plastic femoral arterial catheter, and the administration of fluids and [Morris, 1956]. Blood samples were obtained from an indwelling

Transcription

1 THE EXCHANGE OF PROTEIN BETWEEN THE PLASMA AND THE LIVER AND INTESTINAL LYMPH.' By BEDE MORRIS.2 From the Kanematsu Memorial Institute of Pathology, Sydney Hospital, Sydney. (Received for publication 14th December 1955) LYMPH collected from different tissues throughout the body contains varying concentrations of proteins which are chemically and electrophoretically the same as the proteins in the plasma [cf. Drinker and Yoffey, 1941; Perlmann, Glenn and Kaufman, 1943; Drinker, 1946; Courtice and Morris, 1955]. That the lymph proteins come from the plasma by passing through the capillary wall has been shown by injecting labelled protein intravenously and measuring its concentration in the lymph. Such intravenously injected protein appears in the thoracic duct lymph within 10 to 15 minutes and in other lymph somewhat later [Courtice, 1943; Cope and Moore, 1944; Wasserman and Mayerson, 1951; Forker, Chaikoff and Reinhardt, 1952; Korner, Morris and Courtice, 1954]. These, and other similar investigations, have led to the concept of protein pools in the interstitial spaces of the different tissues. Protein exchange between the plasma and these pools can be studied by collecting and analysing lymph from the region concerned, since protein that escapes from the blood stream is returned by way of the lymphatics. Most studies of protein exchange between the plasma and tissue fluid have been made by collecting lymph from the thoracic duct, which probably represents about 90 per cent of the total exchange in the body. There is, however, little information concerning the protein pools in different tissues drained by this duct. This paper deals with the dynamics of interchange of labelled proteins, particularly albumin, between the plasma and interstitial protein pools in the areas drained by the hepatic and intestinal components of the thoracic duct. METHODS Acute experiments were carried out on cats anxesthetized with intravenous nembutal (Abbott). The techniques of lymphatic cannulation and collection of liver and intestinal lymph have been described [Morris, 1956]. Blood samples were obtained from an indwelling plastic femoral arterial catheter, and the administration of fluids and 1 This work was carried out with the aid of a grant from the National Health and Medical Research Council, Canberra. 2 Present address: Sir William Dunn School of Pathology, Oxford. 326

2 Plasma and Lymph Protein Exchange 327 supplementary anaesthetic was carried out through a similar venous catheter. Portal venous pressure measurements were made using the method of Burch and Winsor [1943]. A plastic catheter was passed into the portal system by way of a splenic vein tributary and tied in place. Sterile Ringer-Locke solution was used in the transfusion experiments, and this was administered at a constant flow rate using a standard Murphy's drip apparatus regulated to deliver approximately 50 or 100 ml. of fluid per kg. body weight in 5 hours. Partial occlusion of the inferior vena cava above the liver was effected by passing a linen thread around the vein, between the liver and the diaphragm. This thread was twisted until the venous pressure in the portal system was raised by about cm. of water. Protein Labelling.-Plasma proteins were labelled with T-1824, or radioactive iodine, Three ml. of T-1824 (2.5 mg./ml.) were injected intravenously to label the cat's own plasma protein. As the T-1824 binds almost exclusively to the albumin [Gregerson and Rawson, 1943], the rate of exchange of the dye is considered to give an estimate of albumin transfer rates. Radioactive iodinated human albumin was obtained from the Radiochemical Centre, England. The product supplied was indistinguishable from native human albumin on electrophoresis and contained 1 atom of iodine/molecule of albumin. The amount of radioactivity present in free form was less than 1 per cent of the total activity. Injections of this albumin were made in tracer amounts, 1-4 mg. of albumin containing 3-5 microcuries of Analyses.-Radioactive assay was carried out on 0'2 ml. samples of plasma and lymph using a thin mica end-window G-M tube. The samples were mounted on machined brass planchets and evaporated to dryness in a hot-air oven at 400 C. and counted for 8000 counts. Preparation, mounting and radioactivity measurements were carried out at the completion of each experiment. T-1824 estimations were conducted on 0-5 ml. samples of plasma and lymph, using the extraction method of Allen [1951]. Optical densities of the acetone-water extracts were read at a wavelength of 625 m,u on a Beckman model D.U. Quartz spectrophotometer. Blood samples taken for I131 and T-1824 analyses accounted for a plasma loss of about 5 per cent of the total label injected. The methods used for total protein and albumin estimations have been described [Morris, 1956]. Specific activities were calculated as the concentration of label (J131 or T-1824) divided by the concentration of albumin or total protein. Activities calculated in both these ways gave similar results for plasma-lymph equilibrium times. Any changes in plasma volume during the experiments were corrected from haematocrit readings of blood samples spun in duplicate at 3000 r.p.m. for 1 hour. VOL. XLI, NO

