Survival of Goldfish in Presence of Carbon Monoxide^*

The introduction to a previous paper outlined the manner in which an increasing understanding of mammalian respiration stimulated an interest in the respiratory adaptations of fish. Of the various attempts to relate the biochemical properties of fish blood to behaviour of the intact animal, asphyxiation experiments deserve particular attention. Bohr, Hasselbalch & Krogh (1904) first called attention to the depression in the oxygen affinity of mammalian blood that is caused by carbon dioxide. The sensitivity of fish blood to carbon dioxide varies with the species, but in general it is much more sensitive than mammalian blood. Likewise the sensitivity of the intact fish to carbon dioxide has been shown to vary with the species, and more recently it has been suggested that this sensitivity may depend upon the temperature to which the fish is acclimatized (Fry, Black & Black, 1947). It is this sensitivity of the intact fish to carbon dioxide that is measured by the asphyxiation experiments mentioned above. Initially at least, the experiments were attempts to observe Bohr effect in vivo. Fish sealed in jars of water containing oxygen at atmospheric level and varying amounts of carbon dioxide are left until a cessation of respiratory movement indicates that death has been brought about by asphyxiation. Immediately after death of the fish, the residual amounts of oxygen and carbon dioxide are measured. From a graph of the results, a picture is obtained of the ability of the fish to remove dissolved oxygen from water when various concentrations of carbon dioxide are present. This ability has been called respiratory tolerance (Irving, Black & Safford, 1939), but Anthony (1960) pointed out that carbon dioxide tolerance is a better term, since it does not imply any inference as to the mechanism by which death is brought about. The latter term is used in this paper.

The inverse relationship that has been revealed between Bohr effect and carbon dioxide tolerance (cf. table 1 in Anthony, 1960) has generally been interpreted as evidence that carbon dioxide exerts its asphyxial effect upon the intact fish via the haemoglobin within its blood. Obviously such an interpretation assumes a certain dependence of fish upon haemoglobin. In particular it suggests that asphyxial oxygen tensions are related to haemoglobin content of fish blood. Recent dis coveries in Antarctica drew attention to facts already in the literature that make it doubtful that fish as a class make much use of this respiratory pigment.

Nicloux (1923) reported the survival of three common teleosts, without apparent detriment, for 4 hr. in water equilibrated with air containing 2 % carbon monoxide. In 1926, Schlicher described a yearling carp with no erythrocytes in its blood. Matthews (1931) mentioned the existence of ‘bloodless’ species of fish in the Antarctic Ocean, and Ruud (1954) confirmed their existence with a description of three species, one of which may attain a kilogramme in weight. Commenting on Ruud’s report and drawing attention to the supporting literature, Fox (1954) concluded that haemoglobin is an emergency precaution for fish and that oxygen carried in solution in their plasma serves adequately for normal purposes. Anthony (1956) suggested that haemoglobin acts in the role of a supercharger in fish respiration, but such an analogy leads to incorrect implications and must be discarded.

The suggestion that haemoglobin exists in goldfish blood merely for emergencies raises a number of questions with respect to previous work. Is haemoglobin essential to the normal activities of goldfish over any of their thermal range? How much of their activity depends upon presence of haemoglobin in their blood? Does the ability of goldfish to remove oxygen from water in the asphyxiation experiments depend upon this respiratory pigment—i.e. is asphyxial oxygen tension related to haemoglobin content of the blood? Ruud’s report and Fox’s comment upon it were published while work reported in the preceding paper was in progress. It was immediately apparent that Nicloux’s work pointed to a possible means of answering some of the foregoing questions. Consequently, a series of experiments was set up in which goldfish were subjected to carbon monoxide. The results are presented herewith.

Goldfish averaging 9−10 cm. in length were used in the experiments. The reasons for choosing these animals, the manner in which they were cared for, and the procedure by which they were thermally-acclimatized has been described in a preceding paper (Anthony, 1960).

The value of M, the CoHb/O₂Hb equilibrium constant for goldfish blood was very kindly determined for me by Prof. F. J. W. Roughton, F.R.S.

