Monday, 10 January 2022

Census Censure

 In common with a lot of people with a degree of curiosity about their family history, I recently signed up to get access to the online 1921 Census records, released this week. Sadly, this will be the last census to be made available in the likely lifetime of anyone over 60 (at least under the current 100 year rule) as the 1931 Census of England and Wales was destroyed by bombing in WW2, and there was no census in 1941 for obvious reasons.

The census is being made available exclusively by, who were contracted to do the scans and transcriptions, and I'm still getting to grips with the quirks and shortcomings of the user interface and indexing. However, the fact that the very first record I brought up had major transcription errors does not inspire confidence  – I'm sorry, but my grandfather did not share the middle name "Matilda" with his wife – as these transcriptions will be the basis of the indexing. Assuming this is human error, though, I suppose it should be more accurate, on the whole, than relying on the sort of automatic OCR that plagues newspaper digitisation.

Another problem is that it is impossible to see the street name or even partial content of household records beyond a list of names without paying £2.50 for a transcription or (more reliably) £3.50 for a scan of each household record, which puts severe limits on one's willingness to take stabs in the dark or to establish context – who were the neighbours? etc. – and what possible use is an "address search" that merely gives a list of house numbers in a street, with no indication of who lived there? I cannot imagine anyone would start randomly opening households in a typical city street, at £2.50 / £3.50 a shot.

Most of this came up as an issue with one of my first searches. My mother was born in 1923, but I was able to find my maternal grandparents easily enough and her older sister. To my surprise, they were living in a pub, the Seven Stars in Charlton, a tiny hamlet near Hitchin in Hertfordshire. One of the innovations of the 1921 Census was to include the name and address of a worker's employer, so a further surprise was to discover that my grandfather was working as a fitter in a "Portland cement" factory situated in some large chalk quarries at Arlesey, north of Hitchin. I was curious to know who else was living in the Seven Stars, but an "address search" failed to find anything at all. Despite being recorded as their address, it was not recognised as such by the indexing.

Moving on, then, I searched for my great grandmother, and quickly found her, too. She was living with her second husband George, the son of her first marriage Herbert, 20, her step-son Charlie, 23, plus her two children with George, Edna and Eric, 11 and 9 respectively. All the working age men were labourers, with Charlie working in a tannery (a notoriously stinky job). When I looked at the location, I was amazed to discover that they, too, were all living in the Seven Stars: all nine of them in three rooms, with one shared room, presumably a kitchen space of some kind.

Despite using large-scale contemporary Ordnance Survey maps (provided online by the National Library of Scotland), I could not identify the actual location of the Seven Stars.  As it happens, an old school-friend is a well-regarded researcher, writer, and blogger about the very important subject of beer, so I asked him what he knew. He replied almost instantly:

"In fact the Seven Stars closed that same year, 1921 - it was one of several places in the area that had their licences taken away by the licensing magistrates as "surplus". (This was a nationwide movement that had been going on since the early 1900s. Brewers paid into a fund to compensate pub owners whose pubs were closed.) The Seven Stars was only a beerhouse, that is, licensed to sell beer and cider only, not wines or spirits, and it was leased to the local brewer, Lucas of Hitchin (which closed itself a couple of years later after having been sold to Green's of Luton)."

However, he was also unable to locate it within the village. Not that it really matters: I have no intention of ever visiting the place.

It then occurred to me to look for great-grandmother Eliza's daughter Edith, who had married by 1921. The suspicion had begun to form that she and her family might also be hanging out in the same establishment. Perhaps her husband was the landlord? A search under her married name turned her up, born in – yes – Charlton. But the scan of the original document gave the location as "Great Green", not the Seven Stars. Ah, well; nice try. But where was that? It does not appear on even the most detailed contemporary OS map for Charlton.

Now, the only indication of the geographical location of an address on a scanned census document are the registration district (generally the nearest town), the sub-district (a number), and the enumeration district (another number). It is not entirely obvious how to interpret these, as there is no available index to consult. In fact, what you need to do is look at some supplementary scans that are made available free of charge as "Extra Materials": that is, the document's "cover" and "front", plus a couple of original annotated maps that do explain where a scan is located, but only if you're prepared to give them very close scrutiny; none of this is obvious, and took me quite a while to figure out. In fact, the most obvious way to get that information is to buy (or pay another £2.50 for) a transcript, but these, as my grandfather "Matilda" and, hilariously, Edith's daughter "Town" (Joan) would attest, are not exactly reliable data. No doubt there are "how to" tutorials available somewhere on the website, but with a bit more effort this information could so easily have been provided as metadata on the scans, perhaps as a "mouseover". Anyway, after I'd finally figured this out it transpired that Great Green is not in Charlton at all but in Pirton, another nearby village, and the place destined to become Maternal Family Central in later generations.

