masses, and, as the shadow had not so much greater a length at that time, two months from opposition, as it had when the planet was much nearer opposition, he infers that the true explanation of the appearance has hardly been found. He appears to have overlooked the fact that the assumption made in the explanation is not that Jupiter has a semitransparent atmosphere always equally translucent and penetrable to the same depth by the solar rays. When the shadow was shorter than it should have been, had the atmosphere been in the same condition as when Mr. Brett made his observation, it is probable that a layer of clouds interrupted the rays, and thus the shadow was much closer to the cloud-mass throwing it than it would have been had that layer not been there, Mr, Knobel's paper contains very striking evidence of the variability of Jupiter's atmosphere, or rather of the clouds which float in it, "The greater distinctness of the satellites when near the edge," he says, "is a curious phenomenon which has been repeatedly observed by astronomers, but which seems to require explanation." On an occasion described "the second satellite transited a dark limb which was" (seemed) "most dark near the centre, and fainter toward the edge, yet the satellite was almost invisible when on the darkest part of the belt, and was bright and distinct when the background of the belt was faintest." This practically proved that on the occasion in question the dark, central part of the belt seemed darker than it really was by contrast with the bright belts on either side, while the edge seemed lighter than it really was by contrast with the dark sky on which the planet was projected. In reality the part near the edge must have been darker than the part near the middle, or the satellite could not have appeared brighter when near the edge. No doubt the darkness near the edge (which, by-the-way, my friend Mr. Browning tested photometrically, and demonstrated, at my suggestion, eight years ago) was due to transparency, the darkness of the sky beyond being to some degree discernible through the edge. But this transparency is not always to be observed to the same degree, or through the same extent of Jovian atmosphere as to depth. Mr. Knobel proceeds, illustrating this the more effectively that he does so unintentionally: "The third satellite, on March 25, 1874, appeared as a dark spot when in mid-transit, and on nearing the edge appeared as a bright spot without trace of duskiness. But on March 26, 1873" (observe the difference of years), "the fourth satellite made the whole transit as a dark spot, and was not perceptibly less dark at egress than in mid-transit."
It appears to me demonstrated by the evidence thus far noted that in a semi-transparent atmosphere of enormous depth, surrounding Jupiter, there float vast cloud-masses, sometimes in layers, at others in irregular heaps, at others having well-rounded forms. These cloud-masses undergo sometimes remarkable changes of shape, often forming or disappearing in a very short time, and thus indicating the infe-
