The world's newest and largest windmill has been ordered by the federal government to restrict operations because it shakes cups and saucers and rattles windows for miles around.

"It makes what has been described as an annoying swishing noise, a most unexpected phonomenon," the Department of Energy's Dan Ancona said of the 2,000-kilowatt windmill DOE put into service last year at Boone, N.C. "Noise was one problem that was never foreseen as being significant."

Foreseen or not, the windmill's 200-foot-long rotary blades will no longer turn early in the morning, at night or on weekends, while DOE and its consultants struggle to find out what's making the noise and move to correct it.

For three weeks in March and April, Boone (pop. 3,686) was besieged by engineers from DOE (which built it) and the National Aeronautics and Space Administration (which tested it when it was built).

To their surprise, the experts found what the townspeople had known for six months: the wind machine generates a very low frequency sound that is aggravated by weather conditions, wind direction and terrain through which the sound travels. Though below the range of human hearing, the sound has been described by Boone residents as a "swish-swish" noise that is plainly an irritant.

"The swish-swish noise is the audible component of the low-frequency sound the blades are generating," Ancona said. "But what the people of Boone are really hearing is their windows and other objects in the homes rattling from the vibrations of this sound."

Apparently, the windmill was built in such a way that it strikes up a two-cycle-per-second sound in the air that is 20 cycles below the threshold of human hearing. The beat travels farther and gets more intense in the mountainous terrain around the windmill.

"It relates to the weather, too," Ancona said. "An inversion in the air around the machine can cause the sound to be redirected and focused, which only makes it worse."

Not wanting to scrap the $6 million project DOE's first move was to restrict windmill operations.

Next, it plans to replace electric generators to drop the rotation of the blades down to 23 rpm from 35 rpm. Changing generators will let the windmill generate a 60-cycle electric current even at a slower rotation of the blades.

If that doesn't stop the noise, DOE will replace the steel blades with fiberglas.

DOE doesn't consider these alternations a setback to its plans for wind energy.

The windmill's rotor is mounted behind the tower. The prevailing theory is that when the rotor is turning, it passes into the "wind's shadow" each time it moves behind the tower. Ancona said engineers believe this causes momentary aerodynamic instabilities that generate the noise.

"This is an experimental machine, and that's why we built it, to find out if there would be any problems," Ancona said. "I think that once we understand it, we can design around it and eliminate the disturbance."

Doe may have done that already. The next generation of wind machines will have the rotors mounted upwind instead of downwind.