So, if you're intending to use the 1921 Census, be warned: unnecessary obstacles have been placed in your path, and you will almost certainly end up spending more money than you had anticipated, especially if you're not sure about who you're looking for, what name they were using, or where they might have been living in 1921. "Fuzzy searching" looks to be impossible, and there's not even an "all you can eat" subscription available, which is strange, and surely calculated to annoy the professional genealogists.

But strangest of all, though, is to realise that the 2021 Census – which I must admit I'd forgotten had ever taken place – was arranged under the direction of a guy of exactly my age with whom I used to sit for many years on our local trade union executive, now the National Statistician, Professor Sir Ian Diamond. I suppose we'll have to wait 100 years to find out whether or not he and his people did a good job.


old_bloke said...

In my view this 1921 census has been wildly over-sold. "the 1921 census is a record of a moment of unique trauma" (David Olusoga) - oh no it isn't! The thing I've found most useful in my own attempts to trace back from the present to the start of the twentieth century is the 1939 Register, which also started out as pay-per-record-viewed, but is now part of the FindMyPast subscription. Presumably the 1921 census will do likewise, once they've recouped their costs.

With regard to paying for genealogical information, the England & Wales subscription model is a middle way. The keepers of records in Scotland want you to pay for everything individually, with no sort of bulk discount or subscription available. Then there's Ireland, where the number of records involved is smaller but, if they've digitised it, you just tick a box to say you're a law-abiding person and you can look at it all for nothing.

Maybe I should claim that Irish passport after all . . .

Mike C. said...


All true. In the end, there is no substitute for forking out the cash to buy actual copies of certificates of birth, marriage, and death, using the censuses and the excellent Freebmd database as a guide. The main benefit of censuses (and the 1939 Register) is to establish who lived with who (plus, in the case of certain people with rather more elevated lineages, how many live-in servants they had...).

Scottish online records are superb, despite the cost, and include parish registers. But, Ireland: sadly, some fool decided to trash the entire census returns from 1861-1891, so lots of luck with that!


Chris Rusbridge said...

I've been searching for my parents and their families in the 1921 census, too. I was very surprised to come across an internal database inconsistency. My Uncle Norman, aged 6 (and later a WWII fighter ace and the pilot of the first Spitfire to be shot down at sea during the war; he survived after being dragged down as the plane sank) was revealed as living at Isleworth with Alice, Eliza May +1. My mother aged 10 was living elsewhere in England for reasons I have yet to determine (school?). My maternal grandfather was listed as Army, his wife and youngest son did not appear (presumably overseas, probably in India). I was intrigued to find out who Alice and Eliza May were, preferably without paying for an actual record. After much sleuthing, I found Eliza Palmer and Alice May Palmer living in Isleworth, with Joseph Tovey +1. Tovey turned out to be Eliza's maiden name, and Palmer was my grandmother's maiden name. There are no Eliza Mays living in Isleworth, with any surname, with the correct number of co-habitants. Other records freely available from Ancestry dot com made it clear that Norman was living with his grandmother, Eliza. So somehow they transferred Alice's second name to Eliza in my uncle's record, but not in either Alice's, Eliza's or Joseph's records! That just should not be possible.

Slightly less reprehensible was that my paternal grandfather appeared as Charles R, when we know he was Charles B. I guess a 1921 handwritten B could easily be interpreted as an R to 21st century eyes!

Mike C. said...


If we're talking about the transcriptions, then I'm not surprised. It seems that everyone finds errors in their first few searches, which would suggest it's riddled with them. If not done by AI then I imagine it was mainly done in the Philippines or somewhere similar, where (as in my example) entering a girl's name as "Town" rather than "Joan" wouldn't seem so bizarre. Certainly, enumerators' handwriting conventions can be challenging to modern eyes (though hardly by comparison with early modern scripts).

I'm a big fan of Freebmd as a way of establishing relationships (and, of course, as a way of getting the necessary data to order a copy of a certificate). One you realise you can search for, say, births by using father's surname and mother's "maiden" surname with place and date constraints it becomes an extremely powerful tool.

If you have spotted errors these can be reported directly if you have bought a transcript (there's a button). Otherwise, I quote:

"Transcription errors found without purchasing the transcript itself can be reported by emailing

Please include a link to the record and a brief description of the error, for example, the first name is recorded as Jo when the image shows it as John. Please use "1921 Census transcription update" as the email subject line, this will allow us to correct errors quicker and more efficiently."