The empirical finding of the present study can be stated simply enough : mental patients who have died of other causes than exhaustive diseases include in their number a strikingly large proportion of persons with long, lean, wiry, not merely ectomorphic but to some extent mesomorphic physiques, and a conspicuous lack of those with big-boned, heavily muscled builds. They form a relatively homogeneous group, differing not at all by the nature of their mental illnesses, whether functional or organic in origin : but those who were older at the time of their deaths tend to be bulkier, more rounded, and less thin and fragile than those who died at early ages.To explain these facts is still difficult. Perhaps the patients really differed from an unselected population in these respects before the onset of their illnesses, in which case, one would be led to a hypothesis of a constitutional make-up with nonspecific vulnerability to psychosis. Very probably, the bodily build is not the result of the mental illness, directly or indirectly; if it were, one would expect the effects to show more sharply in cases where the disorder was of long duration. It does not seem plausible that faulty technique on the part of the investigators could have produced such consistent results if readily apparent differences between nosological groups did exist. Possibly mesomorphy is less apparent in a person who is suspended in mid-air instead of standing up, and dead—without muscular tonus—instead of alive; other bias introduced by the fact of death seems less probable, since there were a number of suicides and accidental deaths and none after an exhaustive, lingering illness.It is possible also that some of the differences between our subjects and Sheldon's are attributable to lack of representativeness in his sample as well as in ours. If it is also true that somatotypes change with age, another reason for not using Sheldon's norms as a standard might be found, though such a hypothesis would make the results even harder to explain. The group which is most comparable in age with the 4,000 college students is the very one which is least like them, while advancing age puts enough flesh on these skinny patients to make them appear (perhaps speciously) more like the standardization group. It is credible that American men as a group are not as athletic in their physiques as young college men, but it is very hard to believe that they are as scrawny as the majority of our research population. The general finding of a specific excess and deficit of just two kinds of builds, making up a total of only 35.5% of Sheldon's group, is thus hard to explain away on the basis of sampling error. A possibility which does still remain is that one of these groups—the mesomorphic ectomorphs—is simply less viable, while the other—the predominant mesomorphs and ectomorphic mesomorphs—has greater vitality. It would be easy to test this hypothesis by somatotyping a similar group of patients dying successively in a general hospital, using the same sampling controls as the ones we employed. If significant differences between such a group and ours were still found, it would be possible to say with some confidence that the peculiarities of somatotype were specific to mental illness.In The Varieties of Human Physique, Sheldon has already hedged a little against the possibility of largely negative findings such as the ones presented here, by stating that certain kinds of dysplasia are much more closely related to dementia præcox than are particular somatotypes. The analysis of dysplasia, or disharmony between various regions of the body, requires an additional five sets of judgments for each picture; we decided to forego it. The possibility still remains, then, that more refined and exhaustive somatotyping studies (and we may add, much more time-consuming ones), might bear fruit of positive results if oriented toward the analysis of dysplasia.The difficulties which have plagued this study point dearly to one conclusion: the mere adoption of the technique of somatotyping opens up no royal road to the relationship between physique and psychosis. As it presently stands, the system is not well adapted to application to the very types of research problems to which it seems to have so much to contribute. To have studied with the originator of the technique and to have tried conscientiously to follow the book is not enough to enable one to apply it to subjects who differ appreciably from those Sheldon used and end up confident that the results are not a function of errors in method. Until normative studies have been done upon satisfactory samples of older men, and of women, somatotyping studies will perforce be limited to populations of white college men or others who are similar in sex, ethnic group, and age.Within these limitations, it is still possible to design a study which can result in conclusive findings where we have tried to pioneer. The research population should be a complete or a random sample of hospitalized patients, all physically healthy white males between 18 and 25 years old. They should be somatotyped independently by two or more observers who have no knowledge of the patients or their diagnoses, preferably using the anthropometric technique or at least validating anthroposcopic judgments by measurements and reference to the tables in Sheldon's book in a considerable sub-sample. Groups should be equated for length of hospitalization and for duration of illness. Diagnoses should be clearly and of course independently established, if possible by the use of psychological tests as well as clinical examination. Finally, the data should be analyzed from the standpoint of mean ratings, groups of somatotypes, and kinds and degree of dysplasia. Such a study, which would not be prohibitively difficult to carry out, would furnish the conclusive data which are so badly needed on the relationship between mental disease and physique.Abstract Teaser