I was recently asked if I’d thought about this question. The quick answer is YES – the answer to this question is at the core of my belief that genetic genealogy is valid out to 9 generations back. And I think this question is really two questions: one about the Triangulated Groups (TGs) themselves; and one about the Matches with shared DNA segments within each TG.
How far back do our TGs go?
Using a 7cM threshold for shared DNA segments, I’ve documented 372 TGs, covering over 98% of my DNA. These TGs have natural breaks [recombination crossover points] between them. These TGs represent actual DNA segments, on my chromosomes, which are from my Ancestors down to a parent to me. So how far back do they probably go?
The number of segments we have at each generation of our ancestors is fairly easy to estimate. Using a female to make it easier, she gets 46 segments from her two parents – in the form of 46 chromosomes. Pretty big segments… Using the average recombination rate of 34 crossovers per genome (per parent), she would get 68 additional segments one generation back. In other words she would have a total of 46+68=114 segments from her grandparents. And she would get 114+68=182 segments from her Great grandparents. Here is a handy table I made up for my reference:
This table starts with me at the bottom and shows the generations back, the number of Ancestors at each generation back, the generic name of those Ancestors, the relationship of my cousins who share a Common Ancestor with me at that level, the calculated percentage and cM amount of DNA I got from each of those Ancestors (at any given number of generations back), the calculated average number of segments in my DNA from all the Ancestors in any given generation, the average cMs per TG; and in the last two columns the average and range of cMs collected in Blaine’s cM study. The first column is just for a very rough estimate of the birth year of my Ancestors at any given generation (it helps me).
Highlighted in yellow is the 386 segments expected (roughly) from my 3xG grandparents. That’s roughly the same as my 372 TGs. So I expect some kind of distribution curve around that point. Matches who share the full DNA segment represented by a TG would probably be 4th cousins (4C). Due to the random nature of DNA, I expect a range from 2C to 7C or 8C. My TGs range in size from a few just over 7cM to some around 50cM – it all depends on several variables.
Another aspect of this discussion has to do with what I call “sticky” segments. Per the Table above at 5 generations back we would see 386 segments – or 386 TGs – of about 18cM each. But going back one more generation – one more round of 68 crossover points would result in 454 segments. This means that 64 of the 386 segments were subdivided, and 322 segments were not! This means that 322 segments (TGs) were passed down intact (no recombination). The effect of this is that many TGs will persist, at the same size, for several generations. We could well see the same size TG from a 6xG grandparent to a 5xG to a 4xG to a 3xG grandparent. So it would be possible for a 7C, 6C, 5C and 4C to all share the full size DNA segment represented by the TG. Clearly the probabilities of that decrease as the cousinship increases.
Bottom line from my experience: I think we’ll find most of our TGs to be within a genealogical time frame of, say, 9 or 10 generations. And there is always the opportunity for closer cousins to share a DNA segment within any of our TGs.
How far back do the Matches go?
This is a different, but related, question. The above discussion was all about the full DNA segment represented by a TG. Most of our Matches in a TG will not share the full DNA segment. They overlap us or are wholly included within the TG segment. For example, the Matches in 20cM TG can range from sharing 7cM up to 20 cM. And, in fact, some of our closer cousins may share 35cM and span across more than one TG. It’s very random. However, to the point of the question – many of our Matches who share, say, 7 to 15cM may well be cousins beyond the Ancestors who passed down the full TG. To be sure, the Common Ancestors in this case would be ancestral to the TG Ancestor, but it could be 10, 20, or more generations back.
Bottom line: Matches in a TG are limited to a narrow range of your Ancestors, but they are not limited by how close or how distant they could be. And Matches who share small segments may well be beyond a genealogical timeframe; but some will be within a genealogical timeframe. Witness the Ancestry ThruLines Common Ancestors down to 6cM.
Summary: I think most TGs will be within a genealogical timeframe (using a 7cM threshold for shared DNA segments). The Matches in a TG will range from close Matches, out to Matches on the fringes of our genealogy and on out to Matches who will be beyond our genealogy.
