Is a Rare Surname easy to find? Searching for Barrowclift with Ancestry and FindMyPastPosted: 03 Jun 2013 Filed under: Genealogy issues, Research strategy, Sue's family research | Tags: 50 Marriage Mondays, Ancestry, Barrowclift, Birmingham, Coulson, FindMyPast, search 10 Comments
This week’s 50 Marriage Mondays post concerns finding more information on a bride with an unusual surname. This marriage certificate was acquired during the settlement of Raymond Coulson’s estate, so I already know the ancestry of the groom.
Bride: Gladys Rose Barrowclift, aged 25
Groom: Charles Spencer Coulson, aged 28
Date: 8 June 1935
Location: The Register Office, Birmingham
Father of Bride: Samuel Barrowclift
Father of Groom: Spencer Coulson
Witnesses: W J Sleasby, C L Barrowclift
Barrowclift is a very rare surname. Consequently, neither PublicProfiler nor British Surnames offer statistics on the number or distribution of people bearing this name. Both websites suggest that Barrowcliff and Barrowcliffe may be name variants, but these two names are also very rare. The most common frequency was 9 per million for Barrowcliffe based on 1998 data.
As the Barrowclift name combined with the groom’s middle name, Spencer, had made it easy to identify this couple, you might expect tracing the bride’s family would present little difficulty. Aged 25 in 1935, I calculate that Gladys Rose Barrowclift was born ca 1910, so she should be recorded on the 1911 census. If her age is misreported, I should find her father, Samuel. So, a search on the surname likely will turn up both in a short list, won’t it?
A search on FindMyPast for just the surname yielded 2 results, a John and a Susan. The nearest equivalent search on Ancestry, specifying exact matches on the surname only, yielded 10 results. 6 of these, the family of the John in the FindMyPast result, Charlotte, Florrie, Frederick, Bernard and Ernest had been indexed as Barrowcliff, but later corrected by users. The other 4 results named Mary, Thomas, Joseph and Henry are a second family. But, no Samuel or Gladys Rose. I should fare better by widening the search to include variants of the surname, shouldn’t I?
Alternative Search Tactics
Ticking the ‘include variants’ box on FindMyPast’s search, yielded the same 2 results. Choosing the ‘Use default setting’ option on Ancestry also yielded the same result as before.
After trying some variants (e.g. Barrowcliff, Barrowcliffe) and wildcard searches (e.g. Barrowcli*), and getting hundreds of results, but no seeing any good matches on the first page, I started to wonder if Barrowclift was really Samuel and Gladys Rose’s surname. The FreeBMD indexes confirmed I had the right name. Gladys was born in the Oct-Dec quarter of 1909 (Birmingham district, vol. 6d, p. 6), and Samuel married either Sarah Louisa Higham or Mary Ann R Hughes in the Oct-Dec quarter of 1895 (Birmingham district, vol. 6d, p. 275).
Barrowclift might have been illegible or miss-spelt on the 1911 household schedule, or mangled in the indexes, so a search for Samuel and his wife on the 1901 census might prove more successful. This approach found a Mary Ann R Barrowcliff living with a husband indexed as Tammy. Examination of the census image soon confirmed this as the correct family. I have a mental image of a burly Samuel displeased with being called Tammy.
Another approach, entering only forenames of several family members eventually did the trick with the 1911 census.
How accurate are the census indexes?
Compare the 1911 census above with the names in the indexes below.
|Samuel Burroway||Samuel Barrowclif|
|Mary Ann Rose Burroway||Mary Ann Rose Barrowclif|
|Elise Burroway||Elise Barrowclif|
|Leah May Burroway||Leah May Barrowclif|
|Alice Burroway||Alice Barrowclif|
|Fredrick Burroway||Frederick Barrowclif|
|Glads Rose Burroway||Glads Rose Barrowclif|
Likewise, compare the 1901 census above with the names in the indexes below.
|Tammy Barrowcliff||Samuel Barrowclift|
|Mary Ann R Barrowcliff||Mary Ann R Barrowclift|
|Elsie Barrowcliff||Elsie Barrowclift|
|Leah Mary Barrowcliff||Leah May Barrowclift|
|Alice Barrowcliff||Alice Barrowclift|
What do you make of the census images compared to the indexed names? Is the error rate acceptable?
