Certain correlations between food intake and social status, height, weight and gain of weight during pregnancy were referred to in a previous series of papers (Thomson, 1958(Thomson, ,1959a. In this paper, the same data, derived from a study of pregnant women, are used to investigate further the relationships of height and weight to diet.The most interesting conclusion is that, contrary to popular belief and the implication of physiological teaching (e.g. F.A.O. : Second Committee on Calorie Requirements, 1957), heavy people do not as a rule eat much more than light people. A review of the literature suggests that this conclusion is valid also for men and for non-pregnant women.
M E T H O D SThe dietary data used here were obtained from 489 primigravidae by means of a I -week weighing-inventory survey (Widdowson, 1936). The technique used has been fully discussed and, so far as possible, validated (Thomson, 1958); any record considered by the dietitian concerned to be incomplete or unreliable was rejected. The diets were self-chosen, and no attempt was made by the obstetricians concerned to regulate the amount of weight which the subjects were gaining.All subjects were weighed at the time of the dietary survey (about the 7th month of pregnancy) and most were weighed at intervals throughout pregnancy. Weights at conception could not be assessed reliably, and for 412 out of the 489 subjects 'initial weight' has been taken as weight at the 20th week of pregnancy, by which time all of them had been measured. All weighings were on calibrated lever balances reading to 2 oz (57 g) and subjects wore only underclothes during weighing. Shoes were removed during the measurement of standing height, which was recorded to the nearest 0.25 in.(0.6 cm). Since only 10.6 yo of the 489 subjects were aged 20 years or less, incomplete growth need not be considered when interpreting the height data.Weight-for-height. The significance of a given body-weight ( W ) varies with height ( H ) . T o express weight-for-height, a mathematical index, such as H / J W , is frequently used. Examination of the properties of several such indices showed that none of them discriminates uniformly over the whole range of heights ; that is, the proportion of ' overweight ' and ' underweight' individuals depends on height as well as on the value of the index. Some calculations based on the index H / J W , given on p. 251, show that it can yield misleading results. To overcome the difficulties, we have used a procedure devised during a previous study of weight gains during pregnancy (Thomson i&