I now have over 8,700 Matches at AncestryDNA with a confirmed Common Ancestor (CA) with me between 2C and 8C. See my Common Ancestor Spreadsheet post here. That’s a lot of data, so I thought I’d do some analysis. In 2024 I posted (here) my averages for 3C to 8C which roughly agreed with the Shared cM project.
Below is a table summarizing all of my data (including full cousins, half cousins and removed cousins). For each relationship there are columns for the number of Matches, the average cMs, the lowest cM, the highest cM; plus the number of generations (meiosis events), and average cMs for each. The table is then repeated with a sort based on meiosis events.
A word about meiosis events. They are the count from me up to the CA and then back down to the Match. Like generations… A 1C is 4 events (two up to grandparent (the CA) plus two back down to the Match. The number of meiosis events with a 1C2R is 6 (two up and 4 down). A half relationship adds one to the meiosis events – eg a 4C1R is 11 events; and 4C1Rh is 12 events. These are important because in a mathematical simulation, each event reduces the cM by half. From the Shared DNA Project a 1C (4 events) average is 866cM compared to 2C1R (7 events) is 122cM which is roughly 866 halved three times. Remember, it’s an order of magnitude thing. And, as we shall see, it generally works for close relationships (like 1C and 2C), but drifts away for more distant relationships (like 4C and beyond). Important: this is not biology’s fault, it’s the math’s fault. It’s because we have a LOT of true distant cousins that do NOT share matching DNA with us; and they are not reflected in the averages. This (lack of a normal curve) is highlighted in the second sort (by meiosis numbers) below. This is also reflected in the DNA Painter Shared cM Project tool which shows different groups of Matches for a given input cM value. For example at DNA Painter, plug in 55cM… the 29% group of 3Ch, 3C1R, 2C3R and 2C2R half are all 9 meiosis events; and the second group of 4C, 3c1Rh, and 3C2R are all 10 meiosis events. This also demonstrates that by the time we get down to 3C and 4C levels there is a lot of overlap.
For this first table, the takeaway is that the number of Matches with CAs increased dramatically with each generation. [Note I combine full cousin with cousin 1R because at my age, most Matches will be a generation younger that me] 3C & 3C1R: 196 Matches; 4C & 4C1R: 662 Matches; 5C & 5C1R: 1,406 Matches; 6C & 6C1R: 3,426 Matches. WOW, what an increase in the number of Match cousins. And then we have 7C & 7C1R: 584 Matches; 8C & 8C1R: 363 Matches. What happened? Why the steep decrease in numbers. Well, IMO, the major factor is that AncestryDNA’s ThruLines quits at 6C – ThruLines can “see” into private Trees (I cannot); and it roots out MRCAs with the smallest of Trees (I don’t have that time). I can only dream of how many ThruLines I’d get at the 7C and 8C levels. Some of the ones I have now, were found/recorded when we had Circles at Ancestry.
The point is: there are LOTS of cousins still waiting to be determined. ProTools is helping.
