What part of a train is always moving in the opposite direction relative to the rest of the train?

The flanges on the wheels, where they extend down below the tops of
the rails, always move opposite the main part of the train (except, of
course, if the wheels lose traction and slide on the rails, as I
suppose might happen if the brakes lock up).  That's because the
bearing surface touching the track is rotating "backwards" (relative
to the train) just fast enough not to be moving either backward or
forward against the rails; the flange is a little farther out from the
axle, however, so moves backwards just a little faster than that.

-- David A. Karr <karr@cs.cornell.edu>

[ Editor: an objection to the wording, and Karr's reply, March 1996. ]

I disagree.  The answer takes too much liberty with the statement of the
puzzle.  The puzzle asks for a "part".  Is flanges below the rails
really a part?  It may be an instantaneous "part" but it is stretching
it to say that this thing is a part because material is continuously
entering and exiting this part.  Material only becomes part of the part
when it descends below the rail and then is no longer part of the part
when it ascends again.  This is a really strange part.  The puzzle
requires a part that is "always" moving in the opposite direction
relative to the rest of the train.  It is the always part that makes
this a difficult puzzle.  Note that the part only has to be moving in
the opposite direction as the train "relative to the train" not
relative to the earth.  Sort of like a conductor who starts at the front
of the train and moves backward taking tickets,

What I think we have here is a messed up puzzle statement.  Somebody
took an old puzzle - "What part of a train is moving backwards?" and by
adding his own embellishments, completely screwed up the puzzle.

-- Don Chill <dchill@relay.nswc.navy.mil>

Fair enough.  There were additional hints in the rest of the problem
statement that suggested the poster didn't really mean "relative," and
the words "part" and "always" are just barely ambiguous enough in this
construction to make an argument for the classic answer.  But people
ought to be more careful how they word trick questions such as this.
This is one of those puzzle statements that comes down to, "I know what
question you *meant* to ask; here's the answer to that question."

-- David A. Karr <karr@cs.cornell.edu>