With the thumb serving an important role in the function of the human hand, improving robotic prosthetic thumb functionality will have a direct impact on the prosthesis itself. So far, no significant work exists that examines the ranges of motion a prosthetic thumb should exhibit; many myoelectric prostheses arbitrarily select them. We question this design practice as we expect a significant functional volume reduction for performing certain activities vs. the maximum obtainable workspace. To this end, we compare and contrast four anatomically-accurate thumb models. We quantify their angular ranges of motion by generating point clouds of end-effector positions, and by computing their alpha-shape bounded volumes. Examining the function of the thumb for several grasps, we identify a 76% reduction of the required workspace volume vis-a-vis the maximum volume of a "'generic'" human thumb.