Matching unmatched DNA matches by Casting a Wide Net, Part 6 – Our crazy attempt to leverage 288 DNA matches to expand our tree comes to it’s conclusion

Matching unmatched DNA matches by Casting a Wide Net, Part 6 – Our crazy attempt to leverage 288 DNA matches to expand our tree comes to it’s conclusion

In the five previous parts of this series: We identified a plan to tackle what looked like a large group of DNA matches (Part 1), we went through and tagged all 288 of our Ancestry DNA results that were related to a group of matches that had Woodley/Woodson surnames in their attached trees (Part 2), we then built out a common tree for as many of the matches as we could, to nail down common ancestors, and to gain clues on where these matches link up with our tree (Part 3), we used GEDmatch and DNApainter to target the most likely line of “Mary’s” that leads from her to the group of 12 DNA matches (Part 4), and last week we broke through a brick wall with some old fashioned genealogy (Part 5). In this installment, we wrap up the story of this journey and the lessons we’ve learned. 

This journey also highlights the paradox of genealogical DNA: Your matches will come easiest on lines where you have a complete and accurate tree, but you’ll struggle to match those that are on the lines where you really need the help of DNA…because you don’t have a complete and accurate tree.

As we ended our last installment, we’d identified Sam Caswell’s wife as Annie (Moore) Caswell, daughter of Robert Moore and Henrietta (Bradford) Moore. We were able to quickly identify Henrietta’s mother, Sallie Bradford and five of Henrietta’s siblings. It was amazing, the links came easy, and the tree fell in-place just how you’d hope. The only problem was…we weren’t getting any closer to linking Roman and Mary Jones to “Mary”.  

Screen Shot 2018-08-10 at 3.18.43 PMGoing back to our work with the “What Are the Odds?” tool (Part 3), it’s 48 times more likely that “Mary” and Roman/Mary’s Most Recent Common Ancestor was our “Mary’s” 3x Great Grandparents, than it was her 2xGGP, and 77 times more likely that it was 3xGGP v. 4xGGP. That means Annie (Moore) Caswell’s parent all but needed to be the MRCA. One thing became increasingly clear as we shrubbed out our tree with the new information: Sam and Annie weren’t a link to Roman and Mary Jones 

Roman Jones was born around 1840, and his wife Mary was born around 1838. Annie (Moore) Caswell parents were both born around 1880, and for them to share parents would be…incredible. We looked back a generation (hoping to defy the 48 times odds!), and the lines still didn’t match.  We had good info on “Mary’s” 4xGGM Henrietta Bradford and her siblings…and while we couldn’t rule it completely out, it was very likely she wasn’t a link to the Jones either.

We went back to review everything we had on Annie Caswell, and in the 1910 U.S. Census it jumped out at us: Sam and Annie listed themselves as having no children, despite the fact that Mattie would have been 7 years old. She also indicated that she never had children. 

SamAnnie1910Census

When we looked at our notes, and research we realized we fell in the most basic trap in genealogy research: we had accepted family lore as fact, and built around that “fact”. We had an uncle that had done some basic Ancestry-based research, and when we first built out a skeleton tree, we’d used his info as the bones of the Caswell line. We had all the right facts on Mattie Caswell, we had all the right facts on Sam Caswell and Annie (Moore) Caswell…but we’d never proven their link. We went back and reviewed the transcripts of other family interviews we’d done with Mattie’s granddaughter (and others) about 4 years ago and there it was. They described that Mattie’s mother had died soon after Mattie’s birth, and her father died soon after. Mattie had been raised by others, her parents weren’t Sam and Annie, and the brick wall we’d broken through wasn’t ours…in-fact it wasn’t anyone’s, since they never had children who would be researching their ancestors.  

So what did we learn in all of this?

  • The crazy strategy of casting a wide net across 288 DNA matches worked..even though it was a LOT of work.
  • We identified a key ancestor, and we know where we can expect the MCRA to fall in our line once we know more about our line.
  • In the end, no matter how high-tech genealogy research becomes with DNA, it still comes back to the basics of a solid tree, with strong evidence, supported by old fashioned family history research. Without a solid tree, we can’t take full advantage of DNA links. 

This journey also highlights the paradox of genealogical DNA: Your matches will come easiest on lines where you have a complete and accurate tree, but you’ll struggle to match those that are on the lines where you really need the help of DNA…because you don’t have a complete and accurate tree.

