Dear List, I mentioned earlier that I was working on a cross-surname analysis of some surnames within M222. I have posted my results at <http://mysite.verizon.net/weh8/M222CrossSurname.pdf> Go there and let me explain what I have done ------------------ I had a list of 184 M222 haplotypes of 16 surnames. I generated an RCC matrix from them. I located the 16 surnames in the matrix. They are in column 1 with the number that were in the cluster. For example, there were 13 Dohertys (or spelling equivalents) in the Doherty cluster. They appeared in 16 identifiable clusters. Each cluster had 15 interclusters, one intersection with all the other clusters. Members of those intercluster regions had one testee from one cluster and one from the other. Then I formed a time slice matrix and set the lower RCC to zero and the upper RCC to 15. This assured that all values in the time slice matrix were members of the bonafide cluster(s). Then I looked at the testees who were in each cluster and members of any other intercluster. The names and numbers in the first column were the surname and the number of that surname (or soundex-equivalents) that were in the cluster. Then, I looked at the RCC points in the intersections of two clusters in the intercluster region. If one testee had 50% or more entries in the intercluster region, I listed the Kit No and name and the percentage of times it appeared in the intercluster at an RCC less than 15. For example, ten (71%x14) O'Doghertys were in the Cowan-Doherty intercluster.100% of the McGovern entries appeared in the Howle-McGovern intercluster. 50% of the Griersons appeared in the Dunbar-Greer intercluster, etc. These are the entries in columns 1-4. These four columns were the raw data. The columns to the right of the brown colored vertical line contain the analysis. In the top double-entry table, you read the entries this way: In the intersection of the Cowan cluster with the Doherty cluster, 71% of the 14 testees in the Cowan cluster appear in the intercluster. In the lower double entry table, we find that the name was O'Dogherty and on the left we see that his Kit number was 38173. Another example: When the 13 members of the Doherty clusters were paired with the McGonigals, we see in the top table that six entries had percentages over 50% McGonigals and those percentages are given in the table; and in the lower table we see the spelling of the names. If there are numbers in the lower table, then the spelling was the same as in the heading of the column. You will note that the Ewing cluster had 15 members but they were so tight that any surname that appeared in the intercluster entries had less than 50% of the possible entries. Now, how to use it -- Look at the intercluster regions that have many entries of RCC under 15. Those entries either represent an evolutionary bridge between the surnames or examples of NPEs or some other goof. To try to avoid problems of NPEs and goofs, I took 50% as the criterion. So, how about recent postings --- I went over a couple of them and picked out what some posters were looking at. Here they are, with my conclusions based on the two tables I have just discussed. There are two Milligan-Griersons. You might want to identify them, see where they lived and see if there is an intersection in their pedigrees. They might match on Family Finder, for example. There is only one McAdam-Milligan. No Grierson-McAdams. There is one Milligan-McCord. But two McCord-McGoverns. The Howle-McGoverns that I mentioned in an earlier postings show up again here. There are lots of connections between the Dohertys and the McGonigals and the McAdams and the McCords as well as the McGonigals and the McLaughlins. I will leave it to you all to explore similarities in spelling, which will surely occur at times when RCC is less than 15, about 1100 AD. You should look at the locations of where these surnames lived because they may well lead to insight about 'inbreeding'. But please read my papers and FAQs. If you don't, this will surely open a real can of worms otherwise (grin!). I do think that analyses like this one afford insight into the relationships between and among surnames (and various spellings) that cannot be found otherwise. All this information comes from the RCC matrix, not the phylogenetic tree, so individual pairs of values will have rather large errors. The phylogenetic tree will average out the uncertainties but it still will not be perfect due to unknown mutations. I have tried to minimize those errors by considering only surname pairs that appeared at a level of 50% or more in the RCC matrix. And you can also see the power of using the time slice matrix. Setting it to an RCC interval of 0-10 will tighten the relationships, and setting it at 0-20 will broaden the relationships found. Enjoy! - Bye from Bill Howard