Leeds Method: Why Do I Have More Than 4 Clusters? Or, Time to Consolidate!
The Question
Today I want to discuss one of the most frequent questions I get: Why did I get more than 4 color clusters on my Leeds Method chart?
(I’ll address the issue of getting less than 4 clusters in a future post.)
The Simple Answer
There are basically two answers to this question.
-
- You don’t have enough second cousins. (We will also discuss this in a future post.)
- You need to do another step which is consolidating (or combining) some of your clusters.
Key Concepts
As we discuss this, let’s keep a few key concepts in mind:
-
- The goal of the Leeds Method is to naturally sort our DNA into clusters of matches who are biologically related to each other.
- We hope to get 4 clusters representing our 4 grandparent lines. However, not everyone will get 4 clusters! (In other words, your results may vary!)
- Many people need to consolidate (or combine) clusters to “see” their grandparent lines.
The Leeds Method
Let’s start with a simple, fictional Leeds Method chart which created 5 clusters using these basic steps:
-
- List all of your DNA matches between 400 and 90 cM.
- Assign the color blue to your highest match, Lori, and add an asterisk to her.
- Assign the blue to all of Lori’s Shared Matches: Joe, Rebecca, and Liam. (Note: A Shared Match is a person who shares DNA with both you and your match, Lori.)
- Assign the color red to the highest match that doesn’t already have a color, Meredith, and add an asterisk to her.
- Assign red to all of Meredith’s Shared Matches: Joe and Rebecca.
- Continue steps 4 and 5 until everyone has at least one color. (Note: These steps created the Green (Lynne), Yellow (Lucy), and Purple (Mitzi) Clusters.)
Overlap and Heavy Overlap
We’ve now completed our initial Leeds Method chart. We were hoping to see 4 clusters, although we know some people will have more or less than 4 clusters. Our chart has 5 color clusters.
But our chart shows overlap. In fact, it shows heavy overlap. So let’s define these two terms:
-
- overlap – a person with more than one color. In this example, Joe and Rebecca were assigned both the color blue and red, so they have overlap.
- heavy overlap – most of the people in one cluster are also in another cluster. In this example, 2 out of 3 of the Red people are also in the Blue Cluster. Also, 2 out of 4 of the Blue people are also in the Red Cluster.
Whenever we see heavy overlap, as in this case, my recommendation is to combine those colors into one. In this case, we would combine the Blue and Red Clusters. This would give us a total of 4 clusters.
Next, I’ll explain why I recommend consolidating clusters. This is a more advanced explanation. If you want to keep things simple, you might skip this part! But if you want to understand why these clusters probably formed then continue reading.
Seeing the Segments
As you probably know, we cannot see the segments (or pieces) of DNA we share with our matches at Ancestry. This is because Ancestry does not have a chromosome browser. But let’s pretend we can see the segments of DNA using a fictional chromosome browser. For simplicity, we will only look at the 5 people who make up the Blue and Red Clusters. (Remember, those are the two clusters that have heavy overlap between them!) And, we will pretend that all of the matches showed up on one of your chromosomes: chromosome #1.
In this pretend scenario, let’s also assume that you have identified these 5 matches and know they are all related to you through your Grandpa Fred’s part of the family.
Below are the fictional chromosome browsers for you and your 5 matches we determined were biologically related to Grandpa Fred. Each chromosome is broken into 6 segments which are numbered from 1 to 6. The grey segments are those inherited from Grandpa Fred’s part of the tree.
The Blue Cluster
Now let’s think about how we created our chart. When we used the Leeds Method, Lori, who inherited segments #2-5 from Grandpa Fred, started our Blue Cluster. Her Shared Match list included the following people: Joe, Rebecca, and Liam. They showed up on her Shared Match list because they share DNA with both you and Lori.
With this fictional chromosome browser, we can see that Lori shares the following DNA segments with these matches:
-
-
- Joe at segment #3 & #4
- Rebecca at segment #3
- Liam at segment #2 & #3
-
Note that, although Meredith also has DNA from Grandpa Fred’s part of the family (segments #1 & #6), Meredith and Lori don’t share any DNA segments. So, Meredith would not show up as a Shared Match of Lori’s. And Meredith did not end up in the Blue Cluster.
The Red Cluster
Instead of being a part of the original Blue Cluster, Meredith—at step #4—started the Red Cluster. Using the fictional chromosome browser, we can see she matches both Joe and Rebecca at segment #6. So Meredith, Joe, and Rebecca are the Red Cluster.
