Abstract. New results on particle correlations and event-by-event fluctuations presented at Quark Matter 2004 are reviewed.
Event-by-event fluctuationsNon-statistical event-by-event fluctuations have been proposed as a possible signature for the QCD phase transition [1,2]. The passage of the system close to the critical point of the QCD phase diagram might be indicated by a non-monotonic evolution of fluctuations as function of beam energy. These exciting predictions triggered an extensive study of event-by-event fluctuations at SPS and at RHIC.
Fluctuations of the mean transverse momentumThe existence of non-statistical event-by-event fluctuations of the mean transverse momentum M pt at SPS and RHIC is by now well established. However, these dynamical fluctuations are small in central collisions, typically about 1% of the inclusive mean transverse momentum p t , and only weakly depending on √ s N N , see Fig.1 (left panel) [3,4]. This value is roughly compatible with the extrapolation of fluctuation measurements in p-p collisions [5] under the assumption of an independent superposition of particle sources in A-A. In particular, no indication for a non-monotonic beam energy dependence has been found so far. Despite the absence of a 'smoking gun' signature for the phase transition or the critical point, the systematic study of M pt fluctuations gives valuable insight into the particle production mechanism and the dynamic evolution of the system which cannot be extracted from inclusive distributions. A quantitative study requires an appropriate formalism which facilitates a comparison of results among different experiments and to theory. Presently, an unfortunate situation has arisen: Practically each experiment uses different measures for fluctuations, see [4,6,7,8,9,10]. These measures have very different sensitivities to particular experimental conditions, such as track quality cuts, tracking efficiency, and acceptance. In this sense, measures which are most closely related to single and two-particle densities appear preferable since they are the least sensitive to trivial efficiency effects [9]. In this situation, it is mandatory that