Blood for this purpose was collected from the exposed heart by means of a paraffin-lined hypodermic needle attached to a small plastic vessel. Clotting was prevented by dusting the interior of both needle and vessel with dry heparin. The Hb solution was prepared by laking 1 vol. of whole blood in 10 vol. of distilled water and removing cellular debris by centrifugation. The solution was diluted with teleost Ringer (Young, 1933) to a colour density suitable for observation by reversion spectroscopy when viewed through the small tubes of the spectro-tonometers. The final dilution was about 1:150.

10 ml. portions of the Hb solutions were taken up in each of two tonometers and measurements of the fully oxygenated spectra were recorded. The solutions were then equilibrated with measured quantities of O₂ and CO and the measurements repeated. Finally measurement of the fully carboxylated spectra were recorded. From these measurements, the span—i.e. distance in Angstroms between a-bands of COHb and O₂Hb—and the partition constant, M, were calculated.

The paper by Bailey (1954) proved useful in setting up the gas flow systems, shown diagrammatically in Fig. 1. Driers were not used, but use was made of both the water columns (A) as compensators and the fine control valves (B) described by Bailey for regulating gas flow. The plaster of Paris chokes were found to be unnecessary. The compensator used with CO had to be modified slightly so that excess CO would be conducted out of the laboratory. In addition to the fine control valves, single-way glass taps (C) were used in each gas line. A three-way glass tap (D), inserted beyond the point where the gases mixed, permitted sampling the gas mixture with only momentary interruption of its flow. O₂, CO and N₂ were obtained from cylinders of compressed gas, and use was also made of the laboratory compressed air supply. The desired mixture of gases was initially set by means of CFM (E); the gas mixture was then checked by analysis using a Scholander-Roughton syringe (Roughton & Scholander, 1943), and finally the concentrations of gas in the water containing the fish were measured by Van Slyke analysis.

Each fish was confined in about 400 ml. of water in jars of 500 ml. capacity. The jars (Fig. 1, F) were broad and flat and fitted with plastic screw caps. Glass tubes through which the gas mixtures passed into and out of the jars were inserted through rubber stoppers fitted into holes in the plastic cap. A third hole in the cap permitted sampling of water in the jar without interrupting the gas flow and was closed with a rubber stopper.

The tip of the glass inlet tube was drawn to a moderately small orifice and arranged horizontally near the bottom of the jar. The gas mixture issued as a small jet and exerted a slight stirring effect upon the water in the jar. The jars were filled with water from, and were suspended in, the aquarium containing the experimental fish. A broad rubber band around the outside junction of cap and jar produced a gas tight fit. The flow of air to control jars was adjusted by visual comparison to be approximately the same as for experimental jars.

Straight, uncalibrated pipettes which would contain about 20 ml. were used in sampling the water. They were fashioned by adding the usual pipette tip to glass tubing about 1 cm. in diameter. The unconstricted upper end was fitted by means of a rubber stopper and a short glass tube to a length of rubber tubing and a glass mouthpiece. The tip of the pipette was fitted with a small piece of rubber tubing to enable it to be used in the Van Slyke cup in the usual manner. A small section of glass tubing containing some glass wool was fitted over the tip when sampling. It prevented collection of excreta and other particulate material with the sample and also prevented contact of the tip with air during transfer of the sample from the aquarium to the Van Slyke analytical apparatus.

By holding these pipettes vertical and drawing in the water steadily, they could be filled readily without entrapping bubbles. Suction was applied either by mouth or by means of a ‘Propipette’. A surplus of water was drawn up the rubber tubing to the mouthpiece. A clip on the tubing prevented loss of the sample.

The flow experiments were sampled through a special hole in the cap of the jar. Aeration of the contents of the jar was prevented by temporarily closing the exit so that gas was forced to pass out through the sampling hole. Where a series of samples was taken during the course of an experiment, the water level was maintained by addition of 20 ml. after each collection.

Where duplicate samples were taken, as in the asphyxiation experiments, one had to be stored 15−20 min. while the first was analysed. Otherwise, the samples were transferred within 2 min. to the chamber of the Van Slyke apparatus. Temporary storage was effected by submerging the pipette containing the sample, and with the filter over the tip, in the appropriate aquarium so that no change in temperature occurred from sampling to analysis.

Analyses of water for dissolved gas content were carried out upon 10 ml. samples in the Van Slyke manometric gas analysis apparatus (Van Slyke & Neill, 1924; Peters & Van Slyke, 1932). A number of investigators have called attention to the desirability of manometric methods of analysing water for gas content where studies on aquatic respiration are concerned (Hall, 1923; Bosworth, O’Brien & Amberson, 1935; Zeuthen, 1947).

The Van Slyke chamber was de-aerated in the usual manner. The filter was removed from the tip of the sample pipette and the latter was then inserted beneath mercury in the cup and pressed lightly against the bottom. The clip was then removed from the rubber tubing so that the sample was free to flow. By raising momentarily the tip of the pipette, a few ml. of water were allowed to escape into the cup from the tip which had been briefly in contact with air. 10 ml. of the sample were then admitted to the chamber by lowering the mercury meniscus to the top of the numeral 10 above the 10 ml. graduation line. When the chamber was sealed it was found to contain exactly 10 ml., the difference having been made up from the quantity of sample contained between the top of the chamber and the mercury in the cup. The excess water in the sampling pipette served to protect from aeration the lower portion that was admitted to the chamber. I am indebted to Prof. Roughton for suggesting this simple method of transferring a measured aliquot of the water sample to the Van Slyke chamber.

The mercury was then lowered to the 50 ml. level and the gases were extracted by shaking for 5 min., a time that was found to be adequate for extraction to constant volume. CO₂ was absorbed by 1 ml. of 4% NaOH and O₂ by 1 ml. of alkaline hydrosulphite solution containing sodium anthraquinone-2-sulphonate and prepared according to directions in Peters & Van Slyke (1932). Both of these solutions were thoroughly evacuated and stored under vacuum until immediately prior to use (Roughton & Root, 1945). The water was allowed to rise up and wash down the upper part of the chamber after each absorption and was then extracted at the 50 ml. level for about 30 sec. prior to making the subsequent pressure reading. Pressures were read with the water menisci at the 2 ml. level.

Following absorption of O_2, a bubble of the gas in the Van Slyke chamber was transferred to the syringe and its CO content was determined by absorption with Winkler’s reagent (Roughton & Root, 1945). The remaining gas was expelled from the chamber and the pressure at the 2 ml. level was measured with only the liquid present. Factors for converting these pressure measurements to ml. of gas per litre of water were computed by the formula given, in Peters & Van Slyke (1932).

The method was tested by analysing fully aerated distilled water and the results, as shown in Table 2, indicate that the method was quite precise and reasonably accurate.

The problem of expressing most meaningfully the gas content of water has been discussed by Fry in Brown (1957). He chose to express O₂ in mg./l. and CO₂ in partial pressure (mm. Hg). He was not concerned with expressing the concentration of other gases. Ricker (1934) pointed out that tables on the solubility of gases in water usually present those quantities dissolved from a ‘dry’ gas in contact with water at a pressure of 760 mm. Hg. Obviously this situation never obtains, and the tables are quite unpractical. A more practical table has been drawn up and presented as Table 3.

The results of analyses expressed in ml./l. were readily converted to a percentage of absolute saturation at a given temperature and hence the partial presure of the gas concerned was easily derived. The latter values for O₂ and CO were used in computing the percentage of COHb from the known value of M.

The preliminary experiments were essentially a repetition of Nicloux’s (1923) experiments, except that no attempt was made to analyse the blood of goldfish. Nicloux held fish for 4 hr. in a flask of water through which air containing 2 % CO was bubbled. The fish showed no signs of distress. He then removed the fish, obtained a sample of blood from a branchial artery, and analysed the blood for COHb. He found concentrations of COHb ranging from 86 to 92 %, but thought that some re-equilibration with air may have taken place in the 15 min. or so required for sampling.

