23andMe Genetic Tree Provides Critical Clue to Solve 137-Year-Old Disappearance Mystery

DNA can convey messages from the great beyond – from times past and people that died long before we were born.

I had the most surprising experience this week. It began with receiving an email with the sender name of my long-time research buddy, cousin Garmon Estes.

It’s all the more surprising because not only did Garmon never own a computer, despite my ceaseless encouragement, he passed over in 2013 at the age of 85. So, imagine my shock to open my email to see a message from Garmon. Queue up spooky music😊

As it turned out, Garmon’s nephew is also Garmon. I had communicated with the family off and on over the years since the death of Garmon the elder. Garmon, the younger, had written to tell me that the second “great brick wall” that haunted his Uncle Garmon had fallen – and how that happened, thanks to DNA.

Garmon, the Elder

Estes Garmon

Garmon Estes, the elder

I first met Garmon the elder, via letter, back in the 1970s or maybe early 80s. He was an experienced genealogist and I was beginning.

At that time, Garmon had been chasing the identity of the father of our common ancestor, John R. Estes, for decades, and I was just embarking on what would become a lifelong adventure, or perhaps it could better be called an obsession.

John R. Estes had moved from some unknown location to Claiborne County, Tennessee with his wife and family about 1820. That’s pretty much all we knew at that time. Garmon had spent decades before the age of online records researching every John Estes he could find. I can’t even begin to tell you how many John Esteses existed that needed to be eliminated as candidates.

Garmon lived in California, far from Tennessee. I lived in Indiana, then Michigan – significantly closer. He began caring for his ill spouse, and I began traveling to dusty courthouses, sometimes reading musty books page by yellowed page, extracting everything Estes. Garmon worked from his local Family History Center when he could and wrote letters.

Between our joint sleuthing and many theories that we both composed and subsequently shot down, we narrowed John R. Estes’s location of origin to Halifax County, Virginia. However, there were multiple John Esteses living there at the same time, about the same age, none using middle initials reliably, and some not at all. How inconsiderate!

I began perusing every possible record. I had eliminated some Johns as candidates, most often because they clearly remained in the community after our John had moved to Claiborne County. Late one night, in our local family history center, I found that fateful clue – John R. Estes noted as (S.G.) short for “son of George,” on just one tax list. All it takes is that one gold-nugget record.

It was after 10 PM when I left the Family History Center and even later when I got home. I debated whether I should call Garmon or not, but I decided that indeed, he would want to know immediately, even if I did call at an inconvenient time or wake him up.

The discovery of John’s father, of course, opened the door for much more research, and it solved one of Garmon’s two brick walls that had haunted his genealogy life.

He never solved the second one, but it wasn’t for lack of trying.

What Happened to Willis Alexander Garmon Estes?

Willis Alexander Garmon Estes was born on December 21, 1854, in Lenoir, Roane County, TN. His nickname was Willie.

Willie married Martha Lee Mathis in 1874 and they had 4 children beginning with the first child born the next year in Roane County. Sometime between 1875 and the birth of the second child in 1877, they migrated to Greenwood, Wise County, Texas where their next two children were born in 1877 and 1881.

Martha was pregnant for their fourth child in 1883 when something very strange happened. Willie disappeared, and I do mean literally and completely. Just poof, gone.

Not sure what to do, Martha’s father, who lived in Missouri, went to Texas to retrieve his pregnant daughter and her children and took her and the children home to Missouri where their last child was born that September.

Willie was only 28 when he vanished. The family, of course, had many stories about what happened. Texas at that time was pretty much the “wild west” and the stories about Willie reflected exactly that.

Texas was sometimes the refuge of outlaws and shady characters. One story revealed that Willie had shot a man back in Tennessee and the family fled to Louisiana, then Texas. Of course, that doesn’t tell us why he disappeared in Texas, but it opens the door to speculation and casts doubt on his character, perhaps.

Another story was that he was shot by Indians.

A third story stated that Willie settled in Indian Territory north of the Red River, now Oklahoma, and that he had an altercation with an Indian over the supposed theft of firewood, although who was accusing who was unclear. Willie shot the Indian, then had to flee for his life, leaving his pregnant wife and children as a posse of Indian Police surrounded his house. Willie supposedly promised Martha that he would return, but never did. It was reported that he was shot in Mexico, but no further details emerged.

