Maxwell's book on electromagnetism, it shows his hindsight perspective on interpreting the phenomena. One can look at the papers and works he wrote that led up to it, but they are full of detours and unbalanced attention with small dead ends. When he writes his book he tries to convince the audience of his time in one comprehensive work.
What I love especially, is that he is very careful and systematic about his conclusions, splitting up in cases, instead of simply producing the solution and then proving it is a valid solution.
For example he does not assume that every point in space has a single value for total potential. First he describes how it is at least theoretically conceivable to follow a path and observe the total potential to vary continuously, and arrive back at the point of departure but end up with a different value of the total potential. He is effectively describing the possibility of wormholes (which didn't bear the name "wormhole" yet back then), or "charge without charge" (think Wheeler). But after this part of the book he assumes that the total potential is singlevalued (not because he proved so, but because describing physics in spaces with complicated topologies is far from straightforward, even today).
Which book is this? I believe reading the original masters who came up with the discoveries is the way to understanding. It is said that when the Great Mathematician Gauss was asked as to how he made his discoveries, replied; "By studying the Masters and not their Students". There is something about the process of trial and error, testing various hypothesis', going down dead ends and then doubling back and finally lighting upon the answer which clarifies and provides intuitive understanding that can only be conveyed by the person who went through the experience i.e. The Discoverer himself. Everybody else is just parroting the end results without understanding (there are always a few exceptions of course).
The relevant section is in the preliminaries in Vol I : I assume you already know electromagnetism, I suggest jumping to pdf pages 52 "on potentials" and reading till 57. No modern textbook will explicitly introduce such an assumption, instead modern textbooks assume the student has been indoctrinated with cartesian product spaces i.e. R^3 etc. Such students would have seen in prior analysis courses that the line integral of the gradient of a scalar function does not depent on the path [... in R^n!]. So if curricula and students implicitly and a priori exclude spaces other than R^n, then no student will protest that for example the line integral of the electric field between 2 points could depend on the exact path, making all subsequent math and theorems more tractable, at the "small price" of a priori excluding the (arguably) confusing possibilities of handles (if you are a mathematiciann), wormholes (if your a physicist).
Maxwell on the contrary refuses compromise on investigative accuracy to "ease the job of didactics", theres many examples of that throughout the treatise.
>Which book is this? I believe reading the original masters who came up with the discoveries is the way to understanding.
I agree. There is a great book series: Great Books of the Western World, that accumulates all the fundental texts written by subject area founders. They are green bound tomes. There is also a homeschooling curriculum built on using them, I am jealous of anyone who received such a great education growing up.
If you want something nicer you could find a old used hardcover from the mid-twentieth century. Sometimes big reference works like this have never actually been used and are in effectively new condition.
What I love especially, is that he is very careful and systematic about his conclusions, splitting up in cases, instead of simply producing the solution and then proving it is a valid solution.
For example he does not assume that every point in space has a single value for total potential. First he describes how it is at least theoretically conceivable to follow a path and observe the total potential to vary continuously, and arrive back at the point of departure but end up with a different value of the total potential. He is effectively describing the possibility of wormholes (which didn't bear the name "wormhole" yet back then), or "charge without charge" (think Wheeler). But after this part of the book he assumes that the total potential is singlevalued (not because he proved so, but because describing physics in spaces with complicated topologies is far from straightforward, even today).