How many generations should autosomal DNA test results attach to?

+23 votes


We're starting work on adding the ability to indicate that you have taken a DNA test for genealogy. (See for updates.)
If it's a yDNA test, this information will attach to you, your father, your father's father, etc., and all your father's father's sons, your father's sons, your sons, etc.
If it's an mtDNA test, it will attach to you (if you're a woman), your mother, your mother's mother, etc., and all your mother's mother's daughters, your mother's daughters, your daughters (if you're a woman), etc.
If it's an auDNA test, it will attach to everyone up and down your family tree. I think.
How many generations for the autosomal? I saw Maurice Gleeson mention in a video that it's good for about seven generations. That's probably on the high end. Should we use seven? Six? Five?
We'll probably do all downward generations, since there aren't likely to be more than three. That is, if you're a grandparent, the fact that you've taken a test will attach to all your grandchildren, nieces, nephews, etc.
The number of generations we choose will apply to all the autosomal tests out there, e.g. 23andMe, AncestryDNA, Family Tree DNA Family Finder, etc. I hope that they're all relatively equal and the same number of generations can be used. Would anyone disagree with that assumption? Please speak up if so; this will affect how we structure the program.
in Policy and Style by Chris Whitten G2G Astronaut (1.2m points)
What if numerous males in my family have taken the Y-DNA test so I know the markers up to my g-g-g-grandfather, but being female, I haven't taken the test? If that's not part of the program yet, that's fine. I'm merely curious.
Hi Debby.

Good question.

If the male member of your family that has taken the test is active on WikiTree, the answer is simple. They can just indicate that they've taken the test.

If they're not an active WikiTree member, we probably still want the test associated with their profile instead of yours, in order to get the logic of the connections right. Maybe if you're the profile manager you should be able to say that the person took a test, then you could explain on their profile that people should contact you to compare results.

I have this for my self also. My uncle had the test. I have imput my uncle as a living person but showing his Biography. I have put DNA info in his biography for now

But what I hope for in the future is that I can indicate that he had a y-DNA test at and this will show on all the directly connected paternal males with the Lemons surname. Then anyone can contact me to ask more info.
Kari, that's exactly the plan!

Hi !

I do not know if you had a look at the FAQ of FTDNA's family finder.
The Family Finder program determines relationships for up to five generations. We will return data and matches that are more distant. However, we established the confidence interval ...

If your relationship is within recent generations (2nd cousins or more recent relatives), Family Finder is almost sure to detect your relationship. Testing will also detect many 3rd cousins and about half of your 4th cousins. It will detect a small percentage of 5th and more distant cousins.

Chances of finding a match:
Relationship Match Probability
2nd cousins or closer > 99%
3rd cousin > 90%
4th cousin > 50%
5th cousin > 10%
6th cousin and more distant Remote (typically less than 2%)

For example, if you have 100 of your 3rd cousins test, Family Finder will detect about 90 of them as your 3rd cousins. It will not detect the other 10.

Family Finder only detects a small percentage of 5th cousins and relatives that are more distant. However, the number of such cousins in the population increases exponentially with each generation. This means that if 1,000 of your distant cousins test, you can expect to see a few of them in the Speculative Relative category.

For genealogists, it is best to use Family Finder to prove recent relationships (1 to 5 generations). However, after testing, you may discover distant cousins.

Hi !
I wonder if you plan to come up with an own system to analyse the raw data from auDNA tests? FTDNA does their own calculations and matching within their community. You can download raw data with all the SNPs, but I wonder what to do with is.

You can look at some of it by using the demo user: myFTDNA account demonstration page.


There is a DNA tools website  that is becoming the go to place for uploading raw DNA data from all the labs and being able to compare your results to all in the database.

I have uploaded my FTDNA Family Finder and Mt-DNA and then created an account for my uncles and uploaded his y-DNA.

I am still waiting for my uploads to be fully processed so I can see my matches.

But they have many other DNA tools to use once your raw data is uploaded besides matching.

Register for an account and then follow the instructions for your lab to upload your data. It does take some time (about 4 weeks) for your data to process all the way to be able to use all the tools.

Hi Kari !

Magic. Thanks.

The sad thing is that they suspended uploads  till August.

Thanks for mentioning GEDMatch, Kari. I've browsed their site but am still trying to get my head around it. (I'm very disappointed I didn't upload my FTDNA data months ago. Like Andreas says, they're not currently accepting any new uploads. Darn!)

My thinking is that GEDMatch will be where we point WikiTree members if they want to compare results but haven't used the same testing company. Does anybody else even offer this?
Gedmatch is run by a couple of gentic scientists that started with a blog and then out of the conversations created a couple of tools to raw data, out of their research.