For us, it’s back to the drawing board. We’re spinning off the branch of the Caswell tree for Sam and Annie that we’ve documented so well, and making it Public so others can benefit from our work. We’re attempting to identify more information from family on where/when George Barnes and Mattie (Caswell) Barnes died, so we can get their Death Certificates and begin working backwards again!

Matching unmatched DNA matches by Casting a Wide Net, Part 5 – Rolling up our sleeves and doing some genealogy

Matching unmatched DNA matches by Casting a Wide Net, Part 5 – Rolling up our sleeves and doing some genealogy

In the first four parts of this series: We identified a plan to tackle what looked like a large group of DNA matches (Part 1), we went through and tagged all 288 of our Ancestry DNA results that were related to a group of matches that had Woodley/Woodson surnames in their attached trees (Part 2), we then built out a common tree for as many of the matches as we could, to nail down common ancestors, and to gain clues on where these matches link up with our tree (Part 3), and we finally used GEDmatch and DNApainter to target the most likely line of “Mary’s” that leads from her to the group of 12 DNA matches (Part 4). In this installment, we use take the high-tech leads we have and do some old-school genealogy to try and prove out our theory on who connects us to Roman and Mary Jones.

Given what we knew, it’s most likely that Mary matched the other 12 through a 3x Great Grandmother on her mom’s side. Of course that’s two women…neither of which we know much about: Fannie (Johnson) Barnes and Annie (–?–) Caswell. We had some decent confidence in who Fannie’s parents were, and the family originated in Tennessee. Since Roman and Mary Jones were from NW Mississippi, we decided to focus in on Annie Caswell from Quitman County, MS.

Mattie's Tree
Sam and Annie Caswell, when we started this process

We had almost no information on Annie Caswell. We didn’t know her maiden name, her actual birth year (only Census year), her death year, or her parents’ names. We did have her husband’s death month, year and location, and so we decided to order Sam Caswell’s Death Certificate and hope that would be enough…and that maybe there was more information on Annie listed.

Mississippi is horrible when it comes to Vital Records. They didn’t start requiring any Birth or Death Records to be recorded until 1912, and wide adoption by counties wasn’t complete until the early 1920’s. Additionally, they spent decades trying to dehumanize people of African descent to the degree that the less they were recorded as people, the better. On top of that, what they did record is mostly neither online or microfilmed…or even indexed.

There is no question DNA testing/results are valuable, but in the end it’s still just genealogy and the same techniques that have been used by family researchers for a 100+ years that are going to break through your brick walls.

This is one of the reasons there are so many holes in this branch of our family. We haven’t made the genealogy pilgrimage to Mississippi, we’re not sure what we’ll find when we get there, and there’s next to nothing available remotely. Facing this, we decide to attempt our first MS Vital Record purchase online…which requires “VitalCheck”.

About 2 weeks later, after having to remember just about every address we’ve had in the past 20 years to get past VitalCheck and deciding that selecting “Grandchild” for our relationship when we really meant “maybe 2nd Great Grandchild” was the same thing, we received Sam’s Death Certificate in the mail.

We were hoping that Annie was still alive at the time of his death, and that the certificate was filed out with her maiden name listed (maybe?). The bad news is, none of that was true…but the good news is that it gave us a next step.

We know that Sam and Annie were married in the 1940 U.S. Census but by the time of Sam’s death on 4 Jul 1974, he was married to Emma (Fox) Caswell. Of course, we’re not even 100% sure this Sam is the same Sam from Sam and Annie in 1940, and now we have a different spouse listed on the birth certificate. But, it’s possible that Annie died after 1940 and Sam remarried. The problem was how do we find Annie and tie her to Sam, let alone find her maiden name.

The hint we needed came from Sam’s Death Certificate, and his burial location. He was buried in an African American cemetery in Quitman County, which like most Southern Black cemeteries, is poorly documented. Looking in Find-A-Grave, we saw that only 1% of the headstones here were photographed, and there was no record of Sam’sOakGroveCemetery headstone/burial. However, looking at the other Caswell’s in Oak Grove Union Cemetery, we found a major lead: Annie Caswell, b. 15 Sep 1882, d. 11 Jul 1969. We’re well into speculative territory here, but this Annie first the proper birth range and she died before Sam.

