I've rewritten mpl::math::big_integral_impl so that, instead of being a heavyweight mpl::vector, the mantissa (formerly magnitude) is treated as a nullary metafunction containing the integral constant expressions digit_0...digit_N, where N is one less than the mantissa size, which is now directly configurable. Arithmetic operations are now much faster as a result. The functionality that converts integral_c et. al. to a big_integral_impl is now moved to mpl::math::enlarge_integral. All I need now is a *much* more efficient division algorithm and, of course, mpl::math::big_integral itself. BTW, there may be extraneous files in the <mpl_math.zip>boost/ directory; ignore them, I was using them for study purposes. Cromwell D. Enage __________________________________ Yahoo! Mail - PC Magazine Editors' Choice 2005 http://mail.yahoo.com