The tests I had two maternal second cousins (first cousins to each other) take may be bearing fruit.
My maternal grandmother, Margaret Colvert Allen, had two full sisters, Louise Colvert Renwick and Launie Mae Colvert Jones. K.J. and G.W. are Aunt Launie’s grandchildren by a son and daughter. Per Gedmatch, K.J. and I match 394.4 cMs (46.2 on the X), which is a whopping match for second cousins. G.W. and I match 151.5 (45.6 on the X), which is on the low side of the second cousin range.
Here is matrix showing the matches between K.J. and G.W. and my grandmother’s descendants, me (L.Y.H.), my mother B.A.H., my uncle C.C.A., and my sister K.H.J.
I have ZERO identified matches to my grandmother’s African-American lines. I was startled then to get a new match last week that matched me at 40 cM (per Ancestry, which lowballs estimates), K.J. at 91 and G.W. at 137. It didn’t take long to get a response from him that identified him as E.J., the great-grandson of my Aunt Louise. In other words, a second cousin-once removed to me, K.J. and G.W. My cM total with E.J. is lower than might be expected, but K.J.’s and G.W.’s are right in line with the estimated range. (I’m hoping he’ll upload to Gedmatch so we can get better cM estimates and wider comparisons.) So, finally, descendants of all three Colvert-McNeely sisters have done DNA testing and all match appropriately.
Several months ago, K.J., G.W. and I had a match to an unknown person who hit us all in the estimated 3rd-4th range. I have communicated with his sister, who indicated that three of their four grandparents were Caribbean-born, making her maternal grandfather the most likely connection. Unfortunately, A.P. knows very little about her grandfather. Comparing A.P.’s brother A.R. to me, K.J., G.W. and E.J. yields cM totals of 99, 98, 111 and 27.7. That’s in the half-second cousin or second cousin once removed range for me, K.J. and G.W. and half-third/third once-removed for E.J., indicating that we likely share at least one great-grandparent/great-great-grandparent.
Let’s work with that assumption. Though it’s possible that Caroline McNeely Colvert had a child other than my grandmother and her sisters, presumably older, I’m doubtful. She would not have been the first McNeely sister to bear a child out of wedlock, and there’s no reason to think she would have given up such a child. (Especially to anyone outside her large immediate family.) So, of the two, the more likely shared ancestor is Lon W. Colvert. Another possibility is Lon’s son, John Walker Colvert II. He had only one known child, a son who died in childhood in a car accident, but there could have been others and the cM numbers could work in that scenario. Right now, we just don’t have enough information, but Lon and Walker are the starting point of my working hypothesis.
[Update: 4/4/2017 — Mystery solved. As suspected, Lon W. Colvert is our most recent common ancestor.]