Matthew points to:

See http://www.gowesty.com/starter_story.htm

They recommend not using a rebuilt Bosch starter and explain why.
~~~

Seems to me that the claim that the 'wire' resistance increases with age is
suspect.

Nice to see that Lucas is close but like his namesake perhaps needs a bit
better grounding in the materials science area.  I do not know of a derating
curve for copper vs time.  Not saying that I have all the resources but it
is likely that if that was the case MOST 20yr old electric motors would
exhibit the symptoms of copper aging referred to in the article.

Since the solenoid/relay is a current actuated device it is rated at some
pull force vs applied current, no doubt.  For N amperes of drive current one
should expect P pounds of force.

IF there is mechanical resistance to moving the bendix and closing the
connection to the high current contacts then the force generated by the core
and windings may be insufficient to move the bendix far enough to make the
high current contact.  If there is schmutz gefutzing the mechanism then
applying more current may overcome the mechanical resistance.  For a
sufficient amount of schmutz, the mechanism will not work even with 14VDC
applied directly to the actuator contacts (the ones to which the
starter-ignition signal is connected).

Take a suitable 13.5V source such as a battery charger.  Disconnect the
control wire from the starter contact.  Ground the negative side of the
charger to the case, not the frame, the case of the starter itself.  Now
connect the positive side of the 13.5 V source to the control wire contact.
The starter should spin merrily.  Take two aspirin and let us know how the
test turns out.

This should eliminate the keyswitch, the ignition wiring, and the connectors
AND the ground strap from trans case to frame.

Try it.  Do tell.