----------------------------Original message----------------------------
Dear list (and Nick!)
 
Map projections not based on the poles of rotation of the earth are quite
common, though not necessarily straightforward. In principle, ANY map
projection can be constructed on any arbitrary pole - the mathematical
dodge is to transform the geographic coordinates to a new pole (fairly
simple spherical trigonometry) and then apply the usual projection
routines. An example that is available on the Web is the Logo of the Scott
Polar Research Institute (http://www.spri.cam.ac.uk/), which is an oblique
aspect of the Mollweide Projection.
 
A lot of useful information can be found in the documentation for Proj, a
free projection conversion utility that can be downloaded from
kai.er.usgs.gov/pub/PROJ.4. It has extensive documentation in Postscript
format.
 
The main reason that most projections stick to "natural" projection
directions is that the mathematics for dealing with a non-spherical earth
become very nasty! Spherical versions of projections are not affected, and
for world maps or small scale maps these are quite appropriate.
 
It is worth mentioning that map projections are mathematical entities which
have an existence regardless of whether anyone has used it for publishing a
map. Most so-called "new" projections are minor variations on well known
projections. The (IMHO) dreadful Peter's projection is an example of this -
it is simply a particularly awful version of the Lambert Cylindrical Equal
Area projection, known since the 18th century!
 
A Paul R Cooper
GIS Manager
British Antarctic Survey