I notice the cosh bug has not been fixed in Clean 2.1, i.e. cosh = 0.5 * (exp(x) + exp(~x)) and not 2.0*(exp(x) + exp(~x)) cheers, G