Contents
09 — Mitosis and meiosis Part 2 — Transmission genetics

09 — Mitosis and meiosis

Before this: Ch 03 · Ch 04 · Time: ~45 min

Mendel's laws are not laws. They are the observable behaviour of a mechanical process that nobody had seen when he inferred them. This chapter is that process. Everything in Part 2 — ratios, , maps, , the arithmetic of — is downstream of what happens to a pair on a spindle.

What you'll be able to do

The core idea

Two operations, with different invariants.

Mitosis is a copy. The invariant is exactness: each daughter cell receives one complete copy of every . Nothing is shuffled, nothing is lost, the is unchanged. This is growth, repair, and every cell division in your body that isn't making a gamete.

is a reduction. The invariant is halving: each product receives exactly one member of each chromosome pair. But which member is randomised independently for each pair, and before the split the two members physically exchange segments. So meiosis is a copy operation deliberately corrupted in two controlled ways — and those two corruptions are the entire source of the combinatorial variation that genetics studies.

The trick that makes reduction possible is arithmetic: two divisions, one round of replication. Replicate once, divide twice, and you have quartered the DNA content and halved the chromosome number. Everything mechanically interesting about meiosis is in the service of that one-off asymmetry.


1. The vocabulary that trips everyone up

This section is the reason people find hard. Three words, one of which changes meaning depending on when in the cell cycle you ask.

. One and everything attached to it. In G1 that is a single DNA molecule. After S it is two identical DNA molecules held together — and it is still one chromosome, because there is still one centromere.

Chromatid. One of the two DNA molecules in a replicated chromosome. The two are sister chromatids: made by replicating the same template in S phase, therefore identical apart from the occasional replication error.

Homolog. The other copy of the same chromosome — the one that came from your other parent. Same genes, same order, same length. Not identical: two human homologs differ at roughly one base in a thousand, which over a 100 Mb chromosome is around 100,000 differences. Every heterozygous site you carry is a difference between two homologs.

        G1 — before replication              G2 — after replication

     maternal        paternal            maternal          paternal
       chr 7           chr 7               chr 7             chr 7

        │               │                 │   │             │   │
        │ A             │ a               │A  │A            │a  │a
      ══╪══           ══╪══             ══╪═══╪══         ══╪═══╪══
        │ B             │ b               │B  │B            │b  │b
        │               │                 │   │             │   │
        ↑               ↑                 └─┬─┘             └─┬─┘
    centromere      centromere         2 SISTER CHROMATIDS  2 SISTER CHROMATIDS
                                        identical to        identical to
                                        each other          each other
     1 chromosome    1 chromosome        1 chromosome        1 chromosome
     1 chromatid     1 chromatid         2 chromatids        2 chromatids

     ←──────── HOMOLOGS ────────→        ←──────── HOMOLOGS ────────→
     same loci (A, B), different alleles (A/a, B/b) — inherited from
     different parents, and differing at ~1 base in 1,000 along their length

Count centromeres, not arms. A replicated chromosome that looks like an X under the microscope is one chromosome, not two. The count changes only when the centromere splits — at anaphase of mitosis, or anaphase II of meiosis, and at no other moment.

The standard bookkeeping is n (number of chromosome sets) and c (DNA content, in units of one unreplicated haploid genome). For a human somatic cell:

StageChromosomesChromatidsn, c
G146462n, 2c
G2 / mitotic prophase → metaphase46922n, 4c
Mitotic anaphase (centromeres split)92924n, 4c
Each mitotic daughter, G146462n, 2c
Meiotic prophase I → anaphase I46922n, 4c
After meiosis I2346n, 2c
After meiosis II (gamete)2323n, 1c

Note the row in bold: at anaphase I the chromosome count is still 46, because the centromeres have not split — whole replicated chromosomes are being pulled apart from each other. This is the single most common counting error.

2. The cell cycle is a state machine with guards, not a timer

Division is a small fraction of a cell's life. A typical cultured mammalian cell runs a ~24 h cycle of which mitosis is under an hour; the rest is interphase — G1, S, G2. (Treat those durations as a teaching approximation: they vary enormously by cell type. A cleaving embryo can cycle in under 30 minutes; a hepatocyte may not divide for a year.)

graph LR G1["<b>G1</b><br/>growth · 2c"] -->|"restriction point<br/>mitogens? size?<br/>DNA intact?"| S["<b>S</b><br/>replication<br/>2c → 4c"] S --> G2["<b>G2</b><br/>4c · finish repair"] G2 -->|"G2/M checkpoint<br/>replication complete?<br/>damage repaired?"| M["<b>M</b><br/>segregate"] M -->|"spindle assembly checkpoint<br/>every kinetochore attached<br/>and under tension?"| G1 G1 -->|"no mitogens /<br/>terminal differentiation"| G0["<b>G0</b><br/>quiescent"] G0 -->|"re-entry<br/>(some cell types only)"| G1

The transitions are guarded, not scheduled. Each checkpoint is a condition that must be satisfied before the machinery is allowed to advance, and the mechanism is worth noticing because it is a pattern you already know:

The spindle assembly checkpoint is a barrier that blocks anaphase until every kinetochore in the cell is attached to the spindle and under tension. The signal is emitted by unattached kinetochores — a single one produces a diffusible "wait" signal strong enough to hold the entire cell. It is a barrier where the last worker to report in releases everyone, and where the default is to block. That default matters: it is why mitosis in normal human cells is accurate to roughly one missegregation per 100 divisions — chromosomally unstable cancer lines are closer to one in five — and, as we'll see in §9, why oocytes, in which the checkpoint is comparatively permissive, are not.

G0 is not a stage of the cycle; it is exit from it. Most cells in an adult body are in G0. Some can re-enter, some (neurons, cardiomyocytes) essentially never do.

3. Mitosis: the copy

Replication happens in S , long before the visible drama. Mitosis is purely the step.

StageWhat happens
ProphaseChromatin condenses into visible ; the two centrosomes separate and nucleate a spindle
PrometaphaseNuclear envelope breaks down; microtubules capture kinetochores — the protein platforms built on each
MetaphaseEvery chromosome is bi-oriented: its two sister kinetochores face opposite poles, and it sits at the spindle midplane under balanced tension
AnaphaseSeparase cleaves along the whole chromosome; sisters are pulled to opposite poles
Telophase / cytokinesisEnvelopes reform, chromatin decondenses, the cell divides

Two mechanical facts carry all the weight, and both get inverted in I:

  1. Sister kinetochores face opposite directions, so a captured chromosome is naturally pulled toward bi-orientation.
  2. Cohesin — the ring complex holding sisters together since replication — is cleaved everywhere at once when the checkpoint releases.

Outcome: two daughters, each 2n, each genetically identical to the parent cell and to each other, modulo . That "modulo" becomes the whole of Ch 56.

4. Meiosis: replicate once, divide twice

graph TD A["<b>Primary spermatocyte / oocyte</b><br/>2n 4c — 46 chromosomes, 92 chromatids<br/><i>homologs paired as 23 bivalents</i>"] A -->|"<b>MEIOSIS I</b> — REDUCTIONAL<br/>homologs separate<br/>arm cohesin cleaved<br/>centromeric cohesin protected"| B["2 cells · <b>n 2c</b><br/>23 chromosomes, 46 chromatids"] B -->|"<b>MEIOSIS II</b> — EQUATIONAL<br/>sisters separate<br/>centromeric cohesin cleaved<br/><i>no S phase in between</i>"| C["4 cells · <b>n 1c</b><br/>23 chromosomes, 23 chromatids"]

I is reductional — the halves. Meiosis II is equational — mechanically close to mitosis, but performed on a haploid cell.

Three things make meiosis I different from every mitosis the cell has ever done:

Homologs pair. In mitosis, homologs ignore each other completely. In meiosis I they find each other, align along their entire length, and become a single physical unit — a (or tetrad: two , four chromatids).

Sister kinetochores co-orient. Both sisters of one chromosome are wired to face the same pole. Bi-orientation now happens at the level of the bivalent — maternal chromosome toward one pole, paternal toward the other. This single is what makes the division reductional.

is released in two stages. At anaphase I, cohesin is cleaved along the chromosome arms only; the protein shugoshin shields cohesin at the from separase. So sisters stay joined through meiosis I and are separated only at anaphase II. Two-stage release is what lets one replication serve two divisions.

Notice what that implies: until anaphase I, the two homologs of a bivalent are physically held together by arm cohesion distal to a crossover. No crossover, no arm cohesion holding them, no bivalent.