They are not a company but a couple of people who have an expertise, that want to help people.  So bear with them as they keep updating their website. I think some of the issues they are having  is with their server. As they get more popular they keep having server issues.

If you go ahead and register for an account you will be notified when you can uploads will be alowed again. For awhile now they open it up for a couple weeks at a time then close it while they process that batch, then open again.

I looked yesterday and my data was fully processed an I could see my one to many matches now on my family finder data.
I know my comment is too late but I just wanted to say that I have found several matches that go back 9 generations
I'm definitely thinking that the autosomal link should be attached to at least 2 more generations, possibly with the current wording of

It's also likely that these autosomal DNA test-takers will share significant DNA:

being changed for the additional, more distant, connections to something like

It's also likely that these autosomal DNA test-takers may share detectable DNA matches:

I've found at least one match where the link is 9 generations back (but they cheated with cousin marriages) and also confirmed 6th and 7th cousins without "cheating" by cousin marriages.

I believe we should reconsider taking the generations back another step or two.

If you go to this chart which is being maintained by a lady at 23andme to check and verify percentages you'll see it's actually proving connections are being found further back. 

When I go to my 3rd G GF, it only shows possible connecetions for his 
Great Grandson
Great Great Grandaughter
Great Great Great Grandaughter (me)

It misses his Great Great Great Great Grandaughter entirely, who we all match.

I wonder if there needs to be flexibility with this.

I have tested myself, my mother and my mother's mother - and to at least one of our matches, we each share an identical segment, the same length in the same place - indicating that the segment has passed through two recombination processes completely intact.

What this means is that anyone who matches me at a potential 3rd cousin level - actually matches my grandmother at a potential 3rd cousin level - pushing the connection point back 2 generations than initially thought.

Because my grnadmother's mother died some 35 years ago, her DNA is untested - but what if that segment I have, was identical in her as well - there's another generation back again.

I have located and confirmed a cousin based on a 4th great grandparent using FamilyTreeDNA results/surnames.

So, I think it's fair to say "at least one more generation back from where it's set right now," since nothing on wikitree indicated that a descedant of that 4th great grandparent had atDNA test results on wikitree. I would propose going to at least 5th great grandparents if not 6th. If the tests are yielding matches at 5th+ cousin, honestly I'd vote for 7th gg's.

If, for some reason, a profile has so many DNA tested descendants that it becomes problematic to display, then a "more" button could suffice. But I doubt that's even a real problem today -- is it?


Mind you, in my HENDERSONs, not even the full 2nd cousins show up on each other's profiles.

Surely that's a bug?
Henderson-4289, Henderson-2297, Henderson-4291, Henderson-7497

but they do show up on the mutual ancestor's, and my 1st cousins show up on mine.

Perhaps I'm missing a subtlety here.
Lorna, Do you have all of those profiles set to a privacy level that includes a public tree? If not, the DNA information won't display.
Thank you Kay

yes they all have public family trees - and as I said, they do show up on the profiles for the  ancestors, just not the 2nd cousins

I'm also seeing that cousins aren't showing up at degrees of relationship where there would be very DNA significant matches. Example: Profile for my great-great grandmother lists three likely autosomal matches. I am one of those matches, and the other two matches are living descendants of her brother. These other two people are confirmed significant DNA matches for me, so I'm sure they would be more significant as DNA matches for my great grandmother and her siblings, but I am the only potential match that appears on the profiles for my great grandmother and her siblings.

I have fourth-cousin autosomal matches, with whom our most recent common ancestors are great-great-great-grandparents John Lane and Elizabeth Parrott; as well as other matches, with whom our MRCA are great-great-great-grandparents Samuel Massey and Catharine Farnsworth. In both cases that would be six generations, including myself. Or, five generations removed from me. I have other autosomal matches, with whom our MRCA are William James Parrott & Matilda Ann Tharpe, and James Parrott & his wife Lydia; they are our great-great-great-great-grandparents, so that we are double-fifth cousins. That would be seven generations including myself, or six generations removed. En fin, I have a third cousin once removed autosomal match, with whom our MRCA are our great-great-grandparents Charles Stones and Harriet Hall. That would be five generations including myself, or four generations removed. So, it seems to me, that at least SIX generations removed (or seven including ourselves) would be needed for autosomal matches. Which is within the limits of genetic provability and believability.
As I dig into this, I'm realizing that we never connected second cousins. I almost can't believe it.

We count the starting person in the "six degrees" and really only do five steps in all directions. This means we only go as far as first cousins once removed. We'll work on expanding this.
many thanks Chris, I hope you'll include 3rd cousins as well, they're 95% likely to share substantial bits of dna according to FTDNA's table
Thanks Lorna,  It should include 3rd cousins (and 2nd cousins once removed).  I look forward to that change being made.
Done! We now going a whopping eight steps instead of five.

