Some ideas are so attractive that they don't go away, no matter how impossible they seem. For millennia, humans dreamed of flying like a bird, of living forever, of a life without labor, of freedom from pain.

And for millennia, those desires came no closer to being satisfied. Only in the last couple of hundred years have we made great strides toward achieving humanity's ancient dreams. Many of us fly regularly, though hardly with the ease and freedom of birds. The inhabitants of the prosperous countries of the world live on average almost twice as long as their eighteenth century ancestors, our daily activities are increasingly free of hard physical labor, and while our painkillers and anesthetics are not perfect, they are enormously better than the prospect of surgery without them.

Of course, we seldom pause to think ourselves lucky to be living today, rather than a thousand years ago. Instead we invent new dreams, for things that remain beyond reach. Sure, we can fly through the air safely and cheaply, but that's not enough. Flight to space remains at the very limit of our performance. A three-day jaunt to a different continent can be had for a thousand dollars. A three-day trip to space will set you back at least twenty million, and doesn't begin to cover the real costs.

It doesn't have to be that way. In 1960, a Russian engineer named Yuri Artsutanov described a device which could in principle make access to space inexpensive and routine. If the space shuttle is like a ferry-boat to and from orbit, Artsutanov's idea is like a bridge to replace that ferry. No one except the Russians knew of his work, and the idea was independently discovered in 1966 by a team working at Woods Hole, and "discovered" yet again in 1975 by the American engineer Jerome Pearson.

What follows describes Pearson's method of building the structure described by Artsutanov. Imagine, he said, that you are out in a geostationary orbit, turning at exactly the same rate as the Earth and thus hovering always above the same point on the equator. Such locations are already accessible. They are where we place our communications satellites and many of our weather satellites. Now begin to lower a cable down toward the surface, while at the same time extending another cable outward. The weight of the cable below you tries to pull downward, while centrifugal forces on the cable above you tries to pull outward. If you arrange the rate of cable extension correctly, the forces away from and toward Earth will exactly balance.

Keep paying out cable, upward and downward. Eventually, the lower end will touch down on the surface of the Earth. Anchor it there. You now have a load-bearing structure, similar to the cables that hold up a suspension bridge. Attach cars to run up and down, and you have a space elevator, also known as a skyhook, beanstalk, orbital tower, or cosmic funicular. This forms a bridge to space, available for use twenty-four days a day. You can do away with the whole messy and dangerous business of rockets.

This is such an attractive notion that you may be wondering what's holding us back. One thing is the length of cable needed. Geostationary orbits are 35,000 kilometers above the surface. Since we have to extend cable in both directions from there, even if we attach a large ballast weight at the outer end we are still talking of a structure at least 50,000 kilometers long.

Curiously enough, this is not the big problem. We could, today, manufacture that long a cable. What we couldn't do is make it strong enough to support its own weight, let alone the cars and drive system needed to make it into a transportation system. The strongest material we can produce, carbon dislocation-free filaments, are at least five times too weak. Some people pin their hopes on a new form of carbon known as fullerene tubes, but so far this is nothing more than speculation (and perhaps wishful thinking).

It sounds as though space elevators must remain as dreams until we can make materials far stronger than any we know. However, there are variations on the basic idea which we could build now. One is a rotating beanstalk, or Rotavator, which orbits the Earth and turns as it does so, so that one end touches down several times each orbit. This could be built with today's materials and carry objects to and from space, but I must admit the Rotavator does not much appeal to me. For one thing, the end must descend and immediately ascend at high acceleration. You have only a second to attach your payload.

An idea I like much better is the dynamic beanstalk. In this device, a continuous stream of objects (say, steel bullets) are launched up the center of an evacuated vertical tube. As the bullets ascend, they will be slowed by gravity and also will receive an additional deceleration through electromagnetic coupling with coils placed in the walls of the tube. When this happens, the bullets apply an upward force to the coils. This continues all the way up the tube.

At the top, the bullets are reversed in direction and allowed to drop down another parallel evacuated tube. As they fall they are accelerated downward by coils surrounding the tube. This again results in an upward force. At the bottom the bullets are slowed, caught, given a large upward velocity, and placed back in the original tube to be fired up again. We have a continuous stream of bullets, ascending and descending in a closed loop.

If we arrange the initial velocity and the bullets' rate of slowing correctly, the upward force at any height can be made to match the total downward gravitational force of tube, coils, and anything else we attach to them. The whole structure will stand in equilibrium, and serve as our load-bearing structure for sending payloads up and down. The dynamic beanstalk does not call for super-strong construction materials and it can be of any length, though there would be little point in making one that did not ascend at least to low Earth orbit. Since the dynamic beanstalk will stand there, apparently unsupported by anything, it is like the Indian Rope Trick. Building this would be the biggest engineering project ever.

So when will we build a beanstalk, in any one of its several possible forms? I won't offer a date, but give you the answer offered by Arthur Clarke: "About fifty years after people stop laughing."