Every flat world map distorts something, because a sphere cannot be flattened without stretching or tearing, a fact Carl Friedrich Gauss proved in 1827. Projections differ only in what they choose to sacrifice. The Mercator projection of 1569 keeps angles and local shapes true but inflates areas towards the poles, so Greenland, at about 2.17 million km², looks as large as Africa, which at about 30.3 million km² is 14 times bigger. Equal-area maps such as the Gall–Peters keep sizes correct but squash and stretch shapes. Compromise projections such as the Robinson and the Winkel tripel accept small errors in everything to avoid large errors in anything, and the AuthaGraph folds the globe into 96 triangles to keep sizes nearly right. Below, the five are compared one by one and then in a single table.
Gauss’s Theorema Egregium: Why No Flat Map Can Be Perfect
The root of every distortion is a piece of geometry. In 1827 Gauss proved his Theorema Egregium, Latin for "remarkable theorem": the curvature of a surface is an intrinsic property that cannot change if the surface is bent without stretching. A sphere has constant positive curvature; a sheet of paper has none. So there is no way to lay a globe, or even a small patch of it, onto a flat sheet while keeping every distance correct. Step on an empty eggshell and it must crack before it lies flat.
The consequence for cartographers is a menu of trade-offs. A map can be conformal, keeping angles and the shapes of small areas correct; equal-area, keeping relative sizes correct; equidistant, keeping true distances from one or two points; or a compromise that is none of these exactly. No projection can be both conformal and equal-area. Choosing a map is choosing which lie you can live with.
Tissot’s Indicatrix: Circles That Measure the Damage
To compare those lies, the French mathematician Nicolas Auguste Tissot introduced, in 1859 and 1871, what is now called Tissot's indicatrix. Imagine drawing identical tiny circles all over a globe, then projecting the globe onto a map. Tissot proved that each circle becomes an ellipse whose long and short axes show the directions of greatest and least stretching at that point.
The pattern of ellipses works as a fingerprint. On a conformal map every ellipse stays a perfect circle, but the circles grow or shrink from place to place. On an equal-area map every ellipse keeps the same area, but they flatten into ovals of differing shape. On a compromise map both size and shape vary, but neither runs to extremes. The hero image of this article shows the indicatrix on a Mercator map: identical circles near the equator, swelling into huge discs at high latitudes.
Greenland and Africa on the Mercator Map
Gerardus Mercator, a Flemish mapmaker, published his projection in 1569 on a world map 202 by 124 cm printed on 18 sheets, titled in Latin as a new description of the Earth "corrected for the use of sailors". Its purpose was navigation. On a Mercator chart any course of constant compass bearing, a rhumb line, is a straight line, so a navigator can rule a line from port to port and read off the heading. Mercator never explained how he built it. The English mathematician Edward Wright published the first accurate construction tables in 1599 and, in more detail, in 1610 in Certaine Errors in Navigation. Once the marine chronometer made longitude measurable in the mid-18th century, the projection became the standard nautical chart, and practically every printed marine chart still uses it.
The cost is size. To keep angles true, Mercator stretches each parallel of latitude by a factor called the secant of the latitude, and stretches distances north–south by the same amount. Area is therefore inflated by the square of that factor. At 30° the inflation is only about a third; at 60°, the latitude of Oslo or Anchorage, areas are drawn four times too large; around 73°, a fair middle latitude for Greenland, the factor is about 11.7. At the poles it becomes infinite, which is why Mercator maps simply stop short of them and Antarctica, when shown, smears across the whole bottom edge.
This is how Greenland comes to rival Africa. In reality Greenland covers about 2,166,000 km², similar to the Democratic Republic of the Congo at 2,345,000 km², while Africa covers about 30.3 million km². Other Mercator illusions follow the same rule. Alaska, at 1,717,856 km², looks as large as Australia, which at 7,688,287 km² is about 4.5 times bigger, and as large as Brazil, nearly five times bigger. Madagascar and Great Britain look similar, yet Madagascar's 592,800 km² is more than double Great Britain's 209,331 km².
Mercator never vanished. Publishers reduced its use for world maps through the 20th century, but web mapping revived it: Google Maps, Bing Maps, OpenStreetMap and others use a variant called Web Mercator, cut off at latitude 85.05113° so the world fits in one square tile. At street level, where conformality keeps corners square, it works well; zoomed out to the whole planet it reproduces the old distortion. Turning the cylinder on its side gives the transverse Mercator, which keeps scale nearly constant along a chosen meridian; in ellipsoidal form it underpins the Universal Transverse Mercator grid and most national grid systems, proof that Mercator’s mathematics is excellent for narrow regions even when it misleads at world scale.
Gall–Peters (1855, 1973): Correct Sizes, Stretched Shapes
The Gall–Peters projection sits at the opposite corner of the menu. It is a cylindrical equal-area projection: every country appears at its correct relative size. The Scottish clergyman James Gall described it at a science convention in 1855 and published it in 1885. It attracted little notice until the German historian Arno Peters reintroduced it in 1973 as the Peters World Map, arguing that Mercator made wealthy Europe and North America look too large beside Africa and South America.
The price is shape. Gall–Peters is undistorted only along 45° North and South. Near the equator its formulas draw north–south distances about twice as long, relative to east–west ones, as they should be, so Africa and South America look drawn out vertically, while high latitudes are squashed flat. Cartographers objected to Peters' claim that his was the only area-correct map, since hundreds of equal-area projections already existed, and in 1989 and 1990 seven North American geographic organisations adopted a resolution rejecting all rectangular world maps, Mercator and Gall–Peters alike. The projection still won institutional support: Boston Public Schools began phasing it in from March 2017, and a Nebraska law signed on 16 April 2024 requires public schools to use Gall–Peters, a similar cylindrical equal-area map, or the AuthaGraph.
Robinson (1963) and Winkel Tripel (1921): National Geographic’s Compromises
Most commercial atlases now pick a compromise instead of Mercator. Arthur H. Robinson devised his projection in 1963 at the request of Rand McNally, working, as he put it, backwards from the shapes that looked best to the mathematics. It is neither conformal nor equal-area. Its meridians curve gently and its poles are drawn as lines rather than points, so distortion is severe near the poles but moderate across the inhabited latitudes. Robinson published the details of its construction in 1974, and the National Geographic Society adopted it for world maps in 1988, replacing the Van der Grinten projection.
Ten years later, in 1998, National Geographic switched to the Winkel tripel, proposed by the German cartographer Oswald Winkel in 1921. It is the average of two older projections, the equirectangular and the Aitoff, and its name, German for "triple", refers to Winkel's aim of minimising three distortions together: area, direction and distance. In a comparison by the physicists David Goldberg and J. Richard Gott, the Winkel tripel produced low distance, ellipticity and area errors and the least skew of the projections they studied.
AuthaGraph (1999): Folding the Globe into 96 Triangles
The Japanese architect Hajime Narukawa took a different route in 1999. His AuthaGraph projection divides the sphere into 96 equal triangles, transfers them onto a tetrahedron while keeping their area proportions, and unfolds the tetrahedron into a rectangle. The result shows all continents and oceans at close to their true sizes, includes Antarctica whole, and can be tiled seamlessly in any direction. Strictly it is not equal-area, because area is controlled only for the 96 regions, not within them. Tokyo's Miraikan science museum adopted it as its official map in 2011, and it won Japan's Good Design Grand Award in October 2016. Its drawback is unfamiliarity: the continents sit at unexpected angles and north is no longer straight up everywhere.
Comparing the Five Projections Side by Side
| Projection | Year | Preserves | Main distortion |
|---|---|---|---|
| Mercator | 1569 | Angles, local shapes, straight rhumb lines | Area grows towards the poles (4× at 60°) |
| Gall–Peters | 1855 (Peters 1973) | Relative area | Shapes stretched near the equator, flattened near the poles |
| Robinson | 1963 | Nothing exactly; balanced look | Poles drawn as lines; high-latitude shapes |
| Winkel tripel | 1921 | Low combined area, direction and distance error | Moderate everywhere, curved meridians |
| AuthaGraph | 1999 | Near-correct area, including Antarctica | Unfamiliar orientation; area only controlled per region |
The right map depends on the job. Navigators still want Mercator; anyone comparing country sizes wants an equal-area map; a general wall map is usually best on a compromise such as the Winkel tripel. To see the Greenland illusion settled with real figures, read whether Greenland is bigger than the DR Congo, learn why your phone uses Web Mercator in Google Maps, or try a projection built around distance in our guide to the azimuthal equidistant projection.
