1066 and all that: a millennium of Normans

On this 950th anniversary of the Battle of Hastings, what English chiffrephile could resist a look at the numbers of our Norman cousins over the intervening period – and specifically at the time of the Conquest? The topic has come up with surprising regularity on discussion boards and query sites, understandably as the relative resources of England and Normandy cannot have been without relevance in the events of 1066 and the subsequent unravelling of the joint realm.

It has long been apparent that England was the more populous of the two states whose half-century-old dynastic engagement erupted in the conflict that brought down the Anglo-Saxon monarchy. But by how much? Modern England's 130,000 sq km compare favourably to Normandy's 30,000, while England's 55 million inhabitants now dwarf Normandy's 3.3 million. But the magnitude of today's population gap is of comparatively late origin: when we get our first semi-official figures in the 18th century the Normandy region contained about 30% as many people as its now far faster-growing neighbour across the Channel, and there seems no likelihood that England's lead had ever been greater.

England's population history is now relatively well-documented, with the painstaking work of the Cambridge group showing a near-doubling from 2.8 million in 1541 to 5.2 million by 1700. Going back further we have the 1377 return of taxable adults, indicating a total of 2.3 million or more after the worst ravages of the Black Death, depending on the rate of exemption and evasion. And for the 11th century we of course have the Conqueror's own Domesday survey, variously thought to indicate a population of 1½-2½ million, nowadays generally taken to be around or above the middle of the range since Sally Harvey's identification of the large numbers unrecorded for their lack of manorial obligations.

For Normandy we have little to go on beyond the first French censuses indicating 2.5-2.6 million in the early 19th century, and various returns of 1½-2m in the 18th – the latter certainly underestimates as for the kingdom as a whole, but sufficient to suggest an adjusted 1.65 million around 1700 for the then généralités of Rouen, Caen & Alençon (a figure that as we shall see is itself likely to be too low). Before then there are isolated figures for hearths or tax receipts.

But it is possible to identify a likely trend extending back into the Middle Ages. Today's Normandy covers just over 5½% of the area of metropolitan France, with a similar share of national population. Two centuries ago its population share was significantly higher, as much as 8½% in 1801-11. The 18th-century returns suggest a proportion nearer 7½%, perplexingly as there seems no other evidence of exceptional intervening regional growth, recorded births and deaths instead indicating a slight relative decline from at least the 1760s: part of the answer may lie in the inclusion of some south-eastern districts in the généralité of Paris, or in fuller early census reporting than elsewhere.

What of the earlier period? Here Normandy's share of France's urban population seems to offer a guide to the likely trend. The contribution of Rouen, Caen and le Havre fell steadily over the modern period relative to 35 principal French cities, from 10% in 1550, 8½% in 1650 and 7½% in 1750 to 7% in 1800, 5¾% by 1850 and only 4¼% in 1900 – the abruptness of the 19th-century fall reflecting rapid growth in Paris, Marseille, Lyon and the country's other large industrial cities. Removing Paris from the calculation gives 13% in 1550, 11% in 1700, 10¼% in 1800, 9½% in 1850 and 8% in 1900, despite Le Havre's late start and rapid 19th-century growth.

While heavily influenced by Rouen's relative eclipse from the 16th century and not in itself conclusive – the 18th-century totals indeed suggest a contrary trend, which may in turn indicate a provincial under-estimate – the urban evidence strongly suggests that Normandy's share of French population in earlier centuries was at least no smaller than in 1700 even if the region experienced relatively slow urbanisation to 1830, which is possible as maritime activity shifted to the west coast and higher government functions became more centralised in Paris.

Going back still further, the hearth count of 1328 – after the 1315-17 famine but twenty years before the Black Death – yields around 300,000 hearths (usually considered to approximate to households) in the baillages of Rouen, Caen, Caux, the Contentin and Gisors, indicating a population of around 1½ million out of the 18 million estimated for the area of modern France. A similar figure is implied by the tax return of 1221, or possibly rather more again allowing for exempt groups. The 14th-century estimate suggests a Norman share of a twelfth of France, the 13th-century one a tenth or more as it seems improbable that French population remained unchanged in the interim.

What then of 1066 itself? Here we need to venture further assumptions about French and wider European population growth. It is generally considered that European population roughly doubled over the 11th-13th centuries, with faster increases in the east, centre and north, slower in the south and particularly the Balkans. For France with its Mediterranean attachments and Roman past, growth is likely to have been at best no higher than the continental average, indicating a population in 1066 of around ten million for today's area. This in turn indicates a Norman population of 830,000 or more, perhaps in excess of a million, to which we might add an allowance for Norman emigration in the following centuries.

Taking the population of Harold's England as around two million, Normandy is thus likely to have contained two-fifths to half as many inhabitants. So how did William win given that after a quarter-century of growing Norman influence under the confessor little is likely to have separated the two sides in battle tactics or technology? Part of the answer doubtless lies in his good fortune in facing an enemy who had only three weeks earlier fought off one dangerous invader and had then had to march 250 miles south – mere weeks after a 200-mile march north – to confront another. Harold's force should have been the more numerous, but on the fatal day barely matched his adversary's, with the respective armies amounting to under one per cent of Normandy's population and 0.4% of England's. Perhaps Harold erred in being too ready to give battle when he might have drawn the challenger from his coastal beachhead; but perhaps time was of the essence as many of his best warriors approached exhaustion.

Perhaps more importantly, the exercise suggests why England was so important to William – and so valuable to his heirs that they preferred to keep their new land as the seat of their power, even at the cost to their Angevin successors of losing Normandy itself to the rising French crown after only 138 years of on-off cross-Channel personal union. At a stroke the House of Robert had tripled the people and resources under its rule, winning in the process a territory free of even nominal obligations to the French crown.

For Normandy, 1066 was a high-water mark in the duchy's meteoric ascent, before incorporation as a mere French province and centuries of sluggish growth in contrast to the later explosion of English population, production and commerce. For the Anglo-Saxon state it was of course a more dramatic end, whose consequences remain with us to this day even as England prepares to close another chapter in its often prickly relationship with the Continent.

How many have lived? A short history of everyone

Looking for an appropriate topic to start this blog (which will take over the population side from my other pages), the best candidate seems to be that popular topic, "How many of us have ever lived?" The question appears to interest us beyond the old urban myth of more living today than have ever died. And the answer can tell us a great deal about the long-run evolution of human numbers.

The now classic response to the query is that of the Population Reference Bureau's Carl Haub, who in 1995 ventured a total of 105,472 million anatomically modern human beings born over the past 50,000 years, since raised by an additional couple of billion to cover intervening growth. His sensible conclusion was that only 5½% of those who had ever lived were alive in the present, a proportion that would now have risen to 7%, still far short of the reported claims of up to 75% which he so rigorously debunked.

Needless to say Haub’s estimate has been criticised, sometimes as being on the low side, at other times as too high. The author himself stressed the "semi-scientific" character of his speculation, hazarding a guess that the true number might be higher, though his approach – estimating the population in given periods and assigning a likely approximate birth rate in the light of probable mortality and growth – was a sound one in the absence of reliable data until comparatively recently. His allowance of only fifty millennia of Homo sapiens would today generally be doubled or even quadrupled, though as he observed the number of additional lives involved in such an extension would constitute only a tiny fraction of the total as global human population remained sparse – at most 5-10 million – until the dawn of agriculture little over ten thousand years ago.

