yes they are curved.... one edge is N and other edge is S....and of course
you can turn them end for end to reverse polarity.....
possible to stick them in unglued temporarily slow speed 100 rpm to see which way
works best?
So they are not magnetised through the thickness? (i.e north on inside of curve and south on outside?)
then, if i understand the magnetization correctly, four can be arranged:-
nn
s s
s s
nn
or
ns
s n
n s
sn
my guess would be the 1st one, which would match your 4 pole 90° arrangement. (I suppose the rotor can be wound for 4 poles with just 2 brushes, and the 90° brush positions would suggest that.)
you could try passing current through the rotor and use a compass to see how many poles it has.
m