Using the 1939 Register to analyse birth patterns in England and Wales

[HowardS]

One of my favourite features of the 1939 Register (at findmypast.co.uk) is the ability to search for entries by date of birth. This is particularly helpful if I'm researching a branch of my family which has a quite common surname, such as Smith or Jones (etc). If I'm not sure, for example, whether a certain person died prior to 29th September 1939, or which possible marriage record for an individual is correct, but happen to know their date of birth (which appears on some baptismal records and on school admission records) then I can sometimes solve my brick wall with a fair degree of certainty.

But in fact this post is more about the general fun one can have with the Register, when one has nothing better to do (viz, myself this morning!) Some time ago when I first opened the search form for the Register, I mistakenly clicked on Search before entering any search data. This led me to discover the quite interesting fact that the Register, today, contains a total of 35,538,212 non-redacted entries. (I've no idea what the proportion of "closed" entries actually is - does anyone have an inkling?)

It occurred to me to wonder how "popular" my own date of birth was at that time, so I entered just the day and the month of birth, and learnt that 94,868 people visible on the Register share my birthday! Now multiply that by 365 and you get "only" 34,626,820 which suggests that the number of people sharing my birthday was slightly below average. Of course the only way to be sure was to learn how many people had each possible birthday during the year, so I tried each date in turn. This took about an hour and a half! But hey....

The upshot of this was that the number of people declaring they were born on a particular date varied from 76,986 for 30th July to 114,562 for 1st January (closely followed by 114,281 for 25th December then 113,795 for 14th February!) But the vast majority of dates gave a number of births in a fairly narrow band between 87,000 and 97,000, so one might wonder whether a fairly large number of people really weren't sure and put down a "happy" invented date...

Incidentally, the only persistently "slightly-low" numbers of births (well under 90,000 on most days) occurred from mid-October to early-November, which surprised me, as I was sort of expecting that there might be a procreation lull at harvest time (!) but not in mid-winter.

A few other QI facts came to light. For example, the total number of "open" entries showing a precise day and month of birth is "only" 34,197,608, which means that 1,340,604 do NOT show a precise day and month. Which seems like rather a lot to me and I wonder who all THOSE people were...?

Not to mention the rather amusing fact that (at least according to the transcribers), 497 people claimed to have been born on 30th February, 213 on 31st February, 1066 on 31st April, 1060 on 31st June, 847 on 31st September and 891 on 31st November!!

 

[At home in NZ]

Good fun, if you have the time. According to Ancestry 84,641 people share my birth day and month.

 

[lindy]

Looks like I would have been one of 86,646 back in 1939 on Ancestry and slightly more on FMP. The difference being the number of redacted entries?

I too have found searching by birthdate useful but have found quite a few inconsistencies. Several one year older (or even more!) than declared but the correct day given.

 

[peter]

Bear in mind that the enumerators may have misread the householder's handwriting, and this is especially likely in the case of impossible dates.

For example, when I picked - at random - someone supposedly born on 31st February I was able to establish from their death index entry that they had actually been born on 31st March.

 

[John Dancy]

Bear in mind that the enumerators may have misread the householder's handwriting.

and/or the transcribers. Have found multiple errors since it was introduced. The year for date of birth has generally been calculated by subtracting the age from 1939, and the age was when the form was filled out, so the year quoted always need checking against GRO. Unsurprisingly, all households did not fill out their form on the same day, so there is another variation built in if close to a birthday.. Commonest 'errors' tend to switch Jan for Jun/Jul and Mar with May, but I have found entries with no month or no day given. It is quite possible that given the size of the register a lot of these cancel each other out ? and your figures are possibly quite accurate if you ignore percentage points.

 

[peter]

and/or the transcribers

But we can easily spot those. It's much more difficult to determine whether the enumerator has made an error when copying the data across, or whether the householder got it wrong in the first place.

The year for date of birth has generally been calculated by subtracting the age from 1939, and the age was when the form was filled out

In my experience once people get beyond a certain age they're more likely to remember the year that they and other family members were born rather than their ages, so I doubt that miscalculating the year of birth is a common source of error - but it would be easy enough to check using statistical sampling.

 

[Katie Bee]

It's much more difficult to determine whether the enumerator has made an error when copying the data across, or whether the householder got it wrong in the first place.

My father's cousin was living with my mother and father in 1939.
My mother and the cousin are on adjacent lines and both have birthdays in December.
The enumerator (?) has transposed the day of birth.