1921 census review

[Thunder]

A rather disturbing description of how the transcription was done appears on https://www.findmypast.co.uk/help/artic ... anscribed-. It says:

When transcription of the 1921 Census of England and Wales took place, each digital image had to be broken up into segments so that the person transcribing it could not see a whole record or household. This was to ensure we complied with security and data protection regulations but also why you might see various spellings of the same surname or street address on one record, because it has been transcribed by multiple people without the context of the whole record.

In other words, having hit an excessive number of errors on the 1939 Register by splitting the page into columns, they appear to have gone and done something similar on the 1921!

The prime issue for me is that, by this point in time (January 2022), there are no relevant data protection regulations - the whole thing is open. So why create such an impediment to quality for such a short period? I am assuming that the indexing process takes some months but why could derogations not have been granted for those months? After all, we saw films of people restoring pages prior to imaging - they must have been looking at the whole page, so they must have had a derogation. Did anyone in authority (i.e. TNA) actually talk to the Information Commissioner about what a sensible path forward would have been? Or did TNA, yet again, presume that the Man in Kew knew best?

You assume it was the "Man in Kew", I would say the "person in Kew". In my view TNA are data protection-obsessed and have gone over what is needed, as in a murder case from 1919 where they have redacted the name of the victim even though it is available at ancestry from the open Calendars of Prisoners held at TNA!. You have to trust people doing the work on the Census and there is the Official Secrets Act. Of course the 1901 Census was being transcribed by prisoners in Durham Prison whilst the records were still closed!!.

[Thunder]

I would add that when the 1901 Census was retrieved from Durham Prison it had a second transcription check done by TNA staff and students. Even so there were still a number of people transcribed 'Ditto' and shown as so on the day the records were open and amended later.

[AdrianBruce]

... You assume it was the "Man in Kew", I would say the "person in Kew". ...

I'm alluding to the old saying that "Sometimes the Man in Whitehall really does know best"...

... In my view TNA are data protection-obsessed and have gone over what is needed, ...

Agreed. And that is exactly why they should consult experts like the Information Commissioner and not make decisions for themselves.

... You have to trust people doing the work on the Census and there is the Official Secrets Act. ...

No, you can't just trust people. You have to draw up the procedures and instructions and contract your staff / suppliers to work to them. I very much doubt that the OSA can be applied. It has a specific scope, as I understand it, and stuff about to be made public clearly ain't it. Or so says my non-expert view! (See my middle comment!)

[Guy]

What errors have you found in the images, any or none?

I will take note of you comment when you produce a similar sized data set with no errors. It is often the case that many of those who complain have never attempted to produce something similar.
Cheers
Guy

I spent 16 minutes looking for my grandparents as FMP had dropped the 'h' as shown in their surname (and correct) on the image, also the Barton-upon-Irwell Poor Law Institution was transcribed as Barton upon Derwell if I recall correctly and the institution was transcribed as the Poor Low Institution, all correct on the image. Also for Broadmoor Asylum mis-spelt as Aslyun. One of my uncles stated correctly to be born and living (not in Broadmoor) in Eltham South-East London was transcribed as being born in Hertfordshire. When I did transcribing at TNA these errors never happened.

Were you transcribing open images (i.e. able to view the whole image ) or were you viewing colomuns or sections of an image. It is far easier to transcribe when you can see the context in the page rather than little bits of the page.
Yes errors will always occur but the restrictions the National Archives placed on the bulk transcribers and the checking of unopened records make errors far more likely.

If you have complaints I suggest you direct them to the NA rather than FMP or their reanscibers who have done a good job in very restricted circumstances, however if you do be prepared for the reply that for the transcribers to have open access they would have to wait until 100 years had passed meaning the online release would be delayed by 3 or 4 years due to legal requirements.
Cheers
Guy

[AdrianBruce]

... if you do be prepared for the reply that for the transcribers to have open access they would have to wait until 100 years had passed meaning the online release would be delayed by 3 or 4 years due to legal requirements. ...

Ouch. I had wondered how long the indexing process took. The people preparing and restoring the pages must have had a derogation from the 100y limit, ditto the guys doing the imaging. That being so, I wonder why the same derogation couldn't have applied to the indexers? Maybe that was thought to be too big a risk - that's a perfectly reasonable attitude to take provided it's come from professional experts in Data Protection.

