Hardy-Weinberg Equilibrium Explained in Plain English

The 1908 answer to an awkward problem

Early geneticists kept running into an objection that sounded fatal. If dominant alleles always show and recessive ones hide, shouldn't dominance drive recessive alleles out of populations within a few generations? In 1908 G. H. Hardy, a Cambridge mathematician, answered with a short note in Science (Hardy, Science, 1908): in a large randomly mating population, genotype frequencies settle into a predictable relationship with allele frequencies and stay there, dominance or not. Wilhelm Weinberg, a Stuttgart physician, reached the same result independently the same year, which is why the relationship carries both names. Edwards' reconstruction of the priority dispute (Edwards, Genetics, 2008) makes the point that matters here: the equilibrium was never a claim about nature behaving perfectly. It was a null model: the state a population drifts around when nothing else is acting.

Two equations, one worked example

Fix one locus with two alleles, frequencies p and q. Allele frequencies sum to one, so p + q = 1. Under random mating the genotype frequencies become p², 2pq, and q² for the two homozygotes and the heterozygote, and p² + 2pq + q² = 1 follows automatically: the binomial expansion of (p + q)². Work it with p at 0.6: expect 36 percent homozygous p, 48 percent heterozygous, 16 percent homozygous q.

p + q = 1
p² + 2pq + q² = 1
p = (2 × homozygous-p count + heterozygous count) ÷ (2 × total individuals)

The second equation is the one people actually use. Count genotypes in a sample (say 48, 40, and 12 individuals of the three genotypes out of 100), and you get p = (96 + 40) ÷ 200 = 0.68, q = 0.32, ready to compare against what random mating would predict.

The five assumptions

Equilibrium holds only while five conditions hold, and every assumption names a force that changes allele frequencies when it breaks:

• A large population. In small groups chance alone drifts allele frequencies: sampling noise, not selection.
• No mutation. New alleles arrive at low rates and push frequencies upward from nothing.
• Random mating. Inbreeding and assortative mating shift genotype frequencies even when allele frequencies hold still.
• No migration. Alleles entering or leaving a population change its mix directly.
• No natural selection. When some genotypes leave more offspring, allele frequencies march: the classic case being sickle-cell protection against malaria, where Allison's fieldwork in Africa (Allison, BMJ, 1954) showed the recessive allele persisting because carriers survived malaria better.

What the arithmetic is actually used for

Two everyday jobs dominate. First, estimating carrier frequency from disease prevalence: if a recessive condition affects roughly 1 in 2,500 births, then q² is about 0.0004, q is about 0.02, and carriers (2pq) sit near 4 percent: one in twenty-five, the cystic fibrosis screening figure reported across Northern European populations (Ioannou and colleagues, Genetics in Medicine, 2014). Second, quality control: genotyping studies test observed samples against Hardy-Weinberg expectations, and a locus wildly out of equilibrium usually means a scoring error or hidden population structure rather than evolution in a laboratory sample.

Why the mismatch is the point

No real population sits exactly at equilibrium (large, isolated, randomly-mating groups with no selection are an abstraction), and that is the useful part. A sample with fewer heterozygotes than expected usually signals inbreeding or a mixture of subpopulations pooled together; an excess can flag a typing error; a steady shift over generations points to selection or migration. The equations give you the expected column of the table, and science lives in the residuals. Keep the model one locus wide: Hardy-Weinberg arithmetic says nothing about two genes at once, and linkage, epistasis, and trait continua such as eye color all fall outside it.

Why one generation of random mating is enough

Allele frequencies and genotype frequencies move at different speeds, and that asymmetry is the entire mechanism. Allele frequencies change only when transmission changes: selection, mutation, migration, or plain sampling chance in a small group. Genotype frequencies are reassembled every generation by whichever matings actually happen, so under random mating the proportions snap to p², 2pq and q² almost at once, whatever the parents' generation looked like. That is the shape of Hardy's 1908 answer to the dominance objection. A recessive allele can sit hidden in heterozygotes for generations without disappearing, because hiding changes what an observer sees, not what gets passed on. Dominance describes expression; p and q describe inheritance; the equations track only the second. A population can therefore look completely changed in outward trait while its allele frequencies stand still, which is exactly the state this model predicts when nothing else is acting on it.

Tracing a screening figure back to prevalence

The carrier estimates behind screening programs all run the same three steps backwards through the two equations. Prevalence of an autosomal recessive condition is q²: the chance that a random individual draws two copies. Take the square root to recover q, then evaluate 2pq for the share carrying one copy. On the worked condition above (1 in 2,500 births), that is q² = 0.0004, q = 0.02, and 2pq = 2 × 0.98 × 0.02 = 0.0392, the near-4-percent figure the guide gives for cystic fibrosis screening across Northern European populations. What makes the result feel wrong is the square root: it inflates a rare disease into a common carrier pool, because the equations count carriers whether or not they ever show a symptom. The same arithmetic also runs in the other direction, from observed genotype counts to p and q and on to an expected column, and that pair of directions is what makes the model useful on both sides of a clinic.

Worked example: Hardy-Weinberg Equilibrium Explained in Plain English

Take the counts from the guide: 48 homozygous-p, 40 heterozygous and 12 homozygous-q individuals out of 100. Allele counting gives p = (96 + 40) ÷ 200 = 0.68 and q = 0.32. Random mating then predicts p² = 0.4624, 2pq = 0.4352 and q² = 0.1024, so about 46, 44 and 10 individuals. Observed runs 48, 40, 12: a few more homozygotes and a few fewer heterozygotes than the ideal, a small deficit worth a second look at the data rather than a headline about evolution. Run these counts in the calculator →

Common mistakes: Hardy-Weinberg Equilibrium Explained in Plain English

• Counting alleles as individuals: double the heterozygotes and divide by twice the sample size, or p drifts low every time you run the tally.
• Expecting a real population to sit exactly on the equations: treat any deviation as a prompt to re-check genotyping and sample structure before you call it selection.
• Carrying the model past one locus: Hardy-Weinberg describes a single gene with two alleles, so split two-gene questions into separate single-locus counts.
• Taking the square root of a prevalence you have not matched to an autosomal recessive: confirm the inheritance pattern first, or the carrier figure means nothing.

Put it to work: Hardy-Weinberg Equilibrium Explained in Plain English

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

• Hardy-Weinberg Equilibrium Calculator

Hardy-Weinberg Equilibrium Explained in Plain English: frequently asked questions

Sources & further reading

• Hardy (1908), Mendelian proportions in a mixed population (Science scan)
• Hardy (1908), Mendelian proportions reprint, full text
• Edwards (2008), G. H. Hardy (1908) and Hardy-Weinberg Equilibrium
• Allison (1954), protection afforded by sickle-cell trait (BMJ)
• Ioannou et al. (2014), population-based carrier screening for cystic fibrosis