Department of Nutrition and Food Science
Massachusetts Institute of Technology
Cambridge, Massachusetts 02139
At the outset, it is essential to consider what the objectives of determining a maintenance requirement should be. The term "maintenance" implies constancy of body weight and composition, and is traditionally the concept of the adult as compared to the growing animal. Recent studies on adult human subjects demonstrate that this concept is false. From an analysis of their own and other published data, Forbes and Reina (1970) have concluded that lean body mass declines progressively with age from about 30 years onwards, the rate of decline being somewhat more rapid in the male. It would in fact seem that there are only two phases of life, namely growth followed by aging, and that the latter sets in when or shortly after growth ceases. Many bodily functions (cardiac capacity, muscle strength etc.) attest to this inexorable decline with age (Shock, 1970). If we accept that the adult undergoes continuous age-related loss of bodily functions and body protein, maintenance requirements become those levels of nutrients which minimize aging changes most effectively. From the practical point of view, measurement of protein requirements in this way is not attainable on the basis of present information. Wellknown studies on rats (Ross, 1959, 1961) do not, however, suggest that a high intake of protein prolongs life. We are therefore probably secure enough in taking as the maintenance (adult) requirement for health an amount of protein that is just sufficient to maintain various measures of protein metabolism at an optimal level over relatively short periods of time.
The 1965 Report on Protein Requirements accepted what has come to be known as the factorial approach to protein requirements for, maintenance, in which the obligatory nitrogen losses of subjects receiving a protein-free diet are individually estimated and then added to provide the irreducible minimum loss of body protein that has to be replaced from the diet. Two major questions thus arise: what are the routes of obligatory nitrogen loss that have to be measured and what is the magnitude of N loss through each channel under basal conditions? A third question which is also obvious is when one should
Fig. 1. Endogenous urinary N output (Scrimshaw et al.)
measure the endogenous N output of subjects receiving a protein-free diet since prolonged depletion will reduce output; this problem will be considered in discussing minimum urinary N output. A survey of all available studies on man has recently been assemble by Irwin and Hegsted (1971a).
When a protein-free diet is fed to a human subject or to a rat, there is a rapid decrease in urinary N output for a few days, followed by a plateau or near-plateau (Fig.1). The 1965 Report concluded that the point of inflection between these two phases represents the minimum output of N in the urine before protein depletion sets in. This minimum output was assigned a value of 46 mg N/kg body weight by the following somewhat circuituous calculation. Direct estimates of N output at the point of inflection were not very numerous, but it had been claimed on the basis of studies with several animals (Smuts, 1935) that endogenous urinary N output is close to 2 mg/kcal of basal energy output. In man, basal metabolism has been measured often and with greater precision than basal N output, and if the basal metabolism of a man of 65 kg is taken as 1500 kcal/day, then the figure for obligatory urinary N output of 46 mg/kg is derived by simple computation.
Subsequent to the 1965 Report, there have been further studies of the minimum N output of human subjects on protein-free diets. Young and Scrimshaw (1968) published extensive data for eight male students followed over a 10-day period, and they have since extended these studies to one hundred similar subjects examined for a 16-day period on the proteinfree diets. From the daily urinary N output of this group (Fig. 1) it is apparent that the point of inflection in the N output curve is not sharp enough to be easily identifiable and in addition, at this point there is high individual variability. Second, N output does not continue to decline indefinitely, but is quite constant between the 9th and 16 th days on the protein-free diet and individual variability is much lower. These observations suggest that it is better in practice to accept the plateau level from 9 to 16 days as a more easily identifiable and more reproducible end-point of endogenous N output. Table 1 shows basic and derived data obtained between the 11th and 14th day on the group of 100 subjects. The urinary N output was 37 mg N/kg body weight; since the basal diet contained 6 mg N/kg body weight of vegetable protein, with an assumed NPU of 50, we can subtract 3 mg N from this estimate to give 34 mg N/kg as the obligatory N loss on a truly proteinfree diet. Basal metabolic rate was measured directly on the same subjects, and gave a figure of 1.4 mg N/kcal, which is considerably lower than the figure suggested by Smuts. The other extensive new studies are those of Calloway and Margen (1971) who fed "proteinfree" diets providing 8 mg N/kg to a group of 13 men and obtained an endogenous N output in the urine of 38 mg per kg body weight; on correction for dietary N, this represents 34 mg N/kg. Again, endogenous output per kcal basal metabolism was considerably less than the figure given by Smuts. An unpublished study by Inoue et al. (1971) on 9 Japanese subjects gave a value of 33 mg urinary N/kg between days 7 and 14; the N content of their proteinfree diet is not known.
