1: \begin{abstract}

2: We study the motion of a random string in a convex domain $O$ in $\R^d$,

3: namely the solution of a vector-valued stochastic heat equation, confined

4: in the closure of $O$ and reflected at the boundary of $O$. We study the structure of

5: the reflection measure by computing its Revuz measure in terms of

6: an infinite-dimensional integration by parts formula. Our method exploits

7: recent results on weak convergence of Markov processes with log-concave

8: invariant measures.

9: \end{abstract}

10: