MS> Ah, now I see what you meant. Unfortunately, this approach works MS> only with self-quoting values for BAR: Holger> What Common Lisp are you using ? CLISP, indeed. Holger> USER(8): (setq bar #'(lambda (x) x)) Holger> #<Interpreted Function (unnamed) ＠ #x8cf772> Holger> USER(9): (defmacro foo () bar) Holger> FOO Holger> USER(10): (let ((bar #'(lambda (x) (+ x 1)))) (funcall (foo) 23)) Holger> 23 Holger> This is Allegro Common Lisp 4.3. I think that it should be pretty Holger> close to ANSI Cl requirements. I bet CMUCL is somewhat more Holger> explanative than ACL (note however that CMUCL as of 17f does not Holger> implement ANSI CL, although 18 is said to move towards ANSI Holger> compliance). The fact that one Common Lisp allows this is irrelevant. The fact that the standard doesn't specify (or requires) an external representation for closures is. I believe CLISP is entirely correct in generating the error. Here's another unintended effect which exposes the hackishness of this, this time relying on the fact that symbols are not self-quoting: (setq bar (intern "Foo")) (defmacro foo () bar) (cons (foo) '()) => variable |Foo| has no value Again, the problem is that BAR is not referring to a binding---it's referring to the value of the binding. -- Cheers =8-} Chipsy Friede, Völkerverständigung und überhaupt blabla