The Growth of Brown Trout (Salmo Trutta Linn.): IV. The Effect of Food and Temperature On the Survival and Growth of Fry

The object of the experiments described in this paper was to investigate the problem of the formation of circuli and annuli on the scales by producing trout fry which grew at different rates because of differences in their environment. It is hoped that the results of the examination of these scales will be described later; the present paper is an account of the effect of the conditions used on the growth and survival of the fry.

Gray & Setna (1930) and Bhatia (1931a, b) observed that in rainbow trout narrow circuli were produced by a low level of feeding and wide circuli by unlimited food, temperature apparently having no effect. Van Someren (1950) has deduced that in rainbow and brown trout in tropical conditions in Kenya the zones of narrow rings correspond to spawning periods. Cutler (1918) kept plaice and flounders in tanks at different temperatures with much or little food, and showed that the zones on the scales could be correlated with temperature but not with the amount of food. In the present experiments, trout fry were grown at two different temperatures, on three different types of food and at four levels of feeding. The experimental conditions were comparable with those in earlier experiments on the growth of trout fry and 2-year-old trout under controlled environmental conditions (Brown 1946 a, b, c), but the details of the apparatus and techniques used were different.

Trout ova were obtained from the Midland Fishery, Nailsworth, Gloucestershire. They arrived in Cambridge on 23 January 1950, in two batches; one batch consisted of all the eggs of one female trout which had been fertilized with milt from one male fish, while the other batch consisted of a random sample of the mixed eyed ova used in the hatchery. The ova were kept in hatchery trays at 11° C. until they hatched, and the alevins remained in the same trays until they were ready to feed. Samples of alevins from each set were weighed individually with a torsion balance and measured to the nearest millimetre in order to determine the amount of variation within each set (see Fig. 2 and 3 and Table 1).

Twenty experimental tanks were used; they were aquarium tanks made of glass and angle iron and measured 24 ௗ12 × 12 in. They were arranged in a row on a tray made of corrugated iron which collected all the overflows and discharged into a sink. A wooden frame 4 ft. above the tanks supported six 2 ft. 20 W. Mazda strip lights (warm white) and was made light-proof above with cardboard. The tanks stood in an alcove so that no light could reach them from behind or at the sides, and a black curtain was hung in front of them. The lights were switched on and off by a Venner time-switch which was set so that the fish received 12 hr. of light and 12 hr. of darkness alternately.

The aquarium tanks were supplied with Cambridge tap water which flowed in from a constant-level tank and overflowed through constant-level siphons so that each tank contained 45 l. of water and water flowed through at the rate of 2 l./hr. Compressed air was bubbled into each aquarium to keep the water well stirred. The water in the tanks was maintained at a constant temperature below that of the tapwater supply by using a cooling system designed by Dr R. H. J. Brown of the Department of Zoology, Cambridge, whose advice and assistance were invaluable.

Cold water was circulated by an electric pump through a series of control-valve units, joined by copper pipes, and returned to a tank which was held a few degrees below the required aquarium temperature by a refrigerator. Each control valve unit was opposite an aquarium tank and had connected to it a spiral of three coils of flexible lead tubing which was suspended in the aquarium (see Fig. 1). The valve unit was a brass block with a straight-through hole and two side openings leading to the aquarium coils. Normally most of the cooling water passed through the valve block and only a little circulated through the coils but the throughway could be blocked by a sliding plug so that all the cold water was forced to pass through the cooling coils. This plug was moved to the closed position by the inflation of a rubber balloon in a brass chamber and was returned to the open position by a spring.

