1 International Journal of Sport Nutrition, 1994, 4, Human Kinetics Publishers, Inc. Plasma Glucose Levels After Prolonged Strenuous Exercise Correlate Inversely With Glycemic Response to Food Consumed Before Exercise Diana E. Thomas, John R. Brotherhood, and Janette Brand Miller It was hypothesized that slowly digested carbohydrates, that is, low glycemic index (GI) foods, eaten before prolonged strenuous exercise would increase the blood glucose concentration toward the end of exercise. Six trained cyclists pedaled on a cycle ergometer at 65-70% V02max 60 min after ingestion of each of four test meals: a low-gi and a high-gi powdered food and a low-gi and a high-gi breakfast cereal, all providing 1 g of available carbohydrate per kilogram of body mass. Plasma glucose levels after more that 90 min of exercise were found to correlate inversely with the observed GI of the foods (p <.01). Free fatty acid levels during the last hour of exercise also correlated inversely with the GI (p <.05). The findings suggest that the slow digestion of carbohydrate in the preevent food favors higher concentrations of fuels in the blood toward the end of exercise. Key Words: glycemic index, carbohydrates, free fatty acids, insulin, endurance exercise Plasma glucose and free fatty acids (FFAs) provide substrates for exercising muscles in addition to the glycogen stored in the working muscle. Carbohydrates (CHOs) in the form of muscle and liver glycogen and plasma glucose are a limited source of energy in the body, while there are comparatively limitless amounts of stored fatty acids. Depletion of body CHO is associated with a lowering of blood glucose and the onset of fatigue (1 1). Because CHOs are an essential fuel for strenuous exercise (>65% VO,max), strategies have been sought to maintain the availability of CHOs during prolonged exercise (2,4, 14). Most involve consumption of solutions containing CHOs such as glucose or maltodextrins during exercise in order to keep the plasma glucose levels high. In this situation, glucose from the gastrointestinal tract can be employed by the muscle in addition to glucose derived from gluconeogenesis in the liver and glycogen D.E. Thomas and J. Brand Miller are with the Human Numtion Unit, Department of Biochemistry, University of Sydney, N.S.W. 2006, Australia. J.R. Brotherhood is with the National Occupational Health and Safety Commission (Worksafe), P.O. Box 58, Sydney, N.S.W. 2001, Australia.
2 362 / Thomas, Brotherhood, and Brand Miller stores. Ingestion of maltodextrins during prolonged strenuous exercise has been found to significantly increase cycling time (4). A similar effect might be obtained by eating a slowly digested CHO food before exercise commences (14). These foods produce a flattened glycemic response and release glucose from the gut for many hours after ingestion (12). The rate of digestion and absorption of the CHO is reflected in the food's glycemic index (GI), a ranking of foods based on the postprandial glycemic responses (9). We postulated that slowly digested CHOs (i.e., low-gi foods) eaten 1 hr before exercise would provide a sustained source of glucose to the blood without an associated insulin surge to inhibit the release of FFAs from fat stores. In a previous study, eight trained cyclists cycled at 67% V02max until exhaustion after consumption of foods of high and low GI (13). The low-gi food (lentils) maintained higher levels of glucose and FFAs during exercise and extended exercise endurance by 20 min compared with the high-gi food (potato). In the present experiment, the aim was to determine whether the magnitude of the glycemic response to different foods ingested preexercise predicted the plasma fuel substrate levels late in exercise. Since plasma glucose levels during the recovery phase may be important to glycogen restoration, we also examined their relationship to the food ingested before exercise. Subjects Methods Six healthy, trained male cyclists participated in the study. Their physical characteristics (mean + SD) were as follows: age, 25 f 5 years; height, 176 f 6 cm; weight, 70.6 f 5.5 kg; VO,max, Llmin. Subjects regularly cycled an average (f SD) of 215 f 123 km per week. Selection criteria included a regular training schedule that would remain the same throughout the period of study. Pretrial protocol required the volunteers to follow a specified meal plan and training schedule on the 2 days prior to each test, in order to minimize the withinsubject variation in starting glycogen levels. The aim was to study subjects with a consistent, submaximal level of muscle glycogen. This protocol has been shown to give reproducible levels of muscle glycogen (4). The protocol was approved by the Medical Ethical Review Committee of the University of Sydney, and all subjects gave their written, informed consent. Study Design Each subject performed four trials in random order, in which one of four test foods was consumed 1 hr before the subject cycled at 65-70% V02max for as long as possible. The volunteer arrived at the laboratory in the morning after a 12-hr overnight fast, and a sampling needle was inserted into a vein in the left forearm for blood sampling. The subject rested 15 min before sampling commenced, and then the food was consumed in the 10 min following the time 0 sampling. The subject then remained seated in a chair until 60 min, when exercise on the cycle ergometer started. The subjects were instructed to continue cycling as long as they felt they could comfortably maintain the set pedaling rate, but they were not encouraged to continue to complete exhaustion. The exercise was terminated when the pedaling rate fell by more than 10% of the original pedaling rate. Every 15 min, the heart rate was recorded (Polar tester,
