The size of Mars taught us that we have in it a planet with an atmosphere of but one half the density of that prevailing on the top of our highest mountain; the distance of Mars from the Sun showed us that it must have a mean temperature close to that of freezing mercury. What chance would there be for life on a world the average condition of which would correspond to that of a terrestrial mountain top, ten miles high and in the heart of the polar regions? But Mars in the telescope does not look like a cold planet. As we look at it, and note its bright colour, the small extent of the white caps presumed to be snow, and the high latitudes in which the dark markings--presumed to be water or vegetation--are seen, it seems difficult to suppose that the mean temperature of the planet is lower than that of the Earth. Thus on the wonderful photographs taken by Prof. Barnard in 1909, the Nilosyrtis with the Protonilus is seen as a dark canal. Now the Protonilus is in North Lat. 42°, and on the date of observation--September 28, 1909--the winter solstice of the northern hemisphere of Mars was just past. There would be nothing unusual for the ground to be covered with snow and the water to be frozen in a corresponding latitude if in a continental situation on the Earth. Then, again, in the summer, the white polar caps of Mars diminish to a far greater extent than the snow and ice caps of the Earth; indeed, one of the Martian caps has been known to disappear completely. Yet, as the accompanying diagram will show, something of this kind is precisely what we ought to expect to see. The diagram has been constructed in the following manner: A curve of mean temperatures has been laid down for every 10° of latitude on the Earth, derived as far as possible from accepted isothermals in continental countries in the northern hemisphere. From this curve ordinates have been drawn at each 10°, upward to show average deviation from the mean temperature for the hottest part of the day in summer, downward for the deviation for the coldest part of the night in winter. Obviously, on the average, the range from maximum to minimum will increase from the equator to the poles. The mean temperature of the Earth has been taken as 16°C, and as representing that prevailing in about 42° lat. The diagram shows that the maximum temperature of no place upon the Earth's surface approaches the boiling-point of water, and that it is only within the polar circle that the mean temperature is below freezing-point. Water, therefore, on the Earth must be normally in the liquid state.