have sought the central regions; and let it further be admitted that the presence of large quantities of hydrogen outside the visible surface of the sun points in the same direction: though, in reality, neither of these facts can be regarded as affording satisfactory evidence in this respect. Yet once more, let it be granted that the small densities of the four large outer planets, as compared with those of the four small inner planets, show that in the outermost parts of the solar system the lighter elements are superabundant, and the heavier elements superabundant in the innermost; though all the evidence available, as well as all the theoretical considerations admissible, point to a totally different explanation of this characteristic feature of the solar system. Do these arguments-even if admitted to be valid pro tanto, when considered separately-establish the theory in question? Do they not, when considered together, completely overthrow it? We see that in the sun, at the very centre of the whole system, hydrogen, the most tenuous of all the elements, exists in enormous quantities, even if we assume that the hydrogen outside the visible surface of the sun represents all or nearly all the solar hydrogen, though this is not only incredible in itself, but is disproved by several observed factsas, for instance, by the way in which hydrogen is sometimes ejected to enormous heights from the sun in the form of jet prominences, and still more by the widening of the hydrogen lines in the spectrum of a sun-spot (showing that at the lower level of the spot hydrogen is greater in quantity than at the visible surface of the sun). Again, we see that, though the earth is so much nearer the centre of the system than the giant planets (which travel at distances exceeding hers from five to thirty times), and though she might be regarded as at the centre of the region, originally some millions of miles in diameter, out of which her own mass and the moon's were formed, the lightest of all the elements is one of the most important constituents of the earth's mass. According to the theory we are considering, there should be but a small proportion of hydrogen in the total mass of the earth, and scarcely any at all (relatively) in the sun's mass.

But whether these considerations overthrow the theory above described, or not, one point must in any case be admitted. Since hydrogen is certainly present in enormous quantities in the sun, oxygen, being a denser element and an even more important one if we can judge from the relative quantity of hydrogen and of oxygen in the earth, ought assuredly to be also present in the sun, and in even greater quantity than hydrogen. Both considerations would point to this conclusion if the theory above dealt with were admitted; the latter alone points to it if that theory be rejected. But in either

case this conclusion must be accepted unless we are prepared to abandon the opinion, now admitted by every astronomer and physicist whose opinion is of weight, that the solar system has reached its present condition by processes of development.

Yet among the most striking results of the spectroscopic investigation of the sun's structure was this, that no trace could be recog nised of the presence of oxygen. The dark lines indicating the absorptive action of hydrogen were there, those corresponding to sodium, to iron, to magnesium, and to a number of other elements, many of which might have been expected to be present only in quantities so small that no trace of their presence could be discerned. A strange and probably significant circumstance respecting the elements thus detected is that they are all, except hydrogen, metals. Even hydrogen appears to be a metallic element, though (as I have said. elsewhere1) the idea may seem strange to those who regard hardness, brightness, malleability, ductility, plasticity, and the like, as the characteristic properties of metals, and necessarily fail to comprehend how a gas far rarer, under the same conditions, than the air we breathe, and which cannot possibly be malleable, ductile, or the like, can conceivably be regarded as a metal. But, as I there pointed out, there is in reality no necessary connection between any one of these properties and the metallic nature; many of the fifty-five metals are wanting in all of these properties; nor is there any reason why, as we have in mercury a metal which at ordinary temperatures is a liquid, so we might have in hydrogen a metal which at all obtainable temperatures, and under all obtainable conditions of pressure, is gaseous. Since I thus wrote, however, gas has been liquefied by Pictet and Cailloux (independently of each other); and in one experiment the liquid spray obtained as a jet of hydrogen rushed out of a minute aperture, seems to have been converted into a sort of fine hydrogen hail, which struck with a metallic ring against the glass walls of the receiving vessel. As to the metallic character of hydrogen, however, the most satisfactory evidence we have is that obtained in the experiments of

See the first essay in my "Pleasant Ways in Science." I may take this opportunity of noting that, while I in no single case have ever written two essays presenting the same subject in the same way (that is, for the same purpose, carrying on the history to the same point, and so forth) on the same scalethough often enough giving a full account of a subject in one essay and a condensed account in another-I have not unfrequently found it absolutely essential to the due discussion of a subject to explain certain matters already dealt with elsewhere. In such a case it seems mere affectation to modify the verbiage. As Humboldt has remarked, I think, in the preface to his "Cosmos," the practice of the ancients in this respect-the repetition of the same words-is invariably to be preferred to any arbitrary substitution of paraphrases.

the late Professor Graham (aided by Dr. Chandler Roberts), by which it was shown that hydrogen will enter into such combination with the metal palladium that it may be regarded as forming for the time an alloy of hydrogen and palladium, whence (since alloys can only be regarded as compounds of two or more metals) it would seem to follow that hydrogen is in reality a metallic element. Thus, then, we have this strange result—that metallic vapours only rise above the visible surface of the sun, seeing that the dark absorption lines of such vapours only (the missing tints, that is, which indicate the absorptive action of such vapours) are alone indicated in the solar spectrum. In passing, we may note that several of the elements thus indicated are not such as we should expect to find thus remote from the sun's centre if the theory were true that the denser elements are superabundant towards the centre, the lighter towards and above the surface, of the sun.

Oxygen and nitrogen showed no trace of their presence, nor could any trace be found of the non-metallic elements, carbon, sulphur, boron, silicon, and so forth.

