Illustration by Richard Fairhead

The measure of all things: Britain's census

Establishing statistical truths is never a neutral act. Nothing exemplifies that better than our national census
October 6, 2021

Back in March, those of us in England, Wales and Northern Ireland (Scotland waits until next year) had to fill out our census forms. Although some will have sent theirs off in the post, the government’s “digital-first” approach meant that most did it online for the first time instead—while others still handed over intimate information on their doorstep to an officer with a purple-striped hi-vis jacket and clipboard. In the strange circumstances of the pandemic, we were maybe grateful for the company. Meanwhile, as we sit and wait for the results of this vast operation to be published next March, many of us might catch ourselves wondering: what, really, is this thing all about?

For some, the benefits of a comprehensive count—secured with the threat of a criminal record and a £1,000 fine for non-compliance—are self-evident. How can any government claim to make decisions on behalf of its people if it does not even know who they are, what they believe or what they do for a living? In order to be managed we need to be measured. But such arguments will never persuade some individualists, such as the Conservative MP who in 1950 bemoaned the first census in 20 years (the war had seen the one planned for 1941 cancelled) as “the cause of annoyance to a considerable number of people.” Such sceptics always need specific practical arguments for enduring such annoyance. Fortunately, there are plenty available.

The planning and funding of public services, for example, depends on how many people live in each place. The census also presents a chance to reveal the extent of society’s underlying ills, without spin. Other official surveys exist, some of which dig deeper into areas like income, work and experience of crime. But, unlike the census, they are subject to sampling error—and in any case rely on the census itself to calibrate various parameters. Certainty matters. Once you’ve counted all the people who are unemployed or homeless or doing unpaid care work, you know the real number, and then the question becomes: what are you going to do about it?

When the chaos of life is rendered into orderly truth “which can be stated numerically and arranged in tables,” as the Statistical Society of London puts it, this has a power over decision-makers that anecdote alone does not possess. The census dispenses with hunches, feelings and conjecture, leaving us with that coveted prize: “the facts.”

Or does it?

All kinds of choices are made in deciding what sort of data to carve out of reality—choices that reflect which problems we want to highlight, and which we’d rather not know about. Choosing what we count says a lot about a society’s priorities. This idea is so ingrained as to be idiomatic—what “matters” is what “counts.” But in the 30 minutes spent filling in our census forms, did we stop to think about what questions we were being asked, and why? Are we counting what matters—or does the act of counting it make it matter?

Cooking the books

Collective counting has been around almost as long as “civilisation.” It’s only three books on from Adam and Eve that you get to the Bible’s Book of Numbers, in which the Israelite tribes are tallied before their march to the Promised Land. As Boris Starling outlines in The Official History of Britain (2020), evidence of census-esque experiments dates back to at least Ancient Egypt; the Sumerians and Babylonians too are thought to have adopted some form of “citizen registration.”

The term “census” is derived from the Latin word censere, “to assess,” and the Romans were big on it, with formal surveys of their republic and empire for over 400 years. It was compliance with the census of Judea that—if the Gospel of Luke is to be believed—led to Jesus being born in Bethlehem.

The Domesday Book of 1086 was completed by William I, 20 years on from his Norman invasion. Conducted by Norman commissioners to discern the new realm’s tax base, it acquired its name from the native English who likened its final word on taxation to God’s last judgment. And although it tried to account for every village from which revenues might be extracted, the Domesday Book was neither neutral nor comprehensive: the only people who counted were (almost invariably male) landholders.

Time and again, apparently neutral tallying has in reality been about consolidating political power. In the United States, one clause of the new republic’s constitution mandated a census every decade, beginning in 1790—but due to another clause, the slave residents that it tallied would count only as three-fifths of a “free individual” when it came to determining congressional representation. Since the slaves themselves carried no political weight, the effect was simply to boost the influence of their white Southern masters in Congress and the electoral college. As James Madison put it, the effect of the three-fifths calculation was to give the Southern states influence “on the score of the Negroes.”

This ugly history has had lasting consequences on the way the American state continues to measure and think about its own society. But racial counting isn’t always a bad thing. This information proved to be crucial for civil rights campaigners in the 20th century, who used it to highlight how large the African American community was, and how sorely short-changed it had been on all sorts of statistical indicators.

