1: \begin{abstract}

2:   We consider continuous-time Markov chains which display a family of

3:   wells at the same depth. We provide sufficient conditions which

4:   entail the convergence of the finite-dimensional distributions of

5:   the order parameter to the ones of a finite state Markov chain. We

6:   also show that the state of the process can be represented as a

7:   time-dependent convex combination of metastable states, each of

8:   which is supported on one well. 

9: \end{abstract}