Hey, guys, I'm not a physicist. Fourier analysis - shrugs...
I think this really is Paul Coxwells field, as he knows telephone systems, where you have several frequencies. (From your voice!) Just to hand over the hot potato...
I don't know the answer, but as the question is interesting, let's see if we can reason us to a result:
The wave in an electrical conductor is moving much faster than the electrons. You can send a wave down a conductor almost without moving the electrons at all. A socket outlet where you have not plugged anything in is an example.
The wave which is easiest to illustrate this with is perhaps the wave among the spectators on a stadion. People don't even move in the same direction as the wave. They sit down and stand up waving, whereas the wave moves sideways. In this case there are only two states, standing up or sitting down. If we had more states, like "standing with hands down", we could send two waves with different speeds at the same time. When there is no wave passing, people sit. When one wave pass, people stand without waving. If both waves pass at the same time, people will stand up and raise their hands. You won't see spectators standing up with their hands above their heads mixed with spectators sitting down.
Back to our socket outlet: Let's say we have wired it up to a 50 Hz and a 60Hz generator in series. Are we going to get heating in the wires because of the different frequencies? Not as long as the circuit is open. Otherwise it would violate the basic principle that electricity needs a closed circuit to flow.
(I don't think this holds true if we have very high frequencies ( kHz ), since the wires will essentially have become a radio antenna.)
In reality, what we have done by superimposing the two frequencies is to create a new waveform. What happens if we plug a resistor into the socket? It will heat up proportionally to the average voltage of the supplying source. We could just as well have heated it with a DC current with the same voltage. This does not hold true for other components, like capacitors.
But, as wires are essentially resistors, the heating of the wires will be the same as if we had used them for a DC current. From this, I conclude that superimposing currents of different frequencies does not case more heating of the wires. It's only the current that matters.
This would indicate that the electrons aren't going in both directions, but rather in just one direction. The atoms in a metal share a common electron cloud, which hold the atoms toghether and makes metals good conductors. (Other materials have ionic or covalent bonds, where the atoms share individual electrons.)
Atoms move at random. (This movement is how heat is stored, and what we call temperature is really how much the atoms move.) It is a reasonable assumption that the cloud of electrons move or change properties at random too, totalling zero current. That is just what Hutch wrote in his first reply.
Superimposing a current with any type of waveform on the random movement simply means that the movement of the cloud will follow the waveform, in addition to the random pattern. Superimposing two currents of different frequency will not cause electrons to go in different directions (apart from any random movement) at the same time since the cloud follows the resulting total waveform. There will not be two separate waves that propagate via different electrons.
Well, that's my take on it. I really hope someone steps in and corrects me now
[This message has been edited by C-H (edited 05-10-2003).]