The estimated obligatory urinary N outputs of 34 mg per kg obtained in these two series of extensive studies and Inoue's figure of 33 mg are in fact compatible with the earlier literature if the point-of-inflection concept is abandonned. The experiments of Martin and Robison (1922) that were compatible with the inflection value of 46 mg/kg in the 1965 Report show an average value of 34 mg N/kg when followed to the plateau and 30 mg when corrected for dietary N. Murlin et al. (1946) report 22 subjects on a protein-free diet (4 mg N/kg) for 4 days who excreted 32 mg N/kg. i.e. a corrected value of 30 mg. On the other hand, subjects after a prolonged period of protein depletion were studied by Smith (1926) and by Deuel et al. (1928) and a urinary N output of 24 mg/kg was eventually obtained in the case of Deuel there was probably calorie insufficiency as well as protein lack. In summary, it would seem reasonable to take 34 mg N/kg as the plateau value for the adult.
Even on a diet without protein, there is a loss of N in the feces, representing enzyme
Obligatory N Output in Urine and Feces
|Mean Coeff. of Var. Scrimshaw(a) et al.||Mean Coeff. of Var. Calloway(b) and Margen|
|Number of men||83||13|
|Body weight (kg)||73.5||13||70.8||12|
|BCM or LBM (kg)(c)||35.1||10||60.5||4|
|Basal urine N (g/day)||2.69||18||2.41||12|
|Basal fecal N (g/day)||0.63||24||0.96||15|
|Urinary creatinine (g/day)||1.72||10||1.59||12|
|Urine N (mg/kg)||
|Fecal N (mg/kg)||8.8||23||14||21|
|Urine N (mg/kcal)||1.4||17||1.4||25|
|Fecal N (mg/kcal)||0.4||24||0.3||11|
|Urine N (mg/kg BCM or LBM)||77.2||16||44||23|
Fecal N (mg/kg BCM or LBM)
(a) Scrimshaw et al., unpublished data
(b) Calloway and Margen (1971)
(c) BCM = body cell mass (Scrimshaw et al.); LBM = lean body mass (Calloway and Margen).
- - - - -protein and desquamated intestinal cells that have not been fully digested and reabsorbed. This is sometimes referred to as fecal metabolic nitrogen, since it represents N wastage due to digestion. The 1965 Report allowed 20 mg N per kg body weight for this component of obligatory N loss. However, the data of Young and Scrimshaw (1968) and of Calloway and Margen (1971) give much lower figures, namely 9 and 14 mg/kg respectively (Table I) and Inoue et al. (1971) report 12 mg/kg. Data obtained by Hegsted et al. (1946) have been recalculated in a body weight basis by Irwin and Hegsted (1971a) to show a value for fecal metabolic nitrogen of 6 to 7 mg/kg. Presumably some of the discrepancy in these estimates is due to differences in the types of non-protein diets used, notably the effect of dietary carbohydrate on intestinal flora. As a compromise, we may suggest a value of 10 mg N/kg body weight for obligatory fecal N output.
Nitrogen is lost from the skin in the form of desquamated cells, hair and nail clippings and also as sweat. In early studies of nitrogen balance, it was common practice to regard cutaneous losses as negligible and in any case likely to be unaffected by diet. However, Mitchell (1949) reported a long-term N balance study of adults based on dietary N minus urinary and fecal N that showed a retention averaging 1.38 gm N daily without a corresponding gain in body weight. In a subsequent extensive review of the literature (Mitchell and Edman, 1962), Mitchell equated this positive balance with integumental N losses. This view was reflected in the value of 20 mg N/kg body weight assigned to cutaneous N losses in the 1965 Report. Further evidence suggests that this figure is quite excessive. In an extensive study of 20 young men, Sirbu et al. (1967) investigated the effect of level of protein intake on skin cell and sweat losses, and hair and nail losses under controlled environmental conditions. At a normal level of protein intake (76 gm per day), skin and sweat loss averaged 119 mg N/day, while hair and nails raised the value to 143 mg N/day, which is 2 mg integumental N/kg body weight. The skin and sweat loss was influenced by dietary protein level, being less when the subjects received a diet low in protein. The values obtained by Sirbu et al. are much less than other direct estimates of cutaneous losses indicate. The hair, nail and skin cell losses by a single subject studied by Voit (1930) amounted to 179 mg N/square meter body surface area, that is, a total loss of about 300 mg N for an average subject of 1.7 square meters area. Even if we assume that Voit was measuring insensible perspiration along with desquamated cells, this value is still twice the figure obtained by Sirbu et al.
