The reason cytogenetics works at all is that the genome is not uniform: it is organized into long stretches that differ systematically in base composition, gene density, and replication timing, and these differences are what a stain makes visible. G-banding exploits this. After mild trypsin digestion partially denatures chromosomal proteins, Giemsa preferentially stains AT-rich, gene-poor, late-replicating chromatin — the G-dark bands — while GC-rich, gene-dense, early-replicating segments stain lightly. The alternating pattern is therefore not an arbitrary dye artifact but a reproducible map of the underlying genome architecture, which is why the same band pattern recurs on the same chromosome in every cell and every person.

This is also why band position carries clinical meaning. A breakpoint or deletion falling in a gene-dense G-light band is far more likely to disrupt a dosage-sensitive gene (or several, producing a contiguous-gene syndrome) than the same-sized event in a gene-poor dark band. The cytogeneticist reads a karyotype not just by counting chromosomes but by checking that the expected light/dark sequence is intact — a missing, added, or relocated band signals a structural rearrangement.

The practical limit of the method is resolution. Banding resolution is quoted as the number of bands visible across the haploid set: routine clinical preparations resolve roughly 400–550 bands, and prometaphase (high-resolution) banding pushes this to 550–850 bands by capturing less-condensed chromosomes. Even at the high end, each band represents several megabases, so karyotyping reliably detects only rearrangements ≥5–10 Mb. Anything smaller — the great majority of clinically relevant microdeletions and microduplications — is invisible to the microscope and is precisely the gap that chromosomal microarray was developed to fill.

Centromere position is the other identifying feature. It divides each chromosome into a short arm (p, petit) and long arm (q), and its location defines morphology: metacentric (central), submetacentric (off-center), and acrocentric (near the tip). The five acrocentrics — 13, 14, 15, 21, 22 — have tiny p arms composed largely of redundant ribosomal-DNA repeats, which is why their loss in Robertsonian translocations is tolerated.

The point of the International System for Human Cytogenomic Nomenclature (ISCN), now in its 2024 edition, is to make a chromosomal finding unambiguous and machine-comparable: any two laboratories describing the same result should produce the same string, and that string should be parseable back into a precise genomic statement. ISCN achieves this with a strict, ordered grammar rather than free text. A karyotype is read left to right as [total chromosome number],[sex chromosomes],[abnormalities] — the commas are syntactic, separating fixed fields, so 46,XY and 47,XY,+21 differ in exactly the way the underlying biology differs.

The band-naming scheme is the part trainees most often misread, and the logic is worth internalizing rather than memorizing. An address such as 7q11.23 is built from chromosome (7) + arm (q) + region (1) + band (1) + sub-band (.2), and the final 3 marks a further subdivision of that sub-band. The numbering is anchored at the centromere and counts outward toward the telomere: region 1, band 1 lies immediately beside the centromere, and the digits grow as you move distally. The decimal digits are not a number to be read as 'eleven point twenty-three' — they are an ordered hierarchy ('region one, band one, sub-band two, subdivision three'), added as higher-resolution banding split a coarse band into finer ones. This is why higher-resolution studies append digits rather than renumbering everything.

Understanding the anchoring matters clinically because it lets you reason about a breakpoint's position from the name alone. 7q11.23, for example, sits in the proximal long arm and is the Williams-Beuren critical region; a reader who knows the numbering convention can infer that a band labeled q36 lies far more distally, near the telomere, without consulting an ideogram.

The deep distinction ISCN encodes is between abnormalities that change how many chromosomes there are and those that change how the material is arranged, because these two classes arise by different mechanisms and carry different recurrence risks. Numerical abnormalities — aneuploidies — almost always originate from non-disjunction, a failure of paired homologs (meiosis I) or sister chromatids (meiosis II) to separate, leaving one gamete with an extra chromosome and one short. This is why the count departs from 46: 47,XY,+21 (trisomy 21, Down syndrome) gains a whole chromosome and is written with a +; a missing autosome takes a −. The sign sits before the chromosome number because the event is a gain or loss of an entire chromosome, not a piece of one. Constitutional sex chromosome aneuploidy is the exception: it is written by listing the sex chromosomes present, so Turner syndrome is 45,X and Klinefelter syndrome 47,XXY, without a + or −.

Structural abnormalities, by contrast, result from chromosome breakage and rejoining, and here the total count can stay at 46 even though material has moved. ISCN distinguishes them by symbol and by what is conserved: a deletion (`del`) removes a segment; a duplication (`dup`) adds one; an inversion (`inv`) flips a segment end-for-end within a chromosome; a translocation (`t`) exchanges segments between chromosomes; a ring (`r`) forms when the two ends of a chromosome join, usually after breaks in both arms; some rings, such as ring 20, lose no detectable material. The parentheses after the symbol carry the breakpoint addresses: two breakpoints, as in del(7)(q11.23q11.23), name an interstitial loss between them, and a single breakpoint, as in del(5)(p15.2), names a terminal loss from that band to the end of the arm.

