Since 1960 the US, USSR and Russia have sent 29 missions toward Mars, and only 8 worked well. The score is lopsided: USA at 8 out of 13, the USSR and Russia stand at zero for 16. The Russians apparently have not yet mastered the hard arts of navigation, reliability in spacecraft design, and hands-on computer maintenance. Most embarrassing, their Phobos probe spun out of control because one technician sent a command that was wrong by one digit.

Of the last 5 US shots, 3 failed. The Mars Observer died at Mars rendezvous in the early 1990s, falling silent suddenly---apparently, a panel decided, because a valve stuck when fuel poured through it, and the plumbing blew open, throwing the billion-dollar craft into a tumble it could not correct.

That loss called into question the faster-better-cheaper philosophy that guides NASA now. Craft sent since then have cost about $125 million, most of it for the launch rocket. But when they failed, too, NASA began to worry.

The case of the September 1999 Mars Climate Orbiter is embarrassing, for many reasons. If the Orbiter had been an airplane, its loss would be ascribed in accident investigators' lingo as a "controlled flight into terrain"---the first for a space vehicle. Complex spacecraft have pointed their instruments the wrong way, failed to open antennas, short-circuited, leaked and exploded---but they haven't been mis-flown. Until now---and the reasons why are instructive.

Everybody now knows that "the" cause was a failed translation of English units of thrust (foot-pounds; the force needed to move a pound a foot in a second) to metric (Newtons, which move a kilogram a meter in a second). True, but not the whole story. The metric conversion mistake was a classic hand-off goof, one side believing that the other was thinking the same way, but not checking.

All spacecraft have to keep their bearings against deflecting forces. Their antenna swings, sunlight pressure, and even the tiny push of the solar wind (that gale of particles blown out from the sun's plasma)---all deflect the craft. That's why most spacecraft spin, to keep their direction more or less fixed.

To adjust its "attitude," or orientation in space, the craft uses "momentum wheels"---small gyroscopes that store angular momentum in isolation, until it's needed to turn the craft. These little tops spin at up to 3000 revolutions per minute, one gyroscope for each of three axes in direction. Another way to move a spacecraft is to thrust with gas jets. The ground team turns on a jet for a few seconds, calculated to correct a straying aim. Both methods get used to make the complicated changes along a long interplanetary orbit.

In that translation, from seconds of thrust to change of momentum, lay the error. Lockheed, who built the thrusters, used English units, which differed by a factor of 4.45 from the metric. This may sound big, but it isn't gross. The bursts were only a few seconds, but when you're adjusting several times a day for six months, they can add up.

Why so many adjustments? The Orbiter had two big appendages sticking out of it, the high-gain antenna for talking to Earth, and its solar panels. Those panels were about as long as a living room (5.5 meters), and the pressure of sunlight alone on them pivoted the spacecraft continually. That had to be fixed, daily.

Why such big panels? Because spacecraft now fly without use of the dreaded N-word, nuclear. Those huge panels provided a kiloWatt of power, about what your refrigerator uses. A small nuclear cell would have weighed much less and provided more power. It would have lasted far longer in orbit at Mars, too.

But the staffs of congressional committees are political to the bone, and they're scared of the N-word because the public has been terrified of it for decades. Nuclear power, nuclear weapons, even nuclear medicine; one wonders what they think of the nuclear family. Using isotopes on cancer patients has a tough time in some communities; Berkeley, California banned such treatments, and has big signs up at the city line proclaiming so. But when I asked a friend who is a cop there, he allowed with a laugh that of course they all look the other way when Alta Bates hospital uses them, and brings in more short-lived radioactive isotopes for the purpose.

Some folks in my own town of Laguna Beach tried to follow suit, and the measure lost only after a long hearing where I and one other physicist testified against a hostile crowd of a hundred. (Doctors everywhere ignore such feel-good bans; they have to watch the patients die.)

So a deeper reason for the Orbiter loss lies with the terror of the N-word. Science fiction is hardly blameless here. For decades sf movies featured giant ants and assorted horrifying, quite impossible creatures as the direct outcome of nuclear testing. Even print sf fastened on nuclear matters for scary plots, and not just nuke-war dramas. We writers should mull over the impact of our choices.

Still, the total of all those mis-calibrated thrusts amounted to about fifty miles (80 kilometers) error in the Orbiter's arrival altitude. The craft was never supposed to enter the Martian atmosphere. Instead, it was to swing around the planet, fire its rocket engine for 15 minutes, and settle into an orbit just above the thin carbon dioxide that accounts for a surface pressure of less than a hundredth ours. A few deft skims at this altitude would make the orbit circular, as is suitable for what would be the first true weather satellite around another world.

But as the craft approached, the flight team kept getting the wrong numbers for its position. Unease spread.