Each aquarium was fitted with a mercury-toluol thermostat which was connected through an electronic relay to a magnetic valve ; this in turn controlled the compressed air supply to the appropriate control valve in the cooling line, each tank being separately controlled. In the ‘off’ condition, with the thermostat circuit open, the grid of the 807 valve was held about 40 V. negative from the junction of the 40,000- and 70,000-ohm resistances. No current passed in the anode circuit, and the magnetic valve was unenergized so that the balloon remained deflated and open to the atmosphere. The hole through the control valve was open so that water did not circulate through-the cooling coils. When the temperature rose, the thermostat circuit closed and shorted the grid of the 807 valve to earth. About 50 mA. flowed in the anode circuit, energizing the magnetic valve which then connected the balloon to the compressed air supply. The control valve plug was pushed in to block the through hole and thus to cause the cold water to circulate through the cooling coils. When the temperature fell, the system returned to the ‘off’ condition and the balloon deflated through the magnetic valve. Additional parts of the equipment were: (1) indicator lamps which lit on the relay panel when the movement of the appropriate sliding plugs to the ‘on’ position was complete, (2) a switching system which allowed the thermostats to be cut out and the correct functioning of the relays to be tested with push buttons.

This system was sufficiently sensitive for the water in the aquarium to be maintained at a constant temperature which varied less than 0·1° C. When there were marked changes in the room temperature and in the temperature of the tap-water supply the rate of circulation of the cooled water and the temperature of the refrigerated tank were adjusted. In really cold weather, the tap water in the constant-level tank was heated by an electric immersion heater controlled by a thermostat so that its temperature did not fall below 13°C. The eight tanks nearest to the refrigerator were maintained at 10·5° C. and the other twelve at 12·5° C. ; with occasional breakdowns, these temperatures were maintained throughout the experiments.

The aquaria were stocked with 100 alevins in each on 23 February; 70 of each 100 were taken from the batch of ova derived from a single female and single male trout, and 30 were taken from the mixed batch of ova (see Fig. 3). The alevins began to feed on 24 February, and for 4 days they were all allowed to eat as much as they would of a mixed diet of liver, shrimp meal and live Tubifex. Food remains and faeces were siphoned out daily throughout the experiments.

The proper feeding schedule was started on 1 March and was arranged as follows : the eight tanks at 10·5° C. were divided into two sets, each of four aquaria—the fish in one set of four aquaria were fed entirely with minced fresh liver, while those in the other set were given live Tubifex worms; the twelve tanks at 12·5° C. were divided into three sets, each of four aquaria—the fish in one set of these were fed with fresh minced liver, those in another set were fed with live Tubifex, while those in the third set were given dried shrimp meal mixed to a paste with water. The shrimp meal was too finely dispersed to be suitable for food when the fish grew larger, so from 24 March onwards these fish were given a paste made from 2 parts of shrimp meal to 1 part of Bemax to 1 part of Farex. Even with this mixture the fish did not flourish and their diet was varied on 28 March, 5 and 19 April by giving them live Daphnia magna.

The amount of food given to the fish was not measured but when they were fed they were always given slightly more than they would eat. In each of the five sets of four aquaria, the fish in one aquarium were fed twice daily (once only on Sundays) ; in the second they were fed once daily; in the third they were fed three times a week; and in the fourth, once a week only. Thus the ratio of the amount of food available to the various groups of fish was as 13:7:3 : 1. By the end of a month it was obvious that the fish which were fed only once a week were receiving less than a maintenance ration, since some had died of starvation. The schedule was therefore altered on 28 March so that these fish were fed twice weekly and those which had been fed three times weekly were thenceforward fed four times weekly. The ratio of food available to the various groups thus became as 13:7 :4 :2.

The following combinations of environmental conditions were thus available, with 100 trout fry living in each aquarium:

Material for studying the development and growth of the scales was obtained by catching three fish at random from each aquarium each week. These fish were measured, killed and preserved. During the 11 weeks of the experiments, twentyseven fish were thus removed from each aquarium. The number of fish which died in each tank was also recorded.

At the end of the fifth week, on 2−4 April, all the fish in each aquarium were caught and anaesthetized with urethane. Their lengths were recorded on a piece of waxed paper attached to a measuring board by pushing their snouts against the perpendicular end of the board and making pricks in the paper at the forks of their tails. The fish were then returned to their tanks and their lengths were read to the nearest millimetre by placing the waxed paper over graph paper illuminated from below.

