"It frequently happens that in studying the properties of a mathematical system S we come to recognize that the results obtained do not depend essentially upon the precise definition of S but only upon certain formal relations holding between its elements. In that case it may be useful to select a suitable set R of such relations and then consider a system Σ of entities of which nothing is presupposed except that they, too, satisfy all relations in R. The system Σ constructed in this way is said to be derived from S by a process of abstraction. If we now investigate the properties of Σ, the conclusions at which we arrive will, of course, hold for S, since S is a special case of Σ. In this way we obtain results more general than those relating to the system S only; and we are able, moreover, to distinguish between the basic structure of S, and its more superficial features."
- An Introduction to Linear Algebra, by L. Mirsky (OUS 1955) (italics by author, links by me (natch))
And they say mathematicians don't have a sense of humour...