The Neurological Basis of the Locomotory Rhythm in the Spinal Dogfish (Scyllium Canicula, Acanthias Vulgaris): I. Reflex Behaviour

One of the crucial problems of animal behaviour concerns the spontaneity or drive behind the motor activities, another the pattern or co-ordination of the various phases of movement. In this connexion the study of some elementary motor patterns, such as respiration, locomotion, cleaning and scratching, has led to far-reaching conclusions, which have also been applied to more complex acts of behaviour (Lorenz, 1939).

One theory presents the animal as an intrinsically active system, which only comes to rest under suitable inhibitory stimuli ; according to another it is normally at rest, and is only activated by suitable excitatory stimuli. Consequently, if removal of all sources of outside stimulation caused perpetual activity in a preparation, this would support the former theory; if, on the other hand, the preparation was in the first place perpetually active and became silent when all sensory influx was cut off, this would favour the latter. At the same time, the existence side by side of excitatory and inhibitory stimuli adds to the complications.

In recent years, however, the investigation of lower vertebrates—notably by Adrian, von Holst and P. Weiss—has produced a very definite picture based on the conception that rhythmically co-ordinated movements are driven and governed by an automatic mechanism within the central nervous system, and are not essentially dependent on any kind of afferent inflow. Since Adrian & Buytendijk (1931) recorded from the isolated brain stem of the gold fish slow changes of electrical potential which were more or less of the same order of frequency as the opercular movements, it has become customary to regard such changes as the expression of a spontaneous, automatic driving mechanism for various rhythmic activities. Under certain conditions, in fish, the respiratory and locomotory movements become synchronous. From this von Holst (1934, 1936) concluded that the electrical changes as recorded in the brain stem represent a process which is the common central cause of origin for respiratory and locomotory movements alike; in the normal, intact fish their frequencies merely become modified by regulating influences from the periphery.

It has often been suggested that the underlying mechanism consists of certain groups of ganglia in the central nervous system, which owing to their physiological make up, and under the influence of their normal metabolism, continuously produce a stimulating substance or an electrostatic field. This is said to elicit patterns of motor discharges as soon as a critical level of concentration is reached. At that point the stimulating agent is destroyed and the process starts anew.

In so far as the locomotory rhythms of teleost fish are concerned, von Holst (1939), after extensive research on the subject, has reached the following conclusions : groups of neurones arranged along the spinal cord are responsible for the rhythmic activity; they operate automatically as long as the central excitatory state is kept at a certain level, which is constantly maintained by the medulla. Hence, a medullary preparation is persistently active, whereas a spinal fish merely responds to reflex stimulation. This theory of an internal automatic rhythm which governs the locomotory movements of the trunk appears to have been widely accepted for teleosts and elasmobranchs such as the dogfish (Gray & Sand, 1936; le Mare, 1936), except that in teleosts ‘the control of rhythmical movements resides in neurones restricted to the brain, whereas in the dogfish cells all the way down the spinal cord are capable of initiating such movements’ (le Mare, p. 437). This additional assumption in the case of the dogfish appeared necessary, because it has been known since Steiner (1885) that a spinal dogfish exhibits a persistent locomotory rhythm.

The dogfish therefore seems to be particularly suitable for an investigation of the problems outlined above. The experiments reported here were designed to throw some light on the two alternative theories, viz. whether the locomotory movements are due to a central drive and a central pattern of activity, or whether they can be considered essentially as moving sites of nervous integration, continuously re-excited from within or without the central nervous system, and ruled by co-ordinating patterns of afferent impulses.

The experiments were performed on fourteen individuals of Scyllium (Scylliorhinus) canicula, varying between 55 and 70 cm., and ten Acanihias vulgaris (Squalus acanthias), 48–75 cm. in length. In the first part of this study the influences were examined which affect the swimming rhythm of spinal preparations. By determining the range of response to various isolated types of stimulation, and by correlating their sequence, one may attempt to build up all phases of the swimming rhythm. For this purpose the spinal animals were fixed by two strong vertical pins through the snout, allowing free lateral movement of the trunk. The persistent swimming and the reflex responses were studied by attaching a writing lever to the anterior end of the base of the first dorsal fin, and kymographically recording the movements. The stimulation employed was mostly tactile: touch with a camel-hair brush, a straw, or a blunt seeker, and diffuse touch to the body by lifting and lowering a glass plate. Other stimuli used must be considered as nociceptive, or involving deep pressure receptors ; they were produced by means of a needle, a sharp metal clip or clamp, by pressure applied with the fingers, or a sharp pinch with forceps.