At that time the idea entertained respecting the sun's constitution, and definitely adopted by Kirchhoff, Bunsen, Roscoe, and others, was simply this-—that the sun, or at any rate his visible surface, consists of glowing solid or liquid matter, while around his visible globe there is a deep complex atmosphere in which the vapours of the metals above referred to are present in greater or less quantity. Kirchhoff, indeed, would not allow that even the solar spots are to be regarded as regions where this state of things is modified. He adopted the opinion that all the phenomena of the spots are due to the presence of clouds floating in the solar atmosphere. Thus we should have the solar spectrum explained as follows :-The rainbowtinted background is formed by the light coming from the solid or liquid surface of the photosphere. Were there no vaporous envelope, the rainbow-tinted streak would be as perfectly continuous as a strip from a rainbow (athwart its breadth); but the various vapours forming the solar atmospheres cut out from the spectrum those special tints over which each of them has absorptive influence, in such sort

In general appearance only would such a strip resemble a continuous spectrum, such as we should have from sunlight in the case supposed; for the latter spectrum would be pure, each tint colouring its own part, and its own part only, of the spectrum. The rainbow spectrum is impure, each tint colouring a considerable part of the rainbow's breadth. No one, indeed, who has ever seen either the solar spectrum, or a pure continuous spectrum, can fail to see how faint are the colours of the rainbow compared with the splendid hues of a pure spectrum.

that tens of thousands of tints are missing from those which constitute the gradations (infinite in number) of the seven prismatic colours. Of course each missing tint is represented by a dark line athwart the breadth of the long rainbow-tinted spectrum.

But since 1859, when Kirchhoff announced his grand discovery of the significance of the solar dark lines, a series of minor (because less general) but yet most important discoveries have entirely modified our views as to the constitution of the sun, and the interpretation of the solar spectrum.

First among these must be noted the recognition of bright lines in the solar spectrum-a discovery which must be clearly understood and carefully studied by those who would appreciate justly the evidence respecting the presence of oxygen in the sun. We have said that the dark lines of hydrogen are present in the solar spectrum. This is always the case when the light of the whole sun is examined (as, for instance, when with a spectroscope we examine the light of the sky), and ordinarily it is the case also when the light of special parts of the sun's surface is thus examined. But, in the latter case, it happens at times that the dark lines of hydrogen cannot be recognised; at other times that the lines of hydrogen are brighter than the rainbow-tinted background of the spectrum; and at yet other times that they are bright but broad, while on these broad lines can sometimes be seen the fine dark lines of hydrogen. The explanation of these varied appearances, according to the principles resulting from the general laws of radiation and absorption, would be as follows: Whereas ordinarily the hydrogen above the sun's surface, though, of course, intensely hot, is cooler than that surface itself, thus absorbing more light of its own special tints than it emits, and so producing dark lines, the hydrogen over special regions becomes at times of the same temperature as the surface, in which case no lines of hydrogen can be seen; at times hotter, in which case bright hydrogen lines are seen; and at times not only hotter than the solar surface, but to such degree increased in density as to have its bright lines widened into bands. When, in the last of these cases, there exists a sufficient quantity of relatively cool hydrogen above the dense and intensely hot hydrogen producing the bright bands, there will be seen on these bands the fine dark lines of hydrogen at less pressure. We learn, then (i.) that an element may be present in the sun, though neither its dark lines nor its bright lines may be seen; (ii.) that, at any rate for a time, an element may indicate its presence in the sun by bright lines instead of by dark lines in the solar spectrum; and lastly, that a dark line even centrally placed on a VOL. CCXLV. NO. 1788.

bright band may not of necessity indicate that the corresponding element is not present in that condition of intense heat in which such an element gives out a superabundance of its own special tints.

Next to be mentioned, in this connection, is the discovery that, though in the case of our own sun hydrogen always indicates its presence (when the whole light of the sun is examined) by dark lines, this is not the case with all the suns. The spectrum of the bright star Betelgeuse shows none of the dark lines of hydrogenthough few physicists would now advance the theory that there is no hydrogen in that star. The spectrum of the star Gamma Cassiopeia shows the lines of hydrogen bright instead of dark. Thus it appears that the presence of an element in a sun may constantly be indicated by the bright lines of that element.

Thirdly must be noticed the recognition of the important circumstance that a vapour which, at ordinary pressures, has a spectrum of bright lines (so that when acting absorptively on light-that is, when such a vapour is interposed between the eye and a source of light hotter than the vapour-it would produce dark absorption lines) has at higher pressures a spectrum of bright bands, and that when the pressure and temperature are sufficiently increased the spectrum becomes continuous. So that a continuous spectrum is not necessarily an indication that the source of light is a glowing solid or liquid body; it may be a gaseous body, if only we have reason to consider that any gas constituting it would exist at a very great pressure and at a very high temperature. As we certainly have reason to believe that this is the case with the sun, we can no longer accept as demonstrated the theory that the visible surface of the sun consists of glowing solid or liquid matter. It may consist of gaseous matter. Nor indeed, when we rightly apprehend the evidence obtained respecting the widening of the bright bands forming the spectra of glowing gaseous substances, are we even forced to assume that the visible surface of the sun consists of gas at a very high pressure. If the multitudinous bright lines which form the spectrum of the sun's complex vaporous atmosphere (as we know not only from theoretical considerations, but from the fact that this beautiful spectrum-tens of thousands of bright lines of all the colours of the rainbow-has been seen during total eclipses of the sun) were severally widened even to a moderate extent, they would form a continuous spectrum, for the bright bands would coalesce and overlap. And precisely as the dark lines of hydrogen have been seen on the broad bright bands of underlying hydrogen, hotter and