In these preliminary experiments, air containing 2 %, and later 4 % CO was used. The gas mixture was not passed through the jars until after admission of the fish. About i hr. was required for equilibration of gas and water at the prevailing rate of gas flow and that period is included in the times recorded in Table 4 for the preliminary experiments. The temperatures were 6, 15, 25 and 30°C. and in all of the experiments, the fish were alive after exposure of 6-8 hr. Sometimes the treated fish appeared less active than the control, but after both had been liberated one could not be distinguished from the other in the aquarium. Thus it was concluded that goldfish over their entire thermal range can survive apparently unharmed for at least 8 hr. in aerated water containing 4 % CO. No attempt was made to measure COHb concentration. It was expected to be near saturation. Subsequent determination of M permitted calculations of COHb concentrations shown in brackets in Table 4. They are lower than anticipated, and indicate that Nicloux’s results probably were more accurate than he thought.

In the next experiments goldfish were subjected to a mixture containing 80 % CO and 20% O₂. This was the maximum concentration of CO that could be used without decreasing the tension of O₂ normally supplied by aeration. To increase the significance of measured survival times, the fish were introduced into solutions already in equilibrium with the gas mixture. The results are shown in Table 5. When the experiment at 5° C. was terminated, no difference in activity could be noted between control and experimental animal. At 15° C. the experimental animal, while quite lively, appeared to fatigue more readily upon handling. At 30° C. the experimental animal was alive and showed no ill effects after 24 hr., although it appeared more quiescent than the control. At 29 hr. it was dead.

The rapidity with which death from anoxia might be expected was tested by placing the fish in a jar of water equilibrated with N₂ that had been washed free of O₂ with alkaline hydrosulphite and rinsed with de-aerated water. Undoubtedly introduction of the fish resulted in some aeration of the water. The fish was distressed in less than 30 min., and dead in 2·25 hr. In a similar experiment the fish died in 1·5 hr. It is possible that traces of O₂ remained in the N₂ and prolonged the life of the fish; nevertheless, these tests indicate clearly that survival in presence of CO is not due to resistance to complete anoxia.

The foregoing experiments indicate that goldfish supplied with oxygen at atmospheric tension can survive for some time in concentrations of CO calculated to convert practically all of their Hb to COHb. Presumably the plasma carries in solution enough oxygen to supply the tissues. In view of previous interpretations of asphyxiation experiments, it was now of interest to see how effectively a fish can reduce the oxygen content of water in presence of CO. With this in mind, asphyxiation experiments were set up with acclimatized goldfish, using techniques similar to those of Fry and his associates.

Flasks of water suspended in the experimental aquaria were equilibrated with a gas mixture containing O₂ at not less than atmospheric tension, CO estimated to block goldfish Hb, and CO₂ at as low a level as was practical. The jars containing the fish in the prepared solutions were submerged in the aquaria. The record of initial and final gas concentrations are results of duplicate analyses of water into which the fish was introduced and in which it later was found dead. Lack of gross respiratory movement was taken as a criterion of death. The diagnosis proved incorrect on two occasions, where fish subsequently recovered, but both had succeeded in reducing the O₂ content of the water to a low level. The results of these asphyxiation experiments are summarized in Table 6 and compared with similar data from experiments of Fry et al. (1947), in which CO was absent.

The asphyxial O₂ tensions in presence of CO tend to be slightly higher than those of Fry and his associates using low CO₂ tensions. This may be a true indication of the effect of CO, or a combined effect of CO and the CO₂ produced during the experiment. It may also be partially attributable to differences in the size of vessels, as suggested by Shepard (1955). In any event, the difference in asphyxial O₂ tensions in the two sets of experiments is not large, and the more general conclusion to be drawn from these results is that CO does not greatly impair the ability of goldfish to remove oxygen from water in asphyxiation experiments. This is evidently so at either 5° or 30°C. and in presence of concentrations of CO calculated to convert at least 99 % of the Hb to COHb.

