BIS HD gave a good answer to what it means to take a pure mathematical approach at a subject. To add to BIS HD's answer, I talked to a professor today and he explained (essentially) that the study of permutations came about before functions were defined in the way that they are today. And, originally in the study of permutation it was natural to to write a permutation, composed of two seperate permutations (say $\sigma$ & $\tau$), on a set $A$ as
$$A\tau\sigma$$
Meaning morph $A$ into $\tau (A)$ and then morph $\tau (A)$ into $\sigma (\tau (A))$. Later when the study of functions became more developed the common notation $f(x)$ came along which wrote the object to be changed (or morphed) $x$ on the other side of the transformer $f$, in this respect it became common to write things like $(g\circ f)(x)$ which can be read "$g$ follows $f$". This new way of writing compositions, eventually made its way to other part of mathematics including the study of permutation groups. Today, it is not uncommon for an author to write the same permutation above as $\sigma\tau(A)$, opposed to $A\tau\sigma$.
This new style of notation is some what unnatural (if you were not taught this at first), because we read and write from left to right (at least in English). So, some Algebraist prefer use the original notation $A\tau\sigma$, and in the same vein the familiar composition $(g\circ f)(x)=g(f(x))$ is written as $x(fg)$ which naturally says let $x$ be morphed be $f$ and then let then morph the resulting object by $g$.