The Mechanism of Gill Ventilation in the Dogfish and Skate

The general nature of the respiratory movements of cartilaginous fishes is well known, especially since the opening and closing of the valves which cover the individual gill slits are very obvious in the dogfish and sharks as are the ’winking’ movements of the spiracular valve in skates and rays. But despite descriptions of the mechanism of ventilation, particularly in the dogfish, in many text-books of zoology most of these descriptions are inadequate in detail. Darbishire (1907) gave a good account of the respiratory movements in the dogfish (Scyllium canicula), paying special attention to the direction of the water current in the spiracle not only in the dogfish but also in the angel fish and skate. In doing this he noted that carmine particles which entered the spiracle escaped chiefly through the anterior two or three slits, whereas water entering the mouth was ejected from the last three gill slits. No adequate explanation of this difference in flow has been suggested. Darbishire also drew attention to the importance of the spiracle for the entry of water in the skate and to the ’spouting’ which occurred through this opening when the direction of the current was violently reversed. This behaviour was similarly described by Rand (1907) who noted that during ’spouting’ the gill openings must be closed by muscular action. As Mines (1913) stressed, this phenomenon seems to be a regular part of the breathing behaviour of flattened cartilaginous fishes, although it may also be induced by tactile or chemical stimulation near the spiracles. Baglioni (1907) drew attention to the differences in ventilation mechanism of pelagic and benthic selachians and incorporated them in a system of classification similar to that for bony fishes. As with the latter group he considered that the respiratory current was largely produced by the action of a buccal force pump. Woskoboinikoff (1932), on the other hand, emphasized the importance of the pumping action of the gill pouches themselves, and maintained that this was the most primitive mechanism, as it is the sole mechanism in the Agnatha. Balabai (1939), however, from measurements of the maximum and minimum pressures in the oro-branchial and parabranchial cavities, suggested that as in teleost fishes (Woskoboinikoff & Balabai, 1936, 1937; Hughes & Shelton, 1957, 1958), water passes across the gills as a result of the combined action of a force pump in front and a suction pump behind the gill lamellae. Teichmann (1959), in a comparative study published since the present work was carried out, has given evidence concerning the relative importance of these two pumps in Scyliorhinus, Mustelus and Torpedo. In the present work attention has been paid to the time-course of the ventilation movements and associated pressure changes in the different cavities of the respiratory apparatus in the dogfish and skate.

The dogfish used in this work was Scyliorhinus (Scyllium) canicula (L.) and the fish referred to as ’the skate’ was in fact the thornback ray, Raia clavata L. Specimens had been freshly caught and kept in the sea-water circulation at the Plymouth Laboratory of the Marine Biological Association, whose aquarium was of great value for watching the normal respiratory movements of these animals. The size of the animals varied, but usually the dogfish weighed about 600 g. and the skates about 400 g. All experiments were carried out under light anaesthesia. First of all the animals were deeply anaesthetized before they were fixed in the experimental tank, where they were allowed to recover to a state of light anaesthesia. In no case was the depth of anaesthesia sufficient to stop the breathing movements completely. The anaesthetic used in the first series of experiments was urethane, but it was found in later experiments that MS. 222 (Gilbert & Wood, 1957) gave much more consistent results, the animals showing a constant pattern of breathing for several hours which was indistinguishable from that of a resting unanaesthetized animal. The concentration used varied according to the size of the animal, but was usually about 1 in 50,000. The dogfish were held in a larger version of the clamp used in studies of freshwater teleosts (Hughes & Shelton, 1958; Shelton, 1959), but in many cases it was found that the body need be clamped only very lightly. Holding the skate was not so easy, but an adequate method was found by lightly bandaging it to a brick which was placed in the experimental tank. Most experiments were done with the skate’s ventral side uppermost in order to make it possible to insert the pressure-recording needles into the gill slits. As the spiracle is on the dorsal side it was difficult to obtain simultaneous recordings of the spiracular and gill slit pressures, but this could be done if the fish was held so that the spiracle did not lie over the brick. The experimental tank of about 25 l. capacity was filled with fresh sea water containing the anaesthetic and was kept constantly aerated throughout the experiments. The temperature of the water was 10–12° C.

The pressure-recording apparatus was similar to that already described (Hughes, 1958; Hughes & Shelton, 1958). Two Hansen condenser manometers, one utilizing a modified high-frequency circuit (Machin, 1958), were used in the main part of the present study. An Ediswan four-channel pen recorder made it possible to record simultaneously the pressures in two places together with the corresponding movements. As the frequency response of the pen recorder was flat up to 90 cyc./sec., which was also true of the electric manometers, the wave-forms were faithfully reproduced. Movements of the mouth and branchial apparatus were recorded by means of RCA 5374 mechano-transducer valves. Simultaneously with the pen recordings it was possible to record any two of the four channels on a Cossor 1049 double-beam oscilloscope, and these oscillograms were used for more detailed analyses of the wave-forms especially when comparing the pressures at particular instants, as this avoided errors due to the alignment of the pens and the curvatures of their recording arcs. In the initial experiments on the dogfish some films were taken simultaneously with pressure and transducer records, and these results proved of value for comparison with the results obtained later using the four-channel recording apparatus. These films, together with results obtained on freshwater fishes, have confirmed that the use of a long light arm attached to the anode of the RCA 5374 gives recordings of the respiratory movements which are essentially the same as those obtained by the more laborious analysis of ciné films. Films were also taken of unanaesthetized dogfish and skates, and also some of animals which though under light anaesthesia were not restrained in any way. Of particular value in ascertaining the time relations of the movements of the spiracular valve in the skate were films taken of young animals about 5 in. in fin span. With these it was possible to photograph both dorsal and ventral surfaces simultaneously by making use of a suitably placed mirror. Indian ink or milk adjusted to the same specific gravity as the water was used to follow the course of the respiratory stream. A Zeiss Movikon 16 mm. camera was used at speeds of 16–32 frames/sec., and lighting provided by a pair of photofloods.

The most characteristic feature of the gill apparatus in cartilaginous fishes, contrasting with that of bony fishes, is that the respiratory current passes out by a number of openings on each side. The presence of these separate gill slits externally is due to the well-developed septa which separate the two rows of gill filaments attached to a given branchial arch. The number of gill slits varies from five to eight, but in the species used in the present investigation there are five pairs, which is the number most generally found throughout the group. The gills themselves are characterized by the formation of gill pouches which communicate with the pharyngeal cavity by relatively large internal gill openings. Functionally the gill pouches together with the bucco-pharyngeal cavity form a single cavity which will be referred to as the oro-branchial cavity. The gill pouches may be compressed by the action of some of the intrinsic musculature of the visceral arches and their expansion is largely due to the elasticity of this skeleton but may be aided by some antagonistic muscles. No attempt has been made in the present study to determine the precise roles of the many different groups of muscles involved in the respiratory movements except in so far as these can be judged from the records and their known morphological disposition. When the water current has passed over the gill filaments it collects into cavities which run dorso-ventrally beneath the lateral walls of the branchial region in the dogfish and mainly horizontally beneath the ventral surface in the skate. These cavities were named by Woskoboinikoff (1932) the parabranchial cavities and his nomenclature will be adopted here. In the dogfish and skate they are only open to the exterior for a small part of their length but in other selachians, e.g. Cetorhinus, the basking shark, their openings are relatively much longer. Pressure measurements were readily made from these cavities and provide useful information about the mechanisms by which the flow of the respiratory current across the gills is maintained. Collectively they are analogous to the opercular cavities in the teleost fish. The buccal cavity of the latter corresponds to the orobranchial cavity. A further difference from the teleost fish is the persistence of the second visceral slit as a spiracle through which water may enter or leave the orobranchial cavity. This opening is guarded by a valve on the anterior side of the cleft which can be actively closed by the action of the first dorso-constrictor muscle.

