Which brings up the question of the vertical temperature profile of the atmosphere, and if the elasticity of the O-rings was still important by the time the shuttle reached a height where the ambient temperature was at or below freezing...
In relation to getting to orbit, it takes very little altitude to "get cold." The average temperature at 30,000 feet is -40. What is the temperature outside of an airliner at cruising altitude?
VIRTUAL CERTAINTY OF FAILURE SHOWN FOR SHUTTLE SEAL - NYT, 1986
The temperature tests bear out, in exacting detail, the general thesis of Dr. Feynman's experiment. The data show that a ring at 25 degrees Farenheit - about the temperature of the joint on Jan. 28 - has less than a fifth the resiliency of the same ring at 75 degrees. Resiliency is of critical importance because the design of the joint calls for the ring to jam into a narrow slot within milliseconds of ignition in order to stop hot gases from escaping. ...In the first set of tests, the two parts of the joint, one containing a metallic tongue, or tang, and the other its fitting groove, or clevis, cleared each other by 20 thousandths of an inch. That put the O ring under relatively little compression and allowed enough room for gases to scoop the ring into the channel, where it could block the gases.
A set of tests conducted at the Marshall Space Flight Center showed that the seal performed perfectly down to 25 degrees, while a parallel set of tests run at Morton Thiokol Inc., the manufacturer of the rocket, showed the first signs of failure at 40 degrees.
In the second set of tests, the clevis and the tang were only 4 thousandths of an inch apart, putting the O ring under significant pressure. In that case, the tests show, the joint sealed only at 55 degrees or above; below that temperature the seal was too hard to seat properly in the channel. Test reports indicate that serious leakage occurred at 40 degrees in a Marshall test and at 50 degrees in a Morton Thiokol test; below those temperatures the seals failed altogether.