DNA survey - please respond

[pjd]

Here goes - the majority of my ancestors come from Yorkshire & Lancashire way back plus 1 branch from Norfolk but Middlesex doesn't feature much so somewhat surprised by that county coming out top!

 

[peter]

As you will see, Yorkshire has come well down in his list while Durham is much higher.

Thanks - I think that's another example of the algorithm being biased against large counties. I'll see if I can come up with a better formula once I've got the newsletter out of the way....

 

[peter]

Here goes - the majority of my ancestors come from Yorkshire & Lancashire way back plus 1 branch from Norfolk but Middlesex doesn't feature much so somewhat surprised by that county coming out top!

It isn't - it's bottom (the table should be sorted from low to high, not high to low). Can you re-sort and this time highlight the main counties in your tree based on the ancestry of your 3G grandparents?

Again this is an example of the bias against big counties. I'll definitely revise the formula!

 

[jorghes]

I'm still working on the algorithm for the rest of my DNA results (I'm getting a slightly different list each time!) - but I had a thought, perhaps the algorithm would work better if it was more in relation to the number of ancestry labelled "4th cousins" according to county?

Anyway here is the breakdown for my 32 3xGs, but this is a little misleading as I will explain, but it is as follows:
Hampshire - 1
London - 3
Yorkshire - 2
Lancashire - 1
Cambridgeshire - 1
Sussex - 1
Staffordshire - 2
Durham - 2
Gloucestershire - 2
UNKNOWN - 2

The rest of the 32 are from Ireland (2); Scotland (6) and Wales (7). The "slightly misleading" part of this list is that my ancestors frequently, it seems, moved around quite a bit, for example my Jewish 3x GG are on this list as London (1) and Sussex! The Lancashire on this list is another whose family kept moving as his parents were born in Cambridgeshire and Lancashire respectively, while the Cambridgeshire on this list also had a parent born in Lincolnshire!
I would have added an extra two for Staffordshire, since that is the most likely place the relatives of my illegitimate great-grandfather came from, but I thought I had better add them to the "unknown" section.

 

[Heather]

Go to your DNA page at Ancestry...............

Thanks Pauline, you are a star, I'll give it a try.

 

[pjd]

It isn't - it's bottom (the table should be sorted from low to high, not high to low). Can you re-sort and this time highlight the main counties in your tree based on the ancestry of your 3G grandparents?

3 x great grandparents:
Yorkshire (West Riding) - 25
Lancashire - 1
Norfolk - 2
Cambridgeshire - 2
Unknown - 2

So really the only 'main' county is Yorkshire

 

[Pauline]

Hmmm When I just search for England I only get 40 pages. I think the search is completely literal, searching for England only searches for the word England and not where England exists in the format, county, England

It does raise the question as to how reliable the Ancestry search is.

My husband's brother has also done a DNA test which is linked to my husband's Ancestry tree. And since most of their ancestry is in Yorkshire, searching for Yorkshire matches in either set of results ought to bring up the other at the top of the list of matches.

However, when searching for Yorkshire matches in his brother's results, yesterday my husband did not show up as a match, despite repeating the search several times. This morning, though, he is showing up as a match. So something is amiss somewhere.

 

[peter]

perhaps the algorithm would work better if it was more in relation to the number of ancestry labelled "4th cousins" according to county?

Perhaps it might, but as I mentioned earlier, there is no way of working out the numbers for a subset of your matches.

 

[peter]

When I just search for England I only get 40 pages

I've discovered that you can never get more than 40 pages of results when carrying out a birthplace search.

My wife, who is three-quarters Welsh, gets 40 pages for England, the same as you and the same as me. But whereas I get only 11 pages for Wales, she gets 40. And she gets 16 pages for Glamorgan, I get 3. I should mention, however, that she has more than 50% more total matches (471 pages vs 308), so you'd expect her figures to be a bit higher.

For many English counties she has the same number of pages as me, or fewer. Exceptions that stand out are Cheshire (10 vs 5), which is right next to north Wales, and Gloucestershire (15 vs 9) which is where her English ancestors came from. Other exceptions are Herefordshire, Shropshire, Devon, Cornwall, and Somerset - also close to Wales.

Had she been adopted it would be absolutely clear from these numbers that she had Welsh ancestry - so clearly this technique has some validity. It's just a question of refining it so that the signals stand out from the noise.

 

[Pauline]

I've discovered that you can never get more than 40 pages of results when carrying out a birthplace search.

Ah, that explains why my husband’s brother, despite having almost twice as many matches overall, still only has the same number of pages of Yorkshire matches - 40.

 

[Sue345]

Hope I manage to upload this...

 

[jorghes]

Perhaps it might, but as I mentioned earlier, there is no way of working out the numbers for a subset of your matches.

Oh I didn’t mean specific 4th cousins, but rather the labelled section “4th cousins” that Ancestry uses which could cover anyone from 3rd cousins to 6th (I say 3rd because my father has a half 3rd cousin who pops up in the “4th cousin” bracket - half because they descend from different wives of the same ancestor.) As opposed to using the mostly “distant cousin” label, which might hold incorrect matches.

 

[peter]

Oh I didn’t mean specific 4th cousins, but rather the labelled section “4th cousins” that Ancestry uses which could cover anyone from 3rd cousins to 6th (I say 3rd because my father has a half 3rd cousin who pops up in the “4th cousin” bracket - half because they descend from different wives of the same ancestor.) As opposed to using the mostly “distant cousin” label, which might hold incorrect matches.

You can't search any subset. No need to worry about incorrect matches because they'll be white noise.