The experiments were ended on 19 May 1950. All the surviving fish were anaesthetized in urethane, their lengths were recorded on waxed paper and they were preserved in formalin for subsequent examination of their scales.

At the beginning of the experiments, the fry would take all three types of food quite readily. They fed only on particles moving or suspended in the water and ignored shrimp meal and bits of liver which sank on to the bottom of the aquaria. The Tubifex worms which fell to the bottom wriggled about but associated themselves into clumps within 2 hr. ; these clumps were ignored by the trout and were siphoned out on the following day and used for food again.

It was apparent by the end of 3 weeks that shrimp meal alone was not a suitable diet, so it was reinforced with Bemax and Farex. The fish took this paste readily but remained in such poor condition that on three occasions they were given live Daphnia which they took eagerly. Towards the end of the experiment the mortality in the groups fed with shrimp meal was very high (see Table 3 and Fig. 4), demonstrating a definite deficiency in the diet.

An interesting qualitative effect of the shrimp meal diet was that the trout whicm were fed with it developed orange pigment in the caudal fin and an orange border to the adipose fin. These colours appeared after 3 weeks and were retained till the end of the experiments, whereas the trout fed with liver and Tubifex never developed any orange pigment.

The fish which were offered Tubifex worms fed eagerly and seemed to be doing well until after the third week, when large fish in these tanks began to die (see Fig. 4A). It was noticed that these fish showed dilation of blood vessels in the skin and bleeding from the tail and fins, but there was no evidence of pathogenic organisms. This phenomenon of the death of well-grown fish continued in these groups of fry fed with Tubifex until the end of the experiments, and when the survivors were killed and measured, they all showed this characteristic dilation of cutaneous blood vessels and bleeding from the fins. Either there is some deficiency in a diet consisting of Tubifex alone or there is some toxic substance present, and this is accumulated by the trout fry. This deficiency or toxicity becomes lethal as the fish grow larger; this is illustrated by Table 2, in which the mortality rates of fish fed once a day are compared with those of fish fed three or four times a week. As will be described later, the former group grow faster than the latter and their mortality peak occurs 2 weeks earlier than that of the more slowly growing fry.

Fresh liver was clearly the most satisfactory of the foods used from the point of view of survival of the fry. This is illustrated by Table 3, in which the total numbers of fish dying in the groups fed once and twice daily are given for the 2nd-6th and 7th−11th weeks inclusive. The figures for the first week have not been included, since deaths during that week are less likely to be the result of diet than of injuries sustained in the transfer from the hatchery troughs to the aquaria.

There is no point in comparing the sizes of the survivors in all the tanks at the end of the experiments because of the high mortality and consequent few survivors of the fish fed with Tubifex and with shrimp meal. The range of size in the various tanks at the end of 5 weeks, however, provides an interesting comparison, as can be seen from Fig. 5, in which the total range of length, the medians and the upper and lower quartiles are given for fish fed with liver, Tubifex and shrimp meal at 12·5° C.

The groups are arranged according to the frequency of feeding per week. The fish fed less frequently (1 or 2 and 3 or 4 times per week) show no significant differences in size which can be related to diet, but those fed once and twice daily with liver attained a significantly greater size than those fed with Tubifex, and the latter were markedly larger than those fed with shrimp meal.

Thus the type of food available for trout fry affects their survival; when plenty of food is available, the rate of growth is determined by the type of food. Of the three diets tested, liver gave markedly better results than Tubifex or shrimp meal.

