"We must assume... a later form of Pythagoreanism... was closely akin to early . [T]he fifth-century doctrine was of the more primitive type..."