Walter de Caen and the Dickinson line

[Dora Smith]

I have an allegation that my Dickinsons (well off merchants of Hull,
Yorkshire, who intermarried with the De la Pole's) are "generally assumed"
to be descended from Walter de Caen, a descendant of Rollo the first Duke
of Normandy (and therefore of the kings of Norway), who came to England
with his kinsman, William the Conqueror. But the allegation says this
isn't proven. I am interested, but I question this, since Dickin or
Dickon is a very common name in Yorkshire, where the Dickinson's lived.
It seems as if Dickinson would be a common surname. Like Johnson and
Williamson and Richardson and Robertson.

My allegation says only that Walter de Caen was the grandson of Walter de
Caen son of Rollo, first Duke of Normandy, who got the Town and Castle of
Caen for an inheritance. Nathaniel Dickinson's lineage is traced only to
the 13th century (a few generations before the De la Pole marriages).
It doesn't even say where he settled in England.

As to the Dickinsons, it says, and proceeds to prove on a generation by
generation basis, that "There is ample evidence that over the ensuing
generations the Dickinsons were well to do builders, merchants, landowners
and men of importance in their communities. They married into well
established and affluent families. Some of the descendants held highp
ositions in government and the church.Others were scholars and writers of
record."

The implication is that this is the sort of family that could be of royal
blood. But Yorkshire was never fully incorporated into the feudal system
that was in its most well developed form brought to England by the Normans
who were able to force it on people closer to London; I don't know
Yorkshire ever knew serfdom, the ordinary man owned land that had been in
his family since it was parcelled out by the Danish Viking rulers of the
area, I think. Ownership of land was the basis of wealth, and became a
reason why people of Yorkshire helped spearhead the Industrial Revolution.
Success as a merchant depended on the means to outfit oneself to begin
with, but it seems as if it was largely a matter of luck, and it was a
path to upward mobility. Not taht every peasant was likely to be a
successful merchant, but many successful merchants had common backgrounds.
There was actually a sizable, thriving class of tradesmen and merchants,
as well as doctors and lawyers, all through medieval times, and this was
one of that economic system's fundamental contradictions. Quite a few of
my ancestors have been established to have belonged to this class in late
medieval times.

What I would like to know is, more about Walter de Caen. Where did he hold
lands, who were his children and who did they marry, and what happened to
them? Is there any known connection to Yorkshire or to the vicinity of
Hull, where it seems the Dickinsons always lived, atleast by 1260,
century, and any of the family of Walter de Caen? They would have had 200
years, about six generations, from 1066 to find their way to Hull.

Yours,
Dora Smith
cl...@freenet.buffalo.edu
dora...@usa.net

[Todd A. Farmerie]

Dora Smith wrote:
>
> I am interested, but I question this, since Dickin or
> Dickon is a very common name in Yorkshire, where the Dickinson's lived.
> It seems as if Dickinson would be a common surname. Like Johnson and
> Williamson and Richardson and Robertson.

Trust your instincts here. There is no reason to assume that any two
Dickinsons are related, and to talk of the "Dickinson family" is like
talking about the Smith family. All the name means is that someone had
an ancestor named Richard who used the common nickname Dick, with the
diminutive ending -> Dicken.

> My allegation says only that Walter de Caen was the grandson of Walter de
> Caen son of Rollo, first Duke of Normandy, who got the Town and Castle of
> Caen for an inheritance.

No such person can be shown to have existed. Rollo had two, and only
two children documented in the historical traditions of normandy -
William, later duke, and Gerloc, who married the Duke of Aquitaine.
(There is an additional daughter who appears in Icelandic tradition,
this information is not independently supported, and perhaps false.)
There is no son Walter.

> What I would like to know is, more about Walter de Caen. Where did he hold
> lands, who were his children and who did they marry, and what happened to
> them? Is there any known connection to Yorkshire or to the vicinity of
> Hull, where it seems the Dickinsons always lived, atleast by 1260,
> century, and any of the family of Walter de Caen? They would have had 200
> years, about six generations, from 1066 to find their way to Hull.

As to the younger Walter, there may have been a Domesday baron of that
name, but it is unlikely that his male-line descendants would have used
the surname Dickinson, and anyhow, your source couldn't trace it, or
they would have. This whole thing is likely an unsupported tradition,
perhaps based on nothing more than Walter holding at Domesday the land
that some Dinkinsons would hold 300 years later, but not the stuff of
reliable genealogy.

taf

[Leslie Mahler]

>I have an allegation that my Dickinsons (well off merchants of Hull,
>Yorkshire, who intermarried with the De la Pole's) are "generally
assumed"
>to be descended from Walter de Caen, a descendant of Rollo the first
Duke
>of Normandy (and therefore of the kings of Norway), who came to England

>with his kinsman, William the Conqueror. But the allegation says this

>isn't proven. I am interested, but I question this, since Dickin or

>Dickon is a very common name in Yorkshire, where the Dickinson's lived.

>It seems as if Dickinson would be a common surname. Like Johnson and
>Williamson and Richardson and Robertson.

>My allegation says only that Walter de Caen was the grandson of Walter

de
>Caen son of Rollo, first Duke of Normandy, who got the Town and Castle
of

>Caen for an inheritance. Nathaniel Dickinson's lineage is traced only
to
>the 13th century (a few generations before the De la Pole marriages).
>It doesn't even say where he settled in England.