One thing to note: We have lost the original intent of the DNA test connections. When Peter and others originally conceived this, the idea was to be able to say "You can test your relationship to this person by comparing DNA tests with these WikiTree members."

We can still say that with Y-chromosome and mitochondrial tests, but not with autosomal tests.

We had to edit the language that appears on profiles for autosomal test connections. We used to say "It is likely that these autosomal DNA test-takers will share significant DNA" and we meant "with each other." Now we say "It is likely that these autosomal DNA test-takers will share significant DNA with X" where X is the person whose profile it is.

Not that this is bad. This is potentially exciting.  For example, on my sixth great grandfather's profile I now see ten autosomal test connections. I didn't even know about a couple of these cousins.

I do want to get in contact with these cousins (and I'm still thinking we should automate means to do it) but if I'm understanding things correctly I can't expect that our tests will match. We can't expect to confirm a relationship to our ancestor this way, but we can hope to.

To restate this: We now make connections for eight steps in all directions. This means that people appear on the same profiles when they're seventh cousins to each other. I don't appear on my seventh cousin's profile (only third cousins' profile) but we both appear on our sixth great-grandparent's profile.


Hello Chris, Thanks! yes

However, I'm not understanding your step count. 

From my perspective, third cousins would be eight steps from each other.  That is the setting we want because about 90 percent of 3rd cousins are expected to share some auDNA.  50% of 4th cousins should share some, and less than 2% of 5th cousins and greater will share auDNA. 

So if you did take it out to 7th cousins then most of them would have no hope of being a match and it would usually be wasted effort trying to find a match.  GEDmatch, DNAgedcom 23andMe, AncestryDNA, Family Finder already tell us which of those distant cousins are a match.  That is why I believe it would be helpful in WikiTree if we could indicate which of those more distant cousins have been found to be a match.


Most sincerely, Peter

That's right. I could be wrong on my numbers, but my point is that by connecting third cousins on each other's profiles we are also connecting sixth or seventh cousins on ancestor profiles.

Browse the profiles linked from your test connections page to see what I mean.

For example, see

This connects everyone who is within eight steps of you, including third cousins, and also including your sixth great-grandmother

The original purpose of making the DNA test connections on profiles was to tell the viewing user on Barry-634 that they could test their connection to her by comparing DNA test results with you. This is no longer the case.
Hello Chris, See the first graph at

Which shows the probability that an ancestor of a person will pass auDNA to them.  I read that chart as showing there is about a 50% probability of having any auDNA from a particular 5th great grand parent.  A person's 5th great-grandparent lived 7 generations ago.

On that chart, (between 6 and 8 generations) draw a line up until you are between both descending lines and off to the left you see the sharing probability is about 0.5.

The WikiTree wording is currently "It is likely that these autosomal DNA test-takers will share signicant DNA with xyz."  I believe the wording should be "It is likely that most of these autosomal DNA test-takers will share DNA with xyz."

See for example who is my 5th great-grandfather.  Most of his auDNA tested descendants listed are 7 generations from him.  Only about half of them would be expected to have any of his auDNA.

Sincerely, Peter

yes OUTSTANDING! yes  This fix has revealed connections I was overlooking.

Brilliant thanks Chris. I can now see my 2nd and 3rd cousins on my profile but not 3C1R until I go up a generation.
Pity that unlike yourself and Peter this has not as yet revealed any previously unknown test takers, but I live in hope.
In my opinion (as worthless as that is to everyone but me0, you are misleading average Wikitree users by showing autosomal test takers on profiles of ancestors who lived so long ago.   You have to take into consideration both lines of descent to living people.  I think putting the cutoff at 3rd-great grandparents would be best.

I do understand that a snippet of DNA could be retained from an old ancestor but I do no agree that it is probable that the average users are capable of properly decoding the links.  The paper trail trees they post are not well documented or sourced, and therefore not reliable enough for deep auDNA comparisons.

I mostly ignore the information that is displayed, but my concern is that as it is so prominent on the profile, the potential to mislead the less informed is all too likely to result in the internet being flooded with garbage DNA just like the garbage family trees we see everywhere.

Thanks, Mikey, I think we agree.  If you have seen an auDNA test or FamilyFinder test listed on an ancient profile, please let us know which one(s).  I believe the normal cut-off is five to six generations.  Here is my list of surnames that I use to find auDNA common ancestors with my genetic cousins.  It goes back to my G-G-G-G-grandparents. 

