[The foregoing passages, to most of which I do not commit myself as more than possibilities, I republish chiefly as introductory to the following speculation, which, since it was propounded in 1865, has met with some acceptance.]
"But what of the spots commonly so called?" it will be asked. In the essay on the Nebular hypothesis, above quoted from, it was suggested that refraction of the light passing through the depressed centres of cyclones in this atmosphere of metallic gases, might possibly be the cause; but this, though defensible as a "true cause," appeared on further consideration to be an inadequate cause. Keeping the question in mind, however, and still taking as a postulate the conclusion of Sir John Herschel, that the spots are in some way produced by cyclones, I was led, in the course of the year following the publication of the essay, to an hypothesis which seemed more satisfactory. This, which I named at the time to Prof. Tyndall, had a point in common with the one afterward published by Prof. Kirchhoff, in so far as it supposed cloud to be the cause of darkness; but differed in so far as it assigned the cause of such cloud. More pressing matters prevented me from developing the idea for some time; and, afterwards, I was deterred from including it in the revised edition of the essay, by its inconsistency with the "willow-leaf" doctrine, at that time dominant. The reasoning was as follows:—The central region of a cyclone must be a region of rarefaction, and, consequently, a region of refrigeration. In an atmosphere of metallic gases rising from a molten surface, and presently reaching a limit at which condensation takes place, the molecular state, especially toward its upper part, must be such that a moderate diminution of density, and fall of temperature, will cause precipitation. That is to say, the rarefied interior of a solar cyclone will be filled with cloud: condensation, instead of taking place only at the level of the photosphere, will here extend to a great depth below it, and over a wide area. What will be the characters of a cloud thus occupying the interior of a cyclone? It will have a rotatory motion; and this it has been seen to have. Being funnel-shaped, as analogy warrants us in assuming, its central parts will be much deeper than its peripheral parts, and therefore more opaque. This, too, corresponds with observation. Mr. Dawes has discovered that in the middle of the spot there is a blacker spot: just where there would exist a funnel-shaped prolongation of the cyclonic cloud down toward the Sun's body, the darkness is greater than elsewhere. Moreover, there is furnished an adequate reason for the depression which one of these dark spaces exhibits. In a whirlwind, as in a whirlpool, the vortex will be below the general level, and all around, the surface of the medium will descend toward it. Hence a spot seen obliquely, as when carried toward the Sun's limb, will have its umbra more and more hidden, while its penumbra still remains visible. Nor are we without some interpretation of the penumbra. If, as is implied by what has been said, the so-called "willow-leaves," or "rice-grains," are the tops of the currents ascending from the Sun's body, what changes of appearance are they likely to undergo in the neighbourhood of a cyclone? For some distance round a cyclone there will be a drawing in of the superficial gases toward the vortex. All the luminous spaces of more transparent cloud forming the adjacent photosphere, will be changed in shape by these centripetal currents. They will be greatly elongated; and there will so be produced that "thatch"-like aspect which the penumbra presents.
[The explanation of the solar spots above suggested, which was originally propounded in opposition to that of M. Faye, was eventually adopted by him in place of his own. In the Comptes Rendus for 1867, Vol. LXIV., p. 404, he refers to the article in the Reader, partly reproduced above, and speaks of me as having been replied to in a previous note. Again in the Comptes Rendus for 1872, Vol. LXXV., p. 1664, he recognizes the inadequacy of his hypothesis, saying:—"Il est certain que l'objection de M. Spencer, reproduit et développée par M. Kirchoff, est fondée jusqu'à un certain point; l'intérieur des taches, si ce sont des lacunes dans la photosphère, doit être froid relativement. … Il est donc impossible qu'elles proviennent d'éruptions ascendantes." He then proceeds to set forth the hypothesis that the spots are caused by the precipitation of vapour in the interiors of cyclones. But though, as above shown, he refers to the objection made in the foregoing essay to his original hypothesis, and recognizes its cogency, he does not say that the hypothesis which he thereupon substitutes is also to be found in the foregoing essay. Nor does he intimate this in the elaborate paper on the subject read before the French Association for the Advancement of Science, and published in the Revue Scientifique for the 24th March 1883. The result is that the hypothesis is now currently ascribed to him.[26]
About four months before I had to revise this essay on "The Constitution of the Sun," while staying near Pewsey, in Wiltshire, I was fortunate enough to witness a phenomenon which furnished, by analogy, a verification of the above hypothesis, and served more especially to elucidate one of the traits of solar spots, otherwise difficult to understand. It was at the close of August, when there had been a spell of very hot weather. A slight current of air from the West, moving along the line of the valley, had persisted through the day, which, up to 5 o'clock, had been cloudless, and, with the exception now to be named, remained cloudless. The exception was furnished by a strange-looking cloud almost directly overhead. Its central part was comparatively dense and structureless. Its peripheral part, or to speak strictly, the two-thirds of it which were nearest and most clearly visible, consisted of converging streaks of comparatively thin cloud. Possibly the third part on the remoter side was similarly constituted; but this I could not see. It did not occur to me at the time to think about its cause, though, had the question been raised, I should doubtless have concluded that as the sky still remained cloudless everywhere else, this precipitated mass of vapour must have resulted from a local eddy. In the space of perhaps half-an-hour, the gentle breeze had carried this cloud some miles to the East; and now its nature became obvious. That central part which, seen from underneath, seemed simply a dense, confused part, apparently no nearer than the rest, now, seen sideways, was obviously much lower than the rest and rudely funnel-shaped—nipple-shaped one might say; while the wide thin portion of cloud above it was disk-shaped: the converging streaks of cloud being now, in perspective, merged together. It thus became manifest that the cloud was produced by a feeble whirlwind, perhaps a quarter to half-a-mile in diameter. Further, the appearances made it clear that this feeble whirlwind was limited to the lower stratum of air: the stratum of air above it was not implicated in the cyclonic action. And then, lastly, there was the striking fact that the upper stratum, though not involved in the whirl, was, by its proximity to a region of diminished pressure, slightly rarified; and that its precipitated vapour was, by the draught set up towards the vortex below, drawn into converging streaks. Here, then, was an action analogous to that which, as above suggested, happens around a sun-spot, where the masses of illuminated vapour constituting the photosphere are drawn towards the vortex of the cyclone, and simultaneously elongated into striæ: so forming the penumbra. At the same time there was furnished an answer to the chief objection to the cyclonic theory of solar spots. For if, as here seen, a cyclone in a lower stratum may fail to communicate a vortical motion to the stratum above it, we may comprehend how, in a solar cyclone, the photosphere commonly fails to give any indication