before. This branch finally terminates in the constellation Ophiuchus, while the eastern branch, growing narrower, can still be followed toward the south.

Between the equator and the southern horizon we have the brightest and most irregular regions of all. Several round, bright patches of greater or less intensity are projected on a background sometimes moderately bright and sometimes quite dark. If the night is quite clear and moonless we shall see that, after a vacant stretch, the western branch seems to recommence just about the constellation Scorpius. In this constellation we have again a bifurcation, a dark region between two bright ones.

This is about as far as the object can be well traced in our middle latitudes. From a point of view nearer to the equator it can be traced through its whole extent. South of Scorpius and Sagittarius it becomes broad, faint, and diffused through the constellations of Norma and Circinus. It reaches its farthest southern limit in the Southern Cross, where it becomes narrower and better defined. The most remarkable feature here is the "coal sack," a dark opening of elliptical shape in the central line of the stream. West and north of this, in the constellation Argo, is the broadest and most diffused part of the whole stream, the breadth reaching fully 30°. Here we again reach the portion which rises above our horizon.

Returning now to our starting-point, we shall notice that, as we make our observations later and later in the autumn, the southern part, which we have been mostly studying, is seen night by night lower

down in the west, while new regions are coming into view in the north-east and east. These regions rise earlier every evening, and, if we continue our observations to a later hour, we shall see more and more of them above the eastern or south-eastern horizon. By midwinter Cassiopeia will be seen in the north-west, and we can readily trace the course of the galaxy from that constellation in the opposite direction from that which we have been following. South of Cassiopeia we see, near the central line, the well-known cluster forming the sword-handle of Perseus. Farther south the belt grows narrower and fainter; although the irregularities of structure continue, they are far less striking than on the other side. On a On a moonlight evening it will scarcely be visible at all. If the moon is absent and the air clear we shall see that it grows slightly brighter toward the southern horizon, near which will be the narrowest part of its entire course. Below is the broad and diffused region in Argo already mentioned.

One conclusion from the inequalities of structure which we have described will be quite obvious. The Milky Way is something more than the result of the general tendency of the stars to increase in number as we approach its central line. There must be large local aggregations of stars, because, as we have already pointed out, there cannot be such diversity of structure shown in a view of a very widely stretched stratum of stars.

When, instead of a naked-eye view of the belt, we study the photographs of the Milky Way, we find

this evidence of clustering to grow still stronger. It is seen very strikingly in the photograph by Barnard showing the singular rifts in the Milky Way in the constellation Ophiuchus. Yet more singular are three small openings very close together in the constellation Aquila, the positions of which are:

(1) R. A. 19h. 35.0m.; Dec. = +10° 17'.

(2)

=

The fundamental question which we meet in our further study of this subject is: At what magnitude do these agglomerations of stars begin? Admitting, as we must, that they are local, are they composed altogether of faint stars, or do they also include the brighter stars within their limits? We consider this question in a way quite similar to that in which we discussed the clustering of the stars in the last chapter. We mark out on a map of the Milky Way the brightest regions-that is, those which include the densest agglomeration of very faint stars. We also mark out the darkest regions, including the coal sack. For this purpose I have taken the maps found in Heis's Atlas Celestis for the northern portion of the Milky Way and the Atlas of Gould's Uranometria Argentina for the southern portion. In order to enable anyone to repeat and verify the work I give the position of the central part of each patch or region studied. This serves simply for the purpose of indentification. The outlines can be drawn. by anyone when the patch is identified. In the