It would be slightly more satisfying to set theta = f(t), where t is the time variable, but since dtheta/dt cancels out it doesn’t matter.
besides, this would require the dy/dx form of the derivative, and this seems to have gone out of fashion – poor Leibniz
