"Pat D." writes:
I have worked on chassis that have had the incorrect value of caps
installed. (Which is why they weren't working in the first place).
Honest to God, they must have had something like a 220 uf @ 160v cap in
place where a 22uf@50v belonged. The person that worked on it before
must have just stuck whatever cap they had sitting on the bench in
whatever spot was open.

I've also seen a board that was marked wrong from the manufacturer.

Once upon a time, there was a PDP-11 that ran Unix, about 1975. DEC
core memory was expensive, so we had a box of some third-party memory,
maybe Dataram. It interfaced to the Unibus via an assembly that
pretended to be a Unibus jumper between adjacent backplanes, so it
didn't take any slots of its own.

One day, we had it apart, and I noticed there was a little tanalum
capacitor that was clearly dead. I figured I might as well fix it,
replacing it with another tantalum of the correct capacitance and equal
or higher voltage rating, carefully orienting it according to the "+"
marking screened on the circuit board. Put it all back together and
turned on the power, and there was the smoke and smell of a fried
tantalum.

So I looked at it again, and traced circuit foils from the plus and
minus supply pins on the Unibus connector. (This was a 5 V bypass cap,
so it connected directly to the power supply pins). Sure enough, the
"+" in the silkscreened marking was adjacent to the *negative* lead of
the capacitor. I installed another cap "backwards" relative to the
marking, and it was fine for the rest of the life of the machine.

Dave