1: \begin{abstract}

2:   There are several common ways to encode a tree as a matrix, such as

3:   the adjacency matrix, the Laplacian matrix (that is, the

4:   infinitesimal generator of the natural random walk), and the matrix

5:   of pairwise distances between leaves. Such representations involve a

6:   specific labeling of the vertices or at least the leaves, and so it

7:   is natural to attempt to identify trees by some feature of the

8:   associated matrices that is invariant under relabeling. An obvious

9:   candidate is the spectrum of eigenvalues (or, equivalently, the

10:   characteristic polynomial). We show for any of these choices of

11:   matrix that the fraction of binary trees with a unique spectrum goes

12:   to zero as the number of leaves goes to infinity. We investigate the

13:   rate of convergence of the above fraction to zero using numerical

14:   methods. For the adjacency and Laplacian matrices, we show that that

15:   the {\em a priori} more informative immanantal polynomials have no

16:   greater power to distinguish between trees.

17: \end{abstract}

18: