On Monday, Oct 6, 2003, at 15:00 America/New_York, Nils Wagner wrote: > I am interested in the origin of >>>> limits.double_epsilon > 2.22044604925e-16 > > Please can someone explain this value. IIRC, double_epsilon is the smallest representable number by a double (64bit IEEE 754 floating point). Python floats are of this type. -bob