RootsChat.Com

Thrulines does not consider clusters of DNA matches. It just looks to see if any presumed ancestors of yours is a presumed ancestor in your matches' trees.

My grandmother was born out of wedlock, and Thrulines suggested the entire line of her biological mother's later husband (not her father) was her own. Simply because he was from the same area, and among my matches happened to be ones who shared parts of his ancestry - enough to suggest that all of his lines and he himself was my ancestor. But there was no intersection of those matches. The alleged common ancestors were pulled out of multiple matches' trees who weren't related to each other.

This is the big issue for newcomers, there is a process and quite a bit of knowledge to doing this properly. It is easy for newcomers to just pick something out of Thrulines or their matches, where there isn't really a connection.

In particular if most of your ancestry is in a very concentrated geographical area, this can seriously muddy the intersections/overlaps of the matches. I have so many matches that come up with shared matches from multiple clusters, from maternal and paternal sides. I even suspected that one of my lines had infidelity on it. But it's just that almost every DNA match from that line, was also a DNA match from another line. I also have a few closer matches who are matches on both lines, and it's been those that have been key for unsorting those intertwined lines.

Does anyone know when the typical bumps for new matches are. I got a steady clip of new matches from about the end of Jan to the end of Apr, from people getting kits for Christmas. Clearly very few coming in now. I think Mothers' and Fathers' Days are the other periods you get a boost?

Has anybody seen any sign's yet that ancestry is actually delivering this autoclusters feature?

Richard

Been keeping an eye out for it, but not seen anything. It's certainly a much needed feature. And as usual Ancestry are spending months, years rolling out something that should be finely polished within a month (given the scale of their data more time is needd).

Just sharing a few bits and pieces after having worked on a few of my various unaccounted for DNA clusters, which all have primary matches with lower cM.

This has confirmed for me something I already suspected - the smaller the shared cM of the largest cM match in a cluster the more likely it is to be a more distant relation than you'd expect.

If you have a cluster of matches where the largest match is, say, 46cM, the relationship is more likely to closer to the most likely relationship. For 46cM: 4C, H3C, 3C1R.

The smaller a largest match in a cluster is, the more distant the relationship is likely to be. If you have a cluster when the largest match is 23cM, the most likely relationship for that amount of cM is about 5th cousin. However, when part of a cluster with no moderately sized matches, say 40cM+, it seems the relationship will be to the more distant end of what you could expect from shared cM.

This makes sense, as when dealing with matches with whom you share more distant MRCAs, the chance of inheriting large enough segments to match with other closer matches decreases. We will all inherit some small chunks of DNA that could be used to triangulate with 7th-8th cousins, but very few descendants will have just the right amount of the right segment to become a cluster of matches.

To illustrate, a tree:



You may have a cluster of DNA matches with common ancestry from Samuel Ward (1788) and Charlotte Payne. Let's say 40 matches, and on average your 5th cousins. The chances of you and those 40 matches inheriting enough DNA to also match with descendants of Samuel Ward (1726/7) and Jane Williams is quite low. Some descendants may inherit enough, but not many.

But you may be one of the ones who does happen to inherit enough to have matches with the descendants of Samuel Ward (1726/7) and Jane Williams. And it's likely that few if any of the descendants of Samuel Ward (1788) and Charlotte Payne will share the matches. So you will probably have a separate cluster of matches that has no apparent bearing to any of your lines; and in turn making it more difficult to identify MRCA.

***

As one case, I had a cluster of matches where I found nine matches, descendants of Sara Humphrey (1810) m John Price (1799). The only clue as to where it may relate to my tree is one of my 4th cousin matches, descendants of Anna Humphreys (1811) m Richard Evans (1806) matches two people in this "Sara Humphrey" cluster.

Sara Humphrey (1810) was born in a parish near my line, while her husband was from about 30 miles away. So I focused on her ancestry. Sara was an immigrant to the US, but her children had distinctive names and she emigrated in 1856, thus her place of birth was listed in the 1851 census. Her date of birth could also be extrapolated from her headstone, which listed her age at death with months and days. A baptism was found 1 year and 7 days after the listed date of birth. Likely a typical error. And someone has uploaded a hymn book owned by her, in which she listed her mother as Elizabeth Williams (turned out to be a surname from a second marriage). So it was not too difficult to be sure of having the correct baptism.

With the amount of shared cM with the matches with identified common ancestry (between 10 and 22), I was expecting the parents of Sara Humphreys to also be my ancestors. But they were not. Her father was a Thomas Humphreys, born around 1765, and his wife Elizabeth Morgans (later Williams). It turns out the MRCA between all matches was my ancestor Lewis Humphreys, born 1731, father of Thomas Humphreys just mentioned.

One of the matches at 21cM was a 7th cousin. I wasn't expecting so much to be inherited over such a number of generations. But obviously it can happen.

***

I have a lot of clusters like this, when I've found common ancestors among the matches and the matches share low 20-something cM at most. In such cases I'd assumed I was closely related to one of the common ancestors among the matches.

Let's say you have matches with common ancestors Thomas Perry (1790) and Sarah Turner of Tipton, Staffordshire, and at most 22cM shared. There are no matches identified with ancestors from previous generations. It's quite likely that your MRCA with these matches is rather a grandparent of Thomas Perry (1790) and Sarah Turner, rather than a parent.

So when dealing with such clusters, you need to have a good solid tree of them back to about the 1730s. Between your various clusters, you may find some that indicate common ancestry.

***

One thing that might be able to help with this is tests of cousins. If you have access to them, you can use profile pages on Ancestry (accessible by clicking on the match's name on the match page). At the top of profile pages is a drop down, which you can use to toggle between different DNA tests.

The shared matches page only shows some shared matches - not sure on what parameters (?). You'll generally have far more matches in common that shown there, including some that cluster for you both.