THE USE OF TERM-VALUE FITS IN TESTING SPECTROSCOPIC ASSIGNMENTS

The use of a least-squares term-value fit of a spectrum as a means of testing assignments is described. In a direct term-value fit the norma l matrix is singular, with one zero eigenvalue for each set of connect ed levels. These equations are solved by a method which gives the term values of each connected set relative to their mean. The standard dev iation sigma from a term-value fit should be equal to the measurement error. It is shown that, as a test of individual assignments, it is po ssible to calculate from the results of a fit the fractional change ta u(l) = (sigma(l) - sigma)/sigma when each line l is omitted singly fro m the fit. Anomalously large negative values of tau(l) indicate lines that may be misassigned or mismeasured. As an example, it is shown tha t the standard deviation of an unweighted term-value fit of a publishe d compilation of 294 lines of the H-3+ ion can be reduced from 0.031 c m-1 to 0.004 cm-1 by the omission of 7 lines. (C). 1994 Academic Press , Inc.