There was an excellent well documented article in NEHGR about a year
ago by
Clifford Stott which shows the correct origin of Nathaniel Dickinson.

Nathaniel was from Lincolnshire, and his lineage has been reliably
traced
back a few generations. The said article shows that the descents from
nobility and the supposed
connection with Caen were apparently invented.

Leslie

[Richard Borthwick]

At 01:19 PM 17/09/99 +0800, you wrote:
>>X-From_: GEN-MEDIEVA...@rootsweb.com Fri Sep 17 09:27:29 1999
>>Resent-Date: Thu, 16 Sep 1999 18:26:17 -0700 (PDT)
>>X-Authentication-Warning: bl-1.rootsweb.com: news set sender to
>use...@rootsweb.com using -f
>>Old-To: GEN-MED...@rootsweb.com
>>Date: Thu, 16 Sep 1999 21:31:16 -0400
>>From: "Todd A. Farmerie" <ta...@po.cwru.edu>
>>Organization: Case Western Reserve University
>>Sender: use...@rootsweb.com
>>Subject: Re: Walter de Caen and the Dickinson line
>>X-Gateway: news2mail for RootsWeb
>>Resent-Message-ID: <0kX1pB.A....@bl-11.rootsweb.com>
>>To: GEN-MED...@rootsweb.com
>>Resent-From: GEN-MED...@rootsweb.com
>>X-Mailing-List: <GEN-MED...@rootsweb.com> archive/latest/31378
>>X-Loop: GEN-MED...@rootsweb.com
>>Resent-Sender: GEN-MEDIEVA...@rootsweb.com

>>
>>Dora Smith wrote:
>>>
>>> I am interested, but I question this, since Dickin or
>>> Dickon is a very common name in Yorkshire, where the Dickinson's lived.
>>> It seems as if Dickinson would be a common surname. Like Johnson and
>>> Williamson and Richardson and Robertson.
>>

>>Trust your instincts here. There is no reason to assume that any two
>>Dickinsons are related, and to talk of the "Dickinson family" is like
>>talking about the Smith family. All the name means is that someone had
>>an ancestor named Richard who used the common nickname Dick, with the
>>diminutive ending -> Dicken.
>>

>>> My allegation says only that Walter de Caen was the grandson of Walter de
>>> Caen son of Rollo, first Duke of Normandy, who got the Town and Castle of
>>> Caen for an inheritance.
>>

>>No such person can be shown to have existed. Rollo had two, and only
>>two children documented in the historical traditions of normandy -
>>William, later duke, and Gerloc, who married the Duke of Aquitaine.
>>(There is an additional daughter who appears in Icelandic tradition,
>>this information is not independently supported, and perhaps false.)
>>There is no son Walter.
>>
>>> What I would like to know is, more about Walter de Caen. Where did he hold
>>> lands, who were his children and who did they marry, and what happened to
>>> them? Is there any known connection to Yorkshire or to the vicinity of
>>> Hull, where it seems the Dickinsons always lived, atleast by 1260,
>>> century, and any of the family of Walter de Caen? They would have had 200
>>> years, about six generations, from 1066 to find their way to Hull.
>>
>>As to the younger Walter, there may have been a Domesday baron of that
>>name, but it is unlikely that his male-line descendants would have used
>>the surname Dickinson, and anyhow, your source couldn't trace it, or
>>they would have. This whole thing is likely an unsupported tradition,
>>perhaps based on nothing more than Walter holding at Domesday the land
>>that some Dinkinsons would hold 300 years later, but not the stuff of
>>reliable genealogy.
>>
>>
>>taf
>>
>>
>

I don't know whether this is the Walter de Caen under consideration but
Keats-Rohan at I:449 has an entry for a Walter de Cadamo (de Caen). His
father was Alberic. There is no mention of a wife but he had three sons:
Robert (ancestor of the Chesney family), Ralph and Roger (ancestor of the
Huntingfield family). Walter held Horsford, Norfolk, from Robert Malet with
whom he came to England in 1066. Robert's wife was Sibil de Chesney dau. of
Ralph (II) fitz Ralph (I). William, Robert's son, took his mother's name
and was lord of Blythborough, Suffolk.

[K S B Keats-Rohan *Domesday People: A Prosopography of Persons occurring
in English Documents, 1066-1166* (The Boydell Press: Woodbridge, 1999)
I:331, 449; I J Sanders *English Baronies: A study of their Origin and
Descent, 1066-1327* (Oxford University Press: Oxford, 1963 [1960]) 16; CP
V: Chart facing p.72; HKF III:314, 316-317]

[Chris Dickinson]

Re Nathaniel Dickinson

Yes, this descent is bogus. It was invented by an American
genealogist P Wharton Dickinson around 1900 and reissued by
Elinor V Smith in 1978 as a publication of the Dickinson
Family Association.

Some Americans are descended from Nathaniel Dickinson but,
as Clifford L Stott has pointed out, Nathaniel's connection
with those in Yorkshire can't be proved (the descent back to
Yorkshire has been called 'mostly rubbish'); nor did he
have brothers Thomas and John, hence eliminating a number
of American claims of ancestry.