[19H] Segment-ology: How Many TGs From Distant Ancestors? By Jim Bartlett 20191217
segment-ology 
Pingback: Friday's Family History Finds | Empty Branches on the Family Tree
I like how your table is similar to Graham’s Genetic vs Genealogical ancestors in https://gcbias.org/2017/12/19/1628/
LikeLike
Kuba, the list of generations, cousinships, Ancestor relationships are the same for everyone. It’s an easy table to make up.
LikeLike
Jim how does IBS fit into this clustering approach? Obviously a good number of matches at 7cM are IBS… so how do we avoid being thrown off the CA trail because of IBS in clustering?
LikeLike
Douglas, In my experience and analysis, many (I don’t know if all) IBS Matches do not Cluster or Triangulate. If they do, then look to see if they are in agreement with others in the Cluster. The odd-balls usually stick out like sore thumbs.
LikeLike
How does double cousins or even more endogamy complicate things?
LikeLike
I have some double cousins, and they are difficult to untangle. However, I’m finding as the Clusters divide they go up just one ancestral line which is a big help. And some of the additional Matches may be close ones, which also helps. I’m not there yet, but I think this process can involve all of my 119,000 AncestryDNA Matches.
LikeLike
I really like your table as it is something I have been rolling around in my mind for awhile. That we likely cannot have more than 1,000 TG’s because that makes them all be 7cM or smaller. There is only so much non-overlapping DNA we can create. (although we have to be careful to tie that number to a number of generations back because, as you state, your DNA is sporadic with different lines due to the inheritance pattern. And so some segments may be from your grandparent(s) that are 10x or more generations back.)
So all the more impressive at how many TG’s you already have compared to that upper limit. Keep up the great push of our general understanding.
What this continues to point out to me though is the importance of testing all of your siblings if the parents cannot be tested. Because the siblings carry ~50% different DNA than you do. And so the chance they carry more TG’s that you don’t is pretty good. And ones to distant relations that you do not share with them. And so this increases the pool of TG’s possible. Although such a combined pool is not to exceed 2x or 2,000 TG’s as eventually, with enough siblings tested, you are simply recreating both parents DNA and the limit is set by what they each contributed.
LikeLiked by 1 person
Randy, Thanks for your feedback. You raise several different points.
I really believe that with a 7cM threshold, most will only see 350-400 TGs. Since TGs are formed with overlapping shared DNA segments, we probably won’t have any at the 7cM level, and the average will be about 18cM. If, and it’s a big IF, we could find the most distant ancestors for each of our 375 TGs, THEN we could lower the shared DNA segment threshold to, say, 5cM. This would introduce a lot of false segments, which would be harder to determine. It would also subdivide some of the 375 TGs, and we’d get more. I cannot begin to estimate how we would ever get to 1,000 TGs – and even if we could, most of us would be unable to determine the Common Ancestors that far back. My 372 TGs basically cover all of my DNA. So the only way for me to get more is to subdivide some of the existing TGs…
Since you and a sibling share about 50% of your DNA, a sibling would have many Matches that you don’t have and would create a different mix of TGs. Each person, including a sibling, will have their own chromosome map, but rarely would they any identical TG. Try just one TG with a sibling, it will usually be shifted some amount. Your TGs and chromosome map are unique to you, and siblings and close relatives will overlap (but not exactly match) your TGs. By using multiple siblings, you can create overlaping TGs for a parent or other Ancestor, but you’ll never get their whole genome. Hope this helps. Jim
LikeLike
As an only child, I have no siblings to test, so I sought out first cousins. My father was one of four siblings, all four of them deceased. One child of each person in my father’s generation has DNA tested, thus, I feel like we have a reasonably broad spectrum of our mutual grandparents’ DNA to compare and triangulate. We work with what we have . . .
LikeLike
Gail,
A good plan. I have one parent and one uncle, but no true first cousins. But I’m still swamped with data.
LikeLike
Hi Jim,
My OCD kicked in when I saw this was the second “19G” in your outline.
LikeLike
Chris, I can’t find the other one… I’m trying hard to keep them straight.
LikeLike
Hi again,
I have “Unified Theory of Genetic Genealogy” as the first 19G. They came out pretty close together, at least in my email.
LikeLike
You are correct – thanks – I just fixed it.
Jim Bartlett Sent from my iPhone DNA blog: http://www.segmentology.org
>
LikeLike