I translate his words-"Of all the bodies iron has certainly produced the greater number of lines in the solar spectrum. Some of these seem to be common with those of calcium." Thalèn carried on this work, and if one compares the magnificent tables, which we owe to his untiring skill and industry, one is perfectly astonished to find the number of coincidences which he has so carefully tabulated.

2. There was another kind of work, a newer kind of work, going on. Observers began to give particular attention to the bright lines of flames, and the lines thickened in spots. And here I may limit myself to the general statement that the divergence between the spectra of the different substances as observed in the sun and in our laboratories was very much intensified as facts were accumulated. Very many of the lines observed in flames were lines with no terrestrial equivalents, and the spotspectrum often contained lines much thickened, which were either not represented at all, or only feebly among the Fraunhofer lines.

3. Next, among all the metalloids known to chemists only one of them-or one substance classed as such, hydrogen-was present in the solar atmosphere, and that in overwhelming quantity; whereas the efforts of Ångström, Kirchhoff, and others could not trace such substances as oxygen, chlorine, silicon and other common metalloidal constituents of the earth's crust.

4. Then again, the layer which was produced by what was taken to be gaseous magnesium round the sun, a layer indicated by the brightest member of the b group, was always higheralways gave us longer lines-than that other layer which was brought under our ken by the bright line D seen in the spectrum

of sodium.

Here was a distinct inversion of the chemical order. The

SIRIUS

atomic weight of sodium being 23, and of magnesium being 24, the sodium ought to have been higher than the magnesium; but the contrary was the fact, and that fact still remains after twelve years of observation.

5. As the work of tabulating the lines went on, and the more complex outpourings of vapours from the sun's interior were studied, it was found that the lines of iron, calcium, and so forth revealed to us were by no means the brightest lines-by no means the most important, or most prominent lines, but lines which really we had very great difficulty in recognising as characteristic of any particular spectrum. There they certainly were, however, mapped as very fine lines by the most industrious observers. Similarly with the spots, there was an absolute inversion of the thickness of the lines of any one substance in the spot. Surely there was a great screw loose here.

6. Closely allied to these observations we had another extraordinary fact. We could quite understand why in a spot the change of refrangibility of the magnesium lines when there was a storm going on in the sun should be different from the change of refrangibility of, say, the iron lines. The natural explanation was, of course, this: you have the magnesium gas going at one rate, the iron gas going at another rate, and that is all there is to be said about it. But it was soon found that the differences which could be sharply seen between the spectrum of a particular mass of magnesium vapour and a particular mass of iron vapour extended to the iron vapour itself. There were just as many variations in the refrangibility of the lines of iron itself, for instance, as there were between the lines of iron and other substances: that is to say, we had in the one case magnesium going at one rate and iron going at another rate; but when we came to deal with the iron lines alone we found one

VEGA

iron line told us the iron vapour was going at one rate, and another iron line told us that same iron vapour was going at another rate. It will be seen at once that there was a great difficulty in that.

7. Further. The lines on which these observations of the relative motions of the vapour depended were found to go in sets. In a spot, for instance, we would generally see movement indicated by one set of iron lines, whereas in a prominence we would always see a different set-a set in a different part of the spectrum altogether-registering this movement for us. Here again was considerable food for thought.

That was stated very roundly a good many years ago-in 1869. I will read what was then written on this subject :1 Alterations of wave-length have been detected in the sodium, magnesium, and iron lines of the spot's spectrum. In the case of the last substance the lines in which the alteration was detected were not those observed when iron, if we accept them to be due to iron alone, is ejected into the chromosphere."

That caveat with regard to iron arose from the fact that of the 460 lines recorded by Kirchhoff in 1869 only three lines of iron had been seen bright in the solar prominences.