© Sue Adams 2013
 Superintendant Registrar, Birmingham Register Office. Certified Copy of an Entry of Marriage. England, Birmingham county borough, Birmingham district, 1935/06/08, no 116. Coulson, Charles Spencer & Barrowclift, Gladys Rose. issued 1 August 1997, incorporates an image of the original register. Sue Adams, personal collection RWC/1/9/1/3.
I don’t know if it is acceptable; maybe more the norm. People are transcribing handwriting of others. I find that when I use the Heritage Quest Census Browse option I find more than with the index. Yes, it is time consuming going page by page, but the payoff is worth it.The handwriting varies some very readable others like those hastily penned notes we have later look at and said to ourselves, “What did I write?”. Throw in that styles of letters were different script than we are use to using, ie. Nelson sometimes looks like Nelfon or double ss looks like ff. People are human and mistakes are made. I use the index for the initial search; proceed to the browse method for confirmation, especially if I know they lived there at that time. I found my great grandfather when I searched for another family member in the same household; the browse mode showed me that he was listed a Ma Stearns. His name was Mason. The area where “son” should have been was an ink mark. I ranting, so I will stop now. I don’t know if this helps, but I understand your frustration.
Hi Dame Gussie
My intention was to raise awareness of some of the issues that arise from the quality of onliine indexes and search facilities.
As 19th century English census records go, these examples are not especially difficult. As Ancestry and FindMyPast charge for thier services, of which the index and search facilities are the primary added value, I think it is legitimate to assess the quality of the product. Should we as consumers accept poor quality indexes, produced cheaply using volunteers or workers with no paleaographic training? Is the price in line with the quality presented?
In these examples, FindMyPast did a better job. Whose indexes are better overall, or for particular data sets?
Did you try some of the new Advanced Search features on Ancestry.com? I noticed that you mentioned “used default” but I have found some additional results by NOT using the Default settings.
Yes, I tried other search features and got different results. After a while, it gets hard to keep track of just what searches I performed and it is very easy to fail to record the steps taken.
In the context of trying to compare Ancestry with FindMyPast for a search for the surname and variants, I used the default setting on Ancestry. Neither gave the result I was expecting, so I may be mis-interpreting the differences between the various search options. Just what each option does is poorly documented, as are changes to the search features.
However, the point is that fancy search features can’t overcome a mangled index. Better quality index data has a much bigger impact on the usability of the website for everyone than advanced search features. Boring for the software engineer, and more expensive for the company to implement, it would serve the customer better.
The example I was thinking about, when I replied, I went to the “default” setting link, and put selected the 3 options below the Last Name. I hadn’t found this person in one census year with the default setting. I did the same for first name. I don’t usually randomly do both at the same time, but start with the first name.
In the case I was using, I was searching for a person whose first name was “Ready”. The unexpected find was that one census had “Readz” and another “R M”. The Readz is what the census record looked like and was entered that way for searching.
Again, with this specific example, the Location, Adjacent county helped find this guy.
From my experience, the “whose indexes are better….” question is almost case by case. For one record or one surname, Ancestry may garble the transcription, and index accordingly, while FamilySearch gets it right. A different surname or dataset, and this reverses. Ditto for a number of other services I use. (No experience with FindMyPast.)
As much as the errors are irritating, if the sites, paid or otherwise, had depended on transcribers with paleographic training, there would be a tiny fraction of the now-available records out there.
From what I understand, not only are most transcribers not so trained, but many transcription contracts are outsourced to places where English may not be the birth language. And some of the typewritten records/sources are “transcribed” via OCR. In both cases, the index would be created by a program, not a person who double-checks to the image. At least volunteers are motivated, and often very experienced….
