Answer from cs61c-at (soemoe aung 16076861) for Question 2
A subnornmal is a denormalized number.  They have the same epxponent as zero but a nonzero significand, and they allow gradual underflow so numbers smaller than those possible in regular floating point can be represented.