8. Then came a point which has been very slightly alluded to already. How came it that the total chemical composition of this atmosphere of the sun, which we were taught to look upon as the exemplar of what must have once happened to our own planet, varied so enormously from the composition of the crust of our earth? No oxygen in it, no silicon, no fluorine; whereas we get abundance of titanium, nickel, and so on. It was difficult I 1 Proc. Roy. Soc., vol. xviii. p. 74.

to imagine a stronger difference to exist between any two masses of matter than the chemical constitution of the incandescent sun, and of the earth, which is now cooling.

9. There was still another point of view very soon forced upon solar observers by the magnificent success which had attended the labours of Dr. Huggins, Secchi, and other observers in recording the spectra of stars. It was a most interesting inquiry naturally to see whether the stars gave spectra quite like each other, and if it should happen that they did not give spectra like each other, then the points of difference would be sure to give us some excellent working suggestions.

Now what are the facts? Here are three typical stellar spectra (Fig. 27), which show us at once that there is a very considerable difference in the phenomena. In the upper part of this diagram we have a star remarkable for the fewness of lines in its spectrum. From one end of the spectrum to the other there are not above half-a-dozen prominent lines. In the next part however we have a star which is remarkably like our own sun, both as regards the number of lines and their arrangement. In the lower part of the diagram, on the other hand, we have a star in which we get flutings instead of lines; so that we get not only a difference of degree, but a fundamental spectroscopic difference of kind. Now there is a circumstance connected with that first star with the simple spectrum very striking to any one in the habit of observing the sun, and it is this: those lines visible in the star, which, be it remembered, had been independently determined to be hotter than our sun, are precisely those lines, and none other, which we see bright on the disk of the sun itself. I have emphasised the fact that

we have independent evidence that the star with very few lines is hotter than our sun. It is also clear that the other star with the fluted spectrum is a star much cooler than our sun, because it was one of those red stars, the light of which is exceedingly feeble, which, on grounds independent altogether of spectroscopic evidence, are supposed to be stars in the last stage of visible cooling.

So much then for some of the earlier observations on the coin cidence of metallic lines in the sun, with observations on the lines themselves in different portions of the sun's atmosphere. 10. We now come to another part of the work where we also find difficulties. Ångström, in that exceedingly important memoir

which accompanies his Atlas, states: "In increasing successively the temperature I have found that the lines of the spectra vary in intensity in an exceedingly complicated way, and consequently new lines even may present themselves if the temperature is raised sufficiently high.' Kirchhoff, on his part, had seen phenomena very similar to those thus touched upon by Ångström, but his explanation was a different one. He did not agree that the temperature upon which Ångström laid such strong stress was really the cause at work. He attributed those changes rather to the mass and the thickness of the vapours experimented upon-nay, he went further: at a time when scarcely any facts were at his command he broached a famous theorem which went

to prove this; and yet what had Kirchhoff himself done? how had he traversed his own theory? He states that his observations were made by means of a coil using iron poles one millimetre in thickness. Now the thickness of a short spark taken from iron poles one millimetre in thickness would probably be two millimetres. Next Kirchhoff allocated the region where the absorption which produces the reversal of the iron lines took place at a considerable height in the atmosphere of the sun, and he expected the atmosphere of the sun to be an enormous mass represented by the old drawings of coronas, so that on Kirchhoff's view the thickness of the iron vapour which reversed the iron

spectrum must have been, at a moderate estimate, 10,000 miles, and yet he said that the spectrum of that, and of the light given by the coil were absolutely identical; that is to say, that the fart was that the variation of thickness from two millimetres to 10,000 miles made no difference. That was on the one hand; on the other hand he gave us his theorem, showing that a slight variation of thickness would produce all the changes which Angström and others had observed up to that time, and which we have observed since in much greater number. A diagram (Fig. 28) will show the sort of changes to which Ångström referred, changes which have been observed by every

new worker who has taken up the subject. It represents the variations which take place in the spectrum of calcium in the photographic region. At a particular temperature we get a spectrum of calcium which contains no lines whatever in the blue, but when we increase that temperature-the temperature of a Bunsen burner is sometimes sufficient to produce it-we get a line in the blue. When we pass from a Bunsen burner to an electric lamp we get this blue line intensified and reversed, and at the same time we get two new lines in the violet. Using a still higher temperature in the arc, we thin the blue line, and at the expense of that line, so to speak, we thicken the two in the

violet, so that the latter equal the blue line in thickness and intensity. Passing to a large induction coil with a small jar we make the violet lines very much more prominent, and using a larger induction coil and the largest jar we can get, we practically abolish the blue line and get the violet lines alone. Now we have simply produced these effects by varying the temperature, and this diagram enables me to point out one of the things to which reference will have to be made subsequently. The thicknesses of the calcium lines in the spectrum of the sun are also given. The two lines in the violet are really H and K, The

"Recherches sur le Spectre Solaire," pp. 38, 39.

other line the all-important one at low temperatures-is feeble and unimportant. So that both on the solar evidence and on the evidence of all these spectra, whatever the explanation may be, there is the undoubted fact that fundamental changes of intensity in the lines are produced by some cause or other, and if Kirchhoff's statement about the matching of lines is true for one temperature it is false for all the others.

II. In my reference to stellar spectra I mentioned the word "fluted" spectrum. Before Kirchhoff had published his first paper two very eminent Germans-Plücker and Hittorf-were working at spectrum analysis at Bonn, and they found that in the case of a great many simple substances what are called fluted spectra were to be observed as well as line spectra.

The accompanying diagram (Fig. 29) of the fluted spectrum of iodine will show the difference between these fluted spectra and the line spectra, on which we have been exclusively occupied up to the present,

We observe that the chief novelty is an absolute rhythm in the spectrum; instead of lines irregularly distributed over the spectrum, we have groups which are beautifully rhythmic in their structure. The next diagram (Fig. 30) shows us the radia tion spectrum of a particular molecular grouping of carbon vapour, that also is beautifully rhythmic; the rhythm of each of the elementary flutings exactly resembling that of the iodine.

These observations were among the first to suggest the idea that the same chemical element could have two completely distinct spectra. They were eminently suggestive, for if two, why not many?

In my reference to the "long and short" method of observation I stated that it enabled us to note what happens when a known compound body is decomposed. With ordinary compounds, such as chloride of calcium and so on, one can watch the precise moment at which the compound is broken up-when the calcium begins to come out; and we can then determine the relative amount of dissociation by the number and thickness of the lines of calcium which are produced. Similarly with regard to these flutings we can take iodine vapour, which gives us this fluted spectrum, and we can then increase the temperature suddenly, so that we no longer get the fluted spectrum at all, or we may increase it so gently that the lines of iodine come out one by one in exactly the same way that the lines of calcium came out from the chloride of calcium. We end by destroying the compound of calcium in the one case, and by destroying the fluted spectrum in the other, leaving, as the result in both cases, the bright lines of the constituents-in the one case calcium and chlorine; in the other case iodine itself. I have by no means exhausted the list of difficulties which were gradually presented to us when we considered that both in the sun and in our laboratories spectrum analysis brought before us the results of unique, absolutely similar chemical atoms." Not only were there differences, but the differences worked in different ways, whether we passed from low to high temperatures in labɔratory work, or from the general spectrum or the flame spectrum in the sun.

66

But I have said enough for my present purpose; details on the points I have referred to and on others must be gone into afterwards.

How then was one to attempt to grapple with these difficul ties? Was it the time to found new theories? or to rest and be thankful? Was it not better to appeal to what was known-to proceed in accordance with Newton's laws of philosophising, and start no new principle unless one were absolutely bound to do so to appeal in fact to the law of continuity, and to suppose that the explanation of a very large part at all events, of this new matter, lay in the fact that, all unconsciously, spectroscopists had been working under more transcendental conditions as regards temperature than had ever been employed before, and that the natural result was that this higher temperature had done for the matter on which they had experimented exactly what all lower temperatures had been found to do. That is to say, that they had been broken up. In other words, it lent great probability to the view that when we subjected, say iron-because it is a good thing to keep to one specific substance-to one of these transcendental temperatures, we were no longer dealing with the spectrum of iron, but with the spectrum of the constituents of iron revealed to us by a temperature at which no experiments had been made before.

And one was the more struck by the probability of this being at all events an approximation to the truth by those stellar spectra to which I have referred, and by the knowledge we possessed, that in the case of a star of the simplest spectrum we

were dealing with the highest possible temperature. So the idea was thrown out that these stars were really simpler in their structure; that their immense temperature had not allowed a complex evolution of higher complex forms of chemical matter to take place; and that we had there the primordial germs of matter, so to speak, or at all events something nearer to the beginning of things than anything that we had in this cool planet of ours, or anything that we were likely to find easily here, in consequence of the various difficulties which harass every kind of experimentation. It was imagined that we might picture to ourselves a sort of celestial dissociation in the heavenly bodies which would place those stars, the spectra of which have been seen, in a different order; that the first star with lines should be a star of the simplest spectrum, the next star with lines should be that which mostly resembled our sun, and that the last in order should be that one in which the lined spectrum had utterly disappeared in favour of the fluted spec. trum. If this were granted for the stars, why not attach all this to the sun? Because, as has already been mentioned, all these lines which were seen in the spectra of the hottest stars were precisely those lines which were seen most intense in the hottest parts of the sun; and it did really seem as if in that way we could eventually sooner or later-most likely later, for Art is very long-get some light on the subject.

I at once say that this idea which was thrown out in the year 1873 on spectroscopic evidence had been anticipated by the foremost philosopher amongst English chemists of his time; I mean the late Sir Benjamin Brodie. From considerations of a perfectly different kind he had come to the conclusion that our chemical philosophy was not anything like so firmly based as was generally imagined, and that, given a higher temperature, the elementary bodies would cease to be elementary-that the adjective "elementary" applied to them was merely the measure of our inability to dissociate them ; and to watch the progress of dissociation when we got them at a temperature at our command. By a stroke of genius he, before anything was known about the chemistry of the sun, went to the sun for that transcendental temperature he was in search of; thus showing that he had an absolutely pure and accurate conception of the whole thing as I believe it to be-but that is anticipating matters. He suggested that the constituents of our elementary bodies might be found in the hottest parts of the solar atmosphere existing as independent forms. The whole merit of that conception therefore is due to Sir Benjamin Brodie, and dates from the year 1867.

Now we can easily understand, seeing that much of the spectroscopic work which had been done up to 1874 had had stellar, and terrestrial chemistry, that it was not a pleasant thing for its object the connecting-intermingling, so to speak-of solar, to find that the path seemed about to be such a very rugged onethat we seemed after all not to be in the light, but in the dark, and the very practical question was, what was to be done? Would it have been wise to have considered, then, the whole question of the dissociation of elementary bodies? I think it would not have been wise; the data were insufficient. The true thing to be done was, I think, to endeavour to accumulate a vast number of new facts and then to see what would happen when a sufficiently long base of facts had been obtained. What did we want? We chiefly wanted to settle those questions of the variations of spectra seen in our laboratories, and the variations observed when we passed from the spectrum, say of iron on the earth, to the spectrum of iron in solar spots and storms. The coincidence of lires of different bodies which had been referred to by Ångström and Kirchhoff also required investigation. What more ready means of doing that-what more perfect means were there than those placed at our disposal by photography? Photography has no personal equation, it has no inducement to cook a result either in one direction or the other, and it moreover has this excellent thing about it, that the results can be multiplied a thousandfold and can be recorded in an absolutely easy and safe manner. There were other reasons why photography should be introduced. We see at once that it it was quite easy to introduce the process of purification of the spectra to which I have already drawn attention, by merely comparing a series of photographs; the A, B, C of my diagram (Fig. 26) being represented, say, by iron, cobalt, and nickel, or any other substances. Again, it was quite possible by the use of the electric lamp to very considerably increase the

"Ideal Chemistry." Lecture delivered to the Chemical Society in 1867, republished 1880. (Macmillan).