The Official History of Britain

The census continued to become more advanced and complex as it entered its second half-century and beyond. 1851 asked for exact ages rather than quinquennial age-groups, marriage status, and also noted blindness, deafness and dumbness. It also asked about religion – interestingly, as a voluntary question, as it was when it reappeared on the form in 2001. It asked about attendance at services, rather than affiliation, and was aimed primarily at establishing whether there was sufficient provision of churches and other places of worship. The subsequent report opined that, ‘The most important fact which this investigation as to attendance brings before us is, unquestionably, the alarming number of the non-attendants’. It was clear who was responsible for this; not of course the middle and upper classes, but ‘artizans’ and ‘the labouring classes’. 1871 asked for those who were classified as a lunatic, or as an idiot or imbecile. On one level, this seems amusing to us – we all know people of whom we would say, hopefully teasingly, that they fall into at least one of these categories, and in some cases all three – but it is also a salutary reminder that Victorian attitudes towards the mentally infirm now seem at best unenlightened and at worst actively cruel. The term ‘idiot’ survived until 1901, when it was replaced with ‘feeble-minded’.

In 1881, a new dictionary of occupations was drawn up to keep pace with rapid changes in industry and the proliferation of new jobs. Around 12,000 different occupations were listed, almost double the previous amount, though by 1911 this would be up to around 30,000 (much same as in the coding index of the latest 2020 Standard Occupational Classification, though the latter no longer includes ‘wheeltapper’ but does acknowledge ‘YouTuber’ as a recognised occupation). 1881 also saw respondents in Scotland asked whether they spoke Gaelic, a question extended to Welsh-speaking in Wales in 1891. 1901 saw unmarried women differentiated from their married or widowed counterparts for the first time. The 1901 census presciently asked people in certain industries whether they worked from their own homes, though presumably it failed to foresee that in another 120 years the main question for homeworkers would be ‘Are you still on mute?’ as they tried to conduct yet another online meeting with their widely scattered colleagues. In 1911 – the first census to use machines, albeit by our standards rather primitive ones which worked via holes punched in appropriate positions on special cards – the fertility of marriage was included, with questions asked about the duration of existing marriages and the number of living children born to each marriage. However, the outbreak of war in 1914 delayed the processing of the fertility data, the last report of which was not published until 1923, a dozen years after its collection.

The 1921 census in Great Britain was held in the aftermath of the First World War and the Spanish flu – see Part Four. But it was postponed in all of Ireland, north and south, because of the political situation. In Northern Ireland it was not held until 1926, carried out by the registrar general for Northern Ireland, and the timetable for Northern Ireland did not get back in sync with the rest of the UK until 1951. The 1931 census went off without a hitch, but 11 years later all the results were destroyed in a fire at an Office of Works store in Hayes, Middlesex. It also saw clerks downgraded to Social Class III, having been already moved down from Social Class I to Social Class II two decades earlier – this gradual fall seemingly a result of the spread of universal education (and typewriters) devaluing skills that had once been in great demand. 1951 introduced a 1 per cent sample to allow a quick and wide cross-section of data before the actual publication of the full results, and asked each household about their water supply as these began to shift away from communal arrangements and outdoor toilets to individual and internal household ones. A piped internal supply was deemed different from a tap in the yard, and a tap in the yard of course different from a public standpipe: a water closet was distinct from an earth or chemical one. In 1961, an electronic computer – an IBM 705, to be precise – was used to help process results for the first time, and like all computers of the era took up most of a room and yet had less processing power than the SIM card in the smartphone on which you may well be reading this. As the General Register Office did not yet have its own computer, one located in Winchester belonging to the Royal Army Pay Corps was used instead. The 1971 census was the first to be conducted under the aegis of the Office of Population Censuses and Surveys, which had been set up the previous year to combine the General Register Office and the Government Social Survey. And questions on ethnicity and religion were introduced (or reintroduced) in 1991 and 2001 respectively. In 2001 responsibility for the census passed to the Office for National Statistics, which had come into being five years earlier when the Office of Population Censuses and Surveys merged with the Central Statistical Office.

The census has been compulsory since 1921, but at least two years’ censuses have been subject to boycotts by way of civil disobedience. In 1911, the Women’s Freedom League encouraged women to avoid the census (by attending all-night parties or staying with friends) as part of the suffragette movement campaigning for women’s right to vote. Emily Davison, who two years later would die after colliding with by the king’s horse at the Derby, was recorded in two places on census night: her own house in Russell Square and a broom cupboard in the Houses of Parliament. And opposition to Margaret Thatcher’s poll tax (not to mention fear that the census might be used to enforce that tax) led to press reports that up to a million people had avoided the 1991 census, even though Mrs Thatcher herself had stepped down as prime minister the previous year and the poll tax was clearly on the way out.

Of course, for the more recent censuses, one can’t yet see the actual records themselves, only the anonymised data tables that are published. This is because these records, full as they are of very sensitive personal information, remain confidential for a hundred years after they are collected. Once that period has elapsed, census returns are in the public domain, and these days can be easily accessed online, thanks to popular genealogy websites. But until then they remain closely guarded by the ONS. Not so long ago the well-known television crime drama Silent Witness featured the cops logging in to an imaginary ‘National Census Database’ to get the information they needed to crack the case; but, alas, thanks to the hundred-year rule, it would have to be a pretty cold case review indeed for the police to be able to access any useful information from the census (anyone fancy digging around the 1891 census returns to propose yet another suspect for being Jack the Ripper?). Indeed, data confidentiality generally, and the need to avoid accidentally releasing material that could identify individuals, is very important to the ONS, which means that it can’t always release information in the full level of detail that it collects. For example, on baby names, it doesn’t publish those names that are only given to one or two individual babies each year.

