Dear Bill and John,

A student here, John Deke, pointed out to me today that you
both write

        dF^{-1}(p)/dp = 1/f(p)

instead of

        dF^{-1}(p)/dp = 1/f(F^{-1}(p))

[G.3 p 895, J-D.4 p 433].  Since this doesn't seem to have
appeared in your respective errata, 
I thought I would pass it along.  Deke seemed to think
that the agreement was particularly  problematic and had
actually computed some stuff using 1/f(p), 
before recognizing the problem with it.

Best regards,
Roger