When adding two floating-point numbers we can

a)	Get a negative answer even if both the arguments are positive
b)	Obtain the same result as one of the arguments, even if none of them is 0
c)	Obtain different result if we denormalize the other number

not only one can be correct