On this day in 1957, a nuclear test called Pascal-B was carried out in the Nevada desert, as part of the Operation Plumbbob series.
Pascal was part of a set of safety experiments carried out in unstemmed shafts at the Nevada Test Site -- Pascal-A had been detonated one month previously, and the Pascal-C shot occurred in December of the same year. A parallel series of shots, named Coulomb-A -B and -C were carried out on towers rather than in shafts.
This particular test is interesting because of a legend surrounding it. It may have inadvertently launched the first man-made satellite, preceding Sputnik by 38 days. A steel cap welded into the shaft may have been blown out the hole by the shockwave from a concrete collimator vaporized by the detonation. An upper bound on the speed of the departing cap, caught on a single frame of high speed film, is six times Earth's escape velocity - although there are plenty of energy balance reasons to think this is unlikely.
The evidence for this is scant: although I have not been able to see it, a February/March 1992 article in the Smithsonian's magazine Air & Space is often quoted, and started much of the 'manhole in space' legend. The article apparently incorrectly refers to this event as "Project Thunderwell," when in fact the launch was not the prime reason for the Plumbbob/Pascal-B test, and no shot or series was ever named Thunderwell. According to a Lowell Wood of LLNL, there was a project concept called Thunderwell that was a first look at requirements to launch spacecraft using nuclear heated steam cannons.
Enviroweb has the best documentation on the web about this event here and especially here, where an eyewitness, Robert Brownlee, talks about the expectations, calculations, and results of the Pascal-B event.
There is a very interesting tale of an internet search for the truth about this event (and how it led off in other directions, too) here.