Weight can be judged from body measurements most accurately by the use of standards that take notice of the thickness of the chest, for instance, the chest girth.1 Furthermore, it has been shown by modern biometric methods that, in normal men weight is more nearly correlated with chest girth than with height.2 Then, evidence has been presented that the girth is more reliably measured by taking the arithmetical average of the inspiratory and expiratory girths than by taking the resting girth. To

more »

... certain further questions regarding the best method of chest measurement, this paper offers answers. Is the weight as closely correlated with the girth at the xiphoid level as with the girth at the nipple level? If so, the former may be preferable, since it is clearly a more nearly bony measurement, and therefore probably a sounder gage in obese or emaciated persons. Besides, there is some evidence in the literature that at the xiphoid level the girth in the living approaches the girth on the denuded, i. e., bony frame, more closely than is the case at the nipple or at the axillary level. Our evidence is derived from 100 candidates for the varsity crew at Harvard, measured by one of us (D. C. P.). These men were of the usual college age group of 18 to 22. They were all of better than aver¬ age physique, and were organically sound. Mention is made of these facts with the idea that measurements on a group of almost supernormal young adults of this single type should perhaps produce a more accurate coefficient of correlation than the same measurements made on a hundred adults of different types as to physique, age and environment taken at random. The coefficient of correlation of weight to chest-girth at the nipple level in the selected group was + 0.837 with a probable error of 0.0202 versus 0.831 ± 0.0210 at the xiphoid level. The difference in these two coeffi¬ cients, in the light of their probable errors, is insigni¬ ficant ; hence the answer to the question is yes. Is weight correlated with the girth as closely at expiration as at the calculated midrespiratory phase used in all our previous papers? This question is suggested by the preference for the former by army examiners in order to compensate approximately notice¬ able stoutness. The latter wrote: "I noticed that the measurement of the chest girth (which varies with the state of nutrition) taken when the patient was first seen in the state of obesity or undernutrition gave a different value from that noted after the patient had been brought to a normal state of nutrition. Differ¬ ences of five or more centimeters are not rare." Our evidence at complete expiration is a correlation coefficient between weight and girth at the nipple level of 0.818 ± 0.0223 versus 0.806 ¡± 0.0236 at the xiphoid. Again the difference is inconsiderable, and the answer to the question is yes.