BLUF (Bottom Line Up Front) – you may already know this: The average cMs of all your Matches in a Floating Branch back to the Common Ancestor (CA) is a strong clue to about your relationship (use judgment to pick the CA and cull out outliers); and the highest average among the children of that generation would often point to the child you descend from.
This is an insight based on collected data. First, we’ll set up the situation:
A Floating Branch is a group of your Matches who all descend from a CA who is not one of your known Ancestors. Typically, a Floating Branch can contain a lot of Matches (I’ve seen 20, 50, over 100). It’s frustrating – their CA is almost certainly your CA, but the link eludes you. It’s probably one of your Brick Walls, or beyond.
Brick Walls – to me, there are at least two types of Brick Walls – Ancestors in “plain view” and Ancestors “hiding under a rock” (or abducted by aliens). The “in plain view” Brick Walls are out there, with records, but we just haven’t made the genealogy connection. But, in my opinion, most of our hard Brick Walls persist for a reason: there is little to no information about them. My favorite example is a tic mark on the 1840 census (perhaps a young man who impregnated a young woman; and went “west” in 1849 and died along the way). The parents of the tic mark are known, but no other info on the son. There are many other reasons why a person may not be in the records. If they had left records, they probably wouldn’t wind up as Brick Walls.
Think about these two concepts (Floating Branch & Brick Wall)… To my thinking, they should “link up”. But how?
Well… After we’ve done an exhaustive review of the Floating Branch (fill out the descendants of the CA; ethnicity; geography; records, etc.), we probably have at least two things we can do with the collected DNA data. One is to average the cMs of all the Matches (except obvious outliers) and use the Shared cM Project to make an educated guess at the probable relationship (and generation) of our connection to the Branch. Ask: what Brick Walls do we have at that generation? Yes, it gets harder and harder as we go back in time.
However, the gist of this post is to figure out where we tie into the Floating Branch. It’s where the highest cMs are! Probably obvious to Segmentologists…
Take a known Ancestor with, say, 8 children; we would usually see Matches from most of the children having one average cM amount, while the Matches descending from the child who is our Ancestor will have a higher cM average. Again, easier to detect with close relationships. In theory, the difference is a factor of 4. If most of the Matches are 2C averaging about 229cM, we find the 1C Matches with the child who is our Ancestor to be about 866cM. See this table for different generations:
Matches with most of the children vs Matches with just your Ancestor child:
Great grandparents 229cM for 2C vs 866cM for 1C – pretty close to 1:4
2xG grandparents 73cM for 3C vs 229cM for 2C – a little more that 1:2
3xG grandparents 35cM for 4C vs 73cM for 3C – about 1:2
4xG grandparents 25cM for 5C vs 35cM for 4C – still more, on average
5xG grandparents 18cM for 6C vs 25cM for 5C – still more, on average
In other words, the floating branch should have an array of Match cMs just like any of your known CAs has any array of descendant cousins – roughly like the Shared cM table shows.
The point is that if you had a Floating Branch, and didn’t know where you tied in, averaging the cMs for each child at the same generational level, might provide a clue. I have two cases where the probable link sticks out like a sore thumb; and 2 others where the averages are not as clear as I thought they should be…
I’ve had this thought before – see here; and here. The more data (Matches), the better.
I’d be grateful for feedback from anyone who works with Floating Branches and has enough data to indicate whether one path sticks out like a sore thumb… And/or if you have any thoughts, pro or con, about this kind of analysis to squeeze a little more out of a Floating Branch…
[22DM] Segment-ology: Where Is Your Ancestor in a Floating Branch? By Jim Bartlett 20260629
segment-ology 
Hi Jim, I have added a few “floating branches” to my big DNA research tree over the years. The most recent floating branch includes 77 Thrulines that are generated when I attach the potential CA couple to my tree temporarily. These are matches to my maternal great aunt (born in 1918). I only have 3 Thrulines for this couple through my mother and 20 to my 3/4 uncle (yes, I have an interesting tree). The Thrulines indicate DNA matches to my great aunt through 7 of this CA couple’s children and misses 2 of them but there’s something wonky with them as they’re both called Sophia and were born within a couple of years of each other and married different men. For the large part I’ve done a “quick and dirty” tree using other people’s trees and any mostly obviously correct records so there is the potential for errors of course. When I look at each of the Thrulines for each child of this CA couple I don’t see an obvious group with higher cM values. -One son has a match sharing 35 cM with my great aunt and his descendants share from 6-35 cM with my tester. This 35 cM is one segment. -Another has a range from 11-36 cM with the 36 cM being one segment weighted down from 38 cM. -Another has a range from 11-33 cM with the 33 cM being one segment weighted down from 40 cM. -There is another one sharing 44 cM with my great aunt but this daughter married a man with the same surname as the wife in the CA couple so it’s possible that there’s a second family connection there leading to more DNA being shared. In this situation this daughter’s descendants with the larger cM shared, both share one 35 cM segt with my tester and a smaller segment with at least one of the segments weighted down by the Ancestry algorithm. -The descendants of the other children of the CA couple share smaller segments. Based on these numbers, in this floating branch I would not say that any one child’s descendants “stand out like a sore thumb” unfortunately. I’m not sure if that in itself should be pointing to something I should be picking up on? So then, how to fit this into my tree? Ancestry has indicated that I have DNA matches going back to all of my great aunt’s family lines except for one. For this line for some reason the Thruline is indicating the same names and years but showing my “known” direct ancestors as “potential” ancestors. I don’t know what to make of that. Almost all of the matches from these Thrulines indicate they’re from my great aunt’s Maternal side. If correct, this of course narrows things down. Unfortunately, I’ve found a couple of these matches on other platforms and I’ve identified them as Paternal matches to my great aunt! I have run her data through the GEDmatch “Are my parents related?” tool and they’re not. I do have a couple of DNA segments on different chromosomes that seem to share both maternal and paternal ancestors and match each other but I haven’t found a commonality beyond those ancestor couples yet. I’ve looked at each group of Thrulines I have and it seems that all of them also go back to the next generation where there are more Thrulines through several of the children of that generation. The father on the wife’s side of this CA couple has 111 Thrulines through 12 of his children with 45 through the woman in the CA couple. I suspect that this means that my DNA connection is probably back further than this CA couple. Soooo, I’ve now gone back another generation:The CA husband’s father’s 111 Thrulines are over diverse ranges again for each of his children including 6-24 cM for his first son but then 8-9 cM for his second son. The strongest connections are from his two eldest daughters who both have descendants who share a 35 cM segment weighted down to 15 cM and 28 cM. My main issue is that their mother would have been in her mid to late 50s when they were born so these are more likely to be grand-daughters than daughters or possibly from a second younger wife. As far as I can tell there are no significant ethnicity differences that stand out. They all seem to have lived around the PA, NY, VA, TN areas in the 1700s. I’ve been more successful finding land records to sort out these people than anything (Big shout-out to Family Search Full-Text Search!). As the CA couple were born in the 1770s there aren’t really many census or other records to help distinguish between the families. Based on some things I’ve read it’s possible that the family were Quakers in PA and of the few places the CA couple or their ancestors could fit into my tree, my main one is a potential Quaker family too. The reason they could possibly fit into that line is because I don’t have records confirming it! I’ve sort of worked through this as I typed it so it’s not well-presented I’m afraid. I do see though that for my purposes I should probably tackle some old Quaker records to see if I can clear up some tree issues, especially some potential duplicates. I’ve also realized that maybe I should scrutinize the unweighted segments more than I perhaps have until now. It’s scary to think but I feel like I need to almost start over in assigning Matl and Patl sides for my great aunt as something’s not quite right. I hope this all makes sense to you and it’s in some way helpful in your quest to sort out these segments!
Carolyn Jensen PS I’d be happy to share further information should you think it could be helpful to you.
LikeLiked by 1 person
Carolyn – I’ve read and thought about your story for a while. My summary thought is that the process I laid out is for a “normal” ancestry. One where the Ancestors don’t relate in a genealogy timeframe. I have one place where my two 3X great grandparents are 1C. Clustering doesn’t work well there but using Triangulated segments is better. You might want to try something like Banyan or WATO, which evaluate a lot of cM links and can provide some statistical probabilities. Or contine with FS AI and/or Quaker records using standard genealogy methods. Jim
LikeLike
I have a very large number of floating branches or unlinked family trees (with people connected to one another but not my tree) because I have been building trees for clusters and shared matches for many years. So, I certainly have trees with which to explore this idea on the mathematics of where the connection lines.I suspect many are fairly distant, but I have a few groupings in mind that I think might indicate a discoverable connection within the time frame of the available records for the region. I am especially interested when there is sharing across two or more floating trees, along with triangulation across trees for the few people for whom segment information is known, and the MRCA for each of the related floating trees are on different continents.
LikeLiked by 1 person
tetleyt – It looks like you have some good floating branches to try this on. This process is not as precise as others, but may be helpful – particularly if you have Matches with good Trees. Please let us know how it works out for you. Jim
LikeLike
This scenario of fitting one person into a tree is exactly what WATO is for. Build the floating tree, enter the shared cM amounts, and let WATO work out where you might fit in the tree. It’s much more powerful than using average amounts because the possible relationships for all matches are considered.
LikeLike
armillard I agree with you – I am a big fan of DNA Painter. My blogpost was for several reasons. Foremost was to outline the concept. In addition, not everyone uses DNA Painter. And, on a personal note as a long time genealogist, I prefer to look at my data – the Matches, their Trees, their shared cMs – and try to figure it out. Often this leads to other insights. As we know, the correct answer is not always the highest calculated result. No matter how we arrive at a good guess, there is still the genealogy work that has to be done. Jim
LikeLike
Ciao Jim ma se ho delle corrispondenze dna su mhyeritage tra 8cM fino ha 9,4cM e ho corrispondenze dna tra 13,7cM e fino ha 23,3cM si può vedere che anno ho generazione può essere? E perché mhyeritage mette che sono cugini tra 4-5 gradi?
LikeLike
Kevin, With such small segments, it’s very difficult to figure out years from just the cMs – there is a very broad range. A better approach is to use the Matches’ Trees to find their Common Ancestor – then you will have a much better foundation to estimate the distance back from you. Ignore the MyHeritage estimate of relationship! Jim
LikeLike
Che vuol dire immagino che il mio collegamento e con il figlio del cM più alto? Può spiegarmi bene grazie
LikeLike
Kevin, Please use the DNA Painter tool to get an idea of the full range of possibilities. https://dnapainter.com/tools/sharedcmv4-beta/30 It’s really a guessing game – there is no way to accurately guess. The best method is to look at the Trees of the Matches, and see how many generations they are from a probable Common Ancestor among them. Try to determine if you are roughly the same age as the Matches or a generation older or younger. In other words: try to draw the picture of their Tree and make a judgment of where you fit in. You can also try the WATO tool at DNA Painter. Jim
LikeLike
Si ma queste 19 corrispondenze sono di nazionalità diverse esempio 6 kosovari 4 bosniaci 2 ungheresi 1 greco bulgaro polacca lituana e Romania io unico italiano sud tra loro sono chi alcuni rom alcuni misti rom e altri qualcuno balcano
LikeLike
Si Jimi e mhyeritage dice cugini tra 4-5 gradi ma io con 7 corrispondenze ho un solo segmento tra 8cM e 9,4cM e poi con altre 12 corrispondenze ho tra 13,7cM fino ha 23,3cM tutte triangolare sullo stesso segmento.dovrei contare i cM ho il cM segmento triangolato per vedere dove si potrebbe essere la generazione?
LikeLike
Keven, to use the method in the blogpost, you need to build the Tree for the Matches back to *their* Common Ancestor. Also look at MyHeritage for other Matches who are close to the ones you have already identified and add them to their Tree. I would guess that your link to that line would be through the child with the higher cMs…
LikeLike
When I realised that I had no matches with descendants of my great-grandfather Robert’s 11 uncles and aunts, I started looking for his biological father. Analysing matches shared between me and my relevant cousins, it took 30 minutes to identify the CA, William and Rebecca from West Keal in Lincolnshire. The dates meant that one of their sons was the likely candidate. They had eight sons, and I have DNA matches with descendants of seven. It took a few weeks to carefully research these men, with no luck finding a connection. I then researched an illegitimate son, George, who was born to Rebecca prior to her marriage. George used his mother’s maiden name all his life, and was apparently raised by her parents. It was one of those Eureka moments when I discovered George and his family living in the same street as Robert’s parents in the 1861 census at Ancaster, Lincs. DNA matches proved that George was the biological son of William and Rebecca. In this case I did not have to rely on segments, as proximity provided the clue.
LikeLiked by 2 people
Jean, A great research story. Like you, I had a friend with NPE father – they lived in NYC. Narrowed it down to one family in Philadelphia with 5 sons – they all went off to WWII, and 4 returned to PA; one went to NYC – lived a few blocks aways from my friend’s family… BINGO – and subsequent targeted DNA testing confirmed it. Jim
LikeLike
Sometimes it might be someone from YOUR tree who impregnated a woman in the floating branch and did a flit.
LikeLiked by 1 person
Jean, A very good point. Yes, follow ALL options. I have a Match who has determined one of my Y-DNA BARTLETT relatives, during the Civil War, impregnated her ancestor… In another case, there is a different surname with 3 Y-DNA tester who exactly match my Y-DNA (in Big-Y; with unusual E-V13 haplogroup). They wanted to know where my line split off of theirs… I asked how far back they had their line – answer: late 1700s. I replied mine goes back to Thomas BARTLETT b c1703 with matching DNA from 4 of his sons – where did *their* line split off of mine? The important thing is that DNA is helping us to figure these things out. Jim
LikeLiked by 1 person