among such systems, that even among the host of triple stars already observed by astronomers, not two have been found which closely resemble each other in arrangement, while, so soon as we pass to more complex systems, we find that each fresh object of the class differs utterly from all which have been previously observed.

In considering the actual condition of the region of space occupied by a triple, quadruple, or multiple starsystem, we find ourselves surrounded on all sides by sources of perplexity-so long, at least, as we compare the worlds in such regions with our own earth, or with any member of the Solar system. All the most marked characteristics of life on our earth must be wanting in those worlds which circle around the component suns of multiple systems. There can be no year, strictly so called, no orderly succession of seasons, no regular alternation of day and night, in many cases no night at all, save for brief periods at exceedingly long intervals. Placed at such a distance from any star of one of these systems that that star appears as a sun, the others also must supply an amount of light sufficient of itself to banish night. More commonly, indeed, each star of such a system, while above the horizon, would be a sun shining more brightly than our sun does to the inhabitants of many planets of his scheme. But where there are three or four such suns, the simultaneous absence of all from the sky must be an uncommon event-as uncommon, for instance, as those occasions when none of Jupiter's satellites can be

seldom witness the spectacle of the starlit sky; and the study of any orbs beyond their own system must be a task of infinite difficulty, since it can only be pursued for a few occasional hours of darkness, separated by many months of persistent daylight.

The consideration of these multiple systems leads us on, step by step, to the various orders of star-clusters. For we can point to multiple systems of greater and greater richness, and as it were compactness, until we arrive at orders which we are compelled to regard as veritable starclusters. Yet it is to be noticed that. we might have approached the study of star-clusters in a different direction. For we have already had occasion to consider the various degrees of stellar aggregation in different parts of the heavens; and some of the regions of aggregation, while indubitably features of the galaxy regarded as a whole, are yet so well defined, and so clearly separated from the surrounding more barren regions, that we cannot refuse to regard them as vast star-clusters. We can thus either proceed from the smaller star-clusters—to which we have been led by the study of multiple star-systeins— onwards to larger and yet larger groups of stars, until we have arrived at the aggregations just mentioned; or we can pass from these through the successive orders of a diminishing scale until we arrive at multiple star-systems.

Yet here it is well to remark that a difficulty presents itself which can only be removed by the theory, to which we have been already led by other considerations, that the great aggregations of stars are not (or at least, all of

them are not) to be regarded as formed of orbs necessarily comparable with Sirius and Arcturus, Capella, Vega, and Aldebaran, in real magnitude and splendour. For we can pass from the single suns onwards to double suns, multiple systems, star-clusters, and stellar aggregations, thence to less and less condensed and more and more extended aggregations, until we arrive at unaggregated star-groups, consisting, in fact, of the single orbs with which we set out. Now it is to be observed that we seem to be brought to the single stars by a course which is not a mere retracing of our steps. For supposing we regard star-clusters as intermediate between the least condensed aggregations on the one hand, and single stars on the other, we pass onwards from these least condensed aggregations to single stars, without going through any of the steps of our progress towards the aggregations. Obviously there can have been no true progression here. And we are compelled to believe that by following the course indicated, we arrive at quite another order of star-groups than those which form our constellations, although in appearance the less densely aggregated star-groups may resemble systems of true suns, like Sirius and Arcturus, Aldebaran and Capella.

But the fact is that, again and again, as we contemplate the wonders of the star-depths, we find ourselves thus tracing out perplexing sequences, one class of objects merging into another class seemingly altogether distinct, and this class into yet another, until we are bewildered by the multiplicity of analogies whereof some at least must

For instance, I have spoken of the various orders of star-clusters by which we may be led to the vast stellar aggregations; and there can be no question whatever that an apparently perfect sequence can be traced from sharplydefined clusters such as the magnificent object '13 Messier' in the constellation Hercules, to such groups as the Pleiades, or Præsepe in Cancer, or the splendid star-clusters near the sword-hand of Perseus, these groups being undoubtedly mere condensations in rich star-regions. It also cannot be doubted that we can pass from such a cluster as the one in Hercules, to others less rich in numbers, but equally compact, and so to clusters continually poorer and poorer (numerically) until we arrive at mere multiple systems, and so to quadruple, triple, double, and single stars. But it is equally certain that we can pass from the globular star-clusters to others oval in shape, and more and more closely set,* until at length we arrive at the nebulæ, properly so called—that is, spots of cloudy light which have not been resolved into separate stars by any telescopic power yet applied. Here, then, a progression as real as either of the preceding seems to lead us to objects which have been commonly regarded as wholly distinct from any portions of the galactic system, and probably analogues of the whole of that system.

* The connection between shape and closeness of star-setting is certain, though most perplexing. Sir John Herschel writes:-'It may be generally remarked, as a fact undoubtedly connected in some very intimate manner with the dynamical conditions of their subsistence, that the elliptic nebulæ are, for the most part, beyond comparison more difficult of resolution than those of globular form.

But it may be urged that this progression may relate simply to distance, and that therefore we need not regard it as forming a new and distinct sequence. To illustrate the matter, suppose that we could recognize among the companies of persons proceeding along a road,—many different kinds of groups-and that we arranged these different orders into a perfect sequence; then, taking any given order, we should find the various groups belonging to this order presenting different aspects according to their distance. Say the order comprised sets of persons travelling six together; then a set of six close by would differ in appearance from a set of six far away; and we might form many sets of six at different distances into a perfect sequence, according to their varying appearance.

Now according to this view of the matter the various orders of regularly-shaped nebulæ, even down to those which no telescope can resolve, would be star-clusters lying at great distances. And since the star-clusters, properly so called, must be regarded as belonging to our own galactic system, it would follow that all the orders of nebulæ belong to that system. We should at least find it it very difficult to say up to what point this complete sequence of objects belonged to our star-system, and where external star-systems began to be in question) and still less easily could we explain how complete external systems should thus be linked, as far as appearances extend, with mere portions, and relatively minute portions, of our own star-system.