“…(22), rather than the constrained parameters implied by Dunham's with differences deduced from the only other Raman data, from measurements of Q branches by Bendtsen (5) and by notation (23,24). In the latter case, for example, one conGilson et al (6) (3,8,9,11) for 14 N 2 terms D 10 and D 20 were found to be statistically significant are hindered by the variation in number of fitted parameters and are listed in Table 3. Compared to the two separate fits, in the various fits; in general the major Y kl values are consis-the combined analysis decreased the number of parameters tent but our decreased estimated standard errors imply much from 18 to 12, the values of which can be used for the increased precision.…”