The idea: a population that doesn't change
In 1908 a mathematician (G. H. Hardy) and a physician (Wilhelm Weinberg) independently showed something reassuring: in a large, randomly mating population with no mutation, no migration, no selection, and no drift, allele frequencies don't drift apart on their own. The Hardy–Weinberg equilibrium is the genetic version of an object at rest staying at rest.
Under those conditions two things hold. First, the allele frequencies p and q stay constant generation after generation. Second, the genotype frequencies settle, after a single generation of random mating, into a fixed pattern: AA = p², Aa = 2pq, aa = q².
Why p^2 + 2pq + q^2 ? Random mating = drawing two alleles at random
Egg carries A with prob p, a with prob q
Sperm carries A with prob p, a with prob q
sperm A (p) sperm a (q)
egg A (p) | AA = p*p = p^2 | Aa = p*q |
egg a (q) | Aa = q*p | aa = q*q = q^2 |
Sum the offspring genotypes:
AA = p^2
Aa = p*q + q*p = 2pq
aa = q^2
Total: p^2 + 2pq + q^2 = (p + q)^2 = 1^2 = 1Using it: the carrier-frequency trick
The equation earns its keep with recessive conditions. If you only know how often an aa phenotype appears, Hardy–Weinberg lets you estimate the hidden carriers (Aa) — people who carry one copy but don't show the trait.
- Start from the recessive phenotype frequency: q² = freq(aa).
- Take the square root to get q, the recessive allele frequency.
- Get p = 1 − q.
- Carrier frequency = 2pq. Compare it to the patient frequency q² — carriers are usually far more common.
A recessive condition affects 1 in 2500 people. q^2 = freq(affected) = 1/2500 = 0.0004 q = sqrt(0.0004) = 0.02 p = 1 - q = 0.98 Carriers (Aa) = 2pq = 2 * 0.98 * 0.02 = 0.0392 ≈ 1 in 25 Affected (aa) = q^2 = 0.0004 = 1 in 2500 So ~100 carriers exist for every affected person. (Educational illustration of the math — not medical advice.)
What it's really for: the null hypothesis
No real population perfectly meets all the assumptions — so why bother? Because Hardy–Weinberg is a null model. It tells you what genotype frequencies you'd see if nothing interesting were happening. When real counts deviate from p², 2pq, q², that deviation is a signal: maybe selection, maybe non-random mating, maybe drift in a small group. The equilibrium is valuable precisely because it tells you when to look harder.