The world we live in...

is a round globe. Small areas on it, of course, are so nearly flat, one cannot tell the difference.

But when one is drawing a map of the whole world, or even a whole country, one has to decide how one is going to represent our spherical world on flat paper. Many different solutions have been developed over the years, and a few of them will be examined on these pages.

Map projections can be grouped together in two basic ways; and a third characteristic, although it divides different ways of using the same projection, is sometimes considered important enough that different versions of the same projection varying only in this characteristic are given different names.

The first characteristic is **aspect**.
This identifies the basic layout of the projection.
The most important projections are either *cylindrical*,
*conic*, or *azimuthal*. A flat piece of paper can,
without stretching, be bent into a cone or a cylinder, and in
this way, it can touch a globe along an extended line; if left
flat, it only touches the globe at a point.

The basic mathematics of obtaining several important properties of maps is different in these aspects.

There are other types of projection than these three basic
types. A projection can be *pseudocylindrical*, which means
that although the parallels of latitude are straight lines,
the scale along them varies, so the meridians are no longer
perpendicular to them, or (necessarily) straight lines.
*Pseudoconic* and even *pseudoazimuthal* projections
are also possible. As well, there is the *polyconic* projection,
and there are many *conventional* projections.

Projections are also classified in terms of their
**properties**. Specifically, on the basis of two
very important properties: whether they are *conformal*, or
*equal-area*, or neither. A conformal projection maintains the shape
of small regions, so angles at any point are correct, although
sizes will change. An equal-area projection, on the other hand,
maintains size at the expense of shape. Maintaining both size
and shape, of course, requires a globe.

In general, and this is true for the projections in the three basic aspects of cylindrical, conic, and azimuthal, scale going away from the center of a map increases for a conformal projection, and, for an equal-area projection, scale in the direction away from the center of the map decreases, to compensate for an unavoidable increase in scale in the direction perpendicular to that from the center of the map.

Most projections that
are neither conformal nor equal area have a scale behavior that
is somewhere in between. However, two very important azimuthal
projections lie outside this range: the *gnomonic* projection,
which can be used to find the great circle path between two places,
and whose scale expands more quickly than that of a conformal
projection, and the *orthographic* projection, which looks
like a picture of a globe, whose scale shrinks more quickly than
that of an equal-area projection.

A map designed for a specific special purpose, particularly one related to navigation, can also be called a chart; this is applied to maps on both the Mercator projection and on the Gnomonic projection. On the other hand, a map on the Orthographic projection can be thought of as a picture of the globe instead of a map. Whether one chooses to view all this as a manifestation of an unjustified prejudice against map projections outside the equal-area to conformal range, or a normal response to the fact that such projections are special-purpose in nature, is a matter of opinion.

Finally, there is the **case** of a projection.
Just as you can move a globe you are holding in your hand, so too
one can think of the *graticule*, the lines of latitude and longitude
on the globe, as being movable.

So one can move the globe, or move the graticule on the globe, and draw a map of a shifted world: that is, although the usual rules for drawing a projection place the lines of latitude and longitude on it in a given way, one can shift the world under the projection's graticule, and treat the original graticule of the world like the coastlines and borders on the globe, as simply things to be drawn where they happen to be.

There is the *conventional* (or, in the case of an azimuthal projection,
polar) case, where the projection is drawn in the
normal and easiest fashion. There is the *transverse*
(or, in the case of an azimuthal projection, *equatorial*)
case, in which the globe has been shifted by 90 degrees before
the map is drawn, and there is the *oblique* case
where the globe is shifted by a lesser amount.

Although some azimuthal projections can be drawn in other cases relatively easily, drawing most other map projections in a case different from the conventional one was very difficult before computers came along to draw maps for us. This meant that in some instances, alternative cases of the same projection were viewed as significant enough to be considered a projection in its own right. Thus, the transverse case of the Mercator projection is also known as the Gauss Conformal Projection; the transverse case of the Plate Carré projection is known as Cassini's projection.