Chris, i thought that you already did this some time ago. I was pleasantly surprised to find the DNA tests of new distant cousins attached to several of my forefathers. I believe that at that time you (after posting here for input) decided to place DNA tests at profiles six degrees all ways. That is what i see, anyway. And it is fantastic! Whether any given person has inherited a bit of DNA from a particular ancestor six generations back or not, is irrelevant. The fact is that it is possible, and by placing our DNA tests back and forward six degrees at the profiles of our genealogical ancestors (who may or may not also be our genetic ancestors) we can find ''cousins'', who might confirm lines of descent and help to break through ''brick walls.''
Albertus, we extended from six to eight back in June. This does include many more deep ancestors than we'd originally thought wise. You can't please all the people all the time ...
Thank you Albertus.
I probably haven't properly understood all that you are doing, Chris, but thanks for doing it. In genealogy reseach, it seems to me that ''more is better''. :) And no, you can't please all the people all the time: you can only try to do what seems best at the time, according to the best knowledge available and the majority consensus of those who will be affected.

The website shown above:   is a dead link. Please update if known. 

Thanks Elaine,

The Internet Archive’s Wayback Machine crawled that site:*/

Can someone figure out how to retrieve that backup?

A direct link:  How Many Ancestors Share Our DNA?

The blog posting is interesting, unfortunately it's followed by lots of blog spam (spell casters, love doctors, quack healers, etc).

9 Answers

+20 votes
Best answer

At seven generations you can expect to have 128 people sitting across the top of your tree. Potentially each one of those people has contributed something to your DNA and so you can have a "match" with any of their possible tens of thousands of descendants. You need to know who matches you on a particular segment of chromosome, who else matches them (the ICW or In Common With file) and where are their trees.

Many of the answers here, and perhaps even the question, are looking at this the wrong way around. The autosomal testing companies try to predict where your match may lie on your tree but it is your job to then test that prediction. That is most often done by contact and then questions about the respective trees and hopefully some tool or other that can run the comparisons for you. You are actually trying to find out where that piece (or pieces) of shared DNA came from. Some people have been able to map their DNA to the extent that they can say this segment on this chromosome came from this person who sits here on the tree. This requires extensive testing and analysis AND a very well constructed tree.

The most important information WIkitree can make available (besides providing the robust tree) is the identifying number of the person who has done the test. GedMatch for example requires this to enable comparisons between the trees of any two test subjects and many of the analytical tools provided by FTDNA rely upon the trees stored with FTDNA. In many cases the trees are missing altogether making the second step of the analysis impossible. In my case I have some hundreds of matches but so what if I can't see their trees.

So what is needed? An identifying number of the person who was tested and access to their tree are two. The third is a tool like GedMatch that can do the One to One or One to Many (or several) test that is necessary after that.

An ideal world would have a tool much like GedMatch that sat inside Wikitree rather than rely on the tree that resides with the testing company (or is locked away hidden by a pay wall). The test company tells you that you have a match with these people, you come the Wikitree and key their numbers in and a tool goes off and looks into the respective trees to find anyone listed in common. The report you get points to ancestors in common and isn't that the point of all of this?

A final point, there is a new tool on the block that really needs to be looked at seriously. It is called the Autosomal DNA Sequence Analyzer (ADSA) and has been built by professional genealogists and made available free. It can be found at What is desperately required is the ability to take the "In Common Withs" and run comparisons of their trees, somewhere in there the common ancestor will be hiding and so the numerous trees could then be combined. At the moment the tool relies on the trees stored with FTDNA which are on the whole fairly basic or largely non-existent.

by John Hunter G2G2 (2.8k points)
selected by June Marie Carter Carter

Hi John.

Good post.

You write, "The most important information WikiTree can make available (besides providing the robust tree) is the identifying number of the person who has done the test."

Aren't we already doing that by enabling people to enter their ID at the testing company or What more do you mean than this?
I've tried to talk to the guys at GEDMatch about a closer relationship, but they don't see the benefit. They figure that their internal GEDCOM matching tools perform the service that WikiTree performs.
I've also talked to FTDNA about a closer relationship, but I think they want to grow their tree tools internally as well.
I do need to look closer at DNAGedcom and ADSA and perhaps try to make contact with them.
If we can't build a relationship with someone who does the comparisons, we may need to do them internally, like you say. If you want to talk more about what that would involve I'd welcome your help.
GedMatch simply imports the tree that is on FTDNA and clearly not everyone who has done a test with FTDNA has put their tree up there. The problem extends to the trees that are there are simplified, they only follow the direct line up from you, ie your direct ancestors. Sure, in a perfect world there should be a point of intersection at the common ancestor but that requires a complete tree from both parties and that happens rarely. How many times do we see a person on a tree with no other family members listed at all yet they may be on your tree and you might descend from his or he brother. Wikitree's duplicate tool would find the person without any genetic testing but the test result would help confirm the relationship.

