If, in some problems, one has to deal with the ``product'' of distributions
$\rm f_i$ (also called generalized functions) $\rm\bar T = \Pi^m_{i=1} f_i$,
this product has a priori no definite meaning as a functional $(\rm \bar T,
\phi) $ for $\rm\phi \in S$. But if $\rm x^{\kappa +1} \Pi^m_{i=1} f_i$ exists,
whatever the associativity is between some powers $\rm r_i$ of $\rm x$ ($\rm
r_i \in \Bbb N, \sum_i r_i\leq \kappa +1, r_i \geq 0$) and the various $\rm
f_i$, then a continuation of the linear functional $\rm \bar T$ from $\rm M$
onto $\rm S^{(N)}$ for some $\rm N$ is shown to exist in such a way that $\rm
x^{\kappa +1} \bar T$ is defined unambiguously, and $\rm (\bar T, \phi), \phi
\in S$, significant, though not unique.