 

[Bob Spiers]

Talking of subsets but of a different type, there was one such in my household as a result of commenting on Tim's comment that all his ancestors came from Lancashire, which I thought quite extraordinary (as did Peter to a degree). I mentioned this to my wife who promptly responded with - "well all mine come from Northamptonshire so what's odd about that"? Of course I could not let that lie having researched her Tree, and in passing on a comment that "the bulk may well do so, but by no means all", I was called upon to disprove her assertion - which incidentally she has held for years (and her family before her). So I did the 3 x G exercise explaining it went back to the early 1800's and presented her with the results, and in the end it surprised me as both Huntingdonshire & Worcestershire came into the picture. However it did support a very strong Northamptonshire ancestry as these results show: (percentages rounded)

Northamptonshire =22(68%)
Buckinghamshire = 3.5 (11%)
Bedfordshire = 0.5 (2%)
Huntingdonshire = 2 (6%)
Worcestershire = 4 (13%)

But of course I forgot the female 'never say die' logic when she said, well they all settled in Northampton (not even 'shire') in the end. Game set and match to her me thinks.

 

[peter]

A reminder for anyone still to contribute is that you'll need to know (or guess) which counties the ancestors of each of your 3G grandparents came from. It's not where your 3G grandparents ended up, or where they married, or even where they were born - it's where their ancestors came from, to the best of your knowledge.

 

[Bob Spiers]

It's not where your 3G grandparents ended up, or where they married, or even where they were born - it's where their ancestors came from

I think that advice is a little over complicated Peter -especially the 'where they were born bit' - and especially for the period in question late 18th early 19th Century. By and large people remained were they were born and likely married someone from the same catchment and set up home within a stones throw of their respective families. At the extreme they might stray across a boundary line (such as it was at that time) and this would apply in the majority of cases for ancestors as well.

Ruling out enforced migrations fleeing uprisings/famines (et al) by far the majority of County - and indeed Country - migrations occurred at 2 x G and certainly 1 x G levels mostly in search of work. So, in my book noting 3 x G birth places (hamlets, villages and occasionally towns and in which county they lay) almost certainly will be enough.

 

[peter]

I think that advice is a little over complicated

It's not advice, it's the instructions. Nobody is forced to take part, but it's important that contributors follow the instructions, otherwise it adds an extra layer of uncertainty to the process.

In some cases it will be a reasonable assumption that our 3G grandparent's ancestors came from the same county, but in other cases it won't (eg if the 3G grandparent was born in London). And in other cases we'll know, or strongly suspect, that one or more of their ancestors came from elsewhere - it all depends how successful our research has been, and what clues there are. For example, if I had an ancestor called Alexander McTavish I probably wouldn't assume that all his ancestors came from Essex, and if I had an ancestor with a German-sounding surname who was a clockmaker, a sugar baker, or a pork butcher, I wouldn't assume that his ancestors were born in Surrey.

All of my 2G grandparents lived in London at some point during the 19th century; some of them were born there. I don't know that any of them had ancestors who were born in London, and the fact that I've struggled to trace their ancestors, even though almost all of the parish registers have been online for a decade suggests that these lines originated elsewhere.

(But London is a special case, and is effectively excluded from the trial because searching for Middlesex birthplaces doesn't find London birthplaces.)

 

[canadianbeth]

My Ancestry kit was just shipped on August 31st and apparently might have cleared customs yesterday so it will be a while before I get any results since I have to send it back and then wait. We might be having a postal strike later in the month and that will hamper things even further. If you still want my results by then, I will try to figure them out and post them.

 

[Pauline]

for the period in question late 18th early 19th Century. By and large people remained were they were born and likely married someone from the same catchment and set up home within a stones throw of their respective families. At the extreme they might stray across a boundary line (such as it was at that time) and this would apply in the majority of cases for ancestors as well.

I think this is something of an over-generalisation, and our ancestors were a lot more mobile than we sometimes give them credit for. It may be true for some people, in some areas, but many people did move even quite long distances in this era.

Some of my 3G grandparents were born in the area where they had roots going back several generations, but a fair few weren't. For example, I have Lincolnshire folk with roots in Huntingdonshire, Nottinghamshire and Norfolk . One 3G grandmother was born in Kent, married in London and died in Surrey, but her roots were in Sussex and Berkshire. Others stayed within a county but moved from one end to the other.

 

[Bob Spiers]

I think this is something of an over-generalisation, and our ancestors were a lot more mobile than we sometimes give them credit for. It may be true for some people, in some areas...Some of my 3G grandparents were born in the area where they had roots going back several generations

Yes I think it is perhaps an over generalisation , but can only speak for my own and my wife's Trees, which mirror what you say about roots going back generations (my Irish ancestors excepted). Fortunately I do not have a 'London' problem, although 18th & 19th century place names and county designations for 'Birmingham' and its environs can cause bewilderment without knowledge of the area; which fortunately I do have.

But to prove my point (for my own benefit of course) I have set about analysing 4G ancestors even though I do not at this time have all 64 established. However I have more than enough known to be able to test my theory that there is little location variations within my own Trees.

I am about a third of the way through and so far have found one instance where at 3G level I split between Worcestershire & Gloucestershire, which at 4G level I would opt for Gloucestershire alone. Most likely the same will apply when it comes to Warwickshire/Worcestershire and Warwickshire/Staffordshire depending on boundary demarcations of the time. In all probability it will likely come out 'even Stevens', little different overall to what I have already set out for 3G. But the 'proof of the pudding' means I need to press on with my analysis.