Bose also redesigned the suspension geometry. Because the suspension is programmed to maintain a nearly level ride attitude at all times, there's no need for the camber to change much in order to keep the tire contact patch on the road as it arcs through its range of travel. The upper and lower control arms (front) and links (rear) aren't the same length, but they're close, which reduces tire scrub as the suspension strokes over a bump.

We sampled the ride (sort of) in a 1994 Lexus LS 400 fitted with the new system. It was resting on four electromagnetic rams programmed to pound the tires as a rough dirt road would if traveling at about 45 mph. The Bose struts were programmed to mimic the stock LS 400 suspension at first, the effectiveness of which was demonstrated on an identical stock LS 400 bouncing on a second set of shakers over the same rough-road program in the next stall. When the engineers flipped a switch to the Bose suspension control strategy, the car went totally smooth. Of course, the cynics among us suspected the same computer that was telling the shaker how to shake the wheel might have been informing the shocks about what to expect and how to react, so we remained blase.

Then we went outside to watch both the stock and Bose-modified LS 400s negotiate a tight (30-mph max) lane-change, panic stop, speed bump, and slalom course. The Bose car's performance on the course was impressive: no body roll, brake dive, or acceleration squat at all. Of course, the programming could allow some body roll if desired. Over the speed bump, the body didn't move. It looked as if it had picked its tires up and put them down again (it doesn't do this, the system is purely reactive).

For the grand finale, Dr. Bose staged what he called a braking-precision demonstration: "Watch this 2x6 plank we've placed on edge across the roadway and see how close we can stop in front of it," he instructed. The Bose car roared toward the plank at 35 mph, but the driver never lifted. Instead, as the car approached, it hunkered down, and then leapt over the plank, clearing it like a steeplechase stallion. This was a parlor trick, with no practical application, but it served to illustrate the amount of force these shock units can exert. By this point we had plenty of tough questions for the engineers.

How much power does it consume? The system draws about two horsepower or one-third the load of a typical air conditioner. While it can exert 50 kilowatts (67 horsepower) of energy to leap a 2x6, it recovers 49 kilowatts cushioning the landing, with the shocks working like generators.What does it weigh? Bose claims its torsion bars and shock units weigh about what two conventional springs and shocks plus an (unnecessary) anti-roll bar weigh, but that the wheel damper adds unsprung mass to each corner. The controllers and upsized alternator also add some weight, but the total should be less than that of a hydraulic active suspension.

What's it cost? More than anyone else's suspension does now. The neodymium iron in the magnets is the most expensive part, and its cost has fallen in half during the 24 years of development. Expect to see electromagnetic suspensions only on very expensive cars first, and probably never on cheap ones. What happens when the power goes out? Unlike your doorbell plunger, these neodymium magnet plungers move like softly damped shock absorbers if you short the leads to their electrical windings, which is the fail-safe mode. The torsion bar springs continue to carry the weight, and the shorted struts behave like luxo-barge shocks.

How durable is it? The electromagnetic plunger floats inside its winding, so it's virtually non-contact, meaning wear should be way less than with normal shocks. The wheel dampers presumably would wear like normal shocks. When will we see it? By mid 2005, a manufacturer will be chosen to codevelop a production application for sale after three or four years. GM is expected to be the first development partner, given the long relationship between the companies.

Could this tech be applied elsewhere? Absolutely. We asked whether the same concept could be used to provide infinitely variable valve timing and lift without a camshaft, and were stonewalled, so that's almost certainly under development. Driving simulators, vehicle shakers, amusement park rides, etc. are all possibilities.

The road from the computer model through full 100,000-mile harsh-environment durability testing is a mighty difficult one, and many great ideas never complete the trip. We sincerely hope this one does.