Other estimates of sweat N losses under comfortable environmental conditions are 360 mg N/day for men (Mitchell and Hamilton, 1949), 250 mg N/day for sedentary African men (Darke, 1960) and 250 mg N/square meter for men and 120 mg/sq. meter for women (Kraut and Müller-Wecker, 1960), that is, about 5 mg N, 4 mg N, 6 mg N and 3 mg N/kg body weight respectively. It is, however, difficult to believe that such sweat collections do not also include much of the daily desquamation of cells. Using a total integumentary collection technique including hair and nails, Inoue, et al. (1971) reckon the dermal loss as 7 mg N/kg at comfortable temperatures. In view of these various findings, it would seem that 5 mg N/kg body weight would cover cutaneous losses of nitrogen under comfortable environmental conditions.
When environmental conditions promote heavy sweating, considerably more nitrogen is lost. In short-term studies of exposure of soldiers to high temperatures, Consolazio et al. (1963) calculated from direct measurements on arm sweat that an average of 1.4 gm N (in one subject 2.6 gm N) can be lost daily from the body by this route when working moderately under adverse climatic conditions. Furthermore there was no compensatory reduction in urinary N loss; consequently, protein requirements would be increased 13 to 14% under their conditions. This view can be challenged on two grounds. First, the short period of heat stress may have increased adrenocortical activity and have thus led to a negative N balance. Bourges and Scrimshaw (unpublished data) have demonstrated that, in trained subjects, sweat loss due to exercise is compensated by a reduction in urinary N. Second, these excessive sweat N losses may not be typical of subjects habituated to living and working in hot climates. Ashworth and Harrower (1967) support this latter view with data obtained from subjects acclimated to a tropical climate and working 6½ hr. daily at heavy work. They observed a mean daily loss of 0.5 gm N (7 mg N/kg body weight) in the sweat during the heavy work period. It is thus not surprising that Ashmore and Harrower calculated that their subjects achieved positive N balances at their rather low N intakes (8 gm/day) compared with the negative balances found by Consolazio et al. It is also probable that the higher N intake of Consolazio's subjects (14-15 gm/day) resulted in greater N outputs in sweat because of the elevated blood urea levels which are known to be correlated with sweat N losses.
Finally, Inoue et al. (1971) observed an average dermal and sweat N loss of 6 mg/kg at temperatures of 6-22°C, and 15 mg/kg at 25-30°C.
Factorial Calculation of Protein Needs of Adults
|1965 Report||Revised estimates|
Obligatory N Losses:
|mg N/kg b.w.|
|+ Stress factor (10%)||95||-|
|Protein Replacement:||gm protein/kg b.w.|
|Obligatory N expressed as protein||0.59||0.34|
|Replacement by dietary protein (NPU 70)||1.01||0.64|
|Total protein intake (NPU 70) for 65 kg subject)||66 gm||42 gm|
Evolution of gaseous N has periodically been suspected as a factor to be included in the nitrogen balance equation. Costa (1960) found a large discrepancy in the N output compared with the N intake of several animals and hypothesized that it could only be accounted for by loss of gaseous nitrogen. In subsequent experiments with human subjects in a metabolic chamber, Costa et al (1968) claimed to have obtained evidence that they evolved nitrogen when the gas in the chamber consisted of helium and oxygen. They computed that 8 to 24 mg N/kg were released daily in this way. A similar study on chicks and rats using an argon-oxygen atmosphere failed to demonstrate any release of nitrogen (Lewis and Evans, 1970). Hoffmann and Schiemann (1964) failed to find such a loss when ammonium sulfate-15N was given in large doses to rats. There is, nevertheless, evidence of a small amount of gaseous N formation by intestinal bacteria (Calloway et al., 1966). Steggerda and Dimmick (1966) analyzed the gases of intestinal flatus; from their data Waterlow (1969) computed a loss of 0.25 gm N per day, (4 mg N/kg body weight) on a normal diet and up to 0.87 gm N per day on a diet of Boston beans.