3 Plasma Glucose Levels / 363 Finland) and the rating of perceived exertion assessed. The laboratory was maintained at "C and 6045% humidity. During cycling, a small electric fan was directed on the subject. Subjects were weighed before and after each trial, and we minimized dehydration by encouraging subjects to drink 120 ml water after each blood sampling. Blood samples were taken at 0 min and then every 15 min throughout the experiment. The end of exercise sample was taken immediately after the subject indicated he would be unable to maintain the required pedaling rate. Blood (8 ml) was drawn into an EDTA-coated blood collection tube, an aliquot for triplicate micro hematocrit readings was removed, and the sample was spun at 3,000 rpm for 5 min at 5 "C. Plasma was drawn off, divided into four storage tubes, and frozen at -20 "C until analyzed for glucose, insulin, lactate, and FFAs. Blood (2 ml) was collected in a plain tube for the insulin analysis. The sampling needle, which remained in place for the duration of the test, was kept patent with 3.8% citrate (D. Bull, Australia). Expired air was collected before the food was ingested and at 15 min and 45 min after consumption. Timed expired air collections were made every 15 min during exercise, and respiratory gases were analyzed for oxygen and carbon dioxide for calculation of the oxygen consumption and the respiratory exchange ratio (RER). Four foods that varied from high to low GI, according to published tables, were selected (9). The composition of the test meals is shown in Table 1 (5). The four meals were a flaked low-gi food based on lentils, a flaked high-gi food based on potato, and one low- and one high-gi breakfast cereal, based on bran and rice, respectively. The amount of CHO ingested was equal to 1 g of available CHO per kilogram of body mass. The available CHO was determined from ma~iufacturers' data. The volume of each meal was made the same (600 ml) with additional water where necessary. To improve palatability, the flaked foods were consumed with 100 g tomato (3 g CHO in addition to the CHO in the potato and lentil flakes), while the breakfast foods were ingested with 250 ml of low-fat milk (additional 12 g CHO). The tests were given in random order 1 week apart. The observed GI was calculated as the percentage incremental area under the plasma glucose response curve (AUC) for the first hour after ingestion, Table 1 Average Composition of the Test Foods and the Observed Glycemic Index Food Test serve Observed (g dry wt) Available Energy Protein Fat GI M SD CHO (g) (g) (g) M SEM Potato flakes meal , Rice cereal meal , Lentil flakes meal , Bran cereal meal , Note. Potato was US& as the reference food (i.e., GI = 100).
4 364 / Thomas, Brotherhood, and Brand Miller relative to the corresponding AUC for the reference food (potato in this study). The trapezoidal rule was used to calculate the AUC, and the fasting value was taken as the baseline, with negative areas being ignored (16). The observed GI values agreed closely with published GI values (9). Subjects were not informed of their cycling times or of the hypothesis being tested, and the investigators were blind to the food being tested. Before commencing the food trials, the volunteers performed a maximal exercise test and then a stepped submaximal test on a cycle ergometer to confirm the percentage V02max as previously described (13). Results are expressed as mean f SEM. The data from the four trials were compared using two-way analysis of variance for repeated measures, Fisher's least significant differences, and analysis of covariance (Minitab Statistical Software and Statview) (' <.05). Results Figure 1 shows the plasma glucose changes after the four test foods. The mean fasting plasma glucose concentration was 5.2 f 0.2 mmol/l. After ingestion, all the foods caused a rise in plasma glucose, which was highest for potato (4.8 f 0.5 mmol/l) and lowest for lentils and bran cereal (1.4 f 1.0 mmol/l and 2.2 f 1.0 mmol/l, respectively, p <.05, Table 2). The observed GIs (mean k SEM) of the test meals ranged from 30 f 9 for the bran breakfast cereal to 100 for potato, the reference food (Table 2). Plasma glucose levels fell to a nadir at food Time (rnin) Figure 1 - Mean change in plasma glucose concentration (mmolk) preexercise, during exercise, and 30 min postexercise. Food ingestion occurred at 0 min, and cycling commenced at 60 min. *Significant differences are shown in Table 2.
5 Table 2 Summary of Results of the Four Food Trials (Analysis of Variance for Repeated Measures and Fischer's Least Significant Differences) Significant Potato (a) Rice (b) Lentils (c) Bran (d) differences M SEM M SEM M SEM M SEM 0) <.05) Observed glycemic index (%) Peak plasma glucose change (mmol/l) End-exercise plasma glucose change (mmol/l) Postexercise plasma glucose change (mean of 15 and 30 min) (mmol/l) Peak plasma insulin change (piu/l) Postrecovery plasma insulin change (mean of 15 and 30 min) (piu/l) Peak FFA change (meq/l) FFA (AUC) during exercise (meq/l. min) FFA 30 min postexercise (meq/l) RER (AUC) during exercise Exercise times (min) Percentage V0,max Heart rate (beats per min) Percentage decrease in body mass during exercise a vs. c, a vs. d a vs. c, a vs. d a vs. c, a vs. d, b vs. d, b vs. c a vs. d, b vs. d, c vs. d a vs. c, a vs. d, b vs. c, b vs. d a vs. c, a vs. d, b vs. d a vs. d a vs. d 9 a vs. b, a vs. c, g a vs. d 2 a vs. b, a vs. c, C a vs. d n None $ r None 8 None 2 None 1 W m LA
6 366 / Thomas, Brotherhood, and Brand Miller Glycemic index of pre-event food (%) Figure 2 -The relationship between the mean observed GI of the preevent meal and the change in plasma glucose concentration at the end of exercise. Symbols represent the six different subjects studied. The line represents the mean change for each food versus GI. min into exercise that was not statistically different among the foods, before increasing. By the end of exercise (95 f 12 min), the plasma glucose levels relative to fasting were as follows: bran cereal ( mmol/l), lentil flakes (0.5 f 0.3 mmol/l), rice cereal (-0.2 f 0.5 mmol/l), and potato (-1.0 f 0.6 mmol/l) (p <.05, Table 2). There was a significant inverse relationship between the change in plasma glucose at the end of exercise and the GI (Figure 2, analysis of covariance coefficient , p <.01, Table 3). In other words, for every one unit change in GI, there was a change in the plasma glucose level at the end of exercise. Postexercise plasma glucose changes varied among the meals. The bran cereal was associated with the highest plasma glucose levels, while the potato meal gave the least (p <.05, Table 2). There was an inverse correlation between GI and plasma glucose area under the curve (AUC) in the postexercise period (coefficient , p <.05, analysis of covariance, Table 3). Plasma insulin changes after ingestion of the test foods are shown in Figure 3. The mean fasting insulin was 8.9 f 0.8 piu/l. After ingestion of potato flakes and rice cereal, plasma insulin levels rose rapidly, reaching a peak at 30 min (38.5 f 11.9 and 32.3 f 6.4 piu/l, respectively). After lentil flakes or bran, there was a smaller rise in plasma insulin, peaking at 30 min for bran and 45 min for lentil flakes and reaching levels half those obtained in the other two trials. During the early phase of exercise, insulin levels declined in all four trials, so that by 15 min into exercise there were no differences among the foods. Similarly, there were no significant differences at the end of exercise. However,
7 Plasma Glucose Levels / 367 Table 3 Relationship of the Observed Glycemic Index of the Foods to the Metabolic Parameters Measured (Analysis of Covariance) Result Coefficient SD t value p value Exercise parameters Change in end exercise plasma glucose level (mmol/l) Change in plasma FFA concentration during exercise (AUC 60 to 120 min, meq/l. 60 min) Change in RER during exercise (AUC 60 to 120 min) Postexercise Change in plasma glucose concentration after exercise (AUC 0 to 30 min recovery, mmol/l. 30 min) Change in plasma insulin concentration after exercise (AUC 0 to 30 min recovery, piu/l. 30 min) Change in plasma FFA concentration at 30 min postexercise (meq/l) during recovery plasma insulin levels began to diverge, and postexercise the plasma insulin AUC correlated negatively with the observed GI (coefficient , p <.05, analysis of covariance, Table 3). The changes in plasma FFAs are shown in Figure 4. The mean fasting FFA concentration was 0.56 f 0.13 meq/l. Plasma FFAs decreased after ingestion of all the test meals, significantly faster for potatoes than for the other foods. As a result, the potato trial produced a more negative AUC during exercise than the bran cereal (p <.05, Table 2). The AUC for FFA during exercise correlated negatively with the observed GI (coefficient = -0.02, p <.05, analysis of covariance, Table 3). Postexercise, the bran cereal elicited the lowest levels of plasma FFA while the potato flakes produced the highest (p <.05, Table 2). At 30 min postexercise, plasma FFA concentrations correlated positively with the observed GI of the foods (coefficient = 0.014, p <.01, analysis of covariance, Table 3). Changes in plasma lactate levels are shown in Figure 5. Small rises and falls, without differences among the foods, occurred during the preexercise period. After the onset of exercise, plasma lactate rose rapidly and then fell, again with no significant differences among the foods. The changes in RER during exercise are shown in Figure 6. The mean k SEM baseline value was 0.83 & The RER rose immediately after food consumption, reaching a peak during the early part of the exercise period. The extent of the rise was influenced by the food consumed. During exercise, the
8 368 / Thomas, Brotherhood, and Brand Miller food Time (min) Figure 3 - Mean change in plasma insulin concentration (yiu/l) preexercise, during exercise, and 30 min postexercise. Food ingestion occurred at 0 min, and cycling commenced at 60 min. *Significant differences are shown in Table 2. food Time (min) Figure 4 - Mean change in the plasma free fatty acid (FFA) concentration (meq/ L) preexercise, during exercise, and 30 min postexercise. Food ingestion occurred at 0 min, and cycling commenced at 60 min. *Significant differences are shown in Table 2.
9 - i! i 4 - rice Plasma Glucose Levels / bran cereal lentils cereal potato 0 4 pre- exercise post- -1 I ' I * I ', / / ; I I food Time (min) Figure 5 - Mean change in plasma lactate concentration (mmolil) preexercise, during exercise, and 30 min postexercise. Food ingestion occurred at 0 min, and cycling commenced at 60 min. high-gi potato trial gave the highest AUC for RER while the low-gi bran cereal trial produced the lowest (p <.05, Table 2). Thus, on an individual basis, there was a positive correlation between the GI of the meal and the AUC for RER during exercise (analysis of covariance coefficient = 0.08, p <.01, Table 3). The exercise times varied, but the differences among the foods did not reach statistical significance (Table 2). There was no correlation between trial times and observed GI. As required by the protocol, the values for mean percentage V02max during exercise were similar among the four trials (Table 2). Similarly, the mean heart rates during exercise were comparable for all trials, and there were no significant differences in the ratings of perceived exertion (Table 2). Body mass decreased by less than 1% and was similar for each trial (Table 2). Discussion In this study we demonstrated that the GI of preexercise food influences the blood levels of muscle fuel substrates during exercise. In particular, after more than 90 min of exercise, low-gi foods provided higher levels of blood glucose and FFA than did high-gi foods. Since higher levels of plasma glucose, achieved without depressing plasma FFA, are likely to favor endurance, the results lend support for the hypothesis that low-gi foods offer an advantage over high-gi foods in prolonged strenuous exercise (1 3). Rate of digestion and absorption of CHO in food has been shown to be a major determinant of its GI rather than the fat, protein, or fiber content or gastric
10 370 / Thomas, Brotherhood, and Brand Miller Figure 6 - Mean change in the respiratory exchange ratio (RER) preexercise, during exercise, and 30 min postexercise. Food ingestion occurred at 0 min, and cycling commenced at 60 min. *Significant differences are shown in Table 2. emptying (12, 15). Starch in many foods, such as potatoes, is rapidly digested compared with the slowly digested starches in foods such as lentils (12). The reasons relate to the molecular structure of the starch, particularly the ratio of amylose to amylopectin, and physical accessibility of the starch to enzymes. Our desire to test the effect of real foods rather than artificial mixtures meant that the non-cho component of the meals varied. We standardized the volume of the foods to minimize differences in gastric emptying for this reason, but the energy, fat, and protein content varied according to the food source. Total calories, and fat content in particular, may have influenced gastric emptying, but this is reflected in the food's GI. We cannot discount that these differences in composition per se may be responsible for the differences in blood-borne fuels at the end of exercise, although we