Whenever I get a new good-sized DNA match, I try and figure out how we are related. Ancestry and MyHeritage both have clever tools that search your tree and the trees of other users to come up with the likely relationship. Of course both you and your match have to have a family tree connected to your DNA on those sites for that to work.
GEDmatch has just released a tool for its Tier 1 (paid) members that will search for the common ancestor you have with your DNA matches on that site. This capability requires both you and your match to have uploaded a GEDCOM to GEDmatch and associated it with your kit. I just updated my 2015 blog post about using GEDCOMs at GEDmatch which explains much about them, so click here to read it. By the way, a GEDCOM is a text file that is formatted especially for genealogy programs; it lets you move the facts in your tree from one genealogy web site or program to another.
The new tool found nothing on my mother’s side. She was German and half Jewish. There are almost no Germans at GEDmatch and my one known half 2nd cousin on there has no tree uploaded. As to the Jewish side, very few have their trees far enough back to meet mine. I need to get a few of my known 3rd cousins to upload GEDCOMs.
The above listing, partially repeated below, shows all the common ancestors my first run of this tool found for my Norwegian-American father using the default settings on the form. I have cut off the first 3 columns on the left which have the kit number, match name, and email address for privacy; also that makes the image readable on this page!
Let’s look at the rest of the columns for my Dad’s top match. Clicking the tree icon would take you to the user profile information. The cM shared are listed next; 40.3 is match that can be anything from a 4th to an 8th or even more distant cousin. Then the name of the possible shared ancestor, first in my GEDCOM and then in my match’s GEDCOM; either one can be clicked on to go to that person’s tree entry. The 8G is how many steps down from that ancestor my Dad is. If you click that, you see a descendant list from that ancestor in your tree. Notice that his match is also 8 steps down, so in the same generation. Subtract one to get the cousin level so this is a 7th cousin.
I like the display you get from clicking on “PTH” (Up/Down Path column) the best as it shows both ancestral pathways side by side and lists the actual relationship, as shown in the image above. Plus you can see the place locations which can help to quickly rule out people with very similar names.
The other columns are “CMP” (Compare GEDCOMs) which lets you get a full comparison of your two GEDCOMs. Perhaps there are more ancestors in common? Then the GEDCOM id column which has the id of the matches tree, clickable to the matched person. Using those features is covered in my more general article on GEDCOMs at GEDmatch, mentioned above.
The final column has the match score for how likely this is actually the same ancestor in each GEDCOM. Higher numbers are better. So far no MRCA with a score of 2 or less has been correct for me, so on my subsequent runs I have changed that number in the form requesting this search to a 3. Some of the 3s are good, most are not.
Interestingly since I ran the comparison for Dad on the first day of this tool and left the browser screen up when I clicked two days later the little temporary files to display the pathways expired. So it is a good idea to take screen shots or cut and paste the information to a word document. When I reran the tool for this article it found many more matches with trees, not all recently uploaded; it now shows the closer relatives. I was puzzled when they were not listed on the first run. Obviously there have been improvements already.
A problem for me is that Norwegian naming is not standardized between web sites. I prefer using the farm name as a surname and the patronymic as the middle name as is the standard at GENI.com but the GEDcom I have on GEDmatch is an old one without all the farm names. I hate to overwrite it with my latest GEDCOM because then all the confirmed tree matches in it will be lost. Perhaps I will give my brother’s kit the latest GEDCOM and see what is found.
Going quickly through Dad’s initial list, all the ones where the patronymic is not a very close match are not good. So both the pair Hansdotter and Jansdotter as well as Jonsdotter and Olsdotter are not the same person. Amazingly that 16 step descent from Tore Torkellson Koll is accurate but the common ancestor is really his daughter Karen which we have listed differently in our two GEDCOMs. Note that there is no guarantee that the 11.9 cM DNA match is actually from this ancestor. Likely there are other common ancestors that far back.
The Koll/Skiftun family is a well known noble family from Hordaland so lots of people have them in their trees; in fact, the second time I ran this tool there were 2 more matches with that ancestor in the list. It is easy to trace your ancestors back to the 1600s in Norwegian records but before that, it was mainly the noble families who were recorded.
The 12 step Brigit Skiftun (same Koll family) match is correct, we just spelled the father’s name differently. Looking at the path, again the actual common ancestor is the next step down.
