They can use any measure they like as long as then can quantify it, and they can come up with an experimental design (i.e., they each take turns seeing if they can lift sequentially heavier rocks until one of them can't) that eventually turns up a difference in the measured variable.