"Clearly, the century-old Dickinson pedigree was at best the
result of poor research - at worst an outright fraud"
Stott]

What seems amazing to me is that anyone could have believed
Wharton's work in the first place. It shouts absurdities. It
makes for very funny reading though .... people keep on
'removing' from county to county. He seems determined to
include every Dickinson he can find (he includes, by the
way, my g-g-g-grandfather, Joseph Dickinson, in his grand
scheme, giving him an entirely incorrect lineage). Not worth
the paper it is written on, I'm afraid.

Chris

[pdic...@gmail.com]

On Thursday, September 16, 1999 at 12:00:00 AM UTC-7, Dora Smith wrote:
> I have an allegation that my Dickinsons (well off merchants of Hull,
> Yorkshire, who intermarried with the De la Pole's) are "generally assumed"
> to be descended from Walter de Caen, a descendant of Rollo the first Duke
> of Normandy (and therefore of the kings of Norway), who came to England
> with his kinsman, William the Conqueror. But the allegation says this

> isn't proven. I am interested, but I question this, since Dickin or

> Dickon is a very common name in Yorkshire, where the Dickinson's lived.
> It seems as if Dickinson would be a common surname. Like Johnson and
> Williamson and Richardson and Robertson.
>

> My allegation says only that Walter de Caen was the grandson of Walter de
> Caen son of Rollo, first Duke of Normandy, who got the Town and Castle of

> Caen for an inheritance. Nathaniel Dickinson's lineage is traced only to
> the 13th century (a few generations before the De la Pole marriages).
> It doesn't even say where he settled in England.
>

> As to the Dickinsons, it says, and proceeds to prove on a generation by
> generation basis, that "There is ample evidence that over the ensuing
> generations the Dickinsons were well to do builders, merchants, landowners
> and men of importance in their communities. They married into well
> established and affluent families. Some of the descendants held highp
> ositions in government and the church.Others were scholars and writers of
> record."
>
> The implication is that this is the sort of family that could be of royal
> blood. But Yorkshire was never fully incorporated into the feudal system
> that was in its most well developed form brought to England by the Normans
> who were able to force it on people closer to London; I don't know
> Yorkshire ever knew serfdom, the ordinary man owned land that had been in
> his family since it was parcelled out by the Danish Viking rulers of the
> area, I think. Ownership of land was the basis of wealth, and became a
> reason why people of Yorkshire helped spearhead the Industrial Revolution.
> Success as a merchant depended on the means to outfit oneself to begin
> with, but it seems as if it was largely a matter of luck, and it was a
> path to upward mobility. Not taht every peasant was likely to be a
> successful merchant, but many successful merchants had common backgrounds.
> There was actually a sizable, thriving class of tradesmen and merchants,
> as well as doctors and lawyers, all through medieval times, and this was
> one of that economic system's fundamental contradictions. Quite a few of
> my ancestors have been established to have belonged to this class in late
> medieval times.
>

> What I would like to know is, more about Walter de Caen. Where did he hold
> lands, who were his children and who did they marry, and what happened to
> them? Is there any known connection to Yorkshire or to the vicinity of
> Hull, where it seems the Dickinsons always lived, atleast by 1260,
> century, and any of the family of Walter de Caen? They would have had 200
> years, about six generations, from 1066 to find their way to Hull.
>

> Yours,
> Dora Smith
> cl...@freenet.buffalo.edu
> dora...@usa.net

As a Dickinson, I have been interested in the De Caen story for awhile. I have been able to trace my line back only to a George Dickinson of St. Kitts (1800-1842). However, I also am intrigued by the 20% Scandinavian (Viking/Norman?) result in my DNA test. Although I do have other English lines on my mother's side (Nunley/Cooper/Whiteman/Cobley et al), I have no direct Scandinavian ancestors in the past 200+ years. So, there seems to be a strong possibility that my Dickinson line has some Scandinavian roots.

[Andrew Lancaster]

On Thursday, April 12, 2018 at 7:26:14 AM UTC+2, pdic...@gmail.com wrote:
> As a Dickinson, I have been interested in the De Caen story for awhile. I have been able to trace my line back only to a George Dickinson of St. Kitts (1800-1842). However, I also am intrigued by the 20% Scandinavian (Viking/Norman?) result in my DNA test. Although I do have other English lines on my mother's side (Nunley/Cooper/Whiteman/Cobley et al), I have no direct Scandinavian ancestors in the past 200+ years. So, there seems to be a strong possibility that my Dickinson line has some Scandinavian roots.

I do not think you can use your autosomal DNA tests that way. Those ethnic components are very rough tools which often lead to strange results, but when working well they are talking about big chunks of ancestry, and so it should be a recent admixture. The male line, indeed, is not special in this type of test and so there is no apparent reason to pick as the probable cause of an unknown ancestral component.

For anyone looking for a Scandinavian in their family tree though, I would suggest double checking anyone in the family tree with a "-son" name, as these often look English but can easily be derived from Dutch or Scandinavian immigrant. Furthermore, because -son names are so common, they are the types of ancestors were people can often be wrongly identified.

[pdic...@gmail.com]

Thanks for the feedback. Perhaps interestingly, I have found no other "son" surnames in my ancestry on either side.

