How many classes

The eye's limit

Choropleth maps depend on a reader's ability to glance at a patch of colour and assign it to a class. Testing bears this out consistently: human vision reliably separates somewhere around five to seven distinct hues or tonal steps in a sequential scheme before adjacent shades begin to blur together. Push past seven, and the reader is no longer reading a map — they are hunting a legend, matching swatches like paint chips.

This is not a personal failing. The constraint is cognitive, rooted in working memory's limited channel for simultaneous distinctions. George Miller's 1956 paper on the magic number seven formalised what cartographers had already learned empirically: the mind holds roughly seven chunks at once, and colour classes are chunks. A map that demands ten discriminations is a map that will be read incorrectly, however accurate the underlying data.

Three or four classes often do more work than seven. A sequential scheme in three steps — low, medium, high — broadcasts its message at a single glance. The cost is compression: real variation inside each class disappears, and the breaks you choose determine what story survives. Four or five classes usually strike the practical balance for a general audience: enough resolution to show spatial pattern, few enough tones to read without effort.

Diverging schemes — where the palette fans out from a neutral midpoint in two directions — impose a related constraint. Each side of the divide has its own hue and its own lightness ramp. Readers must distinguish within each arm as well as between them. Effective diverging maps rarely exceed five total classes (two on each side plus the midpoint), and three-class diverging maps are often the clearest of all.

Plate 2

A surveyor and a total station. The numbers a map is drawn from begin here, not on the screen.

Photo: Ferhat Kocakaya / Pexels

Qualitative schemes, where hue rather than lightness encodes category membership, can tolerate a handful more — up to eight or nine if the categories are conceptually distinct and the hues are well separated in perceptual colour space. But qualitative classification is a different operation; there is no ordered sequence for the eye to follow, only identity. Push it to twelve categories and the legend becomes the map.

The practical rule: start with five, justify every class you add above that, and treat seven as a ceiling rather than a target. Below the floor — fewer than three classes — the map reveals almost nothing the raw data could not show in a table. The goal is the narrowest range that captures the pattern without hiding it.