Target independence of the EMC-SMC effect

\noindent{\bf Abstract} \vskip0.2cm \noindent An approach to deep inelastic scattering is described in which the matrix elements arising from the operator product expansion are factorised into composite operator propagators and proper vertex functions. In the case of polarised \m p scattering, the composite operator propagator is identified with the square root of the QCD topological susceptibility \sqrt{\chi^{\prime}(0)}, while the corresponding proper vertex is a renormalisation group invariant. We estimate \chi^{\prime}(0) using QCD spectral sum rules and find that it is significantly suppressed relative to the OZI expectation. Assuming OZI is a good approximation for the proper vertex, our predictions, \int_{0}^{1}dx g_1^p (x;Q^2=10\GV^2)= 0.143 \pm 0.005 and G^{(0)}_A \equiv \Delta \Sigma = 0.353 \pm 0.052, are in excellent agreement with the new SMC data. This result, together with one confirming the validity of the OZI rule in the \hp radiative decay, supports our earlier conjecture that the suppression in the flavour singlet component of the first moment of g_1^p observed by the EMC-SMC collaboration is a target-independent feature of QCD related to the U(1) anomaly and is not a property of the proton structure. As a corollary, we extract the magnitude of higher twist effects from the neutron and Bjorken sum rules.