High-end performance or outlier? Evaluating the tail of scientometric distributions

The present paper attempts to shed light on outstanding research performance using the example of citation distributions. In order to answer the question of how the analysis of outstanding performance, in general, and highly cited papers, in particular, could be integrated into standard techniques of evaluative scientometrics. Two general methods are proposed: One solution aims at quantifying the performance represented by the tail of citation distributions independently of the “mainstream”, the second one, a parameter-free solution, provides performance classes for any level. Advantages and shortcoming of both methods are discussed.