On 13/10/2019 16:05, J. Hugh Sullivan wrote:
On Sun, 13 Oct 2019 12:39:02 +0100, Richard Smith wrote:
We don't match any other tested Sullivans.
When you say you don't match, is that at a haplogroup level (e.g. you
are something unusual like T and all other Sullivans are R), or at a
marker level within a haplogroup? If the latter, how many markers are
you away from the nearest Sullivans, and how many markers have you had
tested?
I tested 67 markers. I am R1a1a1a, L-664. Other Sullivans (except one
cousin) are R1b (Irish).
And is this known Sullivan cousin a perfect 67 out of 67 match? I'm
assuming the answer is yes.
I am almost certainly an Anglo-Saxon from
Germany with a female Irish ancestor by following L-664 from the Black
Sea.
It's a clue, but not one to pay too much attention to because you don't
know when this happened. It could be that a Viking raider visited
Ireland in the Early Middle Ages and fathered a child with a married
woman in the O'Sullivan clan.
Our DNA is also an exact match with a Wyatt and there appears to be no
way I can link to him.
Did you discover this Wyatt match by searching some DNA database for
likely matches, or did you have some other reason to investigate this
particular surname or person?
Family Tree DNA notified me - exact match at 67 steps.
Okay.
I did his genealogy and they were in MA, never in NC.
North Carolina and Massachusetts are certainly some way apart. How far
back did you trace the line in Massachusetts?
Do you have any reason to believe either your male-line ancestors or his
were likely to have travelled long distances as work? For example, was
anyone a mariner, or a worker on the railways? The purpose of this
question is to see how likely a more recent relationship is.
If this Wyatt testee is still contactable, he might be willing to do an autosomal DNA test. If you and he are third cousins or closer, there's
a 90% chance or better of this being detected by modern autosomal DNA
tests. Maybe you've already done this.
So we apparently have the same ancestor but probably in England where
there were two Wyatt families that have never been linked.
The Sullivan clan is from Munster (i.e. south-west Ireland), while Wyatt
is primarily found all over England, and may well have been arisen more
than once. I've research quite a few Wyatt lines in England, and there
are certainly a lot more than two families which no-one has managed to
link. There does seem to be a tendency to assume that people emigrating
to America must have come from one of the famous, well documented lines,
like that of Henry Wyatt of Allington. Doubtless some of emigrants were
from such families, but a lot more were from obscure English families
that cannot be traced back beyond the 17th century.
You've traced your line in America to the late 18th century, and have a possible line back at a few more generations, so you're probably talking
about an ancestor settling in America in the late 17th or early 18th
century. There were very few Irish in England this far back, so if a
man of English ancestry had a child with a woman of Irish ancestry, my
gut instinct is that this more likely happened in America. I know
little about levels of early Irish immigration to America, but the fact
that you had nine possible fathers for Russell suggests to me there were
a lot more Irish in America than England at this time.
This doesn't necessarily mean the man who fathered a child with a
Sullivan woman was the ancestor of the Wyatt testee. Possibly the MRCA
was a little further back, and could be in England as you say.
A MRCA indicates a 95% possibility of a match with Wyatt at 6
generations.
I'd like to know more about this. First, is this based on Y-DNA or
autosomal DNA? I.e. is it saying there's a 95% probability that there
is a common make-line ancestor within six generations?
Y-DNA.
Okay. Let's step back and remind ourselves how DNA works. Each time a
man has a son, there is a small chance that the DNA will mutate in a way
that is detectable in a 67-marker Y-DNA test. With each passing
generation, the overall chance that a mutation has occurred increases.
This is like throwing a pair of dice. It's fairly unlikely you'll get a
double one if you throw once, but if you spend an evening playing craps,
it will eventually happen. And exactly as with throwing a dice, what
happens in one generation has no effect on what happens in the next
generation – a long run with no mutations does not mean a mutation
becomes any more likely in the next generation.
Having looked up the statistics for 67-marker tests, I think what they
are saying is that in ten generation there is a 95% chance a detectable mutation will have occurred. It is ten because a mutation on the
descent from the MCRA to you would result in a difference, but so would
a mutation on the descent from the MCRA to the other testee. And it's
ten rather than twelve because there are two generations between you and
your grandfather – i.e. it's the number of father-son relationships, not
the number of people in the sequence.
