(1995) of women in Army basic combat training compared the ability of the four female equations currently in use to predict changes in body composition over the 8-wk period of basic training and to assess the women's compliance with body fat standards. Each of the four equations resulted in significantly different predicted body fat for the same women. The Marine Corps equation predicted the lowest body fat, suggesting that although the Marine Corps upper limit for body fat is lower than those of the other branches, use of its equation would ultimately result in the retention of a population of women who had the same approximate maximum body fat as that of women in the other services. The Navy equation overpredicted body fat of women with greater upper body strength (those able to lift more than 100 lb [45 kg]), in part because of the increased abdominal circumference of these women. Comparing the circumferential measures of all four services, Westphal and coworkers (1995) found wrist and abdominal circumference to be the best independent correlates of body fat; however, comparing the individual components of the equations used by each service showed hip circumference (used by the Army, Navy, and Air Force) to be the best. Both the Navy and Air Force perform a measure of waist circumference, the Marine Corps measures abdominal (navel) circumference, and the Army measures neither. In this study, waist and abdominal circumference were not strongly correlated with measured body fat; other research with Army subjects has shown that the correlation between abdominal fat and total body fat is significant only in the fattest women (Vogel and Friedl, 1992a).
An important aspect of developing anthropometric body composition equations is standardization against a reference or criterion method. Ideally, the selected criterion method should be accurate and precise in heterogeneous populations that vary in age, gender, and ethnicity. In the past, all military branches have used underwater weighing as the reference method to partition body weight into fat and FFM. The underwater weighing two-component model is based on assumed constant densities of fat and FFM. However, as discussed further below, evidence is mounting that water, protein, and mineral FFM fractions may vary considerably among individuals. They may also vary systematically between genders and among age- and ethnic groups (Lohman, 1992). Because the percentage of female and non-Caucasian soldiers is increasing (see Table 2-3 for current ethnicity distribution), and the average age of female soldiers is increasing, the subject population used to develop and validate the military equations (a predominantly young, Caucasian population) is becoming increasingly less representative of the military population whose body composition the equations are being used to predict. Furthermore, questions persist regarding the ways in which underwater weighing is performed and technical problems are overcome (for example, determination of residual volume) (McArdle et al., 1996). Finally, longitudinal studies have shown that underwater weighing does not reliably measure small changes in body composition.
An improved three-compartment model that accounts for the water component of fat-free body mass was suggested by Siri in 1961. Siri's three-compartment model is based on measured body weight, body volume, and body water. Fat estimates using this model are less vulnerable to hydration variation than are those using the classic two-compartment model.