Almost everything you would call normal about modern life depends on output per person rising year after year, and for most of recorded history it did not do that at all.
This course is about the sharpest change in how humans live, so it has to start with the condition that changed. That condition is not poverty as such. It is the absence of a trend: across the four hundred years between a peasant in Oxfordshire in 1300 and a labourer in the same county in 1700 there is no direction in the data, only oscillation. The first thing to establish is that this is a fact about the evidence rather than a slogan, and the second is why it should have been so.
What the numbers do, and do not, do
Angus Maddison's reconstruction of world income, the most widely used long series there is, puts output per head at roughly 450 international dollars around the year 1000 and roughly 666 in 1820, both in the same 1990 prices. That is a rise of 48 per cent spread over eight hundred and twenty years, which works out at 0.048 per cent a year. Between 1820 and 2003 the same series rises by a factor of 9.8, at 1.25 per cent a year. The second number is about twenty-six times the first.
These are estimates built from fragmentary evidence and argued about vigorously, so treat the precise values as indicative and the contrast as robust. The contrast does not depend on the level of the early figures at all, only on the fact that pre-modern series have no slope and modern ones do.
England has the best long series in the world, because English institutions happened to write things down and keep them: manorial accounts, cathedral building records, Oxford and Cambridge college accounts, wage assessments by justices of the peace. Stephen Broadberry, Bruce Campbell and colleagues used them in 2015 to rebuild English national income annually from 1270, and Gregory Clark has assembled day wages for building craftsmen and labourers over much the same span. Both series wander. Neither climbs.
The most quoted single fact from this literature is the one that startles people: on the Phelps Brown and Hopkins index of builders' real wages, begun in the 1950s and still the reference point, the highest real wage before the nineteenth century is in the 1450s, and English building workers did not reliably exceed that level again until the 1870s. Four centuries of Renaissance, empire, banking, navigation and printing left the man who carried bricks no better off.
Malthus states the mechanism
The reason was written down before anyone had the data, by a clergyman who was arguing about something else. Thomas Robert Malthus published An Essay on the Principle of Population anonymously in 1798, aimed at the optimism of Condorcet and Godwin, who expected human society to improve without limit. Malthus said the limit was arithmetic.
He starts from two postulates he takes as obvious: food is necessary, and the passion between the sexes is necessary and will continue. From these he claims that population, when unchecked, grows geometrically while subsistence grows at best arithmetically. His own worked illustration is worth doing exactly as he does it. Take population doubling every twenty-five years, which is 2.81 per cent a year, and take food supply increasing by one unit per twenty-five year period. After nine periods, which is two hundred and twenty-five years, population stands at and food at . Malthus prints that ratio, 512 to 10, and rests his case on it.
The doubling figure was not invented. Malthus took it from the American colonies, where land was abundant and the check was weak. The United States census, which he could see the first two rounds of, gives 3.93 million in 1790 and 5.31 million in 1800, a growth rate of 3.06 per cent a year, which doubles a population in twenty-three years. By 1820 the count was 9.64 million, two and a half times the 1790 figure in thirty years. Malthus's premise was, for the case he chose, correct.
Example. Malthus's argument is often dismissed on the grounds that food supply obviously did not grow by a fixed increment. Does the argument survive if food grows geometrically too, at 1 per cent a year?
It survives completely, and this is the point most often missed. Suppose population grows unchecked at 2.81 per cent and food at 1 per cent. The ratio of people to food then grows at about 1.79 per cent a year compounding, which doubles the number of mouths per loaf in thirty-nine years and multiplies it by 5.9 in a century. Nothing in the argument requires food growth to be arithmetic. It requires only that unchecked population growth be faster than food growth, for any reason, and since population growth is a percentage of a stock while food growth depends on land, that is a safe bet in an agrarian economy. Malthus chose the weakest version of his premise and it made him easier to refute than he should have been.
Now you. If food supply could be made to grow at 2.81 per cent a year, matching unchecked population, would the trap be broken?
Answer
Not on its own, and seeing why is the whole lesson. Matching the growth rates would hold output per head constant, not raise it, so the population would simply be larger at the same living standard forever. Breaking the trap requires food, or output in general, to grow faster than population for a sustained period, and to keep doing so after living standards have risen. The historical escape has two halves: productivity that grows without limit, and a population response that weakens rather than eating the gain. Malthus knew only economies where the second half failed, and he had no reason to expect either.
The trap as a feedback loop
Stated as a mechanism rather than as arithmetic, the model is a loop with a stable resting point, and the loop is what matters.
Suppose the wage rises above the level at which a family reproduces itself exactly. Two things then happen, and Malthus names both. The positive check weakens: better fed people die less, especially infants, so more children survive. The preventive check relaxes too: couples with land or a wage marry earlier, so more of a woman's fertile years fall inside marriage and more children are born. Population therefore grows. But land does not, so each worker now farms a smaller share of a fixed quantity of the scarce input, and output per worker falls. The wage falls with it, until it returns to the level where births exactly balance deaths.
Run the loop from the other side and it works the same way. A wage below replacement raises mortality and postpones marriage, population shrinks, land per worker rises, the wage recovers. The resting point is stable in both directions, which is what makes it a trap rather than a tendency.
The consequence that shocks a modern reader follows immediately. In this world, a permanent improvement in technology does not raise living standards in the long run. It raises the population. A better plough, a new crop, a drained fen: each raises output per acre, raises the wage briefly, and is then absorbed entirely by extra people, leaving the wage exactly where it was and the country more crowded. Technical progress in a Malthusian economy is converted into population, not into income. The model also inverts the sign of things we think of as good and bad: a plague raises wages, and peace and hygiene lower them. Malthus, who was a decent man and knew how this read, said so anyway.
The Black Death, which is the experiment
The awkward thing about a model like this is that you cannot run it. Except that in 1348 something ran it.
Plague reached England in the summer of 1348 and killed, on the best current estimates, something close to half the population within three years, with further outbreaks in 1361, 1369 and 1375 preventing recovery. England had roughly 4.8 million people on the eve of the plague and something near 2 million by the middle of the fifteenth century, which is a fall of about 58 per cent. Land, buildings, mills and livestock were mostly untouched. This is as close to a controlled shock to the labour-to-land ratio as history offers.
The model predicts that wages should rise sharply and stay high while the population is small, then decline as it recovers. That is what happened. Real wages roughly doubled over the century after the plague, reaching the fifteenth century peak already mentioned. Landlords tried to legislate them back down: the Ordinance of Labourers in 1349 and the Statute of Labourers in 1351 fixed wages at their pre-plague levels and made refusing work at those rates a criminal offence. The statutes were enforced, extensively, and they failed, which is a useful reminder of what a price control is up against when the underlying scarcity is real.
Then the recovery. English population climbed back through the sixteenth century, and real wages fell as it did, roughly halving between the 1450s and the 1610s on the Phelps Brown and Hopkins index. Nothing about English agriculture got worse. There were simply more people on the same land, exactly as the loop requires.
Example. A landlord in 1360 complains that his labourers demand double the customary wage and will leave if refused. His neighbour argues that a stronger statute, properly enforced, will fix it. On the model, who is right?
Neither, in the sense each intends. The wage has risen because the ratio of workers to land has changed, and no statute can change that ratio. What enforcement can do is redistribute: a landlord with the local justices on his side holds his own wages down and pushes the shortage onto someone else, which is why the statutes were both vigorously used and nationally ineffective. The real test is what a landlord does once the law has failed, and the record answers it, which is to convert arable land to sheep, since sheep need far less labour per acre than grain. That is a genuine adjustment to the new price of labour, and exactly the kind of substitution that will matter enormously later, when English labour becomes expensive again for a different reason.
