the shells being due not to the outside substance existing only outside, but to the outside substance extending to the bottom of the sun's atmosphere, and finding in it at a certain height another shell, which again formed another shell inside it, and so on; so that the composition of the solar atmosphere as one went down into it, got more and more complex: nothing was left behind; but a great many things were added.

The recent work, so far as I am acquainted with it, has not in any way upset that notion; but what it has done has been to add a considerable number of new elements to this reversing layer. Instead of consisting of 14 elements, as it was then found to do, it may be, I think, pretty definitely accepted now to consist of about thirty.

The metals considered to be solar as the result of the

labours of Kirchhoff, Ångström, and Thalen together with the considerations brought forward regarding the length of the lines, were as follows:

Those more recently added, with the evidence by which their existence in the solar atmosphere is rendered probable, are as follows (in the Tables, pp. 167–169).

It is important to bear in mind that the lines recorded in these Tables are in most cases the very longest visible in the photographic region of the respective spectra; in some cases they are limited to the region 39-40, which I have more especially studied; so that the fact of their being reversed in the solar spectrum must be considered the strongest evidence obtainable in favour of the existence in the sun of the metals to which they belong, pending the complete investigation of their spectra.

Where, however, there is only one line, as with Li, Rb, &c., the presence of these metals in the sun's reversing layer can, for the present, only be said to be probable. Neither must it be forgotten that, in addition to the long lines which a spectrum may contain in the red, yellow, or orange, long lines may exist in the hitherto unexplored ultra-violet region; so that the necessity for waiting for further evidence before deciding finally upon the presence or absence of such metals in the sun will be rendered obvious.

It will be thought remarkable that, if the long lines of such metals as lithium and rubidium are found in the photographic region of the spectrum, the long lines Li W.L. 6705, Rb W.L. 6205 and 6296 should have escaped detection.

Metals the presence of which in the Sun is confirmed.

A line at 4604-5 in spectre normal assigned to Ca

Thalén..

Pd

4817.0

3539

To this it may be replied that, although these red lines may be apparently the brightest to the eye, it by no means follows they are the longest, since they are situated in a part of the spectrum which affects the visual organ more strongly than the photographic region does. It is possible also that the reasoning I have lately used in a paper communicated to the Royal Society, on the spectrum of calcium, may be applied in these

cases.

Since a sensitized film is affected by some rays more strongly than by others, in determining the lengths of lines from a photograph it is not fair to compare together portions of the spectrum separated by too great an interval.

Furthermore, the fact of these red lines having been overlooked in the solar spectrum is not conclusive proof of their absence, inasmuch as this portion of the spectrum is both brighter and less refrangible, and a greater degree of dispersion would be necessary when prisms are employed to render visible faint dark lines which are easily detected in the photographic region.

At present, then, out of the fifty-one metals with which we are acquainted here, more than thirty are known to exist in the sun with more or less certitude. Now it was a very remarkable thing that although such metalloids as carbon and sulphur, iodine, bromine, and the like, had been very diligently searched for, no trace whatever had been found of them, giving any evidence that they existed together with the metals in these zones (these shells) to which I have referred.

Some years ago evidence was brought forward of the possible existence of the metalloids as a group outside the metals; and the evidence for this suggestion was of the following nature:-Independently of any questions connected with solar physics, I think all students of science now agree that the vapours of the various elementary bodies exist in different molecular states; if these different molecular states are studied by means of the spectroscope, perfectly different spectroscopic phenomena present themselves. If we use a large coil, we can drive every chemical substance with which we are acquainted, including carbon and silicon, into a molecular grouping competent to give us what is called a line spectrum, the spectrum with which we are most familiar when we use metals or salts of metals in the electric arc.

If, however, other conditions are fulfilled; if these bodies are not so roughly handled-if, in other words, we employ a lower degree of heat, or if we use electricity so that we get quantity instead of tension, then these line spectra die away altogether, and we have a spectrum, so called, of channelled