There is also a small loss of body N due to menstruation; during the menstrual period, this amounts to 1.5 to 3 gm N (Gillett et al., 1918). Computed over the whole month, this would increase the daily obligatory N output by about 50 to 100 mg N/day, or 1 mg N/kg body weight. In the case of the male, the loss of N through seminal ejaculations has not been computed, but must be negligible since the average 3 ml ejaculation contains no more than 50 to 100 mg N and thus would contribute about 1 mg N/day to bodily loss of N.
The thesis used in the 1965 Report was that protein requirements are the amounts needed in the diet to replace obligatory losses of nitrogen. In order to translate N loss data into dietary protein, the calculations shown in Table II are necessary. The Table gives the figures used in the 1965 Report and also the revised estimates of obligatory N provided by the literature survey presented above. When all routes of obligatory N loss are taken together, the 1965 Report data add up to 86 mg N/kg, whereas the revised data amount to 54 mg N/kg. (Evolution of gaseous N has been put at 4 mg and menstrual or seminal losses at 1 mg, to provide an additional entry for the revised estimate in Table II). Direct measurements of urinary, fecal, dermal and sweat N losses made by Inoue et al. (1971) on 9 subjects between the 7th and 14th day on a protein-free diet add up to 53 mg/kg., compared with 49 mg for the same three entries in Table II. The 1965 Report considered that an additional 10% should be added to the estimated N losses, to allow for increased N output arising from minor stresses of living and infections, giving a final obligatory N loss of 95 mg N/kg body weight. However, these minor stresses also apply to normal subjects undergoing metabolic studies if they are not confined to a metabolic ward; this applies to the data of Scrimshaw et al. obtained on regular students (Table I). In such cases, there should be no additional stress correction of the estimates of obligatory N losses. Accordingly, no such increases have been added to the revised estimate of obligatory nitrogen losses (Table II).
These estimates of obligatory N loss can be expressed as amounts of body protein that have to be replaced daily from dietary sources, using the conversion factor of N × 6.25 to give the weight of protein. However, the computed estimates of protein requirements (0.59 and 0.34 gm protein/kg body weight) are average figures and we must allow for individuals who have needs in excess in these mean figures in order to achieve N balance. From a survey of previous studies on protein needs, mainly those of Sherman (1920), the 1965 Report considered that the standard deviation of the estimates is about 10%, so that addition of 20% to the estimated mean would cover the protein requirements of all except the 2.5% of the population whose needs lie in excess of two standard deviations above the mean. The data of Scrimshaw et al. (Table II) show a coefficient of variation of 15% for the combined obligatory N losses in the urine and feces; some of this variability may represent errors in technique, but on repetition of the studies on some of the subjects, most individuals assumed their previous level above or below the mean value for obligatory N output, this suggesting real individual variability. The data of Murlin et al. (1946) for endogenous urinary N output also display some of this reproducibility (for individuals studied twice, 33.5 and 31.4; 33.2 and 32.0; 32.3 and 29.0; 31.4 and 30.6; 30.4 and 29.9 mg N/kg b. wt.).
Consequently, the revised estimates in Table II include the addition of 30% (twice the S.D. of 15%) to cover individual variations. The amount of body protein to be replaced thus becomes 0.71 gm and 0.45 gm/kg body weight respectively for the two estimates. If the dietary protein has a net utilization (NPU) of 70, the amount required to replace these estimates of protein needs would be 1.01 and 0.64 gm/kg body weight respectively, or 66 gm and 42 gm per day for a person weighing 65 kg.
The use of body weight as a reference standard requires some comment. It is well known that, over a range of mature mammals of increasing body size, basal metabolism and most other metabolic phenomena are related to the 0.7 power of body weight and not body weight itself (Munro, 1969) and indeed in their studies of minimum protein needs for maintenance Hegsted et al. (1946) found a slightly better correlation between these and surface area (r = 0 71) than body weight (r = 0.61). Observation on obligatory N losses made on human subjects do not span a sufficient range of weights to test various relationships to body size rigorously Table I displays the coefficients of variation for the basic and derived data of Young and Scrimshaw (1968) and of Calloway and Margen (1971). Figures for fecal N tend to show a higher variance, presumably because of the difficulty of collections. It is noteworthy that the variance for basal urinary N output is not appreciably reduced by expressing it per kcal basal energy or per unit of metabolic tissue (LBM or BCM). Since common sense dictates that the size of a person must affect his need for nutrients, and since weight is the most easily measured or computed reference standard, its continued use would appear to be justified for adjustment of protein requirements within the narrow weight range presented by adult man.