Now you. Explain why the same model predicts that improved sanitation, on its own, would make a pre-modern population poorer.
Answer
Sanitation reduces mortality without touching the quantity of land or the productivity of farming. Fewer deaths at a given wage means the population grows, which lowers land per worker and therefore output per worker, until mortality rises back to the level where births and deaths balance. The population ends larger, the wage ends lower, and the eventual death rate is the same as before, because the resting point is set by the balance of births and deaths rather than by the disease environment. The conclusion feels monstrous, so be precise about its limits: it holds only in a closed economy with fixed land and no sustained productivity growth, and it fails the moment output grows faster than population. That is what makes the model useful. It tells you exactly which assumption a real escape has to break.
Subsistence is not the same as starvation
One correction is needed before leaving the model, because "subsistence wage" invites a false picture of a continent permanently at the point of death.
The resting point of the loop is wherever births equal deaths, and that level depends on what a society does about marriage and childbearing, not on physiology. Northwestern Europe had an unusual arrangement that historians call the European Marriage Pattern: couples formed independent households at marriage rather than joining the husband's family, which meant they had to accumulate the means to run one first. Women married in their mid-twenties, late by world standards, and a substantial minority never married at all, which held the birth rate well below its biological maximum.
A society with a strong preventive check settles at a higher wage than one without, because it takes a better living standard to trigger the population growth that pushes the wage back down. That is why England could be Malthusian and comparatively well off at the same time, and why English labourers ate wheat bread and drank beer while their equivalents further east ate rye and worse. The trap fixes the absence of a trend. It does not fix the level.
Where the model starts to fail
The last section is the hinge of the course, and it is a piece of evidence rather than an argument.
Reconstructed from parish registers by Edward Wrigley and Roger Schofield, the English population stands at about 5.77 million in 1751, 8.66 million in 1801 and 16.74 million in 1851. That is a population nearly tripling in a century on an island whose acreage did not change, and the second half of it runs at a rate no English century had come near.
Example. Turn that English population growth into a doubling time and compare it with the American rate Malthus called unchecked.
From 8.66 million in 1801 to 16.74 million in 1851 is a factor of 1.93 in fifty years. The annual rate is , so 1.33 per cent a year, and a population growing at that rate doubles in years. Malthus's unchecked American benchmark was a doubling every twenty-five years, so England was running at roughly half the frontier rate. That is the useful comparison, because it shows England was not at the biological maximum: the checks were still operating, just far more weakly than in any previous century. A country at half the unchecked rate for a hundred years still ends with four times the people, and England very nearly did.
Now you. Do the same for the previous half century, 5.77 million in 1751 to 8.66 million in 1801, and say what the two figures together suggest about when the change began.
Answer
The factor is 1.50, so the rate is , or 0.82 per cent a year, doubling in 85 years. So growth in the second half of the eighteenth century was already historically fast and then accelerated by a further two thirds after 1801. The acceleration matters for dating: population growth is well under way before the machinery in the middle lessons of this course is doing anything at a national scale, which rules out any story in which factories cause the population growth directly. Something loosened the checks first, and a later lesson goes after it.
The loop says that wages should have collapsed. They did not. They were flat for a long and contested stretch, which a later lesson takes apart carefully, and then they rose, and by the 1870s they had passed the fifteenth century peak and kept going. Population and living standards went up together, which is the one thing the model forbids.
Malthus published in 1798, at the precise moment the mechanism he described was ceasing to apply to his own country. He was not wrong about the past. Nothing in the previous five hundred years of English data contradicts him. He was writing the definitive account of a world that was, just then, ending.
That is the fact to be explained, and it is worth being clear about how strange it is: not that people got richer, but that they got richer while multiplying. Before explaining it, the next lesson does something less exciting and more necessary, which is to find out exactly how fast the escape actually happened, because the answer turns out to be much slower than the word revolution suggests, and that slowness is itself a clue.