Consolidating Clusters
By using this fictional chromosome browser, we can see why Lori, Joe, Rebecca, Liam, and Meredith should all be in one cluster. They all came from one part of your tree! They are all related to Grandpa Fred! They only ended up in two different clusters because Lori and Meredith didn’t happen to inherit identical segments of DNA.
In this case, we also know they all belong in one cluster because we were able to identify these 5 matches and determined they were all related through Grandpa Fred’s line.
But, even without a browser or identifying these 5 matches, we can see the heavy overlap. When you see heavy overlap, you should probably combine the clusters!
A Chart with 4 Clusters
After consolidating the Blue and Red Clusters into one cluster, the Blue Cluster, we end up with 4 clusters. Each of these 4 clusters likely represents one of your 4 grandparent lines:
Do we need a chromosome browser?
In real life, we do not need to be able to see these chromosome segments to create these clusters! Instead, this example is meant to show you why people who are related to each other—like Lori and Meredith—might end up in different clusters. This happens because they don’t share segments of DNA. However, we can still combine clusters—and thus have Lori and Meredith in the same cluster—based on heavy overlap.
Questions?
I hope this helps! Please let me know if you have any questions or comments.
Just want to say that your explanations are very user friendly. Thank you for this.
Thanks so much, Kathy! I work hard to make it user friendly. 🙂
So do you do this just on an excel spreadsheet or similar or is there a particular form you need to use?
Hi, Sue. You can use an Excel spreadsheet or any spreadsheet or a piece of paper! If you’d prefer to print out a form and fill it in, my friend Fiona created a form you can use. It’s available here for free: https://memoriesintime.co.nz/collections/family-history
You might also want to watch a 20-minute video where I go over the basic steps: https://www.rootstech.org/video/you-can-do-dna?lang=eng (My part starts at about 40 minutes in, but I suggest watching the entire thing!)
Hope this helps!
Hi Dana,
Digging down from top to bottom. Thank you for your help! No biggie, but the link to rootstech you-can-do-dna gives a 404 error.
Dave
Its because rootstech videos are taken down after a period (2yrs I think)
Thank you! That makes sense. 🙂
Dana
Five columns? How about 15?! (71 matches on Ancestry between 90cM and 400cM.)
Hi, Jo. I’m going to share an example that has 9. And, yes, you can get many more. But the principles are the same.
I will talk about another possibility in a future post, but I’ll share the basics now. If you don’t have any 2nd cousins – or even don’t have 2nd cousins for each grandparent line – then you will get a lot more clusters. Do you know how these matches are related to you and whether they are 2nd or 3rd cousins? If so, you can check that. If not, do you have many matches between 300 and 400? Or are they more at the lower end of the range? That could be the “problem.”
Lastly, it could be that you just have a lot of consolidating to do! Again, I’m always happy to take a quick look and provide feedback to anyone who wants to share their chart with me. Just email me at drleeds@sbcglobal.net
On Ancestry, I have 64 2nd & 3rd cousin matches just on my paternal side (by contrast, I only have 8 2nd & 3rd cousin matches on my maternal side). When I made my Leeds chart, I ended up with 7 separate colored clusters just on my paternal side with heavy overlap between the columns. I’m unclear how to combine these clusters, since I only have a handful of matches who are only in one colored cluster. Do you have any suggestions for me? Thank you.
Hi, Lisa. First of all, how much do you know about your family and your matches? If you know your matches, feel free to include matches who share only ONE grandparent with you even if they share more than 400 cM. Also, do NOT include any match – even if they share less than 400 cM with you – if they share 2 grandparents with you.
Then, if ALL of the clusters overlap quite a bit, I would probably combine them all. In this case, it is possible that your paternal grandparents are related to each other. Do you know if they are? For example, did 1st cousins marry 1st cousins several generations ago? This is called pedigree collapse. My uncle, for example, only gets one cluster for his mom’s side because his mom came from a small community and her grandparents were related to each other.
Another possibility is that one of your paternal grandparents does not have any descendants who have tested at this 2nd and 3rd cousin level. (Do you know if this is the case?)
I’m also happy to take a quick look at your chart and give you some feedback if you’d like. Just email me at drleeds@sbcglobal.net
Dana
Ok, It looks like I have 15 matches in the 2nd-3rd cousin range who share both of my parental grandparents. They are my known cousins in the 1C1R-1C2R range.
