Thanks to Art Stone in particular and also Mike Barr, Max Kelly, Dusko Pavolovic and Walter Tholen for their helpful comments, which I still have to study. I wonder whether any specialists in group theory or commutative algebra might be able to find a single example of non-composing regular monos in the category of groups or of commutative rings. Something like: (1) a normal subgroup of a centraliser in a simple group, or (2) A chain of subrings R<S<T such that s # 1 = 1 # s in the tensor S *_R S implies s in R t # 1 = 1 # t in the tensor T *_S T implies t in S but s # 1 = 1 # s in T *_R T does not imply s # 1 = 1 # s in T *_S T This would avoid "theological" objections to discussing on-the-nose colimits in the 1-category of categories. Paul Taylor +++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++