The first physics book to hold my interest was the first edition of Edwin F. Taylor and J. A. Wheeler's Spacetime Physics. (Our university library still doesn't have a copy of the 2nd edition.) I was a sophomore when I first encountered the book at the hangout of the theory group. (My memory may have failed me here-- the only other possibility is after a random walk down the library shelves.
It was at the right level for me, and when I seriously started working on it (near the end of the sophomore year or was it the beginning of the junior year?), I fell in love with it. Although I tried reading other books, I disliked them. The usual approach is an algebraic one; in other books, one had to endure algebra without being able to see, at a glance, what one was working towards.
I did not understand, before Spacetime Physics, what an inertial observer was. The careful description of a latticework of clocks and rods freed me from the belief that an observer was a person stationed at the origin. Due to this misconception, I had trouble with Lorentz transformations. I also laboured under the belief that things like time dilation, length contraction and so on were consequences of the finite speed of light.
Spacetime physics had an unusual emphasis on spacetime diagrams and the geometric interpretation; this geometric interpretation, coupled with algebraic methods, gave me a better grasp of what relativity is about. I remember many later problem sets ---my classmates would struggle with lengthy algebra, and I would get the desired result in a few lines.
Later, when I was assigned to teaching a modern physics course for engineers, I chucked out the usual books and substituted Spacetime Physics for the relativity portion of the course. Since that part of the course relied mainly on the first chapter, we used an electronic copy of the first edition. (You can download the first chapter, with exercises, from Edwin Taylor's website .). The feedback I get is that relativity is the most enjoyable part of the course.
I've even had fun with the undergraduates of the theory group. I had two of our best undergrads take the special relativity exam; the exam was a 45 item multiple choice exam, and I told them that I would give a peso for every correct answer. I also appealed to their self-respect: would they really allow engineers to claim that they knew special relativity better than physics majors? (They've had the modern physics course for physics majors, so I teased them that they ought to have an advantage over "mere engineers". They used another book for relativity though. )
The result? The engineers had them beat. They scored in the early thirties range, while a "mere engineer" was able to get over 40 items right. The average engineering student was able to get around 30 items right, so I teased them without mercy afterwards.
The summer after that, our undergrads sat through the relativity portion of the course I taught, read Spacetime Physics, and solved problems I assigned. They did a lot better this time around; most of them got 40 items or more. (Appealing to self-respect is a good teaching strategy!)
Starting this year, all our undergraduate apprentices are required to sit-in during the relativity portion of the engineering modern physics course. They are also required to take the exam; self-respect should take care of the rest. I'm hoping to get some changes to seep in from below; Our students will eventually replace us, and I hope that when that time comes, the spacetime approach will become a standard part of the undergraduate toolkit.