Thursday, September 21, 2023

Rabbit Holes With Randy - New AncestryDNA ThruLines

 I've been digging in this rabbit hole for over three weeks, ever since Marshall Clow wrote Guest Post: What Happened at AncestryDNA This Week?  

I received 64 new ThruLines with common ancestors defined by AncestryDNA after 25 August.  My usual process is to evaluate each one of them by checking the line of the DNA Match provided in the ThruLines.  That means looking at the data provided by the family tree(s) supposedly supporting the ThruLine claim - are there sourced birth, death and marriage records for each parent-child relationship, do the dates and locations make sense, is the claimed common ancestor the same person as my research results, etc.  

If the line from the Common Ancestor to the DNA Match provided by Ancestry meets my criteria, I then add the line to my RootsMagic family tree, which gets TreeShared with my Ancestry Member Tree, which generates Record Hints for each person, which I use to add more content to my tree for those profiles.  Often, I will do more research and add the spouse(s) and children of each profile to my tree in order to build my tree with descendants of my 4th great-grandparents. 

All of that is relatively easy to do when I get one or two new ThruLines a week, but doing all 64 in three weeks has been a struggle.  I've evaluated the 43 top new ThruLines so far, and the remaining 21 are for matches of 13 cM or less.  I have added the lines of only 6 of the 64 DNA matches to my tree, and the other 37 have incomplete information (e.g., either "Private" generations) or have, in my judgment, erroneous connections in the DNA Match's line (wrong Common Ancestor, wrong relationship, errors in dates, etc.).

One thing I've noticed so far is that these "new" ThruLines are more likely to have errors in identifying the purported Common Ancestor.  For example, the ThruLines for my 5th great-grandfather  Samuel FitzRandolph (1730-????), who married Martha Gach (1730-????) in 1754 in Woodbridge, New Jersey, shows 28 ThruLines.  

Ten of the ThruLines go through my 4th great-grandmother Tabitha Randolph (1752-1841), and I have judged all of them to be correct.  Samuel and Martha (Gach) Fitz Randolph had only two children (Tabitha and Mary) - born in 1752 and 1754 according to New Jersey church records.

The 18 other ThruLines from Samuel and Martha (Gach) Fitz Randolph go through 10 purported children of Samuel and Martha, with birth dates from 1761 to 1805 (when Martha would be  75 years old).  

All of the purported children in the 18 "other" ThruLines are actually children of Samuel Fitz Randolph (1738-1825) and Margaret Fitz Randolph (1738-1832) of Piscataway, New Jersey (except for the one born in 1805), who married in 1761.  If my Tabitha Randolph were the daughter of Samuel and Margaret (Fitz Randolph) Fitz Randolph, the mother would be only 14 years old at the time of her birth.  I  do have the nine children (born 1762 to 1781) of the latter Samuel and Margaret in my RootsMagic and Ancestry trees connected to their parents.  

How did this happen that AncestryDNA got it so wrong?  It appears that they have grown and pruned the BIG Tree and used it to generate the newest ThruLines since August 25th.  I've written a few blog posts about the so-called "Ancestry BIG Tree" that they use to determine the relationships to support the AncestryDNA ThruLines.  This BIG Tree was built based on the family trees submitted by users to Ancestry over the past 20-some years.  So, it appears they have conflated the two couples Samuel and Martha and Samuel and Margaret into one couple.  

This is why I do a fairly rigorous evaluation of each ThruLine, starting with the question "is the purported common ancestor actually my ancestor?"  

Of the 18 "wrong" ThruLines for Samuel and Martha (Gach) Fitz Randolph, at least ten of them were found in this "new" batch of AncestryDNA ThruLines that appeared in late August 2023.  When I write a Note for each DNA Match on my Match list, I denote the "wrong" ones (in my judgment) with "XXX" at the beginning so that I easily can see that I've looked at the ThruLine and judged it accordingly.  

This example is the "worst" example of AncestryDNA conflation in my ThruLines.  I do have several other examples that either have conflated ancestors or erroneous potential ancestors according to my research.  

This rabbit hole is pretty deep, and will always be with me, at least until Ancestry fixes their BIG tree and shows two separate Samuel FitzRandolph families in the same place and same time.

I need some carrots to give me some energy!


Disclosure: I receive a complimentary all-access subscription from, for which I am thankful. has provided material considerations for travel expenses to meetings, and has hosted events and meals that I have attended in Salt Lake City, in past years.

The URL for this post is:  

Please note that all Comments are moderated so they may not appear immediately.

Please comment on this post on the website by clicking the URL above and then the "Comments" link at the bottom of each post. Share it on Twitter, Facebook, or Pinterest using the icons below. Or contact me by email at

1 comment:

mbm1311 said...

As always, it is an excellent explanation. If it ever made financial sense, I would like to see the "Big Tree" at ancestry verified using AI and records. I know some people are worried about AI, but all information has to be verified by the user, so I am not afraid of AI.