As described in Sect. 2.2, as a consequence of the adopted telecentric mounting, the final focal plane is spectrally dishomogeneous, with wavelength fluctuations of the instrumental profile amounting to 13.3 mÅ peak to peak (728 ms-1) at 5500 Å. The Cd spectral lamp, however, can be used to measure such fluctuations, allowing to account for them a posteriori.
Referring to Fig. 4 (click here), when the mirror M10 is on the optical path and M5 is out, the lens L4 forms on CCD1 an image of the diaphragm D2, lighted by the lamp radiation. A series of images taken while the interferometer is scanning the Cd red line allows then to measure on each pixel the line peak position; as reference in this case the PMT2 photomultiplier is used. By evaluating on each pixel the line wavelength shift relative to the mean on the overall field, a map of the spectral dishomogeneities is obtained. A map like this, as a sort of spectral flat field, allows then to correct the solar images for flatness and parallelism errors.
Ten maps have been sequentially obtained, and a
mean error of 
0.2 mÅ ( 9 ms-1)
in evaluating the wavelength shift has been found.
We may conclude therefore that this procedure allows
to measure and correct the spectral dishomogeneities of
the image plane with a precision better than 10
ms-1.
The same maps, which can be considered as a representation
of the interferometer cavity errors, can be also used to
evaluate the surface accuracy. A peak to peak variation of
16 mÅ on the line position has been found, corresponding to flatness
errors of 
/146 at 5461 Å, very similar to the
/150 value, claimed by the manufacturer.