Many writers refrain from starting an introductory work with a definition of the subject and do so for good reason. An accurate and concise enough definition at the start of a work would probably have to be too general, abstract or technical to communicate anything of much value to the beginning reader; while one which avoids these features would tend to become so elaborate and burdened with illustrations as to seem redundant, if the main body of the work is to follow anyway. There is indeed a real difficulty here and its source lies deep. For talk about any subject is, in general, not part of that subject. A definition of science would not be science; just as a definition of music would not be music. Even a practicing scientist attempting to write about his subject would soon find himself using language not ordinarily called for when writing in his subject. Thus, in talking about physical theory he might, for example, take up such matters as the general structure of abstract deductive systems, the relation of such systems to physical subject-matter, the nature of scientific models, and much more - topics which have come to be thought of as more properly belonging to semantics, applied logic or, more generally, to the philosophy of science rather than to natural science itself.