Monday, 2 June 2014

A look at the genetic homeland case reports from English Origenes, Irish Origenes and Scottish Origenes

I've had a number of people writing to me in the last few months to express concerns about the genetic homeland case reports offered by Tyrone Bowes through his English Origenes, Irish Origenes and Scottish Origenes websites. The reports have also been the subject of much discussion in various groups and forums (see the links at the end of this article). I've now spent some time reviewing the methodology and the case reports, and I share the concerns of my correspondents so I thought would take this opportunity to highlight some of the problems.

Tyrone Bowes claims to be able to use "modern science" to pinpoint the place of origin of one's patrilineal ancestors one thousand years ago based on the results of a 37-marker Y-DNA test from Family Tree DNA. The distribution pattern of the surnames that appear in the person's match list is used to identify the "genetic homeland", which is defined as "the area (within a 5 mile radius) where one's ancestors lived for hundreds if not thousands of years. It is the area where one's ancestors left their mark in the history and place names of that area and in the DNA of its current inhabitants."  I understand that the reports cost in the region of  $300 a time. A free consultation is provided to determine whether or not it is possible to pinpoint a "genetic homeland". An 80% success rate is claimed for the Irish and Scottish reports, but no claim is made for the English case studies.

As far as I am able to establish the methodology used for the reports has not been legitimised by publication in a peer-reviewed scientific journal. An article by Tyrone Bowes entitled "Using Y Chromosome DNA Testing to Pinpoint a Genetic Homeland in Ireland" was published in April 2013 in the Surname DNA Journal. This journal was founded in January 2013 by Brad Larkin, a genetic genealogist who runs the Larkin DNA Project. Only five articles have been published in the journal to date, two of which have been written by Brad Larkin, and a third article has been contributed by a Larkin cousin. Although I am given to understand that the articles are reviewed prior to publication, the journal has no named editor and no editorial board so we cannot be certain that the people who are reviewing the articles have any expertise in the subject matter they are commenting on. Brad Larkin and Tyrone Bowes appear to have a business relationship. They jointly founded the Genetic Homeland website in March 2013, and this website is advertised prominently on the home page of the Surname DNA Journal.

The article in the Surname DNA Journal outlines Tyrone Bowes’ hypothesis that "Using commercial Y chromosome DNA testing, the Family Tree DNA database, the 1911 census of Ireland, Microsoft Excel, customized mapping software for surname distribution mapping, ordinance [sic] survey Ireland maps, and Google Earth it is possible to explore the relationships between a test subjects surname, and the surnames of his genetic matches to pinpoint the Genetic Homeland of an individual, and find evidence of their ancestors presence in the placenames and DNA of the areas current inhabitants." Eight case studies are presented to illustrate the methodology. However, the author makes no attempt to test the hypothesis scientifically. In the discussion it is stated that "the pinpointed 'Genetic Homeland' can be verified by Y chromosome DNA testing of people with the test subjects surname in the identified area". It is surprising, therefore, that such testing was not done for all the surnames in question to establish whether or not the hypothesis had any validity. It is claimed in the article that one person with the surname Bowes was tested to confirm the "Genetic Homeland of Clan Bowes" which is supposedly pinpointed to "an area centered upon modern day Abbeyleix in county Laois". However, this conclusion is disputed by Martha Bowes, the administrator of the Bowes DNA Project.

Despite the lack of scientific evidence in support of the methodology it seems that a number of people have ordered the genetic homeland case reports, and selected reports have been made available on the various Origenes' websites. It seems to me that the methodology is fundamentally flawed, and is based on a number of false assumptions. I've summarised below some of the main problems:

1) While matches with other surnames can often provide valuable genealogical clues, the claim to pinpoint a genetic homeland with such precision is not supported by the evidence provided. The technique does not account for the inherent uncertainty in the estimates of the time to the most recent common ancestor (TMRCA). The calculation of a TMRCA is based on probability, and there are many variables such as sample sizes and mutation rates. Consequently TMRCA is not an exact science. It does not provide a precise point in time (eg, seven generations ago) but a probability distribution  – a range of time in which the common ancestor might have lived. For example, a match on 34/37 markers could indicate a common ancestor who lived 200 years ago or two thousand years ago. It is therefore important that DNA evidence is considered in combination with genealogical evidence and not in isolation.

2) The Origenes' method takes no account of the biased nature of the Family Tree DNA database. Around 70% of FTDNA customers are thought to reside in the United States, and matches with other surnames may often reflect non-paternity events in the US rather than the origin of a surname in the British Isles. This point is nicely demonstrated in Howard Mathieson's critique of the Irish Origenes' case study of the Kiely surname. Furthermore, the FTDNA database has inevitable gaps, and many British and Irish surnames, particularly lower frequency surnames, are not yet represented.

3) The Origenes' reports do not appear to make any attempt to verify the SNP status of the people in the match lists whose surnames are used to pinpoint the "genetic homeland". SNP testing is important when investigating matches with other surnames in order to ensure that the matches are not false positives. Two men can have matching Y-STR results, but if they do not share the same SNPs they will belong on different branches of the Y-DNA tree and will not share a common ancestor within the last few thousand years. This problem occurs as a result of convergence. Although 37-marker results are most commonly affected the problem can still occur with more distant matches at 67 markers.  Convergence is a particular problem in haplogroup R1b, the most common haplogroup in the British Isles which is found at a frequency of about 70% in England and over 80% in Ireland.

4) The assumption has been made that the geographical overlap of the matching surnames is an indicator of the "genetic homeland". However, it is conceivable that many groups of randomly selected surnames would overlap purely by chance but no attempt has been made to rule out this possibility.

5) The methodology is based on the hypothesis that "genetically recurring surname matches" provide a snapshot of one's "medieval ancestor’s neighbours". If the theory is correct, then the methodology needs to be applied consistently, but a close examination of the case reports reveals many inconsistencies. Sometimes recurring surnames are omitted for no apparent reason whereas in some of the case reports singleton matches are included, even though the technique requires that singletons should be omitted because they are claimed to be indicators of non-paternity events. For example, in the Patterson case study on the Scottish Origenes website it is apparent that Mr Patterson has recurring 67-marker matches with a wide variety of different surnames. Only the closest matches with a genetic distance of three or less are considered important enough for consideration for this report. Furthermore, only a few select surnames (Henderson, Stewart, Chisholm, McKay, McLean, Logan, and McDonald) are cherry-picked for inclusion while many of the other recurring surname matches (Norton, Stephenson, Turner, Johnson, Edwards, Tate, Rock) are excluded, presumably because the distribution will not support the proposition that Mr Patterson's "ancestral genetic homeland is centred upon Knockbain on the Black Isle", just north of Inverness. The opposite problem applies to the case report for Mr Henderson who inconveniently only had matches with other Hendersons at 67 markers. His matches at 37 markers were used instead in combination with more distant 37-marker matches found on Ysearch. For this report all the matching surnames were taken into account, including those surnames which only occurred once in the match list. Similarly in the Bennett case study on the English Origenes website the surname French is included in the report despite the fact that the surname only appears once.

6) The reports show no understanding of the evolution of place-names, and the evidence used in support of the genetic homeland stories is implausible at best. For example, in the Bennett case study we are told that "The Bennetts of Somerset cluster near the town of Burnham on Sea close to the Bristol Channel". We are further told that "the local placenames that reflect the ancestral link of the Bennetts and their genetic cousins with the surrounding area" include Bennett road, Lockswell, Locksway, Seymour road, Seymour Court, Coat (village), and Coate farm. Place-names can of course be very informative but most roads have been built and named in the last century or so, and road names are therefore highly unlikely to correlate with the presence of a surname in the locality one thousand years ago, especially in a seaside town which has probably had an influx of residents from many different parts of the UK.

DNA testing and surname distribution mapping are both very powerful tools for a surname study but they should always be used in combination with genealogical and historical records, and the results should be interpreted with caution. In order to determine the origin or origins of a surname it is necessary to map a surname at different points in time to establish continuity. Ultimately there is no substitute for a carefully conducted worldwide surname study which makes use of all available records from the beginnings of surnames to the present day.

Further reading
For further discussions on the limitations of the Origenes' methodology see:
- My letter in Family Tree Magazine about "genetic homeland" stories (August 2014 issue)
- A civil discourse on Irish Origenes' methods, a discussion on the Anthrogenica forum
- A Review of Irish Origenes' Bowes Case Study‎ by Martha Bowes, administrator of the Bowes DNA Project and one-name study
- Irish Origenes  and Irish Origenes – the next generation, discussions on the Guild of One-Name Studies' mailing list
- An analysis of the English Origenes' report on the Tucker surname by Howard Mathieson. Note that the Tucker case report has since been removed from the English Origenes website.
 - Bowe research methodology queried - a letter from Fíona Tipple in the June 2014 issue of the Genealogical Society of Ireland's Gazette (page 2).
Can the Distribution of 19th Century Farmers Be Used To Identify a Surname's Genetic Homeland?
 by Howard Mathieson

A list of surname mapping resources can be found in the ISOGG Wiki.

I have discussed the available sources and the techniques for surname mapping at length in my Surnames Handbook (History Press, 2012).

For information on the methodology used in a one-name study see the website of the Guild of One-Name Studies.

See my Sense About Science blog post Sense About Genealogical DNA Testing for a summary of the legitimate inferences that can be made from DNA tests.

© 2014 Debbie Kennett

10 comments:

Howard Benbrook said...

As expected, a devastating analysis of the spurious claims by Tyrone Bowes; thank you, Debbie.
I can claim a little experience in some of this, particularly in respect of surname distribution. It's now widely accepted that, if your surname shows a significant concentration using Victorian census data, it illustrates that this will indicate the geographic origin of the name. Even the academics engaged on a revision of the definitive work on English surnames seem to accept this theory.
But there's a problem. And, as with DNA, it's a statistical one. To make confident claims, the sample size needs to be substantial. If you have something like 100 people bearing the same surname in the Victorian era, all concentrated on a particular area, it may seem conclusive but there is no way of refuting the suggestion that the originating family may have moved to that place, lock, stock and barrel, many hundreds of years earlier. It's, quite simply, not PROOF.
By the way, using Victorian evidence, the homeland of the BOWES surname appears to be the North Riding of Yorkshire!

Anonymous said...

I agree that the use of 37 marker matches doesn't tell you much, but if you have a lot of 67 marker matches, you can get some idea of your surname's place of origin (not necessarily within 5 miles, though). The Patterson study (not Paterson as you indicated, although it is the same name) was purchased by a group of Pattersons (of which I am a member) with decent 67 marker matches. If you revisit Bowes report, you will see he states the "genetic homeland" 1,000 years ago seems to be the north shores of Loch Fyne, with an outmigration in the 14th century with matches (primarily Logan) on the Black Isle. There is historical evidence of a Pat(t)erson group on the north shores of Loch Fyne and there was a depopulation of Argyll due to economic conditions at this time. You mention the close Norton matches were ignored, but Norton is a form of MacNaughton and this name appears on the Clan maps on the north shores of Loch Fyne. A MacNaughton was granted these lands in the mid 14th century. So I don't think Bowes is totally out to lunch.

Brian Swann said...

A very nice summary of all the issues, Debbie. The truth is out there (somewhere). There are no short cuts in this sort of research.

Debbie Kennett said...

Howard

Many thanks for your insightful comments. The Family Names in the UK team are looking at surname distribution over time so it will be interesting to see what they come up with. The problem is that none of the earlier records give such full national coverage as the Victorian censuses. The distribution of a surname in the nineteenth probably does often correlate with its origin, though I suspect there are many surnames which, from census data alone, look as though they might have originated in London, Bristol or Manchester! As you say, particular caution needs to be exercised with low-frequency surnames and of course the surname mapping evidence should always be used in combination with other sources.