This site is chiefly concerned with map projections from a mathematical point of view, although heavy-duty mathematics has tended to be avoided where possible (thus, the conical projections have only been treated with one standard parallel, and the spheroid has been ignored entirely) but included where necessary (complex numbers and elliptic integrals are mentioned in connection with certain exotic conformal projections). Thus, the complexities excluded are those of interest primarily to professional cartographers; those included are those that may fascinate those outside that field but with a mathematical or technical background.

Of the projections described here, some are projections that you would not be likely to see in an atlas, but many of them have been so used.

In an old-fashioned atlas, the choice of projections was usually fairly limited.

It would likely start out with a map of the world on Mercator's projection. The individual continents might be drawn using projections intermediate between an equal-area projection and a conformal projection. Canada, the United States, Australia, and Europe might be drawn using the simple conic. South America might be drawn using the polyconic projection. Antarctica on the azimuthal equidistant projection.

Russia and China could also be drawn on the simple conic projection. Typically, though, what such an atlas would include is a map of Asia, and there the pattern would break down. A very old atlas might opt for depicting Asia on the Stereographic projection. Later, Bonne's projection tended to be favored for that purpose, and it, rather than the simple conic, would likely be used for Russia as well.

Africa might end up on the Polyconic projection, the Stereographic, the Sinusoidal, or even the Mercator.

Later on, Lambert's azimuthal equal-area projection became the preferred choice for maps of Asia, and the convention in more recent atlases has tended to be to use Lambert's azimuthal equal-area for all the maps except world maps.

Some modern atlases have favored the conformal instead, tending to use Lambert's conformal conic for most of the maps. An oblique Mercator projection was used to very good effect in the Oxford atlas for a map of the Americas, and I illustrate this application of the Mercator within these pages. (I have not attempted to illustrate the scheme of projection involving two oblique Lambert's conformal conic projections of differing orientation patched together which was used for a much-acclaimed map of the Americas referred to in many texts on cartography, however.)

Even when the atlas maps are nearly all in Lambert's azimuthal equal area, the convention of using the Mercator for the world map has often been retained, particularly in the traditionally-minded English-speaking world. But other choices have definitely been used, such as Winkel's Tripel, the Miller Cylindrical, the Mollweide projection (with or without interruptions), the Van der Grinten projection, or even the Van der Grinten IV projection. And many atlases will use an interrupted Sinusoidal or some other equal-area projection for a series of informational maps in the back of the atlas (or, in some cases these days, the front).

Also, an atlas might have a Mercator projection as its main world map, and then include a second one for contrast, such as a Mollweide, or the Eastern and Western hemispheres on the Stereographic projection or the Globular projection, the Northern and Southern hemispheres on the Azimuthal Equidistant, or even an Azimuthal Equidistant centered on the North Pole or on London. The intent might be to show a true equal-area projection, to show a conventional projection that at least improves upon the Mercator's depiction of areas, or to illustrate great-circle routes of air travel. And then some atlases have used an oblique (or even transverse) Mollweide or Hammer to both provide an equal-area world map and illustrate great-circle routes.

And I haven't even mentioned Bartholemew's Regional projection, which appeared in one form in Bartholemew's Atlas, and which was changed for an appearance in the five-volume Times Atlas of the World. As in these cases, sometimes the main projection will be used for a political map of the world, and an alternative projection for the physical world map.

And then there are the atlases that don't use the Mercator projection for the main world map, but sneak one in for a map of world time zones. The Miller Cylindrical projection, the cylindrical equal-area projection, or even the Eckert IV projection, not cylindrical, are other possible choices.

I've written a little BASIC program that draws maps of the world in the form of .xbm graphic files. I then view them in my web browser, press Alt-Print Screen, paste the result in a paint program, and save the images in a more compact format I present on these web pages.

