GENEALOGY-DNA-L ArchivesArchiver > GENEALOGY-DNA > 2012-02 > 1329442174
From: (John Chandler)
Subject: Re: [DNA] Out of Africa
Date: Thu, 16 Feb 2012 20:29:34 -0500
> I have explained here many times a very simple and basic thing: in DNA
> genealogy you operate not with generations per se, but with a product "kt",
> which is a product of the generation length and the mutation rate constant.
Unfortunately, the father-son mutation studies have shown that the
relevant time-like unit for Y-STRs is the literal generation. That
means the rate constant "k" must be calibrated in mutations per
literal generation, not per year. The corresponding value of "k" in
mutations per year isn't a constant after all, but rather a
socio-geo-econo-political variable that must be determined
independently for each situation.
> For example, the mutation rate constant 0.12
> mutation/haplotype/generation for the 67 marker haplotypes is valid ONLY for
> 25 years per (conditional) generation.
In fact, that isn't a constant at all and is not valid EXCEPT when the
actual average literal generation length is equal to that of the
specific set of data used to come up with that figure. Since you don't
know the actual average literal generation length for that figure, you
> Yes, it is a problem with the SNP-based dating, because they (e.g.,
> Cruciani) make two independent assumptions - on the mutation rate constant
> (huge margins of error) and generation length (huge margins of error,
> actually such as between 16 and 35; in short, we do not know).
For Y-SNPs, we are all stuck because there are no direct measurements.