Hugh Herr and I are navigating the snow-covered sidewalks of Cambridge, Massachusetts, in two distinctive styles. I test my steps and vary my stride as I make my way among clumps of slick wet snow. He plows straight ahead. We cross a shiny black street, watching for Boston's notorious cabs. On the far side, I balance on the ball of my left foot, stretch my right leg, then hop across a slushy puddle to the curb. Herr strides right through. The cold doesn't bother his feet, although he complains that they squeak when the weather is wet: They're made out of carbon fiber.

When he was 17, Herr got trapped in a blizzard during a mountain-climbing trip and lost both legs below the knees to frostbite. Shaken, the former C student focused on his studies, eventually earning a master's degree in mechanical engineering from MIT and a Ph.D. in biophysics from Harvard. Now, as a codirector of MIT's Leg Lab, he's one of a number of researchers developing radically improved prosthetics. The field is still closer to Captain Ahab's peg leg than Captain Picard's Borg implants, but researchers are making remarkable progress toward creating artificial limbs as good as human ones.

One reason Herr walks so well is that he still has his own knees. People whose legs have been amputated higher have trouble stopping suddenly or recovering from a stumble, let alone navigating icy sidewalks. Human knees adapt to changes in speed and terrain by bending at varying rates and stopping at different angles. Traditional artificial knees swing free while the leg is in motion, then lock as weight is put on that leg. The result is a stiff, awkward gait.

Otto Bock Orthopedics Industry of Duderstadt, Germany, is using computer chips to help prosthetic knees behave more like real ones. The company's new C-Leg contains a lithium-ion battery and a microprocessor that measures the angle of the knee and the rate at which it's bending 50 times per second. The computer uses that information to project what the amputee is trying to do and adjusts valves to change the flow of fluid within hydraulic chambers inside the knee, increasing or decreasing resistance as necessary. "Every step an above-knee amputee takes, as he comes down he has to be concerned that the leg is stable," says Todd Anderson, a prosthetist at the Minneapolis branch of Otto Bock. "But if we could develop a prosthesis that does exactly what an anatomical leg does, theoretically the brain is already programmed to respond to that."

To match each individual's natu-ral gait, the C-Leg must be custom-programmed by a prosthetist when fitted. The Leg Lab is working on an auto-adaptive knee incorporating sensors that determine the correct operating parameters without any help. Researchers in the Smart Integrated Lower Limb program at Sandia National Laboratories in Albuquerque, New Mexico, aim to go further. They are working on prosthetics that measure not only knee position but also the forces exerted on the foot as it strikes the ground. Those inputs would allow a more accurate simulation of normal limb motion.

"This isn't going to be the bionic man or anything like that; it's going to be something much simpler," says Diane Hurtado, the former manager of the Sandia project. Even a small improvement would mean a lot to amputees, because plenty of routine activities are beyond the scope of artificial legs. "Going from standing to walking up a set of stairs requires your leg to completely change configuration," says Hurtado.

Hands are even more complicated. Most artificial replacements are little more than pincers covered with plastic shaped to look like skin. About all they can do is grasp and release. Human hands, by contrast, can pivot, twist, grasp, and pluck. Fingers are capable of moving in at least 22 different ways.

Bill Craelius, an associate professor of biomedical engineering at Rutgers University, is testing a hand that restores some of that finesse by tapping into the ability of certain amputees to sense the limb they lost. If they try to move a missing finger in a particular way, the tendons in the residual arm move the way they would if the finger were still there. The Rutgers "myo-pneumatic" hand contains foam sensors that connect to truncated tendons. Changes in the shape of the tendons squeeze the sensors and transmit data to a computer chip that controls electronic actuators in the arm. "Our vision is the user is going to more or less adopt this thing as part of his body," says Craelius.

Amputees testing the tendon-activated hand have been able to tap out words on a keyboard and to play the piano— slowly— as was demonstrated at the 1999 Discover Technology Awards expo in Orlando, Florida. A commercial version should be available later this year. So far Craelius's subjects can use three separate fingers. In the future he hopes to provide some wearers with five movable digits. Still, his approach will aid only the comparatively small number of patients who have tendon control. "We're a long way from really duplicating the human hand," he says.

One major limitation of current prosthetics is that they rely on mechanical connections, which cannot match the intuitiveness of the body's own neural wiring. Attempts to control prostheses with electrical sensors on the skin have not had much success, mainly because it is difficult to distinguish among the many electrical changes that activate muscles. A number of researchers are therefore trying to tap straight into the nervous system, but that work remains at an early stage of development.

Another problem is power. An ideal prosthetic should move on its own, but conventional motors are heavy and consume too much electricity. In the Leg Lab, Herr and his colleagues are experimenting with a robot driven by animal-derived muscle tissue that burns glucose, just like human muscle. An experimental "biomechatronic" fish built in the lab can propel itself with a tail driven by real muscle fibers. Herr thinks it may someday be possible to replace missing limbs with artificial skeletons powered by bioengineered muscle and hooked into the nervous system.

Even current technology, Herr says, has some advantages over flesh and bone. His rock-climbing abilities have actually improved, because his lightweight prostheses fit into crevices where a human foot cannot. Still, most amputees yearn for a prosthesis that is far more intimate with the body.

"If I can feel my ankles, or better still, if I can think and just move that foot around, it's going to be almost as if my legs were never amputated," he says.