bright close to the sun. And though Kepler considered that his own observation of a broad reddish corona satisfactorily removed Clavius's difficulty, it seems tolerably clear that the corona seen by Clavius must have been very unlike the corona seen by Kepler. In fact the former must have been like the corona seen in July, 1878, much smaller than the average, but correspondingly increased in lustre. Now with regard to the sun-spot period we can go back to the year 1567, though not quite so securely as we could wish. Taking the average sun-spot period at eleven years, and calculating back from the minimum of spots in the year 1610, we get four years of minimum solar disturbance, 1599, 1588, 1577, and 1566. We should have obtained the same result if we had used the more exact period, eleven one-ninth years, and had taken 1610·8 for the epoch of least solar disturbance (1610·8 meaning about the middle of October, 1610). Thus the year 1567 was a year of few sun-spots, probably occupying almost exactly the same position in the sun spot period as the year 1878. Clavius's observation, then, is in favour of our theory.
But another observation between Clavius's and Kepler's may here be noticed. Jensenius, who observed the eclipse of 1598 at Torgau in Germany, noticed that, at the time of mid-totality, a bright light shone round the moon. On this occasion, remarks Grant, the phenomenon was generally supposed to arise from a defect in the totality of the eclipse, though Kepler strenuously contended that such an explanation was at variance with the relation between the values of the apparent diameters of the sun and moon as computed for the time of the eclipse by aid of the solar and lunar tables. The corona, then, must have resembled that seen by Clavius, and since the year 1598 must have been very near the time of fewest spots, this observation accords with the theory we are examining.
The next observation is that made by Wyberd during the eclipse of 1652. Here there is a difficulty arising from the strange way in which the sun-spots behaved during the interval from 1645 to 1679. According to M. Wolf, whose investigation of the subject has been very close and searching, there was a maximum of sun-spots in 1639 followed by a minimum in 1645, the usual interval of about six years having elapsed; but there came a maximum in 1655, ten years later, followed by a minimum in 1666, eleven years later, so that actually twenty-one years would seem to have elapsed between successive minima (1645 and 1666). Then came a maximum in 1675, nine years later, and a minimum in 1679, four years later. Between the maxima of 1639 and 1675, including two spot periods, an interval of thirty-six years elapsed. There is no other instance on record, so far as I know, of so long an interval as this for two spot-periods. In passing, I would notice how little this circumstance accords with the theory that the sun-spots follow an exact law, or that from observations of the sun, means can ever be found for forming a trustworthy system of weather prediction, even if we assumed (which has always seemed to me a very daring assumption), that terrestrial weather is directly dependent on the progress of the sun-spot period. But here the irregularity of the spot changes affects us only as preventing us from determining or even from guessing what may have been the condition of the sun's surface in the year 1652. This year followed by seven years a period of minimum disturbance, and preceded by three years a period of maximum disturbance; but it would be unsafe to assume that the sun's condition in 1652 was nearer that of maximum than that of minimum disturbance. We must pass over Wyberd's observations of the corona in 1652, at least until some direct evidence as to the sun's condition shall have been obtained from the papers or writings of the observers of that year. I note only that Wyberd saw a corona of very limited extent, having indeed a height not half so great as that of many prominences which have been observed during recent eclipses. If the theory we are examining should be established beyond dispute, we should be led to infer that the year 1652 was in reality a year of minimum solar disturbance. Perhaps by throwing in such a minimum between 1645 and 1666, with of course a corresponding maximum, the wild irregularity of the sun-spot changes between 1645 and 1679 would be to some degree diminished.
We are now approaching times when more satisfactory observations were made upon the corona, and when also we have more complete records of the aspect of the sun's surface.
In 1706 Plantade and Capies saw a bright ring of white light extending round the eclipsed sun to a distance of about 85,000 miles, but merging into a fainter light, which extended no less than four degrees from the eclipsed sun, fading off insensibly until its light was lost in the obscure background of the sky. This corresponds unmistakably with such a corona as we should expect only to see at a time of many sun-spots, if the theory we are examining is sound. Turning to Wolf's list, we find that the year 1705 is marked as a year of maximum solar disturbance, and the year 1712 as that of the next minimum. Therefore 1706 was a year of many sun-spots—in fact, 1706 may have been the year of actual maximum disturbance, for it is within the limits of doubt indicated by Wolf. Certainly a corona extending so far as that which Plantade and Capies saw would imply an altogether exceptional degree of solar disturbance, if the theory we are considering is correct.
In 1715 Halley gave the following description of the corona:—'A few seconds before the sun was all hid, there discovered itself round the moon a luminous ring about a digit' (a twelfth) 'or perhaps a tenth part of the moon's diameter in breadth. It was of a pale whiteness or rather pearl colour, seeming to me a little tinged with the colours of the Iris, and to be concentric with the moon.' He added that the ring appeared much whiter and brighter near the body of the moon than at a distance from it, and that its exterior boundary was very ill-defined, seeming to be determined only by the extreme rarity of the luminous matter. The French astronomer Louville gave a similar account of the appearance of the ring. He added, however, that 'there were interruptions in its brightness, causing it to resemble the radial glory with which painters encircle the heads of the saints.' The smallness of the corona on this occasion corresponds with the description of the corona seen in July 1878; and though Louville's description of gaps is suggestive of a somewhat different aspect, yet, on the whole, the corona seen in 1715 more closely resembles one which would be seen at a time of minimum solar disturbance, if our theory can be trusted, than one which would be seen at a time of maximum disturbance. Wolf's list puts the year 1712 as one of minimum disturbance, with one year of doubt either way, and the middle of the year 1817 as the epoch of maximum disturbance, with a similar range of uncertainty. The case, then, is doubtful, but on the whole inclines to being unfavourable. I may remark that because of its unfavourable nature, I departed from the rule I had set myself, of taking only the cases included in my treatise on the sun. For the corona of 1715 is not described in that treatise, as indeed affording no evidence respecting this solar appendage. The evidence given in this case is probably affected in some degree by the unfavourable atmospheric conditions under which Halley certainly, and Louville probably, observed the eclipse. In any case the evidence is not strong; only I would call attention here to the circumstance that if, as we proceed, we should come to a case in which the evidence is plainly against the theory we are examining, we must give up the theory at once. For one case of discordance does more to destroy a theory respecting association between such and such phenomena, than a hundred cases of agreement would do in the way of confirming it.
In 1724, Maraldi noticed that the corona was broadest first on the side towards which the moon was advancing, and afterwards on the side which the moon was leaving. From this we may infer that the corona was only a narrow ring on that occasion, since otherwise the slight difference of breadth due to the moon's eccentric position at the beginning and end of totality would not have been noticeable. Now, the year 1723 was one of minimum disturbance, with one year of doubt either way. Thus 1724 was certainly a year of few sun-spots, and may have been the actual year of minimum disturbance. The corona then presented an appearance according with the theory we are considering.
Few eclipses have been better observed than that of the year 1733. The Royal Society of Sweden invited all who could spare the time to assist, as far as their ability permitted, in recording the phenomena presented during totality. The pastor of Stona Malm states that at Catherinesholm, there was a ring around the sun about 70,000 miles in height. (Of course these are not his exact words; what he actually stated was that the ring was about a digit in breadth.) This is the exact height assigned to the coronal ring by the observers of the eclipse of last year. The ring seemed to be of a reddish colour. Another clergyman, Vallerius, states also that the ring was of this colour, but adds that at a considerable distance from the sun it had a greenish hue. This suggests the idea that the outer corona was seen also by Vallerius, and that it had considerable breadth. The reddish colour of the inner light