Nearly a decade ago, the science community was introduced to the $h$-index, a proposed statistical measure of the collective impact of the publications of any individual researcher. It is of course undeniable that any method of reducing a complex data set to a single number will necessarily have certain limitations and introduce certain biases. However, in this paper we point out that the definition of the $h$-index actually suffers from something far deeper: a hidden mathematical incompleteness intrinsic to its definition. In particular, we point out that one critical step within the definition of $h$ has been missed until now, resulting in an index which only achieves its stated objectives under certain rather limited circumstances. For example, this incompleteness explains why the $h$-index ultimately has more utility in certain scientific subfields than others. In this paper, we expose the origin of this incompleteness and then also propose a method of completing the definition of $h$ in a way which remains close to its original guiding principle. As a result, this "completed" $h$ not only reduces to the usual $h$ in cases where the $h$-index already achieves its objectives, but also extends the validity of the $h$-index into situations where it currently does not.