The signals produced by galactic cosmic ions through ionization processes will be used for independent monitoring of the stability of the energy scale at high
energies: the energy deposit per crystal ranges from ~500 MeV for carbon ions to ~8 GeV for iron.

The spectrum on Fig.~14 has been obtained by selecting crystal hits in low-multiplicity layers (to get rid of the bulk of nuclear interactions) and by
correcting hit energy from the track path-length in the crystal. As can be seen, very narrow peaks (the carbon peak has a 5\% only width) corresponding to
individual species can be identified with good statistics. The statistiscs decreases with the atomic number because of the abundances and of the decreasing
efficiency of the onboard heavy-ion filter.

Figure~15 shows that the carbon peak position is stable within 1\% or less (for the whole calorimeter) on timescales of days. In order to improve the accuracy
of this method and to apply it to all available species in the future, one needs a tighter selection to better reject nuclear interactions, which will decrease
the event rate. Therefore, we will monitor groups of crystals having the same layer and tower position (central, corner or side), instead of individual
crystals. Light elements will be used to monitor the high-energy scale calibration on timescales of weeks or months, and heavier elements for longer timescales.

---

Fig.~14 -- fig1.eps
Fig.~15 -- fig2.eps

Caption:

(a) Spectrum of crystal hit energy, obtained for all calorimeter crystals and 4 days of on-orbit operations. The hit energy is defined as a vertical equivalent
energy, which is obtained as the hit raw energy multiplied by the ratio of the crystal height to the measured trajectory path-length (as given by the main track
reconstruction). The first two peaks (Be and B) are secondary peaks produced by nuclear interactions.

(b) Time history of the carbon peak energy over several days.