Mapmaking in ancient times was not a job for the faint-hearted.

Even without the early worries of going too far and sailing off the edge of the world, anyone interested in determining the positions of land masses and shore lines had to face the dangers of reefs, shoals, storms at sea, scurvy, shipwreck, and starvation. Perhaps even worse were the hostile natives met along the way, who killed, among others, the famous explorers Ferdinand Magellan and Captain James Cook.

Mapping the interior of a country was just as difficult. The hardy surveyor had to face deserts, glaciers, avalanches, impassable rivers, infectious diseases, dangerous animals, and still more hostile natives.

And yet maps were early recognized as vitally important. Within settled countries they were needed to define property ownership, set taxes, measure land use, and establish national boundaries. Farther afield, the lack of good maps and accurate knowledge of position led to countless shipwrecks. In 1707, an English fleet commanded by the splendidly-named Sir Cloudesley Shovel made an error in navigation, ran ashore on the Scilly Isles which they thought were many miles away, and lost more than two thousand sailors.

Why was mapmaking so hard? It sounds easy. All you need to define a point on the surface of the Earth uniquely are three numbers: latitude, longitude, and height above some reference surface (usually sea level). Measure a few thousand or tens of thousands of such points, and you have an accurate map of the Earth.

Unfortunately, it was difficult verging on impossible to determine absolute locations. The fallback position was to measure relative locations. Starting with a baseline a few miles long, a distant point was identified, and accurate angles from each end of the baseline were measured. This allowed the other two sides of the triangle to be calculated. From these as new baselines, new angle measurements led to more triangles, which led to still more triangles, until finally the whole country or region was covered by a network. In practice, because there could be small errors in each measurement, all the angles and lengths in the network were adjusted together to produce the most consistent result.

What we describe sounds straightforward, but the amount of measurement and computation in a large mapping survey was huge. The calculations were, of course, all done by hand. A survey of this type could take years, or even decades. There was also no substitute for going out and making ground measurements. Even fifty years ago, it was possible for a leading expert on maps to declare, with perfect confidence, "there is only one way to compile an accurate map of the earth - and that is to go into the field and survey it."

Today, that is not the case at all. The new generation of mapmakers sit in their offices, while far above them, satellites look down on the Earth and send back a continuous stream of pictures revealing details as small as a few feet across. In perennially cloudy regions, spaceborne imaging radar systems see through to the ground below. The location of the images is known fairly well, but not accurately enough to make good maps. However, they can be cross-referenced by identifying common ground features on neighboring and overlapping images. Also, the position of selected points on the ground can be found absolutely, to within a few tens of meters, using another satellite system known as GPS (Global Positioning System).

Finally, all the image data and all the cross-referenced data can be adjusted simultaneously, in a computer calculation of a size that would have made all early mapmakers blench. The result is not just a map of the Earth - it is an "accurate" map and a "recent" map, in which a date can be assigned to any observed feature.

As the people involved in this will tell you, it is still hard work; but it sure beats cannibals and shipwrecks.

"Borderlands of Science" is syndicated by: