GENEALOGY-DNA-L Archives
Archiver > GENEALOGY-DNA > 2006-01 > 1136851266
From: "Eli d S" <>
Subject: Re: [DNA] Out-of-surname matching (was: Assyrian Heritage DNA Project)
Date: Mon, 09 Jan 2006 19:01:06 -0500
In-Reply-To: <REME20060109182018@alum.mit.edu>
An excellent explanation, thank you.
case 1 is what would apply here.
>Case 1. 24/25 match, different surname, ancestry from Spain in one
and Germany in the other. Ancestry traced to 1100 in one and
1725 in the other.
>I would assert that this match is indeed a wild-goose chase. The
>German line is unlikely to be traced back before 1500, let alone
>1100, and the 95% confidence interval on TMRCA extends all the way
>back to 500. In this case, the pessimistic scenario is the
>instructive one.
For instance in Sorenson you can see how I have 34/36 match with a man by
the last name of Brazil they trace their genealogy to the early 1800's in
the US http://tinyurl.com/dfv9f searching the same haplotype in Ysearch
gives me other matches I spoke of earlier. Clearly attempting to make the
genealogical connection would be a waste of time, however, I had assumed
that a relatively recent common ancestor must have existed, one because
34/36 seems pretty close to me and two their site points a likely ancestor
13 generations ago. Although the Ysearch folks do not appear in Sorenson I
thought the same principles should apply, instead of being two mismatches at
13 generations (400ya), being 4 mismatches at 26 generations (800ya) as it
is portrayed in Sorenson even though we don't share the surname.
If I understand you correctly, what you are saying is that in this case a
relationship may exist although it is improbable, and it would be almost
impossible to track it down for genealogical purposes.
Now Im just thinking out loud. Not certain what to make out of this, if the
certainty of a kinship is so improbable despite a 34/36 match what makes it
more likely that somebody that I share the surname with in reality is
related to me? Even if the match is 25/25 because then we have 31/36....
only the fact that we share the surname, but we knew that already.... any
of those first ten matches would've fallen in that category, and had they
shared the surname (for whatever reason) we would've had the certainty of
kinship.
Eli
From: (John Chandler)
Reply-To:
To:
Subject: Re: [DNA] Out-of-surname matching (was: Assyrian Heritage DNA
Project)
Date: Mon, 9 Jan 2006 18:20:20 -0500 (EST)
Eli wrote:
> This confuses me, if most of the out-of-surname matches don't indicate a
> relationship, wouldn't that be an indication that the whole thing is
> pointless?
What you have to consider is the *very* wide probability distribution
that comes from the slow (in human terms) mutation rate of STRs. The
net result is that everyone can and should follow both an optimistic
and a pessimistic scenario at the same time for any given comparison.
Almost any reasonably close match can (and should) be viewed as both
"close kin" and "not related" as far as the DNA testing is concerned.
I admit this sounds confusing, or at least paradoxical, but this type
of flexibility is needed to take into account whatever other info you
may have.
The other thing to keep in mind is that this is the GENEALOGY-DNA
list. Most people with western European ancestry can lay claim to
descent from Charlemagne, among others, but we don't include him in
our pedigrees, as a rule, because we don't know where to put him.
Perhaps you actually know of a link in your own case, but that's rare,
and you can rest assured that there are other links you don't know
about. This mathematical near-certainty of a link to Charlemagne is
*not* genealogy. The same principle applies to DNA testing -- it's
not genealogy until and unless the links can be laid out with names
and places.
Case 1. 24/25 match, different surname, ancestry from Spain in one
and Germany in the other. Ancestry traced to 1100 in one and
1725 in the other.
I would assert that this match is indeed a wild-goose chase. The
German line is unlikely to be traced back before 1500, let alone
1100, and the 95% confidence interval on TMRCA extends all the way
back to 500. In this case, the pessimistic scenario is the
instructive one.
Case 2. 22/25 match, known MRCA born 1725.
No problem. This is not the most likely genetic distance for a
connection this close, but it falls near the edge of the 95%
confidence interval. The genealogy is in place already, and so the
optimistic scenario is the instructive one.
> For instance I know I descend from a Visigoth in Iberia, and I find close
> matches with people in Germany, Ireland and England. Up until now Ive
> assumed that those 3-5 mismatches would fit nicely within the time frame
> allowed of 1,400 - 3,000 years for a common ancestor. Yet here I see
people
> that share a gg grandfather and have 6 mismatches, I have been saying it
is
> an average so some will take three times longer some will do it in a
couple
> hundred years. But you saying out-of-surname matches are worthless really
> throws me out for a loop.
If you want to "allow" 1400-3000 years for a common ancestor, you have
stepped outside of genealogy into some other realm. The rules are
obviously different there. In fact, the rules are whatever you want
them to be, but I don't see the point in it myself. However, there's
always...
Case 3. 24/25 match, different surname, but both lines lived in the
same locality between 1700 and 1850, and there were a number of
recorded intermarriages along the way.
This case is the starting point for a further investigation. Both
parties need to expand their sights to include a variety of other
descendants of their earliest known ancestors. If one or the other
finds that a different haplotype is common among his supposed kin, he
would conclude that his paternal line very likely crosses over into
the other family, and extensive testing can perhaps pin down when the
cross-over occurred. Out-of-surname matches are not all worthless.
Case 4. (This one isn't made up.) 21/21 match discovered in the SMGF
database before surname searching was implemented. The input
haplotype was that of Edmund Rice (c1594-1663), reconstructed
from testing his descendants. The (three) matches were all
named King, and their submitted pedigrees showed them to be
a father and two sons. Further investigation showed them to
be descendants of a Samuel Rice who was adopted as an infant
by his deceased mother's brother and wife and raised as a
King.
Again, out-of-surname matches are not all worthless.
John Chandler
This thread: