THEORETICAL UNCERTAINTIES IN THE DETERMINATIONOF s FROM HADRONIC EVENT OBSERVABLES

I review the theoretical status of hadronization corrections to hadronic event ob-servables and discuss their impact on the determination of s. Precise determination of the strong coupling has become one of the most important tasks of the QCD theory and phenomenology 1. From the theoretical point of view the cleanest measurement comes from the total e + e ? annihilation cross section at high energies. The accuracy is dominated in this case by statistical errors; high statistics is needed because the eeect proportional to s is a few percent fraction only of the total cross section. Thus, going over to various event shape observables which are proportional to s at the leading order presents a clear experimental advantage. The price to pay is that the QCD description becomes more complicated and making a defensible estimate of the theoretical accuracy presents a nontrivial task. The uncertainty due to nonperturbative eeects is a particularly delicate issue. From the experimentalist's point of view this is the uncertainty of`hadronization corrections' which are applied to uncover the structure of the event at the parton level from the observed structure at the hadron level. The hadronization process is modelled in a certain way, and diierences between models (say, Lund string fragmentation or parton showers) are taken to estimate the error. In statistically important regions the hadronization corrections are of order 10% (at the Z peak) while the claimed error is of order 2{3%. This is less than the uncertainties of existing perturbative calculations (estimated by the scale dependence) so that for an experimentalist the hadronization could be considered under control. This procedure is very successful phenomenologically, but it is unsatisfactory from the theoretical point of view. In particular | I will return to this point later | the separation of perturbative versus nonperturbative alias 1