Isn't the precision of a float 8 decimal digits? .dig reports 6 Why isn't "%.100g" cropped to the max decimal digits of the accompanying type? floating point types are printed with 20 digits.