The worlds that time forgot

Like many more recent design ideas, the medieval mappa mundi may not be very useful, but it's certainly rational. At its most basic, it hits what's called a T-O format. A circular world is bounded by a circular ocean, and divided into three parts by a T-shaped sea. The top semicircle is Asia (east is always at the top), the bottom right section is Europe, and the bottom left Africa. Jerusalem is set bang in the centre. The T stands for, roughly, the Med.

This neat geometry may be elaborated to include rivers and islands, and to represent coastlines with a little more accuracy. The Red Sea is included, coloured bright scarlet. Adam and Eve in Eden appear at the very top. Weird humanoids are shown along the edge of Africa. Though mappae mundi were an English speciality, information about the British Isles isn't much more solid. Wales and Scotland often snuggle into the bottom left- hand corner - as islands.

Geographically, the mappa mundi knows very little. Conceptually, however, it knows everything: it admits no incompleteness. There are no terrae incognitae lurking at its borders, no half-charted coastlines trailing off into doubt, and (whether or not these map-makers were really flat- earthers) no awkwardness about projecting the round globe on to a flat surface. Not until the 19th-century, when every land-mass was at last reckoned, would there again be such confident images of the whole world.

You can find several mappae displayed in "The Earth and the Heavens", a British Library exhibition at the British Museum, which presents an extensive collection of world-maps and star-maps, ancient and modern, though predominantly Western. The arrangement is chronological, but the organisers are keen that we shouldn't simply see a story about the progressive improvement of geographical and astronomical knowledge. Rather, every map is a conceptual scheme, an imaginary picture, a work of art and a cultural document.

This is quite true: the history of maps incorporates the histories of fine-penmanship, semiotics, zoology, geometry, sea-faring, religion et al. Mapping motives are always so mixed and so large. Very few of the world-plans here would be of much practical use to a traveller. Each sets out a view of the world in a more than geographical sense. That now-familiar photo-image of our planet taken from the moon is as much an icon - Spaceship Earth - as any of the mappae.

What makes maps interesting is that you can't be wholly relatavistic about them. However artificial any projection system must be, there is something like a right answer. We see old maps through later ones, and some idea of progress has to shape our viewing. And when that progress is complete - as it is, mostly, from around 1800, after Cook had finally cleared up Australia - the interest does fall off.

Three hundred years previously there were several sources for world geography. The circular mappa mundi still had some mileage in it, and could be considerably refined if you didn't stray too far afield. The world picture worked out by the second-century Greek Ptolemy had been reconstructed, and it told much more about Africa and Asia, though it also posited a vast land-mass filling most of the southern hemisphere. There were Marco Polo's journeys, which were good for place-names if not for locations. There was news coming back from the great sea voyages.

What you find over succeeding centuries are fascinating borderline cases, as the more the map-makers know, the more the gaps in their knowledge also become clear. These holes have to be dealt with somehow - by resorting to tradition or guesswork or by frankly admitting ignorance. The impulse to set down accurate information is continually in conflict with the impulse to provide a complete picture.

Take North America. You have the eastern seaboard marked down for certain. What then? Do you square it off tidily into a very thin continent? Do you extend the land-mass right round to join east Asia? Do you stop the coastline where knowledge runs out, or block further questions with the map's edge, or gracefully blend the unknown territory into the blue sea? All these solutions appear. Map-makers will happily give a definite, realistically wiggly contour to terra the caption confesses is incongita.

Traditional knowledge keeps its authority even when it's been superseded: thus, on several maps, Ptolemy's "India", a very large island, protrudes from south Asia alongside an obviously real India. Meanwhile, rumours and confused sightings raise the new Atlantises like "Company's Land" in the North Pacific, a phantom terrain claimed by the Dutch East India Company, which is sometimes shown with continental dimensions. The inclination, while options remain open, is usually to believe in as much dry and habitable land as possible.

Any map drawn boldly enough carries its own conviction. Hindsight can play in two directions. One sees these old charts not only as primitive versions of the world, but also as fantastic mutations, strange but strangely true. Other possible worlds heave into being. You half-believe there really was a "Company's Land". You see not just the arbitrariness of maps but the contingency of the globe, too - alternative plate tectonics, so to speak.

The star-maps don't present such exciting revelations and metamorphoses. The stars have always been exposed to everyone's gaze. The Copernican revolution didn't radically alter how the heavens looked, though telescopes filled in details and travel extended the view to the southern skies. Moreover, with the stars, observation and imagination don't become interestingly tangled as they do with land-maps. You easily separate the constellations themselves from the colourful astrological menagerie that is so often imposed on them. Nor is there much you can do with these creatures artistically. Their layout is fixed. It's always a jumble. The more lively you make them look, the more wilful their invention seems. Astrology may be imaginary, but it's not much of an advertisement for the imagination.

It has stuck, though. No re-naming has ousted the old Babylonian images, whether the Counter-Reformation's attempt to Christianise the heavens, or AP Herbert's A Better Sky (1944), which tried to construct an improving modern pantheon with constellations such as Science, Children's Corner and Europe Regained (Herbert solemnly anticipated the artist Simon Patterson's recent games with maps and classification).

It's perhaps surprising, now these charts have done their knowledge-work, that they haven't become more of a motif for artistic play. Jasper Johns painted maps, but he never messed with the contours. Yet when you see well-known geography going through multiple transformations - as you do here - you realise what expressive characters the lands' outlines wear. The British Isles, so anthropomorphic anyway, seem especially fertile ground for variation. I think there might be a project here.

n 'The Earth and the Heavens' continues at the British Museum, London W1, to 22 Oct. Details: 0171- 636 1555