Nigrl Calder has the following diagram in his book:
(Sorry its not a great picture!).
I am a bit surprised that he is so adamant that position 2 is the only "correct" place for the GI.
A. If the GI is working either position is fine.
B. If the GI fails closed ie conducting freely galvanic protection is lost in both cases
C. If the GI fails open ie no conduction at all then in position 1 (the "wrong" position) we loose the earthing of the DC negative/bonding system (bad) but in position 2 (the "correct" postion) we loose all earthing of AC and DC.
Am I missing something here? Are any of the above assumptions (A - C) wrong?