GENEALOGY-DNA-L ArchivesArchiver > GENEALOGY-DNA > 2008-09 > 1220614916
From: Vincent Vizachero <>
Subject: Re: [DNA] Age of R1b
Date: Fri, 5 Sep 2008 07:41:56 -0400
On Sep 5, 2008, at 12:07 AM, Anatole Klyosov wrote:
> I hope you have noticed that my calculations of some 17,000 R1b
> haplotypes were accompanied by words "Naturally, it is not a clean
> experiment. O.K., lets move to more specific sets of R1b haplotypes".
Then we agree there are limitations, and can move on from this point.
> Now, in case of Asian haplotypes, they cannot give 16,000 years to
> a common ancestor without a large variations in their mutations.
> You cannot get such a large variations by just taking various
> European subclades of R1b. Clearly, there are some large difference
> in Asian R1b haplotypes compared to European ones. That is what
There ARE large differences, and they stem from the fact that R1b in
Europe represents a very small subset of all R1b. Only the
downstream clade R-
S128 (that is, R1b1b2a) is represented to any significant degree.
Any sample drawn from Europe will therefore underestimate the age of
R1b because, in truth, that's not what you are measuring with such a
> Just throwing names of subclades "(R-M73?, R-M18?, R-M335?, R-P25?,
> R-M343?) without understanding how they differ by mutations would
> not add anything in our knowledge. Show me that their mutational
> differences can account for the 16,000 year difference between them
> (or would bring the 16,000 years to a common ancestor), and I will
> thank you. So far - not. Empty words.
I understand well how the subclades of R1b differ from each other, or
I would not have brought up the subject. You yourself produced an
estimate for "Central Asian R1b" and I am merely pointing out why
your estimate for that sample was so much greater than the age of
your other "R1b" samples.
> I completely disagree. We should use whatever is available, and
> formulate a working hypothesis. Then we can examine it further,
> when new haplotype sets are available. That is how science is
> developing. We cannot sit and wait until everything is just
> perfect. Of course, a working hypothesis cannot allow us to make
> categorical claims, however, it clarifies where to go further.
I think you misunderstand. If some author merely says "this
population has an age of 6,000 years" and provides no haplotypes,
then we can not put that estimate on equal footing with an estimate
we have calculated ourselves because we can not replicate it.
Further, if we want to compare a sample from one source with a sample
from another source, intending to figure out which is older, we can
not do it unless we calculate the ages of both samples using the same
>> Contu mistakenly classified a R-M269 chromosome as being R-M18. In
> other words, they labeled a guy as M18+ who is really M269+. It is
> either a genotyping error or a clerical error. Either way,
> including that mistaken chromosome in their TMRCA analysis led them
> to severely overestimate the TMRCA of R-M18.
> Wrong again. There were 64 R1b1c haplotypes and 8 R1b1a haplotypes
> in the Contu's paper. Are you saying that those 8 haplotypes . . .
> distorted the bulk of other 64 haplotypes (M269) . . . that much??
> That the 5,000 years to a common ancestor moved to 22,800 years?
> Come on... Contu et al had much more serious reasons to screw up
> their numbers.
I agree that all the Contu estimates are wrong, because they used an
improper procedure to calculate them.
However, the R-M18 estimate is doubly wrong, because one of the
eight so-called R1b1a haplotypes is not actually R1b1a. One of the
eight is actually R-M269, and including haplotype in a TMRCA
calculation for R-M18 leads to an overestimate.
Imagine if you had seven haplotypes from J2a and one you thought was
J2a but was really R1a. If you use all eight to estimate the TMRCA
for J2a, you will have a serious problem. This is what Contu did for