital, arriving with a weir, a leat, a pond and a train of gearing, and none of it can be cut into pieces and delivered to cottages. The machines had to come to the power.
What Arkwright built around that necessity is the genuinely new part. The mill ran day and night in two twelve hour shifts, and employed around 800 people by the end of the 1780s, perhaps two thirds of them children. Because Cromford is a hamlet with no labour force in it, he built the workforce too: cottages, a market, a chapel, an inn, and a preference for hiring large families in which the children worked the frames and the father was found other work. He enforced attendance by rules, fines and dismissal in a way no merchant putting out work had ever been able to.
Example. A mill wheel develops about 10 horsepower and a water frame absorbs about a quarter of a horsepower. How many frames does one wheel serve, and what does that argument prove?
It serves frames, and that is the whole case for gathering: the wheel, the weir and the gearing are one purchase that cannot be divided forty ways, so the forty frames must stand within a few hundred feet of it. What the argument proves is narrower than it looks. It proves that powered spinning had to be centralised. It says nothing about the trades that gathered into large workshops with no prime mover at all, and it does not even show that power is sufficient, because John Lombe's silk throwing mill at Derby, five storeys high, water powered and employing hundreds, opened in 1721 and produced no imitators for half a century. Power made the factory necessary in one industry. Something else has to explain why the form spread to industries that did not need it.
Now you. Josiah Wedgwood's Etruria works, opened in 1769, gathered several hundred potters into one site to throw, mould, fire and paint by hand. What does that case do to the power explanation?
Answer
It shows that power is not necessary, only sometimes sufficient. Nothing at Etruria required a prime mover, and Wedgwood's own reasons were three others: division of labour, since a workman who does one operation all day does it better than one who does six; quality control, since a pot that fails at the glaze has already consumed everything spent on it; and regularity, since an export customer wants a matched dinner service on a date. Wedgwood divided the work into specialised shops, kept a record of who made what so faults could be traced to a hand, and rang a bell. The factory arrives here for reasons of information and organisation rather than of machinery, so an explanation resting on the water wheel is answering a question about one industry and calling it an answer about the economy.
Three explanations, and how to tell them apart
The disagreement is old and unusually sharp, and it sets out as three claims that predict different things.
The technological account, argued most forcefully by David Landes, says the factory follows the machine: where production needs a central source of power, or machines too large for a household, work must be centralised and everything else follows. It predicts that factories appear in an industry at the same date as its prime mover and not before.
The disciplinary account, put by Stephen Marglin in 1974 in an article titled "What Do Bosses Do?", says the factory won not because it produced more per hour but because it let the employer control the pace, the hours and the division of the product. What was being sold was supervision, and its purpose was to capture for the owner a surplus the outworker had kept as leisure. It predicts factories with no new machinery in them, and rules about attendance rather than technique.
The transaction cost account says the factory solves the information problems of the previous section: embezzlement, adulteration, unverifiable quality and undeliverable dates. It predicts that centralisation comes earliest in trades where the material is valuable, concealable and easy to spoil, whichever way the technology is running.
The evidence does not award the argument to one of them. Cotton spinning fits the technological account almost perfectly. Pottery, silk throwing before power, tailoring and the Birmingham metal trades fit the other two, and Wedgwood's rule book fits the disciplinary account so well that Marglin quotes it. The workable position is that the three answer different industries, and that once the factory existed for any one reason it collected the advantages of the other two for nothing. That is how institutions usually spread: invented for one purpose, kept for another.
The clock
Whatever brought people into the building, what happened once they were there was the imposition of time, described first and best in E. P. Thompson's 1967 essay "Time, Work-Discipline and Industrial Capitalism".
His argument is that pre-industrial work is task-oriented. A task takes as long as it takes, the day has no edges, and the line between work and life is weak. Factory work is clock-oriented: the employer buys hours rather than output, so the hour must be measured, and both parties begin to argue about it. Workers who had never owned a watch acquired them, employers who owned the only clock were accused of tampering with it, and the rule books fined lateness in minutes.
There is a hard economic reason underneath, and it explains why the owner cared so much.
Example. A mill holds £5,000 of machinery, on which interest and depreciation come to 15 per cent a year. It works the standard cotton week of the 1830s, twelve hours on five days and nine on Saturday, for fifty weeks. What does an hour of that machinery cost, and what happens if the mill works a forty hour week instead?
The annual capital charge is . The standard week is hours, so the machinery runs hours a year and each hour carries , which is 4 shillings 4 pence. At forty hours it runs 2,000 hours a year and each hour carries 7 shillings 6 pence, a rise of 73 per cent in the capital cost of every hour of production, with nothing about the workers, the cotton or the machines changed. A putting-out merchant whose capital sat in bales of wool lost nothing when a weaver took Monday off, because wool keeps; a mill owner whose capital sits in a building loses the charge on it every hour the building is dark. Time became expensive to the owner before it became expensive to anybody else, and the fines, bells and locked gates follow from that arithmetic.
Now you. The same mill installs gas lighting and runs two shifts, taking its weekly hours from 69 to 138. What happens to the capital charge per hour, and why did the length of the working day then become a matter for Parliament rather than for each employer?
Answer
The machinery now runs 6,900 hours a year and each hour carries , 2 shillings 2 pence, exactly half what it carried before. The incentive to run long is therefore enormous and never diminishes, since every extra hour spreads the same fixed charge further. The political consequence follows from the position of an employer who would like to work shorter hours: if he shortens them alone his costs per piece rise and a competitor who did not undersells him, so he cannot move first even when he wants to. A cost every producer would like to escape and none can escape alone is the standard case for legislation, which is why the working day was settled by statute in the 1830s and 1840s rather than by agreement, as a later lesson takes up.
Who was inside
Who was in the building follows from the machines and from the wage structure, and it is not what a modern reader expects.
Arkwright's frames required attentiveness, small hands and almost no strength, so they were worked by children and young women, who were also far cheaper than adult men. At Cromford and Styal much of the workforce was parish apprentices, children handed over by poor law authorities in London and elsewhere and bound until adulthood. Where the machine did need an adult, as the mule did before the self-actor, the spinner was a man who hired his own assistants, often his own children, out of his own wage. That family hiring system is why so much testimony about child labour describes a father driving his own son at the machine, and why the first attempts to limit children's hours were opposed by some of the parents.
The other half of the picture is that factory work paid. A young woman in a mill earned more, in cash of her own, than she could earn in service or at a wheel, which is why mills that were free to recruit generally could. Pretending the mill was simply worse than what came before makes the history unintelligible, because people walked to these places.
How much of Britain was a factory
The last correction is one of scale, and it repeats the lesson of the revised national accounts earlier in the course.
Around half a million people worked in cotton factories across the United Kingdom at the middle of the century, against an occupied population in Great Britain of about nine and a half million at the 1851 census, so roughly 5 per cent. Agriculture still employed more men than anything else, over a million women were domestic servants, and most manufacturing was still done by hand in small workshops. Raphael Samuel made the point at length in 1977 by counting the trades Victorian Britain actually ran on: building, tailoring, shoemaking, furniture, food processing, the Sheffield cutlery shops and the Birmingham metal trades, almost none of them mechanised, several of them growing because cheap factory-made inputs gave them more to work with.
The factory in this period is therefore a form that had conquered two or three industries completely and the rest of the economy hardly at all, while setting the terms on which everyone would eventually work. It is also where the effects of industrialisation on ordinary bodies first became measurable, and whether those effects were good or bad is the most argued question in the subject. That argument, and the four kinds of evidence that refuse to agree with each other, is the next lesson.