To obtain absolute accuracy better than a few parts in 106 from FT spectra, it is necessary to determine a multiplicative correction to the wavenumber scale from accurately known internal standard lines. This correction is often specified by a factor, keff, such that
We have used seven Th I lines [DeGraffenreid & Sansonetti (2002)] measured by optogalvanic laser spectroscopy in a low current Th-Ne hollow cathode lamp. These lines, which lie between 13360 cm−1 and 14400 cm−1, have a reported uncertainty of 0.0002 cm−1 (1.4 parts in 108). The calibration derived from these lines (Figure 1) gives a value of keff of (7.20 ± 0.14)×10−7. The error bars are dominated by the low signal-to-noise ratio in the FT spectrum at the extreme edge of the bandpass. Because the calibration lines are limited to a narrow region at the short wavelength end of our spectra, it is desirable to have some check on the results in the long wavelength region to ensure that the value of keff is actually constant across the region of interest as expected from the theory of FT spectroscopy. Comparison with results of [Engleman, Hinkle, & Wallace(2003)] provides such a check, and full details are given in [Kerber et al.(2008)]. The spectrum optimized for wavelengths above 2000 nm was calibrated from the spectrum optimized for the region 800 nm to 2000 nm, with an total uncertainty in keff of 2x 10−8.