On 8/26/2017 4:03 PM, taf wrote: > In fact, the smallest divisible segments are probably in the 10s to > 100s of thousands of bases, which is what puts a definite limit on the > number of generations over which autosomal is likely to be informative > without a huge amount of luck (for every ancestor from 1600 from whom > you have a detectable preserved block, you have many more ancestors > from whom you inherit no DNA whatsoever). These blocks pass intact for > an incredibly long time, the block that includes the gene determining > the most common form of blue eyes is about 150,000bp long and to have > passed largely intact for more than 12,000 years. This presents a > problem on two sides - relatively close relative may not share the > block at all. If two people do share the block, it shows they are > related, but perhaps too distantly to be genealogically relevant. Doesn't the "centimorgan" (cM) measurement take this partially into account?  If I understand the method correctly, one cM will sometimes contain a huge number of base pairs, and sometimes a relatively small number (if it is contained in a "hot spot").  So, wouldn't a long segment that passed on intact (or essentially intact) over thousands of years have a centimorgan measurement close to zero?  Do I have this right? With regard to Richard's Y-DNA, have STR-tests been done for enough markers that one could do a global search to see what surnames pop up among his closest matches?  I know that the noise to information ration can be too large if not enough markers have been tested, but it seems like it would be worth a shot. Stewart Baldwin