All That ThruLines Is Not Gold

I began to notice a subtle pattern emerge when looking at cousin matches presented by Ancestry’s ThruLines for mine and other families I have access to view. It stands to reason that any technology that relies on algorithms and other people’s trees would be vulnerable to errors.

I kept asking myself how many cousin matches is enough to accept that an ancestor is correct? Notice I’m not saying prove an ancestor is correct, but accept that an ancestor is correct. The only lineage that can be proven with DNA alone is parent and child. For all other lineage relationships DNA can only support other data. This is especially true when we start working with 4th and 5th great-grandparents.

The time finally came to look at the Lewis line. I’m not overly familiar with this line and haven’t done a ton of research so am looking for clues where to look in the ThruLines results for me and my uncle.

The Lewis surname first appears with my 3rd great-grandmother Martha G Lewis on my father’s maternal line.Lewis, Martha ThruLines

In addition there is evidence of a DNA match through her son James Hamilton Cox. Records support these connections so I feel comfortable with them. Records also support her parents being Richard Lewis and Catherine Jewell. ThruLines for Richard Lewis look like this:

Lewis, Richard ThruLines

Polly, Louisa B and Serenico A are all provable siblings for my Martha G. so these results are easy to accept.

Thus far the only thing I’ve seen suggesting the father of Richard Lewis is one anecdote online stating his name was Thomas Lewis. Here’s what ThruLines proposes for a Thomas Lewis:

Lewis, Thomas ThruLines

I’ve been through all these trees and found nothing that looks like proof of this relationship. I think Martha and Richard’s matches clearly illustrate that I inherited what I can label Lewis DNA. Some Lewis family somewhere contributed DNA. Do I accept that the Thomas Lewis I’m being offered is correct? Given only these results my answer is not yet. What I see here are several people descended from some Lewis ancestor who may or may not be this man.

This is where my uncle’s results come into play. He has another generation of results to view that I don’t have. On these trees the wife/mother that is offered is Mary Bandy. Neither of us have any matches to her, but what he has are matches to the man being offered as her father: Richard Bandy. Here are six of the ten siblings shown as offspring of a Richard Bandy:

Bandy, Richard ThruLines

It would appear that my uncle has both Lewis and Bandy DNA. I saw that a daughter of a Richard Bandy is named in his Will (not shown here) as “Mary Lewis wife of Thomas Lewis”. I’m not offering it as proof because I haven’t done enough work to satisfy myself that these are the people who belong to that document. For now I’m willing to add a Thomas Lewis and Mary Bandy as parents of Richard Lewis to my tree to facilitate further research. So while Richard Bandy might be the father of Mary Bandy, who is the father of Thomas Lewis in the ThruLines universe?

Lewis, Thomas Sr ThruLines

The Honorable Thomas Lewis. Pretty cool guy…hung with George Washington, is in the DAR database as a Patriot…sweet. Right? Love me some Patriots. While perusing the DAR database I noticed something odd. There are more than 150 women who have him as a Patriot and not one of them joined using the child Thomas. The first person who joined using this individual was in 1891! There have been ample opportunities for someone to join using the son Thomas. Why hasn’t anyone done it?

Welp. I found a book on Genealogies of the Lewis and kindred families, edited by John Meriwether McAllister and Lura Boulton Tandy, published by the E.W. Stephens Publishing Company of Columbia, Missouri printed in 1906. Here is an excerpt from page 180:

Lewis, Thomas Sr genealogy

Sorry, y’all, but Thomas Lewis is not the son of the Honorable Thomas Lewis. But, but, the DNA. This is why I keep asking how much is enough? My uncle shares DNA with four of the proven children of this Thomas Lewis.