[ch...@dickinson.uk.net]

On Thursday, 12 April 2018 06:26:14 UTC+1, pdic...@gmail.com wrote:

<snip>

> As a Dickinson, I have been interested in the De Caen story for awhile. I have been able to trace my line back only to a George Dickinson of St. Kitts (1800-1842). However, I also am intrigued by the 20% Scandinavian (Viking/Norman?) result in my DNA test. Although I do have other English lines on my mother's side (Nunley/Cooper/Whiteman/Cobley et al), I have no direct Scandinavian ancestors in the past 200+ years. So, there seems to be a strong possibility that my Dickinson line has some Scandinavian roots.

The likelihood of a Scandinavian origin rather depends on where your Dickinson line is from. If you look at a surname distribution map such as the one here:

https://www.your-family-history.com/surname/d/dickinson/

you'll see that the swathe of red covers both the northwest (where Norse settled) and areas that were Danelaw.

Chris

[ch...@dickinson.uk.net]

The Dickinson family that was most associated with the West Indies was of Kingweston in Somerset:

http://somerset-cat.swheritage.org.uk/records/DD/DN

Chris

[wjhonson]

Twenty percent Scandinavian is a very strong result, that cannot in my opinion be supported by a line 200 years old. It has to be more recent.

If you have not also tested several close relatives on *both* sides of your family, or all four lines, or eight lines of ascent, you cannot make a claim to an ascent that ancient.

I suspect that someone much closer in your ancestry is not who they claimm to be

[joe...@gmail.com]

> Twenty percent Scandinavian is a very strong result, that cannot in my opinion be supported by a line 200 years old. It has to be more recent.
>

This is absolutely not good advice. It depends on the testing company, and it is *common* across many of the testing companies that someone 100% British back a dozen generations is reported back as 20% Scandinavian.

[ch...@dickinson.uk.net]

On Sunday, 17 June 2018 23:07:21 UTC+1, joe...@gmail.com wrote:
> > Twenty percent Scandinavian is a very strong result, that cannot in my opinion be supported by a line 200 years old. It has to be more recent.
> >
>
> This is absolutely not good advice. It depends on the testing company, and it is *common* across many of the testing companies that someone 100% British back a dozen generations is reported back as 20% Scandinavian.

I'm not in any way an expert on DNA testing, but simply in logic a 20% Scandinavian ancestry makes sense.

Take my Dickinson family. Until my g-grandfather, the ancestry was purely Cumbrian. See my chart here:

https://static1.squarespace.com/static/50084817c4aa4b810be9dddf/t/509a9ff9e4b0ca218e63cb0a/1352310777148/Dickinson+Simplest+Tree.gif

Had there been a DNA test in the mid-nineteenth century, the Norse DNA would quite likely have been very significant.

Chris

[wjhonson]

Yes you've pointed out exactly the problem with this type of thinking.

The idea that a "typical" Brit is 20% Scandinavian is complete and utter nonsense :)

Without additional close relative testing, a result like this should be raising red flags.

Those people who refuse to see this, will continue along charting their ten generation chart, half of which is completely unrelated to themselves.

Excellent! Keep your head in the sand.

[ch...@dickinson.uk.net]

Have a look at my chart. I am not a typical Brit in that, until 150 years ago, my paternal ancestry was purely Cumbrian (and thus probably partly Norse), so far as can be traced. Many 'Brits' will have similar stories, in different isolated rural counties.

I suspect that you are misunderstanding what results like this mean, if they mean anything. They aren't saying that you are 20% Scandinavian and 80% something else. Danish/Norse is a subset of most English DNA.

Chris

[joe...@gmail.com]

> The idea that a "typical" Brit is 20% Scandinavian is complete and utter nonsense :)
>
> Without additional close relative testing, a result like this should be raising red flags.
>
> Those people who refuse to see this, will continue along charting their ten generation chart, half of which is completely unrelated to themselves.
>
> Excellent! Keep your head in the sand.

It's not that a "Brit" is 20% Scandanavian, whatever it means to be a percentage of a country (in what year?) It's that the commonly available DNA tests will report the ancestry (frequently) of people who have 7 generations back of their family all born in England to be 20% Scandanavian.

Of course, for some of the tests, like ancestry.com, you look further and underneath it will say "possible range 0%-37%" just to give you an idea of the error margins that are involved here.

Why you use these tests so often and don't even have the most basic understanding of how they work is astonishing

[sba...@mindspring.com]

On Monday, June 18, 2018 at 12:29:31 PM UTC-5, wjhonson wrote:

> The idea that a "typical" Brit is 20% Scandinavian is complete and utter nonsense :)

Not really. It is well documented that Britain had a large influx of Scandinavians a thousand or so years ago. What is misguided is the idea that a specific percentage of "Scandinavian" blood can be determined from a DNA test, or even that the term "Scandinavian" even has any meaning at the genetic level.

> Without additional close relative testing, a result like this should be raising red flags.

There are no red flags here. The problem comes from the belief that genetic tests can reliably distinguish between nationalities which are, on average, more closely related than the typical "noise level" of the statistical calculations. As an example, my two siblings and I (shown to be full siblings by our autosomal DNA tests) have very different (alleged) percentages in our European nationalities, as reported by these tests.

