The differences between the RSRF obtained in the lab and in orbit are twofold:
(1) The overall shape agrees mostly within the current uncertainties of the models and the observations. The only place where the discrepancies are so big that a correction of the calibration makes sense at this stage, is at the short-wavelength edges of bands 1a and 2a. The relative responsivities as measured in the lab appeared up to 60% higher than in orbit. This can be explained as a leakage problem since the temperature of the calibration source was only 300K.
(2) In the lab only band 3 showed fringes, while in orbit also bands 1 and 2 show fringes. The filters in the detector block have been identified as the cause of these fringes. The thickness of these filters matches well with the Fabry-Perot gap calculated from the observed fringe frequency. The width of the fringes is close to the resolution limit of the instrument. This is the reason why the effect was not observed in the lab: the external black body source was extended to ISO-SWS and thus the resolution of these measurement was lower.