Here’s the thing: based on the above mentioned book, the Lewis family was quite large. That Thomas Lewis had brothers who had a boatload of children, too. That DNA is bound to show up in a lot of people, some of whom can readily trace their lineage to the Honorable Thomas Lewis (whose brothers were more famous than him, by the way). For all we know our Thomas Lewis could be a by-blow of any one of those brothers, or even Thomas himself. What we can’t do is prove where that DNA originates. We can guess it comes from that family though, and that is intriguing and might be a fun project for someone more skillful than me.

I’ve seen this in other people’s ThruLines, too. Questionable matches that don’t necessarily belong to the people they are assigned. I think it’s wise to guard against being lulled into complacently accepting matches simply because multiple descendants of siblings appear in our results. This is especially true for the 4th/5th great-grandparents. DNA is a good deal more a finding aid than proof.


Sprinkles on a ThruLines Cupcake

In my previous post about the Haussler immigrant family I cobbled together using records, I speculated that my Francis X Haussler had two brothers: Andrew C and Leonard Haussler. There are DNA cousin matches to descendants of at least four of Andrew’s grandchildren. It pays to check ThruLines often because the results change regularly. Like today when I checked the ThruLines profile for Frederick Haussler I found this:Haussler Update

A cousin match to someone descended from Leonard Haussler! My half first cousin has two more matches on his results. So now we share DNA with descendants of three men who might be related in some fashion; they might be brothers, or at the very least they are cousins. One more data point to put in the column that supports my hypothesis that this is the original immigrant family.  I can be confident that I would not have found these connections without this match engine. So thank you, ThruLines.

This made me happy, so I’m guessing there were some endorphins involved.

Hypothesis + ThruLines = ?

I have the luxury of being able to spend a lot of time playing with Ancestry’s new beta ThruLines. I get to see how it is evolving on a daily basis. Doing so is helping decipher the data in more careful terms. One of those careful terms is ‘work in progress’. There is a tendency to take the results we see as fact when in reality ThruLines is changing too rapidly to weight the results that heavily. This tool is still in the experimental phase. That said, the results still merit attention along with a healthy dose of skepticism. Rushing toward the hit of endorphins our brain supplies us with whenever a belief is confirmed is a sure way to get tripped up by the data presented to us by this match engine.

The farther back in time we go, the more the term hypothesis applies to our research. For me the best way to test a hypothesis with ThruLines data is to choose a line I don’t have an emotional attachment to; one I already labeled as speculation.

Haussler, Frederick profile

Frederick Haussler appears in the New York, Passenger and Immigration Lists, 1820-1850 on Ancestry as having arrived in New York on 23 Jun 1841 on the ship Rose, place of origin is France and he was listed as being aged 49. The same date and ship lists a Leonard Haussler, male aged 21 and a ‘Mann’ Haussler, female aged 16. My speculation is that ‘Mann’ equals M. Ann. Very few other passengers on this arrival were from France; the vast majority were from Deutschland. The only indicator that they might be a family is they share a Family Identification number: 30049159.

The New York Public Library’s Digital Collection has many city directories available to search. Here is Frederick in the 1849 issue.

Haussler, Frederick 1949

Haussler, Frederick 1852
Frederick and Leonard 1852 NYC directory
Haussler, Francis X 1855
Leonard and Francis X (my ancestor) 1855

I have not found Francis X in any passenger list, but he lived long enough to answer the immigration question in the 1900 census (with his 2nd wife) for Secor, Woodford, Illinois, Enumeration District 133.

Haussler, Francis X 1900 cropped

Leonard Haussler was born about 1820 in France. Francis X was born 1824 in France, was a cabinetmaker and arrived  in 1841. Francis X has the claim that both his parents were born in France.

Shortly after appearing in the 1855 NYC directory, Leonard joins another Haussler household in upstate NY.

Haussler, Leonard and Andrew 1855


Andrew C. Haussler is three years younger than Leonard, also born in France. Leonard has been at this location for 2 months and is a shoemaker (bootmaker in NYC); Andrew has been at this location for 6 months and is a watchmaker. Take note of Andrew’s daughter Fanny. Are Andrew and Leonard brothers or cousins? There exists a photograph of Andrew with two of Leonard’s sons apparently as part of a musical trio, but that isn’t proof. Is Francis X a brother, a cousin or is it merely coincidence that they’re all from France, lived near one another in NYC and came to this country in the same year?