I will gladly take the vast amount of records available now, even if the indexing is garbled, and I have to get very creative to find the record. Or I have to browse through images to scan for the name…. Just as we used to have to do film/fiche/books/files on-site.
I am thrilled that the images are coming heavy and fast. I can live with trying to deal with the lousy indexing.
I agree with your observation that the quality of indexes varies greatly between records sets. Other than anecdotal evidence like your observation, I have not seen any systematic comparison published even on a small scale. It would be really useful to know that provider X’s index of the 1911 census (or pick another data set) is better than provider Y’s and whether the difference was large or small. Then I could make infomed decisions about which subscriptions would be of most benefit to me.
Ancestry does note at least some of the data sets produced using OCR, but does not assign categories of transcribers (e.g. volunteer, non-English speaking paid worker) to other data sets. I suspect that OCR and non-English speaking workers are less accurate, but could be wrong. I would like to see companies be more open about the method of index creation for each data set.
I do distinguish between paid for and free websites. If I am paying, I demand a fair service in return. Genealogical data services are no different to any other commercial service, so I am not a grateful puppy.
I differ with you in my opinion of what is more important, quantity or quality. Quality matters. It matters a lot. Poorly indexed records remain effectively unavailable to the less experienced researcher, and costly in time for the experienced researcher.
Images and indexes do not have to be published at the same time. Perhaps the images would keep you happy, while I wait for a better quality index?
Oh, I *completely* agree with you on the quality issue, especially on *paid* sites. And with much of the rest of your post. I think that the big issue is the indexing method, and most indexing is entirely dependent on the quality of the transcription itself.
You seem to be saying that a human being is indexing these transcriptions. Indexing, generally, is not done by humans but is created by some program which depends on specific fields entered by the transcriber. So if the transcription is faulty, the indexing will pick up the faults, and no human comparison with the original will be involved. Same with OCR, where the indexing is by program, and will pick up the faults in the OCR transcription. Again, no human compares.
There may come a time when when this is handled differently, and hopefully better. But not now. The best we have now are the sites that allow/encourage comments/corrections/post-its to note indexing errors, with the corrected information. Or those that let you contact them with such corrections. Response can be slooowwwww. I suspect they are flooded.
You said: “I have not seen any systematic comparison published even on a small scale.” That would be truly marvelous! Who has the time, or the economic impetus to do this, considering the millions of records indexed on various sites?
There are sites that *do* have records transcribed by experienced individuals, that are then proofed, and indexed, by other experienced individuals. They are invaluable. But their output is small. Correct, but small.
For the flood of records available over the last 25 years, I really am a grateful puppy. Errors and all….
I am glad to see you recognise the quality issue, yet you are still somewhat uncritical.
I do not think that the indexes are produced by the two step process you describe, transcription/OCR then automated indexing. For clarity, I will define some terms:
Transcript – an accurate and complete copy of the original, including layout, original spelling, punctuation
Abstract – a shortened version that contains the important information, but is not an exact copy
Index – an ordered list of one catagory of information e.g. names
The demarcations between these definitions are blurred. Online searchable databases are somewhere between an index and abstract.
Ancestry’s World Archives Project is a good indication of how searchable databases are compiled by volunteers. The World Archives Project software expects users to extract just the fields of interest e.g. names in the form of a table ready for inclusion in the database – a one step process. It is not clear what, if any, data processing occurs after the users have keyed in the data, but I have not seen evidence of data checking.
You ask “Who has the time, or the economic impetus” to perform quality control? Companies worth multi-millions can afford to do at least some small scale spot checks, and eliminate the less accurate methods. Isn’t that just plain good customer service. As a customer I would encourage you to be more critical, so the companies see that better quality control is an economic neccesity.
On a lighter note and back to the puppy analogy. Perhaps recognising the quality issue is like learning to sit, but being a critical customer is like learning the more difficult stay. Umm, as a puppy trainer, I would not want anyone’s tail to stop wagging.
[…] Is a Rare Surname easy to find? Searching for Barrowclift with Ancestry and FindMyPast […]