So what happens next? Search me

'Can you . . . ?'

'No, I'm sorry - Yow]'

'Is there any milk here?'

'Ow]'

I'm plucking at a vacuum-packed cylinder of plastic coffee-cups, all jammed together, and covered in a coat of cellophane, and I'm surrounded by a scrum of people; we're all about to go into a conference to watch a demonstration about the unknowability of the world, and the conference is just about to start, so I . . . yes] manage to tear the serrated bit of cellophane at the top of the package, freeing the cups. Now people are grabbing for them, elbowing each other, stretching their arms out to press the tap on the hot-water container, scalding themselves, unable to get the hang of it, because . . .

'Where's the milk?'

'It's here, it's here, it's already in with the coffee . . .'

'No, it's not - look, look]'

. . . Because this is not how people normally get their coffee at conferences. There's a tiny difference. This coffee is free. The organiser has taken a situation and improved it by adding a single, nice-sounding variable - you don't have to pay. And now things are spiralling out of control; we are poised at the beginning of a chain-reaction. Now that there's nobody to take our money, there's also nobody to differentiate one type of cup from another, so people are making mistakes, and trying again, going for a second cup, and this has created a bottleneck, which means that people are barging into each other and scalding their fingers, and putting their cups back on the table, so they can free their scalded fingers and flap them around saying 'Yow]' For a few seconds, the world seems to have strayed from its script, and the table, covered with discarded cups of hot coffee, is beginning to shake . . .

The conference, which is at London University, is called Chaos 93, and it's about Chaos, a scientific concept which is sometimes described as the irregular, unpredictable behaviour of deterministic systems. In other words, if you think you know the facts about something, and that you can predict what's going to happen to it in the future . . . you probably don't. You're probably kidding yourself. Of course, some areas of science are worse, more chaotic than others. It all depends how many factors you have to take into consideration. There are some things - tides, for instance - which are fairly straightforward, because they only depend on one basic variable (in this case, the moon). Many things, though - from the swing of a double-pendulum to the incredibly complex patterns of the weather - depend on so many variables that their behaviour is virtually unknowable. The amount of information you would need is beyond human scope. They might turn out in a predictable way - you never know - or they might display exponential divergence - they might do all kinds of things. This is Chaos.

For scientists, Chaos is a frightening notion. Until a couple of decades ago, the scientific establishment didn't believe in it. People thought: we'll keep plugging away; the mystery will show itself in the end. Ian Stewart, Professor of Mathematics at the University of Warwick, and a leading exponent of Chaos theory, told me that a friend of his, researching the way chicken-feathers grew, started coming up with strange, chaotic results. If he altered the temperature a tiny bit, the feathers grew very differently. 'You couldn't publish that kind of thing 20 years ago,' says Stewart.

Rubbing my scalded hand on the side of my trousers, holding my coffee cup out in front of me, I creep down the stairs. We - around 100 of us, mostly scientists - are about to watch a demonstration which will prove to us that the universe is an unpredictable place, that we needn't worry if we can't work things out. In the lecture hall, Dr Stephen Bishop, an engineer from University College, London, has set up a double-pendulum. We will be able to see Chaos, to watch it happening in front of our eyes. Bishop explains Chaos to us. It is, he says, 'sensitivity due to initial conditions'. In other words, some things are just going to

behave weirdly. Or, they will seem to be behaving weirdly, not because they are supernatural, but because we are incapable of knowing every relevant factor. For a while, things will seem to be going along fairly predictably. And then: wham] - exponential divergence. It makes me think: poltergeists, religious visions, corn circles . . . things I've almost accepted as lies, but not quite, not absolutely.

Bishop begins the demonstration, starting with a pendulum, a stick which swings to and fro on a pin, and then comes to a halt. Nothing abnormal here. And then, to a ripple of appreciation, he unties an elastic band, to reveal a smaller second pendulum, concealed behind the other one, and which can now swing on a pin in the tip of the first pendulum.

Silence. Gently, Bishop swings the double-pendulum. One pendulum moves antagonistically to the other. Bishop swings it a little harder. The same thing, but with greater intensity. 'All this looks normal,' Bishop says, suddenly pushing the pendulum much harder, 'but . . . oh my Lord . . .' What happens next is amazing - the first pendulum swings in what looks like a regular way, and the second one is going crazy, it's flipping round three times to the left, once to the right, twice to the left, doing a half-turn; it's producing apparent Chaos, its movements are unknowable, mysterious. The scientists laugh and burst into applause, as if they've seen the footprint of God.

Outside, as predicted, the weather is moderate and dull, a clear front moving southwards, to be followed by a cold northwesterly airflow. And that's it; that's the limit, the very edge of our knowledge - beyond this point, science is impotent. As I step into my taxi, Elton John is being chased off stage by a swarm of grasshoppers in Melbourne, Australia. Nottingham Forest are winning away. When I get home I'm going to consult my horoscope. -