Aren’t these just maddeningly vague???

Yet another story was that Willie headed for the goldfields of California, struck it rich, and was murdered on the way back home. The details varied, but one version had him murdered by a traveling companion on the trail. Another had him becoming ill and dying in a hospital in St. Louis where his wife went to search for him, to no avail. That might explain why she went back to Missouri, Garmon postulated. And yet a third version was some hybrid of the two where “someone” tried to find Willie’s family for years to reveal what had happened, and where, but was never successful. Of course, how did the family know about this if the mystery person was unable to find the family? But I digress.

Garmon desperately wanted to solve that mystery. He wanted closure.

I didn’t realize that the genealogy bug had bitten Garmon’s nephew too, but it clearly has. Garmon would be so proud.

With Garmon the younger’s permission, I’m publishing “the rest of the story,” Connecting the Dots, as written by Garmon the younger, with a few technical interjections from me involving DNA from time to time.

Connecting the Dots

In 2015, My dad Richard Estes, my brother Corey Estes, and I took a trip to Texas and Oklahoma to see if we could find out more about Willis Alexander Garmon Estes’ disappearance.

Estes greenwood

We visited Greenwood, Texas and nearby Decatur where we looked at historical records at the Wise County Clerk Office. We also went up to Oklahoma City to see the state archives and to Tishomingo to look at any records that might be available.

Estes Oklahoma history.png

Interestingly enough, we did not find any clues as to the disappearance of Willis Alexander Garmon Estes. There were no newspaper articles or criminal records concerning any incidents with Willis Alexander Garmon Estes. The only new information that we found was a couple of land deeds showing that Willis Alexander Garmon Estes’ brother Fielding had bought and sold land in Wise County during the time that Willis Alexander Garmon Estes was living in Greenwood.

We left empty-handed on our trip but our curiosity remained strong and we began talking to each other about going on another trip to Tennessee to speak with Estes family members in Loudon County to see if they might know something about Willis Alexander Garmon’s disappearance.

DNA Testing

In December of 2018, my wife, children, and I had our DNA tested using the service 23andMe. We received test results within a month of sending in saliva samples. The results did not reveal anything unusual.

Fast forward to October 2019. 23andMe introduced a new Family Tree feature that automatically creates a family tree based on the DNA results that you share with relatives in 23andMe. This was a fascinating feature and I noticed that all of my family members were automatically placed into the correct position on the family tree without me having to do anything.

[Roberta’s note – this is not always the case, so don’t necessarily expect the same level of accuracy. The tree is a wonderful innovative feature, just treat family placement as hints and not facts.]

Every few weeks as more and more people had their DNA tested on 23andMe, new relatives were added to the family tree.

In February 2020, I noticed something interesting under the location of Willis Alexander Garmon Estes on the family tree. A woman by the name of Edna appeared as a descendent of Willis Alexander Garmon Estes. The first thing I did was to try and get in contact with her on 23andMe. No luck. Next, I thought maybe she was the descendent of one of Willis Alexander Garmon’s sons (James, John, or George). However, after researching the descendants of each of those lines, Edna’s name did not appear.

The next step I took was to look up as many Ednas by that last name on ancestry.com as I could find and trace their ancestry back to see where it led.

There were two Ednas by that last name in the United States whose age matched the one on 23andMe. I traced both of their ancestry lines back to the 1800’s. Neither one had Willis Alexander Garmon Estes as an ancestor.


During the middle of March 2020, when I was quarantined at home from work due to the COVID-19 virus, I took another look at Edna’s family lines. I noticed there was a gentleman by the name of James Henry Houston mentioned as an ancestor.

The interesting thing about James was that he was born on the same day, same year, and in the same county as Willis Alexander Garmon Estes. James Henry Houston was born on December 26, 1854 in Loudon County, Tennessee. This seemed like possibly more than a coincidence, so I dived into the data a little bit more.

I looked at federal census records to find out more about James Henry Houston’s past. Strangely there were no official records of him until May 12, 1889 when he married Allie Ona Taylor in Erath, Texas. Normally, if someone is born in 1854, they would show up in one of the federal census records of 1860, 1870, or 1880. James Henry Houston does not show up in any official federal census records until 1900.

