FOOD AND AGRICULTURE ORGANIZATION OF THE UNITED NATIONS	ESN:FAO/WHO/UNU/ EPR/81/13 August 1981
	WORLD HEALTH ORGANIZATION
	THE UNITED NATIONS UNIVERSITY

Item 2.2.6 of the
Provisional Agenda

Joint FAO/WHO/UNU Expert Consultation on
Energy and Protein Requirements

Rome, 5 to 17 October 1981

INDIVIDUAL VARIABILITY IN ENERGY NEEDS

by

W.P.T. James

Dunn Nutritional Laboratory
Cambridge

INDIVIDUAL VARIABILITY IN ENERGY NEEDS

The current views on the extraordinary range in energy requirements seem to be based on concepts developed by Widdowson (1962) who noted that in any group of 20 or more individuals one can expect to find one individual requiring twice as much energy as another. This conclusion stems from her careful measurements of food intake in her classic M.R.C. Report on the diet of individual children (1947). She presented data which are reproduced in Table 1. listing the maximum and minimum energy intakes of both boys and girls expressed in terms of either body weight or surface area. On either a weight, surface area (or height) basis, the coefficients of variation were usually 15–25%. This concept of a two-fold variation in energy needs was endorsed by Durnin et al. (1973).

Reproducibility of measures of food intake

In trying to define the reasons for this variability one has to consider whether or not the measurement of habitual energy intakes is so inaccurate that it introduces a major factor into the analysis. If a considerable weekly variation in intake exists, then this could well-mean that in 20 individuals studied for a single week there would be bound to be a wide variation between individuals simply because food intake normally varies markedly in any individual from week to week; the ‘true’ energy requirements of a group would then be very much more uniform. Widdowson and others have considered this possibility to a limited extent and Widdowson tried to counter this argument by displaying Boulton's data on the reproducibility of weekly measures of energy intakes. She also claimed (1962) that in her own studies she had found reasonably consistent intakes. Boulton's data are reproduced in Table 2 and in Fig.1. The variation of one individual from week to week is much less than the variation between individuals. Nevertheless there is a coefficient of variation in weekly intakes which ranges from 3.6 to 12.1%. If a 4-week period does give an indication of true energy requirements, then clearly the range in energy needs should be reduced. Table 3 summarizes some published work on the reproducibility of weekly energy intakes in adults. This analysis introduces another variable - that of genuine changes in energy requirements with time, but nevertheless it is clear that a single week's measurement of energy intake is unlikely to give one a good idea of the values found on a subsequent occasion. If published and unpublished data (Table 4) are used to calculate the accuracy with which individuals can be ranked for their energy intakes within a group, then it becomes clear that even crude ranking cannot be guaranteed from measurements of energy intakes monitored for a week. A further problem with the assessment of energy requirements from measurements of energy intake is that a short-term change in food intake may be induced consciously or unconsciously by the subject under investigation. Intakes may then be reduced to inappropriate levels (see the accompanying paper on adaptation to low intakes) in an effort to persuade the investigator of the subjects' poverty or metabolic efficiency or the intakes may be temporarily increased as a demonstration of the subject's affluence.

I conclude that measurements of food intake for one week are an inappropriate method of estimating the true energy requirements of adults in affluent societies. On this basis I do not consider the classic paper by Rose and Williams (1961) of great significance since no validation of the high and low food intakes of the two groups was obtained. The paper also presents data for the low intake group which are inconsistent. Thus the mean BMR (expressed in absolute units on a daily basis) amounts to 2,110 kcals compared with a supposed intake of 2,380, allowing only 270 kcals for everything else including the specific dynamic action and physical activity. This last was estimated at 590 kcals per day. This paper should no longer, therefore, be used as an argument for an extreme variability to energy requirements. I am unaware of studies in developing countries which would allow the same analysis to be made as has been presented in Table 4.

The choice of energy expenditure as a better measure of energy needs

Possible explanations for the extremes in energy needs may be assessed if one looks not at energy intake but at energy expenditure. With this approach it is still possible to alter the values for energy requirements by an individual deliberately changing the degree of physical activity while under investigation or by recording inappropriate times for various activities. These objections apply less to those studies where subjects are either on a reasonably regimented system of activity and/or where the subject is not responsible himself for recording the exact time devoted to each task. I have found two examples of objective measures of this type other than Norgan et al.'s energy expenditures figures quoted in the paper on adaptation. They are Edholm's data on energy expenditure measurements in army recruits and our own unpublished data from whole body calorimetry studies. These studies provide information on the variability in energy requirements under prescribed conditions and may not give an indication of the full range of energy requirements in free-living adults. They are, however, at least a first step towards understanding the problem of variability in energy turnover. They are also particularly appropriate given the widespread view that there must be some metabolic explanation for the presumed wide range in energy requirements.

