We face nowadays the most serious disagreement in the observations of UHECR between the data of Auger and HiRes/TA. While Auger points toward an heavy composition at the highest energies, HiRes/TA shows a proton dominated composition at all energies.
The most self-consistent conclusions on the nature of UHECR are obtained at present by HiRes/TA: they observe a proton dominated chemical composition with a spectrum that shows both expected signatures of protons, i.e. the pair-production dip and the GZK cut-off. Moreover, in the HiRes data the GZK cut-off is also found in the integral spectrum with the expected value of the E1/2 energy scale. The picture emerging from HiRes/TA data confirms the dip model  and places the transition between galactic and extra-galactic CR at energies around 1018 eV in very good agreement with galactic CR observations. The only discrepant feature of the HiRes/TA observations consists in the absolute lacking of any anisotropy signal, expected at the highest energies in the case of a pure proton composition.
The Auger data are quite different. Measurement of chemical composition shows a steadily increasing mass starting from energies around 3 EeV, observation recently confirmed by independent data on muon content in the showers . The Auger chemical composition excludes the dip model and the observed high energy steepening as the GZK cut-off. At present there is no nuclei based model which explains simultaneously the Auger energy spectrum and chemical composition, apart from the model proposed in  based on an unlikely framework with a nearby source (70 Mpc) that injects only heavy nuclei with a very flat spectrum (g = 1.6).
We can conclude that the key issue to distinguish among different models in UHECR physics is related to the measurement of chemical composition. The best method at present to determine the UHECR composition is given by the measure of the position of the maximum of the cascade developed in the atmosphere, i.e. the elongation curve < Xmax > (E). In figure 10 we plot the elongation curve as observed by Auger  and in figure 11 the same quantity measured by HiRes .
The calculated reference values for proton and iron plotted in figures 10, 11 are those computed in the framework of the QGSJET01 and QGSJET02 models .
Unfortunately, the determination of chemical composition through < Xmax > (E) suffers from many systematics due to the experimental approach and uncertainties in the interaction model. Systematic errors in the < Xmax > measurements can be as large as 20-25 g/cm2, to be compared with the difference of about 100 g/cm2 between < Xmax > of proton and iron. A better sensitivity to distinguish different nuclei is given by the width of the Xmax distribution, i.e. RMS(Xmax) .
We conclude stating that renewed experimental efforts are needed to reach a more reliable determination of the chemical composition of UHECR solving the apparent contradiction in the Auger and HiRes/TA data.
I'm grateful to V. Berezinsky, P. Blasi, A. Gazizov and S. Grigorieva for our joint activity in the field of UHECR physics.