Genetic Ancestors Calculator
Of your 2ⁿ ancestors n generations back, how many actually passed you any DNA? A simulation of how your chromosomes were cut and handed down.
Genetic Ancestors Calculator
Simulating…
A simulation, run in your browser: each figure is the average of 150 simulated family trees, with the 90% range beside it. "Share" is the fraction of that generation's ancestors who passed you any DNA; "≥ 7 cM" counts those whose DNA a consumer test could detect; "Expected each" is one ancestor's average share of your two genome copies. It covers the 22 autosomes; the X chromosome follows different rules — see the X-DNA inheritance calculator.
How to use the genetic ancestors calculator
Slide from 1 to 20 generations. Generation 1 is your two parents, generation 2 your four grandparents, and so on — 2ⁿ ancestors at generation n.
How many of those ancestors passed you any DNA at all, and how many passed you a stretch long enough for a consumer DNA test to see.
Each generation shows its ancestors, the average number who passed you DNA with the range across simulations, and each ancestor's expected share of your genome.
Tick the simple model to see crossovers without interference — the assumption behind the classic formula, which the table shows alongside.
Genealogical ancestors and genetic ancestors are not the same people
Every generation back doubles your ancestors: two parents, four grandparents, eight great-grandparents, and so on. Your DNA does not double with them. You have one copy of each chromosome from each parent, and each of those copies is a patchwork your parent stitched together from their own two copies. Go back far enough and there are more ancestors than there are patches to go round — so some ancestors, who were genuinely in your family tree, passed you no DNA at all.
How the patchwork is cut
When a parent makes an egg or sperm, each chromosome swaps between their mother's copy and their father's copy at a few points, called crossovers — about one per chromosome per hundred centimorgans of its length. Across all 22 non-sex chromosomes that is around 34 crossovers each time. Going back one generation, every crossover in an ancestor's meiosis cuts your DNA from that ancestor into one more piece. So the number of pieces you carry from a given depth grows by roughly 34 a generation, while the number of ancestors at that depth doubles. Linear growth against doubling: the ancestors win, and the share of them who passed you anything falls.
What the simulation does
The page works backwards from you. It starts with your 44 chromosome copies, one from each parent, and climbs one generation at a time, cutting each ancestor's stretches at the crossovers of that ancestor's own meiosis. The crossover positions come from a model with interference — crossovers on the same chromosome tend to keep their distance — fitted to measured human recombination, on the deCODE genetic map. Each answer is the average of 150 simulated family trees, with the range in which 90% of them fell.
The simple formula, and why the simulation is a little lower
The geneticist Graham Coop worked out a quick estimate in 2013: treat each ancestor's pieces as landing at random, and the expected number of genetic ancestors at generation k is 2ᵏ × (1 − e^(−(22 + 33(k − 1)) ÷ 2^(k − 1))). The table shows it beside the simulation. The simulation comes out slightly lower, because pieces do not land at random: a stretch of chromosome alternates between one ancestor's two parents, so neighbouring pieces tend to belong to related ancestors, and a few more ancestors end up with none.
Two kinds of "no DNA"
An ancestor can pass you DNA in pieces too short for a test to recognise as shared with a distant cousin. The table therefore gives two counts: ancestors who passed you any DNA at all, and ancestors who passed you at least one stretch of 7 centimorgans or more — the shortest segment the mainstream consumer tests keep. The second count stops growing long before the first.
What it leaves out
Real family trees fold back on themselves: go back far enough and the same person appears in more than one place, because distant cousins married. The simulation treats all 2ⁿ ancestors as different people — the ancestor count calculator shows how quickly that stops being true. And it covers only the 22 autosomes; the X chromosome follows its own inheritance pattern.
Why fewer than half of your 1,024 tenth-generation ancestors left you any DNA
You have 2ⁿ ancestors n generations back: 1,024 at generation 10.
Your genome arrives as 44 chromosome copies: 22 from each parent.
Each meiosis adds about one crossover per 100 centimorgans of chromosome — around 34 across the 22 autosomes.
The pieces of your DNA from a given depth grow by about 34 per generation; your ancestors double.
By generation 10, fewer than half of your 1,024 ancestors passed you any autosomal DNA.
Every one of your 16 great-great-grandparents passed you DNA in every simulated family tree.
Graham Coop's 2013 estimate treats each ancestor's pieces as landing at random; real pieces cluster, so the true count is a little lower.
Peter Donnelly worked out the exact theory in 1983, without crossover interference.
Pieces under 7 centimorgans are usually discarded by consumer DNA tests, so detectable ancestors level off early.
The X chromosome is inherited differently and is not part of these counts.
Frequently asked questions
-
Not beyond a few generations. All your parents, grandparents and great-grandparents passed you DNA, but further back the number of ancestors outruns the number of pieces your DNA is cut into, and some ancestors passed you none.
-
In the simulation, every ancestor up to the great-great-grandparents (16 people, generation 4) passed you DNA in every simulated tree. From generation 6 or so, some do not.
-
Any DNA counts every piece, however short. Detectable counts only ancestors who passed you at least one stretch of 7 centimorgans or more — roughly the shortest segment consumer tests keep when matching relatives.
-
Crossovers fall at random, so two people with identical family trees carry different pieces. The table gives the average of 150 simulated trees and the range covering 90% of them.
-
It drops crossover interference — the tendency of crossovers on one chromosome to keep apart — and places them independently. It is the assumption behind the classic estimate, shown in the table for comparison.
-
The formula treats each ancestor's pieces as landing independently. Real pieces cluster — a stretch alternates between one ancestor's two parents — so slightly more ancestors end up with none.
-
No. They are fully your ancestors in your family tree; they simply did not pass on any of the DNA that reached you. Your siblings and cousins may carry DNA from different ones.
-
It is not included. The X passes down by different rules — a man gets his only X from his mother — so it is covered by the separate X-DNA inheritance calculator.
-
No. It treats all 2ⁿ ancestors as different people. In real trees the same people recur far back (pedigree collapse), which the ancestor count calculator explores.
-
No. The simulation runs in your browser.
Method & sources
How it computes
Seeded backward simulation (150 trees) of crossovers on the 22 autosomes: deCODE 2019 sex-averaged map, Housworth-Stahl two-pathway interference model with Campbell et al. (2015) parameters. Each ancestor's stretches are cut at the crossovers of that ancestor's meiosis. Shown beside Coop (2013): 2^k(1 - e^-(22+33(k-1))/2^(k-1)).
What this tool implements
- Every figure comes from a seeded simulation in genealogy-kernel.js, reusing the crossover model and genetic map the shared-cM predictor already documents (docs/genealogy-kernel-sources.md); the same seed gives the same answer on every visit.
- The mechanics are checked against what must hold exactly: with no interference the pieces of your two genome copies at depth k average 2 × (22 + M(k − 1)), M the map length in Morgans (test-asserted within 1%), and every ancestor up to generation 4 passes DNA in every simulated tree.
- Coop's (2013) closed form is shown beside the simulation and the gap is explained, not hidden: it treats each ancestor's pieces as independent, real pieces cluster, so the simulation sits 0-12% lower (no interference) and 0-7% lower (with interference) — both bands test-pinned.
- Two counts, because they answer different questions: any DNA at all, and DNA in a stretch of at least 7 cM, the shortest the mainstream consumer tests keep. Pedigree collapse and the X chromosome are excluded and the page says so.
Sources
- Coop, G. (2013), "How many genetic ancestors do I have?", gcbias.org — the closed-form estimate 2^k(1 - exp(-(22+33(k-1))/2^(k-1))): https://gcbias.org/2013/11/11/how-does-your-number-of-genetic-anc…
- Donnelly, P. (1983), "The probability that related individuals share some section of genome identical by descent", Theoretical Population Biology 23:34-63: https://doi.org/10.1016/0040-5809(83)90004-7
- Halldorsson, B. V. et al. (2019), "Characterizing mutagenic effects of recombination through a sequence-level genetic map", Science 363:eaau1043 (the deCODE map lengths); Housworth, E. A. & Stahl, F. W. (2003), AJHG 73:188-197; Campbell, C. L. et al. (2015), Nat Commun 6:6260 — the crossover model, as documented in docs/genealogy-kernel-sources.md.
Pick up where you left off
Stored only in this browser — never sent to our servers.