The numbers of fish which died each week in each tank are shown in Fig. 4. In each of these graphs groups are compared which were fed equally frequently but with different foods and at different temperatures. The groups of fry which were fed twice daily did not differ markedly from those which were fed with the same food once daily, except that there was a higher mortality with shrimp meal as food twice a day than once daily. It was noticeable that the water containing the former group was cloudy with suspended shrimp meal, and this may have caused the higher mortality. The water in the tanks containing fish fed once and twice daily with liver was cloudy with dissolved blood, but this had no adverse effect on the fish. The water containing the fish fed with Tubifex was always very clear.

The figures for fish fed three or four times per week do not differ very much from those fed daily except, as already described, for the delay in the onset of mortality of large fish on the Tubifex diet. The fish fed only once a week apparently did not receive sufficient food, and there was high mortality of small fish during the 4th and 5th weeks. The frequency of feeding was doubled at the end of the 4th week and after that the- mortality was low except with shrimp meal—in this tank all the fish except one died before the end of the experiment.

The behaviour of the fish in groups fed at different intervals of time showed interesting differences. Those fed twice daily showed some interest but no great enthusiasm for their food, while those fed once daily were much more excited and appeared to eat more at each feed. The fish fed four times per week or less frequently showed great activity and enthusiasm whenever they were fed, and they ingested so much food that they bulged markedly after each meal and this bulge lasted for more than 24 hr. The fish fed every day never ate sufficient food to distort their streamlined shapes.

Fig. 6 and 7 are histograms showing the frequency distributions of length among the fish fed with liver at 10·5 and 12·5° C. After 5 weeks, and even more markedly after 11 weeks, there was a direct correlation between the frequency of feeding and the size distribution within the groups, the fish fed most frequently having grown largest. At the end of 11 weeks there was no overlap in the size distributions of fish fed once or twice weekly with those fed twice daily and once daily; those fed three or four times weekly had an intermediate size distribution (see also Fig. 9).

Fig. 8 shows that with Tubifex as food there is a similar effect of frequency of feeding on growth, those fed more frequently growing larger than those fed less frequently. There is no real difference between the size distributions of fish fed once and twice daily with Tubifex, and the differences between the groups are less marked than those between the groups fed with Ever; this is the result of the mortality of large fish which began after 3 weeks in the tanks fed once and twice daily.

The fish fed with shrimp meal showed better growth when fed once daily than when fed twice a day (Fig. 5), and there was no significant difference between the sizes of fish fed once daily and three or four times per week. This was probably because the suspension of shrimp meal in the water adversely affected the growth of those fed daily.

Thus the maintenance requirements of trout fry are not satisfied if they are allowed to eat as much as they will only once per week, but when fed twice per week they are able to grow slowly. The amount of growth depends on the frequency with which they are fed, and this was shown most clearly by the fish living on a diet of fresh liver.

Two different temperatures only were used in these experiments, and fry were fed with liver and with Tubifex at both these temperatures. There was no apparent difference in behaviour between corresponding groups of fish at the two temperatures, but the mortality curves (see Fig. 4) showed a lag of about 2 weeks between the maxima for deaths per week at 10·5° C. as compared with 12·5° C. This is also illustrated by the number of deaths of large fish in the tanks fed daily and three or four times weekly with Tubifex (see Table 2), and it is shown in the mortality curves for fish fed once or twice weekly where there is also a higher mortality in the tanks at the higher temperature (Fig. 4C).

After 5 weeks’ growth there were already differences between the size distributions of fish fed once and twice daily at the two temperatures, those feeding at 12·5° C. being larger than those at 10 5° C. (see Fig. 8). The fish fed with fiver showed these differences even more markedly after 11 weeks of growth, those at 12·5° C. being significantly larger than the corresponding fish at 10·5° C. (see Fig. 9). Fish fed less often than once daily showed no significant differences in length distribution after 11 weeks at the two temperatures. Thus with a restricted amount of food there seems to be no difference in the amount of growth at 10·5 and 12·5° C., but with a higher level of feeding the fish at the higher temperature are able to grow faster than those at the lower one. Since the food intake was not measured, this faster growth may have been the result of better appetite or of higher efficiency of conversion of food into fish, or of both.

In earlier experiments with trout fry at 11·5° C. (Brown, 1946a), liver was the only food used, and the fish were fed twice daily for the first 12 weeks and once daily thereafter. From the point of view of food, therefore, they can be compared with the fish which were fed most frequently with liver in the present experiments. In other experiments with 2-year-old trout (Brown, 1946b, c) the effect of level of feeding on growth was investigated at 11·5°C., and appetite and maintenance requirements were measured at five other constant temperatures. In all these experiments, the growth of individual fish was very much affected by their positions in the size hierarchies. The effect of size hierarchies was also apparent in the present experiments, since the disparity in size between the smallest and largest fish increased as the fish grew larger.

McCay, Dilley & Crowell (1929), using Salvelinus fontinalis, investigated diets of purified protein, carbohydrates and fat and found that with less than 10% protein the fish did not grow and died within 6 months ; with a higher level of protein, the fish grew but died within 3 months. They concluded that there was a store of some essential substance in the trout body which was destroyed during growth and could not be replaced from the purified diets. This same explanation could account for the mortality of large fry found in the present experiments with Salmo trutta using a diet of Tubifex. McCay et al. do not refer to any characteristic symptoms being associated with the death of their fish as was bleeding from the fins and dilation of cutaneous blood vessels with trout fry dying on the Tubifex diet. The deaths on this diet could be explained as the result of accumulation of toxins from the food instead of as the result of depletions of essential substances.

McCay et al. found that brook trout grew and survived normally when raw liver was added to the purified diets. They also discovered that diets of dried buttermilk, dried skim milk, gluten and pea nut meal could only be utilized for growth when raw liver was added. Coleman (1930) compared raw liver, salmon-egg meal, shrimp meal, sardine meal, lactein, kelp and alfalfa as food for trout fry, and noted that Ever was the most satisfactory and that all the others required to be supplemented with liver if good growth was desired. McCay & Tunison (1937) also found that raw meats were more efficient foods than dried feeds, and Tunison (1940) observed that the higher the amount of protein in the food, the better the growth and the lower the mortality. Hewitt (1943) also remarked that the dried foods used in trout hatcheries lacked some essential dietary constituent which is present in liver. There is thus plenty of evidence that liver contains some substance essential for trout growth and absent from such foods as dried shrimp meal and dried milk. The high mortality and poor growth found in the present experiments with shrimp meal, even when supplemented with Bemax and Farex, and the excellent growth and low mortality of the fry fed with liver are therefore in full agreement with the results obtained by other workers.

Myers (1946) stressed the importance of the first year’s growth for the subsequent production of large trout, and stated that natural foods are better than unnatural ones and that zooplankton and dipterous larvae are the best food of all. The present experiments showed that not all species of aquatic animals are adequate trout foods even when alive, since the diet of Tubifex led to high mortality and was not nearly as satisfactory as the ‘unnatural’ food liver. The fish fed with shrimp meal were given live Daphnia on three occasions, and their liveliness was much increased by this addition of ‘natural’ food. Some twenty alevins, not required for experiments, were kept in a separate aquarium and were fed almost entirely with planktonic Crustacea. These fry remained healthy and grew faster than any of the experimental fry fed with shrimp meal. Had it been possible to obtain a sufficiently large and regular supply, planktonic Crustacea would have been used instead of shrimp meal in the present experiments.