3 328 Morris RESULTS The Disappearance of T-1824 and I131-labelled protein from the Circulation The two plasma protein labels, T-1824 and I31-albumin, were injected together intravenously and the plasma levels followed for 8 hours. The disappearance curves are shown in fig. 1. Both labels followed an almost identical course, _ #bx. ^< suggesting no significant difference in the rate of leakage of homo- - ~ logous dye-labelled albumin and heterologous I'31-labelled human albumin during the first 8 hours. The disappearance curves were not simple logarithmic functions and showed a rapid phase during the first 2-3 hours and a subsequent almost linear disappearance up to 8 hours. After 3 hours, approxi- I mately 80 per cent of the injected labelwas present in the circulat- T2 ME HOURS ing plasma, whilst at the end of 8 TIMEHOUM FIG. 1.-The disappearance of T-1824~ and hours about 70 per cent remained labelled human albumin from the An estimate of the importance plasma following their simultaneous injec- of lymphatic return on the form tion intravenously into cats. 0: Il31-labelled human albumin. of the disappearance curve was 0: T made in a series of cats in which Mean results for 3 experiments. the hepatic and thoracic ducts The lower curve (A) shows the mean disappearance for labelled albumin in 5 cats w with hepatic and thoracic lymph ducts with its contained labelled protein, cannulated. collected. In these animals the disappearance curves approximated closely to a logarithmic function, and at the end of 3 hours the mean percentage remaining within the circulation was 78 and at the end of 8 hours 55 (fig. 1). The average amount of label collected in the liver and intestinal lymph of 5 cats during the 8-hour period was 11 per cent of the amount injected. The diversion of lymph from the blood stream, therefore, accounts for most of the increased amount of labelled protein lost from the circulation when the lymphatic ducts were cannulated. Lymphatic return of leaked protein was thus seen to be of considerable importance in reducing the rate of loss from the plasma after the first 3 hours. The calculated mean disappearance rate of the labelled protein from the circulation was 0*14 per cent per minute, and the half-life corrected for sampling dilution was 8-7 hours.

4 Plasma and Lymph Protein Exchange 329 The Appearance of Labelled Protein in the Lymph There were striking differences in the rate of appearance and final concentration of the label in lymph collected from the liver and intestines. Lymph from the liver contained appreciable amounts of label in the first i-hour, with a rapid increase to a maximum concentration about 3 hours after injection. The levels of labelled protein in the liver lymph from this period onwards amounted to about 90 per cent or more of the plasma label levels (fig. 2). TIME HOUR8 FIG. 2.-The appearance of radioactivity (left) and dye (right) in the hepatic and intestinal lymph of 2 individual cats following the intravenous injection of 131-labelled human albumin and T *: plasma. 0: hepatic lymph. A: intestinal lymph. Significantly smaller concentrations of labelled albumin appeared in the intestinal lymph, reaching maximum values of about 50 per cent of the plasma levels (fig. 2). The first '-hour sample of intestinal lymph usually contained only small amounts of label, and in some experiments none was collected in this period. The proportion of labelled albumin to total albumin in the intravascular compartment reaches an equilibrium with that in the interstitial fluid, when this extravascular compartment becomes saturated with filtrate protein. The plots of the specific activities of the plasma and lymph show that equilibration occurs between the plasma and interstitial protein pools of different areas at different times. The interstitial protein pool of the liver rapidly equilibrated with the plasma and the specific activities were identical after about 3 hours. Thereafter there

