In gazing at the successive phases presented by the polar caps as their annual history unrolls itself to view, beginning with vast white cloaks that in winter hide so effectively the planet’s shoulders, to little round knobs that in summer sit like guardsmen’s caps more or less askew upon the poles, the bodily eye sees only the glisten of far-off snow. The mind’s eye, however, perceives something more: the conviction they carry of the presence of an atmosphere surrounding the planet. Elusive as water vapor is to sight for its transparency and to spectroscopic determination for its earthly omnipresence, recognition of its existence elsewhere by deduction raises such reasoning at once to a more conspicuous plane than it might otherwise assume. Especially is this true where the deduction is itself conclusive, as is here the case. For it depends on phenomena not its own, but which are in their turn dependent on it. We are not even beholden to any knowledge of the substance composing the caps for the fundamental inference that an atmosphere surrounds them. Whatever that substance were, the fact that the caps dissipate and reform shows us with absolute certainty that they pass into the gaseous state, to be later solidified afresh. This gas constitutes of itself an atmosphere; while another phenomenon, to wit, their blue girdles as they melt, affirming their substance to be snow and ice, enables us to precise the fact that this gas is water vapor. From such premise given us by the polar caps we are able to infer much more by the help of the kinetic theory of gases. But the speed of parting by a planet with its gases is conditioned by the mean speed of each gas. Water vapor will, therefore, go before nitrogen, oxygen or carbonic acid gas. If, then, we find it present over the surface of a planet we are assured of the possibility that the other three may be there too, and from the similarity of matter in space strong reason to suspect that they actually are. Corroborative evidence of the accuracy of the deduction as to the presence of a Martian air is shown in several other ways; in the existence of clouds to begin with. Rare as they are, these certainly float at times over parts of the planet, although it is doubtful whether they can then be seen. Fortunately for assurance we have other ways of ascertaining their presence than that of obscuration. Nor is it of account to the argument that they should be few and far between, as they unquestionably are. One single instance of such mediumistic support is enough to support the theory of a medium; and that instance has been more than once observed.