It must be emphasized that the results described below all relate to the respiratory movements of animals which are stationary relative to the water. Since these experiments were carried out it has been possible to observe other selachian fishes at the Marineland Aquarium, California. In one of these, the leopard shark (Triakis semifasciata), it is quite clear that during normal swimming the mouth is held wide open and the gill flaps are also open, water passing in a continuous stream over the gills. As the speed of swimming falls, the fish makes occasional pumping movements which become more frequent when it rests on the bottom. As far as could be judged from the films taken, these movements are identical with those described below.

When observed at rest or swimming slowly in an aquarium a most obvious feature of the breathing movements is the rhythmical expansion and contraction of the whole branchial region, and the regular opening and closing of the mouth and of the small flaps which cover the five pairs of gill slits during the expansion phase. These movements occur every 1-2 sec. and have components in all three dimensions. The floor of the oro-branchial cavity rises and falls rhythmically as the mouth closes and opens. The walls of the gill region are compressed laterally and also appear to move in a posterior direction as the water is expelled from the gill slits. The presence of these complex movements makes them difficult to record and to represent adequately in even a diagrammatic way. However, a combination of the analysis of films taken in side and dorsal views and of the movement records gives a fairly consistent picture. The use of milk adjusted to the specific gravity of sea water pipetted near the mouth and spiracular openings made it possible to confirm the observations of Darbishire (1907) on the direction of the current. It is quite clear that this tends to be unilateral, for water entering a spiracle only leaves through the three anterior gill slits on the same side. Normally there is very little reflux through the spiracle although this is strong during the occasional ’spouts’. The valve guarding this opening is an active one and it must be efficient otherwise it would short-circuit the main flow through the gills. There appeared to be a certain amount of reflux through the mouth as it closed and the oro-branchial cavity decreased in volume. The passive maxillary and mandibular valves are not well developed in Scyliorhinus and there must be some loss of water in a stationary fish. Water entering the mouth escapes through the last three gill slits, and when milk is pipetted into one side of the mouth it issues from the gill slits on the same side.

The decrease in volume of the oro-branchial and parabranchial cavities takes place comparatively rapidly during about one-quarter of a cycle and is succeeded by a slower phase of expansion. The latter is fairly rapid to start with but becomes progressively slower. During this expansion phase the extensions of the gill septa which form valves over the gill slits are seen to be ’crinkled’ as if drawn by suction from within. A more or less distinct pause occurs following the expansion phase, during which the gill flaps become slightly free from the body wall, the onset of the next expiratory phase being marked by the synchronous and rapid opening of these valves. The dorso-ventral movement of the buccal cavity always precedes the lateral expansion of the gill region. This has been recorded many times by means of the transducers. A most convenient point for recording this movement with little interference to the gill movement was just in front of the first gill slit, where the head of the dogfish is widest. On other occasions movements of this region were recorded more posteriorly from positions between and just above a pair of gill slits. The latter records repeatedly indicate that when recording close to a valve a brief abduction precedes adduction of the branchial arch. Furthermore, the relative size of this abduction increases in recordings taken from the more posterior slits. This is most clearly shown in Fig. 1, where the four-channel recorder was used with transducers on all channels. These records show that the first movement to be recorded, whether it be an abduction or adduction, is simultaneous throughout the whole gill region but that if one considers the adductor movements alone then these spread from in front backwards. This is also true of the expansion movements of the branchial region, i.e. the first gill pouch begins to expand before the second, etc. Another difference to be noticed in these records and which is connected with the preceding observations is that the rate of the adductor movements is more rapid in the more anterior arches. The same is also true of the succeeding more rapid phase of expansion.

The complete interpretation of these recordings must be deferred until after the pressure changes in the parabranchial cavities have been described.

