At the elevation in our atmosphere at which these delicate clouds are formed the temperature is too low, even in midsummer, for water to exist in the liquid state; and accordingly, the attenuated vapor from which they were condensed passed at once into a solid form. They consist, in fact, of tiny crystals of ice, not of little drops of water. If the precipitation has been hasty, the crystals will, though all small, be of many sizes jumbled together, and in that case the beautiful optical phenomenon with which we are now dealing will not occur. But if the opposite conditions prevail (which they do on rare occasions), if the vapor had been evenly distributed, and if the precipitation took place slowly, then will the crystals in any one neighborhood be little ice crystals of nearly the same form and size, and from one neighborhood to another they will differ chiefly in number and size, owing to the process having gone on longer or taken place somewhat faster, or through a greater depth, in some neighborhoods than others. This will give rise to the patched appearance of the clouds which prevails when this phenomenon presents itself. It also causes the tiny crystals, of which the cloud consists, to grow larger in some places than others. Captain Scoresby, in his "Account of the Arctic Regions," gives the best description of snow crystals formed at low temperatures with which I am acquainted. From his observations it appears--(a) that when formed at temperatures several degrees below the freezing point, the crystals, whether simple or compound, are nearly all of symmetrical forms; (b) that thin tabular crystals are extremely numerous, consisting either of simple transverse slices of the fundamental hexagon or, more frequently, of aggregations of these attached edgewise and lying in one plane; and (c) that, according as atmospheric conditions vary, one form of crystal or another largely preponderates. A fuller account of these most significant observations is given in the appendix to this paper.