Interpretation of the energy spectrum and arrival distribution of cosmic rays is complicated by lack of knowledge of the nature of the primaries. We review claims for the mass composition above 10 17 eV where it can be determined only indirectly from air-shower observables. Difficulties in comparisons between data arise because, inevitably, a set of measurements is interpreted using the best model of hadronic interactions available at the time of analysis. We discuss the situation and conclude that the evidence for a proton-dominated mass composition, even at the highest energies, is unconvincing. However, it may be that there are consistent differences between mass measurements from optical techniques and those based upon other shower observables.
We also find that iron nuclei of ultra high energy can probably escape from the galaxies that host GRBs, possible cosmic ray accelerators. The accelerators must lie nearby.