The very early work on the spectrum of bromine carried out by Exner and Haschek, Eder and Valenta, Asagoe and others is summarized in Kayser’s “Handbuch”, vol. 8. The first extensive investigation on the spark spectra of this element was made by L. and E. Bloch (1927), who gave an accurate and nearly complete list of lines belonging to Br II, Br III and Br IV. The source adopted by them was an electrodeless discharge through the vapour of bromine, and the allocation of the lines to the different spectra was made by a study of the long and short lines produced on account of the radial distribution of the ions in an electrodeless discharge tube. The length of the series spark gap was also varied to vary the intensity of excitation. The wave-lengths thus recorded by them extended from λ 6550 to 2000. Vaudet (1927) measured the spectrum in the region λ 2200-1200. In a recent investigation Lacroute (1935) photographed the spectrum with a metre vacuum-grating spectrograph and published a list of lines extending between λ 2300 and 600. The analysis of the spectrum of doubly ionized bromine appears, however, to have been attempted so far only by Deb (1930), who published the classification of about 200 lines assigned to Br III by L. and E. Bloch. Making an extensive use of the method of horizontal comparison suggested by Saha and Mazumdar (1928), Deb predicted the positions of the important combination lines of Br III and set up a term scheme depending on what he considered to be recurring frequency intervals among the lines of Br III. But it is extremely doubtful whether his classifications and the term scheme are correct, as he based his predictions on extrapolation from erroneous assignments of Se II, made by Bhattacharyya (1929), and as the discrepancies in frequency differences allowed in the classification are very much larger than what the possible inaccuracies in the measurement of the wave-lengths would permit. The discrepancies amount in many cases to even more than 4 or 5 cm. -1 in the near ultra-violet region. In a previous communication (Krishnamurty and Rao 1935; cf. also Martin 1935) the structure of the spectrum of singly ionized selenium was described. It was found to consist of doublet and quartet systems of terms based on the 3 P, 1 D and 1 S states of the Se III core and to resemble in all its general features the spectrum of the iso-electronic atom of arsenic, i. e. As I (Rao 1929, 1932). In the course of the above investigation attempts were made to correlate the structures of the three spectra As I, Se II and Br III, which arise from the same outer electronic constitution. The correlation clearly indicated that the analysis of Br III, published previously by Deb, was completely out of step in the above sequence and must therefore be erroneous, particularly when a similar comparison of the principal features of the two other corresponding sequences, N I, O II, F III, Ne IV, and P I, S II, Cl III, A IV, revealed several striking regularities. A thorough reinvestigation of the spectrum of bromine has been undertaken, and the present paper deals with the identification of the important quartet and doublet terms of Br III. A preliminary report of the results appeared in Nature (Rao 1936). The relative value of these two sets of terms is yet uncertain. A few intercombination lines have been tentatively identified, but the publication of these is withheld pending an investigation, now in progress, of the structure of Br II and of the higher configurations of Br III. These will form the subject of a future paper.