But for myself. I think he did indeed plumb, at the last, the depth of the sea. For I do not know what he was, or what he had become. And presently people came and tore down most of the engines.'

He shifted his weight and looked at the remains of the strange devices, skeletal against the livid sunset. There was something wistful in his expression.

'No one comes now,' he said. 'Hardly anyone at all. This is where the Fates struck and the gods laughed at men. But I remember how he wept. And so I remain, to tell the story.'

THE NEW NARRATIVIUM

The wizards have been trying to find some 'psyence' in Roundworld, but it is proving even more elusive than the correct spelling. They are having problems because they are tackling a difficult question. There isn't a simple definition of 'science' that really captures what it is. And it's not the sort of thing that comes into existence at a single place and time. The development of science was a process in which non-science slowly became science. The two ends of the process are easily distinguished, but there's no special place in between where science suddenly came into being.

These difficulties are more common than you might expect. It is almost impossible to define a concept precisely -think of 'chair', for example. Is a large beanbag a chair? It is if the designer says it's a chair and someone uses it to sit on; it's not if a bunch of kids are throwing it at each other. The meaning of 'chair' does not just depend on the thing to which it is being applied: it also depends on the associated context. And as for processes in which something gradually changes into something else ... well, we're never comfortable with those. At what stage in its life does a developing embryo become a human being, for instance? Where do you draw the line?

You don't. If the end of a process is qualitatively different from the start, then something changes in between. But it need not be at a specific place in between, and if the change is gradual, there isn't a line. Nobody thinks that when an artist is painting something, there is one special stroke of the brush at which it turns into a picture. And nobody asks 'Whereabouts in that particular brushstroke does the change take place?' At first there is a blank canvas, later there's a picture, but there isn't a well-defined moment at which one ceases and the other begins. Instead, there is a long period of neither.

We accept this about a painting, but when it comes to more emotive processes like embryos becoming human beings, a lot of us still feel the need to draw a line. And the law encourages us to think like that, in black and white, with no intervening shades of grey. But that's not how the universe works. And it certainly didn't work like that for science.

To complicate things even further, important words have changed their meaning. An old text from 1340 states that 'God of sciens is lord', but there the word57 'sciens' means 'knowledge', and the phrase is saying that God is lord of knowledge. For a long time science was known as 'natural philosophy', but by 1725 the word 'science' is being used in essentially its modern form. The word 'scientist', however, seems to have been invented by William Whewell in his 1840 The Philosophy of the Inductive Sciences to describe a practitioner of science. But there were scientists before Whewell invented a word for them, otherwise he wouldn't have needed a word, and there was no science when God was lord of knowledge. So we can't just go by the words people use, as if words never change their meanings, or as if things can't exist before we have a word for them.

But surely science goes back a long, long way? Archimedes was a scientist, wasn't he? Well, it depends. It certainly looks to us, now, as if Archimedes was doing science; indeed we have reached back into history, picked out some of his work (especially his buoyancy principle) and called it science. But he wasn't doing science then, because the context wasn't suitable, and his mind-set was not 'scientific'. We see him with hindsight; we turn him into something we recognise, but he wouldn't.

Archimedes made a brilliant discovery, but he didn't test his ideas like a scientist would now, and he didn't investigate the problem in a genuinely scientific way. His work was an important step along the path to science, but one step is not a path. And one thought is not a way of thinking.

What about the Archimedean screw? Was that science? This wonderful device is a helix that fits tightly inside a cylinder. You place the cylinder at a slant, with the bottom end in water; turn the helix, and after a while water comes out at the top. It is generally believed that the famous Hanging Gardens of Babylon were watered using massive Archimedean screws. How it works is more subtle than Ridcully imagines: in particular, the screw ceases to work if it is held at too steep an angle. Rincewind is right: an Archimedean screw is like a series of travelling buckets, separate compartments with water in them. Because they are separate, there is no continuous channel for the water to flow away along. As the screw turns, the compartments move up the cylinder, and the water has to go with them. If you hold the cylinder at too steep a slope, all the

'buckets' merge, and the water no longer climbs.

The Archimedean screw surely counts as an example of ancient Greek technology, and it illustrates their possession of engineering. We tend to think of the Greeks as 'pure thinkers', but that's the result of selective reporting. Yes, the Greeks were renowned for their (pure)

mathematics, art, sculpture, poetry, drama and philosophy. But their abilities did not stop there.

They also had quite a lot of technology. A fine example is the Antikythera mechanism, which is a lump of corroded metal that some fishermen found at the bottom of the Mediterranean Sea in

1900 near the island of Antikythera.58 Nobody took much notice until 1972, when Derek de Solla Price had the lump X-rayed. It turned out to be an orrery: a calculating device for the movements of the planets, built from 32 remarkably precise cogwheels. There was even a differential gear.

Before this gadget was discovered, we simply didn't know that the Greeks had possessed that kind of technological ability.

We still don't understand the context in which the Greeks developed this device; we have no idea where these technologies came from. They were probably passed down from craftsman to craftsman by word of mouth -a common vehicle for technological extelligence, where ideas need to be kept secret and passed on to successors. This is how secret craft societies, the best known being the freemasons, arose.

The Antikythera mechanism was Greek engineering, no question. But it wasn't science, for two reasons. One is trivial: technology isn't science. The two are closely associated: technology helps to advance science, and science helps to advance technology. Technology is about making things work without understanding them, while science is about understanding things without making them work.

Science is a general method for solving problems. You're only doing science if you know that the method you're using has much wider application. From those written works of Archimedes that still survive, it looks as if his main method for inventing technology was mathematical. He would lay down some general principles, such as the law of the lever, and then he would think a bit like a modern engineer about how to exploit those principles, but his derivation of the principles was based on logic rather than experiment. Genuine science arose only when people began to realise that theory and experiment go hand in hand, and that the combination is an effective way to solve lots of problems and find interesting new ones.