Originally published 5 January 1987
My word processing computer is on the fritz, so I am writing this column with a pencil on a pad of paper. It took me a long time to find the pencil. I had to buy the pad of paper.
Some little gizmo in my computer went pop and suddenly I’m back in the Stone Age. Or it feels like the Stone Age. It has been so long since I’ve used a pencil that it might as well be a quill pen, or a piece of charcoal on the wall of a cave. It is hard to believe there was ever a time when getting the right words onto paper was such a laborious craft.
I want to write about a book called Weather Prediction by Numerical Process, by Lewis F. Richardson. It is a story that will illustrate our growing partnership with machines. The book was first published by Cambridge University Press in 1922. My copy is a Dover reprint, published in 1965, and for most of what follows I am indebted to meteorologist Sydney Chapman, who wrote the introduction to the Dover edition.
The story begins at a village called Eskdalemuir in the Scottish lowlands. To this place of “bleak and humid solitude,” Lewis Richardson went in 1913 to superintend a meteorological station for the British Meteorological Office. In his off-hours, he developed ideas about a mathematical way to predict the weather.
Calculating the weather
Richardson knew that changes in the atmosphere happen in accordance with well-known laws of physics. From observations of the state of the weather at a certain moment over a broad area of the globe, it should be possible, in principle, at least, to calculate mathematically the state of the weather at some future time. Richardson set down the equations of pressure, temperature, and wind velocity as they apply to the atmosphere, and began to work out a way to test his ideas.
The Great War intruded. Richardson was a Quaker, and his religious convictions prevented him from taking the role of a combatant in the war. He was torn, he wrote, “between an intense curiosity to see war at close quarters and an intense objection to killing people.” He joined an ambulance unit, and in 1916 – 1918 was in France transporting wounded soldiers. During that terrible time, at odd moments snatched “on a heap of hay in a wet rest billet,” he put his ideas of weather prediction to the test.
For the purposes of calculation, he divided a map of part of central Europe into a grid of 25 squares, each 20 kilometers on a side. From published observations of pressure, temperature, humidity, and wind velocity at stations within the cells of the grid, he computed for a place near the center of the grid the state of the weather three hours before and three hours after the time of the observations. These theoretical results he compared to the weather that had actually been recorded. The agreement was not good.
But Richardson knew what was wrong. First, better observations were needed over a wider part of the Earth’s surface. He imagined a grid of 3200 squares covering most of the globe, each with a meteorological station near its center. Many of the stations, of course, would have to be ships. The cost of establishing and maintaining such a network he estimated at $1 billion a year. It was a hugely impractical sum.
The business of computation was even more impractical. For mathematical weather prediction to work, the calculations would have to be ready and distributed well in advance of the time to which they referred. To do this with pencil and paper (the only option at that time), Richardson figured, would take an army of 64,000 mathematical clerks, all calculating in unison.
Here was a man who was way ahead of his time.
Lost in the coal
Nevertheless, the Royal Society, the most prestigious science organization in Britain, contributed $500 toward the publication of Richardson’s mathematical theory. During the battle of Champagne in April 1917, he sent a copy of the manuscript to the rear, where it was lost, to be discovered months later under a heap of coal. With later revisions, this was the manuscript published by Cambridge University Press in 1922.
Richardson’s visionary dream has come true. His mathematical theory is at the heart of present-day efforts of achieve reliable long-range prediction of the weather. A worldwide network of ground weather stations and atmosphere-sensing satellites pours rivers of data into the offices of national weather services here and abroad. High-speed digital computers crunch the numbers through the mathematical equations and predict (with growing success) what the weather will be days and weeks in the future. The current generation of supercomputers accomplishes in minutes what Richardson’s imagined army of 64,000 mathematical clerks might have done in a day.
Even my personal computer could keep up with Richardson’s pencil-wielding army of clerks. It could, that is, if it weren’t on the fritz.