The most consequential clinical idea here is balanced vs unbalanced. A balanced reciprocal translocation such as t(11;22)(q23;q11.2), the most common recurrent constitutional reciprocal translocation, moves material without net gain or loss, so the carrier is usually phenotypically normal — yet at meiosis the rearranged chromosomes segregate unpredictably and can deliver an unbalanced complement to offspring, which is why an apparently healthy parent can have recurrent miscarriages or an affected child. A `der` (derivative) chromosome names the structurally abnormal product and is named for the chromosome whose centromere it keeps; a child with Emanuel syndrome, for example, is 47,XY,+der(22)t(11;22)(q23;q11.2)mat. Because a microarray cannot see a balanced rearrangement, the parents of a child with an unbalanced one are tested by karyotype or FISH. The lesson the notation enforces is that a normal chromosome number does not guarantee a normal chromosome content.

Mosaicism is best understood through its timing: it is a chromosomal change that happens after fertilization, in a single already-formed zygote, usually from a mitotic error or from the loss of an extra chromosome in a trisomic conceptus (trisomy rescue), so two related cell lines coexist. The earlier in development the error occurs, the more tissues inherit the abnormal line and the more severe the phenotype tends to be — which is why mosaic severity correlates with the proportion and distribution of abnormal cells, not merely their presence. Crucially, the fraction seen in a blood karyotype need not match the brain, gonad, or skin, so a low-level or absent blood result does not exclude clinically significant mosaicism elsewhere; this is a common source of genotype-phenotype discordance.

ISCN records this with the slash-and-bracket convention: each cell line is written in full and followed by the number of metaphases counted, separated by `/`. Thus 45,X[12]/46,XX[18] states that of 30 cells scored, 12 were monosomy X and 18 normal female — mosaic Turner syndrome. The bracketed counts are not decoration; they let a reader judge whether a minor line is real or noise, which is why laboratory standards (not ISCN, which governs only the notation) call for counting about 20 metaphases in a routine constitutional study and more when mosaicism is suspected. Even so, twenty normal cells exclude an abnormal line only above about 14% with 95% confidence, thirty above about 10% Hook 1977. In a constitutional karyotype the normal diploid line is written last.

A conceptually distinct entity that produces a similar-looking result is chimerism: two genetically different zygotes fusing (or one twin absorbing another), yielding two cell lines that were never related — the classic example being 46,XX/46,XY. The notation looks like mosaicism but the mechanism (two fertilization events) and the counseling implications differ entirely.

Two special structures round out the vocabulary. An isodicentric chromosome (`idic`) carries two centromeres derived from a single chromosome, typically from a mirror-image fusion; marker chromosomes (`mar`) are small, morphologically uninterpretable fragments whose origin cannot be read by banding at all. Because a marker's clinical impact depends entirely on which chromosome and which genes it contains, it is one of the clearest cases where karyotyping must hand off to FISH or microarray for identification.

Microarray fundamentally changed the resolution of cytogenetics, and the notation had to change with it. Where banding measures position in megabase-wide bands, chromosomal microarray (CMA) measures copy number at the resolution of its probes, often tens of kilobases, by comparing a patient's DNA against a reference across the genome. The clinical payoff is large: CMA detects submicroscopic deletions and duplications that karyotyping cannot, raising the diagnostic yield in unexplained developmental disability, intellectual disability, autism, or multiple congenital anomalies to roughly 15–20%, versus about 3% for G-banded karyotype after excluding recognizable syndromes — the evidence that established CMA as the first-tier test for these indications Miller et al. 2010.

Because the result is now a genomic interval rather than a band, ISCN array notation is built around coordinates tied to a genome build. Since ISCN 2016 the format is `arr[build] band(start_end)xcopies`, with no commas in the coordinates and an underscore between start and end; older reports wrote, for example, arr[hg19] with commas and a hyphen. Reading arr[GRCh38] 22q11.21(18924718_21111383)x1: `arr` flags an array result, `[GRCh38]` fixes the coordinate system (a coordinate is meaningless without its build, since builds renumber the genome), the band locates it cytogenetically, the parenthetical gives the first and last probes showing the change (the true breakpoints lie beyond them, so the size is a minimum), and `x1` is the copy state. The interpretive key is that copy number is read against the expected diploid 2: x1 is a deletion, x3 a single-copy duplication, and the event's size is end minus start, about 2.19 Mb here.

One category needs the diploid framing made explicit because it has no copy change at all. Copy-neutral loss of heterozygosity — a long run of homozygosity, written `x2 hmz` — keeps two copies but they are identical because both descend from one ancestral chromosome: from a single parent (uniparental isodisomy) or from an ancestor the two parents share (parental relatedness). It is clinically important precisely because the count is normal: it can unmask a recessive disease allele in the homozygous segment or, over an imprinted region, signal a uniparental-disomy imprinting disorder. CMA thus reports both what is gained or lost and where heterozygosity has disappeared.

Finally, calling a CNV is not the same as classifying it. Whether a given gain or loss is pathogenic, benign, or of uncertain significance is a separate, evidence-weighted judgment — gene content, dosage sensitivity, inheritance, and overlap with known syndromic regions — governed by professional standards for CNV interpretation and reporting Kearney et al. 2011.