FTDNA has recently added a "Matching" tab to the results pages for Big Y testers. Part of the data table is a list of shared Novel Variants, with most individual testers finding they have shared Novel Variants totaling around 60-70 in a total count that ranges from roughly 75 to 100. These shared SNPs are individually reported with the actual number of individuals sharing the SNP reported in parentheses. There are currently 38 testers reported for M222+ individuals (10 of whom are not members of the R-M222 Project), so if a SNP is common to all of them, each individual will find that he shares it with 37 others. There appear to be 42 universally shared Novel Variants reported to M222+ Big Y testers. Two of these SNPS (23856771 and 8157356) are classified as M222 equivalents. The 40 others are all found in M222- individuals and therefore lie upstream. At this point the problem of the low quality read begins to appear. Two individuals are necessarily positive for the SNP S7073 (26078887) because they are positive for downstream SNPs. But 26078887 is not reported for them as a novel variant because it seems to have been a No Call in their cases. It is known that it is possible to be M222+/S7073- because one such individual turned up in the Chromo2 testing in the British Isles. But on current evidence no FTDNA tester has been found in which S7073 is ancestral rather than a No Call. This occasional SNP drop-out continues down the list as lower numbers of individuals share a particular SNP - 35/36, 34/36, 33/36, 32/36, 30/36 and so on, as we see in one instance. Even so, most of the shared variants continue to be upstream mutations. I used a Z2961 Big Y test for primary comparison, but when a shared variant among M222 individuals was not seen in the Z2961 results, I looked further up the tree in the expectation that the missing Z2961 result was a bad read/No Call in that test, rather than evidence that the SNP was ancestral for the tested Z2961 individual. In all uncertain cases I found the derived value for the SNP further up in DF49, say, or L21. But I made no concerted effort to find the earliest instance of any upstream SNP because however interesting the knowledge might be in its own right, it is irrelevant to building the M222 tree. The number of reported upstream novel variants will vary with individual tests, but the total in a single test seems to lie in the 60-75 range. (No, I have not evaluated every single test - just a few.) That leaves around 15-25 relevant SNPs for each test. Of these, whether they are shared or not, one to five may be associated with the valuable and relevant SNPs that have helped build the tree as we now know it. The remaining unmatched SNPs are the candidates for further definition in the more recent branches of the M222 tree. A round number calculation that multiplies 40 tested individuals by 15 private SNPs per individual leads to a population of 600 SNPs that have the potential to refine the tree as new test results come in. That's a lot of words which I will summarize as follows: roughly 75-80% of reported Novel Variants are irrelevant to further development of the branching structure below M222; the reported Novel Variants for which a tested individual has no matches are the raw material for determining the future development of the tree. A Novel Variant for which one to several (but probably not over 5 or 6 at this point) matching individuals are determined may or may not have classificational validity. Remember that there are several known SNPs that behave erratically and turn up in M222 individuals in a pattern inconsistent with the rigid structure we have reliably determined from Chromo2 and FGC tests. The Geno 2.0 test has been helpful with determining a few reliable SNPs below M222, but several of that test's identified sub-M222 SNPs seem to be unreliable. Singletons from Geno 2.0 may be valid, but if they are private to an individual whose name we don't know, how do they help? David Wilson