For a helpful analogy, contemplate our own Earth. Like a pulsar, it spins on an axis. Like a pulsar, it has a magnetic field. The mag­netic field has an axis too, but it's different from the axis of rotation, that's why magnetic north is not the same as true north. There's no good reason for magnetic north to be the same as true north on a pulsar, either. And if it isn't, that magnetic axis whips round thirty times every second. A rapidly spinning magnetic field emits radia­tion, known as synchrotron radiation, and it emits it in two narrow beams which point along the magnetic axis. In short, a neutron star projects twin radio beams like the spinning gadgetry on top of a ter­restrial lighthouse. So if you look at a neutron star in radio light, you see a bright flash as the beam points towards you, and then vir­tually nothing until the beam comes round again. Every second, you see thirty flashes. That's what Bell had noticed.

If you're a living creature of remotely orthodox construction, you definitely do not want your star to be a pulsar. Synchrotron radiation is spread over a wide range of wavelengths, from visible light to x-rays, and x-rays can seriously damage the health of any creature of remotely orthodox construction. But no astronomer ever seriously suspected that pulsars might have planets, anyway. If a big star collapses down to an incredibly dense neutron star, surely it will gobble up all the odd bits of matter hanging around nearby. Won't it?

Perhaps not. In 1991 Matthew Bailes announced that he had detected a planet circling the pulsar PSR 1829-10, with the same mass as Uranus, and lying at a distance similar to that of Venus from the Sun. The known pulsars are much too far away for us to see planets directly, indeed all stars, even the nearest ones, are too far away for us to see planets directly. However, you can spot a star that has planets by watching it wiggle as it walks. Stars don't sit motion­less in space, they generally seem to be heading somewhere, presumably as the result of the gravitational attraction of the rest of the universe, which is lumpy enough to pull different stars in dif­ferent directions. Most stars move, near enough, in straight lines. A star with planets, though, is like someone with a dancing partner. As the planets whirl round the star, the star wobbles from side to side. That makes its path across the sky slightly wiggly. Now, if a big fat dancer whirls a tiny feather of a partner around, the fat one hardly moves at all, but if the two partners have equal weight, they both revolve round a common centre. By observing the shape of the wig­gles, you can estimate how massive any encircling planets are, and how close to the star their orbits are.

This technique first earned its keep with the discovery of dou­ble stars, where the dancing partner is a second star, and the wobbles are fairly pronounced because stars are far more massive than planets. As instrumentation has become more accurate, ever tinier wobbles can be detected, hence ever tinier dancing partners. Bailes announced that pulsar PSR 1829-10 had a dancing partner whose mass was that of a planet. He couldn't observe the wiggles directly, but he could observe the slight changes they produced in the timing of the pulses in the signal. The only puzzling feature was the rotational period of the planet: exactly six Earth months. Bit of a coincidence. It quickly turned out that the supposed wiggles were not caused by a planet going round the pulsar, but by a planet much closer to home, Earth. The instruments were doing the wiggling at this end, not the pulsar at the far end.

Scarcely had this startling claim of a pulsar planet been with­drawn, however, when Aleksander Wolszczan and Dale Frail announced the discovery of two more planets, both circling pulsar PSR 1257+12. A pulsar solar system with at least two worlds! The way you wiggle when you have two dancing partners is more com­plex than the way you do it with one, and it's difficult to mistake such a signal for something generated at the receiving end by the motion of the Earth. So this second discovery seems to be fairly solid, unless there is a way for pulsars to vary their output signals in just such a complex manner without having planets, maybe the radio beam could be a bit wobbly? We can't go there to find out, so we have to do the best we can from here; and from here it looks good.

So there do exist planets outside our solar system. But it's the possibility of life that really makes distant planets interesting, and a pulsar planet with all those x-rays is definitely not a place for any­thing that wants to be alive for very long. But now conventional stars are turning out to have planets, too. In October 1995 Michel Mayor and Didier Queloz found wobbles in the motion of the star 51 Pegasi that were consistent with a planet of about half Jupiter's mass. Their observations were confirmed by Geoffrey Marcy and Paul Butler, who found evidence for two more planets, one seven times the mass of Jupiter orbiting 70 Virginis, and one two or three times Jupiter's mass orbiting 47 Ursae Majoris. By 1996 seven such planets had been found; right now there are about ten. The exact number fluctuates because every so often astronomers discover problems with previous measurements that cast doubt on some­body else's favourite new planet, but the general trend is up. And our nearest sunlike neighbour, epsilon Eridani, is now known to possess an encircling dustcloud, perhaps like our Sun's Oort cloud, thanks to observations made in 1998 by James Greaves and col­leagues. We can't see any wobbles, though, so if it has planets, their mass must be less than three times that of Jupiter. A year earlier, David Trilling and Robert Brown used observations of a similar dustcloud round 55 Cancri, which does wobble, to show that it has a planet whose mass is at most 1.9 Jupiters. This definitely rules out alternative explanations of the unseen companion, for example that it might be a 'brown dwarf', a failed star.

Although today's telescopes cannot detect an alien planet directly, future telescopes might. Conventional astronomical telescopes use a big, slightly dish-shaped mirror to focus incoming light, plus lenses and prisms to pick up the image and send it to what used to be an eypiece for an astronomer to look down, but then became a photographic plate, and is now likely to be a 'charge-coupled device', a sensitive electronic light-detector, hooked up to a com­puter. A single telescope of conventional design would need a very big mirror indeed to spot a planet round another star, a mirror some 100 yards (100 m) across. The biggest mirror in existence today is one-tenth that size, and to see any detail on the alien world you'd need an even bigger mirror, so none of this is really practica­ble.

But you don't have to use just one telescope.

A technique known as 'interferometry' makes it possible, in principle, to replace a single mirror 100 yards wide by two much smaller mirrors 100 yards apart. Both produce images of the same star or planet, and the incoming light waves that form those images are aligned very accurately and combined. The two-mirror system gathers less light than a complete 100-yard mirror would, but it can resolve the same amount of tiny detail. And with modern electron­ics, very small quantities of incoming light can be amplified. In any case, what you actually do is use dozens of smaller mirrors, together with a lot of clever trickery that keeps them aligned with each other and combines the images that they receive in an effective manner.

Radio astronomers use this technique all the time. The biggest technical problem is keeping the length of the path from the star to its image the same for all of the smaller telescopes, to within an accuracy of one wavelength. The technique is relatively new in optical astronomy, because the wavelength of visible light is far shorter than that of radio waves, but for visible light the real killer is that it's not worth bothering if your telescopes are on the ground. The Earth's atmosphere is in continual turbulent motion, bending incoming light in unpredictable ways. Even a very powerful ground-based telescope will produce a fuzzy image, which is why the Hubble Space Telescope is in orbit round the Earth. Its planned successor, the Next Generation Space Telescope, will be a million miles away, orbiting the Sun, delicately poised at a place called Lagrange point L2. This is a point on the line from the Sun to the Earth, but further out, where the Sun's gravity, the Earth's gravity, and the centrifugal force acting on the orbiting telescope all cancel out. Hubble's structure includes a heavy tube which keeps out unwanted light, especially light reflected from our own planet. It's a lot darker out near L2, and that cumbersome tube can be dis­pensed with, saving launch fuel. In addition, L2 is a lot colder than low Earth orbit, and that makes infra-red telescopy much more effective.