A peculiar prediction of perturbative QCD, with the assumption of Local Parton Hadron Duality, is that the difference in mean hadron multiplicity between heavy and light quark initiated events in e+e- annihilation, δQℓ, is energy independent. In the Modified Leading Logarithmic Approximation (MLLA), the numerical value of the constant is derived in terms of a few experimentally measurable quantities. While the energy independence of δQℓ has been succesfully verified experimentally for b-quarks up to the highest LEP2 energy, the numerical prediction originally published in 1992,
, clearly overestimates the experimental results. In this work, done in collaboration with Yuri L. Dokshitzer, Valery A. Khoze and Wolfgang Ochs, we show that the original MLLA prediction needs a revision, in the light of new experimental results and the improvement in the understanding of the experimental data. We now find
, in better agreement with experiment, and we show that the remaining difference can be attributed largely to next-to-MLLA contributions, an important subset of which are identified and evaluated. The situation for charmed quarks is also reviewed.