According to ancestry records, James Henry Houston married Allie Ona Taylor in 1889 and resided in the Hood County region of Texas until 1910. During this time, he raised 8 children with his wife Allie.

In 1920, the federal census placed him and Allie in Whitehall, Montana. The last federal census he appears in is 1930. He lived in Pomona, California where he died in 1933 at the age of 78.

At this point, I thought it was highly likely that James Henry Houston and Willis Alexander Garmon Estes were the same person. If my hunch was correct then a photo of James Henry Houston would most likely show a resemblance to his son, my great grandfather John Alexander Estes.

Estes James Henry Houston

The photos above show a remarkable similarity in the eyes, nose, mouth, and facial structure between the two men. To me, the photo and historical evidence is enough to conclude that Willis Alexander Garmon Estes is James Henry Houston.

Garmon’s Concluding Thoughts

As I reflect on the fact that Willis Alexander Garmon Estes renamed himself James Henry Houston and moved from Wise County down to Hood County, Texas – approximately 60 miles distance to marry and raise a new family, many more questions come to mind.

What exactly happened to cause Willis Alexander Garmon Estes to leave his wife and children behind? Was it simply a marital dispute or did it involve a criminal offense and running from the law as was mentioned in the family lore?

Did my great grandfather know that his father lived in Pomona in 1930, which was only 6 miles away from where he was living in Rancho Cucamonga? Were there other family members that knew what happened but promised not to tell anyone else? We may never know.

Finally, I want to add one more piece to the story that I found fascinating. On ancestry.com, many of the family trees for James Henry Houston state that the mother and father of James Henry Houston was Jennie Bray and Henry Houston. No information is given for their birthdates or where they came from. The mother and father of Willis Alexander Garmon Estes was Jennie McVey and William Estes. The names Jennie Bray and Jennie McVey are very similar. In order to hide his true identity, James Henry Houston would have to make up a surname for his father since he called himself Houston, not Estes. Willis Alexander Garmon Estes had a brother named John Houston Estes. This might explain why James Henry Houston chose to use the surname Houston rather than another name.

Congratulations Garmon

I know this made Garmon the elder puff up with pride for Garmon the younger’s sleuthing skills and leap for joy at the solve. Garmon, the elder, had two main genealogy goals throughout his entire life. One was solved while he was living, but it took another generation to solve this one.

Great job, Garmon!

About the 23andMe Genetic Tree

23andMe is the only vendor to construct a “trial balloon” genetic tree based only on how the tester matches people and how they do, or don’t, match each other. This occurs with no input from testers in the form of genealogical trees of identifying how people are related to the tester.

Family Tree DNA has Phased Family Matching, MyHeritage has Theories of Family Relativity, and Ancestry has ThruLines which all do some sort of DNA+tree+relationship connectivity, but since 23andMe does not support user-created or uploaded trees, anything they produce has to be using DNA alone.

On one hand, it’s frustrating for genealogists, but on the other hand, there is sometimes a benefit to a different “all genetic” approach.

Of course, the only information that 23andMe has to utilize unless your parents have tested is how closely you match your matches and how closely your matches match each other. This allows 23andMe to place your matches at least in a “neighborhood” on your tree, at least approximately accurate, unless your parents are related to each other and that shared DNA causes things to get dicey quickly.

I wrote about 23andMe’s new relationship triangulation tree when it was first introduced in September 2019, nearly a year ago, here. The launch was rocky for a number of reasons, and if you’ve done genealogy for a long time, your research goals are likely to be further back in time than this 4 generation relationship tree will reveal.

23andMe tree

Click to enlarge

This is what my relationship tree looked like at the time the function was launched. You’ll note that 23andMe places relationships back in time 4 generations, to your great-great-grandparents, meaning that you might have 3rd or even 4th cousins showing up on your genetic tree.

I initially had a total of 18 people placed on my tree, with 3 being close family, 4 being accurate, 4 unknown, 1 uncertain and 6, or one third, inaccurate.

Keep in mind that 23andMe doesn’t make any provision to accommodate or take into account half-relationships, like half-brother or half-sister, either currently or historically. Therefore, descendant placement predictions can be “off” because half-siblings only carry the DNA from one common parent, instead of two, making those relationships appear more distant than they really are.