If you model a matrix of possible descendants from the 128 people at the top of a set of seven generation trees it becomes very complex. For a simple model assume 4 children to each couple each generation, 2 male and 2 female, and the females change their names on marriage. Very few in the matrix will have the same surname as yourself for example and I have seen so many people disillusioned by their FF results not fully appreciatiing this likely outcome.

GedMatch has the facility to test your gedcom against some or all of their stored gedcoms to find possible matches which is much like the duplicate tool in Wikitree. But the issue is the extent and quality of the trees they have access to. The key to it lies in being able to take the list of ICWs and pushing a button to find the common ancestors from that pool. There is a way to do that in GedMatch but it is not simple and not every match has a tree on FTDNA and therein lies the way forward.

A wikitree user has a list of ICWs from FF. Wikitree has a tool into which you add the identifiers for the ICWs (most probably the test company identifier much as GedMatch uses) and then the tool goes away to look for duplicates in the trees of the ICWs and that of the testee. Remember each shared segment has its own list of ICWs so the number of possible tests for duplicates would be huge across Wikitree. Also no need to populate autosomal results beyond the person being tested all you need is an identifier for that person from the test company that means something inside Wikitree. The question is what ancestors do I have in common with this person. What I don't know is whether the ICW file from FF has the kit number attached to the individual ICW, I suspect it would have.

What you are looking at is a refinement of the triangulation tools ("triangulation" now being replaced by "in common with"). GedMatch has a new "triangulation" tool but I haven't got to use it yet as my data is still awaiting final processing - and that is another issue. So far the ADSA tool gives you a comprehensive list of ICWs which you can't copy and paste from the screen. I guess it gets down to how to leverage the ICW file.

One last thought. You will recall that "human" DNA has been found to include reasonably large pieces of fully functioning "non-human" DNA. It seems inter-breeding between early homonids was a common practice which has left an important and measureable DNA footprint. I don't think you want to tag profiles with autosomal details back that far.
Excellent points John.   If DNAgedcom's Autosomal DNA Segment Analyzer (ASDA)  was a relational database then the test taker could add a url to their ASDA data at DNAgedcom which links to their tree in WikiTree  or a url to an improved version of WikiTree's Relationship Finder.  Making ASDA a relational database is being considered.  ASDA is telling us which people really share the same atDNA segment due to shared ancestry.  The missing part is a link to a site that can tell us what ancestry those people have in common.
Sorry John but matching and ICW (in common with) has the probability to be a cousin within the genealogical time frame (GTF) but it's not a proof that this is actually a cousin with GTF!

Many people are mixing this up, as companies like and FTDNA explain it wrongly to their customer and those myths get accepted as truth as people don't know better unless they don't take anything for granted what these big testing companies say for marketing reasons.

The only way to find out if a match or ICW is a cousin within GTF is through triangulation. GedMatch and DNAGedcom is both providing you with that, as does 23andme. FTDNA and Ancestry isn't providing you directly (on their websites) with a method to do triangulation and hence their raw DNA data needs to be downloaded and then upload to tools like GedMatch or DNAGedcom.

Let's please not repeat more what isn't correct as it's already difficult enough to convince users those just blatantly belief those marketing messages from the above said two companies (even better "Spit and we give you your ancestors!).


Secondly, those people (like me) that can identify which part of their DNA segment belongs where on their family tree do so through extensive testing of cousins (1st, 2nd etc.). As you probably also found out yourself many matches can never be appointed to the right MRCA, due to not everyone having a 100% correct family tree for 7 generations (you has this, I don't have one). I would say over 90% of my matches point to a 7 generation (and further back) MRCA. Probably because I'm German/Italian born and most people that do DNA test are from the US, so all go far back when people immigrated out of Germany/Italy.

Lastly I think those sites do all the massive number crunching but for WikiTree there is a different purpose so it's let's not mix both purpose. WikiTree doesn't need an automatic triangulation function based on raw DNA data. Let's leave that to GedMatch and DNAGedcom, both sites have their hands full doing this huge number crunching. WikiTree is and should continue to focus on the paper trail.
+8 votes
I have confirmed what was a 4th cousin match on my FTDNA family finder test to actually be a 8th cousin, we connect 9 generations at the immigrant ancestor. She is from one son and I am from the other.

But out of the 46 matches I have, this person is the only confirmed match I have so far. She is connected to my paternal grandmother's family. This immigrant came in 1709. I do have trees back 13 generations to Jamestown in 1618.

I really think that you can list that I have taken the autosomal test on my profile, but showing a conncection should be a manual input of profile ID numbers between me and the person I have a confirmed relationship too.  It is important to remember that Autosomal connections are between 2 living persons as cousins back to a connecting ancestor.  All those directly between us could be automatically marked in some way as Autosomal DNA test connection. It would encourage me to get the person I have confirmed through testing to join up here and have her add her tree so that our DNA relationship can be connected.

