Abstract: Elsewhere I’ve argued in favour of a ‘mutual parthood’ approach to constitution puzzles. Recently this approach has come under attack (Walters 2019). I respond to the main lines of objection, leaning on the distinction between inclusion, outstripping, and non-identity conceptions of parthood. I argue that the mutual parthood approach is useful to understand failures of extensionality where the inclusion conception and co-location are coordinated by underlying “spaces”, and the dominant mereological metaphor is one of “carving” that space. But the mutual parthood approach is less helpful in understanding failures of extensionality where the non-identity or outstripping conceptions are correlated with “places”, and the dominant mereological metaphor is one of “building” from occupants of those places. Along the way, I propose a formal framework for modelling the three main approaches to constitution, showing where the approaches are definable in terms of the others.