The figure below shows the effect of of the new SRF shape on the 20 cm air gap data from 9 Nov 1998. Only the SRF width was adjusted; the overall SRF shape was fixed apart from stretching for the width. We also fit for a pathlength/optical depth scaling factor (close to unity) to account for uncertainty in the exact gas amount in the air gap path.
The top plot shows the air gap measured data and fit transmittance. The bottom plot shows the error in the fit transmittances using the new SRF shape (red) and old (thick gray) SRF.
The new SRF is our best fit of the "PHS386" M12 Bruker data average SRF, while the old SRF is the May '96 SRF. The new SRF fit is a little better, mainly in the region near the Q-branch at 667 cm-1. The RMS of the two fits are 0.0063 (new) and 0.0075 (old). The new fit also returned a fit SRF width that is within around 3% of the width determined from the average of the Bruker SRF measurements, while the old fit width differs by more than 21%.
We have been told there were small adjustments to some of the optics between the time the air gap and "386" Bruker data was recorded, so this may limit the reliability of comparing results from the two data sets. Still, it is encouraging to see roughly consistent results in two independent measurements.
The air gap data can not tell us much about the SRF shape. At best it can augment the Bruker data to provide some independent information on the channel center frequencies and some rough limits on the SRF wing magnitude and the SRF width. Gas cell data would be better than air gap data, but air gap data is better than nothing.