1Correct separation: haploid, and a mix of both parents

When meiosis works, the rules from 5.1 guarantee two things about each gamete. It is haploid (1n): exactly one chromosome from every pair, because the homologues of each pair went to opposite poles in anaphase I. And it is a mixture: some of its chromosomes are copies of the ones the organism got from its mother, others from its father, because each pair separated independently of the others.

So you did not pass on "your mother's half" or "your father's half". Each of your gametes carries a blend: perhaps your mother's chromosome 1, your father's chromosome 2, your mother's chromosome 3, and so on.

2Crossing over: new chromosomes

During prophase I, while homologues are held together in synapsis, non-sister chromatids (one from each homologue) can break at matching points and rejoin to each other. The point where they cross is visible as a chiasma. The exchange is called crossing over, or recombination.

Look at Figure 1(b). One homologue carries alleles A and B (teal); the other carries a and b (amber). After a crossover between the two genes, two of the four chromatids are unchanged, parental (AB and ab), and two are recombinant (Ab and aB): combinations that did not exist on either of the chromosomes the organism inherited. When the chromatids are sorted into four gametes, each gets one of these four.

Two details are worth holding on to. Crossing over happens between non-sister chromatids; an exchange between sisters would swap identical pieces and change nothing. And it happens in prophase I, the only time homologues are paired.

3Random assortment: new mixtures of whole chromosomes

At metaphase I each homologous pair lines up on the plate, and which homologue faces which pole is random and independent of every other pair. Figure 1(a) shows a cell with two pairs. In orientation 1 both teal chromosomes face the same pole; in orientation 2 they face opposite poles. The two orientations are equally likely, and between them they give 4 kinds of gamete.

The count generalises. With n pairs, each pair has two ways to face, so the number of possible chromosome combinations in a gamete is

$$2^n$$

Worked example. Humans have n = 23. The number of different combinations of maternal and paternal chromosomes one person can put into a gamete, from random assortment alone, is $2^{23} = 8\,388\,608$, about 8.4 million. An invented plant with 2n = 10 has n = 5 and makes $2^5 = 32$ kinds of gamete by assortment. Use n, not 2n: a common slip is to write $2^{46}$ for humans.

4Random fertilisation: two gametes, chosen by chance

Any one of a father's possible sperm can fertilise any one of a mother's possible eggs. Using the laws of probability from the formula sheet, the two events are independent, so the numbers multiply: from assortment alone, two humans could produce $2^{23} \times 2^{23} = 2^{46}$, about $7.0 \times 10^{13}$, different chromosome combinations in a zygote. Crossing over makes the true number far larger, because it creates new chromosomes, not just new mixtures of existing ones.

Fertilisation also restores the diploid number and brings together alleles from two unrelated lines, which is why sexual reproduction makes more varied offspring than asexual reproduction, where offspring are copies of one parent.

5Nondisjunction: when separation fails

If a homologous pair fails to separate in anaphase I, or a pair of sister chromatids fails to separate in anaphase II, the gametes are no longer haploid. One gamete gets an extra copy of that chromosome (n + 1) and another lacks it (n − 1). Fertilisation by a normal gamete then gives a zygote with three copies (trisomy, 2n + 1) or one copy (monosomy, 2n − 1) of that chromosome. In humans, trisomy 21 causes Down syndrome, and a single X chromosome with no second sex chromosome (XO) causes Turner syndrome.

When the error happens decides how many gametes are affected, which is a favourite exam calculation.

• In meiosis I (Figure 2): both homologues go to one pole. That cell's two products both get two copies (n + 1); the other cell's two products get none (n − 1). All four gametes are abnormal.
• In meiosis II: meiosis I was normal, so the error is in only one of the two cells, where the sister chromatids of one chromosome both go to one pole. That cell gives one n + 1 and one n − 1 gamete; the other cell gives two normal gametes. Half the gametes are abnormal.

Nondisjunction in meiosis I makes all four gametes abnormal; in meiosis II, only two of the four.

6Asking a testable question

AP expects you to turn an observation into a question that an experiment could answer. A testable question names a variable you can change or compare (the independent variable) and a result you can measure (the dependent variable), and it can be answered with data.

Observation (invented): in a lab colony of fruit flies kept warm, more offspring than usual have abnormal numbers of one chromosome.

• Not testable: "Why is meiosis so complicated?" (no variable to measure).
• Not testable as written: "Is heat bad for flies?" ("bad" is not measurable).
• Testable: "Does raising the rearing temperature of female flies from 20 °C to 28 °C increase the percentage of their offspring with an extra copy of chromosome 4?"

From a testable question you can write a null hypothesis: temperature has no effect on the percentage of offspring with an extra chromosome. 5.3 shows how a chi-square test decides whether to reject a null hypothesis.