The ISGRI count rate for the Crab calibration source does indeed vary over the years by several percents as shown in the plot on the right produced with OSA 8 (courtesy: Erik Kuulkers). The count rate (count/s) does not need to be constant, but what is important is that the count rate divided by the instrumental effective area (count/s/cm²) is constant. The variations in the raw count rate of the Crab are balanced by different ancillary response functions (ARF) — describing the energy dependence of the effective area — to be used for different epochs during the mission. The ARF correction is based on the assumption that the Crab spectrum is constant over time and limits the observed fluctuations in Crab fluxes to about 1-2%.

These small variations over time will of course also be present for any other source than the Crab. One shall therefore be careful when analysing variable sources over periods of several years. For spectral analysis, the OSA software will do an average of the different ARFs that correspond to the different epochs of the analysed dataset. This might introduce a bias of at most 1-2%. A way to avoid this is to perform separate spectral analyses for epochs of different ARF responses before averaging the results.

Note that recent evidence that the Crab flux has decreased by 7% in about 2 years (Wilson-Hodge et al. 2010, arXiv:1010.2679) implies that the Crab cannot be considered as constant anymore and therefore the use of Crab calibrations to tune the ARF shall be reconsidered. [see also IBIS-019]