Another way to look at it is this:

Let's say you have Q1 and Q2 where the Q1 Emitter is tied to Q2 base. Then you apply a VERY small amount of current to the base of Q1. This is not enough to activate the Q1 transistor fully, but it does create gain/ current flow from the Emitter of Q1 to the emitter of Q2. This current is greater than the current applied at Q1 base, and so the pair can activate at a much lower current (although the voltage required doubles (assuming Q1 and Q2 are identical).

But, don't take my word for it:

-Wes