Keilin & Wang (1946) showed conclusively that the linear relationship between the span (y) and logM (K in their paper) does not hold good for all haemoglobins. Thus they brought to a close speculations initiated by Anson, Barcroft, Mirsky & Oinuma (1924) and continued by Barcroft (1928). The revival of this speculation upon the relationship between affinities and absorption maxima by Haurowitz & Hardin in Neurath & Bailey (1954) suggests that they may have overlooked the significance of the experiments by Keilin & Wang. It may be noted that the measurements made upon goldfish blood in the course of the present experiments also fail to support the suggested relationship and appear to be exceptional for vertebrates in this respect. Consequently it may be of interest to compare M, the COHb/O₂Hb partition coefficient, and y, the distance in Angstrom units between the α-bands of COHb and O₂Hb, for goldfish blood with values compiled by Keilin & Wang. This is done in Table 7. I have expanded the data of Keilin & Wang somewhat, but the source of the information is available in their paper. The present measurements were made for me by Prof. Roughton. Remarkably good agreement was shown by the two measurements from which the recorded value was derived, but it should be recalled that they were made upon one blood sample.

Results of experiments and cases of carbon monoxide poisoning in man have been summarized in tabular form by Pieters & Creyghton (1951). This summary indicates that a concentration of 1 ·28% CO in air is fatally toxic to man within 3 minutes. Whether or not its effective time is ever quite so short, one cannot doubt the toxicity of CO to warm-blooded animals. Yet Haldane showed that this toxicity resided mainly in its ability to prevent Hb from transporting O₂. In one of his experiments, a mouse survived unharmed in an atmosphere of CO containing two atmospheres of O₂ (Haldane & Priestley, 1935). However, these authors point out that CO may have other effects. Wells (1918) found that even low concentrations of CO were extremely toxic to five different species of fish, but his results suggest that the bullhead, Ameiurus, may survive for some time in fairly high concentrations. Survival of animals in presence of high concentrations of CO is rather surprising in view of the fact that it is known to combine with cytochrome oxidase. Keilin & Wang (1946) cite a value of 0 ·1 for the partition constant of that enzyme. From this it may be estimated that about 29 % of the enzyme would be blocked by the highest concentration of CO used in these experiments.

In discussing these experiments in which goldfish were exposed to various concentrations of CO, it will be assumed that the exposure resulted in the concentration of COHb shown by the calculation in the relevant tables. This assumption throws considerable stress upon the one measurement of M, but it does receive some support from the analytical data of Nicloux (1923). The stress is further relieved by the fact that the present value for M is lower than has been previously recorded for vertebrates ; hence it might be expected that the calculated percentages of COHb are minimal values if they are in error.

Perhaps the most promising aspect of these experiments is that goldfish survived for fairly long periods of time in a concentration of CO much higher than is necessary to prevent their Hb from transporting O₂. They were able to do so over a fairly wide temperature range. In the one observation at 30 ° C., the time to death was over 24 hr. ; the effective time at lower temperatures has not been observed, but it is greater than 48 hr. These observations indicate that CO may prove to be a useful tool in further examination of fish respiration. These tests have not been extended to other species of fish, nor to other cold-blooded vertebrates, but the results of such an extension should prove enlightening (cf. de Graaf, 1957).

Survival of goldfish exposed to CO in flow experiments indicates that standard (basal) activity in fully aerated water is independent of Hb over most of the thermal range—possibly even to 30 °C—and that routine activity may have that independence over much of the lower range. By routine activity is meant that degree of natural activity such as one commonly observes in an aquarium (Fry; in Brown, 1957). The present experiments did not measure O₂ consumption, but activity was estimated from observation of control and experimental fish.

Despite concentrations of CO calculated to block all but a minute fraction of their Hb, goldfish were able to reduce the O₂ content of water to low levels in asphyxiation experiments. This was the most enlightening application of CO in the current experiments, for it indicates that the removal of O₂ under asphyxial conditions is not dependent upon the Hb content of goldfish blood. The reduction of the external O₂ supply can be accomplished by the circulating plasma. Thus the effect of CO₂ in the previous measurements of the CO₂ tolerance of this species was not limited to its effect upon the combination of O₂ with Hb. Hence no information regarding in vivo Bohr effect can be deduced from the CO₂ tolerance curves of this species. This observation may also apply to experiments measuring the effects of C0₂ upon 0₂ consumption of goldfish and it may well apply to other species.

In this respect, it is of interest to consider observations of Basu (1959) on the effect of C0₂ upon active O₂ consumption by various species of fish, including goldfish. His results clearly indicate that active O₂ consumption is influenced by CO₂ over the entire thermal range (cf. Basu, 1959, fig. 3E). He also confirmed the observation of Fry & Hart (1948) that active O₂ consumption by goldfish is independent of 0₂ concentration down to a low level (cf. Basu, 1959, fig. 5C, 11). In view of this respiratory independence, as the latter aspect of their physiology has been called, it is interesting to find that O₂ consumption under active conditions is so immediately sensitive to C0₂. Unfortunately there does not seem to be any comparable measurement of the effect of C0₂ upon standard metabolism. It would be of great interest to see both standard and active O₂ consumption of goldfish measured in presence of sufficient CO to block their Hb. Although the extent to which other species of fish depend upon the Hb in their blood has not been tested, the results of Nicloux’s (1923) experiments preclude the possibility that the present observations on goldfish are unique. Whereas goldfish with their remarkable degree of respiratory independence are also remarkably insensitive to CO, dependent species such as the trout, Salvelinus fontinalis, may prove to be quite sensitive. Fox (1954) recalls the observation of Hall (1930) that mackerel (Scomber scombus) must swim in order to obtain an adequate supply of O₂ and points out that this may indicate a constant dependence upon Hb that is probably unusual in fish.

Further reasons for doubting that C0₂ tolerance is an in vivo measure of Bohr effect may be found in the literature. The two curves reproduced in Fig. 2 A are from the work of Fry et al. (1947). They are C0₂ tolerance curves for goldfish acclimatized to 7 ° and 32 ° C. Two features of these curves, the high tension at which CO₂ begins to show effect and the suddenness with which the effect occurs, are scarcely what would be expected of Bohr effect displayed in vivo, nor do those authors believe that the curves can be interpreted so simply (cf. Black et al. 1954). This point may be illustrated by reference to similar measurements upon the squid, Loligo. Fig. 3 is reproduced from a paper by Redfield & Goodkind (1929). Although the pigment in the blood of Loligo is not haemoglobin but haemocyanin, the CO₂ tolerance curve for this mollusc is probably a true picture of Bohr effect displayed by the intact animal. That this is likely is even more apparent from the nomographic analysis of squid blood prepared by these authors. Unfortunately there are not sufficient data with which to make a comparable analysis of goldfish blood.

It is also possible that the apparent effect of thermal environment upon the curves in Fig. 2 A may not reflect a true difference in the sensitivity of the two groups of fish to high concentrations of C0₂, but rather a difference in solubility of that gas at the two temperatures. If quantity of C0₂ in mM./l. is plotted along the abscissa instead of partial pressure, all of the curves derived in the paper by Fry et al. (1947) inflect at about the same point, about 7 mM./l. (approx. 160 ml./l.), and the position of the two groups of fish indicated in Fig. 2A tend, if anything, to be reversed. This point is illustrated in Fig. 2B. A similar criticism has been raised by Doudoroff (in Brown, 1957).

It can be calculated that a concentration of 7 IDM. of CO₂/1., as mentioned above would result in a pH of about 43 if the water were unbuffered. The curves in Fig. 2 A are reminiscent of the sharply inflected curves resulting from Powers (1932) observations on the effects of pH upon the ability of marine species to remove O₂ from water. The inflection in the latter curves occurred at a much lower hydrogen ion concentration—at pH 6 ·6 −0 ·2. Powers used CO₂ to increase hydrogen concentration; hence his work fails to distinguish between effect of CO₂per se and effect of increased concentration of hydrogen ions. Weibe, McGavock, Fuller & Marcus (1934) used HC1 to increase the hydrogen-ion concentration in an extension of the work of Powers and of Pruthi (1927). Their curves (Fig. 4) relating pH to concentration of O₂ at time of asphyxiation of goldfish are less sharply inflected than those of Powers for marine species and also less than the CO₂ tolerance curves in Fig. 2 A. Nevertheless, it is noteworthy that the inflexion begins at about pH 5 ·2 in Fig. 4. This would suggest that in poorly buffered water the concentration of CO₂ at point of inflexion in Fig. 2 A could inhibit O₂ uptake by virtue of hydrogenion concentration. On the other hand, it has been shown that CO₂ has a very marked effect upon certain intact animals even when it is applied in well-buffered solution (Jacobs, 1919), and Hall (1931) demonstrated effects of CO₂ upon the pufferfish, Tetraodon, that could not be ascribed to the associated hydrogen-ion concentration. Even the equivocal results of Powers (1932) suggest that it is CO₂ rather than the concomitant hydrogen-ion concentration that is inhibiting O₂ removal under asphyxial conditions, but it will require further experimental work to estabfish this point for goldfish.

Turning from the physical means by which CO₂ exerts its effect to the physiological route, it appears that our knowledge is equally, if not more, uncertain. At least four activities are involved in removal of O₂ from water by fish. First of all, the water must be brought into intimate contact with the blood by being forced through the gills. This ventilation is achieved by respiratory movements. Secondly, the blood must take up the O₂ from the water. It is generally acknowledged that the transfer of respiratory gases between fish and their environment is by simple diffusion, but O₂ in the blood is held in both simple solution and chemical combinatiom Thirdly, circulation of the blood is essential to continuous removal of O₂ from water to tissues. Finally, the metabolic activities at cellular level must create a demand for O₂.

Since the plasma of goldfish has been shown to be an adequate vehicle for removing oxygen under asphyxial conditions, failure of the blood to take up oxygen is no longer a satisfactory explanation of the relationship between the asphyxial tensions of O₂ and CO₂ for this species. Hence attention is directed to ventilation, circulation, and metabolic activity as possible mediators of the effect of CO₂ upon asphyxial O₂ levels.

Present evidence indicates that CO₂ affects ventilation more immediately than it affects circulation. The literature contains few relevant accounts of circulation in fish (cf. Mott; in Brown, 1957). Most of the investigations have dealt with elasmobranchs; none has been concerned with the effects of CO₂ in high concentrations. Weibe et al. (1934) observed that fish may recover if transferred to well-aerated water after cessation of respiratory movements in asphyxiation experiments. Recovery of two goldfish in the present experiments confirms their observation and suggests in addition that cessation of respiratory movement precedes cessation of circulation. Furthermore, the criterion by which death was diagnosed in asphyxiation experiments favours the view that ventilation may be the activity by which asphyxial levels of O₂ and CO₂ are related. The well known narcotic effect of CO₂ suggests that its effect upon respiratory movement may be exerted via the central nervous system.

On the other hand, the toxic effect of CO₂ may arise from its action at cellular level. This possibility may be obscured by the means of diagnosing death in asphyxiation experiments and by the tacit assumption that O₂ continues to be consumed so long as movement can be discerned. It is not only possible but likely that movement continues for some time in the absence of O₂ consumption as measured by its disappearance from the water—i.e. the tissues go into O₂ debt. Present experiments with asphyxiation in N₂ indicate that a fairly large debt is accumulated. Hall (1931) found that increasing CO₂ tension reduced both O₂ consumption and frequency of respiratory movement of the pufferfish, Tetraodon maculatus. The reduction in frequency was accompanied by a marked increase in ventilation volume—i.e. volume of water passing through the gills per unit time— presumably through an increase in amplitude of respiratory movement. Hence there was, for a time at least, an increase in work during a decrease in O₂ consumption. This may be evidence that high concentrations of CO₂ block the mechanism by which the cellular batteries are normally recharged.

During the course of this work, I enjoyed the inspiring supervision of Prof. F. J. W. Roughton, F.R.S. I am indebted to Prof. Sir James Gray, F.R.S., for accommodation in the Zoology Department and for his encouraging interest. The problem was called to my attention by Prof. F. R. Hayes, F.R.S.C., and I am further indebted to him for advice in preparing this paper.