In continental Europe, by contrast, countries that fancy themselves “racially blind” regardless of how racist their societies might be have developed—as Sunder Katwala put it in our August/September issue—an “outdated allergy” to working with race data. In France, its collection is banned. More generally, when censuses look away from race, states end up pretending that all questions about ethnicity can be dealt with by asking after “country of parental birth” instead. As well as ignoring prejudice about skin colour, this results in practical problems when it comes, for example, to keeping tabs on the race gap in the coronavirus death toll. Information can certainly reinforce the power of those who already have it; but it can also be empowering for those who don’t.

Some sort of compulsion

In Britain, the first official census took place in 1801. All previous attempts to organise one had run into stern libertarian resistance: “a national census would molest and perplex every single family in the kingdom,” and would “divest us of the last remains of our birthright,” said William Thornton, the MP for York during a debate in the 1750s.

The mood changed with the prospect of war with Napoleonic France, whose armies were drawn from what was claimed to be Europe’s most populous country. Two questions were suddenly urgent: are we too few to fight, or too many to feed? But everyone—the government included—was in the dark about the answers, because of all the unprecedented changes that industrialisation had unleashed.

In the absence of organised counting, there was no check on wild—and contradictory—speculations, many captured in Roger Hutchinson’s The Butcher, The Baker, The Candlestick-maker (2017). One prevailing assumption was that Britain’s population was on the decline. In the 1770s, the Welsh theologian and philosopher Richard Price “estimated” (on a basis as sound as dead reckoning) that Britain had “lost, in about 70 years, near a quarter of our people”—with the population now standing, he thought, at a meagre 4.5m. He believed if the trend continued, the fallout would be a barren country dotted with only a few dense cities, where “debauchery and voluptuousness” would run riot. Others insisted that the population was growing to a point where food production would not keep up: witness Thomas Malthus’s infamous An Essay on the Principle of Population (1798), with its forewarnings about overpopulation, famine and plague.

The man who finally let light in on the population question was a little-known agriculturalist and magazine editor called John Rickman. The friend of poets and radicals, Rickman had published an essay in his Commercial and Agricultural Magazine in 1800, which suggested that any enumeration required an Act of Parliament (“some sort of compulsion is necessary”) and persuaded a number of MPs to put a population bill to the House of Commons. Soon after the passage of the first Census Act that December, Rickman found himself on the government payroll, with staff and an office—named “the cockpit” after its former life as a ring for chicken fights—and told to put his theories into practice.

If Malthus and Price went in for grim prognostication, Rickman’s focus was the needs of here and now. “No society can confidently pretend to provide the requisite quantity of food, till they know the number of consumers,” he wrote. In wartime, it was all about enrolling “the highest number of men.” Both of these things required that everyone be counted—not just those with land or money. It was as simple as that.

“All kinds of choices are made in deciding what sort of data to carve out of reality”

Under Rickman’s scheme, people of local repute were held responsible as enumerators for making sure everyone in their “Parish, Township or Place” was accounted for. In England and Wales, this normally meant either the “Overseers of the Poor,” or the clergy—whose intimate knowledge of baptisms, marriages and burials came in handy. In Scotland the Kirk had declined the government’s offer of involvement, leaving “dominies,” or schoolmasters, as the next best option.

Rickman proposed four sets of questions. Alongside baptisms, marriages and burials, he wanted to find out how many people lived in each household, “distinguishing Males and Females,” as well as how many houses stood uninhabited, a major point of contention in previous population estimates. (Men already in the armed forces or merchant navy were not to be tallied, to avoid doubling up on figures already provided by regimental musters.)

Workwise, people could only put their occupation under one of three categories: “Agriculture,” “Trade, Manufacturing or Handicraft,” or what was effectively “none of the above.” Given this had never been done before, it was best to keep things simple.

People were not asked after their name, their age, their familial relationships or where they came from. Rickman’s whole design was brutally matter-of-fact: knowing people’s names and relationships did not help you feed or conscript them.

A very commonplace matter

Although topping 600 pages, Rickman’s first “Account of the Population” was incomplete and greatly hampered by the very conflict that had inspired it. The wartime focus, Rickman reflected in 1821, “rendered [it] somewhat defective” when it came to raising tax. The UK was then entering a grim spell of austerity to pay the bills incurred fighting Napoleon. Rickman, having by this time shed his bohemian ways and settled into the role of bureaucrat, wondered if revenues “might possibly be founded on the results” of a more refined investigation.

Whatever the flaws, Rickman’s original masterstroke was securing parliamentary approval for his census. This enshrined his count in law, and ensured that it became the standard upon which all subsequent surveys would be built. His exercise was repeated in 1811, and—as Britain’s modern bureaucracy developed—every decade thereafter. By 1891, it had become—as one assistant enumerator in Yorkshire noted—“a very commonplace matter.” “People, even the lowest in education,” he said, “better understand its object.”

By this time, the census had done valuable work in dispelling myths about a declining population. In 1801, the population of England alone was declared at 8.3m; by 1901, this had ballooned to 32.5m. As well as recording population growth, the census made plain this was nothing to fear. Thanks to improving agricultural efficiency, many more mouths were being better fed by, as the census also soon showed, relatively fewer hands: in 1851, 22 per cent of the total UK population worked in agriculture; by 1901, this had fallen to just 9 per cent.

However, not everything the census revealed was rosy. By the time it was in full swing in Ireland—having been glaringly omitted or badly managed there previously—the census estimated, in 1841, that the population of the whole island stood at around 8.2m. Only a few years after that the Great Famine took hold, and the next census in 1851 (amended with new information in 1856, so enumerators had time to make “statistical sense” of the devastating events they had unwittingly recorded) showed that the population had already plummeted to 6.5m. As mass death and emigration continued, Ireland’s population would remain in demographic freefall for the rest of the century—all tracked in grim detail by the census. By 1911, Ireland’s population was recorded as 4.4m; half of what it was 70 years previously.

Alien understanding

If the census provided useful warnings, it could sometimes also unnecessarily inflame particular issues. For the most part, the early censuses did not pay much attention to nationality; unlike today, people were not asked what they “felt” their national identity to be. It wasn’t until 1841—often considered the start of the “modern” census, for its inclusion of people’s names—that a question about country of birth was introduced. The query was vague and mostly gauged intra-UK movement: Englishmen in the Highlands, say, or Irishmen in London. Those from further afield—which included not only “real” foreigners but also, as Hutchinson notes, “a large number of children of British imperial administrators, diplomats and entrepreneurs”—simply wrote their country of birth in a box marked “Foreign Parts,” and that was that.

Britain was not much interested in immigrants in those days because it was—owing mostly to empire—a nation of emigrants. If real foreigners were mentioned, it was often as a point of pride. In 1861, the registrar general George Graham cheerfully revealed that Italy had sent us musicians and “image makers”; France provided cooks, language teachers and governesses; and Austria servants, merchants and—presumably due to lack of work in their landlocked home—seamen.

But by the centenary census in 1901, “Foreign Parts” were coming closer to home. That year the census revealed that the UK’s foreign-born population, standing at about one third of a million, was twice as high as it had been 20 years before. But instead of Italian musicians and French cooks, the attention was now on a different sort of migrant. The 1881 census said there were 3,789 people from Russia—by 1901, this had risen to 61,789. Many of them were Jews who had fled the pogroms and persecution of the tsars, but—as they soon found in Britain—could not escape discrimination.

Counting them had come to seem important: previous returns had been met with the charge—stoked by new antisemitic groups, like the British Brothers’ League—that, due to their tax dodging and poor English, Britain’s Jews were being under-counted. The registrar general of the day, William Dunbar, went to great lengths to assuage native fears—reminiscent of today’s housing and benefit scares—that the workhouses were full of Jews depriving Britons of their poor relief. He pointed out that, under his predecessor, local rabbis had created promotional material for the census in Yiddish and German; the census form was also translated.

