by David B. Boles

Bowles DNA Project administrator



There appear to be two distinct subgroups within Group 8 that can be contrasted with the remaining kits.  One, which I will call the "12" subgroup, is distinguished by a value of 12 for marker 22, and consists of kits #126464, 287032, and 341274.  Because the Group 8 standard has a value of 15 for this marker, and there are no intermediate values (i.e., 13 or 14) among the other kits, it seems likely that the change from a value of 15 to 12 occurred through a single mutation event.


The other, which I will call the "11-17" subgroup, is distinguished by a pair of values: 11 for marker 26, and 17 for marker 30.  The kits are #57613, 263785, and 308906.  The two mutations may have occurred either as a single event, or as two events close together in time.


The remaining kits appear to have no distinctive characteristics that allow groupings.  Although three kits show a value of 35 for marker 34, they include two in the "11-17" subgroup, and one outside that subgroup.  Similarly, five kits show a value of 39 for marker 35, but they include kits both inside and outside the "12"  and "11-17" subgroups.  It seems likely that markers 34 and 35 independently mutated at several different times for different kits.


Although the remaining kits (which I will call the "remainder" group) cannot be said to form a subgroup, they do have an interesting characteristic: They show smaller deviations from the group standard than either of the two subgroups.  If confined to ancestors with some form of the surname, and at least 37 markers, they include kits #29401, 106142, 164685, 198726, and 355920.  The average deviation from the group standard is 1.80 markers for these five kits.  That contrasts with an average deviation of 2.33 markers for the "12" subgroup, and 3.67 markers for the "11-17" subgroup.


What the average deviations suggest is that members of the "remainder" group split off the ancestral line at a later date than either of the subgroups, with the "12" subgroup having split off at an intermediate time, and the "11-17" subgroup at the earliest time.  These times can be estimated, first by the number of generations.


1. "12" subgroup: This subgroup deviates by an average of 3.67 out of 37 markers from the "remainder" group.  Assuming the "remainder" group split off the ancestral line at a later date, that average can be converted to generations using the calculator located at  The calculator indicates that the "12" subgroup split has an expected value of 13 generations in the past.


2. "11-17" subgroup: This subgroup deviates by an average of 4.67 out of 37 markers from the "remainder" group.  The calculator shows that the split has an expected value of 16 generations in the past.  However, a second estimate can also be made using the average difference from the "12" subgroup since it more recently split from the ancestral line.  That average is 5.56 markers, with an expected value of 18 generations.  Thus from the two comparisons, the overall expected value can be placed in the middle at 17 generations.


These generation estimates can be converted to years using an average value of 31 years per generation (see the calculator page cited above).


1. "12" subgroup: 13 generations corresponds to 13 X 31 = 403 years.  Estimating the average birthyear of the "12" subgroup testees at about 1960 places the split from the ancestral line at about the year 1557.


2. "11-17" subgroup: 17 generations corresponds to 17 X 31 = 527 years.  Under the same birthyear assumption, this places the split at about the year 1433.


About all that can be said about the "remainder" group is that its members generally split off their common ancestral line more recently than either of the two subgroups, i.e., after 1557.  In other words they share ancestry more recently than that (although note the caveats below).  But because it is not a cohesive subgroup, the length of their shared ancestry likely depends on which individuals are compared to one another.




These conclusions come with some caveats.  The most important is that the generational estimates are estimates only, and can be off to a substantial degree.  While the strength of the subgroup approach is that every subgroup member provides an independent generational estimate, reducing the amount of uncertainty, with only three members in each subgroup the beneficial effect is small at the moment.  My best calculation is that currently, the 95% confidence intervals for each subgroup include a little less than 7 generations in either direction.  In other words, for the "12" group, while the expected number of generations since the split is 13, it could actually be anywhere from 6 to 20.  For the "11-17" group the expected number is 17 but could be anywhere from 10 to 24.


Another source of uncertainty comes from the conversion of generations to years.  Although 31 years per generation may be the mean, this value obviously can vary quite a bit.


The answer to these caveats is to add more subgroup members.  Increasing sample sizes will reduce the size of the 95% confidence intervals, and should bring the years per generation increasingly into line with the 31 year average.  Hopefully the identification of Group 8 with the Kelburn Boyle family will encourage an increased rate of kit submission, and lead to larger subgroups.  It is also possible that new subgroups will be discovered.


Finally, keep in mind that these are my personal conclusions only.  Please don't attribute them to FTDNA.