Atlas of Anatomy, Third Edition, Latin Nomenclature Edited by A. Gilroy and B. MacPherson. Illustrations by Markus Voll and Karl Wesker. New York, Stuttgart, Delhi, and Rio de Janeiro: Thieme Medical Publishers 2017.

I'm glad I'm reviewing the physical version of this atlas instead of the electronic version (the software for which won't run on my linux system anyway, the publisher informs me). For this 3‐kg codex book is gorgeous: 750 28‐by‐23‐cm pages with 1935 color illustrations.

The editors proffer us a short Preface, page xv, which sets the effort of revision in context:

In this latest edition we focused our attention on three major tasks. The first reflects an understanding that anatomy is a changing science. Our readers understand that it is a dynamic part of clinical medicine, itself a science undergoing constant evolution. … Our second task was to add additional examples of sectional and radiographic images to help students apply their knowledge of anatomical structure and relationships. … Finally, we have expanded areas that deserved greater attention, [such as the] newly titled Brain and Nervous Systems unit.

So this book review, by a biologist who is not an anatomist but instead a heavy user of its knowledge base and its nomenclature, will center around my understanding of the dynamics of that science and the atlas's failure to cover some areas that, in my view, deserve ‘greater attention’. But I'll mention only the structures I tend to look up in the course of my own research projects.

One specific anatomical concern of mine is the orbicularis oris muscle, and in that regard this atlas serves someone like me splendidly. Its excellent index (40 three‐column pages) lists seven separate mentions of this muscle in the book, of which one set, page 521, nicely breaks out the various insertions on this muscle that will certainly be relevant when I get around to writing my paper on how people with repaired clefts smile. (See also the cover figure, reprinted as a frontispiece, where each muscle inserting on the orbicularis is drawn twice, once in situ and once as sectioned.)

But my other applied anatomical interests do not come out so well. I am interested in the optic disk, for instance, of which the only appropriately scaled drawing here, page 577, oddly excusing itself as a ‘schematic’ (as if the vast majority of other figures were not equally ‘schematic’ in their stereotypy), unaccountably draws it as a perfectly circular raised annulus (a white ring) that leaves a sundial‐sharp shadow on the retina. Needless to say, the retina incorporates no such structure as what is drawn here. By contrast, this atlas's views of the femoral head are always drawn (correctly) as circles (sometimes in section, sometimes as edges of regression), but I cannot find any acknowledgement that a circularity so persistent over so many different renderings rigorously implies that the actual surface being rendered must, with the exception of its fovea, be a sphere. In my world of applications, concerned mainly with the dysplasias, its sphericity is the most relevant factor about the head of the femur; strange that it is correctly rendered here but otherwise not noted.

Or consider the corpus callosum, the interesting three‐dimensional structure that is the ‘waist’ of the white‐matter bundle of the neocortex. Our atlas has 11 separate images of this structure, but all of them are rendered in two‐dimensional section. There is no figure analogous to figure 733 of the online Gray's Anatomy (1918 edition), ‘corpus callosum from above’, let alone the exquisite whole‐brain dissections of white matter retrievable these days by an easy Google probe. (Myself, I love the Pinterest version.) Or consider the nasal septum, a structure of great concern to many oral surgeons. Only 3% of a typical population's nasal septa are flat (Takahashi, 1987), and yet this atlas presents no information whatsoever about the missing mediolateral information. Instead, as on page 526, it is, unbelievably, not only drawn but even sectioned as a uniformly thin and flat plate. Only the two whole‐head vertical section images, pages 646–647, represent this nonflatness realistically, and even for these figures there is no mention of the asymmetry that is so obvious in the artwork and that in reality fluctuates in laterality, amplitude and profile.

Of course the Gilroy–MacPherson atlas offers numerous clear illustrations of the adult genitalia of both sexes, but nothing about the structures that are intermittently but entirely predictably associated with one of these over the life course of about one‐third of us: no illustrations of a fetus, a gravid uterus, an occupied birth canal, or even a placenta. These are all perfectly normal structures; why are they not depicted? One of the older atlases in my collection, Barry Anson's of 1963, includes two illustrations pertinent to a fetus (ductus arteriosus and fontanelle), but nothing of the gravid uterus or the placenta. Here in 2017, coverage even skimpier seems unwise indeed.