Table 1: 8,799 AncestryDNA Matches Summarized by Relationship
| AncestryDNA | cM | cM | cM | |||
| MRCA | #Matches | avg | low | high | meiosis | |
| 1C2Rh | 3 | 138 | 78 | 200 | 7 | |
| 2C | 1 | 269 | 6 | |||
| 2C1R | 14 | 127 | 34 | 220 | 7 | |
| 2C2R | 8 | 47 | 39 | 162 | 8 | |
| 2C3R | 2 | 34 | 22 | 140 | 9 | |
| 3C | 57 | 63 | 13 | 208 | 8 | |
| 3Ch | 5 | 20 | 16 | 95 | 9 | |
| 3C1R | 139 | 28 | 6 | 148 | 9 | |
| 3C1Rh | 26 | 28 | 7 | 111 | 10 | |
| 3C2R | 106 | 22 | 6 | 68 | 10 | |
| 3C2Rh | 20 | 20 | 6 | 92 | 11 | |
| 3C3R | 34 | 22 | 6 | 58 | 11 | |
| 3C3Rh | 12 | 23 | 8 | 40 | 12 | |
| 3C4R | 1 | 20 | 12 | |||
| 3C4Rh | 2 | 10 | 8 | 12 | 13 | |
| 4C | 128 | 24 | 6 | 220 | 10 | |
| 4Ch | 7 | 12 | 6 | 19 | 11 | |
| 4C1R | 534 | 20 | 6 | 114 | 11 | |
| 4C1Rh | 33 | 16 | 6 | 30 | 12 | |
| 4C2R | 267 | 16 | 7 | 92 | 12 | |
| 4C2Rh | 12 | 12 | 6 | 39 | 13 | |
| 4C3R | 27 | 16 | 6 | 44 | 13 | |
| 4C4R | 1 | 17 | 17 | 17 | 14 | |
| 5C | 469 | 16 | 6 | 62 | 12 | |
| 5Ch | 29 | 17 | 6 | 27 | 14 | |
| 5C1R | 1137 | 14 | 6 | 60 | 13 | |
| 5C1Rh | 7 | 14 | 6 | 60 | 14 | |
| 5C2R | 300 | 14 | 6 | 41 | 14 | |
| 5C2Rh | 9 | 14 | 7 | 27 | 15 | |
| 5C3R | 75 | 14 | 6 | 40 | 15 | |
| 5C3Rh | 1 | 18 | 18 | 18 | 16 | |
| 5C4R | 2 | 10 | 10 | 10 | 16 | |
| 6C | 1922 | 12 | 6 | 56 | 14 | |
| 6Ch | 97 | 11 | 6 | 25 | 15 | |
| 6C1R | 1503 | 12 | 6 | 52 | 15 | |
| 6C1Rh | 58 | 10 | 6 | 22 | 16 | |
| 6C2R | 618 | 12 | 6 | 44 | 16 | |
| 6C2Rh | 47 | 12 | 6 | 29 | 17 | |
| 6C3R | 12 | 15 | 6 | 30 | 17 | |
| 7C | 262 | 13 | 6 | 41 | 16 | |
| 7Ch | 10 | 15 | 6 | 39 | 17 | |
| 7C1R | 322 | 12 | 6 | 43 | 17 | |
| 7C1Rh | 7 | 17 | 6 | 43 | 18 | |
| 7C2R | 17 | 16 | 6 | 25 | 18 | |
| 7C3R | 5 | 15 | 6 | 18 | 19 | |
| 8C | 310 | 12 | 6 | 35 | 18 | |
| 8Ch | 6 | 10 | 7 | 17 | 19 | |
| 8C1R | 53 | 16 | 6 | 37 | 19 | |
| 8C2R | 12 | 17 | 8 | 19 | 20 | |
| 8C3R | 7 | 10 | 6 | 13 | 21 | |
| 9C | 63 | 14 | 6 | 24 | 20 | |
| Total | 8799 | |||||
For the second table; the takeaway is that the average cM tracks pretty close to each other at the same meiosis numbers. And after meiosis level 9 which averages 27cM; the “curve” quickly “flatlines” in the mid teens. This is reflected at DNA Painter with many relationships all in play under 20cM.
Table 2: 8,799 AncestryDNA Matches Summarized by Meiosis Events
| AncesttryDNA | cM | cM | cM | |||||
| MRCA | #M | avg | low | high | meiosis | avg cM | ||
| 2C | 1 | 269 | 6 | 269 | ||||
| 1C2Rh | 3 | 138 | 78 | 200 | 7 | |||
| 2C1R | 14 | 127 | 34 | 220 | 7 | 132 | ||
| 2C2R | 8 | 47 | 39 | 162 | 8 | |||
| 3C | 57 | 63 | 13 | 208 | 8 | 55 | ||
| 2C3R | 2 | 34 | 22 | 140 | 9 | |||
| 3Ch | 5 | 20 | 16 | 95 | 9 | 27 | ||
| 3C1R | 139 | 28 | 6 | 148 | 9 | |||
| 3C1Rh | 26 | 28 | 7 | 111 | 10 | |||
| 3C2R | 106 | 22 | 6 | 68 | 10 | 25 | ||
| 4C | 128 | 24 | 6 | 220 | 10 | |||
| 3C2Rh | 20 | 20 | 6 | 92 | 11 | |||
| 3C3R | 34 | 22 | 6 | 58 | 11 | 18 | ||
| 4Ch | 7 | 12 | 6 | 19 | 11 | |||
| 4C1R | 534 | 20 | 6 | 114 | 11 | |||
| 3C3Rh | 12 | 23 | 8 | 40 | 12 | |||
| 3C4R | 1 | 20 | 12 | |||||
| 4C1Rh | 33 | 16 | 6 | 30 | 12 | 18 | ||
| 4C2R | 267 | 16 | 7 | 92 | 12 | |||
| 5C | 469 | 16 | 6 | 62 | 12 | |||
| 3C4Rh | 2 | 10 | 8 | 12 | 13 | |||
| 4C2Rh | 12 | 12 | 6 | 39 | 13 | 13 | ||
| 5C1R | 1137 | 14 | 6 | 60 | 13 | |||
| 4C3R | 27 | 16 | 6 | 44 | 13 | |||
| 4C4R | 1 | 17 | 17 | 17 | 14 | |||
| 5Ch | 29 | 17 | 6 | 27 | 14 | |||
| 5C1Rh | 7 | 14 | 6 | 60 | 14 | 15 | ||
| 5C2R | 300 | 14 | 6 | 41 | 14 | |||
| 6C | 1922 | 12 | 6 | 56 | 14 | |||
| 5C2Rh | 9 | 14 | 7 | 27 | 15 | |||
| 5C3R | 75 | 14 | 6 | 40 | 15 | 13 | ||
| 6Ch | 97 | 11 | 6 | 25 | 15 | |||
| 6C1R | 1503 | 12 | 6 | 52 | 15 | |||
| 5C3Rh | 1 | 18 | 18 | 18 | 16 | |||
| 5C4R | 2 | 10 | 10 | 10 | 16 | 13 | ||
| 6C1Rh | 58 | 10 | 6 | 22 | 16 | |||
| 6C2R | 618 | 12 | 6 | 44 | 16 | |||
| 7C | 262 | 13 | 6 | 41 | 16 | |||
| 6C2Rh | 47 | 12 | 6 | 29 | 17 | |||
| 6C3R | 12 | 15 | 6 | 30 | 17 | 13 | ||
| 7Ch | 10 | 15 | 6 | 39 | 17 | |||
| 7C1R | 322 | 12 | 6 | 43 | 17 | |||
| 7C1Rh | 7 | 17 | 6 | 43 | 18 | |||
| 7C2R | 17 | 16 | 6 | 25 | 18 | 15 | ||
| 8C | 310 | 12 | 6 | 35 | 18 | |||
| 7C3R | 5 | 15 | 6 | 18 | 19 | |||
| 8Ch | 6 | 10 | 7 | 17 | 19 | 13 | ||
| 8C1R | 53 | 16 | 6 | 37 | 19 | |||
| 8C2R | 12 | 17 | 8 | 19 | 20 | 15 | ||
| 9C | 63 | 14 | 6 | 24 | 20 | |||
| 8C3R | 7 | 10 | 6 | 13 | 21 | |||
| Total | 8799 | |||||||
Sidebar – this evaluation also acts as a Quality Control indicator. Watch for data points way outside the norms. I had three Matches who skewed one of the numbers. I went back to them – they were close to each other and I was sure they were from an NPE. Upon reevaluation, they needed to be a generation closer to our CA. I made the shift, and all the numbers fell back into the norm.
These insights are helping me with a new review of Walking The Clusters Back, where in I need to use judgment when imputing relationships and CAs.
[06G] Segment-ology: Insights into cM Patterns; by Jim Bartlett 20260122
If I had to describe what one of worst nightmare situations would be it would be having to live with this man.
LikeLike