“…The results agree (by design) with the picture that all terms in L 4 are consistent with the values obtained by integrating out light resonances, and are inconsistent with m u = 0[30]. In particular, a conventional estimate of this kind gives[33,34,35,36] 10 3 L 8 ≈ 0.9 ± 0.3, 10 3 L 7 ≈ (−0.4 ± 0.2), while the hypothesis m u = 0 would require 10 3 L 8 = −0.4 ± 0.3, and 10 3 L 7 = 0.2 ± 0.2.Thus any method which can directly determine the L's, even with errors as large as 100%, can distinguish between these two possibilities.3 m u on the latticeSimulating QCD on a lattice is, in principle, the most reliable way of determining quantities which are not predicted from symmetry alone. For instance, although it is only possible to constrain ratios of light quark masses phenomenologically, recent lattice computations of the hadron spectrum have suggested an absolute range for the strange quark mass[37,38,39,40,41,42,43,44,45,46] and the parameter B.…”