The mpl bind doc: http://www.boost.org/libs/mpl/doc/refmanual/bind.html contains: bind is [skipping] a compile-time counterpart of the similar run-time functionality provided by Boost.Bind and Boost.Lambda libraries. However, further down, it looks like mpl bind creates a metafunciton with the same arity as the original. IOW, the g::apply arity is apparently the same arity as the original f metafunction class. OTOH, the boost.bind doc: http://www.boost.org/libs/bind/bind.html#Purpose shows examples where the arity is reduced by the number or arguments passed to bind. Quoting, it says: bind(f, 1, 2) will produce a "nullary" function object that takes no arguments and returns f(1, 2). To avoid this confusion, the mpl bind doc should make explicit mention that the arity of the resulting metafunction is the same as the original, which is in contrast to the boost.bind's affect on the arity of the result functions w.r.t. the original. Or am I missing something? -regards, Larry