Thematic Maps Explained: Choropleth, Dot, Isoline, Flow and Cartogram
Source: Wikimedia Commons
Map Projections & Cartography

Thematic Maps Explained: Choropleth, Dot, Isoline, Flow and Cartogram

Thematic maps show one subject across space. Here is how the main types compare, with the maps by Dupin, Snow, Halley and Minard that made each famous.

Geography Worlds
March 19, 2026
Updated October 6, 2026
7 min read

A thematic map is a map that shows the geographic pattern of one subject, such as rainfall, income, language or disease, rather than the general layout of roads, rivers and towns that a reference map shows. There are several main types, each suited to different data: choropleth maps shade whole districts by a rate (the first, by Charles Dupin in 1826, shaded French departments by schooling); dot maps place one dot per case or per fixed number of people; proportional symbol maps scale a circle to a total; isarithmic maps join points of equal value with lines, as Edmond Halley did for compass variation in 1701; flow maps draw movement as bands, most famously Charles Joseph Minard’s 1869 map of Napoleon’s march on Moscow; and cartograms resize the regions themselves. Mapping raw totals where rates are needed, or choosing a type that does not suit the data, are among the commonest ways a thematic map misleads.

The best way to understand the types is to set them against each other, in pairs that are easily confused, using the maps that first made each one famous.

Thematic Map Against Reference Map: Hondius in 1607 and the Atlas Base Map

A general reference map answers “where is it?”: it locates a varied set of physical features, from coastlines to railways. A thematic map answers “what is it like here?”, using symbols to show properties that cannot be seen on the ground, such as temperature, language or population. In practice the two depend on each other, because every thematic map is drawn on top of a reference base, and it could only become common once base maps were accurate enough. According to the American cartographer Arthur Robinson, almost all modern thematic techniques were invented between 1700 and 1850, when governments began to collect and publish statistics.

One of the earliest thematic maps is the Designatio orbis christiani of 1607 by the Dutch mapmaker Jodocus Hondius, published in the French edition of his Atlas Minor, which used symbols to show where the world’s major religions were practised. In England the astronomer Edmond Halley (1656–1742) made the map into a tool for scientific thinking: his 1686 map of the trade winds is called the first meteorological chart. Several older names for the class, such as special-purpose, statistical or distribution maps, have dropped out of use. Our overview of the types of maps places thematic maps alongside topographic, political and navigational ones.

Choropleth Against Dot Map: Dupin’s Departments, Montizon’s Dots and Snow’s Soho

The choropleth: fast, familiar and easy to misread

The choropleth, from the Greek for “area” and “multitude”, colours each existing district according to a value aggregated within it. Its inventor was Baron Pierre Charles Dupin, whose 1826 map shaded the departments of France by the availability of basic education. France soon produced many more “cartes teintées” on crime, disease and living conditions, and the 1841 Census of Ireland published choropleths in its official reports. The geographer John Kirtland Wright gave the type its name in 1938. It is probably the most common thematic map today, because statistics already come packaged by country, state, county or census tract.

That convenience is also its weakness. A single colour makes each district look uniform, hiding the mix of rich and poor neighbourhoods inside a city, and reading a district average as true of every person in it is the ecological fallacy. Results also change with how boundaries are drawn, a problem statisticians call the modifiable areal unit problem. Finally, large districts dominate the eye whatever their population, as Montana and New Jersey show.

Montana, area380,800 km²
New Jersey, land19,050 km²
Montana, people1,084,225
New Jersey, people9,288,994
Area (New Jersey: land area) and 2020 census population of Montana and New Jersey; each pair of bars is scaled to its own largest value. On a state choropleth Montana gets 20 times the ink for under one-eighth of the people. Source: US Census Bureau 2020, via Wikipedia

Montana covers 380,800 km² (147,040 sq mi) and had 1,084,225 people at the 2020 census; New Jersey has 19,050 km² (7,354 sq mi) of land and had 9,288,994. Shade both by an election result and the eye reads Montana as twenty times more important, when New Jersey has about eight and a half times as many people. Cartographers reduce the problem by mapping rates rather than raw counts, and by using an equal-area projection, since a projection like Mercator also inflates high-latitude districts.

The dot map: showing where things actually are

A dot map avoids the uniform-district problem by placing one point per case, or one per fixed number of individuals, so clusters and gaps show up inside districts. The earliest known district-based dot density map was made in 1830 by Armand Joseph Frère de Montizon, a Franciscan friar, teacher and printer, who mapped the population of France by department at 10,000 people per dot. It had almost no influence until the technique was reinvented in 1859 by Thure Alexander von Mentzer, a Swedish army officer, for Sweden and Norway at 200 residents per dot. An earlier map has also been claimed: Valentine Seaman’s 1797 plot of yellow fever cases in New York City.

The most famous dot map is John Snow’s. After the 1854 cholera outbreak in Soho, London, Snow marked each death on an accurate street plan that also showed the public water pumps, and published it in 1855. The deaths clustered around one pump on Broad Street: 127 people on or near the street died in the first three days, and Snow found only ten deaths in houses clearly nearer another pump. At his urging the parish removed the pump handle on 8 September, although the outbreak may already have been in rapid decline. A choropleth of London parishes would have hidden that pattern completely.

Black and white street map of Soho, London, with small black bars stacked at addresses where cholera deaths occurred, clustered around Broad Street
John Snow’s map of the 1854 cholera deaths in Soho, London, published in 1855: each bar is a death, and they cluster around the Broad Street pump. Photo: John Snow, Public domain, via Wikimedia Commons

Isolines Against Proportional Symbols: Halley’s 1701 Chart and Harness’s Ireland

These two handle opposite kinds of data. Isarithmic maps suit a continuous field that has a value everywhere, such as temperature, pressure or height. They join points of equal value with lines: isotherms for temperature, isobars for pressure, contours for elevation. The oldest known example is an isobath, a line of equal depth, on a 1584 map of the river Spaarne near Haarlem by the Dutchman Pieter Bruinsz. Halley’s 1701 “New and Correct Chart Shewing the Variations of the Compass” drew lines of equal magnetic variation across the Atlantic and was the first chart to show such lines for magnetism, which sailors needed to correct their compass readings. Alexander von Humboldt drew isotherms in 1817 and coined the term, and in 1874 Louis-Léger Vauthier mapped the population density of Paris with isolines.

Engraved 18th-century chart of the Atlantic Ocean with the Americas on the left, Europe and Africa on the right, crossed by curving lines of equal compass variation
Edmond Halley’s 1701 chart of compass variation across the Atlantic, whose curved lines join places of equal magnetic variation. Photo: Edmond Halley (published by W. and J. Mount and T. Page, London), Public domain, via Wikimedia Commons

Proportional symbol maps suit totals tied to points, such as the population of each city. A symbol, usually a circle, is drawn at each place with its area, not its diameter, proportional to the value; doubling a circle’s diameter quadruples its area and badly exaggerates the difference. Possibly the first proportional point symbols appeared in 1837, in a booklet of five maps by Henry Drury Harness for a government report on building railways in Ireland. The same booklet included early flow and chorochromatic maps and possibly the first dasymetric map, which redistributes district totals into the parts of each district where people actually live.

The rule of thumb follows from the data. Use isolines when the thing being mapped varies smoothly through space; use proportional symbols when it is counted at places. Drawing isolines through city totals invents values between cities that do not exist. Every type also depends on a clear legend, which our guide to map legends and keys explains.

Flow Map Against Cartogram: Minard’s 1869 March and Levasseur’s 1876 Squares

Both encode quantity as size, but they resize different things. A flow map scales the width of a line to show how much moves along a route. Charles Joseph Minard, a retired French civil engineer, developed it in maps of goods and people on transport networks. On 20 November 1869 he dated his Carte figurative of the French army’s losses in the Russian campaign of 1812–1813. The band is one millimetre wide for every 10,000 men, coloured for the men marching into Russia and black for those coming out, and a temperature graph runs underneath. It combines six kinds of data in two dimensions: troop numbers, distance, temperature, latitude and longitude, direction and dates. The information designer Edward Tufte wrote that it “may well be the best statistical graphic ever drawn”.

Minard's flow map of Napoleon's 1812 Russian campaign, a broad tan band narrowing towards Moscow and a thin black band returning, with a temperature line chart below
Charles Joseph Minard’s 1869 flow map of the French army’s advance on Moscow and retreat in 1812, the band shrinking as men were lost. Photo: Charles Joseph Minard, Public domain, via Wikimedia Commons

A cartogram scales space itself, resizing each region in proportion to a variable such as population. The earliest known were published in 1876 by the French statistician Pierre Émile Levasseur, who drew the countries of Europe as squares sized by area, population, religious adherents and national budget, as teaching aids. The American cartographer Erwin Raisz later produced rectangular statistical cartograms, and in 1963 Waldo Tobler wrote one of the first computer algorithms for them. Daniel Dorling’s 1996 method replaces each region with a circle of proportional size. A population cartogram fixes the Montana problem at a stroke, but readers must recognise the distorted shapes, so it works best for familiar regions.

Colour decides how well any of these maps reads. Up to 1 in 12 men and about 1 in 200 women have red–green colour blindness, so a red-to-green scale can be unreadable for a large audience. ColorBrewer, launched in 2002 by Cynthia Brewer, Mark Harrower and Pennsylvania State University, offers sequential, diverging and qualitative palettes with colour-blind-safe options. Classification matters as much: the same data split into equal intervals or into quantiles can tell different stories. For real cases where colour, classes and projection were used to mislead, see our collection of misleading maps.

Recently updated guides