I really wish you would fix the descendant list to 10-15 generations for those of us that need it.

I do not think that it is a good idea to just have an auto indication on all of my tree if I indicate that I have taken a autosomal test. That does not help me or anyone else.
by Kari Lemons G2G5 (5.5k points)

Hi Kari,

I don't follow you.
Let me try to trace out the logic of what I have in mind. Maybe I'm missing something.
Since you have taken an autosomal test, every profile genetically connected to you, in every direction, would have an indication that you have taken the test.
Now let's say I land on and think I'm related to her. I see that you have taken the Family Tree DNA Family Finder test. I then know that I could probably prove or disprove my connection (or your connection) to Little-415 by comparing Family Finder test results with you.
Doesn't that help you and me, and everyone else interested in the genealogy of Little-415?
P.S. I added an explanation to your descendant lists question here.

I have taken the Family Finder test at FTDNA, I have 46 matches to others right now. Now it is up to me or them to contact each other and share our ged file or online tree, to see what surname we might have in common and  then try to find what branch and person we might connect to with them.  FTDNA does allow me to upload a ged file to my account that my matches can see, and I can also add surnames in my tree to a comment that my matches can see.  The problem is that not everyone does either of these things. 

Again for me, I have talked too, shared ged files with and confirmed 1/46 of my matches to a verified person on both of our trees. 

I have one match (my closest one at 3rd cousin) that we have talked, shared ged files and still cannot find the connecting person. And we both have very extensive trees. It is frustrating be cause we know we should be able to find the connection because supposedly we are fairly close matches, but we can't.  

Just because I have done the test does not mean that I might match a person who has my John Little in their tree. This might be your missing piece of info.  

I think you need to seperate your thinking about the autosomal test from that of the y-dna and mt-dna and how connections are made with them.  

I take the test and get my matches then I have to contact them and do more research to confirm how we are connected. As I said these matches are 2 living persons that are distant cousins who have a common ancestor. 

So how I think it should work on wikitree is: if I confirm a match over at FTDNA and talk her into putting her tree here at wikitree. I then indicate on my profile that I took the test. Then there is a box to add confirmed Family finder cousin by inputing her profile ID number, Then hit connect. Then your program adds a automatic Autosomal DNA confirmation to all of the direct people between me and her.

Now if someone clicks on Daniel Gee they will see that he has that autosomal DNA confirmation on his profile and they will be able to click on that DNA confirmation ( takes you to the DNA page) and see how Daniel connects to me and my matching person Mrs Lawson.  It would show up as an upside down V, Me on the bottom on one side and Mrs Lawson on the bottom of the other and our connecting person Richard Gee at the top, with the direct lineages of us up to Richard. Daniel Gee would be in the list on Mrs Lawson's side. 

This is real, accurate, verified, helpful information to someone, that has been verified by the 2 people who matched and it is probably the most accurate info because there is the DNA and records proof of the connection. DNA gave us the clue, but records showed how we matched. 

I hope my thoughts on this are clear



It'll be up to you if you want to enter that you've taken a test. You could always just include that in your bio but not in the system.
Part of the benefit of having it attached to profiles on WikiTree is that people who haven't taken the test may be encouraged to do so. Sure, Family Tree DNA can connect you with other Family Tree DNA customers. But we can suggest when taking a Family Tree DNA test might be beneficial for your genealogy.
+8 votes
As I answered on another of your recent questions, I've taken 23andMe, which indicates my maternal haplogroup, so that could attach to my maternal line as far back as it goes.

I haven't yet confirmed any matches, however from the conversations there, it's not unusual for "2nd cousin" to be 3rd or 4th, etc. With recombination, etc., there are no solid rules, so 23andMe, at least, is giving a pretty broad estimate. After a few years, with more data, the science might advance to giving closer estimates.
by Marilyn Langfeld G2G1 (1.4k points)
+7 votes


An improved approach would be to help persons who match on a shared autosomal segment to determine who their shared ancestors are. GEDmatch tells us the GEDmatch IDs of people who share an overlapping segment of our atDNA.  If atDNA tested WikiTreers had a field to input their GEDmatch ID then WikiTree's Find a Relationship feature "could" tell us who the shared ancesors are for other WikiTreers who shared that segment.
Also it is confusing for WikiTree to tell us it may be possible to confirm family relationships with all our "4th" cousins who have had an atDNA test when in fact it will not be possible to confirm a relationship with more than half of our 4th cousins. 
The atDNA testers are also grouped together with the Y-DNA and mtDNA testers under DNA on each profile page.  It is Y-DNA testing which is the easiest to use to confirm family relationships. It would be much more helpful if that grouping was organized into a nested bullet list for Y-DNA, mtDNA, and atDNA.  That would make it easier to distinguish those who have had a Y-DNA test.
Thanks and sincerely, Peter


by Peter Roberts G2G6 Pilot (543k points)
edited by Peter Roberts
I have a confirmed 4th cousin (paper+Y 66/67 match). We do not see any autosomal match on FF, however each share AT matches with other confirmed cousins via gedmatch on ancestry and 23andme. So I agree Peter: the ATDNA tests, Y tests and MT should be nested separatly. How about: Direct Paternal, Direct Maternal and Cousin Matches for nesting catigories.