This leads me to my main complaint about the new Thieme volume: a major domain where it falls short of that Anson predecessor by a full 100%. Anson offered a remarkable number of figures dealing explicitly with normal variation. Some of these illustrate quantitative variability, such as the vertebral level of termination of the spinal cord (his page 295, which also illustrates its ‘fetal position’). Others deal with shape variability, such as the remarkable range of normal forms of stomach (his page 376) and liver (his page 377), and still others deal with what we would now call aspects of topological variability, such as branching sequences of arteries (his pages 404–407, 411–414) or nerves (his page 62), or the complex interweaving of the tendons of the forearm (his page 207). These panels of normal variants were accompanied by their prevalences over hundreds of the cadavers he dissected in the course of his teaching: ‘nonconformity as a salient feature of every bodily system’ (his page v).

But this is a review of Gilroy and MacPherson's book, not Anson's, so I must content myself with noting that it is this species of knowledge – variability, not Platonic typology – that is the focus of the applications of the anatomical sciences today, both within and outside of medicine. And this new atlas offers none of these illustrations of variability. (Perhaps because there is no corresponding glossary for them in Terminologia Anatomica?) Such obvious concerns for the medical student as fetal and neonatal structure, the course of growth over childhood and adolescence (or, conversely, the typical decline into senility), the effects on the adult form of relative height, of body mass index, the variability of breasts or of the course of the intestines – omitted, all omitted.

In short, the information needed to automatically map between any specific cadaver before the medical student, or patient before the surgeon, and the images of this atlas is simply missing. But that brings me back to the Preface that I already quoted regarding anatomy as a dynamic part of an evolving medical science. What is missing from this and many other atlases is precisely the component of clinical anatomical knowledge that will be most crucial when we articulate all this crystallized scholarship to the artificial intelligence now being brought to bear on millions of clinical CT and MR scans in the hands of collaborations between such partners as Google and Kaiser Permanente. Normal variation, both qualitative and quantitative, will be as much a component of tomorrow's anatomical sciences as ideal form and its terminology will be. And yet, to my considerable surprise, I have come across only one explicit empirical quantification anywhere in this book, and it is a pretty obscure one: the angle between the orbital axis and the optic axis, page 576, which is specified tersely as ‘23°’. Really? Exactly? Surely this value has a standard deviation? Or examine the spread on ‘female and male pelvic measurements’, pages 232–233, that doesn't present any actual measured numbers or any instructions for how to extract the numbers from the living patient. There must be many other examples of this innumeracy at least as salient, beginning perhaps as early as page 4, the anteroposterior curvature profile of the normal spinal column. Here in 2017, the authors’ apparent determination to treat anatomy as an essentially verbal/visual science quite baffles me.

This cultivated unconcern with normal variability and its categorization and quantification cannot be attributed to technical limitations peculiar to the anatomy‐atlas genre. Anson figured out how to represent such information of this type in dozens of figures published well before most of the readers of this review were born. It must instead be the trace of some community decision as to what counts as the proper domain of discourse for an atlas, versus what topics remain outside. With all due respect I suggest that here in 2017 we must deem that boundary to be badly misaligned. Even Anson's 1963 atlas had far more quantifications than this one, to wit, all those percentages of topologies across his pedagogic dissections, along with the histogram of spinal cord termini. Nowadays, the corresponding quantification of means and ranges of a huge variety of extents and proportions and the mean shapes and prevalences for most of the clinically relevant branching topologies could be produced almost automatically.

If anatomy is to remain a required component of the basic medical sciences in the curricula of 2040, atlases like these must expand to encompass some descriptions of normal variation, even if at first they are rudimentary. A good start would be a fourth edition of this atlas that incorporated the equivalent of Anson's figures of topological variability together with a new set based on multiscale shape analyses of digitized whole‐body imaging. Such a sea‐change will be necessary if anatomy is to remain ‘a dynamic part of clinical medicine’. That mandate goes unacknowledged, which is to say, unanticipated, in this otherwise excellent and visually fascinating Thieme offering.