There are known at present twenty-two variables of the Algol type, that is, stars having each a dark companion very close to it which obscures it either wholly or partially during every revolution. In these cases the density of the systems can be approximately determined, and they are found to be, on the average, only one-fifth that of water, or one-eighth that of our sun. But as many of them are as large as our sun, or even considerably larger, it is evident that they must be wholly gaseous, and, even if very hot, of a less complex constitution than our luminary. Mr. A.W. Roberts tells us that five out of these twenty-two variables revolve _in absolute contact_ forming systems of the shape of a dumb-bell. The periods vary from twelve days to less than nine hours; and, starting from these, we now have a continuous series of lengthening periods up to the twin stars of Castor which require more than a thousand years to complete their revolution. During his observations of the above five stars, Mr. Roberts states that one, X Carinæ, was found to have parted company, so that instead of being actually united to its companion the two are now at a distance apart equal to one-tenth of their diameters, and he may thus be said to have been almost a witness of the birth of a stellar system. A year later we find the record (in _Knowledge_, October 1902) of Professor Campbell's researches at the Lick Observatory. He states that, out of 350 stars observed spectroscopically, one in eight is a spectroscopic binary; and so impressed is he with their abundance that, as accuracy of measurement increases, he believes that _the star that is not a spectroscopic binary will prove to be the rare exception_! Professor G. Darwin had already shown that the 'dumb-bell' was a figure of equilibrium in a rotating mass of fluid; and we now find proofs that such figures exist, and that they form the starting-point for the enormous and ever-increasing quantities of spectroscopic binary star-systems that are now known. The origin of these binary stars is also of especial interest as giving support to Professor Darwin's well-known explanation of the origin of the moon by disruption from the earth, owing to the very rapid rotation of the parent planet. It now appears that suns often subdivide in the same manner, but, owing perhaps to their intensely heated gaseous state they seem usually to form nearly equal globes. The evolution of this special form of star-system is therefore now an observed fact; though it by no means follows that all double stars have had the same mode of origin. CLUSTERS OF STARS AND VARIABLES The clusters of stars, which are tolerably abundant in the heavens and offer so many strange and beautiful forms to the telescopist, are yet among the most puzzling phenomena the philosophic astronomer has to deal with.