In a dramatic test that drew startled gasps from onlookers, engineers fired a chunk of foam insulation at a mockup of a shuttle wing leading edge today, blowing a gaping 16-inch-wide hole in the carbon composite structure and putting to rest any lingering doubts a launch-day foam strike was responsible for the Columbia disaster.

"We believe we have found the smoking gun, we believe we've established that the foam block that fell off the external tank (during Columbia's launching) was, in fact, the most probable cause, the direct cause of the Columbia accident," said Scott Hubbard, a member of the Columbia Accident Investigation Board. "I've now got a direct connection between foam shedding creating a hole that's the same order of magnitude as what must have been there when Columbia came home on Feb. 1."

Eighty one seconds after liftoff Jan. 16, a 1.67-pound chunk of foam insulation broke away from the shuttle's external fuel tank and slammed into the left wing at more than 500 mph. Enhanced video from the one camera that viewed the impact point indicated the foam struck the leading edge at or very near the lower side of reinforced carbon carbon panel No. 8, one of 22 such panels making up the leading edge of the left wing.

But the grainy video, unable to resolve anything smaller than two square feet, provided no direct evidence of actual damage. While most engineers believed the foam strike must have contributed to the breach that ultimately caused Columbia's destruction, they had no proof.

In the wake of the mishap, the Columbia Accident Investigation Board, working with NASA, decided to conduct a complex series of tests to find out whether impacts by low-density foam could, in fact, cause the kind of damage needed to bring down the shuttle. A full-scale mockup of the shuttle's wing leading edge system was built and shipped to the Southwest Research Institute in San Antonio, Texas, where a nitrogen gas cannon was available to simulate the launch-day foam strike.

Initial tests showed foam impacts could cause damage, but the results were not clear cut. Then again, the initial tests involved impacts at RCC panel No. 6, located closer to the shuttle's fuselage. RCC 6 is not as large as panel 8 and does not feature the same complex curvature. For today's test, a foam bullet was fired at the lower side of RCC panel 8, one taken from another shuttle and one with 27 previous flights to its credit.

This time around, the cannon barrel was "clocked," or tilted, 30 degrees to more accurately duplicate the predicted impact energy. The aim point was adjusted to strike the underside of RCC 8 closer to a seal between panels 8 and 9. And this time around, the results were dramatically different.

"There is a huge hole in panel 8!" one observer marveled moments after the test. "It's gone, I mean the foam didn't make it to the back (containment) curtain. You talk about an impact! Unbelievable. If you wanted a smoking gun, you've got it."

Speaking to reporters after the test, Hubbard said high-speed video showed an initial rip that "tears all the way across the panel and produces the hole. The hole is very ragged, about 16 inches by 16 inches, or about 256 square inches. There are a number of pieces that are inside the wing leading edge as well as pieces that fell outside."

One of the enduring mysteries of the investigation has been radar data indicating a piece of debris of some sort separated from Columbia the day after launch. In today's test, the foam essentially blew into the leading edge, pushing large pieces of RCC inside. Based on the size of the breach and the size of the fragments inside the leading edge cavity directly behind the RCC panels, Hubbard said the "flight day 2 object" likely was a large section of RCC 8 that worked its way free in the weightlessness of orbit.

The object seen by radar drifting away from Columbia could "very likely be part of the carbon panel itself," he said.

The foam was fired at the wing mockup at roughly 775 feet per second, or about 530 mph, at an impact angle of 22 degrees. That's higher than the impact angle during Columbia's launch. But by adjusting the impact angle, engineers were able to account for rotational energy imparted by the tumbling foam. Hubbard said the impact imparted about a ton of force to the RCC panel. All of the test parameters, taken together, represented an "average" set of conditions. The actual impact could have been somewhat worse or somewhat less violent. But the hole that was blown in RCC 8 leaves little doubt the foam strike caused the breach responsible for Columbia's destruction.

"I was surprised, I was very surprised," Hubbard said. "As a physicist conducting a test, I feel gratified that after months of work we were able to demonstrate this connection between the foam and the damage. But I know it was a source of tragedy, so that makes me feel very sad. This whole six months, we've constantly been reminded by pictures of the seven lost astronauts what this all means."

The CAIB already has released a preliminary recommendation calling for NASA to obtain spy-satellite imagery of shuttles in flight to look for possible signs of damage. The board also has recommended NASA develop techniques for repairing thermal protection system damage in orbit. Hubbard said today's test shows more data also is needed to understand how carbon composite materials age and react to impacts.

"We need to have better imagery, the shuttle program should have the capability for on-orbit inspection and repair," he said. "We need to realize this is a vehicle that needs to be looked at very carefully each flight. Aircraft, even experimental aircraft, often go through thousands of flights before they're determined to be operational. This vehicle has only 113 flights."

He said engineers are "working to try to determine whether you can do on-orbit repairs of the reinforced carbon. It is the highest temperature area of the orbiter, so finding materials that will patch that are very difficult. I don't know, personally, any way you can patch a hole this big. But that doesn't mean some expert (can't figure something out).

"The RCC has proven to be a very tough material. I think the first step is understanding RCC panels much better and understanding what kind of damage thresholds there are. Where is the dividing point? That database doesn't exist and I think that's one of the first things the shuttle program is going to have to work on."