Spontaneous fission, says TEPCO, the NISA says they cannot rule out criticality, and no one cares as no one believes either of them.
Here's another reason not to believe them, or take their words at face value: TEPCO tested at different locations and for different durations for the nuclides in the gas that was sucked out of the CV of Reactor 2.
Take a look at the table below, which I compiled from TEPCO's handout for the press on November 2, 2011. For each nuclide, the detected amount and the detection limit are listed for October 28, November 1, and November 2.
I was fooled when I first looked at the result of October 28, when TEPCO started the gas management system in Reactor 2. Krypton and xenon were all below detection limit. Then on November 1, they were all detected, and on November 2 xenon-133 (half life 5 days) seemingly dropped out, but other nuclides increased slightly. I thought "Something must have happened between October 28 and November 1."
That was what some reporters at the press conference on November 2 were wondering, and TEPCO's Matsumoto went along with them, not saying anything definite but hinting that the fission event was recent.
WRONG. These reporters didn't look at the fine print. (Well, neither did I...)
On October 28, the detection limits were HIGH, including those for radioactive iodine and cesium (not in the table above), indicating TEPCO didn't test long enough. That can also be inferred from the "date of sampling" information, because for October 28, no duration is indicated in the TEPCO's press handout. The nuclides may well have been in the gas on October 28 but weren't detected.
Then I also noticed that the samples were taken at different locations. On October 28, the sampling location was "gas sampling system intake"; on November 1 and 2, it was "gas sampling system dust radiation monitor".
For some unknown reason, between October 28 and November 1, someone somewhere in TEPCO seems to have decided to test longer, and at a different location. It is also possible that the dust radiation monitor was simply not ready on October 28, although I doubt that even TEPCO would start running the system before the system components were put in place and functional.
And do not be fooled that xenon-133 stopped being detected on November 2. Look at the detection limit for xenon-133 on November 1 and compare it to November 2: 10^-6 vs 10^-5. Since the amount of xenon-133 found on November 1 was in the order of 10^-6, it may well have been present in the sample on November 2, but not detected.
Here's TEPCO's handout for the press (11/2/2011):
October 28, 2011 (2nd measurement):
November 1, 2011:
November 2, 2011:
Here's the Reactor 2's gas management system overview, as released by TEPCO on October 27, 2011:
TEPCO and the Japanese government don't even know where the corium is or whether the Containment Vessel has been breached (most likely). Sucking the gas out of the Containment Vessel, which may or may not have the corium, to "reduce the amount of radioactive materials" is like trying to scoop water with a sieve.
(H/T anon reader)