In the mid 1990s my protege and I were trying to make a lawn mower engine operate without an exhaust. Our plan was to make the engine run on pure hydrogen and oxygen, for which the only exhaust product would be pure water. We wanted to use a large excess of oxygen to buffer the explosions and absorb the excess heat, and then we were going to cool the exhaust gases and collect the water in a small tank and recycle the excess oxygen back into the engine's intake.

After a few months, we got the engine running, sort of. But by then, because of the hazards of hydrogen, we decided to change over to propane.

The main hazard of hydrogen is that it burns with an invisible flame. A person can walk, unawares, into a hydrogen fire. My arm got burned several times when I accidently reached through an invisible flame to adjust this or that control valve. By contrast, propane is safer, cheaper, and easier to handle.

There was in my basement one last large cylinder of hydrogen gas that was more than half full when I rolled it to my car to return it to the company I leased it from. The cylinder's rated capacity is 239 cubic feet (at atmospheric pressure). That much hydrogen weighs 1.3 pounds yet, when burned in air, it has the energy of 35 pounds of TNT. If you were to empty the cylinder in 10 seconds -- and ignite the escaping gas! -- the flame's heat output would exceed 15,000 horsepower. Would I be able to return a half-full cylinder to its owner without first being tempted to open its valve and light off the escaping gas? No.

I used to race motorcycles and do rock climbing plus other things where my life has depended upon my strong fingers or on this or that thread made of roughly 50 percent skill and 50 chance, though sometimes the mix was more like 98 percent chance and 2 percent skill -- or 100 percent pure stupidity. The idea of venting a cylinder of hydrogen and lighting it off . . . well, I've cited my related experience with such things.

At worst, I figured, I might get painfully singed, or the radiant heat might force me to run from the scene, letting the cylinder fall over and roll burning down the street, its invisible flame swirling around, setting on fire the parked cars of my neighbors.

I needed a helper. My protegee, who called himself Mr. Safety ever since he had caught a piece of hot steel in his larynx from a pipe-bomb "experiment" when he was 15 years old, was at work that day. I'm grateful to Mr. Safety for having seen the potential for accidents in previous, more rationally predicated experiments, but he wanted nothing to do with this "experiment."

One of my neighbors, a guy ten years younger than I who'd done skydiving and motorcycle racing, which were appropriate credentials, agreed to help. Mostly all I needed was a cool- headed person to take photographs and to manage the garden hose if things got out of hand and, if need be, help me into the house and into bed. He was the first-time father of a 5-month-old kid, however, so I wanted him back from the flame.

The challenge of this "experiment" -- and the closest thing to a legitimate goal -- was to photograph the invisible flame of the burning hydrogen. The flame is invisible in the sense that it radiates no visible light, but it might still be captured on film because of the visible distortion caused by the hot gases refracting the light.

The plan was irresponsible, I admit. But not completely. Obviously I was taking precautions, sort of. My main concern was that the flame might somehow get deflected back at me and envelop me in invisible fire. There was no chance of getting killed, like there is in rock climbing, or in skydiving or motorcycle racing. Or little chance.

We did it late in the morning on a weekday, when most of the neighbors were at work or school. The sound was like a military aircraft with its afterburners lit coming down the street. And each time I shut the valve off, the sound reverberated through the neighborhood for several seconds.

Neighbors who were home came out to see. A little boy from Sri Lanka, about 5 years old, came over afterward to ask what we were doing; boys everywhere are alike in their attractions to fire and noise; his sister, a little older, kept playing with her toys on the front lawn.

The photographs showed in the distorted image of the street beyond the invisible the flame the flame's extent: about 8 meters (25 feet). There was no mishap. My only "injury" was the thrill of another adrenaline "fix" that would sustain the chance I'd do some such thing again someday, I hope. Such as . . .

. . . detonate a large airborne balloon filled with pure hydrogen and oxygen.

But first I wanted to do small detonations so as to determine how well the explosion's energy can be coupled to air -- that is, I wanted to measure the power of the concussion and be able to calculate the maximum distance at which the explosion would be audible if no other noises were present.

Water, H2O, is the byproduct of hydrogen/oxygen combustion. That formula, H2O, says that a proper explosion should use two volumes of hydrogen for every one of oxygen.

Mr. Safety was hot to do this balloon experiment. Our first tests were to involve two-foot-diameter balloons that could hold four cubic feet of hydrogen and oxygen. We planned to ignite each balloon electrically from about 30 meters (a hundred feet) away. We also wanted to measure the pressure and wavelength of the shockwave so I be able to adjust the equation I had derived for calculating maximum audibility distances.

We planned that after the small-balloon tests, we would get a balloon large enough to hold a full cylinder of hydrogen and one of oxygen. The balloon would be nearly nine feet in diameter. I calculated that the noise would be audible a hundred miles away, but you'd have to listen carefully in a quiet setting.

Our long-term plan was to find an Air Force base somewhere in the middle of the country, such as Nebraska, a base that was about to be closed. We wanted to approach the Congressional representative from that region and point out the economic advantages of having an annual event, one equivalent to the hot- air balloon festivals in Albuquerque and at the former Donaldson Air Force Base in Greenville, South Carolina.

The annual event or festival we had in mind would bring business to hotels and restaurants that were going to be hurt by the base closing. The first festival would include, say, a rock & roll group, so that the all-day event would consist of music interspersed with high-altitude concussions of increasing size.

The advantage of using an air base would be its size and the presence of radar and air-traffic control facilities, because the final explosion of the day would be at least 2 miles high, and big enough to be heard on both coasts.

The balloon would, of course, have to be BIG. The audience would get to feel the thrill feeling a full-body concussion. And the only environmental effect would be the addition of several cubic feet of pure water to the air -- not counting the fossil energy that would be burned to produce the pure hydrogen and oxygen from water.

This annual festival was to be called, "The Creation of Water Festival," because that's what it would be. That was our goal, to do large-scale art. We wanted to make a whole continent reverberate.

POSTSCRIPT -- About a month after I wrote the above, i.e., on April 19, 1995, Timothy McVeigh detonated a 5,000-pound bomb in front of the Murrah Office Building in Oklahoma City and killed 168 people. The idea for a "Creation of Water Festival" suddenly had all the levity of a balloon filled with rocks.