In a potentially significant breakthrough, engineers fired a 1.7-pound chunk of foam insulation at shuttle wing mockup today, visibly cracking a leading edge panel in a test that strongly supports the widely held hypothesis that a foam strike during Columbia's launch doomed the orbiter and its crew.

With a throng of reporters, engineers and a member of the Columbia Accident Investigation Board looking on, a powerful nitrogen gas cannon at the Southwest Research Institute in San Antonio, Texas, fired the box-shaped foam "bullet" at a mockup of a shuttle wing leading edge at a blistering 523.7 mph (768 feet per second). The foam hit a reinforced carbon carbon - RCC - panel taken from the shuttle Discovery. The foam instantly disintegrated in a shower of debris reminiscent of the powdery debris that fell away from Columbia.

From a viewing site 50 yards away, no cracks or other obvious signs of damage were evident. But a close-up inspection revealed a three-inch long crack in the panel through an interior rib. The crack penetrated the panel and high-speed cameras mounted inside the leading edge cavity showed foam fragments blasting into the interior. The force of the impact jammed the inboard side of the panel up against a T-seal between RCC panel 6 and a Fiberglass panel in the inboard No. 5 position and widened the gap between the outboard side of RCC 6 and the T-seal between panel 6 and 7.

Late this afternoon, engineers discovered a second crack in the T-seal itself, near the location of the RCC crack. no dimensions were immediately available.

While today's test doesn't prove the foam lost during Columbia's launch was the initiating event of the disaster, it's the best evidence yet a foam strike could have been, in fact, the root cause.

"This is the first evidence that we've had that a piece of foam that approximates what was observed in the accident can, in fact, crack and damage a piece of flight reinforced carbon panel," said CAIB member Scott Hubbard.

But Hubbard would not speculate on whether damage similar to what was seen today could have led to Columbia's destruction.

"The obvious question one asks is what does this mean in terms of the thermal and structural characteristics and at this point, it's simply too early to draw a conclusion," he said. We've got a couple of the world's experts in RCC and structures over here and they're scratching their heads and trying to figure out what this might mean. As you saw, the response of this is quite different than the response of the Fiberglass."

He was referring to a test last week in which foam was fired at a Fiberglass leading edge panel taken from the prototype shuttle Enterprise. In that test, the impact caused visible shock waves to ripple across the structure, pushing the outboard T-seal out of place and leaving a large gap between the seal and the edge of the target panel.

"Given that the response of the Fiberglass was so dramatic, I think many of us felt there was going to be something perhaps equally dramatic (today)," Hubbard said. "But as I've said, this is why you do the experiment. Drama, of course, wasn't the point of the test, although reporters made friendly wagers on its outcome. As Hubbard pointed out later, the Fiberglass shot, while dramatic, did not cause any cracks or other major damage.

"We didn't show any damage at all on the Fiberglass," Hubbard said. "We had a lot of movement and a lot of high stresses, but here we demonstrated for the first time that foam at the speed of the accident can actually break RCC.

"To me, that's a step forward, maybe a significant step forward in our knowledge."

Eighty one point seven seconds after the shuttle Columbia's liftoff Jan. 16, a suitcase-size piece of foam insulation broke away from the ship's external tank. Columbia was traveling at better than twice the speed of sound at the time - 1,568 mph - and the foam underwent a sharp deceleration when it fell into the airstream. Even so, engineers believe it slammed into the lower side of the left wing's leading edge at some 775 feet per second, or 528 mph while tumbling at 18 revolutions per second.

Columbia was destroyed during re-entry by a breach in the lead edge of its left wing. Analysis of sensor data and evaluation of recovered debris indicates the breach occurred on the lower side of reinforced carbon carbon panel No. 8, one of 22 U-shaped heat-resistant panels making up the leading edge of the wing.

Detailed analysis of launch day footage indicates the external tank foam hit at or very near RCC panel 8. But in the absence of any hard data, investigators have been unable to make a direct connection between the foam strike and the failure.

To find out if foam is at least capable of causing enough leading edge damage to result in a catastrophic breach, the Columbia Accident Investigation Board and NASA, working with the Southwest Research Institute, decided to fire foam bullets at a leading edge mockup in an attempt to duplicate the failure mode. While such tests cannot prove the hypothesis, they can at least show whether it is possible and thus help confirm or rule out various scenarios.

The nitrogen cannon normally is used to fire gel-filled "chicken simulators" at aircraft components to study the effects of bird strikes. For the shuttle tests, a heavily instrumented leading edge mockup was built at the Johnson Space Center and shipped to San Antonio. The mockup includes RCC panels 5 through 10 as well as the T-seals between each panel that effectively lock them in place.

The tests were designed before engineers pinpointed the general location of the breach. As a result, RCC panels 6 and 9 were taken from Enterprise to "bracket" all possible breach locations. For today's test, the foam was fired at RCC panel 6. Upcoming tests will aim at panel 9, both a Fiberglass stand in and the real thing.

For today's experiment, all of the leading edge panels but one - RCC-6 - and all of the T-seals except the one between panels 6 and 7 were made of Fiberglass. Real T-seals and RCC panels, valued at some $800,000 each, are considered too valuable to be risked in any but the most significant of tests. Only one complete "ship set" of spare RCC panels is available, along with the panels already in use by the shuttle's Discovery, Atlantis and Endeavour.

As a result, the first test last week involved firing foam at a Fiberglass panel in the No. 6 position. At least four such firings were planned, but the first test provided dramatic results: The foam lifted and permanently deformed the T-seal between panels 6 and 7, leaving a visible quarter inch gap in the leading edge. Stain gauges measured impact forces up to seven times higher than predicted by computer analysis.

"What we saw was a system response," Hubbard said in a teleconference Wednesday. "We expected to observe an impact and measure that on panel 6, of course. What we found was that we had ... created a response that involved at least three of the panels, maybe more. It was clear the force went from panel 6 to the T-seal to panel 7 and even some measurements in panel 8. So in the accident, then, we have to think of not just the impact point but also the response of the entire wing leading edge."

Worried additional test firings could damage RCC support hardware and make it more difficult to ensure accurate results, Hubbard decided to forego any additional Fiberglass firings and to press ahead with the real RCC panel. To make the test as accurate as possible, a panel was taken from Discovery that has flown in space 30 times, making it comparable to the panels on Columbia's wings.

The foam fired last week and today measured 22.5 by 5.5 by 11.5 inches and weighed 1.67 pounds. The mass, impact angle and velocity were chosen to duplicate the actual impact conditions that occurred during Columbia's launching. Because the nitrogen cannon could not impart the kind of rotational energy seen in the real foam strike, the impact angle was increased, from the actual 15 degrees to 20 degrees, to achieve the same level of kinetic energy.

An initial attempt to fire the cannon failed when an electrical component malfunctioned. Engineers fixed the problem and successfully fired the gun at 3:19 p.m. The blast could be felt by reporters, who were clustered behind a rope watching the test through binoculars.

Speaking to reporters later, Hubbard said "the first significant finding is that we cracked a rib on the structure. Let me emphasize this is the first finding. We will do a complete set of evaluations on this panel, including some specialized non-destructive evaluation, perhaps later today or tomorrow, and we'll be looking at the sensor readings."

The crack, while relatively small, "is not anything you would fly with," Hubbard said. "This would be a piece of hardware that would be rejected. Whether or not this was sufficient to cause structural damage to the whole leading edge or whether it represents a thermal problem is yet to be determined."

Over the next week or so, engineers will prepare the gun to shoot at RCC panel 9.

"We have about a week break here while we reposition the gun and we'll be taking shots at first Fiberglass panel nine and then a shot against the RCC panel nine as well," Hubbard said. "Between then and now we're going to closely evaluate (RCC panel 6), we're going to look at the internal structure and integrity, we're going to look at the sensor readings and see where the forces were and we'll be thinking very carefully about where we want the aim point on panel 9 to be. Nobody has ever characterized this material in this way to see where the point of maximum flexure or maximum strength is."