Left: COSAC measurements #1 to 5 (Table 1) corrected for instrumental backgound. Spectrum #0 was used for background correction. The noise is about 10 ion counts per integer m/z bin. Right: same for Ptolemy. Corrections for instrumental background and CO₂ fragmentation were applied (see text for details). Exponential slopes were fitted to the data (Table 3). As for Fig. 2, the time axis is set to zero at Philae’s touchdown.