Y-DNA (Direct Paternal)

mtDNA (Direct Maternal)

atDNA (Reatives: 4th cousins & closer)  

In a short time some families will have many atDNA tested relatives and it will be quite confusing and not so helpful to see the long list without knowing who shares which segments.  Allowing WikiTreers an input field for a GEDmatch ID  can tell us who the shared ancestors are for shared segments.
I agree, certainly add a field for the GedMatch number (I'm a regular user, and donate to their site).

I do believe that 4th cousins is much to restrictive however as so many of us have found matches much further back due to Colonial ancestors and/or cousin marriages where the alert to older common surnames would be very useful.
Currently the best tool for working with matches on an atDNA segment is the Autosomal DNA Segment Analyzer (ADSA) at

It works for Family Tree DNA's Family Finder and should soon work for 23andMe and perhaps AncestryDNA.
+9 votes
My recommendation would be at least to 7th cousins, considering the occurrence of cousin marriages means that such matches might not be complete shots-in-the-dark. I can usually confirm 5th cousin matches by reviewing results from a panel of close cousins; and I'm starting to see the same thing for 6th cousins. So I suspect that 7th cousins could be confirmed through gathering a large group and closely comparing the raw results.

At any rate, the key would be to identify the strength of the downstream autosomal DNA result for all of the ancestors. If I knew that there was a potential 4th cousin who had tested, I would certainly look to them before any 4th once or twice removed, or 5th cousins.

In addition, whenever someone has found a likely match through atDNA at a particular ancestral couple, it would be good to highlight that, since that would point to particular DNA segments that were passed down.

Finally, it may be good to add be a notie when people report autosomal results if they suspect adoptions on particular lines--I don't know whether wikitree has protocols to identify lines through adoption.
by Anonymous Long G2G1 (2.0k points)
+6 votes
I must second John Long's recommendation of 7th+ cousins based on my experience with Gedmatch, AncestryDNA, 23andMe, and FTDNA.  With 7 autosomal kits and one mtDNA, I have primarily identified 4th through 8th cousins.  I have one 2nd cousin and one 3rd once removed.  

I have two surnames (Jordan & Austin) on my maternal side that extend beyond the 7th cousin break, implying segments that retained sufficient length for an incredible number of generations.  Therefore, I ultimately prefer the flexibility to indicate those matches beyond the guaranteed or anticipated generations indicated on testing websites.
by Sonia Haga G2G Crew (630 points)
We have far too many 6th and 7th cousins for showing them to be useful since we will only match less than one percent of our 6th and 7th cousins on an autosomal DNA test.

A better approach is to encourage matches to upload their ancestral tree to WikiTree and upload their atDNA results to and in GEDmatch's shared segment tool note the GEDmatch IDs for your shared (overlapping) segments.  WikiTree could be tweeked to show any shared ancestry for those GEDmatch IDs.

Sincerely, Peter
I would like to add my voice to see if a place to enter one's GEDmatch id number in such an manner that it would show on the person's need to populate ancestor profiles.

Unitl there is a standard place to look for a GEDmatch id those connections won't be highlighted.

Peter R. has more expertise than I do and therefore his suggestion may have more merit, but some modest step on the WikiTree side might help, such as just having a spot near the large type DNA banner on one's profile.

+6 votes
Its next to impossible to answer this question. Though truthally id be hard pressed to say the answer would be any more then 4rth cousion . because bacically each generation cuts it half  parents/ siblings essentially 50% same with first first cousions aunt and uncles.


second cousin about 25% 3rd  about  1/16ad so one
by Matt Pryber G2G6 Mach 4 (48.3k points)

We should focus on triangulated groups (TGs) as revealed with the triangulated segment tool at or the ADSA tool

If we could know the WikiTree IDs of members of those trangulated groups then we could automatically know the shared ancestor or ancestral couple for those groups.
+6 votes
I have been sucessful in using the AutoDNA results I got back from Ancestry, in combination with completed trees (10 generations or more) on both parties, and have been able to go up my tree to an 8th gr gr - with absolute confidence *noting* the line of children, to get to my current cousin.  Which in most every case, I never had close to a clue (without the DNA match).  

