DNA As A 3-in-1 Solution for Family Trees: Fertilizer, Root Stimulator, and Weedkiller
Don't overthink the metaphor, but do think about it.
The first part of my genealogy “origin story” ended on the cold, hard slab of uncertainty after I explained that my second great-grandmother, Ella (Latimer) Henry, constituted the mistaken half of a case of identical names. Her father was, in fact, not Albert H. Latimer, a signer of the 1836 Texas Declaration of Independence, the husband of three wives (that we know of), and the father of nineteen children (that we know of).
I ended that first part with a pretty clear hint as to how I softened up that uncertainty about Ella’s parents:
Hint: think DNA.
Not for the faint of heart, short of time, or intolerant of attitude. But after 36 days of being held hostage by this story, I’m breaking free and letting go of it. It’s the story of what I did and, as the kids say these days, I have receipts.
But before we get into segments and centimorgans …
Please note that the Hint above did not say:
think only DNA.
Before I could seriously concentrate on solving the who question, I needed to better understand the where and when considerations. And that understanding would require some plain old genealogy gumshoe detective work.
To establish the geographic context for investigating Ella’s unknown parentage, here’s what I refer to as Ella’s Known World. (Or if I wanted to explicitly make this a drama, which I don’t, you could think of it as the scene of the uncertainty.)
See the inset at lower right to situate the area within the state.
Location can be important; sometimes it can be determinative. Of the seven children in Burr and Ella (Latimer) Henry’s pre-nuclear family, all but one were born, lived and died in the area highlighted above. The lone exception was my great-grandfather, Lorenzo, the first child, who ventured out west ending up just across the Texas border with New Mexico.
The only documented facts we know about Ella reside within the rounded square on the map above:
1. Per Texas county marriage records, she was married to Burr Henry in 1880 in Paris, Lamar County.
2. In the 1900 and 1910 census (plural), she, Burr and their family lived near Lone Oak, Hunt County.
3. Burr died in 1916.
4. In 1920, two weeks after the U.S. census was taken, Ella died in Celeste, near Leonard, Fannin County, in the home of her second son, A.H. (Hamilton, or “Uncle Bud”).
5. Ella and Burr are buried together in the East Shady Grove Baptist Cemetery two miles west of Leonard.
Clarksville in Red River County is shown to the east of Paris to indicate the location of the other Ella’s city of birth, life and death.
Mount Pleasant in Titus County, south of Clarksville, is shown because it holds a clue for Ella’s inferred, highly-probable mother, who doesn’t make an appearance in this part of the story.
Throughout the summer and into September of 2019, I completed quite a bit of “field” research and study. I drove to libraries and cemeteries in Paris (Lamar County), Clarksville (Red River County), Bonham (Fannin County) and Greenville (Hunt).
Fortunately, towns like Leonard and Lone Oak had celebrated their centennial years in the late 1800s and had published genealogy-friendly books explaining the history and families throughout the first hundred years. That’s where I found these:
A page from my notebook for my September 11th outing to Fannin and Lamar counties:
My primary focus for all of the library work was the Latimer family and, to a lesser extent, Burr’s Henry family. By mid-September, I had read and absorbed enough of the general history about the Latimer’s to finally convince myself that Albert H. Latimer was not Ella’s father (nor Mary Ann Gattis her mother).
At that point, I deleted Albert and his third wife, Mary Ann Gattis, from my family tree. Ella was left an orphan so far as Ancestry knew.
Back to segments and centimorgans
I mentioned previously that I received my Ancestry DNA results in August 2018. Skipping a lot of details that could rationalize and even justify my inactivity on the DNA front, I’ll just say it and own it – I didn’t do anything but glance and wave at DNA matches for almost a year.
On the 19th of August, 2019, I wrote an email to my sister that documented my first engagement with Ancestry DNA tools. Six months earlier (February 2019), Ancestry had released the beta version of three brand-new DNA application tools:
Common Ancestors
ThruLines
Groups (aka, the little colored dots)
It goes without saying, but of course it goes better with saying … timing-wise, I could not have been more fortunate than to be able to say I’ve never not-known Ancestry Common Ancestors, ThruLines and Groups. That is, once I began to apply them.