The behaviour of the spinal dogfish, Scyllium canicula and S. catulus has been described by Steiner (1885), Bethe (1899), Bauer (1926), ten Cate (1933), ten Cate & ten Cate-Kazejewa (1933), Gray & Sand (1936) and le Mare (1936). A general survey of the observations suggests that the time has not yet come when they can be brought into satisfactory agreement. As the arising controversial points concern the excitatory and inhibitory stimuli which affect the swimming rhythm, they have been subjected to re-examination. However, no extensive analysis of the various reflexes has been attempted in the present paper, and it must be admitted that many questions are left unanswered. Some of the more striking discrepancies in the results of previous authors and the present observations may perhaps find an explanation along the lines suggested through the interesting observations of Bauer (1926), who describes (i) individual variations, (ii) differences arising from the different age of the specimens (cf. Polimanti, 1911), (iii) different behaviour of preparations according to the length of time elapsed since the operation, and (iv) reflex reversal after fatigue (cf. Gray & Sand, 1936). All observations reported below were made within the first 10 days, usually 2–5 days, after the operation.

While there is considerable disagreement about details, and while the general conclusions reached here are the opposite of those reached by other authors, the most important experimental finding in connexion with our problem has been confirmed for Scyllium, and has been found to apply equally for Acanthias vulgaris, i.e. a spinal animal of either species, mounted for recording as described above, displays uninterrupted swimming movements for a period of at least some days.

When swimming movements are initiated through exteroceptive stimulation in a spinal teleost fish two distinct processes appear to take place, as has been described by Gray (1936) and von Holst (1935 b) : (1) a rapid development of an undulatory posture over the body, and (2) a propagation of the posture along the body. Although it is not quite obvious in an actively swimming dogfish, there is evidence to show that conditions are essentially similar in both the teleost and the dogfish.

Before attempting to analyse the factors which govern the swimming movements, i.e. the passage of the undulatory posture over the body of the dogfish, it seems profitable to study in the inactive preparation the initial spreading of the posture. For this purpose animals can be used which are recovering from the anaesthetic, moribund preparations which have ceased to swim, or animals whose swimming movements have been inhibited (see below). A localized stimulus of short duration, applied to almost any part of the surface, causes the body to be thrown into an S-shape, which may persist for a short period before the body returns to its original outstretched position. The amplitude of the excursion, and the duration at which the reflex figure is held, clearly depend on the strength of the stimulus, and the time of application. The reflex posture itself, i.e. the position of the wave on the body, depends entirely on the site of stimulation. Thus if the base of the first dorsal fin be attached to the recording lever, and the body stimulated laterally at different levels, it will be noted that a touch to the left at the level of either the pectoral or the first dorsal fin will cause a movement towards the right of the median, while if the body is stimulated at the level of either the pelvic or the second dorsal fin the response will be reversed, and the region of the first dorsal fin will move towards the left (Fig. 1). At a point between the first dorsal fin and the pelvic fin no visible reaction will be recorded. It is interesting to note that von Holst (1934) has described a similar reaction in goldfish in which the spinal cord was transected at different levels; a tail reaction towards one side became progressively smaller the more anteriorly the cut was situated. It is not clear, however, from the report whether the stimulus was also applied more anteriorly, or whether the effect is due to the structural difference of the reacting system and the resulting different integration.

It is also noteworthy that when two opposing stimuli are applied simultaneously, each of which singly causes a monophasic response, no integration or superposition takes place in the sense that they cancel each other out, but the response becomes biphasic, i.e. the movement takes place in both directions in succession (Fig. 2). For instance, if stimulation of the left pectoral fin causes the region of the first dorsal fin to be moved to the right, and stimulation of the right pectoral fin movement to the left, then stimulation of both produces a movement left-right, right-left or left-right-left (Fig. 2b). The change from one position to the other is accomplished by a movement of the reflex posture along the body. This type of response recalls similar phenomena in the tetrapod locomotory system, e.g. if the hind leg of a spinal toad be passively extended it responds by flexion ; if both legs are simultaneously and passively extended by placing them on to a revolving drum, they flex alternately (Gray & Lissmann, 1940).

