As part of our earlier discussion of car lightning arresters, Dan Mulvihill has provided some new information.

This comes from an ARM reprint of a New Orleana Public Service equipment book.
This is a nice, clear explanation of how a spark gap arrester works. (Although there seems to be a misprint in Design Principles #1. Should be "Arrester must be inoperative on trolley voltage.")

I was struck by the reference to the "left hand three-finger" rule. In physics we always use a right-hand rule, although there are various versions of it.
And I still don't understand the statements in the Westinghouse AMM manual about leaving the light circuits on during an electrical storm. In any case, thanks, Dan!