The first set of data were obtained from the Army Personnel Research Establishment's Report of 1956 to the M.R.C. and are presented in Table 5. The degree of organization involved in this study was immense, with measurements on the 14 subjects being obtained directly for 20 hours per day and a continuous time and motion study being employed throughout the 24 hours for the whole 14-day trial period while the men were either resting or engaged on combat exercises, long marches, parades and other duties in Northern Ireland. It is the most impressive piece of field work which I have ever seen recorded and involved a 30 strong team of drivers, observers and scientists operating in two shifts under radio control from a command centre! They were investigating the value of a combat ration of 3,000 calories per day given during phase 2 while the men were carrying out an arduous field exercise. Note that the variability in expenditure under more sedentary conditions in phases I and III is less than under combat conditions in phase II. These sedentary periods did include a 10 mile route march on day 3 of each period, but in both phase I and III the men were otherwise “confined to barracks”! The surprising feature of the results is the remarkably small variation in energy expenditure between the individuals under these circumstances. Table 5 includes values for body weight; expressing energy output in either phase I or III in absolute terms gives an interindividual coefficient of variation of 5–6.7% and expressing output on a body weight basis does not reduce the variation. It is therefore clear that in this group of young men of reasonably similar age and size there is far less variability in energy expenditure than one would predict from Widdowson's food intake data on children, or from her subsequent work on adult men and women (Widdowson, 1936; Widdowson and McCance, 1936). This strongly suggests that the variability in energy requirements has been exaggerated unless there are marked differences between people in the degree to which they spontaneously take exercise or in the energetic cost of any physical activity.

The role of physical activity

Phase II shows that when a degree of flexibility in exercise patterns is allowed, then the variability in energy expenditure does increase and this variability is more clearly demonstrated by calculating the change in energy output during phase II, as shown in the last column of Table 5. This increment in energy output has a coefficient of variation of 24.4% which shows that individuals do vary substantially in the amount of energy which they expend when undertaking a range of activities which are not rigidly controlled. Measurements of exercising individuals confirm the appreciable differences to be found between subjects in the apparent energy cost (Mahadera et al., 1953). Nevertheless this variability must not be exaggerated because Table 5 shows that even when expressed in absolute terms (i.e. without regard to body weight differences) the actual coefficient of variation in energy output between individuals during phase II amounted to only 7.6%.

Further support for the role of physical activity in the variability in energy output comes indirectly from our whole body calorimeter studies. Much of the data included here is unpublished and to understand the significance of the studies some general background information is needed. The principle in all our studies is that subjects occupy the calorimeters for a period in excess of 24 hours, the additional time being needed either to allow equilibration of gases and/or heat in the chambers or to ensure that the time of starting the 24 hr measurement is appropriate in relation to the defined pattern of food intake. All our studies therefore provide information on 24 hr energy expenditure. The distinctive feature of our studies which ensures a coefficient of variation in repeated measurements at weekly intervals of 2%, is that the diet and physical activity of our subjects are controlled. Physical activity is controlled by simply defining the posture, e.g. lying, sitting, standing, walking or cycling during a set number of minutes in each 24 hr period. We are therefore able to identify the quantitative importance of minor physical activity, basal metabolism and the thermogenic response to food but not to assess, except very crudely by inference, the importance of spontaneous physical activity.

