Hoi, last night I got about the same plots like you, Ric, with two small changes (lambda1 +/- 0.12 which is what Sven said (I think), and taking mB = mb - LambdaBar): http://www.slac.stanford.edu/~ursl/talks/021803/lbl1-mba.eps.gz which shows the chi2. This is also with rho = -0.8, but I don't think that Eckhard's rho is positive, after all the original ellipsis is tilted in the right way. It must be some "-" sign somewhere. Operationally (or in the near future), I guess we might just as well use mb = 4.65: BRBR = 0.0228688 +- 0.00292074(stat) +- 0.00117709(MC stat) mb = 4.80: BRBR = 0.0197019 +- 0.00253248(stat) +- 0.0010378(MC stat) mb = 4.95: BRBR = 0.0163534 +- 0.00215787(stat) +- 0.000838845(MC stat) which is a 16.8% variation. If I ran the mb variation fit correctly... This is still "too good". I don't yet know whether going to leading order will change the correlation between LambdaBar and lambda1 Cheers, --U.