1. The concentration of protein in the serum and plasma of normal dogs is given. Analyses of serum from 38 animals yielded the following averages and standard deviations, (a) for albumin: 3.26 ± 0.48 gm. per cent, (b) for globulin: 2.72 ± 0.76 gm. per cent, and (c) for total protein: 5.98 ± 0.67 gm. per cent. Analyses of plasma from 19 animals showed, (a) for albumin: 3.38 ± 0.38 gm. per cent, (b) for globulin: 2.98 ± 0.55 gm. per cent, and (c) for total protein: 6.36 ± 0.71 gm. per cent. 2. A diet for dogs is described, the feeding of which results in a progressive decline in the concentration of protein in the serum. A composite curve constructed from the findings with 21 animals discloses a rapid initial fall and a slower subsequent decrease in albumin and total protein and an approximately constant level for globulin. The course of the globulin curve was subject to wide variation in separate experiments, both increases and decreases being recorded. 3. With five dogs the nitrogen balance was followed through a total of 42 metabolism periods of approximately 7 days each. The average daily loss of nitrogen was 1.15 gm. Approximate calculations disclose that only 3 or 4 per cent of the nitrogen eliminated is accounted for by the decline in circulating protein, the remainder being represented by loss from the tissues. 4. An episode is described with one dog when, during a period of self-imposed fasting, the serum albumin regenerated to a normal level, apparently at the expense of catabolized tissue protein. 5. The course of serum proteins is described during the recovery which follows interruption of the low protein diet and return to a regime of adequate feeding. 6. A discussion is given of the relationship between tissue proteins and plasma proteins. The data permit one to entertain the hope that a way will be found for stimulating an internal readjustment to provide temporary relief from hypoproteinemia, a way which will depend upon the potential ability of the tissues to provide sufficient protein for the needs of the plasma.

In the course of experiments undertaken to discover the effect of the flow and the composition of lymph on the formation of edema an opportunity has been afforded for comparing the albumin and globulin concentrations of blood serum with those of lymph serum obtained from lymphatics of the extremities of dogs. The comparison forms the basis of this communication. METHODSThe dogs used were of mongrel breeds and varied in weight from 15 to 26 kilograms. Lymph was obtained through cannulas inserted into one of the lymphatic trunks just above the ankle of either a front or a hind leg. Cannulization was performed sometimes under ether anesthesia, sometimes under nembutal anesthesia and sometimes with the aid of a local anesthetic (novocaine) only. With animals under general anesthesia the flow of lymph was promoted by gentle massage of the foot and ankle and passive motion of the extremity. The non-anesthetized dogs were prepared by the method described by White, Field and Drinker (1) and allowed to walk about the corridors, the normal motion of the extremity furnishing the stimulus for lymph flow. A fine fibrin clot which formed in the lymph on standing was removed before analysis. Blood was obtained by puncture of the femoral artery usually at the end of the period of lymph collection and the serum separated as soon as possible. In several instances blood serum obtained both before and after the collection of lymph failed to indicate any change in composition during the experiment.Methods of chemical analysis have been described in a previous paper (2). In some instances, because the quantities of lymph were limited, nonprotein nitrogen was determined on the serum only and the value so obtained used in calculating the lymph proteins. The error involved in such an assumption is small. In nine instances in which parallel determinations were made of lymph and serum nonprotein nitrogen, the average value for lymph was 24.5 mgm. per 100 cc. and for serum 22.7 mgm. per 100 cc.