5. Prophase I: where the exchange happens

Prophase I is not a stage so much as a programme, and it can last decades (§10). Its sub-stages:

Sub-stageEvent
Leptotene condense; the -like protein deliberately cuts the genome, making roughly 200–300 programmed double-strand breaks
ZygoteneHomologs align and the synaptonemal complex begins to assemble between them — a proteinaceous zipper holding them at a fixed ~100 nm spacing along their length
PachyteneSynapsis complete; breaks are repaired using the homolog as template. Only ~20% mature into crossovers — roughly 50 of the ~250 breaks per meiocyte; the rest resolve as non-crossovers (), invisible to a map
DiploteneThe synaptonemal complex dissolves; homologs stay attached only at chiasmata — the cytological manifestation of a crossover
DiakinesisMaximum condensation; the are ready for the spindle

The cell breaks its own genome in hundreds of places, on purpose, and repairs the breaks using the other parent's chromosome as the template. That is what recombination is mechanically; Ch 18 does the biochemistry.

Crossover placement is controlled, not random. Two rules:

Human totals, as ranges because estimates vary with the method: roughly 25–30 crossovers per male gamete and 40–45 per female gamete — about twice that per meiotic cell, since each crossover involves only two of the four chromatids (§6). Females recombine about 1.5–1.7× more than males, and the sexes place their crossovers differently along the chromosome — a fact any genetic map has to accommodate with sex-specific maps.

The obligate crossover is not there to generate variation. It is load-bearing. Arm cohesion distal to a is the only thing holding a bivalent together between diplotene and anaphase I. A bivalent that receives no crossover — an achiasmate bivalent — has nothing connecting its two homologs, so they arrive at the spindle as two independent univalents and segregate at random. Recombination is therefore a structural requirement of accurate , and a shortage of it is a direct cause of . Small chromosomes get fewer crossovers and are likelier to get none, which is part of why chromosome 21 is over-represented among surviving trisomies.

6. Tracing one pair all the way through

One pair, two heterozygous loci A and B, one crossover between them. ▓ marks maternal-origin sequence, ░ paternal.

G1                       2n 2c
   ▓▓▓▓▓▓▓▓▓  A———————B          maternal homolog
   ░░░░░░░░░  a———————b          paternal homolog

S PHASE — each homolog replicates; sisters are identical
                         2n 4c
   ▓▓▓▓▓▓▓▓▓  A———————B  ┐ sisters
   ▓▓▓▓▓▓▓▓▓  A———————B  ┘
   ░░░░░░░░░  a———————b  ┐ sisters
   ░░░░░░░░░  a———————b  ┘

PROPHASE I — synapsis; ONE crossover, between A and B,
             involving one maternal and one paternal chromatid

   ▓▓▓▓▓▓▓▓▓  A———————B          chromatid 1   parental
   ▓▓▓▓▓▓▓░░  A———————b          chromatid 2   RECOMBINANT
                    ✕
   ░░░░░░░▓▓  a———————B          chromatid 3   RECOMBINANT
   ░░░░░░░░░  a———————b          chromatid 4   parental

METAPHASE I — the bivalent bi-orients; the two homologs
              face opposite poles, held together by the chiasma

              ◄── pole            pole ──►
                 [A—B / A—b]  ✕  [a—B / a—b]

ANAPHASE I — homologs separate. Centromeres do NOT split.
             46 chromosomes still, now 23 per side.

   cell 1:  A———————B          cell 2:  a———————B
            A———————b                   a———————b

MEIOSIS II — centromeres split; sisters separate.
             Note the sisters are no longer identical.

   ┌─ A———————B ─┐   ┌─ A———————b ─┐   ┌─ a———————B ─┐   ┌─ a———————b ─┐
   │  parental   │   │ RECOMBINANT │   │ RECOMBINANT │   │  parental   │
   └─────────────┘   └─────────────┘   └─────────────┘   └─────────────┘

Two results fall straight out of that diagram.

A single crossover produces two recombinant and two parental products — because it involves only two of the four chromatids. So even in a that is guaranteed to have a crossover between A and B, only half the gametes are recombinant. This is the mechanical reason can never exceed 0.5, and it is the foundation of the entire mapping framework in Ch 14.

Sister chromatids stop being identical the moment a crossover happens. After prophase I, "sisters are identical copies" is true only proximal to the exchange. This matters for diagnosing (§8).

7. How this produces Mendel

Mendel inferred two rules from counting peas in the 1860s. The theory — Walter Sutton on grasshopper spermatocytes and Theodor Boveri on sea urchin embryos, 1902–1903 — proposed that his abstract "factors" were carried on the chromosomes people were watching down microscopes. Roughly four decades separated the inference from the mechanism.

. Take a heterozygote Aa. A sits on one homolog, a on the other. Anaphase I sends the homologs to opposite cells; anaphase II splits each into two. The four products of one are exactly A, A, a, a.

Look carefully at what that means. The 1:1 ratio is not a probabilistic outcome of a single meiosis — it is exactly 2:2, deterministically, every time. The randomness Mendel saw in his ratios comes from sampling which gametes participate in fertilisation, not from the segregation itself. In fungi, where all four products of a single meiosis stay together in an ascus, you can see the deterministic 2:2 directly; that is tetrad analysis (Ch 14 §6), and it is why fungal genetics was so powerful so early — an ascus that departs from 2:2 is a caught in the act.

. Two loci on different chromosomes, AaBb. At metaphase I each bi-orients independently — which pole the maternal homolog faces is a fair coin, and the coins are uncoupled. Two equally likely configurations:

   orientation 1              orientation 2
   ◄─ A      a ─►             ◄─ A      a ─►
   ◄─ B      b ─►             ◄─ b      B ─►

   gametes: AB, AB, ab, ab    gametes: Ab, Ab, aB, aB

Averaged over meioses: AB : Ab : aB : ab = 1 : 1 : 1 : 1. The 9:3:3:1 dihybrid ratio is just this, squared and collapsed by dominance.

Mendel inferredThe mechanism
Factors come in pairsChromosomes come in homologous pairs
The pair separates so each gamete gets oneAnaphase I separates homologs
Which one is a fair coinThe bivalent's orientation at metaphase I is random
Different factor pairs assort independentlyDifferent bivalents orient independently
Pairs are restored at fertilisationA haploid gamete fuses with a haploid gamete

The correspondence is exact, which is why the hypothesis was compelling — but correspondence is not proof. The proof came from breaking the correlation on purpose. Morgan's white-eyed Drosophila (1910) showed a gene whose inheritance tracked the X chromosome specifically (Ch 13). Bridges (1916) then found rare exceptional flies whose implied had occurred — and confirmed cytologically that those exact flies had the abnormal chromosome constitutions predicted. A genetic anomaly and a visible chromosomal anomaly, co-occurring in the same individuals. That is what settled it.

The caveat Mendel got away with by luck. Independent assortment applies to loci on different chromosomes, or on the same chromosome far enough apart that crossovers randomise them. Loci close together on one chromosome are linked and violate the rule flagrantly. Mendel's seven characters happened to fall in configurations where he never saw it. Had they not, Part 2 would be organised entirely differently.

The scale of the shuffle. With 23 independent bivalents, one human meiosis can produce 2²³ = 8,388,608 distinct combinations of whole parental chromosomes. Two parents: 2⁴⁶ ≈ 7 × 10¹³ zygote genotypes — before recombination. Recombination removes the ceiling entirely: each transmitted chromosome is a mosaic whose ~1–3 switch points fall at positions drawn from a continuous distribution, so the number of distinct gametes a person can make is effectively unbounded. Siblings share 50% of their genome in expectation; the variance around that is set by how many crossovers there are and where they land.

8. Nondisjunction: meiosis I versus meiosis II

is failure to separate — of homologs at anaphase I, or of sisters at anaphase II. The two produce distinguishable products.

  MEIOSIS I NONDISJUNCTION            MEIOSIS II NONDISJUNCTION
  both homologs go to one pole        one cell's sisters fail to split

        [M P]      [ ]                     [M]           [P]
          │         │                     ╱   ╲         ╱   ╲
        MII       MII                  [MM]   [ ]     [P]   [P]

   products:                          products:
     n+1  M P   ← BOTH parents'         n+1  M M   ← two copies of the
     n+1  M P      homologs                        SAME homolog
     n-1                                 n-1
     n-1                                 n    P
                                         n    P
   2 disomic + 2 nullisomic            1 disomic + 1 nullisomic
                                       + 2 normal

The genotypic signature is the diagnostic tool. Consider markers close to the , where crossovers are rare:

Only pericentromeric markers work for this. Distal markers have been scrambled by and will show a mixture. That single fact is what makes the classification tractable and is how the epidemiology below was measured.

