In the introductory chapter of the Typographic Desk Reference, Theodore Rosendorf, the author, says his goal was to satisfy a need of the industry: a book that is "solely devoted to quick reference across the entire craft." In my opinion, it fails to do so because of three factors.
Firstly, his use of the word, solely, is not without a reason. That's because the celebrated Elements of Typographic Style by Robert Bringhurst has 147 pages of reference -- I'm not even including the sections on page and text block proportions. Bringhurst's coverage of specimen is both broader, including cyrillic and greek, and deeper, as it sometimes delves into the story and development of type. Since the TDR is meant to be quick, its terseness can be excused.
Secondly, there are misstatements, conscious or not, in many places. I enumerate a few: Helvetica's original name is spelled "Neue Hass Grotesk" instead of "Neue Haas Grotesk." Quadratic splines are said to be more complex than cubic splines -- actually, type created with quadratic splines tends to have more points than the type constructed with the cubic counterpart exactly because the former is a simpler, less expressive tool. The umlaut/diaeresis entry says that it is employed in Portuguese. No longer.
Finally, some editorial choices are quirky and have unsettled me. In the meatier chapter, for example, the author enumerates almost every possible latin diacritic letter (e.g. e-acute, e-circumflex, e-circumflex-tilde, &c.) and gives them all the exact same perfunctory description: "inflected latin letter e used in such and such language." It reads like a table, only it's not. The sole bit of information besides the appearance of the accented letter is the list of languages that employ it, which generally is too short. Bringhurst's approach is much more sensible: only the diacritic marks are listed, along with a rational amount of related information. It's both briefer and more complete. In the last chapter, dedicated to specimens, Rosendorf includes the hideous Verdana (and even its unsightly slanted italic) and Consolas, a font for software programming. The pleasant and recently popular Georgia is not mentioned, however. (Actually, the colophon reads "Typeset in Decatur, Georgia," which had me thinking for a couple of seconds, blame everyday comma splices.)
Not all is lost, however. The book as an object is a little piece of art. It is wonderfuly bound. The hardcover is sturdy and handsome, with just a relief "TDR" identifying it. The paper is stunning and Caslon looks crisp on its textured, just off-white surface. Margins are ample, leading is just right and -- oh, surprise -- the page format is a golden rectangle. The abuse of c_t and s_t ligatures throughout the text can be a nuisance to the modern reader, though. The black satin bookmark is unnecessary, too: this is a reference, not a novel where you must pick up where you left before going to sleep. Overall, the book is itself a good example of pleasant (if a bit ostentatious) typography and bookbinding. Interestingly, the metalinguistic foreword and introduction seem to reinforce this.
I might rephrase my initial statement: the book fails to achieve its goal because of only one reason, I got The Elements first and it is enough. Bringhurst's work has a delightful prose, includes lots of invaluable historical information and numerous guidelines, costs less than half the price and has more than twice the number of pages. It might not be as chic as the TDR but, as Bringhurst says, "[the] satisfactions of the craft come from elucidating, and perhaps enobling the text, not from deluding the unwary reader by applying scents, paints and iron stays to empty prose."