OK, thanks to your off-list note, I have looked at Blaine Bettinger's table. And it really just emphasizes what I found with the GEDMatch Generations table. For definitively determining whether Mary or her daughter was Sam's mother, using the averages is simply not going to work. There is just too much of a range for evey one of those boxes in Blaine Bettinger's table. So a matching cousin who shares 331 cM could be any one of the following: 1C2R, 1C1R, 2C1R, 1C, 2C, 3C (though just barely), 1C1R, 2C1R, 3C1R (though just barely), 1C2R, 2C2R Using only the average, the 331 comes closest to 2C or 1C2Rm with 1C1R a bit further away. But restricting any conclusion to just those two or three is simply not supportable by the ranges in the table. You cannot rule out any of the other ones listed above. So the bottom line is the same: the chart does not suffice. And looking at the chart leads me to think that method 2 (below) is not going to resolve this, no matter who else we find to test, since the variabilty is so high (even in just 2 generations, a grandchild with average 1760 can have an actual range of 875 to 2365). So Method 1 is the only method by which we could answer the question -- with testing of enough descendants of Mary's husband Joe's sibling to assure that a non-match with Joe is real and not just a fluke of one or two descendants who happened not to inherit a particular segment. What are the cM's they share with each other and the suspected generations/family relationships? With Blaine's table and other available averages for genetic distance there might be some clue (all calculated given the standard settings at Gedmatch). Andreas > On Nov 23, 2015, at 00:49, Wesley Johnston via <genealogy-dna@rootsweb.com> wrote: > > I want to share what I found out in a first step toward method 2 below. > I generated a GEDMatch generations matrix for Sam's son and six of his documented cousins who have tested (closest of whom is either a 1st cousin once removed or 2nd cousin, depending on who was the mother of Sam's father -- and most distant of whom is either a 2nd, 2x or a 1st, 3x). > It turned out to be a calibration of the GEDMatch generations matrix against the two forms of the actual generations matrix (one form for Mary as the mother of Sam and one form for Mary's daughter as the mother). It clearly showed how wide the normal natural variability can be even in close generations: two siblings showed as 1.3 generations to MRCA for example. And one pair (the most distant one) known to be actual MRCA 4.5 was estimated by GEDMatch at 7.2 -- WAY OFF. > So it is clear that in-generation variability even within one generation already was off and that as the relationships became more distant, the range of variation widened rapidly, resulting in the very large error noted above. > And the bottom line for our effort to identify Sam's mother is that it simply cannot be done from the cases we have. > This became extremely obvious in comparing how two siblings (who are either Sam's son's 1st cousins 2x removed or 2nd cousins 1x removed) were estimated in comparison of Sam's son's closest cousin in the group. One of the siblings was estimated by GEDMatch to be exactly the same (3.5) for both Sam's son and Sam's son's closest cousin, which would indicate that Mary's daughter was Sam's mother. But the other sibling was estimated by GEDMatch to be exactly a half generation apart for the same pairings (3.2 and 3.7), which would indicate that Mary was Sam's mother. > So two siblings' estimates resulted in exactly opposite conclusions. And it was entirely due to the normal range of variation of inherited DNA. > So the bottom line is that the GEDMatch generations matrix alone does not suffice with the currently tested six cousins: it does not tell us who Sam's mother was. > So we now have to start identifying key people to test, both for method 2 and for method 1. >    On 21 Nov 2015, at 16:58, Wesley) Johnston via <genealogy-dna@rootsweb.com> wrote: > > > > Thanks very much for your responses on this. This has allowed me to see two different ways forward on this. > 1 - Elizabeth and Tim's method - Find descendants of Mary's husband Joe's siblings who are willing to test and see if they match Sam's son. If they do, the Sam's mother is one of the daughters of Joe and Mary. If they do not match, then it is evidence for Mary being Sam's mother (although we would really need to test several people and have them all not match in order to be sure it was not some fluke of atDNA inheritance). > 2 - Belinda's method - See what the existing cousins' level of relationship is to Sam's son in terms of shared DNA, maybe identifying and testing a few other relatives who could shed more light on this. > I am going to try method 2 first, since we have 7 relatives on Sam's mother's side who have already tested. So we might have enough to make a preliminary determination that way. But I ultimately see that method 1 is something we have to eventually do. >    Wesley > > > ------------------------------- > To unsubscribe from the list, please send an email to GENEALOGY-DNA-request@rootsweb.com with the word 'unsubscribe' without the quotes in the subject and the body of the message ------------------------------ Message: 4 Date: Mon, 23 Nov 2015 20:24:23 +1300 From: AJ Marsh <ajmarshnz@gmail.com> Subject: [DNA] What are typical SNP/ novel variant rates in YElite? To: genealogy-dna <genealogy-dna@rootsweb.com> Message-ID: <88D8E0A3-D6DE-4520-8130-B2947DF23457@gmail.com> Content-Type: text/plain;    charset=us-ascii Sorry list, I know this has been addressed on list before, but my memory fails me. YFull suggest that after the SNP L617 occurred, the tree started branching out about 3,400 years ago.  L617 may have occurred much earlier than this. I have access to 4 YElite results for L617s.  One of these has 68 SNPs/ novel variants which Full Genomes feel confident enough about to name.  There are many more lower reliability mutations discovered which for the moment I have put on the back burner.  But my first observation is that if it is 68 reliable mutations in 3,400 years that would be a nice simple 50 years per mutation.  I have not yet counted the total mutations in other branch lines, it is on my to do list. I believe some have looked closely at rates of mutations found in YElite, does the 50 years per "reliable" mutation sound consistent with what others have found? Of course, the speculated 3,400 years to common ancestor in my example is not firm and final, so it is a suspect figure to use for calibration of SNP rates. John. Sent from my iPad ------------------------------ To contact the GENEALOGY-DNA list administrator, send an email to GENEALOGY-DNA-admin@rootsweb.com. To post a message to the GENEALOGY-DNA mailing list, send an email to GENEALOGY-DNA@rootsweb.com. __________________________________________________________ To unsubscribe from the list, please send an email to GENEALOGY-DNA-request@rootsweb.com with the word "unsubscribe" without the quotes in the subject and the body of the email with no additional text. End of GENEALOGY-DNA Digest, Vol 10, Issue 639 **********************************************

Yes, individuals show a wider range of cMs the more distant they are. See also my graphic at www.segmentology.org for the concept of "tails" on the range of possibilities. However, as you get several more tests, they will generally tend to average out. Unless you have some ancestry on one alternative vs some different ancestors on the other alternative (allowing you to check for Matches on the unique ancestors), you are pretty much stuck with analyzing the cMs between pairs. Then use judgement for the best fit. The more tests you can get, the clearer the picture should become. Jim - www.segmentology.org > On Nov 23, 2015, at 9:42 PM, Wesley Johnston via <genealogy-dna@rootsweb.com> wrote: > > OK, thanks to your off-list note, I have looked at Blaine Bettinger's table. > And it really just emphasizes what I found with the GEDMatch Generations table. For definitively determining whether Mary or her daughter was Sam's mother, using the averages is simply not going to work. There is just too much of a range for evey one of those boxes in Blaine Bettinger's table. So a matching cousin who shares 331 cM could be any one of the following: > 1C2R, 1C1R, 2C1R, 1C, 2C, 3C (though just barely), 1C1R, 2C1R, 3C1R (though just barely), 1C2R, 2C2R > Using only the average, the 331 comes closest to 2C or 1C2Rm with 1C1R a bit further away. But restricting any conclusion to just those two or three is simply not supportable by the ranges in the table. You cannot rule out any of the other ones listed above. > > So the bottom line is the same: the chart does not suffice. And looking at the chart leads me to think that method 2 (below) is not going to resolve this, no matter who else we find to test, since the variabilty is so high (even in just 2 generations, a grandchild with average 1760 can have an actual range of 875 to 2365). > So Method 1 is the only method by which we could answer the question -- with testing of enough descendants of Mary's husband Joe's sibling to assure that a non-match with Joe is real and not just a fluke of one or two descendants who happened not to inherit a particular segment. > > > > What are the cM's they share with each other and the suspected generations/family relationships? With Blaine's table and other available averages for genetic distance there might be some clue (all calculated given the standard settings at Gedmatch). > > Andreas > >> On Nov 23, 2015, at 00:49, Wesley Johnston via <genealogy-dna@rootsweb.com> wrote: >> >> I want to share what I found out in a first step toward method 2 below. >> I generated a GEDMatch generations matrix for Sam's son and six of his documented cousins who have tested (closest of whom is either a 1st cousin once removed or 2nd cousin, depending on who was the mother of Sam's father -- and most distant of whom is either a 2nd, 2x or a 1st, 3x). >> It turned out to be a calibration of the GEDMatch generations matrix against the two forms of the actual generations matrix (one form for Mary as the

Wesley, you are at an advantage to most of us. You know the relationship between the testers. So focus on one assumption, eg real mother = mother and then check all amount of shared DNA vs the relationship they have under this assumption. I had a similar case and we could easily proof that someone's great grandma cheated. So in your case I would put up a spreadsheet, put both assumptions next to each other, name the relationship and give the total cM (remember to not change the standard setting of Gedmatch for one-to-one) next to it. Then do the same for the alternative assumption (real mother = adult daughter) Then upload that spreadsheet so that you can post the link here (attachments aren't allowed on this mailing list) and get the wisdom of the crowd as to which assumption is more likely. Without looking at that spreadsheet I can't rule out that it's not working (my method I explained in the private email and above) because the shared DNA with the testers should be different for both assumptions. Oh and a little overview of the testers relationships (just some simple cells showing who is descendant of whom) in the spreadsheet would help as well (to visualize). I know you're very familiar with the case but for everyone else it's hard to get a grip on it without looking at a simple family tree. Andreas > On 24 Nov 2015, at 09:42, Wesley Johnston via <genealogy-dna@rootsweb.com> wrote: > > OK, thanks to your off-list note, I have looked at Blaine Bettinger's table. > And it really just emphasizes what I found with the GEDMatch Generations table. For definitively determining whether Mary or her daughter was Sam's mother, using the averages is simply not going to work. There is just too much of a range for evey one of those boxes in Blaine Bettinger's table. So a matching cousin who shares 331 cM could be any one of the following: > 1C2R, 1C1R, 2C1R, 1C, 2C, 3C (though just barely), 1C1R, 2C1R, 3C1R (though just barely), 1C2R, 2C2R > Using only the average, the 331 comes closest to 2C or 1C2Rm with 1C1R a bit further away. But restricting any conclusion to just those two or three is simply not supportable by the ranges in the table. You cannot rule out any of the other ones listed above. > > So the bottom line is the same: the chart does not suffice. And looking at the chart leads me to think that method 2 (below) is not going to resolve this, no matter who else we find to test, since the variabilty is so high (even in just 2 generations, a grandchild with average 1760 can have an actual range of 875 to 2365). > So Method 1 is the only method by which we could answer the question -- with testing of enough descendants of Mary's husband Joe's sibling to assure that a non-match with Joe is real and not just a fluke of one or two descendants who happened not to inherit a particular segment.

