mind of man can dwell upon. Those long beams have (then) a real extension compared with which the volume of our Earth, nay even the volume of the Sun himself, sinks into utter insignificance; and that inner radiance which encloses the Sun on every side indicates a luminous region of inconceivably vast extent, while the problems suggested for our consideration by the aspect of this region, and by the physical state of the material distributed through it, are of the most interesting character.

It will conduce perhaps to clearness of conception if we consider where that portion of our atmosphere is situated which, according to what may be called the 'atmospheric glare' theory-advocated by M. Faye, Mr. Lockyer, and possibly by others—is illuminated by solar light during total eclipses. We can then inquire at our leisure by what path the solar light reaches this region of our atmosphere.

Let the observer be at o (fig. 86) in the centre of the Moon's shadow bb', which forms an elliptical dark space on the Earth's surface. We need not concern ourselves with the shape of this ellipse, which will vary in different eclipses and at different stations. We need only note that in a considerable total eclipse the least diameter of b b' will be greater than 100 miles. Now, let a b' represent a portion of the Moon's shadowcone, forming within our atmosphere a figure not differing greatly from an oblique cylinder. Assigning to the atmosphere a height of about 200 miles this cylinder will have a shape such as a b′; but if we assign to the

*

atmosphere an elevation of 500 miles, then we should have to assign to our shadow cylinder the figure a b'. Now, let lines drawn from the observer's eye to the boundary of the Moon's disc enclose the black cone shown in part in o m, while lines drawn to the boundary of a circular corona extending one degree on every side of the Moon's disc during totality form the cone shown

in part in o c c. Both cones are shown well within the cylinder a b', because as a matter of fact we find that the lines o a, o a' would contain an angle considerably greater than the angle c o c'.

* We are here considering, be it remembered, the atmosphere which is effective in reflecting solar light to the eye; and it will scarcely be admitted by most meteorologists that an atmosphere of this sort extends even to so great a height as 100 miles from the Earth's surface. Our best estimate (so far as this characteristic of the atmosphere is concerned) is undoubtedly that founded on the height of the twilight curve when observed from suitable stations, for this height depends on the very quality we are considering. Now Bravais, from a discussion of Lambert's observations of the crepuscular curve, deduced a height falling short of 100 miles, while his own observations made from the summit of the Faulhorn gave a height of about 66 miles. So far as the real extension

Now, if the atmospheric glare theory is true, all the cone o c c'in our atmosphere is illuminated at the time of central eclipse except only the core belonging to the cone om. This is certain, because we see the Moon dark and the corona bright; so that we require om to be dark and the remainder of oc c' to be bright. Now, so far as undeflected solar rays are concerned, the whole region a b' is in shadow. The light from the prominences can get into this region, and so perhaps can solar light deflected by some possible action at the Moon's surface. But the problem which the supporters of the atmospheric glare' theory have to solve is to get the light into the cone o ce', growing brighter and brighter up to the very boundary of the dark cone mo (to correspond with the increase of the corona's light up to the Moon's limb), and there suddenly ceasing. This done, they must show further that if another observer is stationed somewhere else

6

of the atmosphere is concerned, we may accept the opinion of Dr. Balfour Stewart that observations made on the aurora supply the best means of forming an opinion. From such observations made in 1819. Dalton estimated the extreme height of the auroral arch at 102 miles. Sir John Herschel estimated the height of the auroral arch seen on March 9, 1861, at 83 miles. Observations of meteors afford another means of solving the problem. A height greater than any of those above mentioned has been deduced from observations of this sort. Lastly, polariscopic observations have led Liais and others to the conclusion that our atmosphere extends to a height of more than 200 miles from the Earth. The rarity of the atmosphere at such an elevation assuming the law of diminution of density which prevails lower down to continue unchanged, would be altogether inconceivable. A quantity of air which a healthy person of average height could draw into his lungs at a single inspiration would suffice, when so reduced in density, to fill a sphere exceeding in diameter the orbit of Jupiter.

within b b' as at o', the cones formed by lines from his eye to the Moon's limb and the corona's boundary, are respectively dark and illuminated in exactly the same way that is, they must show that the same regions of the air are at once illuminated and in darkness. This may fairly be regarded as impossible.

6

Yet even if this could be demonstrated, much more would still remain to be done before the atmospheric glare' theory could be regarded-I will not say as established-but as worthy of consideration. Until something of this sort has been done-and nothing of the sort has yet been attempted-we need not inquire how far those spectroscopic observations can be explained away which Professors Young and Harkness justly regard as of themselves demonstrating the non-terrestrial character of the coronal light.*

It appears, then, that whatever view we are to form of the actual constitution of the corona, we can at least have no doubt that it is a true solar appendage. There may not be any closer bond of union between the

*It will be observed that the above considerations dispose of that modified form of the atmospheric glare theory which Mr. Lockyer has recently put forward. Without inquiring here whether (as he asserts) a possible lunar action formed a part of the theory as originally propounded (see preface to the second edition of my Other Worlds), I must point out that such action does not render the theory at all more satisfactory. We have, if the theory is true, a certain region of our atmosphere illuminated and a certain other region dark, and the theory gives no explanation whatever of how this comes about. Moreover, supposing it did explain the matter for one observer at one moment. the explanation would not avail to show how in the case of another observer that same illuminated region would be dark and that same dark region illuminated at the very same moment.

material substance (of whatever sort) which emits the light forming the corona than exists between the nucleus of a comet and the comet's tail. But certainly the evidence seems to force on us the conclusion that a relation as unquestionable associates the corona and the Sun as that which compels us to regard the tail and coma as real appendages of a comet.

The corona thus viewed becomes one of the most important and interesting of all the phenomena of the solar system. We no longer have to deal with sunbeams shining through our atmosphere, or with mirages in some lunar envelope, but with luminous spaces of inconceivably vast extent. Let us consider what evidence we already have bearing on the nature of this wonderful solar appendage. We shall then more justly appreciate the interest attaching to those efforts which are being made to gain fresh information.

The general aspect of the corona, as described in the preceding pages, does not suggest the idea that we have to deal with a solar atmosphere. Those radial projections are not the appearances we should expect to find in an atmospheric envelope. Nor again is it easy to understand how the irregular masses of light, the spots resembling hanks of thread in disorder, and other peculiarities of a like nature, can be accounted for on the theory that the corona is a solar atmosphere.

But this view of the corona may be regarded as disposed of completely by the nature of the lines seen in the spectra of the coloured prominences. We have seen