In Garmon’s case, his great-great-grandfather is the ancestor who was MIA, so the genetic tree has the potential to work well for this purpose.

Estes 23andme tree today

click to enlarge

Today, my tree looks somewhat different, with only 14 people displayed instead of 18, and 6 waiting in the wings to see if I can help 23andMe figure out how and where to place them.

Since the initial launch, customers have been given the opportunity to add their ancestors’ names to their nodes. This works just fine so long as nobody married more than once and had children from both marriages.

Estes Willie Alexander today

click to enlarge


Here’s a closer image of the left-hand side of my tree where I’ve super-imposed the location of Willis Alexander Garmon Estes and Edna, as they are related to Garmon the Younger, at bottom right. Ignore the other names – I only utilized my own tree for an example tree structure.

One more generation and it’s unlikely that 23andMe would have made the connection between Edna and Garmon the younger.

Not only does this illustrate the perfect reason to test the oldest generations in your family, but also never to ignore an unknown match that seems to be within the past 3 or 4 generations. You never know what mysteries you might unravel.

Four generations actually reaches back in time quite substantially. In my case, my great-great-grandparents were born in 1805, 1810, 1812, 1813, 1815, 1816, 1818 (2), 1820, 1822, 1827, 1829, 1830, 1832, 1841 and 1848.

If you have mysteries within your closest 4 generations to unravel, the genetic tree at 23andMe might provide valuable clues, but only if you’re willing to do the requisite work to figure out HOW these people match you.

You can’t transfer your DNA file TO 23andMe, so if you want to have your results in the 23andMe database, you’ll need to test there.

Acknowledgments: Thank you to Garmon Estes, the younger, for generously sharing this story and allowing publication. My heart was warmed to see your generational research trip.

Thank you to Garmon Estes, the elder, for being my research partner for so many years. You can finally RIP now, although somehow I suspect you already have these answers.



I receive a small contribution when you click on some of the links to vendors in my articles. This does NOT increase the price you pay but helps me to keep the lights on and this informational blog free for everyone. Please click on the links in the articles or to the vendors below if you are purchasing products or DNA testing.

Thank you so much.

DNA Purchases and Free Transfers

Genealogy Products and Services

Genealogy Research

39 thoughts on “23andMe Genetic Tree Provides Critical Clue to Solve 137-Year-Old Disappearance Mystery

    • Yes, and that’s certainly possible with their chromosome browser. I would also paint it with my other DNA at DNAPainter too, but they aren’t doing that.

      • Roberta, if there is someone who tested at 23andme but is not showing up on your match list. You can add them by using the email address they are using on 23andme. This works for those who will match you or those that do not. For example I had a match on ancestry that test at 23andme but not on my match list. I sent an invite using her email address and now she is on my match list. In the beginning you could search the db of all testers. So you could contact and add people who had a common surname. So if you are working on a surname project or for all descendants of an ancestor you can add them. You can then compare to others that you are sharing with.

    • I assume 23andMe’s placement of DNA relatives on our trees there is based on triangulation. I don’t have any DNA matches on my tree that do not triangulate with at least one other match on the same branch. I do have a few matches on my tree whose connections still need a research breakthrough (they aren’t from the US so it’s harder to trace them, and they don’t appear to have Ancestry trees).

  1. Fascinating –

    I am still trying to figure out where some of my 23andme people go on that tree who have already been auto-placed. One person had to be moved from my dad’s branches to my mom’s because I have no parents on 23andme so a swap must have occurred – as you experienced – but I already expected that was an issue by looking at where the placements were for some ethnicity, wrong side of family on chromosomes

    I also have a great grandpa who went missing – yes a “poof” but I also wonder if some were keeping secrets, and I have suspects that I can’t quite place on my tree.

    This story makes me hopeful that I will one day find him – a traveling salesman he was


  2. I love this story! It resonates with me as my grandmother had a full (brother) born about 18 months after her. For years I wondered what happened to him as there was no death certificate and he never appeared on a census. All we had was a birth certificate. My great grand parents separated and my gg grandmother actually ended up raising my grandmother and my g-grandmother remarried and had a new family.

