I consider two logically independent definitions of (mereological) sum identity when x is a sum of the ys and w is a sum of the zs. Def 1 x=y: every part of every y shares a part with some z, and every part of every z shares a part with some y. Def 2 x = y: all the ys are zs, and all the zs are ys. Neither allows a sum to change its parts. Peter van Inwagen tells a (...) story about a brick house that persists through the loss of a brick (The Journal of Philosophy 2006: 614–30). His claim that this story illustrates the persistence of a sum through the loss of a part conflicts with other unassailable assumptions. An understanding of sum identity that does not allow a sum to change its parts if helpful in thinking about material composition. (shrink)