5 330 Morris was a steady and uniform decline in the specific activities of the plasma and liver lymph, but they remained identical throughout the 8-hour experimental period. The specific activity of the intestinal lymph had not equilibrated with the plasma by 8 hours, and extrapolation of the activity curves suggested an equilibrium occurring somewhere between 12 and 14 hours (fig. 3). I1-51 T-_8Z4 a. 'I) I" O0 4 6 TIM4E HOUR8 FIG. 3.-The specific activities of the plasma, hepatic and intestinal lymph of 2 individual cats following the intravenous injection of I131- labelled albumin (left) and T-1824 (right). *: plasma. 0: hepatic lymph. A\: intestinal lymph. The Size of the Hepatic Interstitial Pool When an equilibrium has been reached between the specific activities of the plasma and liver lymph, then the protein in the interstitial protein pool of the organ will have been completely replaced by protein derived from the capillary filtrate. If the liver lymph is collected during the period between the injection of the labelled protein and the time of equilibration, then the total protein and the labelled protein can each be measured. The protein derived from the interstitial pool and the capillary filtrate can each be calculated, and also the volume of the pool. This involves three assumptions: (1) that there is a random leakage of labelled protein from the circulation, (2) that no differential breakdown of the tagged molecules occurs, and (3) that the size of the interstitial pool remains unchanged. All of these seem justifiable. The size of the fluid pool was calculated from the protein concentration of the lymph, which is considered to have the same protein content as the average tissue fluid in the area from which the lymph is derived. The calculated

6 Plasma and Lymph Protein Exchange 331 mean interstitial protein was 015 g./100 g. of liver tissue, whilst the interstitial fluid was about 3 per cent of the total liver weight (Table I). TABLE I. THE HIEPATIC INTERSTITIAL PROTEIN AND FLU-ID POOLS IN THE CAT Pr otein Fluid I Experiment Liver wei'lt, 9. mg. mg./100 g. ml. per cent liver tissue Weight Alean The rapid replacement of the hepatic interstitial protein pool by capillary filtrate proteini is shown in fig. 4, where the proportions of pool protein to capillary filtrate protein in the liver lymph have been plotted at different time intervals throughout the equilibration phase. After the first hour more than 50 per cent of the original interstitial protein was collected in the lymph and replaced by capillary filtrate material T-& I a3 2 a TIME HOURS FIG. 4.-The relative contributions of the interstitial and filtrate protein to the hepatic lymph. Results are shown for 2 cats, using I131-labelled albumin (left) and T-1824 (right). *: interstitial protein. 0: filtrate protein. The Effect of Increased V'enous Pressure on Plasma-Lymph Protein Exchange Lymph Flow and Protein Concentration. Experiments on plasmalym--ph protein exchange were carried out to assess the influence of an increased venous pressure on the rate of exchange of labelled albumin

7 Morris between the plasma, liver and intestinal lymph. The pressure in the portal system was increased by partial occlusion of the vena cava between the liver and the diaphragtm. In fig. 5 is shown the effect of the increased :2 5 0_ A A a oe 0 - I A - A LL to.-a j - t4 18 i E i= TIME MIMUTES Fic. 5. The changes in lymph flow (ml./hr.), portal pressuire (cm. H20) and plasma andl lyinph protein concentrations (g.p.c.) following partial occlusion of the inferior vena cava above the liver. 0: plasma. 0: hepatic lymph. A: intestinal lymph. venous pressure on the liver and intestinal lymph flow and protein concentration. The liver-lymph flow began to increase about 5 minutes after the venous pressure was raised and ran parallel with the increase in pressure. An increase in portal pressure of about 12 cm. of H20 produced a ten- to twelvefold increase in lymph flow. After a period of 1. to 1 hour the venous pressure began to fall and the lyiuph flow declined. Simultaneously with the rise in lymph flow the lymph protein concentration increased, and at very high flow rates and portal pressures the protein content of the liver lymiph and plasim-a were approximately equal. The increases in intestinal lymph flow were much smaller, and there was little change in the protein concentration (fig. 5).

8 Plasma and Lymph Protein Exchange Disappearance of Labelled Protein fromn the Plasma and its Appearance in the Lytnph.-At the end of 3 hours, 70 per cent (mean of 5 experiments) of the injected tagged protein remained in the circulation. The shape of the plasma disappearance curve was altered from that seen in the normal cat and no longer approximated a simple logarithmic function. The early part of the curve showed a rapid disappearance phase for the first 1 to 2 hours after injection. I-535 T TIME HOURS FiG. 6. The effect of partial occlusion of the inferior vena cava on the specific activities of the plasma, hepatic and intestinal lymph. Results are shown for 2 separate experiments. 0: plasma. 0: lhepatic lymph. A: intestinal lymph. Labelled protein appeared in the liver lymph more rapidly than in normal animals, and could be detected in appreciable amounts in the first 15 minutes after injection. The concentration rose to a maximum usually in the first j- to 1 hour, by which time the specific activities of the plasma and lymph were the same and remained equal thereafter. There appeared little change in the rate of appearance and concentrations of labelled protein in the intestinal lymph from the normal animal, and the equilibration timne of the specific activities was not greatly reduced (fig. 6). The Effect of Intravenous Infusions on the Exchange of Proteins between Plasmna and Lymiph Lymph Flow aitd Venous Pressur-e Changes during Transfusion.-The effect of intravenous infusions was quite different from that of partial occlusion of the inferior vena cava. The intestinal lymph flow increased

9 334 Morris after the first i-hour and rose to a maximum level after about 3 hours, and thereafter an even plateau was maintained throughout. The increase was about 10 to 15 times the resting level, with infusion volumes of 100 ml./kg./5 hrs. The hepatic lymph flow, on the other hand, rose to only about double the pre-transfusion level. During these infusions the portal pressure increased about 2-3 cm. of water (fig. 7). :22 a. ~~16 14 cre. _a a g TIlM MIMUTEO FIG. 7.-The changes in lymph flow (ml./hr.), portal pressure (cm. H20) and plasma and lymph protein concentrations (g.p.c.) during the intravenous infusion of Ringer-Locke solution 100 ml./kg./5 hrs. *: plasma. 0: hepatic lymph. A: intestinal lymph. Disappearance of Labelled Protein from the Plasma and its Appearance in the Lymph.-The disappearance curves again showed rapid initial slopes lasting for about 2 hours, after which the form of the curve was much flatter. At the end of 3 hours, 60 per cent of the labelled albumin was present in the plasma, whilst at the end of 5 hours 51 per cent remained (mean of 7 experiments). The rate of appearance of labelled protein in the lymph was increased by transfusion and the time of specific activity equilibration shortened. With transfusion volumes varying between 50 and 100 ml./kg./5 hrs., the plasma and liver lymph specific activities equilibrated in about

10 Plasma and Lymph Protein Exchange 50 ML/KG/5 HR. 100 ML/KG/5 H R 335 TIME HOURS FIG. 8. The effect of intravenous infusion of Ringer-Locke solution on the appearance of radioactivity in the hepatic and intestinal lymph. Left, infusion volume 50 ml./kg. body wvt./5 hrs. Right, inifusion volume 100 ml./kg. body wt./5 hrs. 0: plasma. 0: hepatic lymph. A: intestinal lyniplh. 50 ML/KG/5 HR 100 ML/KG/5 HR. 1 2 a TIlE HOURS FIG. 9. The effect of intravenous infusion of Ringer-Locke solution on the specific activ ities of the plasmiia, hepatic and intestinal lyimph. Left, infusion- volurne 50 ml./kg./5 l rs. Right, infusion x oluine 100 rnl. kg./5 hirs. *: plasma. 0: hepatic lymph. intestinal lymph.

