I am thus led to consider what I hold to be far the most important result of the eclipse expeditions of December 1870.

When the photographic records, the best drawings, the verbal descriptions, and the spectroscopic observations are carefully compared together, two relations are revealed which, whatever their interpretation, are undoubtedly full of meaning. They may be thus expressed :

(1) Where any great gap or rift appears in the outer or radiated part of the corona, there a depression is seen in the inner and brighter portion.

(2) Where the inner portion of the corona is depressed, there the coloured prominences are wanting and the sierra is very shallow.

The former relation is very markedly shown in Lieutenant Brown's drawing, fig. 92 (and he referred to it as a fact he had specially noticed), and the Spanish and Sicilian photographs (figs. 93, 94) illustrate the relation in a very striking manner. In confirmation of the second, I may quote Professor Roscoe's words respecting Mr. Seabroke's spectroscopic views of the prominences, and Professor Watson's drawing of the corona: On comparing the two drawings thus independently made, a most interesting series of coincidences presented themselves. Wherever on the solar disc a large group of prominences was seen in Mr. Seabroke's map, there a corresponding bulging out of

disposing completely per se of Oudemann's theory (untenable, as we have seen, for other reasons).

the corona was chronicled on Professor Watson's drawing; and at the positions where no prominences presented themselves, there the bright portions of the corona extended to the smallest distances from the Sun's limb.

Of course the two relations involve an agreement between the greatest extension of the outer corona and the regions richest in prominences. Mr. Brothers distinctly notes this agreement; for he says, there can be no question that there was more coronal light on the west side of the Moon than at the other points;' and then he calls attention to the fact that the prominences are more numerous on the side where the corona is brightest.' To this I may add Father Secchi's description of the western range of prominences. He tells us that towards the close of the totality (when the western prominences were disclosed) the protuberances formed a beautiful small semicircular corona, full of the well-known rosy jets, of admirable shape and beauty.'

Now, at first sight, one might be led to recognise in these relations some evidence for the atmospheric glare theory-in that modified form of the theory which makes the outer corona due to atmospheric glare produced by the inner corona and the prominences. Indeed Mr. Lockyer adopted even so late so late as last August the following line of reasoning: The outer corona was observed to have a rosy tinge over the prominences, and the spectrum of the prominences was detected many minutes above them, as well as on the

dark Moon. It could not have got this colour at the Sun, for its intrinsic colour was green, and the red light of the hydrogen supplied at the Sun was abolished altogether was absorbed and could only reach the corona at the Sun, so to speak, as dark light.'*

Here the reasoning as to light is sound, except as respects the conclusion that the intrinsic colour of the corona is green-a result by no means necessarily to be inferred from the green colour of the bright line in the coronal spectrum.† But a brief consideration of the geometrical relations involved shows that the redness of the coronal light over the prominences cannot possibly be explained by the atmospheric glare theory. The part of our atmosphere which lies towards the rose-tinted extensions of the corona is not more illuminated by the corresponding rich prominence region than those neighbouring parts of the atmosphere which do not show this rosy light. If the rich prominence region were the source of the red tint, there could be no reason why there should be an extension in that particular direction, and not in others. We have, in fact, in the above reasoning a repetition, but in a modified form, of the misapprehension which characterised the form of the glare theory dealt with in pp. 308, 309. It is only necessary to consider a rich red prominence

* From a paper read before the British Association last August. + The light concentrated on a bright line may be far less in quantity than light forming a continuous spectrum, or a spectrum of broad bands. The light of the aurora, for instance, has sometimes shown a green line as the most conspicuous line, even when the colour of the auroral curtain has been markedly red. Similarly the zodiacal light has a red tinge, yet the spectrum shows a green line.

region as a source of light illuminating our atmosphere, to perceive that the law according to which various parts of the atmosphere would be illuminated, must utterly preclude anything like radial extension from the region towards which the eclipsed Sun lies. There would also be a gradual increase of brilliancy (instead of diminution) with distance from the Sun's place.

We are bound then to find an explanation of the phenomenon in some solar relations. In fact, we find ourselves brought by a new line to the difficulty already dealt with at p. 378, and considered as almost necessitating the idea of some form of radial repulsion exerted by the Sun. The meteoric theory per se, as pointed out at p. 378, will not avail to explain the straight radial bars and streaks seen in the corona; neither will it serve to interpret the relations we have now to consider. Regarding the prominences as phenomena of eruption, the occurrence of great solar eruptions cannot affect meteoric or cometic matter travelling myriads of miles above the regions of outburst. We cannot suppose for a moment that a sufficient superiority exists in the light-giving or heat-giving powers of the prominence regions to explain so marked a difference. Nor is it easy to suppose that some special form of electrical or magnetical action exerted above the prominence regions is the true explanation of the peculiarity. That electricity is involved in some way in the production of the coronal light may well be believed; and further, it is far from improbable (as indeed has already been suggested) that there is some special

electrical action in progress above the prominence regions. But to explain a coronal extension by the assumed extension of this electrical action, is to explain one problem by another of greater difficulty. The sole explanation which seems available is so startling, that at first sight many will be disposed to reject it on that account alone. It is this—that a portion of the coronal light is due to the ejection of matter from the Sun, and either chiefly or wholly from those bands of the solar surface which we have thus far termed the spot

zones.

Let it be distinctly understood, however, that this theory (in favour of which I propose to cite further evidence) is not to be regarded as intended to displace the meteoric theory, or even the theory that a solar atmosphere of extreme tenuity exists as far as, or beyond, the observed extension of the corona. On the contrary, we may be led by the emission theory to attach more weight than heretofore to the theory of a solar atmosphere extending very far from the Sun; while the meteoric theory, as accounting for some slight part at least of the coronal light, can scarcely but be regarded as demonstrated.

The first and most obvious objection to the theory that matter is projected from the Sun to such vast distances as the theory requires, resides in the enormous initial velocities which such ejection would imply. From what has been shown in Chapter II., it will readily be seen that a velocity of projection of about 300 miles per second would be required to carry matter