Archiver > GENEALOGY-DNA > 2009-12 > 1260408500

From: (John Chandler)
Subject: Re: [DNA] R-U152 and R-L21 on the European Continent
Date: Wed, 9 Dec 2009 20:28:20 -0500
References: <><A93225B377724B83BF5250F368878955@anatoldesktop><><><><><><><>
In-Reply-To: <>(message from David Faux on Wed, 9 Dec 2009 12:30:04 -0800)

> So the solution would seem to be a type of catatonia - not do anything

No, David, that's apparently your solution -- you call it
"fence-sitting", but it's really just an unwillingness to accept the
facts. I will repeat what I said yesterday because you completely
ignored it. The concept of calculating the age of a haplogroup by
measuring the variance of the genetic markers of its members is
fundamentally flawed. It cannot ever work satisfactorily because it
depends critically upon supplementary data that are not available. In
particular, it cannot be patched up by replacing the known mutation
rate by an ad-hoc rate derived from a handful of quite low-time-depth
case studies.

> whether you call it the use of "fudge factors" or whatever to address a very
> difficult if not insoluable problem. I applaud ZUF for taking a step in
> what would seem to be the right direction.

On the contrary. That's a step in exactly the wrong direction. It is
an attempt to keep around a flawed and inapplicable model for the sole
reason that it seemed simple to begin with. The variance method is
fine for placing lower limits on the age of a clade, but it requires a
correction that cannot be computed without knowing all the begats in
the clade's history. (Well, not absolutely all begats -- just the
ones connecting the present-day members with the MRCA.) In fact, it
is a tool that could be used in the other direction -- after deriving
the age of a clade by other means, such as the interclade variance
method, you can infer something about the demographic history of the
clade by examining the rate of accumulation of variance.

John Chandler

This thread: