Selection was applied on the basis of light radius, L90, age and Ns averages over the different runs. Almost 25% of the runs were thus disregarded. As the mentioned quantities seem to suffer from sudden changes from time to time, we have classified the runs in sub-groups for which the averages keep rather stable. A further comparison can be carried out among these different sub-groups.
Following studies were performed:
1/ rlight average vs shower size for 10 and 25 degrees (see for instance postscript file ). The obtained dependency was compared with the MC predictions. Agreement was satisfactory at small shower sizes. At large ones, statistics are missing but a clear tendency upwards is observed.
2/ rlight average vs L90. Similar results.
Both 1/ and 2/ cannot provide us with direct information about chemical abundances near the Knee since energies are mixed for the several primaries, but the real data curve must necessarily move to the MC Fe-like one at large enough shower sizes/L90. Thus we consider 1/ and 2/ as fine checks of consistency.
3/ rlight average vs reconstructed E (Ne+rlight). Similar results. This plot can directly give us information on chemical composition. At low energies (under 1 PeV), it confirms a proton composition, while at high ones missing statistics and strangely high light radii don't allow us to say anything conclusive.
4/ Xmax vs shower size/L90. Agree well with results presented by R.Plaga.
5/ Xmax vs age for different L90 or shower size slices ( get postscript example . Xmax vs age has been plotted for 4 different intervals of log10 L90 -indicated below. ). At low energies, the Xmax vs age law agrees well with that found in MC. But at the region where rlight starts moving up to the MC Fe law in 1/, 2/, 3/ or 4/ the age gets too high. These events clearly don't agree with MC.
We conclude that nothing reliable can be said at such high energies until we solve this problem with the age. In the low energy region, down to the threshold and up to ~1 PeV, results agree pretty well with the MC predictions and a light chemical composition.