However the situation here is more complicated because there are three
people involved: you, your known Sullivan cousin, and the Wyatt testee.
Assuming you have a 67/67 match with your Sullivan cousin, this tells
us that no mutation occurred between you and Russell Sullivan. This
means you know what DNA Russell had at these 67 points because it was
the same as yours. So when you compare yourself with this Wyatt testee,
you're really comparing Russell Sullivan with the Wyatt testee, and you
should be counting the generations between these two. To return to the
dice analogy, imagine you throw the dice a few times as a practice and
then starting betting on them. The fact you avoided getting a one in
the practice throws does not make it any more likely that they will come
up later. So it is with DNA. The fact that we know no mutation
occurred between you and Russell does not make a mutation any more
likely in the preceding generations, so we have to reset counting
generations for the 95% chance at Russell.
At this point it would helpful to know how many generations you have to
go back on the Wyatt line to reach someone of roughly the same
generation as Russell. I think most people alive today will have great
great grandparents who were born significantly more recently than 1789,
so I'm guessing this Wyatt testee might have an extra generation or two
back to Russell's time. Let's estimate six, excluding himself, i.e. one
more than you. The statistics tell us there is a 95% chance that
Russell and this Wyatt testee are related within ten generations. If we
need around six to get back to a contemporary of Russell, this leaves
four. So the statistics tell us there is a 95% chance that the Wyatt
tester's ancestor was a first cousin or closer of Russell. Allowing for
a bit of uncertainty over what the testing company mean by "related
within six generations", this could be a second cousin rather than a
first cousin.
It probably seems perverse that by gathering more information the
relationship becomes less certain, but the field of statistics is full
of this sort of counter-intuitive result and it is one of the things
that make DNA results very hard to interpret. You know that you and the
Wyatt tester share a common male-line ancestor, and therefore in at
least one of your lines, the surname did not pass down the biological
male line. You want to find out when this was. The 67/67 match between
you and the Wyatt tester says it was relatively recently, and almost
certainly not, for example, in Early Middle Ages. By discovering a
67/67 match between you and this Sullivan cousin, you eliminate quite a
few possibilities and therefore the less likely options become more
likely, pushing back the most likely date for the MRCA.
If the Wyatt tester had a distant male-line cousin who could be
persuaded to take a Y-DNA test, and he is found to be a 67/67 match too,
this would eliminate another swathe of possibilities involving
non-paternity in the documented Wyatt line, and so the most likely time
for the MRCA would be pushed back further. It is as Sherlock Holmes
said: "Once you eliminate the impossible, whatever remains, no matter
how improbable, must be the truth."
At the moment, I would suggest the DNA evidence suggests the MRCA was
most likely born in the early 18th century.
Have you found anyone who is a 66/67 or even a 65/67 match that might
help pin down the lineage from the other direction? (Of course, a 66/67
match can be a closer relative than a 67/67 match, depending where the
mutation occurred.)
Do you have a feel for how many Sullivans there were in the county, or
even the state, at the time?
I have a reconstructed 1740 VA census and I have tracked the recorded Sullivans. There were 2 Sullivan families in Amelia Co. VA at the time
- one from York Co. - female and I know her genealogy. Also a John
Sullivan was on the 1740 Tax Rolls. That's why I presume the three
Sullivan females were his daughters. The LDS Church records a marriage
for him.
Is your theory that Charles, the possible grandfather of Russell, is the illegitimate son of one of John's daughters? If Charles's father is an ancestor of the Wyatt family in Massachusetts, this would be consistent
with the DNA evidence as I understand it. However, pushing the MRCA
further back is starting to stretch the interpretation of the DNA evidence.
You say there's a Wyatt in Amelia Co. at the right time. Have you tried tracing his known children? Is there any possibility he may have had a
son or grandson who moved to Massachusetts was ancestor that family?
On a related note, is there any evidence documenting the family's move
from Virginia to North Carolina? Or is it simply the case that a man disappeared from the record in Virginia at about the same time as a man
with the same name and age appeared in North Carolina? There's nothing
wrong if that is the case, but it is more circumstantial and raises more question marks in an already shaky lineage.
Richard
--- SoupGate-Win32 v1.05
* Origin: fsxNet Usenet Gateway (21:1/5)