In my third block (second block last year), there is this lamp (105.png - What it looked like about when EOL) that went EOL in about early May, late April of last year. It had some blackening and some swirling and orange flashes. But one day when I came to class, it looked like this (106.png), now at first sight it may appear to be completely dead, but it actually is glowing slightly pinkish (more at the less blackened end). These days the more blackened side is the standard EOL brown but the other side stays that pinkish color, you just can't really tell well since the other lamps work. This has happened three times, but the lamp has actually randomly came back on: Twice during class last year, and for a few days this year! (107.png) The lamp is slightly 'mercury starved' now as it is slightly dim once it has warmed up (it takes a long time to warm up too). It also swirls a lot at both ends when it's on. Originally the less blackened end had way less blackening, but while it was 'sleeping' that end has gotten a lot more blackening too.

Does anyone know what causes it to work every once-and-awhile?

 I have seen T12 rapid start do weird things.  First to F40T12 are wired in series on magnetic rapid start ballast. When lamp Eols
 It will do all snaking and swrling,  and color flashes  unti cathode gef broken.   But never drove tube to Vaccume loss
  The dead tube would sit there dim and bit of dim flicker. 
  Every time there a voltage spike come though I would struck
 The lamp and connect them self back in series to run other F40T12.  Cause lamp to run again breifly run again hunting for
Every last emitter on  tube till cant sustain an arc. 

Perhaps that is what is going on here too...

It would be funny if both ends went EOL, the less blackened end does have a decent amount of blackening...