ThruLines has a hint that at the very least there is a relationship of some sort that goes beyond coincidence.

Haussler, Frederick ThruLines 01

Haussler, Frederick ThruLines mine

There’s Fanny and four of her known children, all of whom have at least one descendant with a cousin match to me and similar results appear for my half first cousin. ThruLines is suggesting that those cousin matches are related to us through our shared Haussler lines.

Does this prove Andrew and Francis are brothers? No. Does this prove that Frederick is their father? No. What statement can I make about these results? I feel confident saying the following: There are five people with whom I share small bits of DNA that the ThruLines tool has placed in the Haussler family suggesting a relationship between Francis and Andrew.

No endorphins for me. This doesn’t confirm my hypothesis. It suggests my hypothesis, in the strictest sense, has not yet been disproven. It’s one more data point that can be placed in the likely column along with birthplace, age, immigration year and proximity. Unfortunately, Andrew didn’t live to be included in the 1900 census, so it might never be known when he arrived. Francis and Andrew were probably related in some way but the DNA can’t tell us how they were related. I cobbled together a possible immigrant family using records and I might have been right, or at the very least on the right track.




ThruLines Quick Update

Yesterday I watched this video from RootsTech 2019 on the cool things Ancestry is doing with DNA results. From it I learned that they read all the comments people make in the feedback from their beta testing. Well, I thought, I have something that needs fixing, so let’s test that.

My g-g-grandfather Jacob Myers was adopted by his step-father Daniel Myers. His biological father was John Naylor. Close to 50% of all the matches for me and my uncle are Naylor/Carpenter, so it was frustrating to me that ThruLines kept showing Daniel Myers as Jacob’s father instead of John Naylor. I clicked on the little link at the bottom of the page where we can give feedback on Beta testing and filled out their four-slide survey. On two of the slides there are boxes where text can be entered to augment the answers to the Strongly Disagree > Strongly Agree choices for questions in the survey. On the last slide I explained the problem, mostly just hoping it would help them improve their algorithm for choosing parents. I also mentioned that my William Wood kept showing up as a potential ancestor from other people’s trees even though he’s on my tree. I submitted the survey and gave it no more thought.

This morning I checked ThruLines (I do that every day because it’s not static, it changes) and lo and behold John Naylor was shown as an ancestor! It gets better. Now that they have the correct ancestor, John Naylor’s parents showed up as well and BOOM my uncle has 56 possible matches spread out over 10 children and I have 37 over 6 children!

AND William Wood is a real ancestor now.

All this happened in less than 24 hours. Now that is one amazing customer experience.

ThruLines Thoughts

Ancestry recently introduced a new feature for visualizing DNA cousin matches called ThruLines. Rather than spend time explaining how it works, I recommend reading this blog where the author shows how she broke through a brick wall using this feature.

ThruLines is an excellent tool, but like all tools has its limitations. The results are shown as a redux of the existing tree on Ancestry, with plenty of ‘potential ancestors’ sprinkled in to fill the gaps. The relationships are visualized in a grid from parent to 5th great-grandparents, which is the limit of how far back we can get with DNA.

Something I noticed right off is that they added half relationships: half-cousin, half-aunt, etc. That is an improvement over the profile version of the tree. Sometimes the half-relationships are correct, and sometimes not so much. Because the data is dependent upon trees, the genealogy still has to be done to benefit from this new data.

Here is an example of half relationships that are correct.ThruLines Half relationships 01

Angeline, Jeannette and James are all children of different wives of Jacob Myers, so they’re all half-siblings. That is reflected in their stated relationships to me of half-great-aunts.

