The reduction of echelle data (and fiber-fed echelle data especially) is a quite complicated process since there are many unpredictable effects; they may evoke some instabilities in exposed spectra. As consequences of these, there are problems with merging spectral orders - merged continuum exhibits characteristic wave-like patterns. Of course, these structures complicate broad liner measurement (Hα etc.) - thus it is advisable to describe all complications and effects and minimise them, if possible. We have put main accent to the temperature and torsion influences. We will continue with experiments described in this paper; final results and some summary will be published in future.