So, should I use the Leeds method starting with my highest 3rd cousin match (she is 174 cM) who is not in my shared parental grandparents group, and continue to go down to my 90 cM matches?
My paternal grandparents were not related to one another. My grandfather was a MPE, and a majority of my matches are through his previously unknown father’s line. I know there might be a little pedigree collapse in that line, but it was a 1C marrying a 1C1R several generations back.
Thank you!
Hi, Lisa. Interesting case! Thanks for sharing!
This is very clear and I appreciate you addressing this issue. I have stated elsewhere that clustering my DNA matches using the colored dots at Ancestry based on your original method of using a spreadsheet is my most important tool in understanding my DNA matches. Now I think I may just go ahead and utilize a spreadsheet what with the changes that are rolling out at Ancestry which are greatly harming the use of the colored dot groups. Thank you for developing and sharing the Leeds method.
Thank you for your kind words and I’m glad you find the Leeds Method so useful!
According to GEDMATCH my Mother’s parents were related. Endogamy was common in her Mennonite-Germans from Russia community. So now I have clustered her side of the family and…of the 57 cousins between 90 & 400 cM, 22 are NOT overlapped. Is there a next step?
Hi, Joyce. I’m not familiar with that specific population, but in general people with endogamy have matches who show up unusually high. This is because they are related to the test taker in multiple ways. For example, a match might show up with 400 shared cM of DNA, but this DNA actually reflects multiple relationships. For example, they could be a 3rd cousin in several ways as well as a 4th cousin.
In these cases, the pieces of DNA are small and might not overlap as much. I think that might be what is happening with your case. You might look at how many segments you are sharing with these matches. In my case, those I shared about 300-450 cM of DNA with share around 15 to 20 segments. If you are seeing much higher numbers of segments, they are small segments.
From what you wrote, I’m not sure how many clusters you found. Feel free to send me a screenshot of your chart and I’ll be happy to provide a bit more feedback. My email is drleeds@sbcglobal.net
Dana
Would there be a reason to adjust the range (90-400 cM) because of this endogamy?
Unfortunately, the Leeds Method does not work with endogamy. I would suggest watching Paul Woodbury’s talk titled “Dealing with Endogamy” on Legacy Family Tree Webinars.
Pingback: Leeds Method: Consolidating Clusters, An Example - Dana Leeds
Pingback: Leeds Method: Lack of 2nd Cousins Can Cause Lot of Clusters - Dana Leeds
I’ve been trying to learn your method. It works pretty well for my In-Laws but they don’t have many matches in the 90-400 cM range. For my parents, who each have over 100 matches in the 90-400cM range, it is quite a mess to read. They both end up with twelve groups. Some, but not all, of the groups have heavy overlap. All of the groups have overlap. My parents aren’t related but they come from a small community where a lot of their relatives married people that were related to my parents in a different way. I think this is making it very difficult to interpret the form and decide which columns should be merged and which shouldn’t.
Hi, Michael. When people come from small communities that intermarried, we call that pedigree collapse. Instead of sharing one ancestor or ancestral couple with a DNA match, these people share multiple connections. Working with DNA matches in these situations is a lot more difficult and, unfortunately, the Leeds Method doesn’t help. You might read this post: https://www.danaleeds.com/can-the-leeds-method-help-me/
Dana
I have many cousin 26 cousins between 400 – 90 cM. I end up with seven columns after I combine two sets of them into one (started with 9 columns) based on two overlaps in those groups. Is there a reason I’m getting 7 instead of 4? Am I just getting the g-grandparent groups instead of the grandparent groups? I don’t know the relationships at this point.
Hi, Dan. Yes, it is likely you don’t have enough 2nd cousins to gather the clusters together so you are getting more groups. Whatever clusters you end up with, the next task is to determine what they mean. How are the people in the cluster related to each other? And how are they related to you?
Best wishes!
Hello, how about if you end up with five columns, with very little overlap? (only one: the other two had no overlap.) I have four very robust columns, and one puny fifth column with three people on it, only one of whom overlaps with anyone else; the other two are just solos; empty sad sacks just sitting there. What do I do with those two? Thank you! Tracy
Tracy, Whatever clusters you end up with, the next task is to determine what they mean. How are the people in the cluster related to each other? And how are they related to you? If they don’t match anyone else in the 90 to 400 cM range, hopefully they match someone who shares more or less DNA. Or hopefully you are missing some part of your family and can hypothesize that this cluster represents that group.