    But what happened to William? No one [living] knew he had even existed until I found the birth record. My grandmother, long since deceased, had never mentioned a brother and it was news to my father as well. When my father and I matched on ancestry with someone who, by cM estimates, appeared to be my father’s first cousin, I contacted them and was finally able to figure out that he was the son of my great uncle (my grandmother’s long lost brother we never knew existed). My missing baby mystery wasn’t 130+ years (!) in the making but it but it may have ended up being that without the DNA. The Gamon Estes story really makes you appreciate the legwork that goes into genealogy and I admire everyone’s perseverance. I always feel a little guilty at how ‘easy’ it is for relative newcomers like myself.

  3. When I was the list admin for the Rootsweb MOSES mailing list, I had a subscriber who kept asking repeatedly about her ancestor. Eventually I decided to try my hand at searching for information. My first suggestion to her was one I later realized was false because the person in question had married a different person as I later discovered. But the second time I looked, I found a different person who seemed to match most of the criteria she was looking for (Birdie Moses, married a man named Archie, died shortly after giving birth to a son who was the subscriber’s ancestor). What I discovered was that Birdie’s husband Archie had a different last name (Clay rather than Gibbs), and the son’s name was Gerald Clay rather than Harry Gibbs. The descendants of Birdie and Archie’s other children had been searching for Gerald Clay for many years because he just disappeared like you say above. It turned out that Gerald Clay and Harry Gibbs were one and the same. I put my subscriber in touch with the descendants of the other children of Birdie and Archie. They exchanged pictures, which clenched it.

    I have two cases in my own tree of men having two families with neither family knowing about the other family, at least not at first anyway. One was my own grandfather. He had a family in Montreal, Canada, then abandoned them, came to the US by way of Cuba, stayed here, and then lied about his date and place of birth for the rest of his life — at least on official records anyway. My father knew that his father was likely born in Montreal, but none of the records (census, social security, birth records for his children, his US marriage record, etc.) show that — except of course for my grandfather’s baptismal record in the Drouin collection.

    The other case was my grandfather’s first cousin who immigrated to Australia, initially Tasmania, where he had a wife and child using his actual name (surname Dresser). Then he abandoned that family and later married another woman in New South Wales, and had several children with her there, all using a different surname (Dymock). At one point his first wife became aware that he was still living and using a different surname. As his first wife had essentially married another man in the meantime, they were both charged with bigamy. One of the daughters of the second marriage needed to get a copy of her birth certificate as an adult, but couldn’t get one because there wasn’t one under her name. I don’t know how she and the clerk were eventually able to determine that her birth certificate was filed under the surname Dymock rather than Dresser (to which he returned after the bigamy charge).

  4. 23andme has made more changes to their tree. Now they are giving me a list of family members that they are asking me for help in placing them in the tree. And a lot of information and instructions on how to make changes to the tree.

  5. There is a post in this group about 23andme today. This is what some it says

    23andMe is rolling out a premium subscription service for access to additional health reports (pharmacogenetics) and enhanced genealogy features (ability to access four times as many matches). At the moment it’s invite only.

      • I just need the triangulation tree to go back one more generation than it does. I realize they probably stopped where they did because things begin to be less … stable?… after that, but I am searching for my last unknown 3rd great grandparent – who’s name appears no where on any document ever for for my 2nd great grandmother. Maybe if I find out who he was, I’d find out what happened (I don’t think my 3rd great grandmother ever married him, and I don’t know if my 2nd great grandmother’s sister was a full sister or a 1/2 sister or what even happened to her).

        More matches would be nice too, but (amazingly) I haven’t topped out the 2000 yet (though I’m starting to get closer now I think).

      • because the sub includes health features then not surprised it would need to be v5. I am disappointed about the 5k limit though. If I pay for something I want more. for a db that is supposed to be 2nd largest behind ancestry then I would expect to get more matches. I look at the number of matches I have at familyltreedna and myheritage. more than 5k.

  6. That’s fascinating and gives me hope. We have never been able to figure out what happened to my great-grandfather. There are several family stories just like in the Estes case. In our case, my great-grandfather was not an upstanding citizen. My second great-uncle was the sheriff in Wise County twice. He married in Hood County where his mother lived. He was later a judge.

    The photo on the right just looks like an older version of the man on the left. No doubt that is the correct ancestor.

