[andrepd]

That's not at all the point I was trying to make. Indeed, if you check my other comment, both the books I give as an example have precisely the feature of being self-contained. Landau's Mechanics starts with the basic concept of a particle. Cohen-Tannoudji's Quantum Mechanics starts with the simplest possible description of the double-slit experiment. Again, they assume no prior knowledge about what they are trying to teach (wouldn't be much of a textbook about the subject if it did require previous knowledge about the subject, now would it? :))

[pcein]

The counterparts of these books in the mathematical domain may be self-contained with respect to the subject matter, but you fail to consider the fact that most of them require a high degree of mathematical sophistication and are really targeted at people who have already acquired the required level of intuition. The "self-contained" nature of these books is deceptive - they might start with some innocent looking stuff about counting, integers etc and very soon drill deep into abstract stuff through complex chains of reasoning which a beginner will find extremely dry, bewildering and unfathomable. When you are a total beginner and trying to learn a subject through self-study, you are more likely to appreciate a book in which the author tries to recreate the experience of sitting in a classroom and listening to a professor holding your hand and showing you stuff from various angles.