The Sinister Problem of Convergence

As you know from my last post, Charles Hayes, #138, matches all the English Colony Berrys 25/25, and is 36/37 with Robert, #43 (DYS456), and Alan, #112 (CDYb), the only other two who have tested 37 markers.

Lawrence Mayka, has just posted the following note to the Rootsweb DNA list:

In a highly populated haplogroup such as R1b1b2, almost every possible combination of values on fast-changing markers is occupied by _someone_, usually from the British Isles.    A member of R1b1b2 is thus almost entirely dependent on the slower markers for differentiation.    If, sadly, that member happens to have modal values for almost all the slower markers, he may very well get coincidentally convergent matches at 37 markers.    Luckily, of the additional 30 markers in a 67-marker upgrade, 28 of them are slow, and they generally do a good job of widening the gap.

An excellent example on Ysearch is that of 4ZKEU of Poland and M6PHZ of England.    At 37 markers, their GD is 3; at 67 markers, the GD is 16.    The Englishman has a relative who is even closer to the Pole:   a GD of 2 at 37 markers, a GD of 15 at 67 markers.

Coincidental convergence at 37 markers is not restricted to R1b1b2.    It can occasionally happen in other haplogroups, even G2a.    In one spectacular G2a example, a GD of 7 on 37 markers became a GD of 28 on 67 markers.

This is why:

- A putative cross-surname match at 37 markers must _always_ be extended to 67 for verification.

- At larger genetic distances, it is generally best to ignore the faster-changing markers and focus on the 54 slower markers.    Otherwise, your closest matches will be those (usually British Isles) individuals whose fast-changing markers _happen_ to match yours, purely due to coincidental convergence.

For the reasons stated, I would encourage each of you English Colony Berrys to expand your test results to 67 markers for the purpose of minimizing 'The Problem of Convergence'.