Embed Size (px)
Citation preview
NOVEMBER 15, 1930 PHYSICAL REVIEW VOLUME 36
LETTERS TO THE EDITOR
Prompt publication of brief reports of important discoveries in physics may be secured by addressing them to this department. Closing dates for this department are, for the first issue of the month, the twenty-eighth of the preceding month; for the second issue, the thirteenth of the month. The Board of Editors does not hold itself responsible for the opinions expressed by the correspondents.
Polarization of Mercury Lines in Stepwise Radiation
The polarization of several mercury lines appearing in fluorescence when a mixture of nitrogen and mercury vapor is radiated by light from a quartz mercury arc has been investigated. Wood and others have shown that when a mixture of mercury vapor and a few millimeters of nitrogen is radiated by a quartz mercury arc giving an unreversed resonance line (2537), mercury atoms in the normal ( l 1 ^ ) state are raised by absorption of this line to the excited (23Pi) state where they collide with nitrogen molecules causing a large fraction of them to revert to the metastable (23JP0) state. These metastable mercury atoms, having a long mean life, may absorb other lines from the arc, thus carrying them to various higher states from which they may radiate a diversity of lines. We shall call this process "stepwise excitation."
In the present experiments a quartz tube containing mercury vapor at room temperature and 3 mm pressure of nitrogen was radiated by a water cooled and magnetically deflected quartz mercury arc. The fluorescence produced was observed at right angles to the exciting beam, and was tested for polarization by means of a Savart Plate, a Quartz Glans Prism and a small quartz spectrograph previously described.1 By this method any line of the fluorescence showing more than a few percent polarization will appear on the spectra plate crossed by fringes perpendicular to the line, which will be more or less distinct depending on the degree of polarization of the line. The fluorescence tube was placed in a magnetic field parallel to the exciting beam. Mercury atoms reaching the 23Si state from the 23P0 by absorption of the line 4047 from the arc, radiate 4047, 4358 and 5461 (not registered on plate used). The 33Di state is reached in the same manner by absorption
of 2967, from which 2967, 3131, and 3663 are radiated. A reproduction of a typical plate is seen below (Fig. 1). The lines 4047, 4358, 2967 and 3131 are all crossed by fringes, while the line at 3660 is not. The fringes on 4047 and 2967 are very distinct, indicating a large degree of polarization while those on 4358 and 3131 are indistinct showing only partial polarization. Moreover, the maxima
t—
\o Os CN
T-1 CO
CO
© vO VO CO
l>-
^ o ^
00 LO CO
^ I I I I [
M HI
FIG. 1
of intensity of fringes on 3131 are at the same position as the minima on 2967, showing that they are polarized perpendicularly to each other. This is easily seen on the original plate but may not show distinctly on the reproduction. The line 4047 is strongly polarized corresponding to 2967, and 4358 is partially polarized perpendicularly to it, analogous to 3131.
These results can be explained by a consideration of the Zeeman levels for each line in question and their relation to the polarization of resonance radiation. The polarization of the lines 4047 (28Si-»2*P0) and 4358
1 A. C. G. Mitchell, Journ. Frankl. Institute 209, 747 (1930).
1589
1590 LETTERS TO THE EDITOR
(235i-^23P1) have been explained by Hanle and Richter2 who showed experimentally that with a magnetic field parallel to the exciting beam, 4047 was 100% polarized with its electric vector perpendicular to the magnetic field and 4358 3 3 % polarized with its electric vector parallel to the field, in agreement with the theory. From the known Zeeman levels of 2967 (33£>!~23P0) and 3131 (33J91-23P1), it is easy to show theoretically that 2967 should be 100% polarized corresponding to 4047 and 3131 3 3 % corresponding to 4358. These experiments are in qualitative agreement with the theory.
It has further been shown, by using a large quartz spectrograph (Hilger E\), that the line seen at 3131 on the small spectrograph is actually that line and not a composite of 3131 and 3125, the latter line being absent or at any rate very weak compared to 3131 in the fluorescence. The line at 3660 is a com-
A. L. Hughes and G. E. M. Jauncey (Phys. Rev. 36, 772, (1930)) describe some experiments intended to detect the self-scattering of light bundles due to collisions of photons. Some years ago similar experiments were performed and published by me in Russian (Jour. russ. phys.chem. 60, 555, 1928) with the same negative results. Light of condensed sparks was used, its momentum intensity being much greater than that of condensed sun light as used by the American authors. At the same time it was pointed out that experiments of this kind are unnecessary.
Phenomena in the neighborhood of the sun give us much more information about the subject. Very intense bundles meet and intercross near the sun's surface. In case collisions of photons exist—light in the neighborhood of the sun must be powerfully scattered. We know that near the sun some scattered light really exists—it is the solar corona.
Many Zeeman patterns are quite unresolv-able, even with powerful apparatus, and appear as spurious triplets or quartets, according as AJ— ± 1 or 0. Shenstone and Blair,1
assuming that in this case the measured position coincides with the theoretical center of intensity of the unresolved pattern, have derived formulae which are of great utility in
posite of 3650, 3654 and 3663 and therefore shows no polarization.
The fact that the line 2967 is largely polarized (practically 100%) when 3 mm of nitrogen is present means that the 33Di state has a short mean life, for a calculation based on the time between collisions of mercury atoms and N2 molecules shows that the mean life is probably 10~8 sec or less. The work is being continued with a view to measuring the mean life of this 3zDi state by a magnetic depolarization experiment and will be reported in detail at a later time in this Journal.
ALLAN C. G. MITCHELL
Bartol Research Foundation
of the Franklin Institute.
October 24,1930.
2 W. Hanle and F. Richter, Zeits f. Physik 54, 811, (1929).
From data about the intensity of the corona and from the law of distribution of its intensity as a function of the distance from the sun it is easy to calculate that the coefficient of the scattering near the sun is of the order 10~17. Theories advanced about the corona explain this scattering as a scattering of sun light by atoms or electrons.
But even if we had some reasons to ascribe the corona to the hypothetical self-scattering of photons, its value (10~17) must be so small that it is hopeless to detect it with terrestrial experiments. The effective radius of photons must be smaller than 10~20 cm.
The principle of superposition of the incoherent light bundles is also fulfilled with great accuracy.
S. VAVILOV
The States Electrotechnical Institut U.S.S.R. Department of Physics.
October, 1930.
determining g-values from observations of complex spectra. The present note calls attention to the very simple values which the unresolved shifts assume when the g's follow Lande's formula.
If Ba, BT are the mean displacements of
1 Phil. Mag. 8, 765-771 (1929).
On the Attempt to Detect Collisions of Photons
A Note on Zeeman Patterns
O N CO "O >* IT) vo o er> cr> •*? •*?
II
FIG. 1