This is supported by the visualization of James’ mother Amarilla Cox. The other two half-sisters don’t appear as her children.ThruLines Half relationships 02

Next is an example of a missing half relationship.

ThruLines Half relationships 03

Margaret Jane Smith is the daughter of John Clark Smith and his second wife, where Maria Ann and Anna Marie are the daughters of his third wife.

This example shows a half-relationship that can’t be explained and shouldn’t exist. To adequately show this I’m putting John D Allen and his wife Martha Clark together to show how Francis Clark only appears on John D Allen but not on Martha Clark

ThruLines John D AllenThruLines Martha Clark

The problem I have with this is that the descendant of Francis C Allen only shares 7 centimorgans. How is it possible for this new system to determine if those 7 cMs came from the father or mother? I check these every day. At first Francis was appearing then disappearing then reappearing, sometimes as a half-relation and sometimes as a full relation, but has settled into the half-relation and no longer appears on Martha Clark’s results. Clearly it would require many more matches to sort this out, but the variability has to be taken into account. If half relationships were always correct, then I’d put some stock in this result. The system still has glitches that need to be sorted out.

Here is another half-relation that ThruLines get right, which is surprising because it’s so nuanced.

ThruLines John Holmes

Sarah and William Holmes are offspring from John Holmes’ first wife and Lourena Holmes is from his second wife. To go back that many generations and still get it correct is impressive.

Here is another nuanced relationship that ends up not being correct.

ThruLines Jeriah Wood

Jeriah Wood is most definitely the son of John Wood; that’s not in question. And the descendant listed for him is very probably descended from Jeriah Wood. The problem is that when I click on that person’s DNA profile and look at the shared matches, the DNA shared between this person and my uncle is actually from my uncle’s maternal line, not his paternal line. ThruLines currently lacks the ability to distinguish, which is why I question the above issue with Francis Clark. If this system can’t detect that her 17 cMs are from my uncle’s Cox (maternal) line, then how can it know if the 7 cMs from the descendant of Francis Clark is from the father or mother?

Like with so many of the amazing tools we now have to use online to assist us with solving our heritage, nothing takes the place of old-fashioned hard work and critical thinking skills. ThruLines is an excellent finding aid. To Ancestry’s credit, they don’t claim otherwise. They’re very careful to use words like ‘suggests’ and ‘may be’ in the presentation of the data. Overall I’m quite pleased with this feature and hope it sticks around and continues to improve.



How Did He Do That?

I have two documents that create a time conflict for my understanding of when Levi Houston and his family migrated from Maryland to Kentucky.

The first one is a deed of gift from Joseph Schoolfield II to his daughter Sarah Ewell, where he is giving to his daughter and son-in-law three young slaves. One of the witnesses is Levi Houston; the date is 7 Sep 1812.

Worcester County Deed Book AD pg 27

The second document is a deed dated 15 Sep 1812 in Bracken County, Kentucky where Levi Houston is purchasing land from Lawson Dobyns and his wife Mary.

Bracken County, Kentucky Deed Book D pg 319

In 1812 it was physically impossible to travel from Worcester County, Maryland to Bracken County, Kentucky in one week. So how did this happen?

Because I was puzzling over this conflict, the dates stuck with me. Then one day I noticed that Levi and Dolly’s son Joseph Houston got married on 27 Sep 1812 in Bracken County to Delilah Weldon. ( Kentucky, Compiled Marriages, 1802-1850). That strikes me as a short engagement. Then I realized that Levi and Dolly’s daughter Nancy married Thomas Dix on 6 Jul 1811 ( Kentucky, Compiled Marriages, 1802-1850). Their daughter Sarah Houston married John Secrist on or about 10 Jul 1812. Oh look, here’s Levi as a witness to the marriage license. Kentucky, County Marriage Records, 1783-1965

Ah, see Joseph was already in Bracken County in July, so his marriage to Delilah Weldon in September is looking more realistic. Now I know that at least one member of the Houston family, Nancy, was in Bracken County as early as 6 July 1811. She was only 23 at the time; it is highly unlikely her parents would have allowed her to come alone. Unless this was a whirlwind romance, I think the Houston family was already in Bracken County before that marriage took place.

