LVI. Some Spectroscopic Notes. By ROBERT A. HOUSTOUN, M.A., B.Sc., Research Student in the University of Glasgow*. § 1. N a communication made to the Royal Society (" On a Field," by Prof. A. Gray, F.R.S., and Walter Stewart, D.Sc., with Robert A. Houstoun and D. B. Macquistan, Proc. Roy. Soc. vol. lxxii. No. 477) certain measurements were given of the wave-lengths of the satellites of the green line in Hg. These measurements disagree with those of Fabry and Perot; they have since been verified, and there seems evidence to show that the satellite system of the green line varies considerably under different conditions. According to Michelson (Phil. Mag. (5) xxxiv. p. 280, 1882) the green line of Hg is fourfold. It consists of a narrow bright line in the middle of a broader weaker one. Then follow three weaker lines towards the red (fig. 1), the values of dλ being about +0.075, +0100, and +0.125 Ångström units. Fabry and Perot (C. R. 1898, p. 409) represent the line as having two attendant lines +008 and +092 (fig. 2). Later they make it fivefold (C. R. 1899, cxxviii. p. 1156), one very weak component on the violet side, 0044, and three on the red side, 0·009, 0·082, and 0.136 (fig. 3). Fig. 1. •100 0 .075 .125 Fig. 2. 0 ⚫092 ·008 Finally, they make it sevenfold (P. Zeeman, Astroph. Journal, xv. p. 218), the differences being -0.224, -0076, -0.052, 0·008, 0.082, and 0.136 (fig. 4). * Communicated by Prof. A. Gray, F.R.S. Our own measurements represent the line as sixfold when most highly developed, the distances of the satellites Fig. 5. A B C 11 -0.208 -.059 -.096 E F ⚫032 being -0-208,096,059, +0.032, and +0.067, with the condition that these figures are right only to a multiple of 464 the distance between two successive orders. When a faint line is seen in the field the echelon spectroscope was the instrument with which our observations were of the central line D are very close together, and it is impossible to tell to what order the faint line belongs. Considerations of symmetry lead us to suppose that BCE and F are rightly placed. It is quite as probable that A is on the other side of D at a distance of +256. .067 made the different orders These measurements were made on an ordinary Geissler tube with aluminium electrodes, and were verified on a tube the electrodes of which were little mercury cups. The tubes were fed from an induction-coil that gave a 10-inch spark. The E.M.F. in the primary was usually 6 volts. The system did not always appear as above. B and C appeared at times as one, as so also did E and F. In that case the distance of (BC) was 084, and of (EF) 069. Then BC might be a doublet and EF a single line, or the system might reduce to two satellites corresponding to A and the unresolved doublet BC. It was usually thus when a less powerful induction-coil was used. Although Zeeman did not measure the position of the satellites with his echelon (which is slightly less powerful than ours), he apparently obtained the same arrangement, for he says, in the article quoted above: "Using the echelon in a position in which two strong lines of equal intensity corresponding to successive orders of the radiation were visible, I could distinguish also five faint, very narrow lines between the principal ones. The distance between two pairs of these lines was very small." The doublets (BC) and (E F) at times appeared as single lines. The single line was not the doublet badly defined; its width was not greater than that of one of the components of the doublet. Capacity and inductance in the primary and secondary had no effect on the satellites. They were usually seen best when the hammer of the induction-coil was going jerkily when it was screwed up, so that its spring was short. The bright yellow line of Hg 5790 was also investigated, and was found as in fig. 6, B being so close to C as sometimes not to be distinguished from it. Calculating A the distance dλ from an imaginary line between C and D from observations made at three different times, we get On one occasion while we measured the differences, the frequency of the hammer changed and the line A was seen to move further out. This was seen only once. The satellite systems are of course exactly the same when viewed across and along the direction of discharge; and no component is ever plane-polarized. Fig. 7. § 2. These two lines were also examined in the electric arc, the soft core of the lower carbon being bored out, a rubber tube being fastened to its lower end and connected to a mercury reservoir, by raising and lowering which the supply of Hg vapour in the arc could be altered, when some extraordinary cases of reversal were observed. These remained steady for long. Figs. 7 & 8 represent what the green line reversed into. Here the A B C light maxima were rather bands than lines. A is given as if the lines were satellites, for which they were at first taken. Fig. 8. The numbers represent distances from B. Similar measurements for fig. 8, taken from a point mid way between L and M, are Finally, the satellite system was obtained as in the vacuum tube (fig. 5). The results were :-— A -0.227 Fig. 9. Ę, A B C DE B -0.106 -0.056 0.046 0.084 A striking reversal effect (fig. 9) was also obtained for the yellow line in the E was very faint; E, is E in the next order. Measuring from a point midway between B and C we have arc. A B C D E -014-005 +0·05 +0.16 +0.29 The current varied from 5 to 7 amperes during the above measurements. Readings were also made on the lines of zinc in the electric arc. The zine was introduced into the are by boring the lower carbon which was the anode, and packing strips of thin zinc sheeting into the hole. Again some very striking forms of reversal were obtained. They did not maintain themselves so long as in the case of Hg, and passed out of the one form into the other frequently when I was making measurements on them. It was easier to recognize the zinc lines as being reversed. Changes took place in the reversal system when the arc began to hiss. Also the distance between the different bands varied in the different parts of the arc, being greater and better defined near the anode A; being small but well-defined at the edges B, and being as a rule ill-defined towards C. Each of the four lines, the red and the three blue (6361, 4812, 4721, and 4681) reversed in three types, figs. 10, 11, & 12. roughly speaking Fig. 10. Let us take the measurements for the case of 4812, the bright blue line. Measuring the wave-length differences in the three cases from B, E, and L, we obtained C... +0.05 Fig. 10. +0.08 +0.16 DE F GHL M P P... P may, however, be G in the next order. Using a different arc-lamp and a smaller quantity of Zn, I was able to get the lines without reversals. The red line (6361) was single. The three blue lines (4812, 4721, 4681) are all doublets, the components being of equal intensity. For 4812 dλ is 0.09, for 4721 0.08, and for the third line something similar. Examining Hg vapour under the same conditions, the green line appeared with two satellites, one corresponding to A (fig. 5) and the other to the doublet (B C), while the yellow line appeared as a double line with a weak component For the green line dλ=-0.24, -0.08, and for the yellow -0.06, +0.12. towards the red. The nickel line (5476) and the Na D lines were single. The red C line of hydrogen was double, da being =0.065, or about one-half Michelson's value. This explains why the line does not appear double in the first-order spectrum of a Rowland's grating. It seems on the whole tolerably certain that the satellites of the green line of Hg vary in number and position with the conditions of the source. What these determinative conditions are it is difficult to say. We can look on the multiple reversals of the lines as being made up of simple reversals of the components of the complex spectral lines. This seems more probable than alternate layers of radiating vapours at different temperatures, as it is difficult to see how these alternate layers can arise. The above measurements were made with a micrometer eyepiece and not on a photographic plate. This explains probably why such reversal effects are not more frequently seen. § 3. The echelon used for the above observations was made and mounted by A. Hilger. The telescope and collimator were both fixed and mounted on the one stand. The echelon rested on a platform between them; by moving this about a vertical axis, the different images could be brought into the field of view. The rays of light from the collimator do not fall normally on the echelon. In the discussion usually given it is assumed that they do so, that the collimator and echelon remain fixed and that the telescope moves. The formulæ are the same for both cases, though the discussion is slightly different. I have derived the formula for the case of the fixed telescope, using the graphical method given in Preston's 'Light, and as the echelon spectroscope illustrates this method very neatly the discussion is given here. It will be well to describe the appearances seen in the field. First of all, if the instrument is focussed on a welldefined line we see several images of it in the field. As the echelon is rotated these images cross the field, and as each crosses the centre of the field it gets very much brighter. In fact it is only the two or three in the centre of the |