Two homologous chromosome pairs (each already duplicated into sister chromatids) are shown going through both meiotic divisions. Meiosis I separates the two homologs of each pair to opposite poles — one maternal, one paternal copy per daughter cell. Meiosis II then separates the sister chromatids of each remaining chromosome, exactly like mitosis, giving four haploid gametes.
Nondisjunction is a failure of that separation: instead of splitting 1-vs-1, both copies migrate to the same pole. It can happen at either division, with different consequences:
MI nondisjunction (prob p1):
→ one daughter cell gets BOTH homologs, the other gets NEITHER
→ after a normal MII split, ALL 4 resulting gametes are aneuploid
(two get n+1, two get n-1) for that chromosome
MII nondisjunction (prob p2), given MI was normal:
→ one daughter cell's sister chromatids fail to separate
→ only that cell's 2 gametes are affected (one n+1, one n-1)
→ the other daughter cell's 2 gametes stay normal (n)
For a single chromosome, the probability that any one resulting gamete ends up with the wrong copy number works out to:
P(aneuploid for this chromosome) = p1 + (1 − p1)·p2
With two independent chromosome pairs tracked here, a gamete's total chromosome count can differ from the expected n=2 through either pair — that's the count the "chromosomes per gamete" readout reports directly from the simulated outcome, no shortcuts. Real aneuploidies produced this way include trisomy 21 (Down syndrome) and Turner/Klinefelter syndromes for the sex chromosomes, and the risk rises sharply with maternal age because human oocytes can sit arrested in meiosis I for decades, giving cohesion between sister chromatids far longer to degrade.
Chromosome shape (metacentric / submetacentric / acrocentric) reflects where the centromere sits relative to the two arms — a real, fixed feature used to identify each human chromosome on a karyotype; it plays no role in whether nondisjunction happens, only in what the chromosome looks like here.