More striking still is the motion of the planetary nebulæ, in excess of 25 kilometers per second, while type A stars move 11 kilometers, type G 15 kilometers, and type M 17 kilometers per second. Can the law connecting speed of motion and spectral type be so general that the planetary nebula is to be regarded as the final evolutionary stage? Stars have been seen to become nebulæ, and one astronomer at least is strongly of the opinion that a single such instance ought to outweigh all speculation to the contrary, as that stars originate from nebulæ. In his discussion of stellar proper motions, Boss has reached a striking confirmation of the relation of speed to type, finding for the cross linear motion of the different types a series of velocities closely paralleling those of Kapteyn and Campbell. Concerning the marked relation of the luminosities of the stars to their spectral types, there is a pronounced tendency toward equality of brightness among stars of a given type; also the brightness diminishes very markedly with advance in the stage of evolution. There has been much discussion as to the order of evolution as related to the type of spectrum, and Russell of Princeton has put forward the hypothesis of giant stars and dwarf stars, each spectral type having these two divisions, though not closely related. One class embraces intensely luminous stars, the other stars only feebly luminous. When a star is in process of contraction from a diffused gaseous mass, its temperature rises, according to Lane's law, until that density is reached where the loss of heat by radiation exceeds the rise in temperature due to conversion of gravitational energy into heat. Then the star begins to cool again. So that if the spectrum of a star depends mainly on the effective temperature of the body, clearly the classification of the Draper catalogue would group stars together which are nearly alike in temperature, taking no note as to whether their present temperature is rising or falling. Another classification of stars by Lockyer divides them according to ascending and descending temperatures. Russell's theory would assign the succession of evolutionary types in the order, M_{1}, K_{1}, G_{1}, F_{1}, A_{1}, B, A_{2}, F_{2}, G_{2}, K_{2}, M_{2}, the subscript 1 referring to the "giants," and 2 to the dwarf stars. In large part the weight of evidence would appear to favor the order of the Harvard classification, independently confirmed as it is by studies of stellar velocities, Galactic distribution, and periods of binary stars both spectroscopic and visual, where Campbell and Aiken find a marked increase in length of period with advance in spectral type. At the same time, a vast amount of evidence is accumulating in support of Russell's theory. Investigations in progress will doubtless reveal the ground on which both may be harmonized.