The individuals before entering and while living in the calorimeters are usually on a defined diet related to their previously documented food intake. In a few studies, e.g. Group A, Table 6A, the subjects have received instead their estimated energy needs based on the 1973 F.A.O. Committee Report, values being calculated on an ‘ideal’ body weight basis from Metropolitan Life Insurance tables. In these groups there is a preliminary period on the diet during which time the subjects may then have their intake increased or decreased in 1 MJ amounts to take account of their appetite and any change in weight. Table 6A shows the intake during the study and the energy expenditure of predominantly young men confined to a calorimeter for 24 hours during which time they are allowed to stand and wander around the 11c. metre chamber for a total of 1 hour per day. They also cycle gently for 1 hour on a bicycle ergometer at 50 watts. This degree of activity we would consider to be equivalent to that of the sedentary individual, and perhaps close to the maintenance requirement calculated from the previous 1973 Committee's report as 1.5 times the BMR. The figure we have is 1.38 from direct measurements in both men and women. It could be argued that normally there might be slightly more physical movement than we achieved, e.g. in gesticulating when a subject is in company rather than isolated in a calorimeter and that most people do more tasks while sitting or standing and moving than those demanded of our volunteers. Nevertheless, the studies can give a very good picture of the variability in this maintenance requirement figure.

The 24 hr energy expenditure varied from 8.34 MJ to 11.4 MJ and the coefficient of variation in absolute terms was only 10.3% despite a range in weights from 56 to 97 kg. When expressed on a weight basis the coefficient of variation falls to 8.1% with a range in energy expenditure of 118 kJ.kg^-1 to 159.8 kJ.kg^-1 (mean 140.2 kJ.kg^-1). Table 6B presents data for lean and normal women. A similar variability in values is apparent, with a reduction in the variability of both the 24 hr expenditure and the basal metabolic rate when values are expressed per kg body weight. There is no further reduction in the variability if the BMR is expressed in terms of the fat free mass despite the recognized link between the fat free mass and BMR.

Sex differences

A comparison of Tables 6A and 6B brings out some of the energetic differences between the sexes. It should be remembered that both sexes were studied under similar conditions with a constant period of defined exercise. The lower total expenditure of energy of women seems primarily related to the body weight difference since expenditure on a weight basis was 139 kJ.d^-1.kg^-1 compared with 140 kJ.d^-1.kg^-1 in men. Durnin (1976) in his analysis of sex differences in energy requirements, notes the previous evidence which led the 1973 Committee to conclude that the energy requirement of men undertaking light activity was 176 kJ/kg and in women was 151 kJ/kg. These values, based on many hundreds of studies, are not confirmed by our calorimeter results obtained on a very limited number of people. Explanations for the smaller values found in the calorimeter have already been presented but in addition we may need to recognize that there may be a small component relating to environmental temperature since our studies are conducted either at 28°C in light pyjama-type dress or at 26°C in normal everyday clothing. Our recent calorimeter studies (Dauncey, 1981) have confirmed that quite subtle falls in environmental temperature can increase an adult's metabolic rate. Perhaps a 5% increase in total energy metabolism would have been produced at the lower temperature. A lower temperature may also lead to greater spontaneous activity and this is yet to be tested objectively.

Durnin, on the basis of his calculations rather than by direct observation, concludes that on a fat free mass basis total energy requirements are similar in men and women. He considers the result to be spurious since the use of the FFM as a basis for reference is too naive a concept; FFM comprises a mixture of metabolically active and inactive tissues, whereas adipose tissue he considers very active. This last conclusion I consider to be incorrect since the measurement of fat free mass automatically includes the non-triglyceride portion of adipose tissue and relating values to FFM may be much more logical than relating energy to body weight unless physical activity and its relationship to body weight dominates the picture of total energy requirements. This I doubt. Tables 6A and 6B show that the BMR is equivalent on a weight basis in men and women, and this accords with the data collected by Durnin (1976). Given this agreement it would hardly be surprising if the BMR per unit FFM were higher in women and simply reflected the smaller contribution of muscle with its low resting metabolic rate to the FFM in women than in men. The data presented in Table 6A and 6B are similar to 24 hr measurements recently reported by Webb (1981) for energy expenditure anf fat free mass in men and women.

Indirect assessment of the energy cost of physical activity

Most of our subjects were students (Groups A, C and D) or scientists (Group B). The subjects maintained their weight on these diets so the ‘excess’ intake was presumably normally dissipated in additional physical exercise outside the calorimeters. Clearly the scientists were more sedentary than the students, who were more active outside the calorimeter. Table 6A gives some indication of the degree to which the variability in energy requirements for a group of subjects can be ascribed to differences in spontaneous physical activity. The differences between the normal intake and the energy expenditure in the calorimeter suggest that additional physical activity accounts for 2–31% of the energy intake. The subjects with the higher energy intakes were in practice engaged in University sports. The coefficient of variation for apparent spontaneous physical activity amounts to 45% but this is unduly restricted by the method of calculating energy intakes for Group C and a figure taken by excluding Group C is 62%. This coefficient is, however, large because it includes an error measurement from energy intake data and compares with a coefficient of variability between individuals of 24% in the additional energy expended during controlled activities involved in combat training in the Edholm study set out in Table 5.

Table 7 provides data from an unpublished study by ourselves on 12 women in Cambridge. In this case there is a much greater variation in energy expenditure, which is reduced by taking the range of body size into account. Nevertheless the greater freedom of movement which these women had compared with the army recruits may explain the 30% range in energy output, a value which is not very dissimilar to our estimated range in requirements from Table 6.

The validity of weekly measures of energy expenditure

It could be argued that measurements of energy expenditure of free-living individuals, although less likely to be affected by behavioural changes during the investigation, may still be appreciably in error, because of changes in the pattern of physical activity, inappropriate timing of activities, errors in measuring metabolic rates, or because of genuine physiological changes in the metabolic rates. These problems may be assessed from the repeatability of energy expenditure measurements. Table 7 is presented in such a way as to exclude the measurement errors or physiological changes in metabolic rate from one week to another.

One may conclude that in this particular group of women the variability in physical activity from week to week must be small since the values from the non-consecutive weeks are surprisingly similar. This implies that although individuals may differ considerably in the amount of physical activity undertaken or in the energy which they expend for a given task, there are not normally substantial changes in physical activity from week to week.

More detailed data on the reproducibility of expenditure data are given in Table 8, because it seems to be the only information on this point in the literature and is difficult to find. Further analyses of this data, which include daily food intakes and daily energy expenditure, will be submitted as an appendix once computer analyses have been completed. Twenty eight army recruits had their energy output measured for three separate weeks and also had their weights and heights presented. The paper also records the basal metabolic rate of some of the men measured in each week. It is evident that there is an appreciable variation in energy expenditure in these men who were all involved in reasonably standardized 10-week initial training programmes as they entered the army. Nevertheless, the data suggest that energy expenditure values are much less variable than energy intakes on a daily basis and also on a weekly basis as noted by Edholm et al. (1970) in their presentation of the results of this large study.

The basis for the variation in energy requirements

It has been argued so far that the variability in energy requirements has been exaggerated and that once body size is taken into account then there is not a two-fold range but a range of about 50% in energy needs. Measurements of energy expenditure for only a week may exaggerate the true variability in the same way as food intake measured for a week exaggerates the range. In determining the variability within a homogeneous group of individuals it would seem that physical activity accounts for about half the variation. This in turn implies that metabolic factors must account for the rest and emphasis has usually been given to individual variability in the programming of metabolism for genetic or other reasons.

Genetic control of energy metabolism

I am unable to find any documentation of the genetic programming of metabolism which can account for the variability in energy metabolism in man. However, recalculating data presented by Durnin et al.. (1957) reveals surprisingly (Fig.2) that there is a close relationship between a mother's energy needs and those of her adult daughter. Again once body size has been taken into account there is only a 50% variation in the older woman's energy needs, the daughters having a higher energy turnover than their mothers. The paper does not provide information on the body fat content of the subjects, nor data on the basal or postprandial rates of thermogenesis in the two groups. The data, if still available could, however, be analysed to find whether this striking familial aggregation in energy requirements depends on mothers and daughters having similar basal metabolic rates or similar patterns of physical activity.

Variability in basal metabolic rates

Table 8 shows that the variation in BMR is not very great if the values are related to either surface area, as in the careful study by Robertson and Reid (1952) of the BMR of a large number of British adults, or to the fat-free mass. Robertson and Reid found a coefficient of variation of 6.6% in men and 6.7% in women when the BMR was expressed in terms of surface area. The data shown in the earlier tables also suggest that a normal range of 30% for the BMR is appropriate in well-fed subjects even when measurements are related to fat-free mass. If subjects are adapted to low energy intakes then, as suggested in the accompanying paper on adaptation, a 15% fall in BMR could be expected. On this basis a 45% range in BMR might occur in communities where food intake was poorly distributed between individuals so that some would have adequate amounts whereas others could not.

