# Thread: Is amount of shared DNA accurate?

1. These are the relationships among my family with cM count:

I share 719 cM with my great aunt.
My aunt has a first cousin on Ancestry and she shares 953 cM with him.
My dad's aunt has two first cousins (two sisters) on Ancestry. She shares 1,067 cM with one of them and 886 cM with the other.
My mum shares 1,403 cM with one first cousin and 649 cM with another first cousin.

2. I have a fair number of fairly close relatives at Ancestry, so I'll also chime in:

Two 1st cousins: 831 cM with one and 690 cM with the other

Six 1st cousins once removed: 633 cM; 593 cM; 590 cM; 491 cM; 379 cM; and 344 cM

Two 2nd cousins: 265 cM and 261 cM

Two 2nd cousins once removed: 196 cM and 193 cM

My wife has three 1st cousins at Ancestry: 910 cM; 714 cM; and 642 cM

My daughter has two 1st cousins at Ancestry: 1,058 cM and 876 cM

Since her mother's 1st cousins and my 1st cousins are our daughter's 1st cousins once removed, I'll list those for comparison purposes. In the same order as for her parents, they are:

Paternal - 428 cM; 307 cM / Maternal - 348 cM; 224 cM; 339 cM

You'll note that my daughter shares a fair amount less than half as much DNA with her mother's 1st cousins as her mother does. In fact, with the cousin who shares 714 cM with her mother -- which is definitely in 1st cousin territory, she only shares 224 cM. That really looks more like a 2nd cousin to our daughter, rather than a 1st cousin once removed. But she is definitely a 1st cousin once removed, since this is her mother's 1st cousin.

It's possible that these numbers have actually been "messed with" by Ancestry, due to their moronic software Timber. Timber may have a useful function, but the DNA matching should all be reported rather than hidden; if Ancestry feels it's necessarily, simply flag the DNA matching that Timber finds iffy. Then we can check for ourselves.

Comparing the same people at Ancestry versus GEDmatch, I have on a couple of occasions found a difference of 50-100 cM (or more) for close relatives.

I could also show cM for various relatives at 23andMe, but I'm not feeling industrious enough at the moment.

EDIT: I'll throw in one more of my wife's relatives at Ancestry. She and a 1st cousin once removed share 465 cM, and our daughter shares 254 cM with this person -- which is pretty much what you'd expect for 2nd cousins, although the relationship is actually 1st cousin twice removed.

3. ## The Following User Says Thank You to geebee For This Useful Post:

Rufus191 (08-27-2019)

4. I was looking at this recently, in part because I was comparing my FTDNA matches with Ancestry.

My closest matches at Ancestry (the parenthesis is their match with my sister for comparison):

Paternal first cousin: 916 cM (1090 cM)
3 maternal first cousin 1xR: 422 cM (379 cM), 339 cM (353 cM), 316 cM (329 cM)
2 maternal 2nd cousins: 306 cM (249 cM), 277 cM (328 cM)
Paternal 1st cousin 2xR: 222 cM (240 cM)
Maternal 2nd cousin 1xR: 208 cM (226 cM)
3 paternal 2nd cousin 1xR: 188 cM (280 cM), 178 cM (91 cM), 177 cM (92 cM) -- my numbers are all about the same, my sisters are way different

5. Frankly, I hope someone can tell me what the cut off is that Ancestry uses for the different categories. The last person in my "3rd cousin" list shares 91 cM with me across 5 segments. Since this is over 90 cM, that means that Timber theoretically didn't eliminate any matching -- since the Ancestry white paper on Timber claims that Timber is not used when sharing is over that amount.

But, the first person on the "4th cousin" list shares 72 cM with me -- and this is actually close to the average (74 cM) for 3rd cousins. In fact, while "The Shared cM Project 3.0 tool v4" (https://dnapainter.com/tools/sharedcmv4) gives a 31.65% probability that this is a 3rd cousin once removed or half 3rd cousin and only a 28.41% probability that it is a 3rd cousin, it also gives a 19.18% probability that this is a half 2nd cousin or similar relationship. In other words, there is a little better than a 50% chance that this is at least a 3rd cousin, not a 4th cousin.

I actually know what the relationship is -- this match is a 3rd cousin, not a 4th cousin. So why does Ancestry put her in the "4th cousin" group? (Or to be more precise, they say 4th-5th. The first number should really be "2nd".) To make matters worse, since Ancestry is showing our matching as 72 cM, I have no way of knowing whether it really ought to be just under 90 cM -- the cut off for Timber not "messing with" the amount of cM reported.

The bottom line is, you can probably generally presume that you have at least as much sharing as Ancestry says. But if the amount is lower than 90 cM, you can't trust that it might not actually be higher -- and underreported due to Ancestry presumptuousness. At least be thankful is a match is reported, since it is clear that Ancestry's Timber completely eliminates some matches.

