It is easy to see how the evidence of the presence of any element which behaved in this way would be weakened, if we consider what would happen in the case of our own earth, according as the air were simply moist but without clouds, or loaded with cumulus masses but without cirrus clouds, or loaded with cirrus clouds. For although there is not in the case of the earth a central glowing mass like the sun’s, on whose rainbow-tinted spectrum the dark lines caused by the absorptive action of our atmosphere could be seen by the inhabitant of some distant planet studying the earth from without, yet the sun’s light reflected from the surface of the earth plays in reality a similar part. It does not give a simple rainbow-tinted spectrum; for, being sunlight, it shows all the dark lines of the solar spectrum: but the addition of new dark lines to these, in consequence of the absorptive action of the earth’s atmosphere, could very readily be determined. In fact, we do thus recognize in the spectra of Mars, Venus, and other planets, the presence of aqueous vapour in their atmosphere, despite the fact that our own air, containing also aqueous vapour, naturally renders so much the more difficult the detection of that vapour in the atmosphere of remote planets necessarily seen through our own air. Now, a distant observer examining the light of our own earth on a day when, though the air was moist, there were no clouds, would have ample evidence of the presence of the vapour of water; for the light which he examined would have gone twice through our earth’s atmosphere, from its outermost thinnest parts to the densest layers close to the surface, then back again through the entire thickness of the air. But if the air were heavily laden with cumulus clouds (without any cirrus clouds at a higher layer), although _we_ should know that there was abundant moisture in the air, and indeed much more moisture then there had been when there had been no clouds, our imagined observer would either perceive no traces at all of this moisture, or he would perceive traces so much fainter than when the air was clear that he would be apt to infer that the air was either quite dry, or at least very much drier than it had been in that case. For the light which he would receive from the earth would not in this case have passed through the entire depth of moisture-laden air twice, but twice only through that portion of the air which lay above the clouds, at whose surface the sun’s light would be reflected. The whole of the moisture-laden layer of the air would be snugly concealed under the cloud-layer, and would exercise no absorptive action whatever on the light which the remote observer would examine. If from the upper surface of the layer of cumulus clouds aqueous vapour rose still higher, and were converted in the cold upper regions of the atmosphere into clouds of ice-crystals, the distant observer would have still less chance of recognizing the presence of moisture in our atmosphere. For the layer of air between the cumulus clouds and the cirrus clouds would be unable to exert any absorptive action on the light which reached the observer. All such light would come to him after reflection from the layer of cirrus clouds. He would be apt to infer that there was no moisture at all in the air of our planet, at the very time when in fact there was so much moisture that not one layer only, but two layers of clouds enveloped the earth, the innermost layer consisting of particles of liquid water, the outermost of particles of frozen water. Using the words ice, water, and steam, to represent the solid, liquid, and vaporous states of water, we may fairly say that ice and water, by hiding steam, would persuade the remote observer that there was no water at all on the earth—at least if he trusted solely to the spectroscopic evidence then obtained.[2]