This range in BMR has been exceeded by Edmondson (1979) who selected 11 subjects for further study from a group of 57 farmers from East Java, six with a previously documented low energy intake and six with a high intake, the selection being arranged so that the average height and weight of the two sub-groups were the same. Edmondson also claimed that the two sub-groups had similar active tissue masses and this may be true if, as suggested in Table 9, there was a printing error in the paper. These data are included because they are increasingly quoted as evidence for adaptation and different efficiencies of metabolism. Yet by recalculating the data for the BMR in terms of MJ.d^-1 it is clear that the value for subject S in the high energy group must be incorrect since it approximates to the daily intake of energy. Subject S in the low group also has an unbelievably low BMR which amounts to only 2.59 MJ (619 kcal) per day. If these two individuals are excluded, then there remains a two-fold variation in metabolic rates expressed in absolute terms, but in no other study, e.g. Table 7 with data obtained by authorities in energy metabolism, is such a range found in a group of adults of similar sex, age and size. It seems, therefore, that at most a 50% range in basal metabolic rates is what one can expect in a uniform group of adults.

Variation in postprandial thermogenesis

The increase in energy expenditure after a meal and its amplification by exercise is considered by Miller to be one key to the individual variability in total energy turnover. This aspect has been dealt with in the accompanying paper on adaptation but in summary I can find no clear data from any of Miller's papers which allow one to assess his claim. A recent abstract in the Proceedings of the Nutrition Society from York's group (1980) suggests that individuals with a greater than normal energy intake have a higher thermogenic response to food. However, recalculating the basal metabolic rate values of the “low intake” groups shows that they approximate to the supposed energy intake so the validity of the data on intake must be suspect and no evidence is presented which allows a quantitation of the importance of the difference in thermogenesis between individuals with high and low intakes.

One of the difficulties in estimating the role of postprandial thermogenesis is to monitor the metabolic rate for long enough after a meal for the values to return to base-line. Thermogenesis after a protein meal may continue for 6 hours and the absorption, distribution and metabolism of fat is even slower. On this basis it is difficult to infer from short-term studies the energetic significance over a 24 hr period of thermogenesis after food, e.g. Shetty et al. (1981).

Another approach to the problem of estimating the variability in postprandial thermogenesis is to assume that the variability in energy expenditure in individuals on a prescribed physical activity regime in a calorimeter is entirely due to variation in postprandial thermogenesis once the BMR figure has been subtracted. This assumption is not valid since the variability will be affected by minor degrees of movement as well as different efficiencies during the short exercise periods. Nevertheless, it may help to give a figure for the range in energy output which can be ascribed to postprandial thermogenesis. From Table 6A (last column) it is clear that the 24 hr energy output can vary from 25 to 53% above BMR values in the calorimeter. This means that the range in energy output from the various non-BMR factors amounts to 28% of the BMR or about 20% of the total energy output.

A more specific way of assessing the quantitative role of postprandial thermogenesis is to overfeed subjects just for a single day after maintaining food intake constant for the previous week. If 24 hr measurements are made in a whole body calorimeter both while on a normal food intake and then while being given an excess of food for a single day, then the difference in calorimetric responses can be expected to be solely due to diet. This type of study has been undertaken by Dauncey (1980) and recalculating her data shows that the response may vary from 0 to 25% of the intake (James and Trayhurn, 1981). This range is therefore similar to that inferred from the other calorimetric studies. This range may be exceeded by prolonged over or under feeding but there are no experimental data to support extending the range in postprandial thermogenesis to more than 25% of the intake unless one accepts Stock's data given in the accompanying paper on the interaction of exercise and food.

Other thermogenic responses

The thermogenic response to coffee (Jung et al., 1981) or smoking (Dallosso, unpublished) is also variable but unlikely to be greater than 5% of total energy turnover.

Physical Activity

The old literature provides ample evidence, e.g. Mahadera et al. (1953) that there is variability in the energetic cost of standardized exercise which relates in part to body weight. However, when this is taken into account there is still a range in energy expenditure during the exercise which amounts to about 4.2 kJ/min, i.e. 6 MJ per day if people exercised continuously (see Fig.1, Mahadera et al., 1953). If sedentary individuals exercise for about 2 hours daily, then the difference in energetic efficiency would amount to 0.5 MJ per day or perhaps 7.5% of intake. In Norgan et al's study in New Guinea only 10% of the time is used in walking so the variability in efficiency might not be much greater in agricultural societies - perhaps reaching 10% of intake. Table 9 suggests that medium work at 100 watts can lead to an appreciable difference in work efficiency, the variability and the increment in metabolic rate being greater in those who have a higher energy turnover. These studies, however, need confirming with more precise techniques.

Conclusions

It is difficult to estimate energy requirements because contrary to current scientific opinion the choice of a week for measuring food intake leads to an overestimation of the true range in energy requirements. Measuring intakes is more likely to lead to errors induced by the measurement technique than measures of expenditure, but a week's measure of energy expenditure is insufficient to define the normal energy output pattern of the individual. Body size is an important factor explaining substantial differences between people in energy needs, but even when this is taken into account there still appears to be about a 50% (not a 2-fold) range in energy requirements.

There is a familial component to energy requirements which seems to account for just over 40% of the variance in energy requirements when these are expressed on a surface area basis (Fig.2), but it is not yet known if this has a genetic or social basis. The normal variability in basal metabolic rate can be reduced by expressing it in terms of the fat-free mass of the individual but a range of up to 20% of the total energy requirements of sedentary individuals can still be ascribed to differences in basal metabolic rates. The variability in postprandial thermogenesis may also amount to 20–25% of the energy requirements and the variability in the metabolic efficiency of physical exercise may amount to 10% of the intake. The time taken in physical activity probably accounts for the majority of the remaining variability in energy requirements. It is not yet clear which factors predominate. The data presented here and in the accompanying paper on adaptation suggest that the basal metabolic rate makes a substantial contribution towards the variability of individuals on their usual energy intakes whereas postprandial thermogenesis may play a greater role in buffering the effect of abnormally large energy intakes on body energy stores. The evidence for this conclusion is limited, however.

Data taken from Widdowson (1947)

From Boulton (1945)

From Marr (1971)

Data based on repeated 7-day weighed food records of individuals monitored either consecutively (2) or at specified intervals of time (1,3,4).

1) Huenemann and Turner (1942)

2) Yudkin (1951)

3) Unpublished data on individuals monitored on 3 alternate weeks for 6 weeks.

4) Unpublished data on individuals monitored for 3 weeks (every 4th week) for12 weeks.

Table taken from James, Bingham and Cole (1981)

Data taken from document APRC 57/1 (Adam et al. 1957a).
(n = 4 or 5) refers to the number of days in each phase of the study.
In Phases I and III the men were not involved in severe physical activity.

Groups A and B had their intakes measured for 7 days before the study and were then given food which provided approximately (Group A) or exactly (Group B) the same energy as eaten before the study. Group C were provided with an intake of 46 kcal.kg^-1 ideal body weight with some adjustment for hunger before the study; the subjects in practice maintained a constant body weight while on a carefully controlled diet for at least a week before the measurement in Groups A and C.

Data from Groups A and C are unpublished work from H. Dallosso's Ph.D. Thesis.

Data includes that from Dauncey 1980 and 1981 (Groups A and B) and unpublished data from C. Zed.

The weekly estimates of energy expenditure have not been calculated from the individual measurements of energy expenditure monitored during each week. Instead the average energy values for each individual's activity measured while undertaking a series of up to 12 categories of activity of increasing severity has been taken and applied to both weeks. The variation from week to week in this study therefore reflects changes in the recorded time taken in each activity.

Data from APRC 58/3 (Adam et al., 1957b).

* Claimed in the original table to indicate average of triceps and subscapular skinfold thickness in mm. but in the text as the sum of the two. Yet low energy intake subjects “considered very lean by Western standards”. This suggests odd errors in study since if taken with a Harpenden caliper most of the subjects would be obese. More likely that data out by a factor of 10 with group mean data of 15.4 mm for the high intake subjects and 8.4 mm for the low intake group. Work considered light amounting to 50 watts and medium work to 100 watts on a bicycle ergometer.

Data from Edmondson (1979).

FIGURE 1
Daily and weekly variations in the caloric intake of a 4-year-old girl

From Boulton, 1945 as displayed by Widdowson (1947)

Figure 2
ENERGY TURNOVER OF MOTHERS AND DAUGHTERS

The values for energy turnover have been calculated by taking the average of the figures given for energy intake and expenditure (From Durnin, Blake and Brockway, 1957).

REFERENCES

Adam, J.M., Best, T.W., Edholm, O.G. and Wolff, H.S. (1957a) Combat ration trial July 1956. Northern Ireland. Food intake and energy expenditure.

Army Personnel Research Committee. Med.Res.Council, London, Report No.APRC.57/1.