The time-course of pressure changes in different parts of the respiratory system is very similar in its general form (Figs. 2 and 5). It includes times when the cavities are positive or negative, in addition to periods when the pressure is the same as that of the outside medium, which will be referred to as zero pressure. The relative size and duration of these different parts of the cycle vary according to the position of the cavity. There are also variations between different individuals, but the present account is based upon the pattern found in about 70 % of the experimental animals. Pressures recorded from the oro-branchial cavity with a needle inserted into the mouth or through the spiracle both showed that the maximum positive pressure attained (0·5−1·5 cm. water) was greater than the maximum negative pressure (0·2−0·5 cm.). The opposite was true of the pressures recorded from any of the parabranchial cavities. In the latter the negative part of the curve is predominant and reaches a maximum of 1−2 cm., whereas the maximum positive pressure attained does not usually exceed 1 cm. Some differences were observed in the pressure curves recorded from different gill slits. For instance, Fig. 2 B shows simultaneous recordings from the first and fifth gill slits on the same side. It can be seen that the general wave-forms are similar, but closer inspection reveals that while the positive phases are about equal in magnitude, the maximum negative pressure in the more anterior slit is greater. Furthermore, it is noticeable that the rate of fall of this pressure is much more rapid than in the case of the pressure recorded from the fifth parabranchial space. The oscilloscope records (Fig. 3B) confirm these features, and also show a slight phase shift between the two curves, since pressure changes in the more anterior slits precede the corresponding changes in the posterior slits by a short but perceptible time.

This phase difference between the wave-forms is even more obvious when simultaneous recordings from the oro-branchial cavity and one of the parabranchial cavities are compared (Fig. 3 A). The oro-branchial cavity becomes positive a little sooner than the parabranchial cavities and likewise it usually becomes negative before the parabranchial cavities. This latter feature means that, just as in the teleost fishes, there is a brief phase (4) during a cycle when the two curves cross and the direction of the differential pressure reverses, thus tending to produce a reversal in the direction of the respiratory current. The differential pressure curve (Fig. 4) shows that a gradient exists from the oro-branchial to parabranchial cavities except during this transition when the oro-branchial cavity begins to increase in volume.

The pressure curves make it quite clear, therefore, that the flow of water is maintained by the action of two pumps as in teleost fishes. One of these (the force or pressure pump) depends on a greater positive pressure from the oro-branchial to parabranchial cavities and the other (the suction pump) involves a greater negative pressure in the latter cavities.

As mentioned above, the general form of the pressure curve recorded when the needle was placed within the spiracle was similar to that recorded from the oro-branchial cavity by needles inserted through the mouth (Fig. 5). The two pressure curves are similar in that the positive part is predominant in both, but they differ in detail. This is seen in the rate of fall of the positive pressure, which is much more rapid and occurs a little sooner when recordings are made from the spiracle than from the mouth or any of the gill slits. As the spiracular pressure returns to zero that of the oro-branchial cavity is more negative. Thus the pressure gradient drawing water into the oro-branchial cavity is greatest first of all via the spiracle and later via the mouth. This would suggest that the first water to enter this cavity will be through the spiracle, and it is reasonable to suppose that it will enter the anterior gill pouches. A further factor contributing to this must be that water entering the spiracle will tend to take up a more lateral position in the flow through the oro-branchial cavity, which is likely to be laminar on account of its relatively low velocity. When the needle was gradually withdrawn from the spiracle a point was reached where the positive phase disappeared. In this position the pressure was being recorded just outside the spiracular valve where only a negative pressure is present. If the needle was carefully manoeuvred until it was directly opposite the valve it was possible to obtain a small positive deflexion as the valve closed.

In general this is fairly clear and what might be expected. Thus as the mouth closes the pressure in the oro-branchial cavity rises and remains positive until the orobranchial cavity begins to expand again. It then becomes negative, water is drawn in through the mouth, and the pressure rises to be the same as that outside. The pressure recorded from the spiracle has similar relationships, but it usually becomes positive a little later than that recorded from the mouth. The spiracular pressure becomes positive when the spiracular valve closes. The form of the curves recorded from the parabranchial cavities and the corresponding movement records are not always so easily interpreted. Movement recordings from the first slit can be related to the pressure changes in the parabranchial cavities quite easily as during the adduction movements of this arch the pressure rises and during abduction it falls, becoming negative. Movement recordings from the more posterior slits, however, give the impression that the arch is maximally adducted as the pressure in the corresponding parabranchial cavity becomes negative and similarly the commencement of the positive phase seems to coincide with the brief abduction which precedes the main adductor movement.

