# Calculating segment cM from 23andMe matches?

Going through my list of matches I'm working on a triangulation analysis. (Such that if person A matches from position 10-15 and person B matches from position 10-13 I can break out position 10-13=A&B, 14-15=A.

I want to weight each segment by cM and I figured I can derive some segments using subtraction. For example the cM for 14-15 would be the cM of 10-15 minus the cM for 10-13 as reported by 23andMe.

However I see that 23andMe is giving me different cM values for the same segment in some cases. Example below:

What does this mean? In the same test why would there objectively be a different distance for 2 people between 8,307,849 and 12,628,471?

Well that is indeed very peculiar.

I checked my own 23andMe matches. Of 268 matches that have two or more identical start-stop positions with another, 266 had the same cM values, but 2 had different values. My differences are:

This is definitely incorrect on 23andMe's part. They should be consistent in calculating the cM between two matches with the same base pair positions for the start and end of the segment.

The # of SNPs could feasibly be different, because some people may have no-calls, where a reading for a SNP could not be determined. No-calls are treated as matches, but are likely not counted as one of the SNPs compared.

The cM differences is a problem that 23andMe might be unaware of and might want to fix, so it would be worthwhile to report it to their customer service.

• Blaine Bettinger made a recent post on Facebook (closed group of which we;re both members): facebook.com/… Is that relevant here?
– user6485
Jul 8, 2019 at 12:20
• @ColeValleyGirl:. No. Blaine is saying not to rely on the start or end positions being exact. This question is about using the start and end positions to calculate cM. Using start and end positions, even though they are never exact, will give you the best cM estimate. Jul 8, 2019 at 14:37
• Hmm, I can pass the question to 23andMe. Do you think their formula discounts the no-matches when they calculate cM? The ratios aren't exactly the same but I know cM calculation isn't linear... Jul 9, 2019 at 3:00
• Scott: No, no-calls should not affect cM calculation.The cM calculation should be calculated based on just the starting and ending base pair positions. Jul 9, 2019 at 15:35