Needless to say, the next step was to order our 2nd Mississippi Vital Record from the Mississippi Department of Health. But it highlights something we’ve known for quite some time about Genetic Genealogy, but it’s easy to forget: There is no question DNA testing/results are valuable, but in the end it’s still just genealogy and the same techniques that have been used by family researchers for 100+ years that are going to break through your brick walls.

We’ve shown how we can use things like DNA testing and GEDmatch to give us leads researchers would have NEVER had 20 years ago…but in the end, only the basic work of gathering and confirming Birth/Marriage/Death records will turn the leads into family members.

We received Annie Caswell’s death certificate, and it was the goldmine we were hoping for! She was married to Sam Caswell at the time of her death, and her mother’s maiden name was listed. We had gotten back a generation, both parent’s names, birth date and location…everything you could hope for! Unfortunately, her mother’s maiden name wasn’t Jones…it was Henrietta Bradford. That means Mary’s 3x Great Grandmother wasn’t a Jones, by name. It would have been much easier…but it’s possible that the 4x Great Grandparent we expect will be the link to the Jones family was Henrietta’s mother.

AnniesParentsNameWe’ll conclude this series next time, as we shrub out Henrietta’s tree…and reach the end of this journey!

Matching unmatched DNA matches by Casting a Wide Net, Part 4 – Proving the matches, and establishing a theory of connection

Matching unmatched DNA matches by Casting a Wide Net, Part 4 – Proving the matches, and establishing a theory of connection

In the first three parts of this series (Part 1, Part 2, Part 3), we went through and tagged all 288 of our Ancestry DNA results that were related to a group of matches which had Woodley/Woodson surnames in their attached trees. We then built out a common tree for as many of the matches as we could, to nail down common ancestors, and to gain clues on where these matches link up with our tree. In this installment, we leverage GEDmatch, and deductive reasoning, to identify where we think our tree will link up with their trees.

The largest DNA matches (by centimorgans) we identified in Ancestry had also uploaded their DNA results to GEDmatch, so we were able to do tests to confirm they all truly matched. The “One-to-One” matches for each of them confirmed they were all related to “Mary”. It’s not scientific to say that all 288 were actual DNA matches, but we know the core group of matches are and that a good number of the matches-of-matches are likely also legit.

X DNA is tricky, but the important use of it identifying people who you CAN’T be a match if you share X DNA.

We were also able to use GEDmatch to identify the “true” cM match amounts between various matches, and from there we leveraged the International Society of Genetic Genealogists’ table showing cM ranges and averages between various family relationships (Shared cM Project – V3). The closest match for Mary was “W.W.”, and we settled on 133cM as their match level. The most likely relationship for that level of match was with a shared Great Grandparent, with W.W. likely being a 2nd Cousin, or a 2nd Cousin Once Removed. When we fleshed out the other 12 matches on paper, they all roughly fit this notion that they matched either Mary’s Great Grandparent or Great-Great Grandparent.

female-x-chart
Inheritance pattern for females (X DNA)

The other thing that jumped out at us, unexpectedly, from GEDmatch was that some of the 5 matches there had X DNA matches as well. X DNA is tricky, but the important use of it identifying people who you CAN’T be a match if you share X DNA. For example, a person will only inherit X DNA from their mother, so if you have an X DNA match that you’re theorizing is related to someone, but there are two male relatives in a row between the two matches, that isn’t possible.

Once we built out the theoretical map between all the matches and Mary it all fit that her GGP’s could be Roman and Mary Jones, and with the DNA levels and inheritance pattern of X DNA it’s likely that Marie’s relative was a daughter of Roman and Mary. It also pointed strongly to the matching being on her mother’s side.

 

Screen Shot 2018-08-10 at 3.18.43 PM
“What are the Odds?” gives you the chances of various hypothesis’

The day after we did our work on paper with the ISOGG chart as our guidance, DNAPainter introduced a new tool called “What are the Odds?”  that does the same work we just did on paper! It’s easy, it’s awesome, and we’ll cover it in more detail in a future post. But, most importantly, it showed us that it was 77 times MORE likely that Roman or Mary Jones’s parents are our Most Recent Common Ancestor, than anyone else. It’s technically possible that our Mary is directly descended from Roman and Mary Jones, or that they are connected by 4xGGP’s…but it’s much, much more likely we’re looking for Mary’s 3xGGP’s, the parents of Roman and Mary.

