This comes from one of my physics course at university (many years ago!), I hope I recall it right.
We were studying optics, and were given some polarizer filters to experiment. They were disc- shaped (like a photographic filter). I stacked two polarizers, face to face, and I verified that, when rotating one of them, a particular orientation (and its 180 degrees rotation) blocked the light. That seemed to make perfect sense to me, because I pictured the polarizer as a wire grid (conceptually like this image)
that only let pass light in a given polarization direction: then, adding a second polarizer to the other behind had no effect when the direction of the wires coincided, and result in high light blocking if they were orthogonal.
We were all satisfied with this explanation... but then I tried to reverse the front filter (I mean, I turned it face down, always keeping it in front of the other, and keeping its "orientation"). To my surprise, keeping the orientation did not keep the block-pass behaviour. Say, if originally the filters were at the "let it pass" position (paralells wires, in my mental picture), putting the filter face down (which would let the imaginary wires in the same orientation) resulted in light blocking, and I had to turned it 90 degrees to let pass the light. This behaviuor puzzled me a little, as also puzzled my teaching assistant (who didn't come with any explanation). I just assumed that my "grid wire" picture was not very apt, but that was all (not enough scientific spirit on my part, I guess).
Can anybody make some sense of this?