I describe recent work with V. Cirigliano and E. Golowich on the effect of dimension eight operators on weak nonleptonic amplitudes. The basic mesage is that there is an inconsistency in the way that many calculations are traditionally performed. If one calculates matrix elements involving only physics below some scale μ, then one needs dimension eight operators explicitly in the weak OPE. On the other hand if one wants to use dimensional regularization throughout, then one needs all scales to be included within the matrix element, and this results in the same net effect. A numerical estimate indicates that this is important below μ = 2 GeV, and this calls into question many of the models that have been used to predict var epsilon′/var epsilon.