There are two approaches to thermal physics, the large scale or macroscopic approach of Classical Thermodynamics, and the atomistic or microscopic approach of Statistical Mechanics. A traditional way of teaching the material, that our curriculum still contains, is a semester of Classical Thermodynamics followed by a semester of Statistical Mechanics. Many of the older texts concentrate on one or other of these approaches, or they present the two methods in two clearly separated parts. Newer books have mixed up the two approaches.

Oldies but goodies

Zemansky (and Dittman). Zemansky's "Heat and Thermodynamics" was the standard undergraduate text for many years. The first edition was published in 1937, the fifth edition in 1968. For the sixth edition of 1981, the publishers added a second author, Dittman. By the time of the seventh edition, in 1997, Zemansky had died. The authors are still listed as Zemansky and Dittman, but the book is much more Dittman than Zemansky. I prefer the older editions. Perhaps in an effort to keep the length down, Dittman cut out several topics that I like to teach. The book deals with Classical Thermodynamics in a traditional presentation (you could say that it established the tradition) and then deals more briefly with statistical mechanics.

Sears (and Salinger). Sears wrote a whole series of textbooks in the 1950's, covering Electricity and Magnetism, Optics, and Mechanics, Heat and Sound, as well as Thermodynamics. His book on thermodynamics is quite similar to Zemansky's. It gives a presentation of Classical Thermodynamics in the traditional order (e.g. it uses heat engines to introduce the Second Law) and then gives a slightly shorter version of statistical mechanics. The version of the book that is still available, although it dates from 1975, is co-authored, "Thermodynamics, Kinetic Theory, and Statistical Mechanics" by Sears and Salinger.

Crawford. I have a soft spot in my memory for "Heat, Thermodynamics, and Statistical Mechanics" by Franzo H. Crawford, because it was the first text book that I ever taught from. It covers the subject in a very traditional way, first Classical Thermodynamics, using heat engines and with enormous use of partial derivatives, and then Statistical Mechanics, more thoroughly than either of the previous books. It is comprehensive, accurate, and boring. It is also long out of print. But I still go back and look at it occasionally.

Adkins. I also have a soft spot for "Equilibrium Thermodynamics" by C.J. Adkins, because I sat through what I think was the first time Adkins lectured on thermodynamics. As a result, I have always tended to use his notation and presentation. This book differs from the previous three in that it deals only with Classical Thermodynamics, with no coverage of Statistical Mechanics. Adkins was a student of Pippard (mentioned below) and it is reasonable to describe Adkins' book as Pippard's book, expanded and made more readable. This is not a bad thing to say about a book.

More advanced oldies

Pippard "Classical Thermodynamics" is one of the seminal books on thermodynamics. It is short (160 pages) and difficult. There are very few applications to illustrate the ideas, but it attempts to present the bare theory with something approaching mathematical rigor. As far as I know, there was only one edition, in 1957, although it was reprinted many times.

Callen "Thermodynamics" is the second seminal text. I know the first edition, from 1960, but there was a second edition, much later. Like Pippard, Callen deals almost exclusively with macroscopic thermodynamics, but he uses a completely novel approach, that he calls a postulational approach. I think this is not a good first book on thermodynamics, but, as a way of shaking up your ideas when you think you have understood thermodynamics, it is excellent. I have usually borrowed some of his ideas on equilibrium conditions in my lectures.

Morse "Thermal Physics" is similar in scope and order of presentation to Crawford, but there is less discussion of the phenomena and the arguments are presented at a higher intellectual level. I view it as a graduate text.

Reif "Statistical and Thermal Physics" is the first book on this list that is purely a book on Statistical Mechanics. (There is one short section in which the laws of thermodynamics are discussed.) Published in 1965 (I don't think there was another edition), it is still widely used, but most often as a graduate text. Reif also wrote a much more elementary book "Statistical Physics" as the fifth volume of the Berkeley Physics series.

