Those of us who have DNA-tested multiple family members may notice some unexpected results among our shared matches. Among known family members, our level of shared genetic material may vary widely, especially as the relationships grow more distant.
Third and fourth cousins who show a strong level of relation to me, while their full siblings barely match me.
As Lisa suspected, this situation may partially be connected to test analysis capabilities, though this is more apparent when we check out this chart comparing results between companies. It seems Ancestry.com claims better results than other DNA companies when it comes to detecting relationships at the level of more distant cousins.
That, of course, does not reveal the entire picture of why one sibling of a distant cousin DNA match might show a stronger relationship to us than that match's sibling at the same testing company. A strong part of the answer involves the concept of recombination. But even within that process of recombination, males and females have differing rates of recombination. Thus, there could be an impact to the comparison if the distant-cousin siblings you match are brother and sister rather than, say, brother and brother.
I have observed examples of those differing results between matches, in particular when preparing to take a class given by Blaine Bettinger. One of the course prerequisites was to have DNA test results from myself and two siblings. Since I don't have two full siblings—the others in my family are half-siblings—I asked my two sisters-in-law, along with their brother (my husband) to participate in this preparatory phase for my class. In comparing results showing each participant's matches, this very situation was quite evident.
One way to see this demonstrated is to look at data provided in Blaine Bettinger's own crowdsourced "Shared centiMorgan Project," now accessible in an interactive format at DNA Painter. Let's take, for example, the website's readout for the DNA match relationship of third cousin.
I've taken this example from the current version of the Shared cM Project at DNA Painter. The top line—"3C"—of course is shorthand for third cousin. Underneath that, the 73 refers to an average centiMorgan result for third cousin matches based on the sixty thousand submissions provided to the Project by volunteer participants.
However, take a look at the final line in this third cousin square from the chart. That entry refers to the range of shared centiMorgans observed between two matches who are third cousins. In the experience of those reporting their test results for this project, two third cousins can share up to 234 centiMorgans. More to our point today, though, notice the lower number in that possible range shared by two third cousins: zero. Yes, zero.
In other words, it is quite possible for two bona fide third cousins to share absolutely no family-linked genetic material—other than, of course, the genes commonly shared by all humankind.
There have been several articles which explain this same issue demonstrated in the Shared cM Project chart at DNA Painter. One such example, from LegacyTree Genealogists, examines a potential "nonmatch" between two individuals who were supposedly third cousins, and explains the process used to determine that the two were family members, after all.
When we consider that, at least on mathematically-determined averages with the assumption of two to three children per generation (a conservative estimate), we could have 190 third cousins, our chances of encountering such a distant cousin who, while matching us, does not match our sibling is a strong possibility. I know I have seen wide variations in shared centiMorgan counts in my own exploration of test results for my husband and his two siblings.
However, no matter how helpful we may find the Shared cM Project to be—or its easily usable interactive version featured at DNA Painter—we need to remember that it is less likely that a value of zero shared centiMorgans for a non-matching cousin will be submitted to a crowdsourced database. For those readers who are fans of number-crunching, this article leads one to think that perhaps such non-matches may be more prevalent than previously assumed.
It was an eye opener to compare the matches showing on the lists of my husband and his two sisters. I actually found the variances I discovered to be quite helpful. One sister's results seemed to reach far back through the generations to her paternal roots in County Tipperary, Ireland, while the other sister's matches pointed more clearly to another branch of the paternal line nearer County Cork. If I only had the one sister's DNA results to work with, I would have missed the matches from the other part of their ancestry.
Because we use DNA test results to suggest directions for further genealogical research—and possibly lead to filling in the blanks on a pedigree chart—this same anomaly of the missing third cousins can work in our favor when we have siblings willing to help out by submitting their own DNA test for our use.