The selection of the most reliable transition frequencies posed few problems since there were relatively few cases where a variety of measurements were reported for the same transition. In cases where this did occur, the selection was based on both the overall consistency of the measurement in question with the other spectral data available, and on the reported uncertainty in the measurements. In nearly all cases the measurements with the smallest reported uncertainties were found to be the most reliable.
The determination of the most reliable molecular constants posed more severe difficulties. Inconsistencies arose in cases where data were reported by several independent workers who studied quite different regions of the spectrum, e.g., molecular beam measurements vs. microwave measurements or centimeter vs. millimeter wave measurements. Since all of the available data had not been analyzed simultaneously in these instances, a complete reanalysis was carried out to eliminate the discrepancies. A second source of difficulty involved the number of parameters determinable from the spectral data available. Many recalculations were carried out to determine the minimum set of molecular constants necessary to fit the observed spectra with the result that in a few cases a set of molecular constants, consisting of fewer parameters than were reported in the original work, was sufficient for the analysis. Such recalculations were performed on all of the Group IV/VI diatomic molecules from CO through PbTe, on most of the Group III/VII molecules, and a number of the heavy alkali halides.
These recalculations allowed the detection of a number of misprints in the literature which had not been detectable by simple inspection of the data set. In addition, several misassignments in quantum numbers were corrected. This was primarily a problem in hyperfine structure assignments on the species AlCl, GaC1, GaBr, GaI, InCl, InBr, and InI where the hyperfine structure was only partially resolved. In these cases accurate calculation of the relative intensities was necessary to determine the spectral assignments.
As a cautionary note, the authors recommend that users of this compilation with a particularly critical application should always refer to the original work referenced here in order to double-check the data of interest. Since it is not feasible to produce a detailed account of the misprints and other errors corrected here, it is also advisable to utilize the equations of Sec. 2, Sec. 3, and Sec. 4 in order to verify that an earlier error has been rectified rather than generated here.