My email explained to my sister that I had just started grouping my known DNA matches (mostly from Common Ancestors). I proudly attached my (meager) work product:
It wasn’t much, but it gave me a focal point from which I could start to show some progress and start learning. I was soon fully immersed in analyzing my DNA matches, having been baptized by some soon-to-be well-known-to-me luminaries in the field. In another email to my sister ten days later:
The Barefoot Genealogist (apologies to Crista Cowan for articulating my first impression, since tempered by acclimation), Blaine Bettinger, DNA Painter (Jonny Perl, who, to a newbie like myself, could be easily mistaken as “a whole organization”) … in hindsight I probably couldn’t have chosen a better group for my initiation into DNA and shared matches. Within a couple of weeks I came to appreciate others like Dana Leeds, Leah Larkin, Kitty Cooper, Evert-Jan Blom with Genetic Affairs (foreshadowing foot-stomp!), and several others on YouTube.
I also discovered FindAGrave.com. By searching the site for Latimer graves, I developed a sense for the concentration of where Latimer ancestors were buried. Searching both maiden and surnames who died before 1960, there were almost 200 Latimers forever residing in the counties northeast of Dallas. About eighty percent of those were concentrated in the three northern counties bordering the Red River – Red River, Lamar, Fannin – and Hunt, bordering Fannin to the south.
Not coincidentally, they overlapped nicely with Ella’s World.
September 26th and Genetic Affairs
For a variety of reasons, I had mothballed my Facebook account in 2014.
In September 2019, I reactivated it for one and only one reason – the Genetic Affairs Facebook Group.
Genetic Affairs is the creation of Evert-Jan Blom from the Netherlands. Like Jonny Perl, EJ Blom had a personal interest in his own family history, especially from a DNA standpoint. Whereas Jonny’s background was computing and programming, EJ has a “PhD in Molecular Genetics with a strong focus on Bioinformatics,” per his GeneticAffairs.com about page.
In 2018 he launched his first genetic genealogy tool, AutoClusters, that
groups shared DNA matches into clusters, helping users to better understand their genetic connections.
Once again, my timing was fortuitous in that by the time I realized I had a need, there was a solution waiting for me.
After learning about the AutoCluster tool from several YouTube videos, I joined geneticaffairs.com in order to run my own cluster scenarios. After connecting my Ancestry tree to Genetic Affairs, I could go onto the geneticaffairs.com website, enter the range of centimorgans I wanted to cluster, identify a family group or groups I was interested in, then press a button to order the cluster.
After a few minutes, I’d receive an email that included one HTML file with a folder of supporting data at the chromosome level. When you opened the HTML file, it displayed this mesmerizing animation that showed the actual clustering process. It’s hard to explain or show the animation, but here’s what the completed output looks like, anonymized:
The blurred left and top axes list all 76 matches I had in my tree with whom I shared from 50-250 cM. Each list is sorted first into the predicted group (or cluster), then sorted by most centimorgans shared to fewest.
Here’s the full Latimer group cluster, number 5 in brown, that includes 19 shared matches that meet the 50-250 cM criteria:
NOTE: The solid diagonal line of darker blocks simply shows each person as their corresponding match along each axis. The top axis isn’t shown in the lower image.
The fact that these nineteen matches are all grouped into one cluster suggests they all share at least one common ancestor. The matches might share one or both 3rd great-grandparents (4th cousins), or 2nd great-grandparents (3rd cousins).
The AutoCluster tool also provides tabular details about the nineteen matches such as that below. The number in the “people” column indicates how many people each match has in an attached Ancestry tree; “% Confid” indicates the degree of confidence the tool’s calculation has that the match belongs in the cluster – not whether the Range is THIRD_ , FOURTH_, half or removed COUSIN).
To me, this was all … interesting? I wasn’t sure how the pieces all fit together – did they all represent just one solution, i.e., one set of parents, or multiple pairs, or could they even be wrong for some reason? About the best I could do was to look closely at the matches, check their Ancestry trees (only four of which proved to be helpful), and start looking at the matches I shared with these nineteen and, as applicable, add them to my Latimer Group/dot.
If I knew then what I’ve learned since, I could’ve interpreted this differently. More on that later.
The most benefit I derived from the AutoCluster tool at the time was realizing that I needed to understand the ascending trees associated with as many of the shared matches as possible. And when I say understand I mean not only accept what’s in my shared-match-cousins’ trees, but possibly to even build out their trees solely for my purposes to reach their equivalent generation to Ella’s parents, my 3rd great-grandparent generation. And beyond if need be.
It dawned on me that, rather than solely searching for Ella’s parents, I was actually looking for an entire family – who were Ella’s DNA siblings?
Putting it all together …
On October 24th, I looked at my Ancestry ThruLines again. Ella was still parent-less at my 3rd great-grandparent generation. In the absence of a father and mother in my tree, ThruLines apparently had no basis to project who my 3rd GGPs might be. (Now six years later, this is not necessarily the case on Ancestry.) A confounding factor was that my cousins who shared Ella as our Common Ancestor all still showed Albert H. Latimer as her father and Mary Ann Gattis as her mother.
For the next couple of weeks I burrowed back into my Ancestry matches. I made an extensive list of shared matches, starting with the nineteen matches in my GeneticAffairs cluster. As I started doing my own manual ‘cluster’ (in the form of an Excel spreadsheet), it became obvious that my second and third cousins who shared Ella also shared DNA with a sizeable group of other matches.
Unfortunately, not many of those matches had trees that extended as far up the branches as I needed, but I had to start with what was known.
In my notebook I sketched out some of the trees of my “Cluster 5” shared DNA matches. I started on October 25th with these four (which turned out to be only the first three):
The Fran match was my best lead in that she had a tree that went up to Ella’s generation plus the next older – through Katherine Mary Latimer who married George W. Welborn. Katherine’s parents were shown as John Latimer and Elizabeth (Snoddy) Latimer. That sounded promising, and the information for Elizabeth was that she was born in Tennessee and died in Lamar County. Fran didn’t have much on John Latimer, however.
Next, tiffanynicole6 had what was likely the same couple in her tree although the names were slightly different - John Lattimore and Eliza Snoddy. The dates made sense and her tree indicated John was born in Arkansas and died in Red River County, Texas – next door to Lamar County.
Then Rebecca, who also descended from Eula Lee Welborn – same as Fran – showed a completely different set of parents for Eula Lee. Something was off.
The next day I sketched out my notes a little differently, trying to reconcile the generations for Fran, tiffanynicole and me (top half), then with Rebecca and a fourth match, Leah, on the bottom. Fran, tiffanynicole and Rebecca all descended from Katherine Mary Latimer (b. 1854 in Pike County, Arkansas). Leah, however, descended from Hiram Latimer (b. 1875 in Texas). Could Hiram and Katherine Mary be siblings, or more likely first cousins given the twenty-one year age difference?
I found two other matches that I similarly sketched, marygary and B.S., both of whom descended from James Robert Isom Latimer (JRIL, b. 1857 also in Pike County, Arkansas). So now it seemed as though the Latimer-Snoddy union had at least two, possibly three lines of succession.
Like the good folks from “Jaws” needed a bigger boat, I needed a bigger piece of paper. A roll of brown butcher paper worked nicely.
And about this time, as an experiment I added John Latimer and Elizabeth/Eliza Snoddy into my tree as Ella’s parents just to see what the consequences would be. (bold for emphasis)
Within a couple of days, I sketched out Ella’s known descendants on the bigger canvas. I noted the amount of centimorgans (cM) I shared with each second and third cousin, noted by the red dots and green numbers. Of seven siblings, I was fortunate to have shared DNA matches that descended from six. (The only sibling for whom I did not have any shared matches was, ironically, Ella Marion Henry.)
I had to wait for Ancestry’s ThruLines to propagate the effect of adding John and Elizabeth/Eliza as Ella’s parents. Once I had those ThruLine results, I manually added them on the butcher paper after doing a quick confirmation on each separate line of descent. Several of the ThruLines projected cousins overlapped with the GeneticAffairs predicted cluster names.
Again, the red dots and green numbers noted my own shared matches, although “SES” isn’t visible on this bottom part. ThruLines had identified two more children of Latimer-Snoddy, and also showed Hiram as a sibling, not a cousin.
On this first pass, I didn’t note that the first two Latimer-Snoddy children were born in Arkansas in 1857 and 1859. The other three were born in Texas in 1866, 1870, and 1875.
This was looking promising. Ella was born in November 1861.
November 2019
Now I was moving out of the discovery phase and tiptoeing into confirmation, verification and if need be, falsification. A first step was to manually cross-reference the matrix of matches, inspired by the Genetic Affairs AutoClusters.
Note that these relationships are independent of any tree; whether or not I had any idea where they belonged on the tree doesn’t affect the fact of these shared DNA matches.
*Caveat: This doesn’t necessarily mean that all identified matches share the exact same DNA – it just means they all share some common DNA. For example, although I haven’t yet proved it, I suspect my higher match with “LL” below (97/7) may be the result of shared DNA through a different family line.
In the top section, I show the shared matches of myself, my daughter and sister to our 2nd and 3rd known cousins who share Ella as a common ancestor. The numbers show the amount of cM and number of segments we each share.
You can see the evidence of random inheritance at work among myself, daughter and sister … I’m the only one who shares DNA with “LL;” my daughter and I share DNA with “AB” but not my sister; and my sister and I share DNA with “JC” but my daughter doesn’t.
