1: \begin{abstract}

2: We derive a new representation for $U$- and $V$-statistics. Using this

3: representation, the asymptotic distribution of $U$- and $V$-statistics can

4: be derived by a direct application of the Continuous Mapping theorem.

5: That novel approach not only encompasses most of the results on the

6: asymptotic distribution known in literature, but also allows for the

7: first time a unifying treatment of non-degenerate and degenerate $U$- and

8: $V$-statistics. Moreover, it yields a new and powerful tool to derive the

9: asymptotic distribution of very general $U$- and $V$-statistics based on

10: long-memory sequences. This will be exemplified by several astonishing

11: examples. In particular, we shall present examples where weak

12: convergence of $U$- or $V$-statistics occurs at the rate $a_n^3$ and

13: $a_n^4$, respectively, when $a_n$ is the rate of weak convergence of

14: the empirical process. We also introduce the notion of asymptotic

15: (non-) degeneracy which often appears in the presence of long-memory

16: sequences.

17: \end{abstract}