Looking again at the deed for the land in Bracken County, Levi paid $1400 for 430 acres on the Ohio River. From where did he get this money? Part of it might have come from two land sales on 28 Dec 1810 in Somerset County, Maryland. In Deed Book U, pg 4 is recorded a sale of land for $580 and in the same book on pg 15 is another sale for $256. I didn’t have these documents before because Somerset County, Maryland deeds can only be found by going to the Family History Center, not from home.

It would seem that some time between 28 Dec 1810 and 6 Jul 1811, the entire Houston family moved from Maryland to Kentucky. This still doesn’t solve the time conflict, but it suggests that one of Levi’s sons was entrusted with purchasing the land in their father’s name while Levi was in Worcester County, Maryland witnessing a deed of gift.

Migration is fluid; travel back and forth was common even across the Atlantic Ocean. It pays to be open-minded about family members staying in contact and conducting business over great distances. I’m glad to have more information to support at least a plausible reason why Levi was one place while business was being conducted in his name in another place.

Long Lost Daughter

Sometimes the only path to a solution is to wait for it to present itself. Many people with Irish ancestors learn fairly quickly that the first place to start learning more about the ancestor’s family is to find out in which County the ancestor was born, then narrow it down to a Parish. I had given up believing even the County would become known for my John Clark Smith born circa 1814 in Ireland. He was married three times that I know of. His first wife was Catherine. Nothing is known about her except that she is listed as the mother of John’s son Richard in his baptism on 10 Oct 1847 at St. Peter’s Church in Reading, Pennsylvania.

Smith, Richard original baptism

The second known child of John Clark Smith was his daughter Margaret. I’ve written about trying to learn more about her before (2015). I was frustrated by the lack of a baptism record for her and my inability to find a marriage record for her. She just disappeared.

No one in my family knew of the existence of Richard or Margaret. They both appeared in the 1860 census in Reading, Pennsylvania and then the 1870 census in Jackson Township, Nodaway County, Missouri. Both were named in the Will of John Clark Smith in 1875. As covered in the previous post, the 1860 census was confusing; at first glance it appeared John’s wife, my ancestress, Elizabeth McIntyre was the mother of Margaret, aged 5 in 1860. Then I discovered the marriage record for John and Elizabeth dated 27 Jan 1859 (the date appears in the transcription, not this cropped version of the original.)

Smith, John married McIntyre, Elizabeth original
St. Peter’s Church Reading, Berks, Pennsylvania

Margaret’s mother was most likely someone else then. Poor Margaret became the victim of my neglect. Then one day came a miracle in the form of a DNA cousin match on  At first I paid no attention to the common surname Smith in this individual’s tree that she had attached to her results. I see the Smith name in the list of common surnames frequently, as would most people of British/Irish extraction. Then I scrolled down her tree and saw this:

Smith, John on cousin match

So I clicked on Margaret and found this:

Smith, Margaret on cousin match

My first thought was: She survived!!! I am pretty darn happy about that. I have enough experience researching people in Missouri that the next thing I did was go looking for Margaret Blakeley’s death certificate. Missouri has an awesome database of original death certificate images from 1910-1967. [Insert Hallelujah Chorus here:]

Smith, Margaret dc cropped

Now I know who John’s second known wife was and…the County where he was born!!!!!!!

I still can’t find Margaret’s marriage record. I did find her future husband Robert Blakeley in the 1880 Census living a few miles from the Smith family; he was a saloon keeper and 23 years older than Margaret. The family seems to have done well; four of their eight children survived to adulthood. Apparently they moved to Cass County, Missouri then on to Lafayette County, Missouri.

It is very nice to know what happened to her; the added bonus of finding what County John Clark Smith came from is pretty great, too.