I have a minority view: that testing beyond 37, or even 25, markers is usually not helpful. I am the administrator of the Willis project at FTDNA. We have about 115 members. About 70% of us match someone else in the project. We have about a dozen groups of Willises (the name is derived from the name William and originated in many different locations in Great Britain). By far the biggest benefit of Y-DNA testing for members of the Willis project is that they find out which group of Willises they belong to. There is no instance where testing beyond 25 markers would have been needed to figure out which group to assign the member to. Interestingly, out of the dozen groups in the Willis project, only a couple of them know who their common ancestor is. Solving this problem by analyzing mismatched markers has not helped much, at least for me. When, and in whose line, did the mismatch occur? Dave Willis

Hello Dave, I have a particular interest in WILLIS because my maternal grandmother was a WILLIS. Unfortunately, I haven't been able to find a living cousin to test. You have the advantage of high numbers in your project, for which I congratulate you. For any project, the number of unmatched members is going to go down as the size of the project increases. As for the biggest benefit of Y-DNA testing being that people find out which group of WILLISes they belong to, yes, I think most of us would agree. As for not needing more that 37 markers, or even 25, you know I'm going to disagree with you, but I would also have to say, "it depends." And it depends mainly on the rarity of the haplotype -- and the rarity of the surname. I have a project that has both a rare haplotype and a rare surname: CARRICO. It appears all CARRICOs in the U.S. prior to 1900 descend from one immigrant to Maryland in 1674 (none has a GD from the family's modal haplotype of more than 3). They are an uncommon haplogroup, J2a4b, and the haplotype for the family is so rare, their highest matches outside the family are 5/12, 17/25, and 21/37. They have a value at one marker in Panel 4 (38-67), possessed by no one else ever tested, at least the last time I checked. If someone is surnamed CARRICO (or variation), and all they want to do is support that they belong to this family (and don't have an NPE), all they need is a 12-marker test. I would, however, consider this project quite exceptional. Surname BIDDLE and it's variations are also rare, but it's haplotypes are not. Of the ten test subjects, nine are R1b1b2: http://dgmweb.net/DNA/Biddle/BiddleDNA-Results-R1b.shtml Two of the four families match 12/12 and a third 11/12. Two families even match 22/25, with all three differences at fast-moving markers. But at 37 markers, any resemblances between the families falls away, and at 67 markers disappears completely. I would not have wanted to base the connections here on only 12 or 25 markers. I'm not the admin of the THOMPSON project, but I do have a brick wall with a THOMPSON ancestor. I've had a cousin tested, and he turns out to be Haplogroup I1, so I extracted the I1 THOMPSONs from the project to do my own analysis of them: http://dgmweb.net/DNA/Thompson/ThompsonDNA-results-HgI1.shtml I would draw your attention first to groups A and B. They match 12/12, but only 22/25, which is not a strong match but still possibly the same family, however their match drops to 29/37 indicating a non-match. I would not have wanted to make the call, either way, with just 25 markers. I would next draw your attention to groups F, G, and H, which match 12/12. We can't tell where F belongs because he has tested only 12 markers. Looking at G and H, they match 23/25, which certainly could mean they're related in genealogical time. However, at higher levels, they are only a 29/37 and 55/67 match, clearly indicating no relationship. If people have a paper connection, especially if their surname is uncommon, and they match 12/12, you could stop with 12 markers if all you wanted was to support that they belong to the family. But I submit it is risky to stop at anything less than 37 markers when their is no paper connection, especially if the surname and/or the haplotype is common. The more you make assumptions, instead of testing, the more likely you are to have made an error, and isn't this what DNA testing is all about: to fix old errors and avoid making new ones? Besides, I'm lazy. Decisions about connections that are difficult to make at 25 or 37 markers usually become slam dunk easy at 67 markers, and I like easy. Diana > -----Original Message----- > From: y-dna-projects-bounces@rootsweb.com On Behalf Of David Willis > Sent: Saturday, June 12, 2010 9:49 PM > To: y-dna-projects@rootsweb.com > Subject: Re: [Y-DNA-projects] testing to 67 markers > > I have a minority view: that testing beyond 37, or even 25, > markers is usually not helpful. > > I am the administrator of the Willis project at FTDNA. We > have about 115 members. About 70% of us match someone > else in the project. We have about a dozen groups of Willises > (the name is derived from the name William and > originated in many different locations in Great Britain). > > By far the biggest benefit of Y-DNA testing for members of the > Willis project is that they find out which group of Willises they > belong to. There is no instance where testing beyond 25 markers > would have been needed to figure out which group to assign the > member to. > > Interestingly, out of the dozen groups in the Willis project, > only a couple of them know who their common ancestor is. > Solving this problem by analyzing mismatched markers has > not helped much, at least for me. When, and in whose line, > did the mismatch occur? > > Dave Willis