For distantly related larger groupings (say European vs. Native American), such percentages will tend to be closer to the mark.

Stewart Baldwin

[taf]

On Monday, June 18, 2018 at 3:40:47 PM UTC-7, sba...@mindspring.com wrote:

> For distantly related larger groupings (say European vs. Native American),
> such percentages will tend to be closer to the mark.

This is about the level that can be distinguished. It is pretty good at looking at a Caribbean Islander or Appalachian Tri-Racial Isolate and determining the proportions of African, European and Native American (which cannot be distinguished from East Asian), but all the nationality-level results that are reported are rubbish.

It fails primarily because of two factors. One is that most populations from the same continent just aren't that different, and likewise, with very few exceptions, nationalities are modern geopolitical constructs, not genetic ones - almost all nationalities, including those that we think of as genetically uniform (such as the Japanese), are actually complex and variable mixtures of ancient strains.

There are people from northern Britain whose families have been there for a milenium who are 20% Viking. There are also British people who have one grandfather from Stockholm who are 20% Viking. At the level of these ethnicity tests, DNA can't tell the difference, and it is muddled all the more by the fact that their statistical conception of the average genetic Brit includes a certain amount of Viking DNA. Add to this that the Vikings themselves were a mixed people, almost all of the genetic markers they carried also being found in other Germanic populations, and you have a real mess. You are using a computer to throw a dart at three-dimensional intersecting Venn diagrams.

The take-home is that these ethnicity tests are vastly overinterpreted and are not genealogically useful on an intra-continental scale. My own uncle's test seems to have reported his odd mixture of inbred-German, French and Irish as if it were part Scandinavian, of which there is none in the paper-trail genealogy. I do not find this curiosity to be a reason to take a hacksaw to my family tree.

taf

[duanej...@gmail.com]

I had my DNA tested through Ancestry a few years ago. My pie chart included 36% Irish, 31% Europe West and 26% Scandinavian. A few weeks ago Ancestry updated my pie chart. It is now 52% England and Wales and 48% Ireland and Scotland. My Scandinavian DNA is now down to zero. My updated England and Wales changes include Benelux Union and small parts of Germany, France and Ireland. My updated Ireland and Scotland changes include most of Wales and small parts of Scotland and England. I don't know why my DNA pie chart changed so much. I have several ancestral lines going back to the North of England. They were all associated with the Whitfields, co: Northumberland hundreds of years ago. I don't know how far back Ancestry DNA goes. Maybe Ancestry has tested a million more people in the past few years and are able to achieve greater accuracy?
Duane Jones

[joe...@gmail.com]

On Monday, June 18, 2018 at 9:00:10 PM UTC-4, duanej...@gmail.com wrote:
> I had my DNA tested through Ancestry a few years ago. My pie chart included 36% Irish, 31% Europe West and 26% Scandinavian. A few weeks ago Ancestry updated my pie chart. It is now 52% England and Wales and 48% Ireland and Scotland. My Scandinavian DNA is now down to zero. My updated England and Wales changes include Benelux Union and small parts of Germany, France and Ireland. My updated Ireland and Scotland changes include most of Wales and small parts of Scotland and England. I don't know why my DNA pie chart changed so much. I have several ancestral lines going back to the North of England. They were all associated with the Whitfields, co: Northumberland hundreds of years ago. I don't know how far back Ancestry DNA goes. Maybe Ancestry has tested a million more people in the past few years and are able to achieve greater accuracy?

This Is exactly it; yes.
The more people tested the more 'accurate' these get, although they still have very large uncertainty bars
They (ancestry.com) are aiming to identify country of residence in the 300-400 years ago or so time frame really. Other tests, such as National Geographic, aim quite a bit further back based on their training database.

--JC

[wjhonson]

I disagree with your characterization that such a result is "frequent"
That is the basis of my claim. Not that it doesn't occur, but that *when* it occurs, it should push people to have other close relatives tested.

I really don't know why you are arguing that people *shouldn't* be pushed to test other relatives. That really is the only way forward with some brick walls and odd results.

Two hundred years ago, is not one thousand years ago also.
The only solution is to test more relatives.

[sba...@mindspring.com]

On Monday, June 18, 2018 at 6:44:39 PM UTC-5, taf wrote:

>. . . but all the nationality-level results that are reported are rubbish.

I think that some qualification is needed for this statement to keep it from being misunderstood. For example, my own reported percentages are pretty close to what my "paper trail" ancestry would indicate, and I have reasonably good (but not ideal) evidence that all eight of my genetic great-grandparents and at least 13 of my genetic great-great-grandparents are the same as my paper trail ancestors. However, my two siblings both have results which are very different from the "expected" result (and also from each other).

Because these estimates are based on statistical averages, it is to be expected that a significant percentage of testees will have a result that is reasonably close to the "real" result (whatever that means when the definitions themselves are ambiguous). The problem is that the expected error rate is so large that a large percentage of testees will have reported results which are way off, and most importantly, there is no good way of telling the reasonable results from the bad ones without tracing a well-documented paper trail ancestry for the testee.