OK, I have uploaded a spreadsheet to http://www.wwjohnston.net/famhist/SamMom.xlsx There are two worksheets. The first worksheet ("Map") shows how all of the seven tested people descend from Joe & Mary, with the two alternate scenarios for who is Sam's mother. The generations are aligned, so that they are all clear. There are 2 pairs of siblings (Belle & Bill, Dan & Don) and one pair of 1/2 siblings (Alice & Arnold), all of which are identified as such, so that these relationships are clearly understood. The second worksheet ("Matrix") has six different matrices of the seven people tested. The top two are the actual known relationship matrices, one for Mary as Sam's mother and one for one of Mary's daughters as Sam's mother. Below these is the GEDMatch-generated Generations Matrix. This is where Belle shows at 3.5 from both Stan (Sam's son) and Ann but Belle's brother Bill shows at 3.2 from Stan and 3.7 from Ann -- the conflicting data supporting opposite conclusions about who Sam's mother was. Then there is a black bar, to signify that below it there are no assumptions about who is Sam's mother. These three matrices are all based on GEDMatch. The top two are the total cM values for (a) the values show in the GEDMatch Match List and (b) the GEDMatch-generated atDNA matrix. Most of the values in the matrices are the same but not all. I am not sure why they are different -- probably something to do with GEDMatch's different thresholds for the two different presentations. Both of these matrices seem to support the opposite conclusions when looking at Belle and Bill's values compared to Stan and Ann. The bottom matrix shows the "largest cM" values from the GEDMatch Match List. This is the only place were Belle and Bill compare in the same way with Stan and Ann -- 38.9 with Stan for both Belle & Bill but 28.4 and 23.2 with Ann, respectively. So this matrix (unlike all the others, which support both scenarios for who was Sam's mother) seems to support Mary as the mother of Sam. I just do not see anything in all this that conclusively resolves the question of whether Sam's mother was Mary or one of Mary's daughters. The only matrix that does show a clear indication is the largest cM matrix. But the other matrices all have evidence to support both conclusions -- and thus neither one conclusively. Am I not seeing something here that makes things more conclusive than I am seeing them? From: Andreas West <ahnen@awest.de> To: Wesley Johnston <wwjohnston01@yahoo.com>; genealogy-dna@rootsweb.com Sent: Monday, November 23, 2015 9:36 PM Subject: Re: [DNA] Real Mother - the Mother or adult Daughter? Wesley, you are at an advantage to most of us. You know the relationship between the testers. So focus on one assumption, eg real mother = mother and then check all amount of shared DNA vs the relationship they have under this assumption. I had a similar case and we could easily proof that someone's great grandma cheated. So in your case I would put up a spreadsheet, put both assumptions  next to each other, name the relationship and give the total cM (remember to not change the standard setting of Gedmatch for one-to-one) next to it. Then do the same for the alternative assumption (real mother = adult daughter) Then upload that spreadsheet so that you can post the link here (attachments aren't allowed on this mailing list) and get the wisdom of the crowd as to which assumption is more likely. Without looking at that spreadsheet I can't rule out that it's not working (my method I explained in the private email and above) because the shared DNA with the testers should be different for both assumptions. Oh and a little overview of the testers relationships (just some simple cells showing who is descendant of whom) in the spreadsheet would help as well (to visualize). I know you're very familiar with the case but for everyone else it's hard to get a grip on it without looking at a simple family tree. Andreas > On 24 Nov 2015, at 09:42, Wesley Johnston via <genealogy-dna@rootsweb.com> wrote: > > OK, thanks to your off-list note, I have looked at Blaine Bettinger's table. > And it really just emphasizes what I found with the GEDMatch Generations table. For definitively determining whether Mary or her daughter was Sam's mother, using the averages is simply not going to work. There is just too much of a range for evey one of those boxes in Blaine Bettinger's table. So a matching cousin who shares 331 cM could be any one of the following: > 1C2R, 1C1R, 2C1R, 1C, 2C, 3C (though just barely), 1C1R, 2C1R, 3C1R (though just barely), 1C2R, 2C2R > Using only the average, the 331 comes closest to 2C or 1C2Rm with 1C1R a bit further away. But restricting any conclusion to just those two or three is simply not supportable by the ranges in the table. You cannot rule out any of the other ones listed above. > > So the bottom line is the same: the chart does not suffice. And looking at the chart leads me to think that method 2 (below) is not going to resolve this, no matter who else we find to test, since the variabilty is so high (even in just 2 generations, a grandchild with average 1760 can have an actual range of 875 to 2365). > So Method 1 is the only method by which we could answer the question -- with testing of enough descendants of Mary's husband Joe's sibling to assure that a non-match with Joe is real and not just a fluke of one or two descendants who happened not to inherit a particular segment.