If they are not on an overlapping segment (overlapping at least 5cM, preferrably more), it doesn't count. The group of folks who take the atDNA tests is relatively small and many have deep ancestries - you can easily have a match without the atDNA. If you and I share a 10cM segment on Chr 4, it means nothing if I match John Doe on Chr 5 and you match him on Chr 6. That does not create a 3-way. And even if you and I find a Common Ancestor, it may not be the one who have us the matching atDNA segment. It would be amazing if John Doe shared the same Common Ancestor with us. Make sure the overlap is larger than the cutoff you use. Jim Bartlett On 05/17/12, [email protected] wrote: Margaret (and Greg) see my post (just before yours in the digest mode). Ã All the people with whom I share that segment have each other somewhere on their DNA and except for some obvious cases where the 2 peoples' segments are shorter at opposite ends of the segment, they all match each other on that segment on c 18. Ã And everyone has multiple segments with at least one of the other members of the group.

Margaret, I downloaded all of my atMatch segments from 23&Me (thru Ancestry Finder) and at FTDNA (5 at a time from Chr Browser) and combined them all into one spreadsheet, which I then sorted by Chr and start location. This arrays every matching segment I have from the beginning of Chr to the end of Chr X. I've then inspected the list and assigned group codes (eg 01n) to each overlapping group. I've then emailed the groups to see if they match each other. In several cases an atMatch and I have already determined a Common Ancestor. The thrid party in the group was amazed that I know her ancestor in the first email. FTDNA and 23&Me have determined the cutoffs (FTDNA is typically very conservative, so I believe virtually all their matches are good ones). I'm finding Common Ancestors for the groups. I'm also finded that some of the Common Ancestors I had previously identified, just can't work - they result in two different ancestors from one parent on the same segment. So the group is looking for other Common Ancestors, and several times we've found more than one from Colonial VA. I also belive that, particularly for 7.7cM segments, on one side or the other of that segment will be another of your ancestors who are a close cousin to the one youyou started with. These rules, like in geometry, will be powerful tools, to help us fill in our Ancestors. If I were you, I'd follow up on every lead - these are probably your cousins. Jim Bartlett On 05/17/12, Margaret Waters<[email protected]> wrote: Jim, The matches range between 6.1 and 9.7 cM with anywhere from about 700 to 1000 SNPs. Even if half of these matches were to my dad's lines and half to my mother's, that still seems like a lot of "cousin matches" (at least 25) in that one small location. It may be worth checking on the few larger ones but I suppose many of them are from the atDNA soup. This is all fascinating. Thanks for your input. Margaret

Margaret, No, I don't think so. I have 30 overlapping matches on Chr 8 between 106 and 127 Are the matches on large segments (23&Me uses 5cM as cutoff, FTDNA uses 7.7cM as cutoff, some advocate 10cM as a cutoff)? It they are smaller segments, most of us chalk it up to atDNA soup, not tied to any specific ancestor. If they are large segments, then count your self very lucky. If you can contact them, see how many of them relate to each other - if you get a 3-way match (each person has a match with the other two within the same large segment), then almost certainly the 3 of you share the same Common Ancestor. This is a relatively easy step that all adoptees should be taking - look for 3-way matches, or greater: if two can find a Common Ancestor the 3rd has a very strong clue. And in any case such a group can pool their information and look for place/time matches. If you match one other person and find a Common Ancestor, you don't really know for sure if the two events are connected - it could be a happenstance. I could probably find a Common Ancestor with almost anyone with real deep roots in Colonial VA (whether or not they tested DNA). But when 3 widely separated cousins all share the same atDNA segment, AND agree on a Common Ancestor - that's a very strong indication that the atDNA did come from that ancestor (usually, of course, our Common Ancestors are husband/wife couples - even if we don't know their names - and we don't know which one provided the atDNA). A 4-way is almost a guarantee. Another point you raise is about a lot of matches on the same segment area. If you take a single point on a chromosome, or some segment length that is common to all of the matches, there can be only two of your ancestors involved - one paternal, and one maternal. The atDNA you got from your Mom or Dad had only one of their ancestor's atDNA at that point - and that's it! Actually to get a little more technical, it's an "ancestral line" at any point on the chromosome - one person could match at one generation, and another person could match at one of their parents level. So pick any 7.7cM segment. Either your Dad or your Mom gave it to you - probably as part of a much larger segment. Let's say Mom. Well she got that segment from either her Dad or Mom; whichever one it was, that ancestor got it from one of their parents, etc. Until we get back to the earliest ancestor who had that whole 7.7cM segment - for that person, the segment was made up of smaller pieces from his/her parents - but we'll never know that because those segments were smaller that 7.7cM and those ancestors wouldn't pass the filter (actually the very distant living cousin/descendants of those ancestors whose small atDNA segments wouldn't pass the filter - so they wouldn't show up as a match). You could have a Common Ancestor at any of these levels, but always on the same ancestral line. Perhaps getting too deep - the bottom line is that at any Chromosome location - you have atDNA from an ancestor of your Mom AND atDNA from an ancestor of your Dad. A very valuable piece of information. So if you have 50 segments that overlap the same location on a Chromosome, some of them are from one maternal line, and the rest are from one paternal line. Memorize this! Jim Bartlett On 05/17/12, Margaret Waters<[email protected]> wrote: While playing with my data at GedMatch, I noticed that I had nearly 50 matches within a small range along chromosome 15. (I was using the Chromosome Segment Comparison.) This is way more matches than I have on any other chromosome and none so tightly bunched. Is this perhaps an area where many unrelated people might have the same segment lineup (or whatever the proper terminology might be)? Margaret

While playing with my data at GedMatch, I noticed that I had nearly 50 matches within a small range along chromosome 15. (I was using the Chromosome Segment Comparison.) This is way more matches than I have on any other chromosome and none so tightly bunched. Is this perhaps an area where many unrelated people might have the same segment lineup (or whatever the proper terminology might be)? Margaret

Hi Margaret (and other interested readers), I have the same thing on chromosome 15, but not just on chromosome 15. Furthermore, I had my mom and dad both tested, and this "stripe" appears on chromosome 15 for my mom, but not so much for my dad. Other stripes appear elsewhere, seemingly haphazardly. We had some discussion about this on the genealogy-dna list. I made graphics for all the matches of my mom and my dad. You can see these "IBS hotspots" (i.e., stripes) here: http://freepages.genealogy.rootsweb.ancestry.com/~gbonner/DNA/FF/filmstrip.html I am guessing that it is due to inadequate mixing of solution near these spots on the chips of those effected, but am not sure how they are doing that, so don't know for sure. I'd still like some better explanation if someone has one. Best, Gregg >________________________________ > From: Margaret Waters <[email protected]> >To: [email protected] >Sent: Thursday, May 17, 2012 3:42 PM >Subject: [AUTOSOMAL-DNA] Strange observation > >While playing with my data at GedMatch, I noticed that I had nearly 50 >matches within a small range along chromosome 15. (I was using the >Chromosome Segment Comparison.) This is way more matches than I have on >any other chromosome and none so tightly bunched. Is this perhaps an >area where many unrelated people might have the same segment lineup (or >whatever the proper terminology might be)? > >Margaret > > >______________________________ >For answers to Frequently Asked Questions about mailing lists, please see: >http://dgmweb.net/MailingListFAQs.html > > >------------------------------- >To unsubscribe from the list, please send an email to [email protected] with the word 'unsubscribe' without the quotes in the subject and the body of the message > > >

23andMe is looking for GEDCOMs from customers for testing purposes: https://www.23andme.com/you/community/thread/13334 If you are a customer and have one, please consider participating. Thanks! CeCe www.yourgeneticgenealogist.com www.studiointv.com

If you want to add one more consideration to the debate, it has become evident that besides both company's "predicted/suggested" 3rd and 4th cousins being more distant and that most people should give greater consideration to the range of relationships suggested, it is also clear that FTDNA's algorithm is less accurate when applied to 2nd-3rd cousin range relationships with it placing greater emphasis on the largest segment than on total share of segments over 5 cM (23andMe's methodology). I have several examples where FTDNA is actually suggesting a more distant relationship than actual due to comparatively smaller largest segments resulting from primarily female meiotic events separating two individuals who would be predicted accurately using the total cM for segments over 5 cM methodology. And I have had a few "unknown" 2nd-3rd cousins appear in results for my several accounts with both companies.

Jim's method is what I use except on 23&Me.  In my experience every set of matches has at least one segment where a lot of people match.  I haven't figured out if this is a real effect or not.  The one  projected 4th cousin (actually 5th once or twice removed since I match to his grandmother also and at the 5th cousin level) thinks our match is something about an ancient segment which makes no sense to me. But I'm really writing about my biggest log jam (about 30 people on c 18).  EVERYONE at that site matches several other people from that segment somewhere on their DNA and usually multiple times.  The longest match to me is 15 cM; the shortest is 7.  23 projects the nearest cousin to be a 4th cousin and the most distant a Distant cousin.  I looked at the overlaps yesterday; and the failures of matching each other at that same segment could be entirely accounted for by one person's true ending being at an unfortunate place .  Occasionally one of those segments would appear to be lengthened by the IBS mechanism by about 10% but only very occasionally.   About a third of the people have reported somewhere Ashkenazi roots.  All have roots within about 300 miles of the  3 country point of Ukraine, Poland and Belarus (I'm about the furthest). The one 4th cousin had been briefly downgraded from 5th to Distant when 23 implemented their first algorithm for lowering estimates for Ashkenazi rooted people but then back upgraded to 4th. Bottom line: I now know for sure that all 30 can match each other consistently on that segment although with that 10% lower estimates of length for some. I haven't figured out yet the total number from  the rest of my matches who have multiple cousins in this list.  But all my first 4 pages (at 25/page) have cousins in 1 of 3 lists.

While you were all debating, I actually found a previously unknown confirmed third cousin today at 23andMe! :-) Sent via BlackBerry by AT&T