Looking at Mary’s tree, and she of course has 2 maternal GGM’s. One, we have some documentation (mostly Census info with Ancestry Member Trees), but the other we had almost no information. We’re guessing the one we have little information on, Annie Caswell, might be the best lead, so we’re going to dig into her.

Samuel and Annie Caswell were born, married, and died in and around Crowder, MS. Family lore has Sam and Annie as Mary’s grandparents, but we only have Annie’s Census birth date, and no maiden name for her. About the only piece of hard information we had was that Samuel might have died in July 1974 (based on the SSDI).

Time for some old-school genealogy, to hopefully prove out the high-tech theory that points to Annie Caswell being on the Jones line.

Next in the Series: Matching unmatched DNA matches by Casting a Wide Net, Part 5 – Rolling up our sleeves and doing some genealogy

Matching unmatched DNA matches by Casting a Wide Net, Part 3 – Building a single tree using all of our DNA matches Public Trees

Matching unmatched DNA matches by Casting a Wide Net, Part 3 – Building a single tree using all of our DNA matches Public Trees

In Parts 1 (Casting a Wide Net, Part 1) and 2 (Casting a Wide Net, Part 2) of this series, we went through and tagged all 288 of our Ancestry DNA results that were related to a group of matches that had Woodley/Woodson surnames in their attached trees. In this installment, we get to work!

After we took a few days off, we downloaded the list of “Mary’s” (our Grandmother we’re trying to build matches to) DNA matches, and converted it to an Excel spreadsheet. We filtered the list to only show the 288 matches we’d tagged as related (DNA Line 47!), and deleted the rows that weren’t tagged. Next, we filtered on Private Trees and color-coded those 128 rows in Red since there’s no further work we could do on those lines.

That left us with 160 matches that had at least a small Public Tree. Sorting the list by centimorgans, we started reviewing each match one-by-one. The first match, with the most DNA in common, had a large Public Tree, so we started a speculative tree using the match as the root, and building back to the oldest Woodley ancestor. We followed the steps we outlined earlier in How to Build a Good Public Tree (Building a good Public Ancestry.com tree), so we had a decent foundation of facts supported by documentation. Going to the next match, we attempted to build a link between the first tree and this second tree. We found the link, and could tie these two DNA matches to each other in our speculative tree! We flagged these linked matches Green in the spreadsheet, and moved on to the next.Screen Shot 2018-08-05 at 9.08.17 AMWe soon had a small skeleton tree building out nicely. We quickly found 4 matches we could link up, 2 others that we couldn’t build out from the data they had in their Public Tree (we flagged those as Yellow), when we hit the tree that brought it all into focus. “W.W.” had a 72cM match with Mary, and a Public Tree with over 3000 members. As we built this link, we found the most important data we’d learn in this project: The shared ancestor wasn’t a Woodley/Woodson, it likely was Roman and Mary (Stewart) Jones.

Roman and Mary Jones were born into slavery on the Eastern Seaboard in the 1830’s, and the Public Trees files available for them seem to have good Probate data showing the trail of ownership for them through to 1860. It appears they have 8-10 children that we know of (3 of which are confirmed with this DNA work), and there’s even an amazing picture of the couple. It appears someone has done quite a bit of work on this couple.

As we built this link, we found the most important data we’d learn in this project: The shared ancestor wasn’t a Woodley/Woodson, it likely was Roman and Mary (Stewart) Jones.

We spent weeks reviewing the 160 matches with Public Trees, and we eventually were able to link 12 of matches together in one speculative tree. All 12 share Roman and Mary Jones as a common ancestor. We also found 4 other matches that we could build out quite a bit, and we believe they are close to being linked to this grand tree, but we couldn’t find that missing piece. We called these “Orphaned Trees” and tied them into a placeholder.

The remaining 145 matches had data in their trees, but we couldn’t find good patterns to match with just what was online. For these 145 we built a spreadsheet of the listed family names, so that as we continue our research we can hopefully match those family names to new matches, and expand the tree.

