This paper presents a systematical comparison of pointing gestures in the context of controlling home appliances in smart homes. The pointing gestures were conducted with head, hand, arm and a computer mouse serving as baseline. To the best of our knowledge, we are the first to report on such a systematical comparison of the mentioned modalities. Our results indicate that although being overall slower and more inaccurate than hand and arm gestures, head gestures are more suitable for mid air input than previous research indicated. We show that disadvantages like slowness and inaccuracy can be compensated by a larger target size. In addition, head gestures have the largest learning effect. Considering our results and the possibilities head gestures would provide in daily life, we recommend thinking of head gestures as a feasible input modality besides hand and arm gestures.