Super Bowl Sunday 2019 started like most of our Sundays. Felice got her breakfast in bed, the kids all got pancakes, and breakfast was complete. I sat down at the computer with a nice cup of coffee hoping to kick the mini-migraine that was resisting drugs and enjoy a few hours of genealogy.
As I sat down that morning, for some reason a Twitter post we’d made 7 months earlier about Felice’s mother “Susan” popped in my mind. Susan matched her 1st Cousin “Charles” with 2122 cM…enough that he was almost a full sibling. It was a head-scratcher. We’d tweeted out our confusion, and a follower explained to us that it might mean he was a “3/4 siblings”. 3/4 siblings are where the same person parents children by two siblings, for example, when one man has children with two women who are sisters. But the tweet came during a busy time and it fell out of our minds…until this morning when it hit like a lightning bolt.
Susan’s paternity was THE big “brick wall” of our family history research. The man listed on her birth certificate, Roger Homes, was likely not her father. Family history held that Susan’s mom Dealia had at least 1 of her 2 other children with Roger, but Roger was on Susan’s birth certificate because Dealia’s father him on there. He didn’t want his Grand daughter’s Father left blank. Family interviews had given us a couple of leads on Susan’s father, but finding “Big Ed” from a neighboring town in Mississippi seemed like a significant long shot. DNA was always our best hope to solve this mystery.
Going into that Super Sunday we had recently finished our series about the tools we used to go from a handful of Ancestry DNA matches to connecting them in a tree. In the “Casting a Wide Net” series (Link) we took a group of over 5000 matches to Susan that were shared between themselves and built them into a mirror tree. Ultimately we mapped out 17 of those DNA kits to each other and identified the MRCA for them and Susa.
In of our research, we’d noted Susan’s maternal cousin Charles also matched the 17. This led us to focus on her mom’s side of the tree to find the link, but the evidence hadn’t lined up with that theory. We ended the series without being able to establish the direct link between the MCRA and Susan.
As the computer fired up that morning, the tweet, the MCRA, and the unknown father all slammed together at once: What if the maternal cousin wasn’t only a cousin? What if Charles’ father was also Susan’s, and what if the 17 matches were on their father’s line!
We’d never built out Charles’ father’s line because he was an Uncle who didn’t feed much information into our line. We’d added his parents, so knew their names and not much else. We’d interviewed Felice’s Aunt “Ann” and she explained about how she’d married Luther White at 13 years old and almost immediately kicked him out. Despite that, he would still go on to father each of her 10 children while Luther’s parents supported her and her children, including giving them a place to live.
I was shaking a little as I opened up Ancestry and started to build out the Father’s line. This theory perfectly clicked together, but if we were right we were about to be swimming in deep waters. It was always our hope to breakdown the brick wall of Susan’s paternity and to help her fill out the picture of her life. We envisioned a happy moment where we put to bed a lifelong secret and expanded our family tree. Now, this was taking a very sudden turn and we were likely unearthing a painful family secret.
All of this before my first cup of coffee on a Sunday…and little did we know at the time how deep this would go.
In early 2018 we made a series of posts on how to use the multiple “Shared Matches” in AncestryDNA to narrow down the DNA line that connects you to them. The challenge was that often they have no trees, or small trees that don’t come anywhere close to matching your (much more complete!) tree.
This strategy was a way to use mirror trees to match them to themselves, which should indicate a Most Recent Common Ancestor for them, and in all likelihood to be your MCRA as well. For this series we broke down a large set of matches (5000+) to Felice’s mother, to try and establish her first DNA link outside of the immediate family.
There were all of the challenges we all face with African American genealogy (fewer family histories to draw off of, smaller trees, difficulty with 3x/4xGGP’s due to the “1870 Wall”, etc.), and in this series we found the MCRA…but we failed to find the link between them and our family. However, about a year later we broke through that wall, and we’ll be following up on that shortly. In the meantime, here’s the complete series in one page:
It’s rare that we can spot trends emerging, but we’re going to take a quick victory lap this week because we saw Ancestry.com’s ThruLines coming. We called it out…twice! And now that it’s arrived it feels like the “killer app” for genealogical DNA.
The good news is that we as serious users can avoid the downfalls, and use the predictive part of this feature to do the research for us, but we must immediately attach the citations to any newly added ancestor.
Ancestry has finally harnessed predictive technology in a very effect feature with the release of ThruLines.
At it’s essence ThruLines a new graphical way to show HOW you’re related to your DNA matches. This is long overdue, and while the old way of clicking on a direct match and seeing the path from your kit to the most recent common ancestor (MCRA) worked, it was limited (we had to dig into each match with 3 clicks to see the path) and it was totally devoid of any context. Was there a brother of this tester that also matches? Did they have a 1st cousin that also tested that in-turn matches our tests? You just had to click through each test to find out.
But, the real power in ThruLines isn’t the graphical change, it’s that it’s using predictive algorithms to scour both Public AND Private trees, to greatly narrow down where another DNA match is likely to fall in your tree. That’s right, those close relatives we see in our “DNA Matches” screens that are just dead ends because they have Private trees and don’t reply to Ancestry messages are now likely to be mapped in ThruLines.
Our first discovery
“Lynne” is solid 3-4th cousin match (67cM) to Michael’s father, and through all of the techniques we’ve discussed previously, we’ve narrowed down that our MCRA is likely to be Wesley and Jane Tradewell. This is the same Tradewell line that is one of our large brickwalls, and so the more data we have for this branch the better. Lynne entered a small tree at some point, but she left all of her named ancestors living, so they appear “Private” and we have almost nothing to go on when trying to link her to our tree.
When we opened ThruLines for the first time, we understood right away that each of the photos we were seeing were MCRA’s, and clicking on Wesley Tradewell, we immediately understood the power of this tool. There was Lynne, mapped for us, with some information still private but it greatly narrowed down the line through which we matched. Knowing that she likely matches through William Humphry Tradewell removed 2 generations of likely matches, and narrowed down our search to children from William’s daughters. Even more powerful however is that we’re in regular contact with family who is likely Lynne’s 1st Cousin, Once Removed. We sent a quick email to that person, they confirmed Lynne’s lineage, and we’d filled in another DNA match. Actually, it was two, because we had the same issue with “Jonas667” (2 relative tree, both living/Private), and it was resolved in the same way.
All of our efforts to shrub out William Humphry Tradewell’s children had failed up to this point (no census, no obituaries, no Public Trees, etc.), so we had only one daughter tentatively identified. It would have taken significant work to break down the matches to Lynne/Jonas667 by building out each of those trees, and since we’d already tried diligently and failed, it might never have been done. But ThruLines broke through on the first click.
Now, this is cannot be stated more clearly: ThruLines are at best speculative “hints” that can guide your work in very effective new ways, but they do not create evidence nor can we be sure they even contain proof of anything. They are like user trees in that manner…and for good reason, they are built entirely on user trees.
We expect one day brickwalls will be broken down by these tools while we’re sound asleep.
But the way the tool narrowed down these matches makes it much easier to prove them out. We know exactly where to start now, where last week we were stumped.
Half of our tree traces their roots to enslaved African ancestors, and the second discovery we made was that ThruLines would give us suggestions even if there were 2 generations separating supposed ancestors. In the example we saw as we clicked through Michael’s maternal grandmother’s MCRA’s showed us an African-American GGM that was born in 1879, and no evidence of who her parents were. This is very common…to get to/near the 1870 wall for those of African decent, but no good leads on the generation previous. In this case however, we have a new hint: two identified generations of European ancestors, and two unidentified generations after them, leading to a link to the known Fanny Johnson. This is a highly speculative connection, and it will take considerable work to prove/disprove it going forward, but it’s at least a lead and it’s based on at least a little bit of conjecture. It may not be anything, but it also could be one of those rare finds that links one of our African ancestors back a few generations closer to their enslavement, as well as identifying the slave owner that contributed to our genetic make up.
Going forward, this could literally revolutionize both genetic genealogy as well as standard genealogy. Artificial Intelligence and matching algorithms can not only see patterns much better than any human can, they can do it faster while analyzing more data than we can ever hope to review in our lifetimes. We can see that in the future, when a new AncestryDNA kit is processed and put online, the user will see a large tree of matches and MCRA’s as their first few of the results, instead of a list of 4000+ matches they have to map one at a time.
Also, imagine a day when they use these tools to validate evidence of each users’ Public tree for instance, looking for clearly incorrect data/relationships and flag it for users. At the very least they can rate that “source” tree as unreliable, and bring better sourced trees to the forefront. These tools could easily allow the power of individual trees, while also bring them all into line with known facts, and start matching them in ways we can’t imagine.
Going back to our Tradewell example, our brickwall is around Reuben Tradewell and the one piece of evidence that his father might be “Jakin” from Connecticut…but the trail goes cold. Tools like this, however, point to the power that’s coming where some other groups of genealogists and family historians have a mystery “Jakin/Jacob” from Connecticut that they have sourced generations back, but don’t know his disposition. We expect one day brickwalls will be broken down by these tools while we’re sound asleep.
We promised to keep this series as science-free as possible, and instead focus on the practical use of AncestryDNA tests to identify your ancestors. We’re going to keep that promise here, but we want to say a few words about other types of DNA tests you can’t get from Ancestry, and how xDNA, yDNA, and mtDNA can be useful! Just remember, we’re generalizing a bit here, and if you want the detailed science behind all of this, Google has many great reads.
yDNA and mtDNA
yDNA and mtDNA come only from your father and your mother, respectively, and change very little over the generations. These tests are often written off in the genealogical community, because they won’t, by themselves, lead you to how you are matched with someone, or how many generations back you might match them.
For example, if our Michael has a yDNA test and he matches “Frank” who shares the same yDNA…it tells us next to nothing. From the test we know that on Michael’s paternal line we have proved he’s matched to Frank…but there’s no way to tell how. It could be 1000 years ago we all had a MRCA, or it could be that Frank’s 4xGGF was a brother to ours, but there’s no way that kind of range will narrow it down by itself.
There are two uses of yDNA and mtDNA, however, that makes these some of the most powerful tests you can take:
Unlike “ethnicity” estimates we see from all the major testers, yDNA and mtDNA can be very effective in pinpointing very accurately the location of your ancestors on the planet. The standard (Autosomal) DNA tests from Ancestry rely on a small global sample of historical DNA (16,000 samples currently), and human created Family Trees, to mathematically try and guess where our ancestors were 800-1000 years ago. They are looking for little shreds of DNA to trace back, and it’s very small amounts because Autosomal DNA gets cut in 1/2 for each side of a lin ever generation. However y/mtDNA doesn’t change over the generations and so we know very accurately where those ancestors were, based on where the bodies were found. This is especially important for African American genealogy, when there are nearly no records of origin before our ancestors were taken from Africa. These tests can be very accurate, and place your ancestral group in to very small physical and/or social (tribal) locations.
Brick wall research
In our example above, we know for a scientific fact that Michael and Frank share an MCRA along their paternal line. The same is true for women who have an mtDNA match. While that again doesn’t help us much if we have no information, it’s invaluable if we have a good guess on how we’re related. Let’s go back to our DNA Painter walkthrough to see how we wish we had yDNA and mtDNA tests.
To recap from last week’s post, we have two lines of DNA tests that we know are connected, and we have narrowed down the MCRA for both a cluster of AncestryDNA matches and on our line, but we don’t know how they connect. So, we have two couples (Jacob/Maria Kupsch, and Joseph/Dorothy Haasl) that we know match, most likely 2-3 generations above them. Each of them have 8 potential match relatives, and we have 4 known relatives, so we’re facing 32 ancestors that might be our MCRA.