Oh we have a debate going. Good, I need something to do on this boring 4 1/2 hour plane fight to Baltimore.  What, no WiFi and no food? I may be late to the debate. Oh well. So now I will have to explain the fuzzy math that led me to believe that the 3rd and 4th cousins are being way over-called at both 23andMe and FTDNA. I specifically was discussing 3rd cousins because there is much more of a likelihood of actually matching a paper trail with one or more of them on a segment, and it is also harder for others to question the statistics if I am slightly off. At first I was figuring everybody only doubles themselves every generation but that may not be realistic, so let me re-figure the calculations as follows. Lets assume every couple has 4 children and every single one of those lives to have 4 more children. Maybe your ancestors had more than this, but some probably did not live to have large numbers of descendants. Most of today's families are not even this large. So the figures can vary quite a bit depending on the situation. Mormons may have more than Eastern European Holocaust survivors, for example. If there is no pedigree collapse, how many great-great grandparents do each of us have? Then figure how many great-great grandchildren these ancestors would have had. I am not counting people like Mitt Romney whose great-grandfather was a polygamist. My husband is editing this message as I am writing it BTW and putting his two cents worth in…He doesn't  do much math or genealogy but he can sure tell me about key political figures. 2 X 2 X 2 X 2 = 16 great-great grandparents How many mating couples = 8 in that particular generation Suppose each of the original 8 couples had four children per couple = 32 children 32 each have 4 = 128 grand children 128 each have 4 = 512 great grand children 512 each have 4 = 2048 great-great grand children Then you need to subtract yourself and your sibs from that number which is insignificant to calculate 3rd cousins. So I guess that comes out to be about 6.5 in a million chance of testing a third cousin instead of one in a million Americans that I estimated previously. I am assuming that many people who do genealogy are Americans also have a lot of Colonial ancestors so the majority might fit this scenario. I can't do the math unless I have a total finite potential number of testing individuals, so 313,000,000 potential testees seems reasonable. If you want to use the entire world population as potential testees, then be my guest. It only further supports my point. Maybe if a lot of your cousins are overseas the one in a million is not so far-fetched.  On the other hand, maybe you have a 1 in 100,000 chance. However, FTDNA has not tested over 100,000 for FF, have they? I might believe their predictions if they had. I figure a lot of those predicted 3rd and 4th cousins are actually 5th or greater cousins that just happened to have larger than predicted segment sizes. That goes for 23andMe as well. Yes, 23andMe has over 100,000 people in their database but how accurate are their predictions? How many times do you get a message in your Inbox from So-and-so, "a third cousin" or from " So-and-so a fourth cousin"? Don't these seems a little unrealistic? The closer ones you want to hear from don't often answer your e-mails. Well, while I am in a calculating mood, lets figure 4th cousins the same way and see how the odds change. 16 mating couples have 4 children = 64 64 children have 4 children = 256 256 grandchildren have 4 children = 1024 1024 great-grandchildren  have 4 children = 4096 4096 great-great grandchildren = 16,384 great-great-great grandchildren which translates into 5.2 in 100,000 chance of testing a 4th cousin in a finite American population of about 313,000,000.;ll6 Now for 32 mating couples 32 couples each have 4 = 128 children 128 each have 4 = 512 grandchildren 552 each have 4 = 2048 great grandchildren 2048 each have 4 = 8192 great-great grandchildren 8192 each have 4 = 32768 great-great-great grandchildren 32768 great-great-great grandchildren have 4 = 131,072 gt-gt-gt-gt grandchildren which translates into 4.2 in 10,000 chance of finding a 5th cousin. It appears 5th cousins are much more likely to test than 3rd and 4th cousins. Now for 64 mating couples 64 couples have 4 children = 256 256 children have 4 = 1024 1024 grandchildren  have 4 children = 4096 4096 great-grandchildren have 4 =  16,384 great-great grandchildren 16,384 great-great grandchildren have 4 =  65,536 gt-gt-gt grandchildren 65,536 gt-gt-gt grandchildren have 4  = 262,144 gt-gt-gt-gt grandchildren 262,144 gt-gt-gt-gt grandchildren have 4 = 1,048,576 gt-gt-gt-gt-gt grandchildren which translates into a 3.4 in 1,000 chance of testing a sixth cousin randomly. Beyond that I would think that pedigree collapse would considerably change the figures. Having more or fewer children would also change the numbers a lot. Nevertheless, there is a big difference in the number of your 3rd cousins versus the number of your sixth cousins who are likely to test at random, regardless of whether they match you on your DNA or not, and regardless of the total potential population size. Kathy J.

Regarding the chances of matching a 3rd cousin, I suspect the odds are much better than 1 in a million. We know far fewer than 1 million people have tested, yet there are multiple reports by people having confirmed 3rd cousin matches. If odds of matching a 3rd cousin were 1 in a million, than odds of matching an unknown 2nd cousin would be even more unlikely. Yet I matched a 2nd cousin at 23AndMe who I did not know (although I knew their grandmother). I have not run out all the lines, but I'm pretty confident I have over 100 2nd cousins, and hundreds of 3rd cousins. My paternal grandmother had 9 siblings who all had children. My maternal grandmother had 11 siblings. Just looking at my maternal grandmother, the second generation descendants of her first cousins are my third cousins, and she had 53 first cousins that I know of. So I probably have hundreds of third cousins just from that grandparent alone! That said, it is my experience also that predicted 3rd cousins (2nd-4th range) confirm to more distant relationships than that when confirmed. Maybe cousin marriages are partly the culprit for making these matches seem closer than they are. Steve