These observations become easier to interpret if the movement recordings from near the gill slits are considered in two components. One of these is due to the activity of the branchial musculature of the arch concerned, and the second is produced indirectly as the result of pressure changes to which the pouch is subject by the action of the force pump. Thus the initial abduction is entirely due to the passive opening of the valves as water is forced out at the beginning of the decrease in volume of the oro-branchial cavity. In confirmation of this it can be shown that this abductor movement decreases and may disappear if the recording lever is moved away from the valve. In some experiments when the pressure needle was inserted more deeply into a gill pouch an extra positive deflexion was found which preceded the normal positive wave of the parabranchial cavity. This extra deflexion was synchronous with the positive pressure in the oro-branchial cavity and also with abduction of the valve. Correspondingly at the other transitional phase when the oro-branchial cavity begins to expand and the pressure differential is temporarily reversed, some movement records indicate an extra adduction which is due to the ’sucking in’ of the valve. Again, such recordings are only found when the transducer arm is placed just in front of a slit where it is most affected by the valve itself.

In a skate resting quietly at the bottom of an aquarium it is apparent that water enters via the two dorsally situated spiracles whose valves open and close rhythmically with a frequency of about 30/min. When the animal is more active and swimming about, this frequency increases and water also enters through the mouth. This was also true under the conditions of the present experiments, and it seems to be normal when the animal is resting with its snout slightly raised (Rand, 1907). This posture can often be observed in the guitarfish (Rhinobatos productos) at the Marineland aquarium. When the skate is almost completely buried it seems unlikely that the mouth will be opened. Rand states that the mouth and spiracular valve close together but that opening of the mouth is slightly later than that of the spiracle ; but other accounts suggest they are synchronous in all phases. Analyses of both films and transducer records show quite clearly that the mouth precedes the spiracular movement during both the opening and closing phases by about one-tenth of a cycle. Expansion and contraction of the branchial region are also delayed with respect to the mouth movements (Fig. 6). Movements of the individual gill pouches are more nearly synchronous than is the case in the dogfish, but a slight delay can be seen between the movements of the first and last gill arch. This is true for both the expansion and compression phases. The spiracle is definitely closed by an active valve, and although the flaps which cover the gill slits appear to be passive, there is evidence for a mechanism which actively closes the parabranchial cavities to the outside. The greater synchrony of action of the gill pouches is to be expected from the structure of both skeletal and muscular systems, and is due in part to the fusion of the basibranchial cartilages into a copula.

The respiratory current leaves through the five pairs of gill slits but during more intense breathing the use of dyes in the water shows a very slight reflux through both the oral and spiracular openings. Under these conditions, which were sometimes found towards the end of an experiment, another very characteristic feature of the respiratory rhythm is the periodic ’spouting’ of water through the spiracle which was observed by Darbishire (1907) and Rand (1907). This activity may be due to a lack of oxygen or to mechanical stimulation of sense organs in the spiracle by mucus or other foreign bodies in the water but it is also a part of the normal respiratory cycle (Mines, 1913). The current entering via the right spiracle or right side of the mouth usually leaves by the gill slits on the same side. There is a tendency for water entering by the spiracles to leave by the anterior gill slits, and for that which enters by the mouth to emerge through the posterior slits, but this is not so marked as in the dogfish.

The pressures recorded from the oro-branchial and parabranchial spaces show that the flow of the respiratory current is normally maintained by the action of both force and suction pumps. The oro-branchial pressure curve has a more positive phase than that of the parabranchial curve, which has the greater negative phase (Figs. 3 C, 7). The positive phase of both pressure curves is associated with a decrease in volume of the cavity and the negative pressure likewise associated with an increase in volume. The latter is about two-thirds of the respiratory cycle and the former one-third of the cycle. As in the typical teleost fish, these two major phases are again separated by phases of transition, but here the two curves come very close together during both transitions. In contrast to the dogfish and most teleost fishes, there does not seem to be any reversal of the pressure gradient which usually occurs as the suction pump takes over from the pressure pump. In fact the opposite is the case here, for the parabranchial pressure falls before that in the oro-branchial cavity. It can also be seen that the two pressures follow the same course during the other transitional phase (2), and in some instances there is a suggestion of a slight reversal. The result of the interaction of these two pumps in maintaining the flow across the gill is shown by the differential pressure curve (Fig. 4). Recordings from any of the five slits seem to be the same except during a ’spout’ when the first slit has a longer and more positive pressure than the more posterior slits (Fig. 8 A). During ’spouting’ the positive pressure in all of the slits is greater than in the buccal cavity and it is due to this that the water current is reversed and forced out through the spiracles and mouth (Fig. 7).