The converse is the idea that Data Protection is never about 100% prohibition - clearly HR clerks can access sensitive data because it's their job. Presumably the derogation for restorers and camera operatives came because it was their job. So an alternative viewpoint might be to say that it was the indexers' job to have full access...

Either way I could accept things if I had confidence that TNA took professional advice. Unfortunately, I'm not convinced...

[Thunder]

What errors have you found in the images, any or none?

I will take note of you comment when you produce a similar sized data set with no errors. It is often the case that many of those who complain have never attempted to produce something similar.
Cheers
Guy

I spent 16 minutes looking for my grandparents as FMP had dropped the 'h' as shown in their surname (and correct) on the image, also the Barton-upon-Irwell Poor Law Institution was transcribed as Barton upon Derwell if I recall correctly and the institution was transcribed as the Poor Low Institution, all correct on the image. Also for Broadmoor Asylum mis-spelt as Aslyun. One of my uncles stated correctly to be born and living (not in Broadmoor) in Eltham South-East London was transcribed as being born in Hertfordshire. When I did transcribing at TNA these errors never happened.

Were you transcribing open images (i.e. able to view the whole image ) or were you viewing colomuns or sections of an image. It is far easier to transcribe when you can see the context in the page rather than little bits of the page.
Yes errors will always occur but the restrictions the National Archives placed on the bulk transcribers and the checking of unopened records make errors far more likely.

If you have complaints I suggest you direct them to the NA rather than FMP or their reanscibers who have done a good job in very restricted circumstances, however if you do be prepared for the reply that for the transcribers to have open access they would have to wait until 100 years had passed meaning the online release would be delayed by 3 or 4 years due to legal requirements.
Cheers
Guy

I was transcribing open images. TNA were quite happy to take the money from FMP and not take responsibility for the errors, but they don't care about researchers. If FMP couldn't do the work properly they shouldn't do it. The aim of transcribing is to allow people to find what they are looking for, not having to spend time trying to find the image.

[Guy]

A rather disturbing description of how the transcription was done appears on https://www.findmypast.co.uk/help/artic ... anscribed-. It says:

When transcription of the 1921 Census of England and Wales took place, each digital image had to be broken up into segments so that the person transcribing it could not see a whole record or household. This was to ensure we complied with security and data protection regulations but also why you might see various spellings of the same surname or street address on one record, because it has been transcribed by multiple people without the context of the whole record.

In other words, having hit an excessive number of errors on the 1939 Register by splitting the page into columns, they appear to have gone and done something similar on the 1921!

The prime issue for me is that, by this point in time (January 2022), there are no relevant data protection regulations - the whole thing is open. So why create such an impediment to quality for such a short period? I am assuming that the indexing process takes some months but why could derogations not have been granted for those months? After all, we saw films of people restoring pages prior to imaging - they must have been looking at the whole page, so they must have had a derogation. Did anyone in authority (i.e. TNA) actually talk to the Information Commissioner about what a sensible path forward would have been? Or did TNA, yet again, presume that the Man in Kew knew best?

I believe the staff at the National Archives restored the pages and would be allowed to do that as part of the archive proceedure to conserve the census, The Findmy past team were not National Archive staff and were therefore not allowed to access the full page, by redacting what could be seen by the contractors the whole procedure complied with the Data Protection Acts. I believe the digitisation etc. took up to three years rather than months.
By the way I would not shout too much about data protection regulations, as there is no time limit imposed, the 100 years is non statutory and could easily be extended to 500 years if it was thought wise to do so.
Cheers
Guy

[AdrianBruce]

... By the way I would not shout too much about data protection regulations, as there is no time limit imposed, the 100 years is non statutory and could easily be extended to 500 years if it was thought wise to do so. ...

Of course, that works in both directions and since the limit is slightly arbitrary, it could have been 100y except where those working on the census were concerned, where it could be set to 97!

[Guy]

... By the way I would not shout too much about data protection regulations, as there is no time limit imposed, the 100 years is non statutory and could easily be extended to 500 years if it was thought wise to do so. ...

Of course, that works in both directions and since the limit is slightly arbitrary, it could have been 100y except where those working on the census were concerned, where it could be set to 97!

Guy

[meekhcs]

An interesting fact came to light today, after discussion with FMP over a problem with buying an original census record. The 1921 census was transcribed in India. Presumably a matter of cost.