The factorial method predicts what amounts of dietary protein should be provided in order to achieve N equilibrium in 98% of the population. It should be possible to test these predictions by feeding different amounts of protein and finding the minimum amount compatible with N equilibrium. Although simple in concept, this approach suffers from the general defects inherent in N balance studies, which tend to overestimate true protein intake and underestimate N output, thus leading to a more positive balance than exists. A second problem that it is more difficult to circumvent is that an allowance has to be made for cutaneous and other minor routes of N excretion in calculating N balance, and these depend on the same published observations as those used in the factorial computations. Consequently, most data on N balance at different levels of protein intake can only tell us the point at which N output in the urine and feces is equal to N intake. In terms of the revised estimates of obligatory urinary and fecal N shown in Table II, this should occur for the average subject at an intake of fully utilizable protein (NPU = 100) of 0.28 gm/kg; with cutaneous and other losses added in, this average requirement rises to 0.34 gm/kg, as shown in the Table.
With this in view, some studies of minimum protein intakes for N equilibrium can be considered. Sherman (1920) determined the minimum amount of protein for N equilibrium (urine and feces) for a series of 109 subjects receiving diets predominantly consisting of cereals, and concluded that 0.5 gm mixed dietary protein/kg body weight was adequate for maintenance of the average adult. Assuming an NPU of 60, this indicates that 0.3 gm high quality protein/kg body weight would suffice for equilibrium. Hegsted et al. (1946) found with an all-plant protein diet fed to 26 subjects that N balance was also achieved with an intake of about 0.5 gm protein/kg; since the BV of the protein of this diet was estimated to be 73, the subjects' protein requirements would be met by 0.38 gm/kg of a protein of maximal BV (100). Hegsted's figures are also based on N balance determined by urinary and fecal N measurements only. Bricker et al. (1945) studied ten young women at different levels of protein intake on a diet with 70% protein of cereal origin (NPU = 61) and found a mean requirement of 0.54 gm protein/kg body weight; for NPU = 100, this would be 0.32 gm protein/kg. Even this low figure may be somewhat inflated, since Bricker et al. used Mitchell's large correction for integumental N loss. Calloway and Margen (1971) report a study in which either egg white protein or whole egg protein was fed in amounts equivalent to the determined endogenous N output of the subjects on a protein-free diet (Table I). In this experiment also, cutaneous losses were included with urinary and fecal N in estimating obligatory N output on a low protein diet, and amounted to 3.50 gm N per day for the group of 13 men. However, nine of these 13 failed to achieve N equilibrium when given 3.5 or 4.0 gm N as egg proteins, and it was computed from regression equations that the achievement of N equilibrium would require 4.8 gm egg protein N per day, which is equivalent to 0.43 gm protein/kg body weight. Furthermore, the computed requirement had a coefficient of variation of 22%, which is considerably larger than the variability obtained in N output on protein-free diets. The observation that much more egg protein was needed to achieve N equilibrium than was predicted from direct determination of obligatory N losses implies that the egg protein was not well utilized and indeed the authors calculate a net utilization of 65%; this suggests that the egg protein used may have deteriorated in some way. A similar observation was made by Young and Scrimshaw (1968), who fed a diet providing dried egg at a level calculated to replace their daily endogenous urinary output of 37 mg N/kg. In fact, urinary N output rose to 59 mg/kg and the biological value of the egg protein was calculated to be 65 instead of 90-100.
Deterioration of the egg protein was confirmed by NPU determinations on growing rats, which gave a value of 79 instead of 93-97. Finally, Inoue et al. (1971) examined the minimum intake of egg protein for N balance and calculated the average requirement to be 0.65 gm/kg body weight at an energy intake of 45 kcal/kg and 0.46 g/kg at 57 kcal/kg. At the lower energy level, the subjects lost weight slightly and the second figure may be closer to normal needs; furthermore, the scatter of N balances in these studies gives rather a high error, so that the error of the estimate of 0.46 is uncertain.
Most of these studies suggest that the revised estimates of protein needs given in Table II must be the right order of magnitude, with an average protein requirement of 0.34 gm/kg body weight and that 0.45gm of fully utilized protein/kg will cover the needs of 98% of adults in a population.