It has not been possible to record the pressures in the spiracles simultaneously with pressures in the oro-branchial or parabranchial cavities. Attempts to insert a needle into the former via the spiracle at the same time as the spiracular pressure was recorded were not successful because irritation of the sensitive spiracular epithelium immediately caused the animal to move. It was possible, however, to record the spiracular pressure on the two sides simultaneously with the movements of the spiracular valves. These experiments showed that the positive phase of the spiracular pressure normally coincides with the time of maximum closing of the spiracle (Fig 8B). As the spiracular valve closes just after the mouth it is probable that these pressure changes in the spiracle occur soon after those recorded in the mouth. The pressure curves from both spiracles were very similar so long as care was taken that the recording needles were in identical positions on the two sides. As in the dogfish, the pressure just outside the valve has a small negative phase only, whereas once the needle has passed the valve a positive phase becomes apparent. During ’spouting’ the spiracular valves only partially close but maintain the normal rhythm. Several times an asymmetry in the action of the two spiracles was observed. Rand (1907) also noticed such unilateral ’spouting’. In one of the spiracles the positive phase was increased, whereas in the other it was less than the normal pressure (Fig. 8B). This decreased pressure still had fundamentally the same time-course as in the normal rhythm and reached a maximum when the spiracular valve was maximally closed, whereas the side with the increased positive pressure showed a delayed maximum which now coincided with the position of maximum opening of the spiracular valve. Direct observation of the water stream showed that during the normal respiratory cycle there was a slight and equal reflux through both spiracular openings as the oro-branchial and parabranchial cavities decreased in volume, but during the ’spouting’ cycle the greatest reflux occurred through the spiracle with the increased positive pressure. Furthermore, water was leaving when the valve was maximally opened and not when it was nearly shut as was normally the case. During these ’spouting’ actions the positive pressure in both the oro-branchial and parabranchial cavities was increased above normal but, as mentioned above, that of the latter became the greater. It preceded and was often more prolonged than the oro-branchial pressure (Fig. 7). The difference in amplitude of these two pressures may be as great as 3–4 times, but in other cases it is only slight. Movement records showed that during this phase the gill pouches contracted a little earlier than normal but maintained the antero-posterior order ; but during the succeeding phase they expand simultaneously. The amplitude of their contraction is also increased during a ’spout’, whereas that of the mouth remains unchanged. Closing of the oro-branchial cavity is interrupted by a brief expansion phase which coincides with the maximum positive pressure in this cavity. Its continued decrease in volume reaches a minimum which coincides with maximum closure of the gill pouches; thereafter the whole system expands synchronously as mentioned above.

The present investigations have substantiated the view that the flow of water through the gills of two species of cartilaginous fish is maintained by the action of two pumps which operate on both sides of the gills. The time-course of pressure recordings in these two positions makes this quite clear and gives a good indication of the nature of the flow across them. As with the teleost fish, the flow is probably more or less continuous. The existence of a pause between two successive cycles in Scyliorhinus (as was also noted by Satchell (1959) in Squalus lebruni) when the pressure throughout the whole system is very close to that of the external medium suggests that any flow during this part of the cycle must be extremely slow. Phase 2 of the differential pressure curve (Fig. 4) indicates this, but it has been pointed out (Hughes & Shelton, 1958) that the effect of possible changes in gill resistance must always be borne in mind when interpreting these curves.