The 6 and 7 step Ole Hansen Fatland match is accurate; I have a number of distant cousins descended from him on GEDmatch but not all with trees. Another one did turn up in the second run. I often blog about Fatland farm (click here for my translated farmbook entry) and even visited there on my trip to Norway
Reviewing that surprising large match of 40.3 cM to my Dad’s 7th cousin Sherril, I discovered I had been in touch with her son Michael before and we had found a different common ancestor. His grandmother was descended from Etne Hordaland farmers, as was my great grandfather and that area is very intermarried. As it turns out, two brothers married two sisters doubling up on the common DNA in their descendants. So we have at least their parents, including the one found, Anna Ormsdatter, as common ancestors. See the images of the extracts from the GEDmatch pedigree tree displays below.
This match has two matching segments which are shared among a number of 4th and 5th cousins and they are assigned in my chromosome map to a different ancestral line from Etne, my Ve line. Looking deeper, I found that Michael’s grandmother’s other Etne side also had a far back common ancestor or two. More research is needed. Have I found an NPE in the 1600s?
More GEDCOMs are being uploaded to GEDmatch every day, so run this tool often.
Thanks for the great article on this new tool, Kitty! I am working with an NPE, looking for her father. We have her kit on all the 3rd party sites. Closest matches (299 cM and below) all tie into the same families. MRCA is either g-grandparents or grandparents.
Is there a way to upload a GEDcom that would help in this search? We really can’t link her completely downstream of the MRCA and are still stuck as to their exact position in her line. I would love to try this with her GEDcom but also don’t want to create problems for anyone else. Thanks!!
You might try it with using first names like FAKE or TEST for the unknown parents and grandparents to see what you get
I was so disappointed when I ran this tool, Kitty, and did not have one match on my maternal (Ulster Irish) side. There were enough matches on my dad’s Colonial/Mayflower line for me to play with it, but I hope more Irish descendants upload their gedcoms soon!
Thanks for this idea, Kitty. Everything points to “John Doe” and “Jane Smith” being either her g-grandparents or grandparents. The 299 cM match was their grandson and he was adopted out. In working on this, I discovered both of his parents for him. However, as far as family knows and I can find, “John Doe” and “Jane Smith” (who were married and divorced) only had 2 daughters. One daughter is the mother of the adoptee. The other daughter swears she only had her 3 sons. Their surname does not come up in any matching records. The adoptee is approximately 30 years older than the NPE.
I’m stuck!!! Should I upload a GEDcom showing “John Doe” and “Jane Smith” as the g-grandparents and then do a “TEST” for the grandparents and father? Unfortunately, the NPE has very few close matches on her paternal side anywhere. This is my longest running case so far and I need to solve it soon. Thanks!!
Sure try it if you think it will help you sort this out. I will send you a private email so you can give me more details if you wish.
Hey there other K Cooper, hope you are well.
I suggest you run the tool every few days as it improves and more gedcoms are uploaded … do you have a kit for your mother? or a Lazarus kit for her to try it on? even today my Lazarus kit finds more than the other day including the missing close relatives
Just a beginner from Australia and not very good at using Gedmatch but I’m trying to help my hubby find his birthparents.
We found his maternal side and he has a Norwegian grandfather who was born in Sandefjord in 1884 surname Palmstrom who emigrated to USA. I just noticed some similar surnames in the box above Olsdatter and Larsdotter from his Ancestry tree. Are these very common surnames in Norway and what would be the best way to research his maternal heritage,
Yes those are extremely common names. Norwegians used patronymics rather than inherited surnames. This is explained in my presentation here (use right arrow key to move along it) about finding Norwegian ancestors
Norwegian research is relatively fun and easy once you get the hang of it. Lots of resources listed here:
You might start by searching for his grandfather by first name, birth year and location in the online Norwegian archives. This story might help you
I’m also working through an NPE and I’m thoroughly stuck. My only match on my mystery line is 144cm but we only share 3 segments. I’ve read lot but can’t decide how close our MRCAs would be. With my other matches with similar numbers the number of segments is much higher.
Hi, Kitty. That’s a pretty neat tool. I spent a few hours playing with it. Most of the hint MRCAs on my tree seemed to be correct. I did detect some of them had equated disparate surnames on a few of them, though. For example “Robinson””Robbins” and “Henderson””Anderson”. But luckily there were only a few of them.
This is an exciting new tool. Thanks for the explanation Kitty. Off I go to try it out.