I'm pretty sure that the "'genius benchmark' of 140" comes from Lewis Terman. No modern test or professional would classify somebody as a "genius" simply based on an IQ score.

IIRC the verbal test used by British Mensa is the Cattell III B, which has a standard deviation of 24. The estimates for Einstein and Hawking (as well as quite possibly being way off) are usually expressed in SD 15.

"The thing that doesn't fit is the most interesting."
-Richard Feynman