The measurements of the lower spin component were taken from
(1) M. C. McCarthy, A. L. Cooksy, S. Mohamed, V. D. Gordon, and P. Thaddeus, 2003, Astrophys. J. Suppl. Ser. 144, 287.
One unresolved HFS doublet was omitted because of a larger residual. All other unresolved HFS doublets were treated with 2 kHz uncertainties instead of the reported 5. The calculation should be viewed with caution above 80 GHz. In addition, the hyperfine structure splitting should be viewed with some caution below 7 GHz.
The partition function takes into account the higher spin component as far as possible at present.
The dipole moment is from a quantum chemical calculation in (1).