Prolonged gentle stimulation (touch with a camel-hair brush or straw) may cause the reflex figure to be held in a tonic contraction, but rarely for more than 5 sec.; adaptation to the stimulus seems rapid, and recovery slow (Fig. 3). More frequently a rhythmic component is superimposed on the tonic contraction ; swimming movements appear, with excursions towards one side if the stimulus is not applied symmetrically. Here again adaptation is rapid and the swimming beats tend to become symmetrical (Fig. 4). Prolonged bilateral stimulation causes in all preparations the appearance of symmetrical swimming beats, which may become more and more persistent as the preparation recovers from the operational shock (Fig. 4).

It seems obvious to associate the initiation of the swimming response with a with-drawal reflex—withdrawal from a tactile or nociceptive stimulus. The nature of the response, however, also depends on the method of applying the stimulus. Thus while one gentle stroke with a camel-hair brush causes withdrawal of the stimulated region, which may be followed by swimming (Fig. 5 B), a more persistent tickling with the brush of a localized area produces at first only slight movements, followed very suddenly by a violent ‘jerk’ (Fig. 5 A), which may be repeated and has then the appearance of a ‘shaking reflex’. It seems that rotational movements along the long axis of the body are associated with it. It is conceivable that this response—which also occurs during swimming—has a functional significance in the removal of ectoparasites which are in the process of attachment.^* The experiments reported below, however, refer mainly to the swimming type of response.

When the spinal cord of a dogfish is transected behind the medulla, the preparation fixed by the snout and free from contact, the course of subsequent events is extremely uniform. Fig. 6 may serve as an illustration. As the animal recovers from the anaesthetic breathing movements are the first to appear; the animal fails as yet to respond to exteroceptive stimulation. In the second stage a strong pinch to the more sensitive parts of the body elicits a single response. Somewhat later this response may be followed by a variable number of swimming beats—the number increasing as time goes on. Finally, either in response to a stimulus or without any accountable reason, the persistent swimming emerges. The time scale of events may take a few minutes or several hours, but not a single case has been observed in which the permanent rhythm failed to appear.

In an active preparation of Acanthias the amplitude of the beat as recorded from the first dorsal fin is about 2 cm., in Scyllium about 4 cm. The frequency is for Acanthias approximately 30–35 beats per minute, and about 40–42 for Scyllium; it is, however, likely to vary with external and internal conditions, e.g. temperature, age of preparation, etc.

If, on the other hand, the preparations were left in the tank after spinal transection, then their behaviour varied considerably, ranging between the type as described by ten Cate & ten Cate-Kazejewa (1933) to others which corresponded more closely to the account given by Gray & Sand (1936); in the former case they remained immobile at the bottom of the tank, and only occasionally began to swim, either ‘spontaneously’ or after excitation; in the latter case they exhibited an incessant swimming rhythm. The reasons for this difference are not quite clear, but attention should be drawn to Bauer’s (1926) discovery of certain areas on the fish body which may have an inhibitory effect on the swimming movements when suitably stimulated. Acanthias tended to be less active, and normally rested motionless on the bottom of the tank, whereas most individuals of Scyllium displayed the swimming movements.

As there appeared to be a marked contrast between the behaviour of these dogfish and others which were suspended from the snout, it seemed reasonable to suppose that the state of inactivity was caused by the touch to the surface of the animal. When the spinal Acanthias, which rested at the bottom of the tank, and the inactive specimens of Scyllium were held by the snout, either in a vertical or horizontal position, they invariably resumed their incessant swimming. However, as soon as a glass plate was gently placed against the ventral side of the animal the swimming ceased, in some preparations almost immediately, in others after some delay (Fig. 7). However, some—notably Scyllium—continued to swim. The fact that this response is the result of inhibitory action, and not caused through mechanical conditions, is clearly demonstrated by the observation that when the glass plate is carefully removed, the fish tends to remain inactive. A clear example of this behaviour is recorded in Fig. 8. This fish had been observed swimming without interruption for many hours. When the glass plate was held against the ventral surface the swimming ceased, but was subsequently resumed when the plate was withdrawn after a contact of about 10 sec.; the swimming then started with some beats of small amplitude and low frequency. If, on the other hand, the glass plate was held against the body for about 20 sec. or more, and then removed, the fish remained inactive, The swimming, however, could again be released by a single, gentle, tactile stimulus, and it persisted until a further inhibitory stimulus was applied. This experiment was repeated many times, extending the states of activity and inactivity to several hours, and only once was the previously inhibited fish found swimming after 5 hr. without the application of any intentional releasing stimulus.

