The key problem I face is when cross-references go outside of the current article, since almost certainly then those cited articles would cite some more themselves and you can never get to the bottom of it. Even worse is when the author references something from outside without stating it, an example of which could be using some variable without defining it.
Stephen Hawking believes what Euclid did in culminating mathematical knowledge (into self-contained 13 volumes) is something that is needed again but has never been done in the modern times. His "God Created The Integers" I suppose is a small attempt in that direction.
I would have imagined Eric Wittgenstein's mathematics online encyclopedia (http://mathworld.wolfram.com/) could be such a book. But it has too much backward and "forward" cross-referencing.
I am currently reading Bertrand Russell's Principia Mathematica in my free time.