The team managers decided to trust the earlier navigation, telling others to prove the probe was not on the right path. This is exactly the opposite of standard practice, which demands that safety be demonstrated, not assumed.

A day before encounter, the Doppler shifts coming from the craft's radio signal were quite far off, suggesting that it was deeper into the Martian gravity well than it should have been. Still, management stuck with their cross-your-fingers philosophy. Several members of the team later said that they were sure the mission was doomed before the craft began its maneuver.

Simulations now show that probably the Orbiter slammed into the atmosphere at high speed, heating up and tumbling, until about one minute into this gyre, the hydrazine fuel on board reached ignition temperature. "There was enough on board to level a city block," an engineer said. The Orbiter simply exploded into coin-sized fragments, to rain down redly into the chilly Martian night.

This experience was sobering enough. Then came December 3, 1999, and the Polar Lander's utter stillness on the communication channels.

Why silence? Because to save weight and money, the telemetry which usually rides on the descent stage was omitted. After all, on previous missions, the signal had simply told the Jet Propulsion Lab that all was going well.

This landing was to feature a new type of descent mechanism, a complicated array of nozzles, so prudence would dictate telemetry. But no---the cheaper part of faster-better-cheaper won out. So in the 12 minutes between breaking contact with Earth and its programmed touchdown, the Lander was supposed to be silently doing its job.

The task was tough. It had to pull away from the structure that housed it for the nine month journey. After a fiery entry, it had to deploy a parachute, drop its heat shield, radar-lock on the surface, drop the 'chute, then pick a landing spot and settle in.

Which part failed? We don't know.

Maybe none. In late January a persistent astronomer at Stanford University thought he got a weak signal from that antenna--- a feat of detection. The antenna was never intended to do more than send housekeeping data about batteries and the solar panels (them again), up to the Orbiter which was supposed to be flying above.

To pick that signal up on Earth would be marvelous, for it would tell us whether the descent mechanisms worked. Even if they did, there was plenty of danger waiting in the cold wastes of the south pole. Geologists suspect that the crust there may be crunchy and unable to bear much weight, because carbon dioxide freezes out every year and then sublimes away when summer comes to Mars, riddling the soil.

Then too, the unlucky craft may have finally done what the engineers feared---land on a big rock, ledge or pothole, tumble, and sprawl so that no antenna could point at the Earth.

But in searches by several radio astronomy groups in February 2000, that signal did not turn up again. Probably we'll never know more about the Lander's fate. It would be an incredibly dim signal from a small antenna on the Lander, and though some will still try, no one is now optimistic.

I was working at the Jet Propulsion Laboratory the day this detection story first broke. I was helping on a completely different sort of experiment, but I have worked in radio astronomy for decades. I was asked to attend a meeting of the Deep Space Network which discussed the possible detection. Guarded elation was a heady sensation, after so much bad news. Engineers are human. They know the odds, but they, too, have a certain faith in technology.

However, no signals ever came from the two small torpedoes the Lander pitched out at high altitude. They were to slam into the Martian soil at several hundred miles per hour speeds, bore in, and report back with slender antennas what fluids they found, mostly looking for water and carbon dioxide ices.

When they, too, never phoned home, the engineers hauled out all the failure scenarios and frowned. Maybe the torpedoes bored too deep to call back to us. Maybe they hit solid rock, shattering. Maybe the whole package failed to separate from the larger Lander assembly . . .

Silence is the worst result you can get from deep space.

This mix of various human errors and unknowns can instruct us.

Flexibility is crucial. So is a habit of mind difficult to cultivate in management structures---always checking your expectations, reminding yourself that however smart and well educated you may be, you can be wrong. We are awfully good at fooling ourselves.

Comparison with the Challenger disaster are instructive here. In the Orbiter case, managers clung to their models when experience said otherwise. Some elements of the whole mission were needlessly complex---especially the navigation program, a motley assembly with updated patchwork changes adorning it at every turn. And staff were told to get in line with the program: "take off your engineer hat and put on your management hat."

People and their chimpanzee hierarchies will always get in the way of sound technical decisions. Politics trumps physics, usually.

But Nature bats last, to mix metaphors. We can fool ourselves, but not Her.

Then too, the difficulties are not entirely those set by nature. Fear of the N-word has made missions tougher, while not protecting any of us from a real threat. Not that this is easy for the public to see, when even physicists, eager for the spotlight, vastly overplay the risks. But the public still bears some responsibility, for living like Chicken Littles.

Finally, the most important facet of technology is that it must fail, eventually. Nothing is 100% safe. Sitting in your living room, reading this, you could be engulfed by fire, or even drilled by a meteorite, at any moment.

Exploring the solar system is much riskier than that. We have to expect failure and be undaunted by it.

Even better, learn from it---for next time.