A more fundamental problem with Haub’s figures is his assumption of an 8% annual birth rate until the start of the Christian era. This is certainly too high for the later part of the epoch as settled societies developed toward the 4-5% mortality of the Classical age, but its plausibility as a norm in earlier millennia is also questionable. As annual population growth over this vast period averaged well under 0.1%, the higher figure implies a crude death rate of barely under 8% with an average expectation of life at birth of just over 12½ years, infant mortality of around a half, and 70% of girls dying before childbearing age, placing an enormous reproductive burden on the survivors. The total fertility rate – live births per woman making it through reproductive age (conventionally 15-45, now increasingly 15-50, but possibly shorter with the poorer nutrition and health of earlier times though modern least-developed country patterns suggest otherwise) – becomes difficult to calculate at such extreme levels, but is likely to be in in the region of 9½-10, at the high end of the range for the strongly pro-natal Hutterite community whose 9-10 births are commonly thought to approach the biological maximum even for such a relatively healthy and well-nourished population. While such numbers may have characterised pre-agricultural populations in particular periods or areas, births at this level seem unsustainable worldwide over hundreds or thousands of generations.

If 8% is improbably high for annual births (and deaths), what might be appropriate in its place? High-fertility populations until recent times have regularly exceeded a birth rate of 5% and sometimes attained completed fertility of 8, with Niger (a society with a strong cultural preference for large families) maintaining a CBR of 5.7% into the 1960s and fertility of around 7½ to the present. The TFR indicated by a mesolithic 6% birth and death rate – still consistent with life expectancy of only 16½ years but now giving around 7¼-7½ births per woman completing childbearing age – is thus not impossible. An average birth rate of around 7% for the long prehistoric period incorporates the notion that childbearing was at times higher, at others lower.

The cumulative impact of a difference between a birth rate of 8% over the millennia before the Roman and Han empires and one descending gradually to 5-6% dwarfs the addition to overall human numbers from an additional 150,000 years of hunting and gathering. But another serious question mark hangs over Haub’s adoption of the common 300 million as his population for the year 1 CE. There is really no foundation for this figure, which still surfaces regularly in UN projections but seems to rest on outdated notions of African and Indian stagnation over centuries. We now know that like Europe and China, sub-Saharan Africa was in the throes of internal agricultural colonisation for at least two millennia before the arrival of Europeans, and there is no reason to assume that India alone of the world’s major regions was becalmed in centuries of demographic inertia. While McEvedy's 170 million is likely to be an underestimate, 210 million seems a reasonable allowance given 60 million in China and still probably under 40 million in Europe, with India at perhaps something around 45-50 million.

This point is important because Haub's 1 CE benchmark underpins the biggest single element of his total, the 46 billion assumed to have been born over the preceding eight millennia. Adjusting both the birth rate and the final population as outlined above reduces that portion by three-fifths, even after raising the number of people – and hence also of births – at the start of the period to take account of progress in the mesolithic. Add to that figures for 1200 and 1750-1900 that seem high by around 50 million (although the intervening estimate for 1650 may conversely be low by 70 million) coupled with excessive birth rates down to at least 1650 (though those for 1750-1900 should be raised, offsetting the overstatement of population), and the case for a lower total seems overwhelming.

The table below illustrates the results of such a reworking, with the latest UN estimates appended to Haub's original estimate to bring the numbers down to 2015. The numbers of births differ little between the two series after 1650, but my revisions for the preceding ten millennia bring the total down by 30 billion to under 80 billion, still a formidable sum to contemplate: on this showing those living today would be 9.3% of all those ever born, up from 1% in 1650, 2% in 1860, 3% in 1930, 5% in 1970 and 8% in 2000: looking ahead, the ratio should peak at 12% toward the end of this century and be back down below 11% a couple of centuries later, still well short of anything approaching a majority as Haub correctly surmised.

The exercise is of use not just as a piece of sterile enquiry, but in revealing population as a complex of dynamic flows rather than mere stocks of people. In particular, it invites us to consider the conditions under which societies of the past survived and eventually grew. And it may explain something of present-day reproductive behaviour: if birth rates of 8% were as unsustainable as calculations of completed fertility seem to indicate, a similar prolonged death rate might have imperilled the very survival even at greatly reduced numbers of communities at the margins of subsistence. Lingering large-family preferences in some regions may go deeper than cultural traditions or the need for children to provide support in older age, perhaps offering an indication of why some fertility levels have been harder to bring down than others.