For 21, the best-characterised case: about 90% of errors are maternal, and among maternal errors roughly 77% arise in I and 23% in meiosis II. A caveat worth carrying: many apparent MII errors show elevated recombination in the nondisjoined , which suggests the underlying lesion was set up in meiosis I and only manifested later. The classification is of the observable, not necessarily of the cause.

outcomes are covered properly in Ch 20. The headline is that most aneuploid conceptions do not survive — roughly half of first-trimester miscarriages are chromosomally abnormal — and that the survivable autosomal trisomies are 21, 18 and 13, all small, gene-poor chromosomes. Autosomal monosomy is uniformly lethal early; 45,X is the sole viable monosomy in humans.

9. The maternal age effect

The strongest known risk factor for is maternal age, and the mechanism is a direct consequence of the female meiotic timetable.

Human oocytes enter in the fetus. They complete prophase I — synapsis, breaks, crossovers, chiasmata — before birth, and then arrest at diplotene (the dictyate stage). That arrest lasts from roughly the fifth month of gestation until the oocyte is ovulated: twelve years at the earliest, fifty at the latest.

The failure follows:

  1. is loaded onto the during the fetal S , and the meiosis-specific subunits (REC8, SMC1B) are not replenished thereafter.
  2. Over decades of arrest, cohesin is progressively lost. Oocytes from women in their forties show markedly reduced REC8 and SMC1B relative to women around twenty.
  3. Arm cohesion distal to the chiasmata weakens, so chiasmata slip toward the chromosome ends and off — the falls apart into univalents before it ever reaches the spindle.
  4. Centromeric cohesion weakens too, allowing premature separation of sister chromatids in meiosis I — which then look like meiosis II errors when the child is genotyped.
  5. The spindle assembly checkpoint in oocytes is comparatively permissive: a single misbehaving bivalent does not reliably arrest the division. The barrier that protects mitosis is weaker here.

This is a two-hit model, and it explains an otherwise odd correlation. Hit one is delivered before birth: a bivalent that got no crossover, or one placed too close to the or to hold well, is structurally fragile from the start. Hit two is age: decades of cohesin decay. Neither alone is usually enough. That is why maternal MI errors are associated with reduced recombination in the nondisjoined bivalent — an event that happened in the mother's own fetal life, cashed in forty years later.

The epidemiology follows the biology. Over 20% of oocytes are aneuploid even at peak fertility; the fraction exceeds 50% by the late thirties to early forties. Live-birth risk of Down syndrome runs at approximately 1 in 1,250–1,300 at maternal age 25, ~1 in 1,000 at 30, ~1 in 350–400 at 35, and ~1 in 100 at 40 — roughly linear to about 30 and rising sharply thereafter. (Treat these as approximate population figures; published tables differ in detail and in whether they correct for prenatal loss.)

Note the counterintuitive corollary: because most pregnancies occur at younger ages, most children with Down syndrome are born to mothers under 35. A high relative risk in a small stratum is not where most cases come from — the same base-rate structure that makes screening programmes hard to design everywhere.

10. Two germlines, two clocks, two error modes

SpermatogenesisOogenesis
When startsPuberty onward, continuouslyIn the fetus, all oocytes at once
Prophase I duration~2–3 weeksMonths to decades (arrested at dictyate)
Second arrestnoneMetaphase II, until fertilisation
Products per meiosis4 sperm1 egg + 2–3 polar bodies
Cytokinesissymmetricgrossly asymmetric — the egg keeps the cytoplasm
Cell divisions before a gamete~30 embryonic, then stem-cell divisions every ~16 days: roughly 150 by age 20, several hundred by age 40+~22 mitotic divisions, all prenatal, then the two meiotic divisions
Cells available~10⁸ sperm produced per day; full cycle ~70 days~6–7 million oogonia peak in fetus → ~1–2 million at birth → ~300,000 at puberty → ~400 ovulated
error modepoint — replication errors accumulate with division count — cohesion decays with arrest duration

The last row is the point of the table. The two fail in different ways because their clocks measure different things.

A sperm's genome has been copied hundreds of times, and each copy carries a small chance of a replication error that repair missed. So paternal age predicts de novo point mutations: about 80% of a child's de novo single-nucleotide variants are of paternal origin, and each additional year of paternal age adds roughly 1.3–1.5 of them (reference/verified-facts.md).

An oocyte's genome has been copied about 23 times and then sat still for decades. Its problem is not copying fidelity; it is that a physical structure had to be maintained, unreplenished, for a very long time. So maternal age predicts number errors.

