I have introduced the relation m_B = m_b + Lambda_bar + lambda/(2*m_B) which changes the a-m_b correlations by a tiny amount: http://www.slac.stanford.edu/~elsen/amb.pdf I would suggest to stick to a variation of +-0.15 for m_b nevertheless. This corresponds to -0.8 for the correlation in lambda and Lambda_bar. Eckhard