    • Roberta,
      Just WOW! That is awesome and this article definitely a great one for sure. Definitely tells you something about chipping away at the brick walls and finding long lost ancestors in using DNA and along with paper trails. Just awesome.
      I did notice that you mentioned Willis’s first wife whose maiden surname was…Mathis, well I happened to have that surname in my own Family Tree as well. My ancestor was Conrad M Mathis who fought in the American Revolution for state of Maryland and sometime after the war migrated to Kentucky. He was a 5th Great Grandfather. Very interesting aspect wouldn’t you think 🤔? Who knows I might be related somehow. Anything is possible…lol
      Well definitely loved hearing about this and reading this article definitely makes we want to dig little deeper into the Mathis lines more thanks again for awesome read.


      Cindy Carrasco

  7. I love this entry. I found it fascinating about this story and seeing how he pieced it all together. Thanks for sharing his story and great job!!

  8. Roberta,
    My Great grandmother died after her two sons were born and my great grandfather re-married and had four more children. I was able to add the second wife and move my half-second cousin to the correct great grandparents. It appears 23andMe does not consider half relationships on their initial tree, but they do provide options where the user can adjust the tree adding second spouses and moving people to correctly depict half relationships.

    • I came here to report that I was able to add my second great-grandfather’s two wives. He has many descendants from his first wife (and many who have tested at 23andMe) and very few from his second wife (including me) so I am able to fit in several half third cousins on my 23andMe tree.

    • I was thinking Red River and Oklahoma, because that’s where my Estes guy wound up, and Oklahoma wasn’t a state. Mistakes happen, really wasn’t “letting facts get in the way.” I don’t do that.

      Thanks for the heads up. It’s fixed.

  9. I knew a Garman, not Garmon, who was from northwestern Arkansas. Do you suppose this is the same family? They were close friends with the Taylor.

  10. Do we wonder if our ancestors who “misbehaved” would have done so if they had any inkling they would be “found out” decades down the road, and their personal history would be re-written ? Thomas Jefferson comes to mind……..

    Roberta, thank you for the article.

  11. Another amazing story! Thank you for this. I love 23andMe’s predicted tree as well. I was amazed at how accurate it was. I think people sometimes nitpick too much at small differences. A 2nd cousin vs a 1st cousin once removed is not that much of a difference. If it said 2nd cousin for someone that is a 4th cousin or further that may be cause for concern. But it offers me that brilliant drop down menu on my matches profile to specify the exact relationship if known. I love it! So far as my elusive Estes distant relation, I am still searching. I am over 80 matches with Estes in their trees. I have color coded them their own group. I have exactly one person with the surname Estes on each: Ancestry, Family Tree DNA, and 23andMe. I will persevere like Garmon until the answer is solved.

  12. Dear Roberta, I was doing some family tree research on Ancestry about my grandfather James Cecil Huston when I stumbled on a link that led me to this article. The picture of Garmon Estes the elder took my breath away – he could be my grandfather Cecil’s twin brother! You have solved the century-old mystery of my great-grandfather James Henry Huston’s past!!!

    In 1938 Allie Ona Huston moved in with her son Cecil and family. My mother, Valerie Huston, told me that when she was a teen she asked Grandma Huston how she met her husband, James Henry Huston. Allie told her when James rode into her small Texas town on a horse, she said to herself “He’s mine if I never have him.” Apparently it was love at first sight on her part. They married when she was 17 and he 34, but he made her promise to never ask him about his past. Our family always wondered if he had killed a man back in Tennessee or left a wife and kids back there. All we ever knew was that he was from Loudon, Tennessee. Allie kept her promise, but she told her curious granddaughter that she was free to investigate the mystery. Living in distant California, Mom never did.

    Thank you so much for solving this family mystery!

    • Sorry, I haven’t looked at the site for over a year. Yes, we are second cousins, I believe. My grandfather was Cecil Huston, Willis A. G. Estes’s youngest child.

  13. “Not sure what to do, Martha’s father, who lived in Missouri, went to Texas to retrieve his pregnant daughter and her children and took her and the children home to Missouri where their last child was born that September.”

    Martha Mathis is my ancestor. I have her father as Stephen Carroll Mathis, who died in 1861 during the Civil War. Do you have different information?

  14. Pingback: 23andMe and GlaksoSmithKline Partnership Ends, Sparking Additional Layoffs | DNAeXplained – Genetic Genealogy

Leave a Reply