What Is a Punnett Square, and How Do You Read One?

Where the grid came from

Reginald Punnett was a Cambridge zoologist working alongside William Bateson when Mendel's paper was rediscovered in 1900 and genetics suddenly needed a teaching language. Punnett's answer was a checkerboard that turned verbal reasoning about crosses into something you could fill in mechanically, and he put it in print in his textbook Mendelism in 1905. Edwards' history of the diagram (Edwards, Studies in History and Philosophy of Biological and Biomedical Sciences, 2012) shows how quickly the device spread from that one textbook into a century of classrooms, while also documenting what Punnett and Bateson already knew in 1905: their crossing experiments in chickens and sweet peas showed that some traits travel together and refuse to sort independently. The square is a tool, not a law, and its inventor was among the first to find where it breaks.

Filling in a single-gene square

Start with two parents and one gene. Write each parent's genotype: for a classic monohybrid cross both are heterozygous Bb, where B is the dominant allele and b the recessive one. Each parent makes two gamete types, one carrying B and one carrying b. Write one parent's gametes across the top columns, the other's down the side rows, then combine each row and column pair into the cell where they meet: one cell per fertilization event, each with equal probability:

That is the entire procedure. The power comes from the fact that allele segregation and random fertilization really do give every cell the same chance, so counting boxes is counting probabilities.

When genes do not sort independently due to physical chromosomal linkage, compute crossover rates and map distance (cM) with our two-point genetic map distance tool.

From boxes to ratios

Four boxes in, two ratios come out. The genotypic ratio reads 1 BB : 2 Bb : 1 bb: one quarter homozygous dominant, one half heterozygous, one quarter homozygous recessive. The phenotypic ratio depends on dominance: if one copy of B is enough to show the trait, the three boxes carrying B look alike and you get the famous 3:1 split, 75 percent dominant phenotype to 25 percent recessive.

When you cannot see a recessive phenotype at all (because carriers look normal), a test cross fixes the problem: cross the unknown individual to a homozygous recessive tester (Bb × bb) and read the offspring. A 1:1 ratio confirms the unknown was heterozygous; all-dominant offspring suggest it was homozygous. This is exactly the logic behind carrier detection for recessive conditions such as cystic fibrosis, where a couple who both test as carriers faces a one-in-four affected-pregnancy risk with each conception.

Two genes at once

Add a second independently assorting gene and the grid grows to four by four: sixteen cells instead of four. For a dihybrid cross of two double heterozygotes (AaBb × AaBb) the phenotypic expectation becomes 9:3:3:1: nine with both dominant traits, three with each single dominant, one with neither. The rule behind the size is simple: each heterozygous parent makes 2ⁿ gamete types, where n is the number of heterozygous pairs, so a trihybrid cross needs 64 cells. That exponential is the honest reason calculators exist. Nobody draws a 64-cell grid by hand, and errors in hand-drawn grids are routine.

One warning belongs in the same breath: the 9:3:3:1 ratio assumes the two genes sort independently. Genes on the same chromosome travel together, the linkage Bateson and Punnett's own poultry crosses uncovered, and linked genes skew the ratios toward parental combinations. Independence is an assumption you check, not a given you inherit.

What the square leaves out

A Punnett square gives expectations, not guarantees. Four offspring will not split exactly 3:1 any more than four coins land two heads and two tails. Each box is an independent event, and small families run hot and cold. The square also stays silent on anything it does not model: how common an allele is in the population (that is Hardy-Weinberg arithmetic), how shade traits such as eye color blend across many genes at once, and how many people in a pedigree carry a silent allele. Use it for what it is (a probability table for one locus in one cross), and reach for the calculator when the grid starts sprawling past a page.

Why dominance hides an allele without erasing it

Dominance is a statement about the phenotype, not about the allele's fate. In the Bb box the recessive b is still present, still replicated, and still packed into half of that parent's gametes; it simply produces no visible difference while B is around. Nothing in the grid deletes it, and nothing about filling the boxes marks it as lesser.

This is why two normal-looking parents can produce a recessive offspring, and why carrier testing matters for conditions such as cystic fibrosis: the silent allele travels through generations of carriers until two of them meet at fertilization. When you read a 3 to 1 split, read it as three boxes where the trait shows and one where it does not, rather than as three copies of one allele against one of another. Read that way, the square explains how a recessive trait can vanish for a generation and reappear in a quarter of the next litter with no change in the alleles at all.

Chaining one square into the next generation

Each square is the starting material for the next one. Cross a homozygous dominant with a homozygous recessive, BB × bb, and every cell comes out Bb: one genotype, no variation, whatever the phenotype looks like. Take those offspring and cross them to each other, and the grid becomes the familiar Bb × Bb with 1 BB : 2 Bb : 1 bb waiting underneath.

The recessive that seemed to vanish in the first cross therefore reappears in a quarter of the next generation, with no new genetic event behind it, because segregation split the pair again exactly as it did before. Two grids in sequence explain why recessive traits skip a generation, and why the second generation rather than the first is where the hidden allele shows itself in the numbers. No new rule is needed between the two steps, because segregation and random fertilization run both grids exactly as they ran the first.

Worked example: What Is a Punnett Square, and How Do You Read One?

Both parents are Bb, so each contributes a B gamete and a b gamete. Write them along the two edges and fill four cells: BB, Bb, Bb, bb. Counting boxes gives 1 BB : 2 Bb : 1 bb, and because one B is enough to show the trait, the three boxes carrying B share a phenotype for a 3 to 1 split. The grid is only bookkeeping for segregation and random fertilization, which is why the same answer appears every time you fill it. Build your own in the Punnett square calculator's worked example

Common mistakes: What Is a Punnett Square, and How Do You Read One?

• Writing parent genotypes along the edges, so put haploid gametes there instead: a Bb parent contributes B and b in separate cells and never the pair Bb.
• Expecting the next litter to land exactly on 3 to 1, when each box is an independent event and the ratio is a per-offspring probability that larger litters settle.
• Assuming a two-gene cross always needs sixteen cells, since the edges come from the gametes each parent can actually make and AaBb × AaBB fills eight cells, not sixteen.
• Trusting 9 to 3 to 3 to 1 for genes on the same chromosome, because linked genes travel together and skew the ratio toward parental combinations, so confirm assortment first.

Put it to work: What Is a Punnett Square, and How Do You Read One?

The calculators below apply exactly what this guide explains: same formula, no mental arithmetic.

• Punnett Square Calculator

What Is a Punnett Square, and How Do You Read One?: frequently asked questions

Sources & further reading

• Edwards (2012), Punnett's square, Studies in History and Philosophy of Science
• Punnett, Mendelism, full text (Project Gutenberg)
• Biography of Reginald Crundall Punnett