Still more remarkable than the irregular clusters are those which condense into a more or less globular form. There are not very many objects of this class in the northern sky visible with a small telescope, but the beauty of those which are visible is very notable. The most splendid of all is the famous cluster M. 13 Herculis. (The M. in these cases refers to the catalogue of such objects drawn up by Messier, the French 'comet ferret,' to guide him in his labours.) M. 13 is situated almost on the line between Zeta and Eta Herculis, and at about two-thirds of the distance from Zeta towards Eta. It is faintly visible to the unaided eye when its place is known, and, when viewed with sufficient telescopic power, is a very fine object. Nichol's remark that 'perhaps no one ever saw it for the first time through a large telescope without uttering a shout of wonder' seems to be based on a somewhat extravagant estimate of the enthusiasm and demonstrativeness of the average star-gazer; but the cluster is a very noble object all the same, consisting, according to a count made on a negative taken in 1899, of no fewer than 5,482 stars, which condense towards the centre into a mass of great brilliancy. It takes a large aperture to resolve the centre of the cluster into stars, but even a 3-inch will show a number of twinkling points of light in the outlying streamers (Plate XXIX.). In the same constellation will also be found the cluster M. 92, similar to, but somewhat fainter than M. 13; and other globular clusters are noted in the Appendix. Most of these objects, however, can only be seen after a fashion with small instruments. Of the true nature and condition of these wonderful aggregations we are so far profoundly ignorant. The question of whether they are composed of small stars, situated at no very great distance from the earth, or of large bodies, which are rendered faint to our vision by immense distance, has been frequently discussed. Gore concludes that they are 'composed of stars of average size and mass, and that the faintness of the component stars is simply due to their immense distance from the earth.' If so, the true proportions of some of these clusters must be indeed phenomenal! A very remarkable feature to be noticed in connection with some of them is the high proportion of variable stars which they contain. Professor Bailey has found that in such clusters as M. 3 and M. 5 the proportion of variables is one in seven and one in eleven respectively, while several other groups show proportions ranging from one in eighteen up to one in sixty. As the general proportion of variables is somewhere about one in a hundred, these ratios are remarkable. They only characterize a certain number of clusters, however, and are absent in cases which seem strictly parallel to others where they exist. [Illustration: PLATE XXIX. Cluster M. 13 Herculis. Photographed by Mr. W. E. Wilson.]