THE NIGHT BEFORE the aviation-safety community torched its own dream of creating jet fuel that would not burn in a crash, a Federal Aviation Administration expert and I walked to our hotel from a restaurant in Lancaster, Calif., the closest civilization to the test site at Edwards Air Force Base.

"I'm worried that we might be in a lose-lose situation," he confided. "We're putting everything on a sample of one. If this works, and there is no major fire, those who oppose the (fuel) additive will say we haven't tested it enough. If it doesn't work, they'll say, 'See? Why bother?'"

That great big fireball on everyone's television screen last Saturday night proves it didn't work as the FAA had hoped. But it would be wrong to conclude that that $11.8 million test was not worthwhile.

It is just as useful to know that something doesn't work as it is to know that something does. If the FAA was planning to require the airlines to use the fuel additive -- called antimisting kerosene -- it needed a strong justification -- a fire-free crash -- because expensive modifications to fuel systems and to highly reliable jet engines would be required to make the additive usable.

Now there is no need. There had been a series of successful tests of the fuel additive at the FAA's technical center in Atlantic City. They were all with smaller planes carrying smaller fuel loads and none of the planes was flown. Instead they were accelerated along a horizontal track before being crashed.

This test was the first with a real jetliner actually powered by jet engines burning the modified fuel. The forces at impact were real airplane- crash forces, with both vertical and horizontal components and with the built-in uncertainties that come from real, rather than laboratory conditions.

Success in the laboratory had given FAA's experts and the manufacturers more confidence than, in retrospect, they should have had. Although they said there would probably be some small fires, they clearly expected those fires to extinguish quickly. Instead, they got your basic aircraft conflagration, the kind that takes more than an hour to put out and that gets so hot it burns through the aluminum hull.

There has been some complaint from the fuel additive's manufacturers that the crash did not follow the planned scenario -- the plane landed left wing low instead of wings level and it did not slide as far as hoped. But it seems a little silly to suggest that pilots about to crash have a lot of control over their scenario.

In the second place, that there was a fire made other experiments on the plane more valuable. The problem with those experiments is that it will take weeks before we know the results. There are no fireballs that can be shown in reverse- angle slo-mo replay.

The other experiments include several that many think will add precious minutes to the time a passenger has to escape a burning aircraft. We know from a series of accidents dating back more than a decade that in crashes similar to this one, people survived the impact only to die of asphyxiation and flames because they could not escape.

The fabrics and plastics used in aircraft interiors -- and the fabrics many of us wear and pack in our synthetic suitcases -- break down under fire into carbon monoxide and cyanide gas. So time is of the essence.

The crashed aircraft contained many seats fitted with so-called "fire-blocking" seat cushions, which will soon be required on airliners. They are supposed to add several minutes to the time it takes a fire to break through the seat. Special windows that are not supposed to burn through as quickly -- and thus protect passengers from outside flames -- were also tested side by side with standard windows.

Onboard high-speed cameras recorded the performance of the windows and cushions. FAA officials believe that film in eight of the 12 cameras survived the crash and will give them a good record of how these experiments worked.

Many pressure and strain gauges were attached to the aircraft, and they continued providing data through telemetry for 10 minutes after impact. Structural experts will study that information for months to see if it shows ways to build things better.

Another experiment of interest was the inclusion in the aircraft of rearward-facing seats, something safety advocates have sought for years because it is thought they would give better support to the back and neck and thus reduce traumatic injuries in a survivable crash.

However, there is this little cosmetic problem. The airlines are not enthusiastic about having the following dialogue ccur between a passenger and a flight attendant:

"Why are the seats facing backwards?"

"So you will have a better chance of avoiding serious injury when we crash."

If the seats won't stay anchored to the floor, it doesn't much matter which way they face. Finding seats that will stay fixed to the floor during a survivable crash has been a tough technical issue and many fasteners have failed. So new types of holders and seat-support designs were also tested. Information on how well they worked will come from both the film and telemetry.

So while the big blast on the desert sent the no-fuel-fire scientists back to the drawing board, there will be many gains.

"I'm feeling a lot better about it today," said my FAA expert Thursday. "We have some wonderful data."