Limits on higher-twist contributions to deep-inelastic scattering

It is shown that second moments of $F_2$ and $x{F}_3$ combined with energy-momentum conservation strongly restrict the magnitude of any single higher-twist contribution or of a combination of arbitrarily many higher-twist terms each contributing with equal signs. These effects turn out to be reasonably small for $0.1\lesssim {\alpha }_s<3\mathrm{}$, leaving all standard radiative QCD $\ln Q^2$ tests of scaling violations unchallenged, in particular those based on second moments of structure functions. Furthermore, a more general and realistic combination of two higher-twist terms (i.e., twist-4 and twist-6 with arbitrary signs) is studied via an $x$- and $Q^2$-dependent analysis of SLAC and CERN-Dortmund-Heidelberg-Saclay data. A combination of physically reasonable twist-4 and twist-6 terms alone cannot explain the observed scaling violations and best fits to the data are obtained by allowing at most 20% higher-twist corrections to the dominant perturbative QCD predictions. Such general higher-twist combinations allow $\Lambda$ also to increase as compared to pure QCD fits or as compared to keeping just a single higher-twist correction, with the result that $\Lambda$ can vary in the range $0.2\lesssim \Lambda \lesssim 0.4$ GeV.