Wild stocks of trout show considerable differences in coloration according to their locality, and it has been suggested that these differences are associated with differences in feeding habits. Orange fins and red spots appear to be characteristic of trout which feed largely on Crustacea, and where these organisms are not normal constituents of the diet, red and orange pigments are not conspicuous. Steven (1948) identified and estimated the amounts of various carotenoids in different parts of the body of brown trout. He showed that in aquaria fish fed with horseflesh and earthworms lost 90% of their carotenoids in 9 months, whereas those fed with natural foods, mainly Entomostraca, retained their wild coloration. McCay (1937) noted that additions of salmon ova, goldfish, Daphnia or carrots to their diet enabled trout to retain their natural coloration, and Steven (1948) found that fish fed on salmon ova were more pink than normal wild fish. In the present experiments, orange coloration was developed only in those fry fed generally on shrimp meal and three times with Daphnia, while the extra alevins fed mainly with planktonic Crustacea, also developed much orange pigmentation. Thus these experiments give further support to the conclusion that the pigments of the lipophores of trout are derived solely from their food. Steven (1948) deduced that trout are unable to convert one type of carotenoid into another, and the trout fed with fresh Ever in the present experiments were not able to produce astacene and lutein in visible quantities in their skins. Thus, the type of food eaten can produce differences in the coloration of the fish as well as in their survival and growth rate.

In earlier experiments (Brown, 1946 a, b, c) it was found that trout allowed to eat as much as they would had definite appetites and would not eat more than a certain definite quantity of food on each occasion of feeding. Moore (1941) observed the same phenomenon with individuals of Apomotis, Helioperca and Perea flavescens. In the present experiments, the food was not weighed, but the fish were given more than they would eat on every occasion when they were fed. The frequency of feeding ratio was as 13:7:3 (later 4):! (later 2). As the fish grew larger they consumed more food, though it probably represented less in proportion to their body weights. The differences in behaviour of the different sets of fish was such that those fed less frequently probably consumed relatively more than the others on the occasions when they were fed because they stuffed themselves to such an extent that they bulged and all rested on the bottom of their tanks for 24 hr. afterwards. Those fed twice daily showed less enthusiasm and probably consumed relatively less at each feed than those fed once daily. Thus the real ratio of amount of food consumed per week per unit body weight was probably more like 4:3:2:1 than 13:7:3 (later 4) : 1 (later 2).

The food consumed by a fish must be used for maintenance of its tissues as well as for growth. The fish which were fed only once per week did not receive sufficient for maintenance, since they were becoming thin and dying 4 weeks after beginning to feed. When their frequency of feeding was doubled this led to decreased mortality and improved condition. Thus, trout fry allowed to eat as much liver, Tubifex or shrimp meal as they will twice per week obtain more than their maintenance requirements. The figures for the fish fed with liver show a clear correlation between frequency of feeding and growth rate, those fed twice daily growing faster than those fed once a day, the latter better than those fed three or four times weekly and those fed once or twice weekly growing least rapidly. Two-year-old fish grow more efficiently if the amount of food given to them is restricted (Brown, 19466), and they may reach almost the same size as comparable fish allowed to eat as much as they would. This phenomenon was not displayed by the trout fry. Myers (1946) states that trout fry do not hunt for their food for some time after they begin to feed, so that they must rely on food coming within their range of vision ; they will swim after it and take it. It is not known how much they could consume in one day, and it is possible that they might obtain more than is possible with two feeds daily of minced liver. Thus, even the most frequently fed fish in the present experiments may have been underfed compared with wild trout fry in an environment rich in food.

Mottram (1936) reported experiments with trout fry which gave very different results from those described here. He used plankton as food and divided his fish into four batches, each consisting of twenty-five brown trout fry. One batch was fed daily, another twice a week, the third batch once a fortnight, while the fourth batch was not fed at all. Mottram states that the fry which were fed twice a week grew as well as those which were fed daily, while those which were fed only once a fortnight remained active and healthy but did not grow as fast as the others. The difference between these results and those described in this paper could be explained if Mottram’s fish were given so much live plankton at each feed that those which were fed twice a week had sufficient for them to eat the same amount every day as those which were fed daily, the latter being provided with considerable excess of food. Those which were fed once a fortnight would have received enough food for them to a eat well for at least several days. Eight of the twenty-five starved fry died during the first four weeks. The temperature of the water is not recorded.