Dear Jim, Karen, and others, In my database of comparisons the average number of shared cMs I have seen for 4th cousins who tested at 23andMe who come from non-endogamous populations is 10.7 cMs. This is from a total of 23 comparisons of known 4th cousins. The range for these comparisons is 0 to 30 cMs. If you include the matching segments between 4 and 5 cMs (only a portion of which are likely to be IBD), then the average increases to 19.9 cMs. In my 23 comparisons of 4th cousins at 23andMe, a total of 8 (34.8%) did not meet 23andMe's criteria for a match. Sincerely, Tim Janzen -----Original Message----- From: [email protected] [mailto:[email protected]] On Behalf Of Jim Bartlett Sent: Wednesday, May 16, 2012 5:13 AM To: [email protected] Subject: Re: [AUTOSOMAL-DNA] Our atDNA matches are our cousins Karen I've tried to calculate this based on about 3380cM which we share with a parent (others with parent matches can post their experience). But at the 4th cousin level it's no feasible to divide two numbers to get a segment length, because a 4th cousin will likely show up with a segment on several chromosomes; and you have to consider that at each generation you and the match will "drift apart" in the amount shared. However, based on my experience so far one segment from a fourth cousin would be about 30cM. Perhaps the best way to learn this number is from examples of others. Jim -

Thanks Jim. Karen > > I've tried to calculate this based on about 3380cM which we share with a > parent (others with parent matches can post their experience). But at the > 4th cousin level it's no feasible to divide two numbers to get a segment > length, because a 4th cousin will likely show up with a segment on several > chromosomes; and you have to consider that at each generation you and the > match will "drift apart" in the amount shared. > > However, based on my experience so far one segment from a fourth cousin > would be about 30cM. Perhaps the best way to learn this number is from > examples of others. > > Jim - Sent from my iPhone - FaceTime! > >

> > Parents and 1st cousins - virtually 100 percent > > 2nd cousins about 99 percent > > 3rd cousins about 90 percent > > 4th cousins about 50 percent > > declining percentages beyond that > Hi Ann Can you tell me what size segments are typical of a fourth cousin, 50% ? Karen

I don't know about FF except my father-in-law only has 26 matches all 5th cousin on FTDNA which is too high for the 1 in six thousand level for 5th cousins.  It is true my wife and I have about 1 page (25) 4th cousin projections on 23 each which does strike me as overly optimistic.  But the # of projected 5th -7th cousins on 23 is about right.  People on average have about 256^2-4 cousins of level 7th or closer which is about the ratio we find at 23.  And if we allow for culture which implies being genealogically interested means having genealogically interested relatives (e.g. Ashkenazi and Mormons), the ratio is even better.

