bz wrote:
wrote in news:430b11f5.38471559@news-
server.austin.rr.com:

We see the optical equivalent of a thunder roll, as the sound from
different portions of the path reach our ears.

We recognize that sight and sound can be missleading and that no wrapping
actually occurs.
Wrapping actually does occur. In the rest frame of the rod and of the
rotating cylinder, we are attaching one end of the rod to the cylinder
a second before we attach the other end. The cylinder rotates 10
times before the other end is attached. Since the speeds are everyday
values can actually do this experiment over a much shorter length.
The short rod will attach to the short segment of the rotating disk,
and if we attach one end at a different time then we attach the other
end (the stationary frame view), the attached rod will spiral about
the cylinder. In the problem I posted because of the lengths involved
the spiral and wrapping occur over large distances, but it still
occurs.
David


Not if the attachment is done simultaniously all along the length using
sync'd clocks.

Bz is again talking out of ignorance. The attachment is
synched in the *moving* frame. So, in the stationary frame
the attachment is not simultaneous, and the rod wraps of
necessity.


We could sync it another way:
At a great distance, we have a signal source. The signal from that source
approximates a plane wave all along the length of our cylinder.

BTW, at that distance, we have a powerful telescope and we watch the
joining take place. No twisting is observed.

None is observed in the frame for which the attachment is
simultaneous. But enough of bz -- my post is mainly about
something else, to wit:

I think dseppala has actually managed to ask an interesting
question here, although as usual he's made it more complicated
than it needs to be. Also I believe he errs (along with most
of his respondents here) in thinking that this can be treated
as a 1D problem when in fact all 3 spatial dimensions must be
considered.

Let me recast it in simpler terms: Imagine an ordinary steel
corkscrew (actually any suitably asymmetric object would do)
rotating along its long axis; and suppose for simplicity that
it has exactly an integral number of turns so that it doesn't
wobble as it rotates. The C.O.M. is fixed; 3-momentum is
conserved in this frame.

Now, look at it from a different frame, one that is moving
along the screw axis. We find that the number of turns in
the corkscrew has *changed* so that there is more mass on one
side of the axis than the other. What happened to momentum
conservation?

Note that in some frame the corkscrew will even be *straight*.
(This is the extreme case that dseppala was considering.) And
yet it will still be going round and round its axis as it zips
along. 3-momentum definitely not conserved.

You can't just say this is impossible; I'm talking here about
an *ordinary* corkscrew, not some 300000km long one. The
inescapable conclusion is that 3-momentum is not conserved.