Itching for a challenging science project, two years ago Thiago Olson decided to build a small nuclear reactor. He had limited funds, limited space in his garage, and little engineering know-how. After all, he was only 15.

With a year of research and another of building, Olson pulled it off, joining a club of fewer than 20 amateurs in the world who are known to have created "fusors," tabletop machines that fuse atoms to produce energy. There's no risk of a mushroom cloud the machine creates barely enough energy to heat a cup of coffee, and radiation officials in Michigan (where Olson lives) have already deemed it safe.

How did he do it? Olson pored over graduate-level physics textbooks, studied vacuum-pump manufacturers' manuals, and scoured the Web for cheap parts. Though mostly self-taught, he occasionally solicited advice on a fusion Web site. Once, he posted photos of a cheap photomultiplier tube he'd bought online because he had no idea how to rig it up. Another fusioneer on the site who had the same model promptly told him which wires went where.

The latest hope for magnetic confinement is ITER, short for International Thermonuclear Experimental Reactor. The U.S., Canada, Europe, Japan, and Russia began designing it in the late 1980s. But when the price tag hit $ 10 billion, the U.S. had second thoughts and yanked its support in 1999. Since then, a redesign has trimmed the size of ITER and slashed the projected investment to $ 4.5 billion. Early this year, the U.S. began to think about rejoining the program. Groundbreaking for ITER is expected around 2003 or 2004.

ITER would cap a 50-year quest — but wouldn't end it. Fusioneers have always said commercialization is at least 20 years in the future, and it probably still is.

The fusioneers want plasma to produce enough energy for fusion to continue without the help of heaters. When this happens, fusion scientists will run around saying that they have achieved "ignition".