Abstract
This paper is a continuation of the work previously reported in this Journal (Phys. Rev. 30, 438 (1927)). Experimental Improvements. The former source for the CO spectrum has been replaced by a Pyrex discharge tube placed longitudinally in the field. By special construction and the use of a forced-draft cooling system it has been possible to increase the intensity of the source about fourfold and to eliminate impurity bands previously the source of much trouble. Field strengths were determined from the Zeeman triplet (five-halves normal) of the W λ4660 line. This pattern could be obtained with great sharpness from a brass-tungsten intermittent arc operating in the open air and was standardized against Zn lines in vacuo. Its magnitude between 15000 and 28000 gauss was proportional to the field within ±0.5 percent. The Hg lines λ4047 and 4358 were used as field standards in later parts of the work. Results. Observations of the Zeeman effect have been extended from the bands at λλ5610, 5198 and 4835 to include λλ4511 and 4394. Measurements on thirty patterns with M (theordinalnumberofthelineinabranch)=1, for fields from 18000 to 36000 gauss, have shown the widths of the Zeeman patterns to be proportional to the field strengths to within 2 percent, the estimated accuracy of measurement. The weighted average of the pattern widths for M=1 is 97.7 percent Δνn and M=2, 66.3 percent Δνn, where Δνn is the normal Lorentz triplet half-width. The quantum mechanics predicts 100 percent Δνn and 66.7 percent Δνn for these widths respectively as against 88.9 percent and 64 percent on the old quantum theory. Thirteen out of the possible eighteen patterns predicted for the first two lines of the P, Q, and R branches have been resolved and measured and found to agree with predictions within the error of measurement. The intensity asymmetries reported previously and later treated theoretically by Kronig (Phys. Rev. 31, 195 (1928)) have been found wherever comparison in both high and low fields was possible to behave qualitatively (though not quantitatively) as Kronig's calculations predict. Eight isolated lines (M=23 to M=35) in four bands have been found which show anomalously larze Zeeman effects. Three (and possibly four) new bands have been observed which from their behavior in the magnetic field must belong to the Angstrom group. Their analysis is reserved for another paper.