Floating-Point Arithmetic with 84-Bit Numbers
A classic and straightforward technique is
presented which is not limited to the size or type 
of number representation used or multiple precision arithmetic.
CACM January, 1964
Gregory, R. T.
Raney, J. L.