But, if we can confirm the y/mtDNA from those 4 relatives, whom all died over 100 years ago, because that DNA doesn’t change between generations. That means a direct male relative from Jacob (say his son’s, son’s, son’s, son’s DNA) will confirm Jacob’s yDNA. Same for Maria, and a direct female relative. If we could yDNA test relatives of both Jacob and Joseph, and mtDNA for both Maria and Dorothy we would have about a 25% chance of finding an immediate match. And, if say Maria and Dorothy share the same mtDNA we just figured out we need to focus our research only on both of their maternal lines to make our match. If we don’t find that match, we just eliminated 25% of our potential match points, so now instead of building out 32 ancestors to find our match, we’re down to 24. But even better, if we can go one level up and do the same thing, we can eventually narrow this down to where we share an MCRA.
It would be highly unlikely we could ever build out a family match with xDNA, and the cM you share with someone tells you almost nothing about close of a match you are with them. The main value of xDNA is if you do match someone, it narrows down your link to that match in a very powerful way. xDNA is inherited in a unique pattern that going back several generations can eliminate more than 50% of your tree as a potential match.
Women will inherit an X chromosome from both their mother and their father, but men will inherit an X from only their mother. Going back to High School Biology, we quickly remembered that women have an XX chromosome, while men have XY!
But, the value for us comes in when we build out our potential ancestor’s chart, using that inheritance pattern. So, if we have a female test subject who has an unknown xDNA match, we know it’s not from her Father’s Father’s line because men only inherit their X from their mothers. Going back 2 generations, we just eliminated 25% of the potential matches. If you know, from other research, that this unknown match is on their father’s line, you just confirmed it’s on the father’s mother’s line.
You won’t see a lot of xDNA matches, but when you do, Google one of the many xDNA inheritance fan charts, and start to see if you can eliminate suspects in how you match. It could bring you much closer to where to hunt for your MCRA.
Just know that all of this work will have to be in GEDmatch however, since AncestryDNA doesn’t provide any information on the details of your genetic matches, and none of the tools needed to view/manage this information.
In this installment we’re going to walk through a key tool to help narrow down where to research when you have AncestryDNA tests that match your family, but despite your research you’re not sure where they match. DNA Painter has a great tool called What are The Odds that gives us the probability of where these unmatched lines link up with our own.
We’re using a real set of unknown matches for this example. Emma Kupps (1879-1953) is a one of our favorite ancestors. She was born and raised the various logging communities that sprang up in North Central Wisconsin in the late 1800’s, but her family settled in Antigo where she graduated from Antigo High School. Within a few years she would married a logger Daniel Leonard (1868-1924), who would soon become Antigo’s Fire Chief, and years later be elected Sheriff of Langlade Coounty, Wisconsin. During his term Dan became ill with cancer, and succumbed with a significant portion of this term remaining. The governor of Wisconsin appointed Emma to the position of Sheriff to complete her late husband’s term, and she became the first woman in Wisconsin to hold the office. (Langlade Co. Historical Society)
DNA Painter has greatly narrowed down where we’re targeting our on-going research to finally break down this brick wall.
But, to family historians, she’s also near the end of a line that is a classic brick wall. Her father died young, and there’s nothing but a couple of records that indicate only the names of his parents. Plus, they are the only lines in our family that come from Bohemia, so it has the combined brick walls of classic genealogy and DNA results.
We’ve identified a group of AncestryDNA matches that have strong Bohemian roots and match descendants of Emma. We used Michael’s Great Uncle as our target DNA match, since he’s the oldest generation tested on that line, and we built a master tree that links as many of the unknown DNA matches as we could. We ended up with 8 AncestryDNA matches that we could link together in a cluster.
The cluster all share Jacob Haasl and Dorothy (Johannek) Haasl as their MCRA, but we haven’t been able to build a link between Great Uncle Leonard and the Haasl’s. So, we’re going to turn to DNApainter’s “What are the Odds” tool, to help identify where we’re most likely linked to the cluster.
When you open “What Are the Odds?”, it will present a box for the most recent common ancestors (MCRA). The options are to “Edit Names”, “Add Child”, or “Add Parent”. In this case, we’re going to edit the name, and add the cluster’s MCRA, Joseph and Dorothea Haasl. When we enter that information, we’re presented with the same 3 choices, but this time we’re going to start building a line to one of the DNA matches but selecting “Add Child” and entering the name of the child that makes up the first step to our DNA match. At first we were surprised how quickly we built out a tree, but it’s because we’re not entering all the data we’d need for a regular tree, just the names!
When we reached a DNA match we entered the cM value that matches our known DNA test. We repeated this step for as many matches as we’ve identified. This works well with a single match, but better with more. In our case we identified 8 matches, so we’re built them all out. Now we’ll really see the power of this tool.
Now that we’ve entered what’s known, it’s time to start mapping out our guesses. In fact, the entire purpose of this tool is to compare the likelihood of at least 2 hypotheses matching the entered cM, and from those likelihoods we can focus on where it’s most likely we all share a most MCRA.
The most likely connection for Jacob and Mary Keips’ line is her parents. We don’t know her maiden name, or birth date, but if we guess that she was born in 1820-1825 it’s reasonable to guess she is a sibling of either Jacob Haasl or Dorothy (Johannek) Haasl, so let’s build that out as option 1. We’ll add an “Unknown 3xGGP” to Joseph and Dorothy, and add a child called Jacob/Mary (because it could be either!). From there we’ll build down to the Great Uncle that is the known DNA match, and select “Use as Hypothesis”.
It shows us a probability of “1” because DNA painter doesn’t show you raw percentages, it shows you comparative probability of one match vs. another. For example, if you enter two hypotheses and one returns “1” and the other returns “2”, we’ll know the second one is twice as likely as the first. In this case, we have no other hypotheses entered, so it shows just a 1.
Given the cM match, it’s most likely that we match the cluster with Great Uncle Leonard’s 3x to 5x GGP’s, so we built out the same line as above, but this time with one more unknown ancestor above Jacob/Dorothy Keips, which would then make Uncle’s MCRA a 4xGGP.
When we built that out, and selected the second “Great Uncle Leonard” as a hypothetical, it soared to a whopping score of “1174” vs. the first “Great Uncle Leonard!! Given that we have 1174 for one possible link and 1 for the other, DNA painter just told us that while not impossible, we’re looking for a 4xGGP as our MCRA, not 3xGGP. Not great news, since now we have to go at least two more generations back, and to build this match back further we’re going to have to dig deep into 18th Century European genealogical records. That’s not our strong suit. But, at least now we have a clear picture of where we’re looking to link these groups.
Since the range of likely Great Grandparents is 3x-5x, we then built this hypothetical out to our match’s 5xGGP, and we see the same score of 1174 from a hypothetical 5xGGP. That means it’s equally likely that our link to this cluster of match is through our Great Uncle Leonard’s 4xGGP or this 5xGGP, but it’s almost certainly NOT through his 3xGGP.
While in some ways this is disappointing, and we’d hoped to come through with a match, this is actually a huge piece of this brickwall puzzle. When we started the work on this DNA cluster we knew that John Keips/Kupps had migrated from Bohemia and, at the time of his death, his wife thought his father was Jacob D. Kupps when she filled out her husband’s death certificate. From their marriage certificate we knew John’s mother, and Jacob’s wife, was Mary. We also knew we had a large cluster of DNA matches who came from the area of Bohemia.
Just by going through that cluster, building out a central tree that links them all, we found a great lead that likely shows John’s arrival information, along with approximate birth years for Jacob, Mary, and John…as well as John’s previously unknown siblings who seem to have a long history together in the US, and left many records. That means instead of having exhausted all the on-site research we could do on the John’s line, we now have a large number of leads to follow and see if we can push back another generation from both Jacob and Mary. We now know enough to start targeting death certificates for both, which may contain critical names, as well as 6 more marriage/death certificates to look for Mary’s maiden name, as well pieces of evidence that link our Jacob to the arrival Jacob. And, DNA Painter has greatly narrowed down where we’re targeting our on-going research to finally break down this brick wall.
We also have about 20 trees integrated into the master tree, and all of their owners are likely working towards the same goal as we are. As they do their research, and new DNA matches are added to the mix over the years, it’s likely one of us is going to have that piece of the puzzle we’re missing, and finally put it all together.