Stewart Baldwin

[taf]

On Tuesday, June 19, 2018 at 3:04:03 PM UTC-7, sba...@mindspring.com wrote:
> On Monday, June 18, 2018 at 6:44:39 PM UTC-5, taf wrote:
>
> >. . . but all the nationality-level results that are reported are rubbish.
>
> I think that some qualification is needed for this statement to keep it from
> being misunderstood.

<snip>

> The problem is that the expected error rate is so large that a large
> percentage of testees will have reported results which are way off, and
> most importantly, there is no good way of telling the reasonable results
> from the bad ones without tracing a well-documented paper trail ancestry
> for the testee.

Yes, exactly what I meant. If the rate of reporting inaccurate results is high enough that you can't be reasonably sure any given result is accurate, then no result is trustworthy.

taf

[Andrew Lancaster]

I think it is also important to emphasize, for those less familiar, the comment you made about the type of test (ie the genealogical testing companies) and also the comment Joe made about the differences between testing companies, which can be very big. Thirdly the point has been made about the fact that there is an important difference between things which are statistically very clear, but not clear for an individual. (The average man, is not a man at all.)

My reason for emphasizing these 3 points, which were already made, is that people who look at claims in some types of academic publications might sometimes see strong conclusions. These are not always badly over-stated (although certainly many population genetics articles do "oversell" a bit). The problems we can have as individual genealogists are not always relevant to some of the amazing progress being made in the study of populations.

[sba...@mindspring.com]

Another big difference is that the most detailed information in the study of populations has come from Y-DNA and mt-DNA, which are easier to trace because they pass from one generation to another largely intact, but have the disadvantage that they only represent one specific line of descent among many. The "nationality" information sold by the testing companies comes from autosomal DNA, from which it is much more difficult to get information about the movements of populations. The fact that the human migrations patterns suggested by Y-DNA differ significantly from those suggested by mt-DNA shows that the big picture is much more complicated than the scenarios suggested by those tests. Also, with chronological error rates still measured in the thousands of years for the early migrations (and quite a few of the later ones), there is still a lot that is vague.

In addition to the different results from different testing companies (and the different results within one company as they revise their statistical models), anyone who uploads their results to www.gedmatch.com can try out a variety of other algorithms, often with very different results. As the amount of available raw data increases, I would expect results to "improve" somewhat, but only in the sense of being "less unreliable" than before. A dependable result would require not only a large enough data set, but also that one not only inherited 50% of their DNA from each parent, but also 1/4 from each grandparent, 1/8 from each great-grandparent, and so forth for many generations, which is simply not the case. So, there are certain statistical limitations which no amount of improved data will ever be able to overcome. (I am ruling out scenarios like digging up graves and testing the remains on a massive scale, and even that would have its limitations.)

Thinking about this stuff made me wonder about the following thought-experiment. Suppose that you created a collection of similar fake DNA raw data files by randomly combining various parts of some genuine ones from several different ethnicities, and then labeled this data with a fake ethnicity (say "Klingon") to be considered with all of the "genuine" ethnicities. Would a significant number of individuals show up as part Klingon?

Stewart Baldwin

[sba...@mindspring.com]

On Tuesday, June 19, 2018 at 5:48:46 PM UTC-5, taf wrote:
> . . . If the rate of reporting inaccurate results is high enough that you can't be reasonably sure any given result is accurate, then no result is trustworthy.

This comment reminded me of the typical undocumented Internet genealogy.

Stewart Baldwin

[Andrew Lancaster]

On Wednesday, June 20, 2018 at 4:48:06 PM UTC+2, sba...@mindspring.com wrote:
> On Wednesday, June 20, 2018 at 1:55:04 AM UTC-5, Andrew Lancaster wrote:
> > On Wednesday, June 20, 2018 at 12:48:46 AM UTC+2, taf wrote:
> > > On Tuesday, June 19, 2018 at 3:04:03 PM UTC-7, sba...@mindspring.com wrote:
> > > > On Monday, June 18, 2018 at 6:44:39 PM UTC-5, taf wrote:
> > > >
> > > > >. . . but all the nationality-level results that are reported are rubbish.
> > > >
> > > > I think that some qualification is needed for this statement to keep it from
> > > > being misunderstood.
> > >
> > > <snip>
> > >
> > > > The problem is that the expected error rate is so large that a large
> > > > percentage of testees will have reported results which are way off, and
> > > > most importantly, there is no good way of telling the reasonable results
> > > > from the bad ones without tracing a well-documented paper trail ancestry
> > > > for the testee.
> > >
> > > Yes, exactly what I meant. If the rate of reporting inaccurate results is high enough that you can't be reasonably sure any given result is accurate, then no result is trustworthy.
> > >
> > > taf
> >
> > I think it is also important to emphasize, for those less familiar, the comment you made about the type of test (ie the genealogical testing companies) and also the comment Joe made about the differences between testing companies, which can be very big. Thirdly the point has been made about the fact that there is an important difference between things which are statistically very clear, but not clear for an individual. (The average man, is not a man at all.)
> >
> > My reason for emphasizing these 3 points, which were already made, is that people who look at claims in some types of academic publications might sometimes see strong conclusions. These are not always badly over-stated (although certainly many population genetics articles do "oversell" a bit). The problems we can have as individual genealogists are not always relevant to some of the amazing progress being made in the study of populations.
>
> Another big difference is that the most detailed information in the study of populations has come from Y-DNA and mt-DNA, which are easier to trace because they pass from one generation to another largely intact, but have the disadvantage that they only represent one specific line of descent among many. The "nationality" information sold by the testing companies comes from autosomal DNA, from which it is much more difficult to get information about the movements of populations. The fact that the human migrations patterns suggested by Y-DNA differ significantly from those suggested by mt-DNA shows that the big picture is much more complicated than the scenarios suggested by those tests. Also, with chronological error rates still measured in the thousands of years for the early migrations (and quite a few of the later ones), there is still a lot that is vague.