Meanwhile, quite distinct from the decennial census, more and more official statistics were being collected, often reflecting particular concerns. For example, the registration of births, marriages and deaths had long been done by the Church of England on a parish-by-parish basis, but with the spread of other denominations that could not carry out legally binding weddings according to their own law, there was increasing pressure for a registration system controlled by the state not the church. This resulted in the Registration Act and the Marriage Act 1836, under which Thomas Lister was appointed the first registrar general for England and Wales. And naturally with the registers came the statistical reports: Lister’s first annual report, published in 1839, revealed that, in the 12-month period of July 1837 to June 1838, there were 107,201 marriages celebrated ‘according to the Rites of the Established Church’ and another 111,481 not according to those rites. Again, the late nineteenth and early twentieth centuries were times of labour unrest, not least the 1888 match girls’ strike and the 1889 London dockers’ strike. Not surprisingly, therefore, the official statistics on labour disputes, first compiled by the Board of Trade and then the Department of Employment but more recently by the Office for National Statistics, go back to 1891 – though the peak year for the number of days lost to strikes did not come until 1926, with the General Strike and the long-running miners’ strike that followed it.

Incidentally, different arrangements were already starting to apply to the different countries of the United Kingdom, even in those pre-devolution days: thus, while England and Wales had a registrar general from 1836, the first registrar general for Scotland was not appointed until 1854 and civil registration in Ireland only began in 1864. To this day, the ONS is responsible for the census only in England and Wales, with National Records for Scotland (set up in 2011) and the Northern Ireland Statistics and Research Agency (established in 1996) now looking after it elsewhere in the UK. This means that sometimes, especially when dealing with old series going a long way back, it can be hard to get long-run numbers for the UK as a whole.

Quite separately, the great advances in computing in recent decades have made possible a number of developments in the collection and analysis of official data. For example, in late 2019 the ONS began publishing, jointly with HM Revenue and Customs, monthly updates on the number of employees on the payroll in the UK. A few months later, when the country was struck by the coronavirus pandemic, forcing many businesses to suspend trading, the ONS was able to accelerate its timetable for publishing these figures so as to get them out quicker than was possible with the old survey-based measures of employment. Likewise, as part of its work to meet the challenge of providing richer and more real-time statistics to inform decision-makers, in 2016 the ONS announced the establishment of a data science campus. This was seen as a way of exploring and using innovative methods and new data sources, including administrative data and big data, to improve existing statistics and develop new outputs for the public good.

The campus continues to play an important role in the data revolution by innovating public sector data science and building artificial intelligence (AI) capability through a range of learning and development programmes, in collaboration with other government departments, industry and academia. Since its launch in 2017, it has researched a large number of different topics – including investigating the use of machine-learning techniques to predict missing energy performance scores in the housing stock; the use of open data sets to develop a better understanding of loneliness; using non-traditional data sources to understand the characteristics of high-growth companies; and using tax returns, ship-tracking data and road traffic sensor data to allow early identification of large economic changes. Moreover, it too was able to respond rapidly to the pandemic; for example, doing analysis of anonymised mobility data released by Google to give insight into the impact of social distancing measures introduced in response to the pandemic.

This book uses the census snapshots over the past two centuries, as well as other figures from the Office for National Statistics and other official sources, to chart the past and present of this country during that period; to see, as far as possible, how much we have changed, and indeed how much we have stayed the same. It is not, and does not aim to be, an overarching or definitive history of the country – such a thing would run to the length of the Encyclopaedia Britannica – but rather a broad and hopefully light-hearted look at what the censuses, surveys and other official statistics have thrown up.

It is divided into three main sections: who we are, what we do and where we live.

‘Who we are’ looks at, among other things, how family arrangements have changed, how we spend our free time, our religious and cultural beliefs, and what the state of education can tell us about what young people do and don’t want to do with their lives.

‘What we do’ examines how our working lives have evolved over those two centuries, a period which has spanned not just an Industrial Revolution but a technological one too. It looks, among other things, at the jobs that exist now but did not in the past, and vice versa; at the effects of automation; and at the extent to which gender roles in work have changed.

‘Where we live’ covers not only changes in local areas but also the wider country too, and in particular the shift to predominantly urban ways of living, and also how housing tenure has altered. This is a small and crowded island, and yet there are still substantial swathes that are, to all intents and purposes, scarcely inhabited, as our look at population density shows.

These are followed by two shorter sections – ‘A Census of Enlightenment’, looking back a hundred years to the 1921 census, held in the aftermath of the First World War and the devastating Spanish flu pandemic, and ‘Back to the Future?’, where I look forward a hundred years and speculate a little about what a 2121 census might look like.

Of course, there are limits to what is, or ought to be, collected in official statistics, which are after all funded out of the taxes we all have to pay to the government. Thus, among the questions that have stumped the normally indefatigable ONS press office down the years were: ‘How many rats and squirrels are there in the UK?’, ‘What’s the number of people in the UK who sing in a choir?’, ‘What percentage of the UK are aristocrats?’ and even ‘Do you have statistics on how many new statistics start getting collected or published every year?’ But while we don’t really need to know how many people sing in a choir, things like the number of people who are born every year, how many get married, how many are in work or what we are dying of, are things that the country and its decision-makers really do need to know.

This book was great fun and rather illuminating to write, and I hope you feel the same way about reading it. If you do, and you’re approached by the ONS at some point in the future to take part in one of its surveys, please bear in mind that, without people like you helping, much of the information here just couldn’t have been compiled.

* A ‘hide’ was an Anglo-Saxon unit of land measurement, originally reckoned as enough land to support one family.

† These were the traditional officials that parishes appointed to see to law and order.