Adam, J.M., Best, T.W., Edholm, O.G., Fletcher, J.G., Lewis, H.E. and Wolff, H.S. (1957b) The dietary intake and expenditure of recruits in the British Army.

Army Personnel Research Committee, Med.Res. Council, London, Report No.APRC.58/3.

Boulton, F.M. (1945) Unpublished data presented by Widdowson (1947).

Dauncey, M.J. (1980) Metabolic effects of altering the 24 h energy intake in man, using direct and indirect calorimetry. Brit.J.Nutr., 43:257–269.

Dauncey, M.J. (1981) Influence of mild cold on 24 h energy expenditure, resting metabolism and diet-induced thermogenesis. Brit.J.Nutr., 45:257–267.

Durnin, J.V.G.A. (1976) Sex differences in energy intake and expenditure. Proc.Nutr.Soc., 35:145–157.

Durnin, J.V.G.A., Blake, E.C. and Brockway, J.M. (1957) The energy expenditure and food intake of middle-aged Glasgow housewives and their adult daughters. Brit.J.Nutr., 2:85–99.

Durnin, J.V.G.A., Edholm, o.G., Miller, D.S. and Waterlow, J.C. (1973) How much food does man require? Nature, 242, 418.

Edholm, O.G., Adam, J.M., Healy, M.J.R., Wolff, H.S., Goldsmith, R. and Best, T.W. (1970) Food intake and energy expenditure of army recruits. Brit.J.Nutr., 24:1091–1107.

Edmondson, W. (1979) Individual variations in basal metabolic rate and mechanical work efficiency in East Java. Ecology of Food and Nutrition, 8:189–195.

Huenemann, R.L. and Turner, D. (1942) Methods of dietary investigation. J.Am.Diet.Assoc., 18:562–568.

James, W.P.T., Bingham, S. and Cole, T.J. (1981) Epidemiological assessment of dietary intake. Nutrition and Cancer (in press)

James, W.P.T. and Trayhurn, P. (1981) Thermogenesis and obesity. Brit.Med.Bull., 37, 1:43–48.

Jung, R.T., Shetty, P.S., James, W.P.T., Barrand, M.A. and Callingham, B.A. (1981) Caffeine: its effect on catecholamines and metabolism in lean and obese humans. Clin.Sci., 60:527–535.

Mahadera, K., Passmore, R. and Woolf, B. (1953) Individual variation in the metabolic cost of standardized exercises: the effects of food, age, sex and race. J.Physiol., 121:225–231.

Marr, J.W. (1971) Individual dietary surveys: purposes and methods. World Review of Nutrition and Dietetics, 13: 105–164.

Norgan, N.G., Ferro-Luzzi, A. and Durnin, J.V.G.A. (1974) The energy and nutrient intake and the energy expenditure of 204 New Guinea adults. Phil.Trans.Roy.Soc.Lond.B., 268:309–348.

Robertson, J.D. and Reid, D.D. (1952) Standards for the basal metabolism of normal people in Britain. Lancet, i:940–943.

Rose, G.A. and Williams, R.T. (1961) Metabolic studies on large and small eaters. Brit.J.Nutr., 15:1–9.

Shetty, P.S., Jung, R.T., James, W.P.T., Barrand, M.A. and Callingham, B.A. (1981) Postprandial thermogenesis in obesity. Clin.Sci., 60:519–525.

Webb, P. (1981) Energy expenditure and fat free mass in men and women. Am.J.Clin.Nutr., (in press).

Widdowson, E.M. (1936) A study of English diet by the individual method. Part I. Men. J.Hygiene, 36:270–289.

Widdowson, E.M. (1947) A study of individual children's diets. M.R.C.Spec.Rep.Ser., No.257, H.M.S.O., London.

Widdowson, E.M. (1962) Nutritional individuality. Proc.Nutr.Soc., 21:121–128.

Widdowson, E.M. and McCance, R.A. (1936) A study of English diets by the individual method. Part II. Women. J.Hygiene, 36:293–309.

York, D.A., Morgan, J.B. and Taylor, R.G. (1980) The relationship of dietary induced thermogenesis to metabolic efficiency in man. Proc.Nutr.Soc.,39,57A.

Yudkin, J. (1951) Dietary surveys: variation in weekly intake of nutrients. Brit.J.Nutr., 5:177–194.