Stewart, possibly getting too far from the topic, but I think your first paragraph is a bit debatable. That Y and mitochondrial DNA show distinctive patterns is logical, but also reminds us that they are not really good indicators of what happened to whole populations. Y lines for example have a rapid turnover, and move quickly around geographical areas. It is always very hard to say where a haplogroup really comes from, but normally it will NOT be where it is most common. Mitochondrial DNA haplogroups are spread out over large areas, and also very difficult to use.

Secondly the problem which we have pointed to for individual genealogists does not apply to population geneticists who really are interested in the " average person" and not individuals. The progress made in defining ancient population movements using autosomal DNA in recent years has eclipsed the older generation of studies which tried to use Y and mito DNA, and they've often created consistent and meaningful conclusions (something which the old studies always struggled to do). I am not a geneticist but reading the literature there seems to be widespread consensus on this.

[sba...@mindspring.com]

On Thursday, June 21, 2018 at 4:19:03 AM UTC-5, Andrew Lancaster wrote:
> On Wednesday, June 20, 2018 at 4:48:06 PM UTC+2, sba...@mindspring.com wrote:

...

> > Another big difference is that the most detailed information in the study of populations has come from Y-DNA and mt-DNA, which are easier to trace because they pass from one generation to another largely intact, but have the disadvantage that they only represent one specific line of descent among many. The "nationality" information sold by the testing companies comes from autosomal DNA, from which it is much more difficult to get information about the movements of populations. The fact that the human migrations patterns suggested by Y-DNA differ significantly from those suggested by mt-DNA shows that the big picture is much more complicated than the scenarios suggested by those tests. Also, with chronological error rates still measured in the thousands of years for the early migrations (and quite a few of the later ones), there is still a lot that is vague.
>
> Stewart, possibly getting too far from the topic, but I think your first paragraph is a bit debatable. That Y and mitochondrial DNA show distinctive patterns is logical, but also reminds us that they are not really good indicators of what happened to whole populations. Y lines for example have a rapid turnover, and move quickly around geographical areas. It is always very hard to say where a haplogroup really comes from, but normally it will NOT be where it is most common. Mitochondrial DNA haplogroups are spread out over large areas, and also very difficult to use.
>
> Secondly the problem which we have pointed to for individual genealogists does not apply to population geneticists who really are interested in the " average person" and not individuals. The progress made in defining ancient population movements using autosomal DNA in recent years has eclipsed the older generation of studies which tried to use Y and mito DNA, and they've often created consistent and meaningful conclusions (something which the old studies always struggled to do). I am not a geneticist but reading the literature there seems to be widespread consensus on this.

I can see that my use of the word "populations" was a poor choice on my part, making it easy to misinterpret the point I was trying to make. Both Y-DNA and mt-DNA are inherited in a manner which allows a relatively well-defined "branching order" among the ancestors of those tested. Thus, even without "ancient DNA" Y-DNA and mt-DNA tests give a reasonable picture of the distribution of Y-DNA (or mt-DNA) among the ancestors of the tested populations (leaving aside for the moment the jumps in logic required to geographically locate these ancestors). In contrast, a particular segment of an individual's at-DNA has not necessarily been inherited from a single ancestor, but possibly (due to crossovers) from many different ancestors spread randomly across the globe. Thus, any analysis of the at-DNA present in ancestral populations has to be much more statistical in nature, with a larger expected error. Autosomal DNA does have the advantage that it is more representative of the ancestry of an individual (rather than concentrating on two specific lines), but properly exploiting this requires huge amounts of raw data from both modern testees and ancient DNA.

Stewart Baldwin

[ch...@dickinson.uk.net]

Well, yes, the use of the term 'population' is always confusing in this context - as it has separate meanings in statistics and data base software beyond the racial one.

I'm wondering whether your Y-DNA argument is too international. For the UK, and I expect for many other European countries, there is still a fairly stable base for local genealogy of the male line, both the paper trail and the DNA one.

Chris

[Andrew Lancaster]

Yes, for sure Y DNA and mitochondrial DNA both give specific types of insight because of their nature. I agree. As an example of how Y DNA has weak points, but also how they can be handled and then compared to the results from other types of DNA, there was a discussion over some years concerning the history of the Y DNA R1b haplogroup.

R1b, as many on this forum will know, is the most common old Y haplogroup in Iberia, Britain and Ireland, and roughly gets more common as you go west, to the extremities of Europe. Initially this led to ideas that it was a pre-Celtic male line that crept back north after the ice-ages. Of course there were many people ready to say this it proved the old stories true of an immigration history from Spain to Ireland. None of these things are believed to be true any more.