The N balance of adult subject on an adequate intake of protein is sensitive to variations in energy intake, whether from carbohydrate, fat or ethanol (Munro, 1964). Contrary to a common misconception, this relationship holds not only for the negative N balance caused by reducing energy intake below requirements, but is also true for increments in energy intake above requirements. Both insufficient and excessive intakes of energy have effects on N balance that are quite prolonged. For example, Keys et al. (1950) reduced the energy intake of 10 subjects from their normal level of 3500 kcal/day to 1600 kcal/day for 24 weeks, and observed a negative N balance of -3.8 gm/day during the first 12 weeks and -1.3 gm during the second 12 weeks (even without allowing for cutaneous losses). The effects of excessive intakes of protein have not been studied over such long periods, but are known to persist undiminished in healthy subjects for at least two weeks (Cuthbertson and Munro, 1937).
This means that, in the usual metabolic study lasting 1 to 2 weeks, N balance will be the resultant of the level of energy fed as well as the protein level. It would therefore seem important to define the adequacy of caloric intake in experiments in which the minimum protein intake for N equilibrium is being examined. For example, Inoue et al. (1971) compared N balance using three synthetic diets, providing first 4 gm essential amino acid N (rice pattern) and 10 gm total N, second 4 gm essential amino acid N and and 6 gm total N, and third 0.6 gm essential amino acid N and 6 gm total N. Caloric intake was varied on each diet, and N equilibrium was obtained at 40 kcal/kg with the first diet. At this caloric intake subjects on the second diet were 1.0 gm N daily in negative balance and about 2.0 gm in negative balance on the third diet; however, at 45 kcal/kg caloric level, they came into positive balance on the second diet, and at 53 kcal/kg on the third diet. This emphasizes the need to define protein requirements in relation to caloric intake.
The effect of short bouts of physical exertion on protein metabolism has been the subject of research for more than a century. Provided the extra energy cost of the exercise is precisely balanced by an increase in energy intake, the changes in N balance are quite small (Cuthbertson et al., 1937). In the present context, a more important question to answer is whether men accustomed to heavy physical labor have a higher requirement for protein. Yoshimura (1955) observed anemia and hypoproteinemia when young men undertook a program of heavy physical exertion on a diet providing 60 gm protein/day, mainly from vegetable sources. In similar studies, Bourges (1968) studied young adults, adapted to a work load on diets containing 0.71 gm/kg or 0.59 gm/kg of egg protein, or an amount calculated from endogenous urinary and fecal N output to cover their requirements. It was concluded that subjects adapted to work loads do not have an additional requirement for protein. A feature of Bourges' studies mentioned earlier is that they demonstrate how N losses in the sweat due to exertion can be an important factor in assessing N balance in relation to exercise.
These studies all represent the need for an adequate protein intake to support muscular development during training. There appear however, to be no data that will answer the basic question of whether the trained manual worker needs more protein to sustain his superior musculature. In practice, in Western countries he does in fact consume more protein along with his extra caloric intake, a fact attested by every dietary survey of food intakes from Playfair (1865) onwards.
The effects of high environmental temperatures on N balance have already been discussed. Cold also influences protein metabolism and its effects on protein requirements must also be mentioned.
In a cold environment, rats on a constant food intake show an increased excretion of nitrogen (Lathe and Peters, 1949; Ingle et al., 1953). However, the change in nitrogen output is not an inevitable consequence of the raised energy metabolism. In short-term studies on fasting human subjects (Voit, 1878) and fasting dogs (Rubner, 1902) and in an investigation of the metabolism of sheep at different planes of nutrition (Graham et al., 1959), the increased heat output evoked by the cold environment came exclusively from the combustion of fat. A recent extensive study on human subjects (Issekutz et al., 1962) shows that, although energy metabolism changes immediately, the increase in N output can be delayed by as much as 2 days after the commencement of exposure to a cold environment, and the increased nitrogen excretion eventually produced persists for several days after returning to the warm environment. The action of cold on protein metabolism probably occurs mainly through an increase in activity of the thyroid gland and adrenal cortex (You et al., 1950; Hardy, 1961), both of which would cause an impairment in nitrogen balance. The independence of action of a cold environment on energy requirements and on protein requirements is also suggested by the finding that a diet too low in protein to support survival of rats in a warm environment will nevertheless permit growth of the rats in a cold environment (Andik et al., 1963). Presumably, the increased food intake demanded by energy needs in the cold environment results in a better intake of protein; if protein requirements were increased in proportion to the greater energy needs, growth rate would not be improved in the cold environment despite the larger amount of protein consumed. The fact that the rats grew in the cold shows that protein requirements were not increased in proportion to energy requirements.