There is one striking feature of the globular clusters which calls for notice; the presence in some of them of enormous quantities of variable stars, while in others few or none can be found. The Harvard Observatory has for several years devoted much time to this class of observations, and the results are given in Professor Newcomb's recent volume on 'The Stars.' It appears that twenty-three clusters have been observed spectroscopically, the number of stars examined in each cluster varying from 145 up to 3000, the total number of stars thus minutely tested being 19,050. Out of this total number 509 were found to be variable; but the curious fact is, the extreme divergence in the proportion of variables to the whole number examined in the several clusters. In two clusters, though 1279 stars were examined, not a single variable was found. In three others the proportion was from one in 1050 to one in 500. Five more ranged up to one in 100, and the remainder showed from that proportion up to one in seven, 900 stars being examined in the last mentioned cluster of which 132 were variable! When we consider that variable stars form only a portion, and necessarily a very small proportion, of binary systems of stars, it follows that in all the clusters which show a large proportion of variables, a very much larger proportion--in some cases perhaps all, must be double or multiple stars revolving round each other. With this remarkable evidence, in addition to that adduced for the prevalence of double stars and variables among the stars in general, we can understand Professor Newcomb adding his testimony to that of Professor Campbell already quoted, that 'it is probable that among the stars in general, single stars are the exception rather than the rule. If such be the case, the rule should hold yet more strongly among the stars of a condensed cluster.' THE EVOLUTION OF THE STARS So long as astronomers were limited to the use of the telescope only, or even the still greater powers of the photographic plate, nothing could be learnt of the actual constitution of the stars or of the process of their evolution. Their apparent magnitudes, their movements, and even the distances of a few could be determined; while the diversity of their colours offered the only clue (a very imperfect one) even to their temperature. But the discovery of spectrum analysis has furnished the means of obtaining some definite knowledge of the physics and chemistry of the stars, and has thus established a new branch ofscience--Astrophysics--which has already attained large proportions, and which furnishes the materials for a periodical and some important volumes. This branch of the subject is very complex, and as it is not directly connected with our present inquiry, it is only referred to again in order to introduce such of its results as bear upon the question of the classification and evolution of the stars.