6. ## The Following User Says Thank You to geebee For This Useful Post:

msmarjoribanks (08-21-2019)

7. Originally Posted by geebee
Frankly, I hope someone can tell me what the cut off is that Ancestry uses for the different categories. The last person in my "3rd cousin" list shares 91 cM with me across 5 segments. Since this is over 90 cM, that means that Timber theoretically didn't eliminate any matching -- since the Ancestry white paper on Timber claims that Timber is not used when sharing is over that amount.

But, the first person on the "4th cousin" list shares 72 cM with me -- and this is actually close to the average (74 cM) for 3rd cousins. In fact, while "The Shared cM Project 3.0 tool v4" (https://dnapainter.com/tools/sharedcmv4) gives a 31.65% probability that this is a 3rd cousin once removed or half 3rd cousin and only a 28.41% probability that it is a 3rd cousin, it also gives a 19.18% probability that this is a half 2nd cousin or similar relationship. In other words, there is a little better than a 50% chance that this is at least a 3rd cousin, not a 4th cousin.

I actually know what the relationship is -- this match is a 3rd cousin, not a 4th cousin. So why does Ancestry put her in the "4th cousin" group? (Or to be more precise, they say 4th-5th. The first number should really be "2nd".) To make matters worse, since Ancestry is showing our matching as 72 cM, I have no way of knowing whether it really ought to be just under 90 cM -- the cut off for Timber not "messing with" the amount of cM reported.

The bottom line is, you can probably generally presume that you have at least as much sharing as Ancestry says. But if the amount is lower than 90 cM, you can't trust that it might not actually be higher -- and underreported due to Ancestry presumptuousness. At least be thankful is a match is reported, since it is clear that Ancestry's Timber completely eliminates some matches.
https://support.ancestry.com/s/artic...egories#fourth

Parent/Child
You will share about 3,275 centimorgans with a parent or child.

Immediate Family
You will share about 2,4002,800 centimorgans with a full sibling and 1,4502,050 centimorgans with a grandparent, aunt, uncle, or half-sibling.

Close Family
You will share about 1,4502,050 centimorgans with an aunt, uncle, or half-sibling and 6801,150 centimorgans with a great-grandparent.

1st cousin: possible range: 1st  2nd cousins
You will share about 6801,150 centimorgans with a first cousin.

2nd cousin: possible range 1st  2nd cousins
You will share about 200620 centimorgans with a second cousin.

2nd cousin: possible range 2nd  3rd cousins
You will share about 200620 centimorgans with a second cousin.

3rd cousin: possible range 3rd  4th cousins
You will share about 90180 centimorgans with a third cousin.

4th cousin and more distant
You will share about 2085 centimorgans with a fourth cousin and 620 centimorgans with a fifth through eighth cousin.

These are just estimated relationships. It can be outside the estimate. The range would be too broad and meaningless if every possible connection was included. Ancestry also provides a percentage chance of a connection. For example, I only share 17 cM with my 3rd cousin. She's estimated to be a 5th-8th cousin but when I click on the relationship, it tells me that there is a 5% chance that we are 3rd cousins. They're not eliminating the possibility but rather suggesting that it is more likely to be a distant connection. The amount of shared cM can vary a lot for people of the same connection so I wouldn't take issue with someone being classified as a 4th cousin when they're actually a 3rd cousin.

8. ## The Following 2 Users Say Thank You to FionnSneachta For This Useful Post:

C J Wyatt III (08-20-2019),  msmarjoribanks (08-21-2019)

9. Originally Posted by FionnSneachta
https://support.ancestry.com/s/artic...egories#fourth

Parent/Child
You will share about 3,275 centimorgans with a parent or child.

Immediate Family
You will share about 2,400—2,800 centimorgans with a full sibling and 1,450—2,050 centimorgans with a grandparent, aunt, uncle, or half-sibling.

Close Family
You will share about 1,450—2,050 centimorgans with an aunt, uncle, or half-sibling and 680—1,150 centimorgans with a great-grandparent.

1st cousin: possible range: 1st – 2nd cousins
You will share about 680—1,150 centimorgans with a first cousin.

2nd cousin: possible range 1st – 2nd cousins
You will share about 200—620 centimorgans with a second cousin.

2nd cousin: possible range 2nd – 3rd cousins
You will share about 200—620 centimorgans with a second cousin.

3rd cousin: possible range 3rd – 4th cousins
You will share about 90—180 centimorgans with a third cousin.

4th cousin and more distant
You will share about 20—85 centimorgans with a fourth cousin and 6—20 centimorgans with a fifth through eighth cousin.

These are just estimated relationships. It can be outside the estimate. The range would be too broad and meaningless if every possible connection was included. Ancestry also provides a percentage chance of a connection. For example, I only share 17 cM with my 3rd cousin. She's estimated to be a 5th-8th cousin but when I click on the relationship, it tells me that there is a 5% chance that we are 3rd cousins. They're not eliminating the possibility but rather suggesting that it is more likely to be a distant connection. The amount of shared cM can vary a lot for people of the same connection so I wouldn't take issue with someone being classified as a 4th cousin when they're actually a 3rd cousin.
Yes, the amount can indeed "vary a lot". That's one reason I don't have any objection to using "2nd", "3rd", or "4th" for the broad categories. But when you say "The range would be too broad and meaningless if every possible connection was included", no one suggested that every possible connection should be included. But, I did point out to you a source that gives a different range and average for 3rd cousins than the one used by Ancestry, which is too high. (Not that they even provide an "average".)

No, instead they say this for every single person in their "4th cousin" category: "4th to 6th cousin". How is this helpful? The first person on my list is much more likely at 72 cM to be a 3rd cousin than to be a 6th cousin, and in fact she is a 3rd cousin. The last person on my list is much more likely at 20 cM than he is to be a 4th cousin, but Ancestry simply states the same range throughout.

How do other companies handle it? Well, for a 72 cM match there it says "2nd to 4th" in my match list -- because anything more distant than 4th is extremely unlikely, and 2nd is quite possible. (It's about a 20% probability, in fact.) If I click on this match, their prediction is actually "3rd cousin", based not only on total cM but on the number of segments shared. FTDNA also lists my 72 cM match as "2nd to 4th", and they certainly don't go all the way down to 20 cM still saying "2nd to 4th" (or even "4th to 6th").

And, to make matter worse, we don't really know if a match said to be 72 cM is 72 cM. It could be closer to 80 cM, because anything less than that may be hidden by Ancestry because of assumptions made by Timber. These assumptions may sometimes be correct, but they may also often be wrong. The remedy isn't to hide the matching on "an ad hoc basis". (https://dnagenealogy.tumblr.com/post...ng-segments-at) The remedy is to simply let customers know when Timber's calculations say that a given match may be "too matchy". Using Timber to flag "suspicious" matches would be preferable to never letting the customers see them to begin with.

Every other company follows the convention for match reporting -- which is to use the start and end points of matches. Not to "jimmy them" using a formula based on an unproven hypothesis.

So yes, the amount of sharing can vary a lot. My point is that this is precisely the reason that Ancestry's stated ranges for relationships need to also vary according to the probabilities. Yet they don't -- everyone in each category except 2nd cousin gets the same range, no matter how much or little DNA they share.

10. Originally Posted by FionnSneachta
https://support.ancestry.com/s/artic...egories#fourth

Parent/Child
You will share about 3,275 centimorgans with a parent or child.

Immediate Family
You will share about 2,400—2,800 centimorgans with a full sibling and 1,450—2,050 centimorgans with a grandparent, aunt, uncle, or half-sibling.

Close Family
You will share about 1,450—2,050 centimorgans with an aunt, uncle, or half-sibling and 680—1,150 centimorgans with a great-grandparent.

1st cousin: possible range: 1st – 2nd cousins
You will share about 680—1,150 centimorgans with a first cousin.

2nd cousin: possible range 1st – 2nd cousins
You will share about 200—620 centimorgans with a second cousin.

2nd cousin: possible range 2nd – 3rd cousins
You will share about 200—620 centimorgans with a second cousin.

3rd cousin: possible range 3rd – 4th cousins
You will share about 90—180 centimorgans with a third cousin.

4th cousin and more distant
You will share about 20—85 centimorgans with a fourth cousin and 6—20 centimorgans with a fifth through eighth cousin.

These are just estimated relationships. It can be outside the estimate. The range would be too broad and meaningless if every possible connection was included. Ancestry also provides a percentage chance of a connection. For example, I only share 17 cM with my 3rd cousin. She's estimated to be a 5th-8th cousin but when I click on the relationship, it tells me that there is a 5% chance that we are 3rd cousins. They're not eliminating the possibility but rather suggesting that it is more likely to be a distant connection. The amount of shared cM can vary a lot for people of the same connection so I wouldn't take issue with someone being classified as a 4th cousin when they're actually a 3rd cousin.
Yes, the amount can indeed "vary a lot". That's one reason I don't have any objection to using "2nd", "3rd", or "4th" for the broad categories. But when you say "The range would be too broad and meaningless if every possible connection was included", no one suggested that every possible connection should be included. But, I did point out to you a source that gives a different range and average for 3rd cousins than the one used by Ancestry, which is too high. (Not that they even provide an "average".)

No, instead they say this for every single person in their "4th cousin" category: "4th to 6th cousin". How is this helpful? The first person on my list is much more likely at 72 cM to be a 3rd cousin than to be a 6th cousin, and in fact she is a 3rd cousin. The last person on my list is much more likely at 20 cM than he is to be a 4th cousin, but Ancestry simply states the same range throughout.

How do other companies handle it? Well, for a 72 cM match there it says "2nd to 4th" in my match list -- because anything more distant than 4th is extremely unlikely, and 2nd is quite possible. (It's about a 20% probability, in fact.) If I click on this match, their prediction is actually "3rd cousin", based not only on total cM but on the number of segments shared. FTDNA also lists my 72 cM match as "2nd to 4th", and they certainly don't go all the way down to 20 cM still saying "2nd to 4th" (or even "4th to 6th").

And, to make matter worse, we don't really know if a match said to be 72 cM is 72 cM. It could be closer to 80 cM, because anything less than that may be hidden by Ancestry because of assumptions made by Timber. These assumptions may sometimes be correct, but they may also often be wrong. The remedy isn't to hide the matching on "an ad hoc basis". (https://dnagenealogy.tumblr.com/post...ng-segments-at) The remedy is to simply let customers know when Timber's calculations say that a given match may be "too matchy". Using Timber to flag "suspicious" matches would be preferable to never letting the customers see them to begin with.

Every other company follows the convention for match reporting -- which is to use the start and end points of matches. Not to "jimmy them" using a formula based on an unproven hypothesis.

So yes, the amount of sharing can vary a lot. My point is that this is precisely the reason that Ancestry's stated ranges for relationships need to also vary according to the probabilities. Yet they don't -- everyone in each category except 2nd cousin gets the same range, no matter how much or little DNA they share.

11. It looks as if I double-posted, but actually the forum seems to be acting up right now.

12. I'll note that in addition to total cM shared, you also have to consider the size of the segments shared. In general, longer segments are more likely to experience crossovers in any given recombination event than are shorted segments. Unfortunately, Ancestry continues to elect not to provide any sort of chromosome browser, nor even to give the length of the longest segment. (Unless there is only one segment.)

They also don't say whether any segment on the X chromosome is shared, even though that would be helpful in determining what side a match may be on -- at least in the case of males. (And even females will be able to narrow down the likely possibilities for matching.) Of course, you do have to remember than it is possible to have more than one path to more distant ancestors, and so a match on the X chromosome doesn't guarantee that any autosomal matches have the same source.

13. ## The Following User Says Thank You to geebee For This Useful Post:

msmarjoribanks (08-21-2019)

14. Originally Posted by geebee
Frankly, I hope someone can tell me what the cut off is that Ancestry uses for the different categories. The last person in my "3rd cousin" list shares 91 cM with me across 5 segments. Since this is over 90 cM, that means that Timber theoretically didn't eliminate any matching -- since the Ancestry white paper on Timber claims that Timber is not used when sharing is over that amount.
I see this question was answered, but for what it's worth.

A bunch of my predicted 3rd cousins or 3rd/4th cousins are really 2 cousins 1xR or sometimes even 2nd cousin.

Some of my known 3rd cousins at Ancestry are (sister in parentheses, line I've been researching lately with stars, as you can see they seem to be lower than average, the others are just the highest cM I've identified by working down from highest cM, I have more known and no doubt more unknown, where my sister is not given that's my laziness, not that she has no match):

213 cM (135) -- predicted 2nd-3rd
138 cM (44 cM) -- predicted 3rd-4th
126 cM (151 cM) -- but I share 147 cM on 23andMe -- predicted 3rd-4th
121 cM -- predicted 3rd-4th
108 cM -- predicted 3rd-4th
101 cM -- predicted 3rd-4th
97 cM -- predicted 3rd-4th (148 cM with my dad at FTDNA, 103 cM with me, so not too off)
95 cM -- predicted 3rd-4th
76 cM -- predicted 4th
73 cM -- predicted 4th-6th
62 cM -- predicted 4th-6th
57 cM -- predicted 4th-6th
*46 cM -- predicted 4th-6th (actually not sure if this is son or father, if father we are 2c1xR)
*42 cM -- predicted 4th-6th
*37 cM (75 cM) -- predicted 4th-6th
*37 cM (42 cM) -- predicted 4th-6th (her sister is 44 cM at MH, so maybe not off, 48 cM with my sister, 98 cM with my mom)
*32 cM (27 cM) -- predicted 4th-6th (her aunt is only 48 cM, predicted the same)
*31 cM (47 cM) -- predicted 4th-6th (her brother is not a match with me, shares 45 cM with my sister)
*23 cM -- predicted 4th-6th

But I do have some higher than expected matches at Ancestry too.

#### Posting Permissions

• You may not post new threads
• You may not post replies
• You may not post attachments
• You may not edit your posts
•