11 336 Morris 2 hours. The larger infusion volumes caused dilution of the liver lymph, and the plasma and lymph label curves became divergent (figs. 8 and 9). The plasma and intestinal lymph equilibrium time was also reduced. Maximum label concentrations occurred after 2 to 3 hours and specific activity equilibration at about 5 hours (figs. 8 and 9). With transfusion volumes of 100 ml./kg./5 hrs. the intestinal lymph label concentrations fell rapidly after the first 3 hours, due to the formation of large volumes of dilute lymph (fig. 8), and the protein content of the lymph reached very low levels. In two transfusion experiments, one in which T-1824 was used and the other 1131, the intestinal lymph and plasma specific activity curves did not meet in the 5-hour experimental period, although by this time these activity curves had become parallel. It was thought that in these animals this was evidence of newly mobilized tissue protein entering the interstitial pool. This did not occur with the plasma and liver lymph specific activity curves, and it appeared that no significant amounts of newly mobilized protein was being transported by the hepatic lymph. DIsCUSSION Whilst both T-1824 and I131 have been used extensively as protein labels in blood volume and protein exchange studies, there is still some disagreement as to the identity of their behaviour in the animal organism. The results of Freinkel, Schreiner and Athens [1953] and Schultz, Hammarsten, Heller and Ebert [1953] suggest that both T-1824 and I131-tagged albumin behave in a similar fashion when injected intravenously into humans and dogs, although Wasserman and Mayerson [1951] found differences in the behaviour of these two tags. The results presented in this paper indicate that for the early equilibrationi phase at least, both T-1824 and I'31 are distributed in a similar manner between the intravascular and extravascular fluids. The Plasma Disappearance Curve of Labelled Albumin The form of the disappearance curve of labelled protein from the circulating plasma in the first 12 hours will be influenced chiefly by the rate of exchange of labelled protein between the plasma and the interstitial pools [cf. Wasserman and Mayerson, 1951]. If this exchange process took place at the same rate with the pools of all tissues throughout the body, this curve would be a simple exponential function. There are, however, apparently very significant differences in the rate of exchange of protein in various parts of the body. Such differences in exchange rates may be due to variations in capillary permeability, effective filtration area or local haemodynamic effects.

12 Plasma and Lymph Protein Exchange 337 In 8 hours about 45 per cent of the injected labelled albumin escaped from the circulating plasma, and about 11 per cent of this was returned to the circulation by the lymphatic system. The principal interstitial pools into which this protein leaked were those of the liver and the intestines. The early steep part of the disappearance curve was due to leakage of labelled protein in these pools, and the later flatter part of the curve was due to its lymphatic return. In particular, it is seen that the hepatic interstitial protein pool equilibrates very rapidly with the plasma, and represents a mobile protein compartment from which "leaked" protein is returned rapidly to the circulation. The total amount of protein leaking into the intestinal pool was greater than the amount leaking through the hepatic capillaries. The difference is probably due chiefly to differences in the available filtration area and filtration pressures in the two capillary beds. The exponential nature of the disappearance curve in cats with cannulated lymphatics suggests that in these experiments the labelled protein passes from the plasma and appears in the lymph at a uniform rate, and the sizes of the interstitial pools do not change greatly. The time taken for equilibration of the plasma and interstitial pool specific activities in any tissue will depend on the rate of protein leakage (a function of capillary permeability, filtration area and pressure) and the size of the protein pool in which the filtrate protein must mix. These parameters no doubt vary greatly in different regions of the body. It appears almost certain from the present experiments that the hepatic sinusoids are more permeable to protein than are capillaries elsewhere in the body. Whilst this greater degree of capillary permeability is important in influencing the rate of turnover of the hepatic pool, the most important factor responsible for the rapid mobilization of the liver interstitial protein is the small size of this pool. The intestinal pool, on the other hand, is considerably larger than that of the liver, and any protein leaking out in this area must mix with a large volume of pool fluid and protein before entering the lymphatic system. This, apart from any considerations of capillary permeability, will delay the equilibration time in this area. Thus when assessing plasma-lymph exchange phenomena, based on thoracic duct lymph collection, these differences should be borne in mind. The Hepatic Interstitial Pool Whilst the total extracellular fluid in the liver represents about 38 per cent of the organ weight [Harrison, Darrow and Yannet, 1936], it is seen that the interstitial component of this is only about 3 per cent. This interstitial component would include all fluid and protein in the liver lymphatics proper and the hepatic interstitial spaces. The lymph vessels themselves will contain a large part of this fluid and protein. As

13 338 Morris the liver lymphatics of the cat do not penetrate beyond the Glisson's capsule [Lee, 1923], it follows that the amount of interstitial fluid and protein within the hepatic lobule itself is very small. It is noted that the pool size estimate is based on collection of lymph draining into the thoracic duct from lymphatics leaving the hilus of the liver. Some lymphatics are known to accompany the hepatic veins, but the volume of lymph drainage in this direction is not known. Previous work on plasma-lymph protein exchange [Wasserman and Mayerson, 1951] has shown the early appearance of labelled albumin in the thoracic duct lymph, following its intravenous injection. It is clear that the tagged protein appearing earliest in the thoracic duct lymph will have been filtered through the hepatic sinusoids. Wasserman and Mayerson [1951] found equilibration between the specific activities of plasma and thoracic duct lymph of dogs occurred somewhere between 7 and 13 hours after injection. The results presented here show equilibration times for liver and intestinal lymph of about 3 and 12 hours respectively. The question of mobilization of protein is difficult to answer. In all experiments there was a significant reduction in the concentration of circulating plasma protein, due to the collection of the lymph. A considerable colloid osmotic stress was placed on these animals after several hours of lymph collection, and the question of replacement of diverted lymph protein arises. In the present experiments there was no evidence for newly mobilized tissue protein in the hepatic lymph. However, in some of the transfusion experiments there appeared evidence of newly mobilized protein entering the intestinal lymph, and this presumably had origin in the cells of tissues from which the lymph was derived. Korner, Morris and Courtice [1954] presented evidence for the mobilization of protein in transfused cats in which the thoracic duct lymph was collected throughout the transfusion period. Protein Exchange Following raised Portal Pressure The vena caval occlusion experiments emphasize the very large volumes of lymph that will form in the liver in conditions of raised intrahepatic venous pressure. That this increased flow of lymph is accompanied by an increase in its protein content was recognized by Starling [1894]. This contrasts with other tissues where an increase in filtration pressure produces a greater volume of lymph, but its protein content falls. In our experiments the raised venous pressure did not permit the passage of cellular elements from the blood, and in most experiments there always remained a small protein concentration gradient between the plasma and lymph. It appears, then, that the capillary walls of the hepatic sinusoids readily permit the escape of large amounts of plasma proteins normally,