Amy, One atDNA process I'm using now might help you - it starts to pay off as you confirm some Common Ancestors. Download your atMatches (yes 5 at a time) from the Chr Browser, and collect them all into one spreadsheet. Next I sort on Longest cM and delete the shortest ones (every atMatch at FTDNA has at least one segment that is 7.7cM or larger, so this might be a good cufoff). Then sort the remaining records by Chr and Start Loc. This arrays all of your large segments from the beginning of Chr 1 to the last one on Chr 22. You can easily scan down this list to see overlapping segments. At any one place on a chromosome, you can only have a segment from a paternal ancestor and a maternal ancestor - no more. So if you already know a segment is attributed to a particular ancestor, then you have a 50/50 chance that any overlapping segments will go back to the same ancestor. You can easily query the atMatch with that segment and see if he/she also has your known cousin as a match. If they match, you can be pretty sure the 3 of you descend from the same ancestor; if they don't match, then it might be that they still have the same ancestor, but not enough matching atDNA to make the cut, OR they share an ancestor that comes from your other parent. In actual practice with overlapping segments, you can make a judgement call - segments which look very close to each other (or one is twice the other) are probably from the same ancestor. If the segments are really staggered (visualize two courses of bricks - one course representing paternal segments, one for maternal segments), they are probably from different ancestors. The dividing places between segments (think brick joints) are random and may or may not line up (comparing maternal and paternal segments) - except, of course, at the beginning and end of a chromosome. It was hard for me to work this concept out in my head, but once I understood it, it has been a powerful new tool with atDNA. Jim Bartlett On 05/16/12, Amy Martin<[email protected]> wrote: I thought it might be helpful to know the following, in response to Karen's question: My 87 year old Mom has deep roots to colonial VA and NC in each of her known lines. She has about 240 cousin matches as of today (FT DNA). She tested 10 months ago. (18) are estimated 3rd cousin matches, (3) of which we have positively confirmed through our paper trails. Of these 3rd cousins, the longest cM segments range from 58-16. She has (30) estimated 4th cousin matches, one positively confirmed through paper trail. Of these 4th cousins, the longest cM segments range from 25-12. We've also positively confirmed a few others more 'distant' as well. In addition, there are dozens more oh so close, but just beyond the evil brick wall. I'm convinced the success of more confirmations lies with 'trench research' and communication. Amy > ---------------------------------------------------------------------- > > Message: 1 > Date: Wed, 16 May 2012 19:07:05 +1000 > From: Karen Hodges <[1][email protected]> > Subject: Re: [AUTOSOMAL-DNA] Our atDNA matches are our cousins > To: [2][email protected] > Message-ID: > <CANvDYaVDBFojMt_-iYcJCLb3XjYz2E[3][email protected]> > Content-Type: text/plain; charset=ISO-8859-1 > >>> Parents and 1st cousins - virtually 100 percent >>> 2nd cousins about 99 percent >>> 3rd cousins about 90 percent >>> 4th cousins about 50 percent >>> declining percentages beyond that >> > > Hi Ann > > Can you tell me what size segments are typical of a fourth cousin, 50% ? > > Karen > > Dr Turner, > > I once constructed a spreadsheet specifically for atDNA. Empirically I found the number to be closer to 10,000 than 1,000,000. I got into a discussion with someone(Tim Jansen?) about a related subject. The thing that we both agreed on was that the order of magnitude of the number should be around 10,000. > > I abandoned the effort as trying to keep the trees synchronized required more effort than the small, 10,000 tree benefit. So, if both parties have all 10,000 ancestors and descendants of those ancestors in both of their trees with no errors, the chances of find 3rd or even 4th cousin matches should be quite good. My main tree is 40,000 plus names and includes numerous tree fragments attached to various DNA and other potential matches. A periodic "janitorial" that resolves duplicate entries finds most of my matches. > > > > Sam > ______________________________ For answers to Frequently Asked Questions about mailing lists, please see: [4]http://dgmweb.net/MailingListFAQs.html ------------------------------- To unsubscribe from the list, please send an email to A[5][email protected] with the word 'unsubscribe' without the quotes in the subject and the body of the message References 1. mailto:[email protected] 2. mailto:[email protected] 3. mailto:[email protected] 4. http://dgmweb.net/MailingListFAQs.html 5. mailto:[email protected]

Ann Sorry I didn't appreciate the limited scope of your message. Intuitively one would expect the chances of finding a match with someone who is a third cousin or closer to be very low, and I would have thought most genealogists would already know who all their third cousins are and would probably have already been in contact with them through their family history research. I would imagine however that the number of third cousins that people have will vary considerably. I have just thirteen known third cousins, but I am sure that some people will have literally thousands of third cousins and their chances of finding a random much will be much higher. As certain populations such as colonial Americans and Ashkenazi Jews are disproportionately represented in the databases, anyone with descent from these populations is likely to get more matches with third cousins by chance. I do wonder if these matching algorithms are over-optimistic. With Family Finder my dad has one predicted third cousin in America. Unfortunately this match only seems to have very limited information on her family, but it seems highly unlikely that the relationship is as close as third cousins. Debbie

I thought it might be helpful to know the following, in response to Karen's question: My 87 year old Mom has deep roots to colonial VA and NC in each of her known lines. She has about 240 cousin matches as of today (FT DNA). She tested 10 months ago. (18) are estimated 3rd cousin matches, (3) of which we have positively confirmed through our paper trails. Of these 3rd cousins, the longest cM segments range from 58-16. She has (30) estimated 4th cousin matches, one positively confirmed through paper trail. Of these 4th cousins, the longest cM segments range from 25-12. We've also positively confirmed a few others more 'distant' as well. In addition, there are dozens more oh so close, but just beyond the evil brick wall. I'm convinced the success of more confirmations lies with 'trench research' and communication. Amy > ---------------------------------------------------------------------- > > Message: 1 > Date: Wed, 16 May 2012 19:07:05 +1000 > From: Karen Hodges <[email protected]> > Subject: Re: [AUTOSOMAL-DNA] Our atDNA matches are our cousins > To: [email protected] > Message-ID: > <[email protected]om> > Content-Type: text/plain; charset=ISO-8859-1 > >>> Parents and 1st cousins - virtually 100 percent >>> 2nd cousins about 99 percent >>> 3rd cousins about 90 percent >>> 4th cousins about 50 percent >>> declining percentages beyond that >> > > Hi Ann > > Can you tell me what size segments are typical of a fourth cousin, 50% ? > > Karen > > Dr Turner, > > I once constructed a spreadsheet specifically for atDNA. Empirically I found the number to be closer to 10,000 than 1,000,000. I got into a discussion with someone(Tim Jansen?) about a related subject. The thing that we both agreed on was that the order of magnitude of the number should be around 10,000. > > I abandoned the effort as trying to keep the trees synchronized required more effort than the small, 10,000 tree benefit. So, if both parties have all 10,000 ancestors and descendants of those ancestors in both of their trees with no errors, the chances of find 3rd or even 4th cousin matches should be quite good. My main tree is 40,000 plus names and includes numerous tree fragments attached to various DNA and other potential matches. A periodic "janitorial" that resolves duplicate entries finds most of my matches. > > > > Sam >

I'd like to toss in my 1.7 cents.   1. I got the impression that the expression something like "even if 300 3rd cousins" was meant to indicate a large number. I would like to point out what a SMALL number this really is. Assume all your ancestors, and everyone in your family had exactly 2 children. Further assume that all of your generation, and all of the generation above you are still alive. In that case you would have living:   A. Sibling: 1 B. Aunt/Uncle: 2 C. First Cousins: 4 D. 1c1r: 8 E. 2nd Cousins: 16 F. 2c1r: 32 G. 3rd cousins: 64   So that's a total of 127 people that you could match - with just 2 kids per couple. If you move that number to 3, you get 378. With just 3 kids per couple, you have very nearly 300 potential matches just with 3rd cousins and 2c1r alone.   Now imagine how many you might have with large families. With 7 kids per, you'd have 300+ SECOND cousins alone, and that's from just ONE set of your sets of great grandparents.   Geometric progressions get big fast. I think people would be astonished to learn how many 3rd cousins they really had, particularly if they are of the generation born prior to the 2nd World War. In my generation, maybe not so much. I have a whopping 3 first cousins. And if you add together all the children of all my siblings and all my first cousins, there are 7 total in that next generation. Do that same calculation with any of my grandparents, and you get a big number.   2. I think the probability of match is like the birthday paradox Ann suggested. And it can be imagined this way...take the 127 matches I outlined in point 1 above, and consider the matching not from your point-of-view, but from the point-of-view of your 3rd cousin #64. By counting the number of cousins you have and comparing it to the number in the database, you are essentially asking what is the chance that ANY ONE (or pair) of your ancestors is the same as the (let's say pair of 2-greats-grandparents) PARTICULAR ONE (or pair) that you have outlined in the common descent tree. But that isn't the only way you could match. You could match any one of your ancestors to ANY ONE of his ancestors. Look at your 3rd cousin's pedigree. You'll see your common ancestor in there, it is true, but what you'll see mostly is just blank space. And the matching could also come from anywhere in that great unknown.   So then it comes back to the point of populations. In a sufficiently isolated and immobile population, the probability of match tends toward certainty.   Gregg Bonner

