That it was marred by glitches has left scientists scrambling to better understand what Deep Space 1 and other missions might encounter when they fly by comets in the next several years. A workshop will be held at NASA's Jet Propulsion Laboratory in Pasadena, California on November 5 to study how to better model what comets look like from up close and how to tailor mission software accordingly.

"If these were just as bright and shiny as icebergs it wouldn't be a problem," said Peter Thomas, a Cornell University senior research associate who is a co-investigator on the Comet Nucleus Tour (CONTOUR) mission. The mission, slated for launch in 2002, will fly by three different comets in 2003, '06 and '08.

Scientists are interested in comets because they are thought to contain, preserved by the deep freeze of space, the original ingredients that made up our solar system, including organic material that may have helped spark life here on Earth.

But comets' nuclei are obscured by abundant clouds of gas and dust. This makes them difficult to study from Earth. And rendezvousing with them -- or any small body, like an asteroid -- in space is an even trickier proposition.

Due to the fact comets are so fast moving, and small -- often just miles or kilometers across -- spacecraft are increasingly being given broad autonomy in navigating the final steps to close encounters. Indeed, to orchestrate such a maneuver from Earth would be impossible: by the time coordinates were received from the spacecraft and final commands sent back, it would have long zoomed past the tiny body.

Even autonomous navigation has its pitfalls, as Deep Space 1 showed during its Braille flyby. The spacecraft's Miniature Integrated Camera Spectrometer -- which gathers the images needed for the spacecraft's AutoNav subsystem -- failed to capture the asteroid during its closest approach. More precisely, the experimental instrument actually imaged the object, but the pictures were too dim to be recognized as such by the on-board computer.

"The pictures would be taken, delivered to the computer and AutoNav said, 'No asteroid here, I'll keep pointing this way,'" said Deep Space 1 chief mission engineer Marc Rayman.

For Deep Space 1, close-up pictures of Braille would have been icing on the cake for the otherwise successful flyby. As it was, the spacecraft was able to image the asteroid only a full 15 minutes after its closest approach.

But similar images are the point of two flybys Deep Space 1 intends to make of the comets Wilson-Harrington and Borrelly in January and September 2001, respectively. And the issue may be the same for the mission Stardust -- which shares near identical tracking software with Deep Space 1 -- for its scheduled flyby of the comet Wild-2 in January 2004.

"The bottom line is you better know how bright your comet is and what your camera can see before you get there," said Shyam Bhaskaran, of NASA's JPL, who wrote the tracking software for both Deep Space 1 and Stardust.

And indeed for a slew of other missions designed to look specifically at comets' nuclei that will soon be underway and that will rely on similar autonomous navigational systems to reach their target, the unexpected problems that befell Deep Space 1 would represent a severe setback.

"Stardust and all other missions going to small bodies are faced with the same dilemma: You don't know what they look like until you get there," Rayman said.

To help ensure missions like Deep Space 1, Stardust, Deep Impact and CONTOUR can best address that problem, Rayman hopes the cometary summit he has convened at JPL will help. The technical workshop will draw together astronomers, engineers, scientists and navigational experts to work on better modeling of what comets may look like - often an issue more of informed opinion than fact.

"There are very good models, but they disagree with other," said Michael A'Hearn, a University of Maryland astronomy professor and the principal investigator on the cometary mission Deep Impact, scheduled for launch in 2004. "That's the problem."

The question underscores the difficulty of pinning down a comet nucleus' size, shape, spin, mass, volume and surface morphology. Scientists have seen exactly one cometary nucleus from any appreciable proximity - but even that was of little help, for no two are alike. Rather than a specific model, scientists are often left with complex matrices that cover a broad range of possibilities.

"We are facing the same problems [as Deep Space 1 and Stardust] and it's not just a question of getting a better model, but a question of understanding the range of possible parameters," said A'Hearn, whose mission will fire a projectile into the comet P/Tempel 1 and then image its interior. "We know very little about cometary nuclei."

For Deep Space 1, time is of the essence: Any changes to the spacecraft's software must be uploaded by March in advance of its January 2001 flyby. That means the programs must be written by January.

"It's not like we have a long time to stew about this," said JPL comet expert Donald Yeomans.