I have documented fifth cousins who share no detectable DNA with me — how is that possible?

The underlying reason is worth understanding because it is counter-intuitive: you do not inherit DNA from most of your ancestors.

Go back ten generations and you have over a thousand ancestral slots. Your genome is broken into a limited number of inherited segments, and with each generation those segments are fewer and larger. Beyond roughly eight or nine generations, the number of distinct segments you carry is smaller than the number of ancestors you have.

The conclusion is unavoidable: many of your documented ancestors from that era contributed nothing to your genome. They are genuinely your ancestors, genealogically, and genetically they are absent.

That surprises people badly the first time they meet it. Genealogical descent and genetic descent are different relationships, and they diverge quickly.

Your records are probably fine. Sharing no detectable DNA with a genuine fifth cousin is not merely possible, it is the more likely outcome.

The mechanism is that inheritance is a lottery at every step. You get half your DNA from each parent, but which half is chosen essentially at random, in large chunks. Each generation halves the expected share from any given ancestor, and — the crucial part — the variance around that expectation is enormous.

Rough expectations for autosomal DNA:

• First cousins: about 12.5%, and always detectable.
• Second cousins: about 3%, essentially always detectable.
• Third cousins: about 0.8%, detected roughly 90% of the time.
• Fourth cousins: about 0.2%, detected around half the time.
• Fifth cousins: around 0.05%, detected perhaps 10 to 15% of the time.

So for fifth cousins, not matching is the normal result and matching is the exception.

Two practical points for the matching you are doing.

Testing company matters. Different companies use different chips and different thresholds for reporting a match, and they have different customer pools. A relative who does not appear on one platform may appear on another, and uploading raw data to a third-party comparison site widens the pool considerably.

Small matches are unreliable in both directions. At the very low end, some reported matches are false — short segments that look shared by chance rather than by descent. So a tiny match with a distant cousin is weak evidence and no match is weak counter-evidence. Neither should move your confidence in a documented line very much.

Which means for anything beyond about third cousins, the paper trail is the stronger evidence and the DNA is the supporting hint, not the other way round.

Worth adding the one line of inheritance that does not behave this way, since it can be useful.

Mitochondrial DNA passes down the direct maternal line essentially unchanged, and the Y chromosome down the direct paternal line. Those do not get shuffled, so they persist over far more generations than autosomal segments.

They only test one thread each through your tree — your mother's mother's mother, or your father's father's father — but for those specific lines they are far more informative at distance than the general test you have taken.