lines appear as dark lines across the continuous spectrum of the electric light, what was more obvious than the conclusion that sunlight also, which shows these same dark lines, must, before reaching us, have passed through the vapour of sodium. Either then, in our own atmosphere, or in the atmosphere of the Sun, this familiar metal sodium, the basis of such commonplace substances as salt and soda, must exist in quantities sufficient to cause the observed absorption-lines.

But more than this, those countless other lines which cross the solar spectrum must each indicate some process of absorption exerted by vapours in our atmosphere or his. Since the presence of sodium has thus been demonstrated, why may not the presence of other elements be in like manner rendered apparent?

Kirchhoff did not long delay the inquiry thus suggested. He compared the spectrum of iron (or, to use the accepted mode of expression, the spectrum of the luminous vapour of iron) with the solar spectrum. Now, the spectrum of iron as known to Kirchhoff contained some sixty bright lines. There was a great difference, then, between this spectrum and that of sodium. Yet, to his astonishment, Kirchhoff found that precisely as the two sodium lines agree with the D lines of the sodium spectrum, so every one of the iron lines had its counterpart in the solar spectrum.† Line for line, and not only so, but strong

* Many more have since been discovered; in fact, there are now more than 450 recognised iron lines.

It is hardly necessary to note, perhaps, that Kirchhoff could not

line for strong line, and faint line for faint line, every line of the iron spectrum appears as a dark line in the spectrum of the Sun!

Kirchhoff in all probability never questioned for a moment the conclusiveness of this relation. He must have felt its significance intuitively. It was probably, therefore, only to satisfy others that he presented a strong argument from the theory of probabilities in favour of the view that the observed relation implied a real association between the two sets of lines. Taking as the chance that a bright line in the iron spectrum would seem to have a counterpart in the richly lined' solar spectrum if accident were alone in question, he calculates that the chance of the observed relation (leaving altogether out of question the relative strength of the lines) is less than

1

1,000,000,000,000,000,000.*

'Hence,' remarks Kirchhoff, this coincidence must

establish this result in the same way as the former, because he could not cause sunlight to shine through the glowing vapour of iron. He employed a method quite as satisfactory however, causing the light from iron vaporised by the electric spark to form a spectrum side by side with the solar spectrum, the solar light being admitted directly to the battery of prisms; that of the iron being admitted after reflection through a small prism near the slit.

To this I may add that the chance of the observed relation, now that 450 lines of iron are recognised, is less than a fraction whose numerator is unity, and whose denominator consists of no less than 136 figures (the first four being 2907). This chance is not very unequal to that which I have shown to correspond to the probability that one of the less marked peculiarities of stellar arrangement detected by me while constructing my large star-atlas, is due to mere chance-distribution.

be produced by some cause, and a cause can be assigned which affords a perfect explanation of the phenomenon. The observed phenomenon may be explained by the supposition that the rays of light which form the solar spectrum have passed through the vapour of iron, and have thus suffered the absorption which the vapour of iron must exert. As this is the only assignable cause of the coincidence, the supposition appears to be a necessary one. These iron vapours might be contained either in the atmosphere of the Sun or in that of the Earth. But it is not easy to understand how our atmosphere can contain such a quantity of iron vapour as would produce the very distinct absorption-lines which we see in the solar spectrum, and this supposition is rendered still less probable by the fact that these lines do not appreciably alter when the Sun approaches the horizon. It does not, on the other hand, seem at all unlikely, owing to the high temperature which we must suppose the Sun's atmosphere to possess, that such vapours should be present in it. Hence the observations of the solar spectrum appear to me to prove the presence of iron vapour in the solar atmosphere with as great a degree of certainty as we can attain in any question of natural science.'

Thus cautiously did Kirchhoff proceed in establishing the great principle on which spectroscopic researches into solar physics were to depend. Nothing can perhaps be required to strengthen the confidence of the reader in the justice of this principle. Yet it may be added that the effect of our own atmosphere on the

solar spectrum is now thoroughly recognised and found to correspond with the density of the layers through which the Sun's rays penetrate at different hours of the day. The reader may further be reminded of what has been already said respecting the evidence afforded by the character of stellar spectra. All those spectra would undoubtedly be similar to each other and to the solar spectrum if our atmosphere contained the source of all the spectral dark lines. It is to be noted also that while the coincidences in the case of iron serve to confirm any evidence we may derive from other coincidences, they are also confirmed by such evidence; because as one after another the terrestrial elements are found to have spectral lines corresponding with the solar dark lines, the general fact becomes more and more firmly established that the Sun is constituted of those elements with which we are familiar.

Kirchhoff found that calcium, magnesium, and chromium exist in the solar atmosphere. The presence of nickel, also, and cobalt was indicated by the agreement of the most conspicuous of their lines with solar dark lines. All the lines of these metals, however, could not be recognised, nor were the coincidences in the case of cobalt sufficient to satisfy Kirchhoff. I consider myself entitled,' he says, ' to conclude that nickel is visible in the solar atmosphere; but I do not yet express an opinion as to the presence of cobalt. Barium, copper, and zinc,' he proceeds, appear to be present in the solar atmosphere, but only in small quantities; the brightest of the lines of these metals correspond to

distinct lines in the solar spectrum, but the weaker lines are not noticeable. The remaining metals which I have examined, viz. gold, silver, mercury, aluminium, cadmium, tin, lead, antimony, arsenic, strontium, and lithium, are, according, to my observations, not visible in the solar atmosphere.'

I have thus far quoted Kirchhoff, not because the account of his investigations exhibits the actual state of our knowledge at the present time, but because they are so associated with the discovery of the great principle on which spectroscopic analysis depends as to have an interest wholly distinct from that-great as it iswhich they derive from their intrinsic importance. It would occupy much more space than is here at my disposal to exhibit the progress of research by which our knowledge of the solar spectrum has reached its present position. For the full history of the subject I would refer my readers to Dr. Schellen's admirable 'Spectralanalyse,'* and to the excellent treatise on spectroscopic analysis by Professor Roscoe. So far as my purpose in this place is concerned it is only necessary for me to sum up the results of the search for coincidences between solar dark lines and the bright lines of terrestrial elements-in other words, to exhibit the elements which exist, so far as is yet known, in the

* A translation of this work by the daughters of Mr. Lassell, F.R.A.S., and edited by Dr. Huggins, the great English master of the subject, will probably be published (by Messrs. Longmans) before this work appears. It will contain many important additions as compared with the first German edition, and cannot fail to be a work of extreme value.