I am just learning how to upload my raw data to GEDmatch.
by Debra Allison G2G6 Mach 2 (20.9k points)
A word of caution regarding AncestryDNA's suggested shared ancestry for an auDNA match - - The match can actually be from unidentified shared ancestry and not the known ancestry which AncestryDNA highlights.  It is when three or more people all match each other on the same segment and you know the shared ancestral couple for most of them that you can be confident of confirming the ancesral line back to the shared ancestor for that group. This also applies to Family Tree DNA and its use of In Common With (ICW).  A group can be ICW but not match on the same segment and so it does not help confirm ancestry back to any ancestral couple the group happens to share (due to a flawed/mistaken genealogy).
+11 votes
I would throw in a request to err on the side of more generations rather than fewer. I am sure there are more than a few folks with genealogy back 9-10 generations where the amount of intermarrying is significant. Anecdotally, the families with a lot of 2nd or 3rd cousins intermarrying across several generations show, not surprisingly, an amplification of the DNA 'signal,' so that the DNA services conclude a more recent common ancestor than is actually warranted.

On another note, I'd love it if Wikitree and GedMatch would fully integrate. All of these DNA services and their walled-garden gedcom uploads - yuck. GedMatch seems to keep crashing from trying to manage all of these resources - (I'm not sure how long their Gedcom + DNA matching has been broken, but it has been at least weeks.) Seems like it would be a win-win for both services if GedMatch built their DNA processing on top of Wikitree.
by Gary Kueber G2G5 (5.1k points)
Hi Gary,

Maybe you could mention integration to the folks at GEDMatch? I've ben trying to get them to communicate with us for years on this. We do a lot on this side to send people to GEDMatch, include GEDMatch IDs, etc., but it's all unilateral.

Hello Gary,

I completely agree with your second note.

On your first note, 5th cousins only have about a 10% probability of matching each other.  6th cousins and greater have less than a 2% chance of matching each other.

If WikiTree noted all our 5th through 9th? cousins who have had an auDNA test then we would likely waste a lot of time searching for possible matches.  If we found a match "with an 8th cousin"  then it would be difficult to know the matching segment was from the shared ancestor who was known and not some other yet undiscovered shared ancestor.

GEDmatch and DNAgedcom do a good job of showing us our significant matches for a segment.  The trick is automatically associating a matching group with their shared ancestry.

It would be different if we were talking about Triangulated Groups and phased results.

Sincerely, Peter
Peter, maybe I'm not understanding, but I wanted to make the point that the various DNA services may match someone as a "2nd to 4th" or "4th to 6th" etc. when the actual MRCA is farther back d/t intermarrying. Trying to track down the actual MRCA with some of these folks (when the intermarrying occurred, say 4 to 5 to 6 generations back) is difficult.

Cheers - Gary
Chris - I sent them an email about this a few minutes ago.

Thanks - Gary
I agree with the difficulty of working with populations where intermarriage was common.  Half my ancestry is from one such group (-;   If WikiTree showed me all my "8th" cousins who had been auDNA tested then I would be overwhelmed and many of us have 6 or more of the same ancestors in our ancestry.

It is better to start with your matches in GEDmatch and/or DNAgedcom who all match each other on the same segment and then look for the shared ancestry each group has in WikiTree.  If some of those matches are not already in WikiTree then invite them to join, or ask their permission to add their information to WikiTree on their behalf.

Sincerely, Peter

Seems to me that GEDmatch needs a big boost of equity capital, or something, so that it can become fully operational in response to demand.

The biggest frustration for the non-scientific bods like me is the lack of genealogical information attached to matches.  This is the only aspect of the Ancestry tests in their favour.   It would be wonderful if it were possible on Wikitree, for example, to link GEDMatch file numbers to specific ancestors.  So for example, take my ancestors John Munro and Annie McDonald.  It would be great if there was an automatic link that showed the GEDMatch file numbers that have linked back to these ancestors (maybe I should just add them in the biography).  

I will send some emails to the GEDmatch people too :-)

Leigh,  I think all that is needed is for GEDmatch or DNAgedcom or Family Finder to give the user the option to enter a url for their ancestry. For example:

or a url to WikiTree's Relationship Finder (if the WikiTree IDs can be associated with he IDs for GEDmatch, DNAgedcom, or Family Finder. etc.)

IF FTDNA permitted what you suggest, would that not violate GPDR and privacy rules.

Just asking.

Related questions

+3 votes
2 answers
+4 votes
4 answers
458 views asked Dec 15, 2017 in The Tree House by Judith Brandau G2G6 (9.6k points)
+7 votes
1 answer
+8 votes
3 answers
+1 vote
0 answers
+2 votes
2 answers

WikiTree  ~  About  ~  Help Help  ~  Search Person Search  ~  Surname:

disclaimer - terms - copyright