Mike,

A request for even more info:

1)Is this a one of a kind board?

2)How is it constructed (i.e. Wire Wrapped, PCB, Proto-board, etc.)?

3)If not "one of a kind" is this the only board of many doing it?

4) Have you "rung" out the board including the "pi" filter caps?

5) Can you "jumper" an alternate clock source to observe any change in the conditions.

You mentioned you swapped chips so it would seem the board is problem and considering how little is required to oscillate the problem should be easily solved. How easy is it to "fly" an xtal and the two required caps of board with a jumper tying the caps to gnd?