According to our source, the flaw is remarkably simple -- and involves the simple "ground contact" switch system designed to turn off the Lander's landing rocket motors the moment one of its three landing legs touched the Martian surface.

After its initial high-speed entry into the Martian atmosphere, the Lander's planned sequence of events was as follows: At about 7.3 kilometers above the surface, while the Lander was still moving at about one-half km per second, it would have deployed its 8.4-meter-wide parachute, which would substantially further slow it, but which -- in Mars' faint wisp of an atmosphere -- would be unable to slow it below about 80 meters per second.

Seven seconds after chute deployment, the heat shield would have been released to crash onto the surface; and 16 seconds after that, the lander's three legs -- previously folded up to fit inside the shield -- would swing down and latch in position.

When the Lander was about 1800 meters above the surface and dropping at 80 meters per second, it would have unlatched itself from the chute -- and half a second later, its three clusters of landing thrusters would have started, modulating their thrust during the remaining descent in accord with data from the probe's multi-beam landing radar to gradually slow it to a gentle 2.4 meters/sec final descent.

America's previous unmanned lunar and Mars soft landers -- the Lunar Surveyors and Vikings -- used their radar to sense when they were three or four meters above the surface and shut off their engines then. But for this mission, JPL decided to simplify things by instead simply equipping each of the Lander's legs with a switch to sense when the lower hinged part of the shock-absorbing leg was starting to flex upward after hitting the surface. As soon as any of the three legs sensed this, the engines would have been cut.

According to SpaceDaily's source, it has now been discovered - that when the landing legs of this Lander design first swing down to lock in to position after the heat shield is released, they do with sufficient force that the flexible part of the leg "bounces" slightly upwards again -- triggering the contact switch and setting a crucial "bit" in the craft's computer memory which indicates that ground contact has occurred.

Apparently this event has been "consistently reproducible" in the Failure Board's tests -- but it was never detected during the MPL's prelaunch tests because the leg fold-down procedure was tested by a team separate from the one which tested the craft's behavior during the remainder of the landing sequence.

Thus the first team never noticed that the ground-contact indicator bit had been set as soon as the legs first unfolded down -- and then the second team, before beginning its own tests, reset the lander's computer and thus erased the erroneous indicator bit, so that the craft operated properly during the tests of its remaining landing sequence.