A small satellite designed to study the sun failed to turn on today after it was released from the space shuttle Columbia. The astronauts then were unable to recapture the spacecraft after the shuttle's robot arm somehow bumped it, causing it to tumble in space.

With the Spartan satellite's program of solar observations written off for the duration of Columbia's flight, mission managers were considering a plan for two astronauts to manually pluck the satellite out of orbit during an already planned spacewalk Monday.

Astronaut Kalpana Chawla, operating Columbia's 50-foot robot arm, released the reusable Spartan-201 satellite at 4:05 p.m. as the shuttle soared 172 miles above the Pacific Ocean southwest of Hawaii. Plans called for the satellite to complete two days of observations.

Release from the robot arm should have set off an internal timer, triggering a self-test maneuver causing Spartan to pirouette 45 degrees, which would have indicated the satellite was operating normally. The pirouette never happened.

Moments later, Chawla drove the robot arm's capture mechanism, a device with rotating snares, over a grapple fixture on the satellite. By rotating the snares to tighten on the fixture, the arm could lock onto the spacecraft.

But Chawla apparently never saw an indication of contact and decided to back the arm away for a second attempt. At some point, the arm apparently brushed the satellite or the grapple fixture, imparting a relatively fast two-degree-per-second spin.

The astronauts only had one hour to recapture the Spartan to preserve any chance for a second launch later in the mission. After an hour, the satellite was programmed to disable its maneuvering system as a safety precaution in this sort of situation.

With time running out, commander Kevin Kregel attempted to fly Columbia so it matched the spin rate of the satellite allowing Chawla another capture attempt. If successful, and if engineers could have determined what went wrong in the first place, the crew would have had a shot at relaunching the satellite and salvaging its mission.

In the end, however, it came to naught. As the one-hour timer ticked to zero, ending any chance for a Spartan science mission, flight director Bill Reeves told Kregel to stop the capture attempt to conserve rocket fuel.

The loss of Spartan's mission was a major disappointment to the solar physics community, which had hoped to use the satellite's two telescopes to study the sun's corona, or outer atmosphere.

Data from Spartan also was to be used to help engineers recalibrate instruments aboard the $1 billion Solar and Heliospheric Observatory satellite, a joint NASA-European spacecraft launched in 1995.

The primary goal of Columbia's mission, however, is to carry out a battery of materials science and fundamental physics experiments with a suite of cargo bay instrument called the United States Microgravity Payload, or USMP.

Those instruments are working smoothly and researchers are optimistic about achieving their objectives before Columbia's planned 16-day mission ends Dec. 5. Another major objective of the flight is for astronaut Winston Scott and Japanese flier Takao Doi to stage a spacewalk to test space station construction equipment and a experimental robotic camera system.

The spacewalk originally was scheduled for Monday, but it may be moved up if flight controllers decide to ask Scott and Doi to manually haul Spartan back aboard. CAPTION: Spartan satellite slowly spins away after space shuttle Columbia's 50-foot robot arm attempted to recapture it.