The data points I use to plot most of these maps are from
GSHHS,
a cartographic data set available at the U. S. National
Geophysical Data Center. The lowest-resolution version of the data
set is used by my BASIC program,
as converted to text format by the program provided.
Note that the data files at that site,
although they bear the ".gz" extension,
are NOT compressed, although they are binary. (This may have been done
to avoid problems with web server software that did not recognize the
.b extension of those files.) The ones with the
".bz2" extension at the Hawaii site *are* compressed, using
the bzip2 program, which is available in a version
for conventional PCs, even though it is mostly used under Linux.

Subsequently, I added to the program the ability to use files produced by pscoast within GMT, which I thought stood for the Generalized Mapping Toolbox, but I see the initials actually stand for the Generic Mapping Tools (that is, when they're not standing for Greenwich Mean Time). However, in order to keep the ability to exclude small lakes, I continue to read the coastlines in the original GSHHS format, which includes more information, and is faster for my program to read (which is not a criticism of GMT, as it is intended to work in the world of the UNIX operating system and compatible operating systems, not with QBASIC programs).

The information for the construction of the projections discussed here
is, for many of the more common projections, derived from any of a number of
standard works, but for some of the more exotic projections, various specialized
sources, such as D. H. Mahling's *Coordinate Systems and Map Projections*,
*An Album of Map Projections* by Voxland and Snyder, *Elements of Map
Projection* by Deetz and Adams, and others, were helpful.

This site, needless to say, does not describe all the map projections in existence. However, even some fairly well-known ones have not been discussed here yet. Although I have encountered formulae for the two-point equidistant projection and even the Chamberlin Trimetric projection, I felt that sort of projection would present some problems in implementation in my very simple BASIC program without a fancy interface. I admit I ignored the Goode Homolosine projection specifically because of a dislike for one feature of it; there is a sudden change in the direction of lines crossing the parallel at which the Sinusoidal and Mollweide projections are joined. On the other hand, while I like the Armadillo projection very much, I haven't seen formulae for the little area with New Zealand on it - and it is a projection in a class by itself, requiring a new chapter.

I have eschewed descriptions of the Bipolar Oblique Conic Conformal projection (used for a noted map of the Americas by the American Geographic Society) and the Robinson projection because of the way in which they are defined. The projections involving conformal representations on equilateral triangles devised by L. P. Lee are so far not described here, though they are the sort of thing I would like very much to include, because I feel unprepared to wade into the waters of the Schwarz-Christoffel formula myself, let alone guide the readers of these pages into such territory. The map projections based on elliptic integrals which I do include are already a venture into very advanced mathematical territory.

As noted above, I've also avoided dealing with the spheroid and the use of two standard parallels with conic projections at this time.

Note, too, that some projections, instead of getting a section to themselves, are merely mentioned (and, usually, illustrated) in passing during the discussion of another projection. These projections are mentioned in the table of contents below so that you can find them as well.

- Cylindrical Projections
- The Mercator Projection (including Gauss)
- The Gall Stereographic (also the Braun)
- The Miller Cylindrical Projection
- The Plate Carrée
- The Lambert Equal-Area Cylindrical Projection

- Azimuthal Projections
- Conic Projections
- Pseudocylindrical and Pseudoconic Projections (the Loximuthal, Fahey)
- Polyconic Projections
- Conventional Projections
- The Globular Projection (also Nell's Globular)
- The Van der Grinten Projection
- The Van der Grinten IV Projection
- The Savard Egg Projection (also Fahey)
- The Ginzburg Projection

- Other Conformal Projections
- The Lagrange Conformal
- August's Conformal Projection (also the Eisenlohr)
- Guyou's Doubly-Periodic Projection (also some related projections by Adams, and the Pierce Quincunctial)

- Other Equal-Area Projections
- The Hammer-Aitoff Projection (also Aitoff's Equal-Area and the Wagner VII)

- Miscellaneous Projections
- Winkel's Tripel Projection (also the Aitoff)
- The Aitoff-Wagner Projection (also the Larrivée)
- The Laskowski Tri-Optimal Projection (also Canters' Minimum-Error Polyconic)