14 Plasma and Lymph Protein Exchange 339 and that any condition producing a raised intrahepatic sinusoidal pressure will result in an increase in both liver lymph flow and protein content. Investigations of protein exchange during transfusion [Wasserman and Mayerson, 1952; Korner, Morris and Courtice, 1954] have shown increased rates of loss of protein from the circulating plasma over those found in normal animals. The experiments described here show that during transfusion most of the increased thoracic duct lymph flow is due to increased filtration in the intestinal capillary bed. The increased lymph forms with the rise in venous pressure, but the volume of intestinal lymph is far greater than with larger venous pressure rises produced by partial occlusion of the inferior vena cava. An increase in filtration area is indicated, and it appears there is an opening up of capillary beds in the splanchnic area under the stimulus of an increased circulating blood volume. There appears also an expansion in the size of the intestinal interstitial protein and fluid pools, as seen by the change in form of the plasma disappearance curves. SUMMARY 1. The exchange of T-1824 and 1131-tagged albumin between the plasma, liver and intestinal lymph has been investigated in cats. Both disappeared from the plasma at similar rates for 8 hours after intravenous injection. 2. In intact, nembutalized cats after 8 hours about 70 per cent of the injected albumin remained in the circulation, whereas only 55 per cent remained in animals with the liver and thoracic lymph ducts cannulated. 3. The tagged albumin appeared first in the liver lymph and reached 90 per cent of the plasma levels in 3 to 4 hours; in the intestinal lymph it reached about 50 to 60 per cent of the plasma levels. 4. The specific activities of the plasma and liver lymph became identical after 3 to 4 hours, whereas equilibration of the plasma and intestinal lymph required 12 to 14 hours. 5. The size of the hepatic interstitial protein pool was calculated as 154 mg./100 g. of liver tissue, and the interstitial fluid as 3 per cent of the organ weight. 6. Following partial occlusion of the inferior vena cava above the liver there was an increase in liver lymph flow which accompanied the rise in portal venous pressure. An increase in portal venous pressure of about 10 to 12 cm. of H2O produced a ten- to twelvefold increase in the liver lymph flow and an equilibration of the plasma and liver lymph specific activities in about 1 hour. There were relatively small changes in the intestinal lymph flow.

15 340 Morris 7. Intravenous infusions of Ringer-Locke solution 100 ml./kg./5 hrs. produced large increases in intestinal lymph flow and reduced the equilibration time for the plasma and intestinal lymph specific activities to about 5 hours. There was a much smaller increase in liver lymph flow, and specific activity equilibration between plasma and liver lymph occurred after about 2 hours. 8. The significance of these results in the interpretation of capillary permeability and protein exchange dynamics is discussed. ACKNOWLEDGMENTS I should like to thank Dr. F. C. Courtice and Dr W. J. Simmonds for their help in the preparation of the manuscript, and Miss Marianne Kearns for technical assistance. REFERENCES ALLEN, T. H. (1951). Proc. Soc. exp. Biol. Med. 76, 145. BURCi, G. E. and WINSOR, T. (1943). J. Amer. med. Assoc. 123, 90. COPE, 0. and MOORE, F. D. (1944). J. clin. Invest. 23, 241. COURTICE, F. C. (1943). J. Physiol. 102, 290. COURTICE, F. C. and MORRIS, B. (1955). Quart. J. exp. Physiol. 40, 138. DRINKER, C. K. (1946). Ann. N.Y. Acad. Sci. 46, 807. DRINKER, C. K. and YOFFEY, J. M. (1941). Lymphatics, Lymph and Lymphoid Tissue. Cambridge, Mass.: Harvard University Press. FORKER, L. L., CHAIKOFF, I. L. and REINHARDT, W. 0. (1952). J. biol. Chem. 197, 625. FREINKEL, N., SCHREINER, G. E. and ATHENS, J. W. (1953). J. clin. Invest. 32, 321. GREGERSON, M. I. and RAWSON, R. A. (1943). Amer. J. Physiol. 138, 698. HARRISON, H. E., DARROW, D. C. and YANNET, H. (1936). J. biol. Chem. 113, 515. KORNER, P. I., MORRIS, B. and COURTICE, F. C. (1954). Aust. J. exp. Biol. and med. Sci. 32, 301. LEE, F. C. (1923). Contrib. Embryol. Carnegie Inst. of Washington, D.C. 15, 63. MORRIS, B. (1956). Quart. J. exp. Biol. 41, 318. PERLMAN, G. E., GLENN, W. W. L. and KAUFMAN, D. (1943). J. clin. Invest. 22, 627. SCHULTZ, A. L., HAMMARSTEN, J. F., HELLER, B. I. and EBERT, R. V. (1953). J. clin. Invest. 32, 107. STARLING, E. H. (1894). J. Physiol. 16, 224. WASSERMAN, K. and MAYERSON, H. S. (1951). Amer. J. Physiol. 165, 15. WASSERMAN, K. and MAYERSON, H. S. (1952). Amer. J. Physiol. 170, 1.