Thursday, September 15, 2016

Working on My Top MyHeritage DNA Match

I wrote I Have MyHeritage DNA Matches on 5 September, and shortly afterwards the manager of the first match on my list - the one I showed in my blog post - contacted me via 23andMe to inquire about my English and German ancestry.  We have continued the conversation in email.

His mother is probably my third or fourth cousin because we share 55 cM with the longest segment 35 cM on Chromosome 20.  That is, I think, significant.  

1)  His story is really interesting and challenging.  The problem is that his mother's great-great-grandmother, born probably in Germany, had an unknown father.  The family thinks that this man was from England or Germany and the great-great-grandmother was born in the early 1850s.  

2)  The problem sounds impossible to solve, but we're going to try to work through it.  

On my end of it, I have only five (out of 32) 3rd great-grandparents who were born in England and migrated to America in the 1830 to 1860 time period.  There were no German-born ancestors in that group of 32 or even the 4th great-grandparents.  

Those English family lines are very likely to be the connection that this match has with me.  I went back one more generation in these family lines to find candidate males who might have been the father and were also related to me through a common ancestor.  

3)  The common ancestors may be:

*  John Richman (1788-1867) and Ann Marshman (1784-1856) of Hilperton, Wiltshire. (my 3rd great-grandparents).  I don't know earlier Richman or Marshman ancestors.  I have some data on their descendants but not a complete tree for three or four generations.

*  John Rich (1791-1868) and Rebecca Hill (1790-1862) of Hilperton, Wiltshire (my third great-grandparents).  I don't know earlier Rich ancestors.  I have some data on their descendants but not a complete tree for three or four generations.

*  John Hill (1765-1825) and Ann Warren (1764-????) of Hilperton, Wiltshire (my 4th great-grandparents).  I have some data on their descendants but not a complete tree for three or four generations.

*  John Vaux (1747-1806) and Joanna Laver (1763-1836) of South Petherton, Somerset (my 5th great-grandparents).  I have a pretty good database through three or four generations thanks to my Vaux cousins and their database.

4)  So how should we attack this problem?  My suggestions have been:

*  We are looking for males who were born in the 1800 to 1830 range who fathered a child in the early 1850s.  Those males have a common ancestor between his mother and myself.

*  Use my RootsMagic database to create descendant reports for the candidate ancestors listed above. I did this today and sent them to him via email.

*  Go through the list and identify candidates from the lists who might be a military man or a business man and was single in the 1850 time frame.  I realize that a married man might be the mystery father, but almost all of those Wiltshire and Somerset males were farmers or laborers and would not be traveling to Germany.

*  Perform more genealogical research on the candidates found and narrow down the candidates.

*  Add the mother's raw DNA file to GEDMatch and see if the DNA results match on the same chromosome with some of my known Richman/Rich/Hill or Vaux DNA matches on GEDMatch.

Here is a screen shot of the first page of one of my narrative descendant reports:

5)  What other tasks should we be performing?  Are my assumptions correct?  I would appreciate recommendations and suggestions for this problem.  Thank you in advance!!


The URL for this post is:

Copyright (c) 2016, Randall J. Seaver

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, Google+ or Pinterest using the icons below.  Or contact me by email at


Debbe in Northern Nevada said...

I'm not sure I would work really hard at finding a connection, because I think "beta" is more like "omega"--and not ready for prime time, to say the least, at least in my experiences.

I uploaded my AncestryDNA matches and here's my closest match on My Heritage:

Possible relationships: 2nd cousin - 3rd cousin once removed
Shared DNA 0.9% (65.9 cM)
Shared segments 4
Largest segment 24.8 cM

Looks real and significant, right?

Well, this match also tested at Ancestry and uploaded to GEDMatch, and here are those results:

AncestryDNA (Less DNA would be expected due to Timber, but still....)

Predicted relationship: Distant Cousins
Possible range: 5th - 8th cousins
Confidence: Moderate
9.2 centimorgans shared across 1 DNA segment

GEDMatch (even more surprising)

Total cM: 15.5
Largest cM 15.5
Est. gen 4.9

I cannot fathom how the total cM could be so different. Based on the relationships in the trees, the GEDmatch and AncestryDNA numbers make a lot more sense than what MyHeritage reported. All I can figure is that their matching algorithm has some serious bugs--and until it gets fixed, I won't even consider anything they report as real.

debbe...whose other matches are even more inexplicable....

Randy Seaver said...

Hi Debbe,

Thank you for the note and the experience. Every bit of information helps, I think.

Obviously, MyHeritage is using some sort of algorithm to estimate the matching amount and number of segments that is different from AncestryDNA.

I heard from my contact, and we match 36 cM on 23andMe, with a largest of 24 cM, so it's significant, but perhaps 4th or 5th cousin now.

He uploaded his raw DNA file to GEDMatch so we'll see what they say.

Hopefully, some of the experts will evaluate the MyHeritage algorithm in the near future and opine on it.

Les said...

I tested with FTDNA and when the results came back several of my top matches and many of the lesser ones had a surname unknown to me. They were all originating from New Brunswick, Canada where my grandmother(father unknown) was born. At this point, I've started a database in RootsMagic with the last known ancestor of some of this matches, looking for candidates near hear hometown.
I also have created in Excel a chromosome data table from FTDNA and added top matches from Gedmatch.
Raw data can be added to FTDNA for a small fee, which would expand your matches greatly.
When I tested, I hadn't even thought about finding my missing great grandfather, this result was a big surprise.