Wannier "Statistical Physics". A fairly conventional development of statistical mechanics. Wannier was an excellent writer. He also wrote a good but out of print book on solid state physics. Wannier gave his name to the Wannier function, heavily used nowadays in density functional calculations, and he came close to solving the two-dimensional Ising model before Onsager did. One merit of this book is that you can buy a Dover reprint for $11.95.

Non-traditional approach

Kittel and Kroemer "Thermal Physics". Kittel wrote the first edition of this, in 1969, and Kroemer co-authored the second edition, in 1980. (This is the same Herbert Kroemer that won the Nobel Prize in 2000.) This has been a very widely used book for the last twenty years, although last year the book store told me it was out of print. It deals almost exclusively with statistical mechanics, with one chapter that derives the laws of thermodynamics. The methods used to derive the statistical distribution laws were very novel when the first edition was published, quite different from the counting techniques used in the earlier books. For me, this is still slightly controversial. The problems at the ends of chapters are excellent.

Stowe "Statistical Mechanics and Thermodynamics". I am less familiar with this than any of the other books listed. It appeared in 1984. In some ways it follows the conventional format, treating Classical Thermodynamics and then Statistical Mechanics, but, before any of this there is a discussion of small systems and ideas from statistics, and the treatment of thermodynamics makes use of these ideas rather than taking a purely macroscopic approach. One reason for looking at it is that Dan Schroeder (see below) credits it as one of his inspirations.

The New Wave

For some reason, there was a dearth of new books on thermal physics in the 1980's and 1990's. Perhaps we all assumed that the last word had been written. For many years, I used Zemansky for the first semester and Kittel and Kroemer for the second. Then there was a flurry of new books. Three appeared in 1999, and 2000, and more are rumored. One characteristic of the new books is that they are tailored to a one semester course on thermal physics, but all the authors try to do justice to both sets of ideas.

Schroeder "Thermal Physics". Written in a very chatty style, that I wish I had thought of. There is an immense number of problems, embedded in the text rather than left to the chapter ends. The way to use this book seems clearly to work a lot of problems as you go. Schroeder says that there is too much material for one semester, but that it is still primarily intended for a one semester course. The presentation starts with microscopic ideas, and freely mixes up microscopic and macroscopic concepts all the way through

Baierlein "Thermal Physics" (there seem to be more new books than new titles). This has much in common with Schroeder. It uses microscopic ideas to motivate and justify the second law. But its treatment both of thermodynamics and statistical mechanics is more formal that Schroeder's. (e.g. there is a chapter called "The Canonical Probability Distribution". Schroeder buries the word "canonical" in a footnote.) However, the treatment of Classical Thermodynamics does not follow the traditional approach through heat engines, and the treatment of the probability distributions follows Kittel and Kroemer's methods rather than, for example, Reif's or Crawford's.

Carter "Classical and Statistical Thermodynamics". This is by far the most traditional of the new books. It follows the pattern of Crawford and of Morse, developing Classical Thermodynamics using a macroscopic approach, and then Statistical Mechanics using the traditional Lagrange multiplier techniques. However, Carter says that you can cover all of this in one semester.

Two very old (and chemical) goodies

Two classic books dealing respectively with Thermodynamics and Statistical Mechanics are

Lewis and Randall "Thermodynamics". Lewis is a giant of chemistry. The first edition of this was published in 1923 and I read it for the pleasure I get from the language. The second edition, not by the original authors, was in 1961, and it is still good, but I would prefer the original.

Mayer and Mayer "Statistical Mechanics". The Mayers developed the theory of an imperfect gas, and why not read it in the original? Incidentally, the book is dedicated to "our teachers, Gilbert N. Lewis (see above) and Max Born". Max Born is best known as the grandfather of Olivia Newton-John, and he also won a Nobel prize for the development of quantum mechanics. I heard him lecture in 1963.

Back to main page