Martha Bowes has a map of the distribution of the Bowes surname in Britain in 1881 using Steve Archer's Surname Atlas CD:

https://sites.google.com/site/bowessurnames/m/england/1881-census-map

Debbie Kennett said...

My response to the comment about the Patterson case report is provided below split into two parts. Here is Part 1:

I agree that 67-marker matches can be informative but they need to be used in combination with the genealogical information provided by the users. Knowing that you match six people with the same surname with ancestry from North Carolina is not going to tell you very much because you've got no idea if the progenitor of the surname in America is descended from a legitimate or an illegitimate line.

The surname Patterson or Paterson simply means son of Patrick, and will have originated independently in multiple locations. It is a very common surname. In the 1881 census there were 20,417 Patersons in Britain who can be found all over Scotland and at low levels in virtually every English county. There were 10,449 Pattersons in 1881. The surname is again found throughout Scotland but is actually most prevalent in Northumberland and Durham. I've changed the spelling in my blog post to Patterson. However, I note that Tyrone Bowes has spelt Mr Patterson’s surname inconsistently throughout his report.

Norton and MacNaughton are completely different surnames. Norton is derived from one of many places called Norton. There were 9,963 Nortons in Britain in the 1881 census. The surname is found only at low frequencies in Scotland but is prevalent throughout England and particularly on the east coast, the Midlands and the West Country. MacNaugton has a completely different derivation. According to Black's Surnames of Scotland the name is derived from the Pictish word Neghtan. The distribution in 1881 is quite different from that of Norton. MacNaughton is found throughout Scotland. It is only found in a few places in England – in Lancashire and neighbouring counties, Oxfordshire and in the south-east.

It’s not just the Norton surname that is ignored but also many of the other matching surnames. The surnames Chisholm and McDonald are used for analysis purposes for the maps. Both of these surnames are at a genetic distance of three from Mr Patterson, but all the other surnames at the same distance (Stephenson, Turner, etc) are ignored. If you’re going make inferences about origins from match lists you need to look at all the genetic matches and not just the closest matches. In any case, genetic distance is not linear as Tyrone Bowes supposes, and you cannot allocate matches to a specific timeframe based on genetic distance as he tries to do in the Patterson report. That is why it's so important to know the SNP status.

The other surnames that match Mr Patterson at a genetic distance of one or two all occur at very high frequencies in the 1881 census and are widespread through Scotland and England:

Logan 7,108 (rank 599)
McKay 17,101 (rank 222)
Henderson 32,554 (rank 100)
Stewart 46,914 (rank 63)

If you map all these surnames there are numerous locations where they all overlap. You could cherry-pick any of the other names in the match list and use them in combination with these surnames to come up with a variety of different overlapping locations to suit whichever story you wanted to tell.

Note that Tyrone Bowes only uses farmers in his maps but by doing so he makes the false assumption from the outset that everyone has a land-owning farming ancestor. You need to look at the surname population as a whole and not just at a subset. The people who emigrated to the United States, whose descendants are the primary purchasers of the Origenes' reports, are probably much more likely to be descended from the tenant farmers who were evicted during the Highland Clearances.

Debbie Kennett said...

This is Part 2 of my response regarding the Patterson case report.

Even when you use Tyrone Bowes’s own data the maths doesn’t make any sense. In Figure 3 in the Patterson report he provides figures for farmers with the matching surnames in the 1841 census and tells us that Mr Patterson's Y-DNA results are associated with Ross and Cromarty in Northern Scotland. However, if you add up the totals in the columns there are 349 farmers in Ross and Cromarty but 682 in Invernessshire, so why is Ross and Cromarty chosen in preference to Invernessshire?

The migration story is only introduced because no evidence of place-names associated with the Patterson surname could be found in Knockbain on the Black Isle, which was supposedly the genetic homeland of Mr Patterson one thousand years ago. However people generally take their names from places rather than the other way round. Where surnames are used in place-names they appear as manorial affixes (eg, Milton Keynes). So we wouldn't expect to find places named after Pattersons denoting origins one thousand years ago. The report actually makes little sense because having told us that the genetic homeland is in Knockbain, he then doesn't believe the results of his own methodology and instead comes up with a story about a migration from LochFyne (which he mis-spells as Long Fyne and Lough Fyne). He arbitrarily chooses a different selection of surnames from the match list (Campbells, McGregors, Buchanans, McFarlane, McLaren, McAskills, McArthur), all seemingly specially selected for their Scottishness and their association with Argyllshire, presumably because of the documented link with Clan Paterson. Many other frequently occurring surnames are omitted with no reason given (eg, Wilson, Ferguson, Livingston, Young, Alexander, etc).

Surnames in any case did not "become common in the 10th century AD" as Tyrone Bowes states in his report. In Scotland surnames were introduced by the Normans but the adoption of surnames was a slow process. The landed families had surnames by the 1300s but in the Scots-speaking areas surnames weren't adopted until the sixteenth century or later. Surnames ending in –son are quite late developing. A search of the POMS database (http://www.poms.ac.uk), which covers the period from 1093-1314 returns no hits for Patterson or Paterson. The earliest reference to the surname Patterson in Black's Surnames of Scotland dates from 1446. Black cites numerous other early references to the surname from the 1500s from all over Scotland. (Incidentally you can download a copy of Black from FamilySearch Books: https://books.familysearch.org.

Tyrone Bowes' lack of even a basic understanding of surnames is evident in the nonsensical conclusion to the report:

"Mr Paterson’s Y-DNA results demonstrate that when surnames became common in the 10th Century AD his Paterson-Adam lived in the area north of Lough Fyne. His Paterson-Adam lived surrounded by relatives who became Campbells, McGregors, Buchanans etc. At some point between 1200 and 1400AD his ancestors migrated north to the Black Isle where some of his ancestors picked new surnames like Logan and Chisholm."

The whole idea of a mass migration from Loch Fyne to Knockbain between 1200 and 1400 in any case seems highly unlikely. I would imagine the Highland terrain would have been virtually impassable back then. As far as I'm aware there was no link by water as the channel through to the Moray Firth was only dredged in 1917. If people want to move for economic reasons they tend to move to big cities like Glasgow or Edinburgh where work can be found rather than obscure parishes on the other side of the country.

Marie Byatt said...

I think you make a great case up to your last paragraph. Then you too slip into generalities. Mass migration might have been difficult in the years mentioned, even unlikely but definitely able to be accomplished. Also people moved out of cities as well as to them. In short, people in earlier times gave in to nagging spouses, poor jobs, religious zeal, lack of a mate, curiosity and numerous other reasons for moving much as they do today - the technology was different but the underlying reasons were as diverse and there were people. Afterall, how did they get to the loch in the first place and from where?

Martha Bowes said...

Hi Howard, Just a little note about the Bowes name in England in the context of this discussion. It is completely separate from the Bowes commonly found in the south of Ireland, who are variants of Irish Bowe, an anglicization of the Uí Bhui or Uí Bhaigh.

Debbie Kennett said...

Marie, I agree that people have always migrated and for various reasons. It is very difficult to re-construct the migratory paths of our ancestors, particularly when the documentary records are scarce. However, to understand migratory patterns you need to look at the road and river networks and the physical geography. People often followed coastal paths or migrated along the route of a river. The proposed migration makes little sense though of course it's not impossible. Nevertheless the point remains that it is a theory which is not supported by the evidence provided.

Mountain Mama said...

Thanks Debbie for another well reasoned analysis. Anyone understanding surnames wouldn't fall for this nonsense. For example, my maiden name means "small farmer". I am just sure we are all related. I also expect that during the middle ages, when a son moved to the city and started a new occupation, there was a good chance he changed his name to reflect his new status.

It's true, there IS a sucker born every minute.