Answer from cs61c-ae (Zhibo Chen 15283355) for Question 2
A subnormal describes the unnormalized floating point numbers that fill in the space between 0 and the next normalized number.  We care about this format because it allows even more precision to be represented.