Tall Fathers, Slightly Shorter Sons

In a large population, fathers' heights ($x$) and their adult sons' heights ($y$) have the same summary statistics:

$\bar{x} = \bar{y} = 68 \text{ in}, \qquad s_x = s_y = 3 \text{ in}$

The correlation between them is $r = 0.5$.

A least-squares line is fitted to predict a son's height from his father's. For a father who is 74 in tall, what height does the line predict for his son — and what share of the variation in sons' heights does the line account for?