From: saunders@math.uchicago.edu Mac Lane / Moerdijk (Sheaves in Geometry and Logic, Chapter 6, S 10.) Reference *much* appreciated, which I am browsing now. (Peter, will Topos Theory be back in print soon?) It is well known that it is easy to go from sets to categories, harder in the reverse. This is disturbing, since it is the opposite of the answer I came up with myself a couple of days after asking my question. (I am reading "harder" as "necessarily harder" here, my apologies if this was not your intended meaning.) As often in these things, at least some of the problem lies with my question, which was certainly fluffy when I asked it, not knowing then quite what I really wanted to ask. I think I can sharpen it now. Is there a category equivalent to Set for which it is easy to recover the membership relation from the category structure? This is still not a mathematical question as it leaves "easy" open to interpretation. But I think the rest of it is unambiguous. Now I thought I'd answered this sharper question in the affirmative, with an interpretation of "easy" that surely *no* reasonable person could complain about, namely a two line construction of the membership relation. Therefore if my answer was only "fluff", I have made an error somewhere, either in my construction of this version of Set or in my choice of problem. With regard to the latter possibility, I freely admit to knowing less than just about anyone on this list what it feels like to work in a general topos, not enough to write even 6 pages about them let alone 600. My experience of toposes is with one topos only, Set, which makes me about the last person qualified to attempt a contribution to topos theory. However, that my question only concerns the one topos Set gives me hope that, if there is indeed a problem with my construction, it is some technical oversight that I need to repair if possible, and not something to do with other toposes besides Set. In passing, let me again draw people's attention to the fact that I described not just the category Set but its (cartesian) closed structure as well, including complete verification of the coherence conditions. (Not that this was particularly difficult in this case. :) Without giving the full closed structure I do not understand how one can claim to have fully specified which topos one is speaking of. Does the topos literature attend adequately to this detail? (It may well, I just don't know where to find it.) I interpret the reference to Isbell at the end of VII.1 of CTWM as (inter alia) a warning that one cannot take the closed structure for granted merely because it is cartesian closed. If this misinterprets the situation for the cartesian closed case, and coherence is in fact routine there, then my apologies for the misunderstanding. Vaughan Pratt