Normal distribution on 32E and 21S

[Dave Britten]

Does anybody happen to know what algorithms are used internally on the 32E and 21S to calculate cumulative normal and inverse normal? The 32E manual only says "The HP 32E computes Q^-1 using a sophisticated iterative algorithm", and I couldn't really find any details in the 21S manual. It sort of leaves to the imagination how many good digits to expect from the result. The various Application Pacs and books have typically used polynomial approximations, but I'm wondering if either of these models integrate numerically, and if so, to what accuracy. Or perhaps it's something else entirely.