This is true, for instance, when using the pressure curves to decide the relative importance of the two pumps in maintaining the flow across the gill resistance. From evidence at present available, however, the areas beneath phases 1 and 3 of the differential curve certainly give a fair indication of the work done by the suction and force pumps respectively. If the resistance remained constant throughout all phases of the respiratory cycle, it would be possible from these areas to express quantitatively the relative importance of the two pumps in maintaining the flow across the gills. Even allowing for some change in resistance the curves provide good evidence that the suction pump plays an important part in both the species investigated and that its importance is greater in the skate than in the dogfish. This result agrees with that of Teichmann (1959) for Torpedo, but is at variance with his conclusion that the force pump does most of the work in Scyliorhinus. He noted, however, that in some individuals the suction pump was more important than in the experiment which he quotes. Such individual variations seem to be very characteristic of dogfish, for Balabai (1939) found quite large variations in the relative magnitude of pressure changes in the two cavities. In some specimens the orobranchial cavity became both more positive and more negative than the parabranchial cavity. Certainly individuals have been found in the present investigations in which the force pump was better developed than that shown in Fig. 4, but in no case is it the only mechanism. Of course if the gill resistance should fall during this phase similar positive pressures will be recorded on either side of the gills and there will be very little differential pressure across them, although the force pump is actively expelling water from the whole system. What matters most, however, is the work done by the pumps in ventilating the gills, and if the resistance of the gills is low it is because a smaller proportion of water is passing between the secondary lamellae. Hence in terms of ml. of oxygen absorbed per unit of work done by a pump, the differential pressure is a good guide to its importance.

Teichmann’s method of assessing the relative importance of the two pumps was to measure the maximum height to which they could force water above the surface when one of the two pumps was functionally disturbed. As he points out, the animal can and usually does compensate for these disturbances. Interference with the force pump by placing a tube into the mouth of Scyliorhinus decreases the height by about two-thirds, but this does not necessarily mean that in the normal respiratory cycle the suction pump plays a very minor role. This method is suitable to show the ability of the pumps to force water out of the system as a whole. As the pressure curves show, this is not the primary function of the suction pump, however, which is particularly adapted to draw water through the gill system. Furthermore, in addition to the maximum pressure achieved, it is also necessary to take into account the duration of the action of the pumps, which is longer for the suction pump, and for this purpose a study of the time-course of ventilation is essential.

One of the most interesting features of the differential pressure curves is the apparent absence of a reversal in the gradient in recordings from the skate. This has also been found in some bottom-living teleost fishes (Hughes, 1960). The simplest explanation appears to be that the opening from the parabranchial cavities to the outside can be actively closed. This would enable a fall in pressure to occur in these cavities before it does in the oro-branchial cavity and hence would maintain the direction of the pressure gradient during this phase. The suggestion of an active control of these apertures is not new, for in an addendum to his paper Darbishire (1907) mentions that Prof. Herdman was of the opinion that the gill covers in the living dogfish were not passive but active agents in determining the respiratory current. Darbishire came to the conclusion that the anterior part of the gill cover which is supported by the gill rays can be moved actively, but that the portion not supported by the rays is entirely passive. Rand (1907) also drew attention to the need for an active mechanism closing the gill slits in order to ensure a reversal in the flow during ’spouting’. The latter is certainly better developed in the flattened forms and correlates with the nature of their differential pressure curve. The value of such a mechanism to the animal probably lies in the fact that it ensures that no water can enter through the external gill slits, which must occur, however slightly, in a system entirely dependent on passive valves. The danger of particles of sand damaging the gills if they entered by such a route is of course much greater in fish which he partially buried on the sea bottom.

The actual details of this mechanism require further investigation but certainly the morphology of the gill septa makes possible an active control of the type suggested by Darbishire. Lighttoller’s (1939) description of the different layers of constrictor muscles also fits in with his suggestion. It would appear that the dog-fish has the structural basis for such control, but that apart from ’spouting’ it is used much more rarely or in a different way during the normal respiratory cycle.

It is a pleasure to record my thanks to the Director and Staff of the Plymouth Laboratory, both for the facilities and material they provided and their friendly assistance throughout the work. I also wish to thank Dr G. Shelton for his collaboration in the early part of this work, and Dr K. E. Machin for his assistance in designing the electronic apparatus.