arithmetic underflow - technical definition

The result from an arithmetic calculation that is too small to be expressed properly. For example, in floating point, a negative exponent can be generated that is too large (too small a number) to be stored in its allotted space.