1: \begin{abstract}

2:  We study asymptotic

3: percolation as $N\to \infty$ in an infinite random graph

4: ${\mathcal G}_N$ embedded in the hierarchical group of order $N$,

5: with connection probabilities depending on an ultrametric distance

6: between vertices. ${\mathcal G}_N$ is structured as a cascade of

7: finite random subgraphs of (approximate) Erd$''\kern -.20cm{\rm

8: o}$s-R\'enyi type. However, the results are different from those

9: of classical random graphs, e.g.,  the average length of paths in

10: the giant component of an ultrametric ball is much longer than in

11: the classical case. We give a criterion for percolation, and show

12: that percolation takes place along giant components of giant

13: components at the previous level in the cascade of  subgraphs for

14: all consecutive hierarchical distances. The proof involves a

15: hierarchy of ``doubly stochastic''  random graphs with vertices

16: having an internal structure and random connection probabilities.

17: 

18: \end{abstract}

19: