The Winkel tripel projection is a world map projection published in 1921 by the German cartographer Oswald Winkel, built by taking the exact average of two older maps: the equirectangular grid of latitude and longitude and the oval Aitoff projection. The word tripel is German for "triple", and it names Winkel's aim of keeping three kinds of distortion (area, direction and distance) all low at once rather than eliminating any one of them. The result is a gently rounded world with curved meridians and curved parallels, poles drawn as short straight lines, and Greenland and Antarctica only moderately inflated. Since 1998 it has been the standard projection for the National Geographic Society's world maps, and in a 2007 study by David M. Goldberg and J. Richard Gott it scored best on several separate measures of distortion among the world maps they tested.
Marinus of Tyre’s Grid and David Aitoff’s Ellipse: The Two Parents
Nothing in the Winkel tripel is invented from scratch; it is built entirely from two older projections. The first parent is the equirectangular projection, the plainest map there is. Every meridian becomes a vertical straight line and every parallel a horizontal straight line, evenly spaced, so a point's x coordinate is simply its longitude and its y coordinate its latitude. Ptolemy credited the idea to Marinus of Tyre around AD 100, and it has never gone away: the version called plate carrée is the standard format for global raster datasets such as Natural Earth and NASA World Wind, because each pixel maps directly to a latitude and longitude. Its weakness is shape. Near the poles, east-west distances are stretched enormously, so Scandinavia and Alaska look squashed flat.
The second parent is the Aitoff projection, proposed by David A. Aitoff in 1889. Aitoff took the equatorial form of the azimuthal equidistant projection, a disc that shows true distances from its centre, and applied a trick: he halved every longitude so the whole world fitted into one hemisphere, projected it, and then stretched the result sideways into a 2:1 ellipse to restore the missing width. The outcome has curved parallels and curved meridians, with the poles as points at the top and bottom of the oval. Three years later, in 1892, Ernst Hammer used the same trick with the Lambert azimuthal equal-area projection instead and produced the Hammer projection, which is why the two ovals look so alike.
Oswald Winkel’s 1921 Averaging Trick
Oswald Winkel (1874–1953) noticed that the two parents fail in opposite ways. The equirectangular map keeps parallels straight and evenly spaced but fattens the polar regions sideways; the Aitoff keeps the polar regions narrow but bends everything into a tight curve. His method was to compute both maps for the same point and plot the midpoint between them: the x coordinate on the Winkel map is half the Aitoff x plus half the equirectangular x, and the same for y. In 1921 he published three variants built this way. Winkel I averages the equirectangular with the sinusoidal projection, Winkel II averages it with the Mollweide, and Winkel III, the tripel, averages it with the Aitoff. The first two are pseudocylindrical, with straight parallels; the tripel inherits curved parallels from Aitoff and is classed as a modified azimuthal projection. Of the three, only the tripel became widely known.
One detail in the recipe is a free choice. An equirectangular map can be drawn true to scale along any chosen pair of parallels, and Winkel fixed that standard parallel at the angle whose cosine is 2/π, which works out to about 50.46° north and south. That choice sets the proportions of the finished map. Working through the formulas, the outline comes out about 1.64 times as wide as it is tall, and each pole becomes a straight line roughly two-fifths the length of the equator, a compromise between the Aitoff's pointed poles and the equirectangular's full-width poles. There is also a practical quirk: no closed-form inverse exists, so converting a point on a Winkel map back into latitude and longitude requires iterative numerical methods. For printing a wall map that hardly matters, but it makes the projection more awkward to use in software that must turn a click on the map back into coordinates.
Area, Direction, Distance: How Winkel Tripel Treats Greenland and Antarctica
No flat map can keep every property of a globe, and the Winkel tripel makes no attempt to keep any one of them perfectly. It is not conformal, so small shapes are slightly skewed, and it is not equal-area, so high-latitude land still looks bigger than it is. What it does is keep each error modest. Measured along the central meridian, the area scale grows only slowly as you move away from the equator: about 17 per cent larger at 40° of latitude, about 52 per cent larger at 60°, and roughly three times larger at 80°. On the Mercator projection of 1569, by comparison, area inflates by a factor of four at 60° and more than thirty at 80°, which is how Greenland can appear as large as Africa even though its land area is about 7 per cent of Africa's.
The polar regions show the trade-off most clearly. Because the poles are lines rather than points, Antarctica is spread along the bottom edge of the map and its coastline is stretched sideways, while Arctic islands such as Svalbard are widened along the top. Towards the left and right edges, where the meridians curve most sharply, land is sheared at an angle. Near the centre, from the Mediterranean down through Africa, shapes are close to what a globe shows. The cartographer's tool for seeing this pattern is Tissot's indicatrix, the ellipse of distortion described by Nicolas Auguste Tissot in 1859 and 1871: on a Winkel tripel map the ellipses stay close to circles across the low and middle latitudes and grow and flatten only in the far north and south.
Washington, 1998: National Geographic Drops Robinson
The projection spent most of the 20th century as a textbook curiosity. Its rise came from one institution: the National Geographic Society, founded in 1888 and headquartered in Washington, D.C., which has used just three standard world projections since 1922. In 1922 it adopted the Van der Grinten projection, invented by Alphons J. van der Grinten in 1898 and patented in the United States in 1904, which draws the whole Earth inside a circle and keeps much of the Mercator look while reducing its exaggeration. Polar regions on it are still badly distorted.
In 1988 the Society replaced it with the Robinson projection, devised by Arthur H. Robinson in 1963 at the request of the Rand McNally company. Robinson worked backwards from appearance, adjusting a table of values at 5-degree intervals until the shapes looked right, and only then fixing the mathematics. His map has straight parallels and stretches the poles into long lines, which flattens the far north and south. Ten years later, in 1998, National Geographic switched again, to the Winkel tripel, saying it reduced the distortion of land masses near the poles. Many educational institutions and textbooks soon followed National Geographic’s example, and most were still using it as of 2012.
Princeton and Drexel Put Numbers on It: Goldberg and Gott, 2007
Choosing a world map had long been a matter of taste. The strongest case for the Winkel tripel came from measurement. In 2007 the physicist David M. Goldberg of Drexel University in Philadelphia and the Princeton astrophysicist J. Richard Gott III published "Flexion and Skewness in Map Projections of the Earth" in the journal Cartographica (volume 42, issue 4). They scored whole-world projections on several separate errors: how far distances are off, how elliptical Tissot's circles become, how much areas are inflated, and two new measures of their own, flexion (how much straight paths appear bent) and skewness (how lopsided shapes become). Across the projections they studied, the Winkel tripel produced minimal distance, ellipticity and area errors and the least skew of any.
Other metrics rank it differently, which is a useful reminder that "best" depends on the yardstick. In a 2001 paper given at the International Cartographic Conference in Beijing, Richard Capek scored a hundred world projections with his "Q" measure, and the Winkel tripel came ninth, behind the Eckert IV and the Robinson. The Kavrayskiy VII projection, created by the Soviet cartographer Vladimir Kavrayskiy in 1939, also scores well against the Winkel tripel despite its straight, evenly spaced parallels, but it was rarely used outside the former Soviet Union. Its reputation rests on doing well across many measures at once rather than winning every one.
Boston, Leipzig and the 2026 UN Resolution: When Equal Area Wins Instead
A compromise map is the right tool for a general reference map, but not for every map. When the point of a map is to compare quantities across countries, such as forest cover, population or the spread of a disease, an equal-area projection is safer, because any visual enlargement of high latitudes misleads the eye. The oldest common choice is the Mollweide projection, published in 1805 by the Leipzig mathematician Karl Brandan Mollweide, which fits the world into a 2:1 ellipse with every area correct. Its shapes, though, are badly sheared near the edges.
A newer alternative came from a dispute in Massachusetts. In March 2017 Boston Public Schools adopted the Gall–Peters projection for world maps so that equatorial countries would appear at their true size. Cartographers objected to Gall–Peters' extreme distortion of shape in the polar regions, and in 2018 Bojan Šavrič, Bernhard Jenny and Tom Patterson responded with the Equal Earth projection, which was inspired by the Robinson's rounded outline but keeps areas exact. One of the first maps published on it was NASA Goddard Institute for Space Studies' map of the July 2018 global temperature anomaly. After a public campaign called "Correct the Map" and a resolution by the African Union, the United Nations General Assembly approved a resolution in September 2026 discouraging the Mercator for most purposes other than navigation and encouraging equal-area projections, naming Equal Earth as one option.
None of this displaces the Winkel tripel from its own niche. A reference map in an atlas or on a classroom wall is read for shapes, neighbours and rough sizes all at once, and that is where averaging Marinus's grid with Aitoff's oval still pays off. If you want to see where each of these designs sits in the family tree, the Robinson projection and the Mollweide projection are its closest rivals, the Mercator vs Peters debate explains why size distortion became political, and the azimuthal equidistant projection is the disc hidden inside the Aitoff half of Winkel's average.