A first reason for caution was the discovery that other pockets of R1b are found in central Asia and even Chad. Furthermore the closest relatives of R1b are mostly central Eurasian and Middle Eastern. Scientists started to remember that in biology new haplotypes often spread out from a starting point in a wave, and can be common on the "front". Just as Y DNA studies were reaching their peak, and people were getting more careful and sophisticated, a major paper proposed that R1b had spread from the east, probably around the time proposed for the spread of indoeuropean languages and bronze technology.

This now seems to match the conclusions of autosomal studies. In this case, but not all cases, a male line was particularly strongly promoted by a population movement and this is interesting in itself. One has to doubt, for example, that it was a matriarchal people.

[Andrew Lancaster]

Chris, can you explain what you mean?

[ch...@dickinson.uk.net]

I assume you refer to my second sentence. Sorry that I was unclear - very badly written!

I was responding really to your argument that "Y lines for example have a rapid turnover, and move quickly around geographical areas". That may be the case from, say, a US perspective, but there are probably still European areas where the Y-lines have stayed put, even if brothers have moved elsewhere, and there hasn't been much migrant inflow, so the population pool is fairly homogeneous. For instance, my grandfather's DNA background would have been entirely 'Cumbrian' (an area with probably not much internal migration for centuries), with a paper trail to 'prove' it for 400 years, unlike mine.

[Andrew Lancaster]

OK thanks for the clarification. It means I also was not clear!

When I spoke of a rapid turnover I was speaking on a large and long-term scale. What I mean is that is over "only" a few centuries the most common Y haplogroups in a population can change surprisingly much. So this is again one of those "average man" statements that applies to no actual man in particular. But it certainly was something intended to apply in the "old world".

What seems to happen historically (and in fact it is harder to know if it is continuing to happen in new populations like the US) is that a small number of Y lines become quite dominant rather quickly, which means that they also don't last long. We can also compare this to surnames, which of course only exist a relatively short time. But even in that period, many have disappeared and some have become very common.

[sba...@mindspring.com]

On Saturday, June 23, 2018 at 4:35:40 PM UTC-5, Andrew Lancaster wrote:

> When I spoke of a rapid turnover I was speaking on a large and long-term scale. What I mean is that is over "only" a few centuries the most common Y haplogroups in a population can change surprisingly much. So this is again one of those "average man" statements that applies to no actual man in particular. But it certainly was something intended to apply in the "old world".
>
> What seems to happen historically (and in fact it is harder to know if it is continuing to happen in new populations like the US) is that a small number of Y lines become quite dominant rather quickly, which means that they also don't last long. We can also compare this to surnames, which of course only exist a relatively short time. But even in that period, many have disappeared and some have become very common.

The statement that "they also don't last long" can be misleading, because as a Y-line becomes dominant, it splits into many branches over time, and a member of one of those branches may become the ancestor of a future rapid expansion. As an example, Y-DNA studies indicate that in the distant past, all men have a common Y-DNA ancestor. If we call the most recent such ancestor "Adam-Y" for convenience (his exact time frame is a matter of debate, but is unimportant for the present discussion), then there would have been many other male humans living at the time who left modern descendants, but always with at least one intervening female generation. At some later (but unknown) point in time, perhaps much later, the last male who was a direct male-line descendant of one of Adam-Y's contemporaries died, and Adam-Y became the most recent Y-ancestor of all of mankind. So his Y-line has lasted a long time, and will continue for as long as the human species exists and reproduces in the usual way. (Various test-tube and other artificial "advances" require a qualification on this point.)

At the time that Adam-Y lived, there would have been a most recent common Y-ancestor of all men then living, who would not have been Adam-Y himself, but one of his Y-ancestors. So, the person who has the status currently held by Adam-Y can change over time. How often the holder of the "Adam-Y" title has changed would be impossible to tell without a completely reliable genetic genealogy of all individuals in the world. For all we know, the last direct male-line descendant of one of our Adam-Y's contemporaries might have died yesterday (but probably not). If such a hypothetical individual's Y-DNA never got tested, we would have no way of knowing.

The same is true for mitochondrial DNA and the common ancestor we might call "Eve-mt." The big difference here is that such rapid expansion is much less likely, because biological limitations mean that women seldom have more than fifteen or so children, while men are theoretically able to sire hundreds of offspring during their lives. Of course, few men fall into that category, but the ones who do can be responsible for big surges in a particular Y-DNA signature. Still, I wonder how many large Y-DNA groups came from patrilineal ruling dynasties which maintained large harems over several generations.

However, even where most individuals are monogamous, some will have mostly sons and others will have mostly daughters. Over several generations, the random factors can result in surprisingly large shift in the distribution of family names, resulting in the disappearance of some surnames. The same kind of drift can happen with mt-DNA, but the generation-to-generation change in surnames in female lines makes this much harder to notice. If a man has left numerous modern descendants, but none of them are in the direct male line, genealogists will often say that his line has "daughtered out" (and the same with surnames). I suppose that people trying to trace mitochondrial lines could complain that a line "sonned out" but I have never heard that term. (In fact, the spell checker complained about "sonned" but accepted "daughtered" without any red underline.)

Stewart Baldwin