Any room with mostly metal walls can behave like a cavity, with Efield peaks and valleys.  I would set up two antennas, one with a sweeper, one with a spectrum analyzer, and sweep the frequency until I saw the signal appear to be "very large" at the spectrum analyzer (ie bigger than the path length would predict).  That is probably a resonance.  You could then move around and confirm that there are standing waves in the room, maybe tweak the frequency a little either way.

BTW, I am not sure what type of "lightbulb" they were talking about.  If it was an unpowered flourescent, It sounds like the RF Power level was dangerously high (tens of watts at least) to make it light!

from the article;
Lewin, says his former student, figured out that MIT's biggest lecture hall, the cavernous Room 26-100, could be turned into a low-energy microwave cavity. "He mounted a light bulb on an antenna," recalls Leeb, "and he'd walk around mapping the microwave field according to where the light bulb would glow or dim. It was terrific."

Sounds like as the antenna radiated and reflected off the walls, the reflections would change from peaks to nulls and change the glow of the bulb. A large cavity can focus alot of energy, but it sounds a bit fishy.
I've heard ex-military types talk about ground based radar lighting up fluorescent bulbs that were hand carried. Those were levels that didn't permanently injure people, though it may be folklore.
 
kch
PS:he might have fooled his professor by having a control in his pocket. you know how those MIT Engineers enjoy fooling Harvardites at the annual football game.