The marked differences in growth rate between trout fry under natural conditions in different localities have been ascribed to many causes, including differences in the quantity and quality of the food. The present experiments showed that differences in type of food can have marked effects on the survival, growth and coloration of trout fry and also that the frequency of feeding is an important factor, since it was only among the groups fed once and twice daily that significant differences in size could be correlated with differences in diet. This implies that it is.only when food supplies are abundant that differences in quality will result in differences in the rate of growth of trout fry. With restricted quantities, the quality of the food should have no effect on the growth rate. With liver, the most successful food used in these experiments, the rate of growth was proportional to the number of feeds per week, while with shrimp meal there was no difference between the growth rates of fish fed three or four times weekly and those fed daily.

Two temperatures only (10·5 and 12·5° C.) were used in the present experiments, but the fry which were fed with liver and those which were fed with Tubifex showed that this difference of 2° C. was sufficient for there to be differences in survival and growth of comparable fish. When fed only once per week, fish at 10·5°C. survived better than those at 12·5° C., indicating that the maintenance requirement must be less at the lower temperature. Fish which were fed once or twice daily grew faster at the higher temperature, but there were no significant differences between those with more restricted rations at the two temperatures.

For 2-year-old trout at constant temperatures, specific growth rates were high between 7 and 9° C. and between 16 and 19° C., and low above, between and below these temperature ranges. Thus the two temperatures used in the present experiments fall within a minimum growth-rate range for older trout. No data are available for trout fry growing under comparable conditions below 10·5 or above 12·5° C., so it is not possible to demonstrate whether they have the same growth rate-temperature relationship as the 2-year-old trout.

Groups of 100 trout fry were grown in identical aquarium tanks at constant temperatures, with 12 hr. of illumination per day, constant rate of water flow, aeration and composition of the water. Two different temperatures, three different types of food and four levels of feeding were investigated. Individual lengths were recorded for the first 11 weeks after the beginning of feeding.

The trout fry took live Tubifex worms eagerly and grew well up to a certain size, when they began to develop bleeding from the fins and dilation of cutaneous blood vessels and to die. It is suggested that Tubifex either lacks some chemical substance essential for trout survival or contains some substance which is accumulative poison.

The trout fry which were fed with shrimp meal lost condition and showed poor survival and growth, suggesting that their diet lacked some necessary constituent. These fish developed orange pigment in their caudal and adipose fins; such orange pigment was not developed by fry fed with Tubifex or with liver.

The trout fry which were fed with liver had the lowest mortality and showed the best growth, supporting the view that liver contains all the substances necessary for trout growth.

Fry fed twice daily with liver showed less enthusiasm for their food than those fed once daily but the former grew slightly faster.

Fry fed three or four times per week with liver gorged themselves whenever they were fed but grew at lower rates than those fed daily.

Trout fry allowed to eat as much liver as they would once per week gorged themselves, but obtained less than their maintenance requirements and began to die of starvation after 4 weeks at 12.5° C. They were able to grow slowly when fed twice per week.

Fry fed once or twice daily with liver grew larger than those thus fed with Tubifex, and the latter grew larger than fry fed daily with shrimpmeal. There were no significant differences in size which could be correlated with diet among fish fed less often than once daily.

Trout fry fed daily at 10.5° C. grew more slowly than those fed daily at 12.5° C., but those fed less frequently showed no differences in growth rate which could be associated with temperature. Those fed with Tubifex at 10.5° C. showed a delay of 2 weeks in the onset of high mortality compared with those at 12.5° C.

I wish to thank Dr R. H. J. Brown for his advice and assistance in the construction and maintenance of the temperature-control apparatus and Miss C. M. Sullivan for her kindness in looking after and feeding the fish on occasions when I was away from Cambridge.

Since this paper was written the aquarium temperature control system has been redesigned by Dr R. H. J. Brown to work over a wider temperature range. Individual tanks can be controlled at any temperature between 4 and 35° C. The rubber balloons which were liable to develop leaks are replaced by reinforced rubber diaphragms.