In other words, for every value x, the value of f (x) is found by summing f (x, y) over all possible values y of Y. We illustrate why this is true in Fig. 3.14. In that figure, the set of points in the dashed box is the set of pairs with first coordinate x. The event {X=x} can be expressed as the union of the events represented by the pairs in the dashed box, namely A = {X= x and Y = y} for all possible y. The A events are disjoint and Pr (A ) = f (x,y). Since Pr (X =x) = ∑_y▒pr(Ay) , Eq. (3.5.1) holds.