Finding Your Roots: How DNA and Heraldry Tell Your Family Story

An autosomal ancestry test reads several hundred thousand points across your genome and compares the pattern against reference panels of people whose grandparents all came from one region. The output is an estimate of how much of your DNA resembles each panel.

That is a genuinely powerful thing, and it has two parts of very different reliability.

Ethnicity estimates are the weaker half. They are statistical inferences against reference panels, not measurements, and this is why your percentages change when a company updates its panels — the DNA did not change, the comparison did. They are reasonably good at continental scale, decent at broad regional scale, and poor at distinguishing neighbouring European populations that have been intermarrying for a thousand years. "43% Irish, 28% English" is a model output with error bars, not a fact about your ancestors.

DNA matches are the stronger half, and the half most people underuse. A shared segment of DNA with another living person is a measurement, not an estimate. Close matches are reliable; distant ones fall away fast, because autosomal DNA is diluted by half each generation and inherited at random. Past about six or seven generations, a genuine ancestor may contribute no detectable DNA to you at all.

Two specialised tests behave differently and matter here. Y-DNA follows the direct male line, which is the same line surnames and arms descend along. Mitochondrial DNA follows the direct female line. Both are far deeper in time and far narrower in scope than an autosomal test.

• It produces no names, places or dates. Ethnicity percentages name no ancestor and no county.
• It cannot tell you which line a percentage came from. Knowing you are 43% Irish does not say whether that is one Irish grandparent or four Irish great-grandparents.
• It fades. Beyond a few generations, a real ancestor can be genetically invisible to you.
• It has nothing to say about heraldry. There is no genetic signature of a coat of arms, and no test will tell you what your family bore.

Heraldry works on documents, and its unit is not a population but a person.

A coat of arms in the record is a grant or an assumption by a named individual, borne in a named place, at a date, and inherited by his descendants. Where a record exists it can be specific in a way DNA never is: the arms of a particular family, in a particular parish, recorded in a visitation in a particular year.

What it also does is date things. A shield bearing a label tells you the bearer's father was alive. A quartered shield asserts a marriage that either happened or did not. An augmentation of honour was granted for a named act on a named day.

The limits are just as firm, and they cut against most of what is sold on the subject.

• Arms belong to people, not to surnames. Everyone called Murphy is not entitled to one Murphy coat, because Murphy was adopted independently by unrelated families in different provinces. An armorial lists several distinct Murphy coats, and that is the correct answer.
• The symbols do not describe your ancestors' character. The tables assigning virtues to charges and colours come from a seventeenth-century literary fashion, not from heralds. A. C. Fox-Davies, whose Complete Guide to Heraldry is still the standard reference, called the symbolism of colours "absolutely and entirely fiction". Our guide to what heraldic symbols actually record sets out what they do mean instead — mostly puns on the name, marriages, and specific granted events.
• Most families never had arms at all. That is not a defect in your family; it is a fact about medieval Europe.

Neither DNA nor heraldry can reach the other. The bridge between them is paper — parish registers, censuses, wills, land records, immigration manifests.

DNA gives you a region and a set of living matches. Heraldry gives you arms attached to a named family in a named place. The only thing that can join the two is a documented line of descent, and that is built one generation at a time out of records.

This is not a limitation to be worked around. It is the work.

1. Test, and use the matches rather than the percentages. Your matches' trees are the useful output. If you are tracing a surname line specifically, a Y-DNA test on a man of that surname is far more targeted than an autosomal one.
2. Build the documented tree backwards, one generation at a time, from what you can prove. Never work forwards from a surname you like the look of.
3. Get to a place. A parish, a townland, a village. This is the step that decides everything, and for a common surname it is the only step that makes the name mean anything.
4. Now look at the armorials — Burke's for Britain and Ireland, Rietstap for the Continent, Siebmacher for the German lands — for arms recorded to that family in that place.
5. Be prepared for the answer to be no. It usually is, and a source that always says yes is not reading a record.

Our guide to verifying a coat of arms covers step four in detail, including which sources are free.

One real point of contact worth knowing about, because it is the closest thing to a link between the two fields.

Surnames and Y-chromosomes both descend along the male line, so in principle men sharing a surname should share a Y haplotype. Surname DNA projects test exactly this — and what they consistently find is instructive in both directions.

Common surnames split into many unrelated genetic clusters, confirming what the armorials already say: the name was invented many times. And even within a single documented line, a small percentage of transmissions per generation do not match the paper trail — through adoption, remarriage, or paternity not being what the record says. Over ten generations that compounds.

Which is the honest summary of this whole subject: a surname is a strong clue and a weak proof, and so is a percentage.

Worth being exact, since this article is about what each method can and cannot establish.

A Heritage Volume is a book about a surname, produced by a system we built. The same surname always produces the same volume. Parts of it are genuine public record, shown with their source and a link — U.S. Census Bureau surname figures, digitised newspaper pages from the Library of Congress, documented people from open reference databases, and the etymology of the name. Other parts, including the coat of arms image, are produced by our system rather than retrieved from an armorial.

It is a documented account of a name, not a genealogical investigation of your family and not a DNA test. How Your Report Is Made sets all of that out line by line, before you buy anything.