Ethnicity Inheritance Calculator

What this calculator does

It averages pedigrees. Mode 1 takes two parents described by components and percentages (the labels are yours: regions, countries, peoples, in your family’s own terms) and computes the child’s expected fraction for each component as the simple mean of the parents’ fractions, because a child inherits, in expectation, half their ancestry from each parent. Mode 2 evaluates the same halving along a single line: an ancestor n generations back contributes an expected (1/2)^n of a descendant’s ancestry, 50% for a parent, 25% for a grandparent, 3.13% for a great-great-great-grandparent. Both modes report expectations from the pedigree model, and the content around them explains, with the genetics stated correctly, why real DNA results differ from these numbers and why neither kind of number speaks to appearance or identity.

How to use it

• Percentages describe family knowledge, not measurements. A parent entered as 60/40 is stating what the family knows of its origins (two grandparents from one place, one from another, say). The model averages what you state; it cannot check it, and a DNA test’s estimates are a different kind of statement altogether, with their own reference panels and error ranges.
• Use consistent labels across the parents. A component present in both parents must carry the same label in the same position to combine; the page pairs components by row. Splitting one background into different labels on each side double-counts it into two smaller components instead of one true one.
• Let “Other or unspecified” be honest. If a parent’s shares total less than 100, the remainder appears as other/unspecified, which is usually the truth about family records: incomplete. Forcing the known parts to 100 by inflating them manufactures certainty the pedigree does not contain.
• Mode 2 is about expectation over many descendants. The halving is exact as an average and increasingly loose as a prediction for any one person as generations grow, because DNA is inherited in chunks, not blended. By seven or so generations back, some genealogical ancestors contribute no detectable DNA at all to a given descendant; the page says this where the number appears.
• Never read a fraction as a trait forecast. The model contains no biology of appearance at all. Skin, hair, features and everything else a person is cannot be averaged out of pedigree percentages, and any tool claiming otherwise is selling the average as a prophecy.

How it is calculated

• Mode 1: for each component, expected child share = (Parent A share + Parent B share) / 2. Unlisted remainder for each parent = 100 – the sum of that parent’s entered shares, combined the same way into an “Other or unspecified” child share. A parent whose shares total more than 100 is refused, because a pedigree cannot exceed itself.
• Mode 2: expected share = 100 x (1/2)^n percent for n generations, also expressed as 1 part in 2^n. This is the pedigree coefficient of relationship halved per meiosis along one line, the standard expectation of quantitative genetics.

Worked examples

Two parents combined. Parent A is described as Irish 60, Italian 40. Parent B as Irish 10, Italian 30, Japanese 60. The child’s expected fractions: Irish (60 + 10) / 2 = 35%, Italian (40 + 30) / 2 = 35%, Japanese (0 + 60) / 2 = 30%. The shares total 100, as they must, and every figure in the sentence is pedigree bookkeeping: it records where the child’s grandparents came from, weighted by line, and nothing else.

One ancestor five generations back. A great-great-great-grandparent: expected share 100 x (1/2)^5 = 3.13%, one part in 32. That same person’s expected share was 50% in their child and halves down the line regardless of who they were, which is the quiet arithmetic underneath every family tree: ancestry is wide (32 ancestors at that generation) precisely because each share is small.

The boundary case: seven generations. Expected share 0.78%, and the honest commentary changes character: across a population of descendants the average still holds, but for any individual descendant the actual DNA share from that ancestor is a lottery of chunk inheritance whose outcomes include exactly zero. The mode prints the expectation and the warning together, because at depth the warning is half the answer.

Reading your results

Read Mode 1’s output as a summary of the family record you entered: useful for genealogy notes, family histories and understanding what a DNA test’s percentages are roughly responding to. Read Mode 2 as an average with widening error bars. If your DNA test disagrees with either, both can be right: the pedigree expectation assumes every ancestor contributes exactly their average, while meiosis deals actual segments unevenly, favouring some ancestors and skipping others. The disagreement is the biology asserting itself over the bookkeeping.

Limitations

• Expected is not actual. DNA inheritance is random at the segment level. Siblings with identical pedigrees inherit measurably different mixes; a child’s actual share from a grandparent ranges widely around 25% (roughly 18-32% in practice). No pedigree calculator can narrow that, and this one does not pretend to.
• Ethnicity is not a genetic substance. The labels on this page name family origins as families describe them. Ethnic identity additionally involves language, culture, community and self-understanding, none of which halving arithmetic touches. A person with a small expected fraction from a background may be fully of its culture; a person with a large one may not be. The calculator computes the fraction and claims nothing else.
• Deep pedigrees overlap. Far enough back, ancestors repeat (pedigree collapse) and the same person arrives by several lines, so naive halving overcounts the family’s distinct origins. The modes assume distinct ancestors, a good model for the few generations families usually enter.
• Inbreeding and relationship coefficients need the full pedigree. Where ancestors are related to each other, exact coefficients require the complete pedigree with common-ancestor paths (Wright’s method). That computation needs a structured family tree rather than fractions, and it is out of this page’s stated model rather than approximated inside it.

Frequently asked questions

How is a child’s ancestry fraction calculated?

By averaging the parents, component by component, because a child inherits half their ancestry from each parent in expectation. The worked example: parents at Irish 60/Italian 40 and Irish 10/Italian 30/Japanese 60 give a child expected at 35% Irish, 35% Italian, 30% Japanese. The words to keep hold of are “in expectation”: the average is exact, and any real child’s DNA will scatter around it by the randomness of inheritance, which the limitations section quantifies rather than hides.

Is the expected fraction the same as what a DNA test will show?

No, for two separate reasons. First, actual inheritance is lumpy: you receive segments, not a blended half, so realised shares differ from pedigree expectations, increasingly with generations. Second, a DNA test does not measure your pedigree at all; it compares your segments against modern reference populations and assigns probabilities, with its own revisions as panels improve. The pedigree figure is the prior the test roughly refines; agreement within a few points is good agreement, and identity with your result is not a standard either number should be held to.

Can this predict what a child will look like?

No, and it is built to refuse the question. Appearance is polygenic, dominance-structured and partly environmental; it does not average between parents the way pedigree fractions do, and no responsible model reduces it to ancestry percentages. Tools that offer look predictions from ethnicity fractions are performing, not calculating. This page’s outputs are deliberately confined to the arithmetic that exists: expected ancestral shares, with their limits printed where the numbers are.

What share comes from a great-grandparent?

A great-grandparent is three generations back, so the expected share is (1/2)^3 = 12.5%, one part in eight. Mode 2 gives the figure for any depth, and pairs the deeper ones with the caveat that matters most there: by around seven generations the expected 0.78% is an average over descendants, some of whom inherit nothing detectable from that ancestor at all, because DNA segments are finite and not every ancestor’s segments survive every meiosis down a given line.

Related calculators

• Culture & Recreation Calculators
• Numerology Calculator
• Calendar Conversion Calculator

Method: pedigree averaging of user-stated ancestry fractions and single-line halving (1/2)^n, exactly as defined on this page. Expected fractions only: not DNA measurement, not trait prediction, not identity assignment. Last reviewed: 14 October 2026.