Two organisms, two failure distributions, one mechanism each. It also disposes of a common framing: "egg quality declines with age" is not an explanation. The explanation is that is loaded once, before birth, and is never topped up.


Common misconceptions

What people thinkWhat's actually true
A is one DNA moleculeOnly in G1 and in a gamete. After S a chromosome is two molecules sharing one , and it is still one chromosome. Count centromeres
Sister chromatids and homologs are both "pairs", so much the same thingSisters are copies made in S phase and are identical; homologs come from different parents and differ at ~1 base in 1,000. Confusing them makes every later calculation wrong
halves the DNAIt halves the chromosome number, once, at anaphase I. DNA content goes 4c → 2c → 1c, a quartering over two divisions
happens between sister chromatidsChiasmata form between non-sister chromatids of . Sister-chromatid exchange does occur but is genetically silent — the sequences are identical
Meiosis II is just mitosisMechanically similar, but there is no preceding S phase, the cell is haploid, and the "sisters" being separated are no longer identical distal to a crossover
Crossing over exists to generate variationThat is a consequence. Its immediate mechanical job is to hold the together until anaphase I. A bivalent with no crossover segregates at random
applies to all genesOnly to loci on different chromosomes, or far enough apart on the same chromosome — well over 100 cM — that is indistinguishable from 0.5; at 50 cM the expected RF is still only ~0.32. Linked loci violate it, which is what makes mapping possible (Ch 14)
is a coin flip within each meiosisEach meiosis produces exactly two of each from a heterozygote. The sampling noise Mendel observed comes from which gametes fertilise, not from segregation
You are born with all your eggs, so eggs are finished productsYou are born with all your primary oocytes — cells frozen mid-meiosis, having already made their double-strand breaks and crossovers, waiting decades for anaphase I
is a rare accidentMore than 20% of human oocytes are aneuploid even at peak fertility, and is the leading identified cause of miscarriage. Accurate meiosis is the exception, not the rule, in our species
Older parents are riskier in the same wayDifferent mechanisms entirely: maternal age → aneuploidy (cohesion decay); paternal age → point (division count)

Worked example: diagnosing the meiotic origin of a trisomy

A child is born with 21. You have SNP for the child and both parents at three 21 markers: M1 proximal on the long arm (~15 Mb, GRCh38), M2 mid-arm (~30 Mb), M3 distal (~45 Mb). are labelled arbitrarily per marker.

M1 (proximal)M2 (mid)M3 (distal)
Mothera / bc / de / f
Fatherg / hi / jk / l
Child (trisomic)a / b / gc / d / ie / f / k

Step 1 — parent of origin. The child carries three alleles at each marker. At M1 those are a, b, g. Alleles a and b are maternal-only; g is paternal-only. So the child has two maternal and one paternal copy of chromosome 21. The extra chromosome came from the mother. (The same reading at M2 and M3 confirms it.)

Step 2 — pick the informative marker. Only markers close to the report on the meiotic division, because crossovers in prophase I exchange distal segments between homologs and destroy the correspondence. M1 is the proximal marker; use it. M2 and M3 are used later, for a different purpose.

Step 3 — read the mechanism off M1. The mother is a/b. The child inherited both a and b — two different maternal alleles at a pericentromeric locus.

The observation is a/b. Maternal I nondisjunction.

Step 4 — sanity-check against the prior. Roughly 90% of trisomy 21 is maternal, and about 77% of maternal errors are MI. The prior probability of maternal MI is therefore ≈ 0.90 × 0.77 ≈ 0.69. But the prior is barely doing any work here: a/b at a pericentromeric marker is essentially impossible under a clean MII error, so the likelihood ratio dominates and the posterior is close to 1. The prior would only matter if the proximal marker were uninformative (mother homozygous), in which case you would fall back on it — and report a probability rather than a diagnosis.

Step 5 — use the distal markers for something else. At M2 and M3 the child is also heterozygous for the maternal alleles (c/d, e/f). Full maternal heterozygosity along the whole arm means no detectable exchange between the two nondisjoined homologs at these positions — consistent with the reduced-recombination signature associated with maternal MI errors, and with the achiasmate- model in §5. With denser markers you would count exchange points directly and compare against the expected crossover count for chromosome 21.

Step 6 — state what you have not shown. You have identified the division at which the error became visible, not its cause. Premature separation of sister chromatids in meiosis I produces a genotype indistinguishable from a true MII error. The classification is of the observable pattern.

Connections