GENEALOGY-DNA-L Archives

Archiver > GENEALOGY-DNA > 2005-06 > 1118120164


From: "Andrew and Inge" <>
Subject: RE: [DNA] Lancaster Project
Date: Tue, 7 Jun 2005 06:56:04 +0200
In-Reply-To: <REME20050606150727@alum.mit.edu>


Andrew wrote:
> Wouldn't this mean that you are saying that the null hypothesis should be
> that a haplotype with a particular surname will be *different* from modal,
> and not just for example 50% likely to be different?

John Chandler wrote:
Quite right. Consider a haplogroup that is 10,000 years old. Call that 400
generations. The probability that a descendant line has had no
mutations at all over that time span on 25 markers is about 1e-10.

Perhaps I wrote this wrongly, but clearly we are talking about the
probability of a particular marker being modal, and not a whole "signature".
Sometimes it helps to change the terms and make an analogy. Let's say that
archeaologists find two new alphabets in Central Asia, both of which seem to
come from the Aramaic ones, like ours. They find clay tablets showing that
at least most of the letters were taught to pupils in something like the
same order as in the other Aramaic alphabets ('alef beth, alpha beta, etc).
However no complete tablet is found proving that a "b" sound is second. You
are saying that we can assume that the second letter is *not* a "beta" type
of letter, but something else. This is very odd and I can only assume that
you've lost track of what we were discussing. To stick to the analogy, the
probability you seem to be talking about is the probability that the new
alphabet will be exactly like the original Aramaic one, which is not what we
are discussing.

Andrew:
> Obviously you *can* say that the existance of a link is proven or proven
to
> be likely, and *not* be understood to be claiming to know what the link
is.

John:
Wrong. We may be dealing in genetics, but this is still genealogy. A link
is a specific relationship. What you have is merely a vague
feeling that there "must" be a common ancestor somewhere and at some time,
if only you could find evidence. Go look for evidence, and you
may then be able to prove a link, but not until then. Note that the
probability of a common ancestor rises monotonicaly to the limit of
100% if you just push back the time window enough. Therefore, by your
understanding of a "proven link" everyone already has it, and the
discussion is moot.

Well, I think you know I was never proposing that I had discovered the
specific relationship and so the question becomes: where does it become
"moot"? That was my question I suppose. If you have a 37 marker match for
example, what is the chance that the link did not happen in the last 500
years? I know that FT DNA and others have calculators but (a) I was
specifically talking about a case where the markers will be a mixture of
Sorensen and FT DNA markers and perhaps some of those markers are less
useful than others? and (b) I wonder what people like yourself think about
those calculators.

Andrew:
> Clearly in fact this is the whole basis of why people even discuss
> comparisons of people with different surnames - because they are
considering
> looking for evidence such as paper trails in a particular direction.

John:
Mostly, the reason they discuss it is desperation -- the lack of anything
else concrete to discuss. In practical terms, it's a very
difficult quest and rendered even trickier by the alternative possibilities
of non-paternal events all along the way or simple
lineages that join before the availability of documentation.

That something is difficult and tricky is not necessarily a good argument
against it.

Best Regards
Andrew


This thread: