|I||Categories, Functors and Natural Transformations|
|1||Axioms for Categories|
|5||Monics, Epics, and Zero|
|II||Constructions on Categories|
|III||Universals and Limits|
|VI||Monads and Algebras|
A metacategory is to be any interpretation which satsifies all these axioms...as if that made sense independently of any specific set theoretic context.
any interpretation of the category axioms within set theory.so the notion of category (unlike that of metacategory) is aknowleged to be relative to some set theory.
First he points out that Category Theory is "to discuss properties of totalities" such as the "set" of all groups. These means that category theory would like to be able to use an unrestricted principle of comprehension, were it not that this is known to give rise to problems of consistency (e.g. Russell's paradox).
Consequently we have to settle for separation (subset formation) instead of comprehension and we end up with many interesting metacategories which are not categories. Mac Lane ends up plumping for either Zermelo-Fraenkel set theory with a single universe U thrown in for good measure, or else Gödel-Bernays set theory (where classes serve the same purpose as U does in ZF). He then draws the distinction between small and large categories, the former being the ones whose sets of morphisms and of objects are members of U (or in NBG are sets rather than classes).
So here we have exposed classical set theory as cumbersome for category theory because it lacks unrestricted comprehension. We then put a bit of sticky plaster on it and carry on, making little of the missing categories.