I think the answer below misses the point. "... am I likely to randomly match any true 3rd cousins?" was the question. I take this to mean that I know who my third cousin is; we both take the same atDNA test; then what are the odds that we will show up as a match? The odds do not depend on how many third cousins you have in your data base, because you already know the two of you are 3rd cousins! The odds DO depend on which and how much atDNA each of you got from your 2G-grandparents. One factor is that the total amount of atDNA is roughly half with each generation. You get full (100%) coverage over all of your 22 autosomal chromosomes from each parent. So your coverage would be 1/2 coverage from each grandparent, 1/4 from each G-gp and 1/8 coverage from each 2G-grandparent. It's random, but not uniform - the segments your and your 3rd cousin get could be anywhere. And certainly they are different. The key to ask is: are there any 7.7cM or larger, identical, segments that were passed down from the 2G-grandparent to both you and your 3rd cousin? FTDNA says that 90 percent of the time there are. In other words if 10 of your known 3rd cousins were to take the same atDNA test as you, 9 of them would probably match you, and 1 of them would not. (this all presumes all the players are in fact biological 3rd cousins). So if you go back 4 generations, are you likely to randomly match any true 3rd cousins? YES!! 9 out of 10 times they will show up as an atDNA match; 9 out of 10 real 3rd cousins who take the Family Finder atDNA test will be listed on your FTDNA Family Finder page. Will any of these 10 3rd cousins be included in your own version of your family family Tree (in your genealogy database)? Probably not for most of us (for one thing, it's very difficult to trace living people whom you don't already know). But the good news is you don't have to have these 10 3rd cousins in your Tree to use the atDNA tool. You don't need to have any 3rd cousins in your tree (whether it's 10,000 or a million). All you need to have is the 16 2G-grandparents! Well, actually the other 10 3rd cousins would also need to know their 16 2G-grandparents, too. I don't think that's very much to ask of most genealogists - to have their 16 2G-grandparents documented... Granted that adoptees don't; and some beginners don't, and a few folks have brick walls early in their trees and don't know ALL 16 2G-grandparents. But I think most of our atMatches will know most of their 16 2G-grandparents. [this is one full page of a 5-gen Ancestry pedigree chart] So I think it is reasonable to assume that 7 or 8 times out of 10 most of us will determine the Common Ancestors with our true 3rd cousins. Two problems: 1. Not many of our true 3rd cousins take the atDNA test. 2. FTDNA over estimates, IMO, and in my experience. Most of the matches that they estimate at 2-4th cousins will wind up being 5th or 6th cousins - IMO. Jim Bartlett On 05/16/12, Sam Eaton<[email protected]> wrote: Dr Turner, I once constructed a spreadsheet specifically for atDNA. Empirically I found the number to be closer to 10,000 than 1,000,000. I got into a discussion with someone(Tim Jansen?) about a related subject. The thing that we both agreed on was that the order of magnitude of the number should be around 10,000. I abandoned the effort as trying to keep the trees synchronized required more effort than the small, 10,000 tree benefit. So, if both parties have all 10,000 ancestors and descendants of those ancestors in both of their trees with no errors, the chances of find 3rd or even 4th cousin matches should be quite good. My main tree is 40,000 plus names and includes numerous tree fragments attached to various DNA and other potential matches. A periodic "janitorial" that resolves duplicate entries finds most of my matches. Sam ------------------------------------------------ "5. Even if I go back 4 generations, am I likely to randomly match any true 3rd cousins? NO!! because you need a database of over a million to find that one in a million 3rd cousin. Even if you have more than 300 3rd cousins alive today, there are over 300 million people living in the U.S. Does FTDNA test that many people? No."