I think Andy is right here, although I can't quite visualise all the
C-shapes contorted into doughnuts and figures of 8. Let's suppose that
you only weld where you would in order to fix the laminations together,
i.e., any welds are perpendicular to the plane of the laminations. If
you put welds on the outer surfaces of both the outer legs of the E,
that's fine, because the loop you create encloses two equal and
opposite quantities of flux: the flux in the central leg is equal and
opposite to the flux in the two outer legs, so there will be no voltage
induced in the loop. But if you made two welds on one of the outer legs
of the E (one on the outer surface and one on the inner), or
alternatively one on the outer surface of each outer leg, and two on
both surfaces of the inner leg, you'd have a problem as the net flux
enclosed isn't zero. I think you get the idea.

Sorry I've taken a while to respond to this thread. My own computer
(faithful Sun Ultra 2) has broken down. The clock stopped working, so I
tried to replace the NVRAM chip and now I can't get the machine to
boot. If there are any Sun hardware experts here let me know - I'd be
very glad of the help. I asked the question at comp.unix.solaris but so
far haven't received much advice.

Best wishes,

Chris