The lower section shows thirteen matches with our newly-predicted 3rd and 4th cousins. Note that my sister and I each share measurable amounts of DNA with all thirteen, but not the same DNA. In fact, the only match we even share the same amount of DNA with is “JH” at 51 cM across 4 segments.
And moving over to our 2nd and 3rd cousins, each of the six shares DNA with at least three of the 3rd and 4th cousins. One, “LS” (with whom I’ve corresponded, we share the same great-grandparents) matches all thirteen.
Now I felt comfortable enough to formalize the butcher paper trees into print. On November 3rd, here were the results, this time with my actual cousin relationships annotated in green:
On November 8th, I ran another Genetic Affairs AutoCluster that produced similar results for the two highest-value matches among my 3rd and 4th cousins. This time, it projected (or confirmed, I’m not quite sure which) a tree relationship.
Everything seemed to fit. Ella’s birth in 1861 was entirely possible given the nine-year gap between Elizabeth/Eliza’s second child (James R.I., b. 1857 in Arkansas) and third (Julia Ann, b. 1866 in Texas, post-Civil War).
But I had some nagging questions that I might be missing or overlooking something obvious in my analysis, especially concerning my AutoCluster tool results and interpretation.
Inquiry to Genetic Affairs User Group on Facebook
Although I’m not a regular Facebook user, on November 11th I contacted the user group that is, or was, the go-to place for understanding the analysis tools at geneticaffairs.com. I introduced the “Ella issue” and then shared an abbreviated version of the merged pedigree diagram.
I asked if there was anything obviously wrong, or if there was any more analysis I could do with available tools to confirm Ella’s fit into this family.
I received two pertinent replies:
Commenter 1: I think you are making this too complicated. If you, JC, AB, LL, KM MC and LS all show up as common matches in most or several of the various match lists of KD, EJ, FJ RA CT ACA, AG, etc… AND not in any (or many) other cousin groups, then you are done.
Commenter 2: If you found 3rd cousins of MC and LS on the Snoddy side (descendants of Elizabeth’s siblings) that would be slam dunk.
I verified that was the case to Commenter 2.
Commenter 2’s reply: That’s pretty conclusive that you descend from Elizabeth’s parents, which is independent of John Lattimer (sic).
November 14th
As further confirmation, I ran another AutoCluster report. Below is a re-creation of the “Latimer” cluster. I’ve edited the left column to replace the names of shared matches with either “Ella Cousin-” or “J&L Latimer-” preceding the initials and shared centimorgans for each match. (Don’t ask me why I used J&L rather than J&E … that statute of limitations expired long ago.)
As conclusive as this seems, there’s still the lingering mystery of … why? what? who? when? where?
And as for the title metaphor … applying DNA tools can grow your tree, develop a broader network of related cousins, and help correct not-so-obvious errors in specific branches.
The story continues; the proof marches on. Thanks for your indulgence.
Postscripts
(1) I asked my only surviving maternal uncle to provide a sample to AncestryDNA. Results came back in December 2019 and provided even more confirmation of this analysis. Here’s my uncle’s ThruLine report of shared matches with Common Ancestors decending from Elizabeth Snoddy’s father, John Robert Snoddy (my uncle is the good-lucking young guy under Elizabeth H. Snoddy):
Note: J.R. Snoddy is my uncle’s 3rd great-grandfather, my 4th. And Ella’s inferred maternal grandfather.
As a bonus, I get a ThruLines prediction for my 6th great-grandparents on both my Latimer and Snoddy lines, as well as all other of my maternal lines.
(2) Six months after I completed the DNA portion of Ella’s story, Ancestry pulled the plug on allowing Genetic Affairs to pull data from Ancestry trees and match lists. Dana Leeds herself lamented the impact:
(3) But then just like that, six years later ... it doesn’t quite justify the 36 days it’s taken me to produce this post, but somewhat ironically, a fitting way to conclude it; from Dana Leeds’ 7 March 2025 newsletter:
Link to Ancestry video on YouTube announcing new Matches by Cluster. If the link doesn’t work, search on the Ancestry YouTube channel for Using Ancestry DNA to Power Family History | RootsTech Demo. The Clustering Tool is the last two minutes of the video.
SCREENSHOT ONLY - not a live link. But it looks familiar, doesn’t it? If there’s any justice in the genealogy world, Evert-Jan Blom should be banking it.
But did you marry your cousin?
I thought the clustering tool was very useful. I was pleased MyHeritage adopted it. I certainly hope ancestry gives credit to EJ Blom. I look forward to it being rolled out.