Once the pawl was reassembled and inserted into KG88II the force required to displace the spring was measured and found to be 100gms. The garter carriage was then set in operation and the left-turn pawl worked perfectly for a while before failing once again. So our next step will be to re-test the springs.
An oddity: Since I wrote the above I've noticed that the problem never happens on 'ordinary' rib, just while knitting patterns ...
Problem solved! See post of 25th Nov 2013