Two years ago in a paper from this laboratory (1) some observations were presented on the development of nutritional edema in the dog. The observations were based on experiments with seven animals which developed hypoproteinemia and edema after subsisting for variable times on a diet low in protein. Since the time of that publication the experimentally induced edema has been utilized as a means of studying particular aspects of the edema problem (2) and a considerable array of additional data has been accumulated. In the light of this further experience it appears desirable to describe in more detail the events which have been observed. The present communication will deal with the behavior of the serum and tissue proteins during the experiments. MethodsAnimals.--25 dogs of mongrel breeds used in these exi~riments varied in weight from 13 to 22 kilos and averaged 17 kilos. Young adult dogs were selected by preference; their ages, not definitely known, were estimated to be between 1 and 3 years.Diet.--The first eight dogs in the series were maintained either on a synthetic diet or a natural food diet, both of which have been described previously (1). For the remaining seventeen dogs the composition of the natural food diet was modified by removing the butter fat and supplying its nutritional equivalent with lard and cod liver oil. The composition of the modified diet follows: gm, Carrots

1. The experimental observations have been summarized at the end of an earlier section. The more important facts only will be recapitulated here. The capacity of the lymphatics for removing fluid from the tissues greatly exceeds the rate at which freshly formed tissue fluid can be made available for removal. Edematous regions can be rendered non-edematous by the application of measures, such as massage, passive motion, or normal exercise, which activate the lymphatics. During continuous activity the rate of lymph flow is first variable and later relatively constant. Constant rates of flow must correspond to the production of fresh lymph. A study of the constant rates indicates that lymph formation in the edematous animal is certainly only slightly greater, and possibly not greater at all, than under conditions of normality. When the protein of plasma decreases, the protein of lymph is also lowered. The loss of protein from lymph takes place at a faster rate than from plasma, so that the ratio of serum protein to lymph protein is greater in the edematous than in the normal animal. In edematous animals the concentration of protein in lymph is of the same order of magnitude as the concentration in edema fluids. The two fluids are not, however, identical in composition. Minor fluctuations in the protein content of lymph always occur during a period of continuous collection. 2. The factors involved in the circulation and accumulation of tissue fluid are discussed. Reasons are given for offering the following suggestions. Significant differences in tissue pressure or tension exist between the states resulting from quiescence and activation of the lymphatics. The differences give rise to variations in the relative areas of capillary wall, functioning for filtration and reabsorption. When the lymphatics are activated it is possible that capillary reabsorption may be completely in abeyance. A decline in the proteins of plasma may be associated with a diminished permeability of the capillaries. Such a lowering of capillary permeability would account for two features, both of which have been demonstrated: (1) failure to observe an appreciably increased rate of lymph formation in the edematous animal, and (2) the extremely low concentration of protein in lymph from edematous animals. Although the difference between the protein concentrations of edema fluid and lymph from the same region is small, the conclusion is not yet justified that a similarly small difference exists between normal tissue fluid and normal lymph.

Clinical edema of several types and its relation to low serum proteins have been adequately discussed in recent publications (1,2,3,4,5). Experimental edema in animals has resulted from two methods which have in common the lowering of serum protein, that is, plasmapheresis in dogs (6,7,8,9,10,11), and low protein diets in rats (12) and in the dog (8). The experiments to be discussed in this paper were performed by utilizing both methods of experimental approach. Dogs were maintained on a diet deficient in protein. Edema was produced sometimes acutely by plasmapheresis, sometimes more slowly as the result of dietary deficiency alone, and in several instances after recovery from the edema of plasmapheresis the edema of protein starvation was allowed to supervene in the same animal. Nine dogs have been followed over periods varying from 59 to 107 days and averaging 84 days. It has seemed worthwhile at this time to present together the results of both methods of study and to compare the events attending the onset of edema in both groups of animals and the relationships found for plasma protein concentrations and plasma specific gravity. METHODSThe animals were kept in metabolism cages throughout the periods of observation, the daily urine collections being measured and aliquots saved for analysis. The importance of washing the cages to insure complete urine collection was not appreciated at first. Later observations indicated an average loss of 1.9 per cent through neglect of this precaution and accordingly records of total urine have been increased by this amount. Stools were not saved regularly but only over short periods often enough to determine approximately the necessary balance corrections. Since the substances studied (sodium, chloride and nitrogen) are excreted chiefly in the urine, these corrections were not large. The diets were offered in weighed amounts and corrected for refusals National Research Council Fellow. 193