Best wishes!
Hi Dana,
Thank you for another great post.
First, I want to tell you that I love creating these charts!
Second, the one I am currently working on has me puzzled. We do not know the relationship of any of these people to the tester. There are no matches between 300 and 400 cM (and only 3 above 400 cM). There are 7 matches between 200 and 300 cM. There are 31 matches between 90 and 200 cM.
I have 8 columns in this chart. The last 4 have no shared matches above 82 cM. There are only 2 instances of overlap. A 72 cM match is in 2 columns. A 61 cM match is in 2 columns. There is no overlap between these two matches either.
What does it mean when there is no overlap (or only two, below 90 cM) across 8 columns? I have no clue which columns to combine in this instance.
Thank you for any assistance you can offer. Keep up the great work!
Hi, Suzanne. I’m glad you’re enjoying making these charts! They can be quite helpful. But I know they are sometimes confusing, too. I also saw your follow-up question about this possibly representing 8 great-grandparents. I think sometimes clusters are created that represent a great grandparent, but I think it would be very rare (or impossible?) for someone to have 8 clusters representing 8 great grandparents.
You asked about the 4 matches who don’t have Shared Matches above 82 cM. Since those are at about the 3rd cousin level, it is possible that anyone else who has tested who is a descendant of that particular ancestral couple just shares less DNA with you than 82 cM.
Also, you don’t always have to combine columns/clusters. If you don’t see any that should be combined, then work with them as they are! Just start looking at the people in that cluster and try to determine how they’re related to each other and then how they’re related to you.
Best wishes!
Dana
Hi Dana,
Thank you for your response. I am sorry that I didn’t see it sooner.
What would cause it be rare or impossible for the clusters to point to each of the 8 great-grandparents?
Have a blessed week. 🙂
PS Upon further pondering of this situation mentioned in my previous comment…could it mean that the 8 columns on the Leeds chart I have created point to each of the great-grandparents?
Hi, I have several 1C1R/2R/3R that are over 250cM matching with me. This is somewhat of a surprise, although I know that a few of them are actually DOUBLE first cousins 2x removed, 3x removed, etc. Should I use them as a starting point? I did not include them in my spreadsheet I created. And, from which company should I use the matches– Ancestry, GEDMatch, MyHeritage, Wiki, ftDNA, or all of them?
Linda, I do not have much experience with double cousins. But, as far as who should be on the chart, use those who share only one (not two) grandparents with you. I work with the company results separately. And I would start with Ancestry.
Best wishes,
Dana
I’m new to this so it could just be my inexperience but I am having such a tough time with mine. After grouping, I am left with 7 columns (3 from paternal side, 4 from maternal side). It was very obvious that two of my maternal columns could be consolidated as they were carbon copies of each each other minus one cousin. All three columns on my paternal side were similar so I don’t which to consolidate with which. One of those paternal columns only has three cousins. I am also left with with 3 columns for my maternal side, none of them have any common matches…
So I am stuck with 6 columns…or 5 if I combine two of the paternal but I don’t know which of the two to combine…
Hi, Kathy. Sometimes it’s a tough call on whether or not to consolidate. You can ALWAYS leave them alone and work with them as is! My rule of thumb is if MOST of the people in one color cluster are also in another then I combine.
But, not matter what, the next step is to figure out how the people in the cluster – any cluster – are related to each other and then how they’re related to you.
Best wishes!
After doing my DNA on Ancestry I discovered that I was not related to my dad’s family. After 2 years work & with the help of one of my stronger matches, I have identified my biological father & his family in NZ.
I have done the Leeds method purely on my paternal matches between 320cM & 91cM. The resulting spreadsheet has 4 columns with some overlapping.
As my mother’s side is not in this spreadsheet, which ancestors do the 4 columns relate to? There is overlapping between clusters 1, 3 & 4 & none in the 2nd cluster which only had 5 matches in it.
Will these clusters be from 2 sets of great grandparents or 2 x gt grandparents?
A few matches in clusters 1, 2 & 3 have some maori ethnicity but none in cluster 4.
Hi, Linda. If there is heavy overlap between columns 1, 3, & 4, then these 3 columns probably represent one great grandparent line and the other cluster represents the other. The only way to know for sure is to figure out how the people in each cluster are related to each other and then how you are related to that common ancestor or ancestors.
Best wishes!
Dana
Dana if a person is looking for their unknown biological father and they are using the Leeds method which suggests using matches with 400 cM to 90 cM making them only have 7 paternal matches to use should they use matches under the 90 cM? My highest paternal matches are 224, 197, 183, 170, 143, 101 & 94 cM. The last two people 101 & 94 are brother & sister for sure and 94 has an extensive tree which I have access to but I am unable to figure our my connection to her or any of the other 6 people. The other 6 do not have much information in their trees. What do you suggest?
Hi, Rachel. I suggest looking at the Shared Matches of those 7 matches and go as low as you need to go. Best wishes!
Dana
Hi Dana.
I think I can see why we have 4 columns in an idealised situation, using only 2nd cousins. But if I am understanding this correctly, if all the matches were 3rd cousins only, would 8 columns actually be the result? Please excuse my dimness.
Best wishes
David
Hi, David. I enjoy the question! I think if we all had lots of 3rd cousins for each branch of our family, we would theoretically end up with 8 columns. But I don’t think, in reality, that happens. But you are understanding the theory!
Dana
Thanks Dana. I asked because my problem is that my Ancestry matches have only 1 second cousin ☹️ Everyone else is 3rd or higher!
Another thing to consider is “only” children. Both my parents were only children – so I have zero first cousins, and no aunts or uncles. (Weird, huh?)
Therefore, my children have no second cousins on my side, and when you eliminate their 1st cousins, all that is left is just 3cs and above.
I’m actually working on a Leeds chart, and stopped by for a refresher.
Hi, PJ. First of all my apologies. Somehow I stopped receiving notifications of comments so I’m quite behind.
Yes, that’s a great reminder! Our family has been “small” in recent generations. This can definitely affect the Leeds Method and working with DNA in general!
Thanks for stopping by!
Dana
Hi Dana, I tried the Leeds method for my friend and ended up with 10 groups with no overlap whatsoever!
I’m pretty sure I did it correctly. I only used matches that shared at least 90 cM and no more than 400 cM. I also double-checked to see who was matching with who. Could this be because my friend doesn’t have many 2nd cousin matches?
Hi, Alicia. Yes, a large number of clusters without any overlap is often a sign of few (or no) 2nd cousin matches in this range. Good call!
Dana
Hi Dana, firstly, thank you for sharing your excellent method. I’m new to the process so please excuse me if I am missing something obvious, but I ended up with 5 colour groups: pink, blue, yellow, green, orange. When I attempt to consolidate them I find that pink and green overlap significantly, but so do orange and blue, so I fear I may have gone from 5 groups to 3?
Hi, Siobhan. If there is just some overlap, I’d leave them separate for now. Then work with each cluster following these steps: https://www.danaleeds.com/next-steps/
Hope this helps!
Dana
Hi Dana. I’ve done the Leeds chart for a cousin. I ended up with 9 colours, so I consolidated, and ended up with 2. So, I started back with my original chart and added in the higher matches (there were only 6 of them, and between 787cM and 425cM.) Then I ended up with 8 colours, so I consolidated, and have ended up with 2 again. I’m at a loss what to do as this is the first time I have done a chart for someone and not ended up with 4 clear grandparent lines. Thanks.
Hi, Sharon. One reminder is that you are looking for MOST of the people in one color to be also in the other. So if there is just 1 or a few people, I don’t combine them. See if that helps. If not, feel free to email me and send me an image of your chart and I’m happy to provide feedback!
Dana, drleeds@sbcglobal.net
Hi Dana,
This is such an interesting way to look at ancestry! My father was adopted and we know nothing about his biological family. I did the colour coding, but since we dont have the paid version it doesnt let me see all the matches for each person, just their top 3 (which usually include myself and my sibling)
So will this still work then? There has always been one other match visible, so I’ve been colouing based on that. IE there is what appears to be a half sibling and/or nephew that 70ish% of the other matches match with and then a potential 1st cousin that the remaining matches match with.
So that has only resulted in two colours, between 26 matches.
So does this imply that there are just no other cousins for other sides of the family that have done ancestry DNA… or an implication of potential incest…?
Hi, Rachel. You would really need a paid account, at least for a short time, where you can see all of the matches to do the Leeds Method. I also have a YouTube channel where I have quite a few videos about the Leeds Method – including last week’s video about not having 4 clusters. You can check it out at http://www.youtube.com/@danaleeds