Al Gietzen <alventures@email.msn.com> wrote:

> ...  It may more than double
> the failure probability because the return spring in the starter solenoid is
> much more robust than the one in the contactor; so which is more likely to
> fail closed?  
> ...

   This comparison overlooks the relative masses that each spring has to move.  The tiny mass of the contactor core represents a much smaller load (perhaps even proportionally) on its spring.  It could reasonably be expected to endure more cycles of being energized/deenergized before failure.  
   Moreover, it is a small matter to mount the contactor such that gravity adds its small piece to the dis-engaging force - the way Ford did during the '50s.  Of course, such an arrangement could mask incipient failure of the contactor - but in that case, if one hears the starter attempt to engage during a negative-G maneuver, one would know to check the contactor, right?

My $.02

Dale