Just to give you the update on the technical matters I have modified the Maple-code to reflect the changes in chi2=1 contours when correlations are included. The result is posted at http://www.slac.stanford.edu/~elsen/amb_new.pdf Look at the last plot that shows the contours for no correlation and 80% correlation. If I remember Daniele's slides correctly this agrees well with what he presented in the meeting. The conclusions are quite favorable: the nonlinearity for a is much reduced with large correlations. Mind you this does not address the questions which uncertainties we assume for Lambda_bar and lambda_1 which will be addressed tomorrow. Eckhard