Beginning with nebulæ, we pass on to stars having banded or fluted spectra, indicating comparatively low temperatures and showing bands or lines of iron, manganese, calcium, and other metals. They are more or less red in colour, Antares in the Scorpion being one of the most brilliant red stars known. These stars are supposed to be in the process of aggregation, to be continually increasing in size and heat, and thus to be subject to great disturbances. Alpha Cygni has a similar spectrum but with more hydrogen, and is much hotter. The increase of heat goes on through Rigel and Beta Crucis, in which we find mainly hydrogen, helium, oxygen, nitrogen, and also carbon, but only faint traces of metals. Reaching the hottest of all--Epsilon Orionis and two stars in Argo--hydrogen is predominant, with traces of a few metals and carbon. The cooling series is indicated by thicker lines of hydrogen and thinner lines of the metallic elements, through Sirius, to Arcturus and our sun, thence to 19 Piscium, which shows chiefly flutings of carbon, with a few faint metallic lines. The process of further cooling brings us to the dark stars. We have here a complete scheme of evolution, carrying us from those ill-defined but enormously diffused masses of gas and cosmic dust we know as nebulæ, through planetary nebulæ, nebulous stars, variable and double-stars, to red and white stars and on to those exhibiting the most intense blue-white lustre. We must remember, however, that the most brilliant of these stars, showing a gaseous spectrum and forming the culminating point of the ascending series, are not necessarily hotter than, or even so hot as, some of those far down on the descending scale; since it is one of the apparent paradoxes of physics that a body may become hotter during the very process of contraction through loss of heat. The reason is that by cooling it contracts and thus becomes denser, that a portion of its mass falls towards its centre, and in doing so produces an amount of heat which, though absolutely less than the heat lost in cooling, will under certain conditions cause the reduced surface to become hotter. The essential point is, that the body in question must be wholly gaseous, allowing of free circulation from surface to centre. The law, as given by Professor S. Newcomb, is as follows:-- '_When a spherical mass of incandescent gas contracts through the loss of its heat by radiation into space, its temperature continually becomes higher as long as the gaseous condition is retained._' To put it in another way: if the compression was caused by external force and no heat was lost, the globe would get hotter by a calculable amount for each unit of contraction. But the heat lost in causing a similar amount of contraction is so little more than the increase of heat produced by contraction, that the slightly diminished total heat in a smaller bulk causes the temperature of the mass to increase.