regard this as unlikely. Further studies with foods of varying GI but identical nutrient composition are warranted. One reason to consider the practical implications of consumption of food before exercise is that it may be difficult to drink glucose solutions during the event, especially in sports such as swimming or in heavy occupations such as firefighting. Low-GI foods could provide a continuing source of glucose at physiologically favorable rates from the start of exercise. Furthermore, consuming 70 g of CHO in the preexercise food is equivalent to drinking 700 ml of 10% glucose solution during exercise. Our findings have practical implications for the choice of foods eaten prior to prolonged strenuous exercise. Foods vary by more than threefold in their GI, lentils having one of the lowest values and potatoes one of the highest (9). In fact, potatoes produce a glycemic response similar to that of an equivalent glucose load. Breakfast cereals vary by twofold, with porridge having a GI about half
11 Plasma Glucose Levels / 371 that of cornflakes (9). We found that a 10-unit difference in GI was associated on average with a 0.2 mmol/l difference in plasma glucose concentration at the end of exercise (Figure 2). Thus a food with a GI of 50 would be predicted to give plasma glucose levels 1 mmol/l higher at the end of exercise than a food with a GI of 100. Both glucose and FFA are important oxidative fuels during prolonged exercise. Blood glucose contributes significantly to the CHO energy required during prolonged, strenuous exercise, and evidence indicates that it becomes the dominant CHO energy source during the latter stages (2). CHO ingestion during exercise can delay fatigue and improve performance, and this effect appears to be partly due to prevention of a decline in blood glucose levels (4). Blood glucose utilization increases as the duration of exercise increases, and rates of over 1.2 glmin have been observed at the end of exercise to exhaustion at 67% V0,max (1). As muscle glycogen decreases, blood glucose represents a progressively increasing proportion of total CHO oxidation (4). The rate of FFA oxidation by muscle is approximately proportional to the concentration of fatty acids to which it is exposed (6, 7). Increased availability of FFA has been found to delay exhaustion in rats subjected to prolonged running, and this has been attributed to a glycogen-sparing effect (8). Elevating the plasma FFA concentration prior to exercise was also shown to improve endurance performance in humans (3). Maintained plasma FFA levels therefore may be an advantage in prolonged exercise; however, there is still controversy regarding the role of plasma FFA, and further research is needed. High-GI foods will depress plasma FFA levels more than low-gi foods. Thus, high-gi foods result in lower blood fuel substrate levels, that is, lower blood glucose concentration and lower FFA after more than 90 min of exercise, compared with low-gi foods. Changes in RER were related to GI, being highest for the food with the highest GI (potato) and lowest for the food with the lowest GI (bran cereal). This effect was apparent (Figure 6) even before the start of exercise and suggests that RER is influenced by the prevailing glucose and insulin level. High RER has been associated with extended endurance in studies employing maltodextrin feedings (10). Our results might therefore be interpreted to mean that high-gi foods are more advantageous than low-gi foods. However, RER is an indication of the proportion of CH0:FFA in the fuel being oxidized and is a function of exercise intensity as well as substrate availability. Since CHO stores are limited, it is not necessarily desirable to be burning CHO when FFA would suffice. Hence, if the same intensity of exercise can be adequately maintained at a lower RER, then this will conserve CHO stores, thus favoring extended endurance. Postexercise In the recovery period, plasma glucose and insulin concentrations were higher with the low-gi foods than with the high-gi foods. This suggests that glucose was still being released from the gut to the plasma even during recovery, 2.5 to 3 hr after the consumption of the low-gi food. A slowly digested starch may take as much as 4 hr to be completely digested and absorbed (12). For this reason, low-gi foods may be an advantage in the outdoor, competitive, prolonged exercise situation especially if food is not available or is limited.
12 372 / Thomas, Brotherhood, and Brand Miller In summary, this study has shown that low-gi foods are associated with higher blood-borne muscle fuel substrate levels after more than 90 min of exercise. Release of glucose from the gut appears to continue into the postexercise period. Drawing on other studies which indicate that higher levels of plasma glucose are associated with delayed time to exhaustion, we interpret our findings to mean that low-gi foods may prolong endurance. Our findings have practical implications for the choice of foods in heavy occupational work as well as in sports, such as hiking, mountaineering, skiing, and cycle touring. References 1. Ahlborg, G., and P. Felig. Lactate and glucose exchange across the forearm, legs, and splanchnic bed during and after prolonged leg exercise. J. Clin. Invest. 69:45-54, Coggan, A.R., and E.F. Coyle. Carbohydrate ingestion during prolonged exercise: Effects on metabolism and performance. In Exercise and Sport Sciences Reviews (Vol. 20), J.O. Holloszy (Ed.), Baltimore: Williams & Wilkins, 1992, pp Costill, D.L., E. Coyle, G. Dalskey, W. Evans, W. Fink, and D. Hoopes. Effects of elevated plasma free fatty acids and insulin on muscle glycogen usage during exercise. J. Appl. Physiol. (Respirat. Environ. Exerc. Physiol.) 43: , Coyle, E.F., A.R. Coggan, M.K. Hemmert, and J.L. Ivy. Muscle glycogen utilization during prolonged strenuous exercise when fed carbohydrate. J. Appl. Physiol. 61: , English, R., and J. Lewis. Nutritional Values ofaustralian Foods, Canberra: Australian Government Publishing Service, Gollnick, P.D., B. Pernow, B. Essen, E. Jansson, and B. Saltin. Availability of glycogen and plasma free fatty acids for substrate utilization in leg muscle of man during exercise. Clin. Physiol. 1:27-42, Havel, R.J., B. Pernow, and N.L. Jones. Uptake and release of free fatty acids and other metabolites in the legs of exercising men. J. Appl. Physiol. 23:90-99, Hickson, R.C., M.J. Rennie, R.K. Conlee, W.W. Winder, and J.O. Holloszy. Effects of increasing plasma fatty acids on glycogen utilization and endurance. J. Appl. Physiol. (Respirat. Environ. Exerc. Physiol.) , Jenkins, D.J.A., T.M.S. Wolever, R.H. Taylor, et al. Glycemic index of foods: A physiological basis for carbohydrate exchange. Am. J. Clin. Nutr. 34: , Mitchell, J.B., D.L. Costill, J.A. Houmard, M.G. Flynn, W.J. Fink, and J.D. Beltz. Effects of carbohydrate ingestion on gastric emptying and exercise performance. Med. Sci. Sports Exerc. 20:llO-115, Newsholme, E.A., and A.R. Leech. Biochemistry for the Medical Sciences. Chichester, UK: Wiley, Thorne, M.J., L.U. Thompson, and D.J.A. Jenkins. Factors affecting starch digestibility and the glycemic response with special reference to legumes. Am. J. Clin. Nutr. 38: , Thomas, D.E., J.R. Brotherhood, and J.C. Brand. Carbohydrate feeding before exercise: Effect of glycemic index. Int. J. Sports Med. 12: , Thomas, D.E., K. Richardson, J.R. Brotherhood, and J.C. Brand. The pre-game carbohydrate meal: Is the glycaemic index relevant? In The Athleteaaximising Participation and Minimising Risk, M.E. Torode (Ed.), Sydney: Cumberland College of Health Sciences, 1988, pp
13 Plasma Glucose Levels / Torsdottir, I., M. Alpsten, D. Andersson, R.J. Brummer, and H. Andersson. Effect of different starchy foods in composite meals on gastric emptying rate and glucose metabolism: 1. Comparison between potatoes, rice and white beans. Hum. Nutr.: Clin. Nutr. 38C: , Wolever, T.M.S., D.J.A. Jenkins, A.L. Jenkins, and R.G. Josse. The glycemic index: Methodology and clinical implications. Am. J. Clin. Nutr. 54: , Acknowledgments We thank Kellogg (Australia) Pty. Ltd. and the Sydney University Nutrition Research Foundation for their support; the volunteers for participating; and Dr. P. Wursch, Nestec, Switzerland, for providing foods. Statistics were compiled in collaboration with Professor E. Seneta, Head of Statistics, University of Sydney.