This wiki page says
In probability theory and statistics, the marginal distribution of a subset of a collection of random variables is the probability distribution of the variables contained in the subset.
Is the subset corresponding to a marginal distribution could be any non-empty set of the original set?
Assume a n-variables joint distribution, How many marginal distribution could be there?
Obviously $2^n$, by this definition. But 2 of those cases, the empty subset and its complement, might raise eyebrows, so one could say there are $2^n-2$ non-trivial marginal distributions.