Ultimately our tree has nearly 500 names with good data supporting each one. We didn’t use Member Family Trees to support any of the facts (other than some relationships), and we were able to shrub out some of the less developed branches with siblings/offspring/etc.

The bad news was, we couldn’t link ANY of these matches to our Mary directly. We are confident the matches are all were related to Roman and Mary Jones, and we can reasonably assume our Mary is too, but the link just wasn’t there yet.

That lead to our next step: leverage GEDmatch to confirm our theory that these were valid matches, and try and come up with an educated guess on where our Mary might link in with this group.

Next in the Series: Matching unmatched DNA matches by Casting a Wide Net, Part 4 – Proving the matches, and establishing a theory of connection

Matching unmatched DNA matches by Casting a Wide Net, Part 2 – Identifying all “Matches of Matches” as a Group

Matching unmatched DNA matches by Casting a Wide Net, Part 2 – Identifying all “Matches of Matches” as a Group

(Note: Before we go further, we rely heavily on Jeff Snavely’s Ancestry DNA Helper tool for this work. It’s a Chrome plug-in that adds a lot of essential functionality to Ancestry DNA results. You can download it here: DNA Helper. For this project, we’re specifically using two features: “Search Test Notes” and “Download Matches”, and if you’re following along at home, you’ll want to get this tool and get acquainted with it.) 

As we discussed in our first post (Part 1 – A crazy, desperate idea), we came up with the idea to cast a wide net on this group of matches that have “Woodley” or “Woodson” surnames, build out one big speculative tree for them all, and see if we can identify a pattern to the matches that would indicate where the DNA grandma in question (“Mary”) might connect.

It is critical that we identify the target group from the nearly 11,000 DNA matches for Mary. It started easily…select the closest match with “Woodley” in their attached tree, and put a string in the Notes field of that match. Our note indicated who’s side of the family this match was on, and broke down if we’d researched it yet, and tagged it as “Woodley/Woodson family, unresearched – Line 47” since this is the 47th separate shared ancestor we’re researching in on DNA trees. By adding this note, we can easily search all of the matches later, and when we have the entire list of matches, we’ll be able to review and confirm the status of each. Have we reviewed the match’s tree? Did we find out how they fit in the large shared tree?

Click, paste, click, click, paste. Times 288. It took over a week, and it was some of the most boring work we’d ever done in our time doing family history.

Once we added the first note, we clicked on “Shared Matches” and clicked on the first match that Mary and this match shared. We then added the same note to that person, and repeated the process. Match, note, Shared Matches, click Match. We had NO idea how many matches we’d have to repeat this before we started. It turns out it was 288 times. Click, paste, click, click, paste. Times 288. It took over a week, and it was some of the most boring work we’d ever done in our time doing family history. We’ve gone through 10 years of unindexed registers of the Educable Children in Tallahatchie County, Mississippi…with about 300 pages per year. That was fun and excited compared to this. We had NO idea how many matches of matches were in front of us.

Line memoTwo things to note. First, yes it was horrible work, but it was great to finally make progress on such an intractable line with so little data on how those matches matched. Second, this is amazingly unscientific so please know that we knew going into this that some one of these people weren’t matches, they merely were tagged as potential matches by Ancestry. There’s no way to confirm the level of confidence in a match in Ancestry, so we took it at face they all were. That way we’d cast that net as wide as we could, because you didn’t know who’s tree/match would be the valuable one that would provide the connection we needed.

Click, paste, click, click, paste. Times 288. Not even knowing if this would pay off.

After over a week of work, when it was finally all done…we took a few days off. Our minds needed the break. But the hardest, worst part of this project was done, and we were about to start the fun, valuable stuff!

Next in the Series: Matching unmatched DNA matches by Casting a Wide Net, Part 3 – Building a single tree using all of our DNA matches’ Public Trees

Matching unmatched DNA matches by Casting a Wide Net, Part 1 – A crazy, desperate idea

Matching unmatched DNA matches by Casting a Wide Net, Part 1 – A crazy, desperate idea

It’s one of the things that slaps you in the face when you jump into DNA Genealogy: Finding a genetic match will rely mostly on other people’s trees being built out to 3x and 4x great grandparents. Since most people who take an Ancestry DNA test don’t have their trees completed to that level, we spend 90% of our time building other people’s trees in order to make matches.

This is amplified when doing African American genealogy, since there are even fewer complete trees available. It is not surprising, given that this country didn’t treat African people as humans for most of its existence, and then we spent the next 100 years or so denying those of African heritage basic rights and access to government across large parts of the US. It resulted in not only devastating impacts, but also simple things like Vital Records not existing, cemeteries segregated and at risk of being destroyed without a thought, etc. Combine that with the cultural hesitance in good parts of the older African American community to either ask or discuss their history with their elders and children (it can be a damn painful story!), and it’s no wonder this fun hobby the European American side of our family has enjoyed for over 100 years wasn’t as nearly matched by our African relatives.

We came up with a strategy on how to break this wall down: cast as big a net as we can, catalog every match-of-a-match that we can identify, and build out all of the trees we can as far as we can to see if we can start building a tree that gives us hints as to where our family might flow through

But, as we build our American Genealogy, we are working to build that side of our tree out and to use DNA as a tool as much as possible. It’s not like we’ll quit just because it’s much harder!

With that, as we went through the DNA matches on Michael’s maternal grandmother’s lines, there are no matches that aren’t from the tests we manage, or from close relatives that tested independently and let us know their results. But as we browsed through her results, we kept coming across two surnames in a bunch of matches: Woodley/Woodson.

Marie's Tree
It’s nearly impossible to build a 3rd Cousin DNA match when you have solid info on only 3 of 16 2xGGP

We have no Woodley/Woodson in our tree, and none of the trees we looked at provided any hint as to the path that might link these matches and the maternal grandmother in question. So, after mulling this for a couple of months, we came up with a strategy on how to break this wall down: cast as big a net as we can, catalog every match-of-a-match that we can identify, and build out all of the trees we can as far as we can to see if we can start building a tree that gives us hints as to where our family might flow through. Once we had the best tree we had, we hoped that using the strength of each match mapped to that tree would give us hints on where to dig in a build the links between our target (“Mary”) and her matches.

We knew this would be further complicated by not entirely knowing/trusting that the man listed on her Birth Certificate is her father. Going into this there was a 50% chance that all of these matches might link to a side of the family where we know only a man’s name…and it might not even be the right name. Again, just because it’s hard doesn’t mean we weren’t going to do it, and the hope was that even if the matches are all from Mary’s father’s line, that will just help us learn more about that brickwall.

We’re going to cover this in a series of posts, and next we’ll talk about how we cast that net over 288 DNA matches and very little other supporting data!

Next in the series: Matching unmatched DNA matches by Casting a Wide Net, Part 2 – Identifying all “Matches of Matches” as a Group

 

Breaking down a brick wall in real-time: Episode 4 – Looks like we’re back at the brick wall…

Breaking down a brick wall in real-time: Episode 4 – Looks like we’re back at the brick wall…

(Note: This is continuation of a series, and we suggest you start at the beginning – Breaking down a brick wall in real-time: Episode 1 – The setup, and the discovery)

Well, it looks like we’ve reached the end of this experiment of live-blogging a search to break down a brick wall. It was kind of exciting starting this series not knowing if we’d be successful, even though it felt like we might be able to link this DNA match.

We’ve gone through every quick trick we know to try and find Excell before 1910, and so far, we’re striking out. We’ve searched for Excell (and XL and Ex and X) in both the 1900 and 1880 U.S. Census. We’ve used Stanford/Standford/Sanford/Stanfor and we’ve looked in the counties we expected him in, surrounding counties, and the entire state. We’ve even looked for him as “White”. Theres nothing so far, although there are a surprising number of people with the first name “Excell” in Mississippi in 1900! We’ve also tried to push through to 1880 using his uncle James, but to no avail. We now know James’s wife’s maiden name, but we can’t find either of them in the 1880 U.S. Census.

So, after a day of searching, we have another set of data points which will someday help us break down this wall, but we didn’t get through it today. When we’re on the ground in Mississippi doing physical research, we’re confident we can put this puzzle together…just not today. Not yet.

But hopefully sharing this day of searching illustrates how we go after these leads, and we’ll follow this up with a more detailed explanation of our approach on these DNA matches, and of course we’ll follow up when there’s a break through!!

Last screen shot 2
Correcting the tree from Episode 3…THIS properly reflects James as Excell’s Uncle. Thanks for one of our readers pointing it out!!