and showing around the limb. After a careful study of the Moon he thought he had discovered several such streaks. One of these is perhaps shown on Plate 3B [2.0, 2.4]. The streaks surrounding the different craters are not all alike. Most of them belong to the class so conspicuously represented by those surrounding Tycho. The streaks surrounding Copernicus, on the other hand, differ from them in several important respects. In the first place, the streaks about Copernicus are yellowish, while those about Tycho are white. Secondly, the streaks about Copernicus are somewhat curved and branching, while the others are straight. Thirdly, the streaks about Copernicus issue in general radially from the crater, while those about Tycho issue tangentially to the crater's rim. Fourthly, the crater of Tycho is surrounded by a dark halo, most conspicuous at lunar noon (Plate 10C); nothing of the kind is seen about Copernicus. Fifthly, under favourable atmospheric conditions we can distinguish the craterlets from which the streaks about Tycho take their origin. If such craterlets exist about Copernicus, they cannot in general be seen, except on the very rim of the crater itself, when under suitable conditions they are fairly conspicuous. While the streaks about Tycho are in general much the brighter, yet when they cross a mare, as do those radiating from Copernicus, the difference in brilliancy of the two systems is not very great.
That the background should affect the brilliancy of the streaks of the Tycho system indicates either that the material causing the streaks is partially transparent or that it occurs in small, separate areas between which the background is seen. The latter suggestion seems on the whole the more probable, and tends to confirm the accuracy of the explanation of the streaks already advanced.
Kepler, 13C [1.9, 6.5], differs from the two craters already mentioned in being surrounded by a bright continuous halo, with one or two dark spots on the northeastern side. This halo retains its brilliancy under nearly all illuminations.
The cause of the dark halo surrounding Tycho at full moon is unknown. If we compare the five plates, 10A, B, C, D and E, we shall see that there is a progressive darkening, reaching a maximum at noon, and then fading out again. The effect reminds us somewhat of Linné. But if the snow melts, which certainly seems to be the case, why should it occur only upon the slopes of the crater? Possibly the effect is analogous to the variable spots described in the next chapter, which always darken toward noon. In attempting to give a satisfactory explanation of the shape of the bright streaks, we may start with the observed fact that those craterlets producing streaks which are
