1: \begin{abstract}

2: Adaptive Monte Carlo methods can be viewed as implementations

3: of Markov chains with infinite memory.  We derive a general

4: condition for the convergence of a Monte Carlo method whose history 

5: dependence is contained within the simulated 

6: density distribution.  In convergent cases,

7: our result implies that the balance condition need only be

8: satisfied asymptotically.  As an example, we show that 

9: the adaptive integration method converges.

10: \end{abstract}

11: