Archiver > GENEALOGY-DNA > 2012-02 > 1328899191

From: Paul Wright <>
Subject: Re: [DNA] autosomal DNA
Date: Fri, 10 Feb 2012 12:39:51 -0600
References: <><>
In-Reply-To: <>


Is you project a collection of documented relatives at the distance noted
being compared, or a group of individuals with segment shares for which you
are identifying or attempting to identify a paper trail connection?
Depending on which perspective you are looking at could create a
completely different response to the question.

If the former, would be interested to know what you are finding for segment
shares over 7 cm for relationships beyond 4th cousin. How many of such
comparisons have you looked at? Unless the samples in question come from
populations with a high degree of pedigree collapse I would expect
relatively few to have segment matches for these more distant
relationships (with 3rd stated as about a 90% chance of having segment
share and 4th around 45%, 3rd once remove would still be more likely to
have one than not; meanwhile 5th @ approx 15% and 6th and beyond at less
than 5%, or less than 2%, depending on the source). If you are accepting
smaller segment shares in the analysis, what certainty do you have that
they originated with the ancestors you think they may have originated with?

I know David Faux has shared data on his 30+ relative project which
appeared to have higher than expected number of shares for some of these
more distant relatives, but also had a number of intermarriages and
multi-cousin situations which might explain the results.

Meanwhile the limitations viewed moving beyond 3rd cousins is more a matter
of biology than technology and no matter how much you improve the
technology, it will not matter if there is nothing there to see. Expanded
databases will increase the likelihood of more recent relatives appearing
in it and larger number (not necessarily percentage) with who you can find
a documented connection. When you get out to these more distant relations
you have a higher likelihood of passing along a segment intact or not at
all rather than passing along a smaller segment. Male vs Female
recombination rate differences are also meaningful as you move through the
transition between 3rd to 5th cousins and beyond as the higher female rates
could lead to higher likelihood of retaining a segment share for an
additional generation or two for persons separated by mostly female events
compared to mostly male events, but will not impact the majority that would
be an close to even mix of male and female meiotic events. May also be
impacted by where in the chain of meiotic events the female event takes
place. Beyond the transitional generations, the net effect is probably


On Fri, Feb 10, 2012 at 12:05 PM, Sam Eaton <> wrote:

> Bill,
> I'm a little cautious about self described experts. As an example, I have
> been in and out of Viet Nam for 42 years, have 21 semester hours credit in
> the language, have been married to a Vietnamese woman for 38 years, live
> here, have sold magazine articles on the subject and have been quoted and
> acknowledged in books on the subject.
> Does this make me an expert? No! But, I am generally less misinformed on
> the subject than most of the self proclaimed experts that I've read or run
> into.
> Now, I am no expert. Still, I have been studying autosomal DNA since Nov
> 2009 and administer atDNA projects at FTDNA and GEDmatch that reach back to
> people born in the 1760's and 1770's. For me this ranges from 3rd cousin
> once removed to 5th cousin. While my project is definitely working and
> providing excellent conformation of existing research, good tools are hard
> to find. has the best tools. Note, these tools are
> powerful, subject to misuse, and if misused can provide any result that you
> wish. Still, the tools are there.
> You are going back around two generations further back than my project. I
> do get the impression that my project is stretching the limits a bit. I'd
> love to see how a project like you propose works out. I have hopes of
> someday extending my projects back a generation or two and would love to
> see how it works for someone else.
> My suggestion would be to start the project a bit more modestly, say late
> 1700's. As technology and database quality improve I'd try to work back
> from there.
> Sam

This thread: