resented in Fig. 7. The yellow light, the wave-length of which corresponds to those of the D lines, is not absorbed, but deflected in consequence of the anomalous dispersion. Becquerel's and Julius' observations had previously shown that the two D lines are anomalous to a different extent. From Wood's experiments we now know that the anomalous absorption of the D lines is larger than that of other sodium lines.

The same effect that is produced by the artificially arranged prismatic mass of a gas is produced by its gradually

variable density, as is the case in the spherical gaseous mass of the sun. In this case the observer with his spectroscope is always at a great distance from the "vapor prism," which disperses anomalously, so that he cannot observe simultaneously the whole phenomena. (In the laboratory we get the same effect by shortening and lengthening the slit of the spectroscope.)

Let us first imagine* an observer at such a great distance

* In this discussion, we follow closely the treatment given by E. Pringsheim in his publication "Ueber Brechung und Dispersion des Lichtes auf der Sonne." Archiv. für Math. und Physik, 4, 316-330, 1903.

from the sodium prism that he can observe only a small part of the whole dispersion phenomenon. Observing only the part between the horizontal lines, aa and a,a,, he sees a spectrum which shows broad, dark D lines. He will naturally, therefore, suppose that these dark bands are broadened absorption lines, or, what is the same thing, that the missing light has been absorbed by sodium vapor of great density. We now know that this supposition is wrong, since the missing light has not been absorbed, but deflected by anomalous refraction to other parts (between lines aa and ce or a,a, and c,c,). Thus: Anomalous dispersion can simulate a broadening of absorption lines.

In the spectrum of sun spots such a broadening of the narrowest Fraunhofer lines is observed, and it was supposed that those substances in the sun spots with broadened lines have a particularly great density. I feel certain that the missing light is not absorbed, but refracted by anomalous dispersion. To prove this, we need only suppose that in the sun spots the densities of the gaseous masses vary to a high degree from point to point. That such should be the case is not astonishing when, with FAYE or SPÖRER, we regard the sun spots as "whirlpools" in the solar atmosphere.

Observing only the parts between bb and cc (or b1b, and c1c1), the spectroscopist sees only two narrow bright lines, so near to the position of the dark D lines in the solar spectrum that the difference would be difficult to measure. Probably he will imagine that he is observing the bright sodium lines emitted by sodium vapor. Thus: Anomalous dispersion can simulate also a brilliant line spectrum.

I feel certain that the observed line spectra of the so-called "flash spectrum" of the chromosphere and of the prominences are not due to the emission of the vapors themselves-for example, sodium, calcium, strontium, etc.-but are derived from the regions nearer to the center of the sun, that has been deflected by anomalous dispersion so as to be seen outside the sharp edge of the sun.

Let us suppose that ZZ (Fig. 8) limits the critical sphere. A ray leaving tangentially from A passes along the slightly

curved path 40 and reaches the observer, 0, on the earth, while a ray leaving the critical zone at B goes in the direction BO'. Let us further suppose that there exists above A a cloudy mass of gas, as, for example, sodium vapor of unequal density. Surely, then, light-waves of a frequency close to the D lines suffer a much greater deflection by anomalous dispersion than all other waves. Thus rays of frequency N may leave the white beam BO' at h (above A), where the gaseous cloud is situated, and pass along h0, parallel to AO, to the observer at 0. Thus the observer at O will see above the sharply defined ball of the sun a region which shows a bright line spectrum. This spectrum of the chromosphere contains, as you know, more bright lines the nearer its origin is to the edge of the sun. According to the old theory, the

different gases must be considered arranged regularly one above the other. Now it is a great advantage that we can consider the gaseous substance a homogenous mixture, a product of complete diffusion. According to Julius' theory, the different lines of the spectrum of the chromosphere are to be observed at different distances from the sun's edge, since they suffer different anomalous dispersions when traversing mixed substances of unequal absolute densities.

According to Fig. 7, these apparent spectrum lines must become broader the nearer the point of observation ap-. proaches the edge of the sun. This is what is actually observed. Furthermore, these bright lines generally correspond to Fraunhofer lines, but show different characteristics with reference to relative intensity and position in the spectrum, which is in accord with Julius' theory.

15-Bull. Phil. Soc., Wash., Vol. 15.

Its most striking consequence was that the bright lines of the chromosphere may appear doubled, since local decreases of density from exterior parts to central parts of the sun may also occur. Indeed, every photograph taken by the Dutch expedition to Sumatra in 1901, on the occasion of the sun's eclipse, showed a doubling of all chromosphere lines. HARTMANN, of Potsdam, however, believes that the doubling of the lines is produced by errors of the apparatus.

Finally, when an observer looks at only the parts between aa and bb (or a,a, and b,b,) of the whole phenomenon (Fig. 7) he observes only two broad bright bands of light, the position of which is different from that of the D lines. Earlier theory assumed this deviation to be produced by the Doppler principle, by supposing gas eruptions with a velocity of 500 kilometers per second. These enormous velocities astonished all astrophysicists, but they were finally believed, since no other explanation existed.

If these enormous velocities are due to a sudden rise of temperature, calculations based on gas theory show that a temperature of five million degrees would be necessary to impart to the molecules of hydrogen a velocity of only 200 kilometers per second. Anomalous dispersion gives an explanation of the observed displacements without any restrictions.

The phenomenon most difficult of explanation of all observed since the spectroscope was invented is the following: In the spectrum of one and the same prominence, or at one and the same place near a sun-spot, the bright lines of one element show distortion, while those of another element are fairly straight, and it is most astonishing that even the different lines of one and the same element show a quite different appearance (Fig. 9). Some are deviated and curved, others are in their proper place and are straight lines.

In order to explain this curious fact, Sir N. LOCKYER introduced the hypothesis that on our sun the elements are dissociated into elementary atoms. This idea sounded at that time somewhat ridiculous; but now, after the discovery of radium and the transformation of its emanation into helium, etc., we must admire Lockyer's imagination and