While some clusters, like those just described, are very easily resolvable into stars with the smallest instruments, others yield with the greatest difficulty, even to the largest telescopes, in which their starry nature is barely suspected. Owing to this peculiarity, star-clusters are usually divided into two principal classes. In the first class are comprised all the clusters which have been plainly resolved into stars, and in the second all those which, although not plainly resolvable with the largest instruments now at our disposal, show a decided tendency to resolvability, and convey the impression that an increase of power in telescopes is the only thing needed to resolve them into stars. Of course this classification, which depends on the power of telescopes to decide the nature of these objects, is arbitrary, and a classification based on spectrum analysis is now substituted for it. The star-clusters are also divided into globular and irregular clusters, according to their general form and appearance. The globular clusters, which are the most numerous, are usually well-defined objects, more or less circular in their general outlines. The rapid increase of brightness towards their centres, where the stars composing them are greatly condensed, readily conveys the impression that the general form of these sparkling masses is globular. The irregular clusters are not so rich in stars as the former. Usually their stars are less condensed towards the centre, and are, for the most part, so loosely and irregularly distributed, that it is impossible to recognize the outlines of these clusters or to decide where they terminate. The globular clusters are usually quite easily resolvable into stars, either partly or wholly, although some among them do not show the least traces of resolvability, even in the largest instruments. This may result from different causes, and may be attributed either to the minuteness of their components or to their great distance from the Earth, many star-clusters being at such immense distances that they are beyond our means of measurement. As has been shown in the preceding section, the star-clusters are found in great number in the Galaxy; indeed, it is in this region and in its vicinity that the greater portion of them are found. In other regions, with the exception of the Magellanic clouds, where they are found in great number and in every stage of resolution, the clusters are few and scattered. [PLATE XIV.--STAR-CLUSTER IN HERCULES. From a study made in June, 1877]