ing upon the liquid surface. Craters (Figures 15 and 16) were formed in the same manner as the first one, excepting that after extinguishing the lamp the tidal action was brought into play, alternately pumping the liquid up to the rim of the crater, where it partially solidified, leaving a little ring of solid parffine, and then drawing it down again into the interior, where it soon partly remelted, preparatory to a renewed elevation. This tidal action was continued until the fluid became quite viscous, solidifying into little hills and ridges inside the crater, and later as the hardening surface was dragged out of shape by the pumping of the liquid below it, little cracks were formed around the edges and across the bottom of the crater, like the rills seen in similar situations upon the Moon.

If now, we raise the piston high enough, and wait for air to get underneath it, we may force this down into the melted paraffine. The result is an explosion, in which the paraffine may shoot up several feet in the air. If care is taken, however, the jets may be confined to the height of a few inches. A cone is soon formed (Figure 17), and the liquid paraffine trickles down the slopes in miniature lava streams. As the cooling process goes on the paraffine comes out in bubbles, like soap suds, which break and rapidly build up the cone. If the process is continued further, partially solid lumps of paraffine are projected into the air, falling down upon the outer slopes of the cone. The crater now gradually narrows, and if care is not taken will soon become clogged. With care many well known volcanic phenomena may be repeated, such, for instance, as the shifting of crater to one side, and the formation of a succession of crater rings and semi-circles. Also the bursting out of new craters near the base of the original

cone. Indeed, the investigator is likely to perform this experiment involuntary, if he permits the main vent to get partially clogged, and applies too much heat below. The introduction of air seems to transport us at once from lunar to terrestrial scenes, although in the case of the Earth the tides of course have nothing to do with the matter, their place being taken by irregularly recurring explosions of steam.

Applying now the results of our experiments to the case of the Moon, we may conceive that the order of formation was somewhat as follows. We will start with the Moon in the form of a liquid or viscous sphere, revolving about the Earth, and close to its surface. Under these circumstances the tides would be of enormous power, and quite unlike in magnitude anything at present existing upon the Earth. Those constituents of the Moon having the least specific gravity would float upon the surface and soon solidify, forming a thin crust. Whether this occurred before or after the solidification of the central core through pressure is of no consequence. As it solidified the crust would contract, forming cracks, which would be enlarged at points into circular holes or craters by the hot liquid interior, while the remaining portions of the crack would become filled with fluid from the interior, which would slowly harden and become part of the continuous surface. We have an illustration this very process arrested in the act, in the case of the great rill of Hyginus, and the small craters distributed along its length. Hyginus is probably a later formation, however, as, if the crust had been thin, the process would have been completed, the craters enlarged, and the rill filled up. When the process first began, numerous comparatively small holes would form one after another. These holes would continue to enlarge, retaining their circular form, as the hot liquid was

forced through them, until the action was stopped by a sufficiently thick crust forming upon the liquid surface. In the mean time the tremendous tides engendered by our Earth, coursing through the imprisoned liquid interior, would fracture the thin and brittle crust in fresh places, where the same process would be repeated. When the crust was thin the enlargement of the crater would proceed rapidly, and the aperture might attain considerable dimensions before the restraining crust was formed, but as the original crust thickened, and the passage connecting the aperture with the liquid interior lengthened, we should find that the craters formed would be smaller, but more numerous. We should thus expect, in general, that the older the crater the larger it would be, and that the smaller craters would impinge upon the larger ones, and not vice versa. An examination of the lunar surface shows this to be the case. The older and larger craters, like Clavius, Albategnius, and many others near the south pole, are pitted and sometimes almost concealed by numerous smaller and later craters, while craters of more moderate size, like Tycho and Copernicus, are comparatively free from such intrusions.

It can be shown that the maximum surface tension exerted by the Earth upon the Moon is exerted upon the great circle forming the limb, and tends to separate the two hemispheres with a force which at the mean distance of the two bodies amounts to a tension of 9.6 pounds on the square inch. When the Moon was at one-tenth of its present distance from the Earth, this tension would have been one thousand times as great, and would have been sufficient to shatter it to pieces had it then existed in the solid form. If, however, it were fluid or viscous, as we have supposed, the effect would have been merely to produce an enormous tide, as has been shown by Professor Darwin. In the mean time, if the Moon revolved rapidly on its axis, so that all portions of its surface were presented successively to the Earth, this maximum strain would be felt successively by all portions of its surface, the tendency being to separate, or crack it, in a meridional direction. We should thus expect to find that the earlier formations would have a tendency to lie in lines in a north and south direction. This we find actually to be the case with the craters that we have been discussing, particularly the larger ones. This fact has been pointed out by Webb, Neison and others.

The craters of this early period, of which Copernicus is the characteristic example, would be moulded by the enormous tides

into forms resembling Figures 15 and 16. The interior surface of one crater, Wargentin, apparently solidified when the tide which filled it reached to its very rim. The aperture connecting it with the interior had in some way evidently become clogged, and the fluid which had formed the crater was caught as it were in the act, to serve as a clue and perpetual illustration of the process of construction to all future generations. Another crater, Mersenius, has a conspicuously convex interior. This was the case at first with the paraffine crater represented in Figure 15, but subsequent cooling caused it to become concave. If the floors of the lunar craters when they solidified were in general convex, it is evident that the subsequent solidification of the fluid beneath them would tend to make the floor level, thereby producing a compression of the surface, which might well result in the formation of a central peak or ridge. If the paraffine model had been constructed upon a larger scale, and the contraction of the fluid beneath it had been allowed to proceed more slowly, it is thought that this result might have been obtained. As it was, a tendency to form small ridges was noticed. In Figure 16 the internal surface of the crater was artificially broken, thereby producing the central mound.

As the cooling process continued, regions deeper down solidified and contracted. The upper layers, having now become completely solid, would not contract at the same rate, with the fall of temperature, and the result would be that the surface of the Moon, instead of being too small, would now be too large for its interior. When this "critical epoch" occurred, the formation of craters would for a time almost cease. The result would be a local subsidence with a considerable local evolution of heat, although the temperature of the Moon as a whole would still continue to diminish. If the heat so developed were sufficient to overcome the latent heat of solidification a considerable portion of the subsident area might be melted, while portions of the original crust carrying their ancient craters with them, would sinkslowly beneath the liquid surface, the process of destruction continuing as long as the supply of heat lasted.

In the mean time the Moon would have receded much farther from the Earth, and the tides would have accordingly greatly diminished in their intensity. The subsident areas would in general be large in extent, such as the Maria Imbrium, Serenitatis, and Crisium. The darker color of their floors would seem to indicate that they were formed from another kind of material, which, coming from a considerable depth, had united and mixed

with the lighter colored molten matter which had formed the original surface. In these maria we often see the outlines of old crater rings which have been partially melted down and absorbed in the subsequent eruption of melted matter from the interior. Since in all cases the melting progresses outward from a centre, we see why it is that these large seas, like the smaller craters, all retain the approximately circular shape. In some cases where the original crust has subsided, it has melted in the thinnest places only, such as the bottoms of the deepest craters. Thus Plato probably had originally an interior like that of Copernicus, but the melting process which destroyed the bottom was not carried far enough to ruin its walls also, as was partially done in the case of many of the older craters. The elevation of the surfaces of the maria probably indicates their relative age, the lower ones being formed last.

(TO BE CONTINUED.)

SPECTROSCOPIC NOTES.

The approaching total eclipse of the Sun May 28, 1900, will be elaborately observed, and should yield some valuable spectroscopic results. As the path of totality passes through such an accessible district in this country the number of well equipped expeditions should be unusually large, while with favorable weather there should be a host of general observers. With New Orleans, Mobile, Raleigh, and Norfolk in the path of totality, thousands can hope to see the corona with no greater inconvenience than going out of doors; while not a few will regard a view of the corona ample reward for traveling a considerable distance.

The probability of a clear sky at the time of totality, as shown by weather observations covering a number of years, varies considerably for different parts of the path The best prospect for good weather is in the highest elevation in northern Georgia and eastern Alabama, decreasing gradually toward the Gulf coast in one direction and the Atlantic coast in the other. The path of totality on land in this hemisphere is so long that even with a general storm it will be quite possible to have clear weather for a part of the path, and with this in view the extreme southwest should not be altogether neglected. Local fog or flying clouds may add interest and excitement. In case of partial cloudiness in any district observers should be scattered; for this, without diminishing the chances of any individual, increases vastly the probability that the eclipse will be seen by some at least of the party.

For the powerful spectroscopes of various types there will be work in plenty, in the accurate determination of the revised position of the corona line; possibly in the measurement of the velocity of rotation of the corona; and in the study of the flash spectrum at the instants of beginning and end of totality. With the most modest spectroscope the gaseous character of the corona may be shown, and the relative brightness of the weak continuous spectrum may be studied.

The European and African end of the path of totality, passing through Por