should be seen streaked with several such bars, is a phenomenon which nothing in the meteoric theory considered per se is calculated to explain.

Now this is a difficulty which must be faced by some better means than a mere attempt to negative the evidence. That erroneous observations are made from time to time is unfortunately true; but when a characteristic phenomenon which does not seem likely to be merely imagined, is attested by trustworthy observers, a theory begins to wear a most questionable aspect which can only be supported by assigning those observations to illusion.

The difficulty in this case is that, even setting aside the objections (overwhelming as I think) which have been brought against all other theories but the meteoric one (not as the sole explanation of the corona, but as an absolutely essential part in any complete explanation), we do not find in any of these theories the means of explaining this particular phenomenon.

Certainly no theory involving the existence of a lunar atmosphere can aid us in this strait; because if the radial bars were due to the passage of the solar rays straight through such an atmosphere in lines touching the Moon's edge (and no other path would account for the observed phenomenon), then the extension of the lunar atmosphere necessary to account for the observed appearances would be about 200,000 miles, or the lunar atmosphere would extend nearly to the Earth.* We cannot for a moment imagine this.

* It is easy to see this by making a diagram showing the Moon and

It will be shown also, further on, that no sensible part of the corona's light can be due to the existence of scattered matter (cosmical dust) between the Moon and the Earth, as well as throughout the whole of that portion of space circled by the Earth in her course around the Sun.

Nor again can the atmospheric glare' theory explain this phenomenon, unless we suppose the Earth's atmosphere to extend nearly to the Moon's distance,— which is altogether incredible. Indeed, even if we admitted this, the bright rays between which the dark rays are seen ought to grow brighter and brighter with increase of apparent distance from the Moon, which is the reverse of what is actually seen.

It remains, then, that we should account for the phenomenon by the theory to which we have been led by other considerations. In fact, since the meteoric theory has been shown to accord so well with other phenomena, while it derives a negative strength from the obvious flaws in all the other theories, we are justified in accepting with a certain degree of confidence any explanation of these dark radial bars which the meteoric theory may point to. In other words, instead of feeling bound to explain these dark bars before admitting the meteoric theory, we may employ the meteoric theory to supply the explanation we require.

Now, there are two phenomena-one belonging to

Earth, with the paths of the imagined rays extending along the boundary of the Moon's geometrical shadow, carrying the rays on until, as seen from the station of an observer on the Earth (and within the shadowcone), they shall subtend an angle four or five times as great as that subtended by the Moon's diameter.

the solar system, the other to the Earth's economywhich seem likely to aid us in this matter. One is the appearance of comets' tails; the other is the aspect of auroral streamers. As respects the former phenomenon, it is to be remarked that the directive forces, whatever they may be, which cause the tails of comets to project from the Sun, reside undoubtedly in the solar globe, and act undoubtedly with very great energy on certain forms of matter near him. Hence, as we have abundant reason for believing that the corona is not free from a certain association with cometary matter, we need not be altogether surprised if we find in the corona evidences of the same sort of action that we recognise in the formation and projection of comets' tails. Yet again, as respects the second phenomenon, we have the striking evidence afforded by the spectroscope to show that a resemblance of some sort exists between the coronal light and that of our auroras; so that we are justified in tracing some resemblance, and even in conceiving that some association exists, between the long straight streamers which form so remarkable a feature of the aurora borealis, and those straight radial bars,* with dark intervening spaces,

seen in the solar corona.

It is worthy of notice that a difficulty exists in the relatively small section of the radial bars, both bright and dark; for this phenomenon would imply that the centre of that action to which these bars are due cannot have the dimensions of the solar globe which we are able to measure. It seems far from unlikely (since indeed we have other evidence corroborating such a view) that the central and more condensed portions of the Sun's mass may be the real seat of this intense repulsive action.

[This was written in 1870, and I have preferred to leave it unchanged

There is a circumstance which seems to render this relation more striking in the fact that the only explanation one can readily conceive of the observed characteristics of the auroral spectrum seems to bring us again upon that subject of meteoric astronomy which has thus far stood us in such useful stead. For among the bright lines seen in the auroral spectrum is one agreeing in position with a line of iron, and it has been thought probable by Stewart, Angström, and others, that the light of the aurora is due in part to electrical discharges taking place in the upper regions of our atmosphere. But how can iron reach those upper regions save from meteoric visitants? and what can be more likely than that iron does actually reach the upper regions of our air in this way, when we consider how largely iron enters into the composition of nearly all the meteoric masses which have been hitherto subjected to analysis?

But it may be reasoned that if this is indeed the case-if solar action in the upper regions of the Earth's atmosphere (or terrestrial action excited in some way by the Sun) can cause these electrical discharges— then solar action exerted directly on similar material in the other parts of the Sun's domain ought to excite a similar luminosity, and that therefore we ought at night to see some traces-faint, it may be, but still

in order to show that by a careful analysis of observed facts the results of later observations may be in part anticipated. What I have here and in the main text described as suggested by what had been noticed before the eclipse of 1870, was strikingly confirmed if not demonstrated by the observations made during that eclipse.]

recognisable of this particular form of phospho

rescence.

This amounts, in fact, to the consideration, that the limits of the corona as seen during total eclipse ought not to mark the real limits of the Sun's light-exciting action. And even supposing that but a small proportion of the coronal light is really due to this form of action—that is, to electrical discharges-it would still be likely that some signs of those meteoric systems whose illumination by the Sun has been here regarded as the cause of the corona, should be seen beyond the observed limits of that aureola of light.

Here again it happens (and I know no surer test of the justice of a theory) that we have been led to see that a certain phenomenon should be manifested, which actually is a familiar phenomenon of the heavens, and which would most assuredly have required explanation if it had not thus been led up to.

For precisely in that region where we should expect to find a faint gleam of light-precisely where the known relations of the planetary scheme would lead us to look for an abundance of meteoric material, there appears that mysterious luminosity known as the Zodiacal Light. And just as our reasoning has led us to regard the meteoric appendage of the Sun—an appendage really extending far beyond the orbits of the most distant planets-as variable in configuration, however constant when regarded as a whole, so we find the zodiacal light varying from year to year in brightness, and extent, and position. Its light, again,