The effect of this careful counting was mixed. The myth of Jews clogging up the workhouses could be dispelled: “European Foreigners” comprised 9 per cent of the population of east London, the highest of anywhere in Britain, but only 1 per cent of those in the local workhouses. Dunbar proved that the foreign-born were—then as now—exceptionally good at raising themselves out of poverty. But at the same time, the rising numbers encouraged a new wave of anti-migrant sentiment. As anxieties about losing an ill-defined “purity” took hold, the census provided anti-immigrant and antisemitic campaigners with statistical sustenance.

By 1905, these fears had reached parliament with the passing of the Alien Act, the first of its kind to implement curbs on immigration, in particular of Jewish people from eastern Europe. Rickman’s original argument that an “intimate knowledge of any country can be the only foundation of the legislation of that country” had been borne out in quite a different way to that he imagined.

Destined to count

It wasn’t just in Britain that the census would be used to justify draconian curbs on immigration: 19 years after our legislation, the United States enacted its own Immigration Act. The act would completely block immigrants from all of Asia while setting quotas for nations elsewhere, proportionally based on the origins of the US population according to its 1890 census—a crude attempt to restore an earlier, supposedly “purer” ethnic mix. The US government was only able to do this in the first place because of the longstanding inclusion of an ancestry question in its census.

The facts always matter in any enumeration exercise, but there remains no guarantee they will be put to good purpose. However, over the years, it’s not just the state but citizens who have cottoned on to the political potential of the census. A boycott of the exercise became a pillar of suffragette protests—they argued if they were to be counted as people, then they should also count as voters. A similar boycott in 1991, this time in protest at the poll tax, saw returns undercount the population perhaps by a million. In 2001, in what was originally a (supposed) protest from non-believers about the new religion question, 400,000 people listed their faith as Jedi, surpassing Judaism, among others, to be crowned Britain’s fourth largest “religion.” (No moral panic this time about the Order drawing poor relief.)

As the census has documented the way society has changed—or acted as a declaration of the way we want it to change—the census itself, in turn, has had to change with it. But the introduction of new questions and how they are framed remains controversial. The arrival of an “ethnic group” question in 1991, for instance, was essential to keep track of our increasingly diverse society—even so, it was laden with fraught judgment calls. For example, are the Irish or Polish populations really best thought of as sub-categories of white? There is no objective way to settle whether skin colour or other dimensions of “otherness” deserve analytical priority. In a necessarily subjective exercise, you can perhaps still achieve a degree of objectivity by drawing on the attitudes of society as a whole. But often even this has proved beyond reach, with the actual approach turning on the prejudices of the government of the day. Further new questions in that same year about physical and mental health were equally controversial, this time for the way they acknowledged only those problems with wellbeing that might require medical intervention—or, in other words, state management.

In many ways, new census questions have continued to be developed in Rickman’s utilitarian vein: they’re still picked to provide information that bureaucrats can “do” something with. But for citizens filling the forms in, it can be much more about feelings. There were heated court battles this year regarding the guidance accompanying the evolving questions about sex: unlike on passports, people could now register as non-binary, and for the first time we were asked whether our gender had changed from that registered at birth. Some people celebrated, while others railed against modifications to the survey which seemed to epitomise changes they were not comfortable with.

But after 220 years of the British census—and thousands of years of enumerating before that—we can say for certain that what we count changes what we think counts. Which brings us back to how the census itself can continue to make sure it counts for something. What do we want the census to do for the future? It might be that the time is fast approaching when we tick the box that says: “nothing at all.” Today, we live in an age of ubiquitous data collection automatically harvested through our phones and computers. Every decade, the argument that we’ve simply outgrown the need for a single, definitive “state of the nation” snapshot gains more currency. We can, runs the argument, use new technologies to do more advanced, more immediate and more interesting things.

Though for all the inescapable biases and subjective decisions that crop up in creating a census, there are surely also preconceptions embedded in algorithms. Census or no census, so long as we’re interested in the state of our society—and, in Rickman-esque parlance, the opportunities for its “future improvement”—we are destined to continue to count. But if the history of the census shows us anything, it is that we can often learn just as much by questioning the questions designed to do that job. The script on the clipboard—or in the algorithm—will sometimes reveal more about a society than the answers yielded. And there’s no list of tick-box forms, however exhaustive, that will ever change that.