Mrst partons and uncertainties

We discuss uncertainties in the extraction of parton distributions from global analyses of DIS and related data. We present conservative sets of partons, at both NLO and NNLO, which are stable to x, Q 2 , W 2 cuts on the data. We give the corresponding values of αS(M 2 Z) and the cross sections for W production at the Tevatron. The parton distributions of the proton, which are currently determined from global analyses of a wide range of DIS and related hard scattering data, are subject to many sources of uncertainty. There are uncertainties due to (i) the experimental errors on the data that are fitted in the global analysis, (ii) the choice of data cuts (W cut , x cut , Q 2 cut), defined such that data with values of W , x or Q 2 below the cut are excluded from the global fit, (iii) the truncation of the DGLAP perturbation expansion, (iv) specific theoretical effects, such as ln 1/x, ln(1 − x), absorptive and higher-twist corrections, and (v) input assumptions, such as isospin invariance, the choice of parameterization, heavy target corrections and the form of the strange quark sea. So far, attention has been focussed on the uncertainties arising from the experimental errors; see Refs. [1, 2] for estimates based on global NLO analyses. However, Ref. [3] concentrates on the remaining uncertainties, (ii)–(v); here we present some results from this forthcoming paper. In principle, if the DGLAP formalism is valid and the various data sets are compatible, then changing the