The analysis of the U III spectrum has been delayed for many years by lack of precision in wavelengths and intensities. As early as 1946, Schuurmans (46.3) had obtained Zeeman data for the lines at 498.99 and 586.04 nm. Later on, the spectra of ionized uranium have been observed from sliding spark discharges by Morozova and Startsev (57.4), H. Crosswhite and H.M. Crosswhite (82.3) and Conway and van Deurzen (unp.14), as well as from uranium plasmas by Korostyleva and Dontsov who have assigned to U III around 1860 lines between 245.9 and 581.8 nm (78.8), by Berg et al. (80.1) and Schneider and Roxey (82.1).
In 1984 the analysis of U III was started by Palmer and Engleman (84.4) from a list of some 750 lines between 277.8 and 900 nm, obtained by comparing the Fourier transform spectra of uranium in hollow cathodes filled either with a helium-argon mixture or with neon. The energies of 22 low-lying odd levels and 36 high-lying even levels were tentatively determined, but only the two lowest levels could be identified as 5f36d 5L6 and 5K5, according to the predictions of Brewer (71.2).
A number of infrared lines have been assigned to U III by Blaise from the comparison of the Fourier transform spectra emitted by an electrodeless discharge tube (no U III lines observed) and by a hollow cathode filled with argon. These transitions led to establish the ground level 5f4 5I at 210.265 cm-1 below 5f36d 5L6 (84.7).
The 238U - 235U isotope shifts and the hyperfine structures of 235U recently measured by Engleman (unp.18) and, between 1.8 and 5.5 µm, by Conway and Worden (84.2), have confirmed the interpretation of several levels of 5f36d and of the five lowest levels of 5f37s based on a parametric study of 5f3(6d+7s) by Wyart, following the Slater-Condon method (87.7).
Unfortunately, a number of J-values are still ambiguous, several level isotope shifts are dubious and among the tentative levels of the original analysis (84.4), 10 odd levels and 5 even levels have not been confirmed.