This behaviour is just one instance of the many interesting analogous reactions observed by Sherrington (1904) in the spinal dog. Sherrington writes: ‘The mere gentle passive arrest of the moving limb often leads to the subsidence almost immediately of the rhythmic reflex either of step or scratch, the reflex being found to be no longer in operation on release of the limb after a brief arrest.’

The observations on the dogfish clearly indicate that (i) a diffuse touch to the ventral surface can inhibit the persistent swimming, (ii) the state of inactivity may persist unless a releasing stimulus be applied, and (iii) once the releasing stimulus has been applied the state of activity is maintained. As the state of either activity or inactivity is maintained for prolonged periods under equal external conditions, and as that state is determined by the initial stimulus, it is hard to avoid the conclusion that in the case of activity the excitatory effect of the releasing stimulus sets up a sequence of re-exciting processes (such as proprioceptive reflexes or central self-exciting cycles) which are the further cause of the persistence of movement.

Gray & Sand (1936) report that some preparations cease to swim when they arc turned on the back, touching the tank, and that these preparations resume swimming when returned to their normal position. No attempt has been made to repeat this experiment, but many specimens of spinal Acanthias have been found in the tank resting on the back without any signs of swimming movements. As has been mentioned before, localized touch to almost any part of the body accentuates the swimming. The only exception which occasionally was found to stop swimming, or more often to reduce the amplitude of the swimming beat, was a gentle touch with a camel-hair brush in the exact dorsal median line at the level of the pectoral fins (Fig. 9). This caused a marked ventral depression of the body in the stimulated region, and a lifting of the pectoral fins. The response is tonic in itself, and owing to the rigidity which it produces in the body is likely to reduce the amplitude of the swimming. According to Bauer’s description inhibition can be obtained regularly through gentle, blunt touch to the anal fin (not present in Acanthias), and a ventral zone slightly anterior to the tail fin, occasionally also by touching the back.

I have not been able to confirm other types of inhibitory action as described by previous authors. Neither Scyllium nor Acanthias ceased to swim after either touching, pinching or pressing in a variety of ways either of the dorsal fins (cf. le Mare, 1936). The response was invariably accompanied by acceleration and augmentation of the rhythm (Fig. 10).

Apart from the superficial cutaneous inhibitory effect, cessation of the rhythm has also been demonstrated to occur after application of strong pressure to the body. On this point there is general agreement, and little new can be added. However, I would hesitate to group these two phenomena of arrest closely together. Cessation of swimming on application of deep pressure is preceded by a rhythm of wider amplitude but slower frequency as compared with the normal swimming rhythm; this reaction often lasts several minutes. The power of the beats is considerable; cessation, when it occurs, has the appearance of fatigue. The general impression conveyed by this response is that it is the composite picture of a number of events, caused by simultaneous but differential contractions of the myotomes at the same level; one might assume reinforcement to take place in response to isometric conditions. It is noteworthy that preparations inhibited through ventral touch, when subjected at the same time to strong pressure, nevertheless show the same succession of reactions as preparations freely suspended in water (Fig. 11). On release accelerated and augmented movements emerge, followed by a period of swimming beats, despite the continuous ventral contact. If simultaneously with the application of strong pressure to the body a further nociceptive stimulus be applied, e.g. a clip attached to the tail, the frequency of the original response is increased; it is, however, not as high, as it is after the attachment of the clip to an otherwise undisturbed dogfish. The frequency of the two responses seems to integrate arithmetically—as if the effects were added and divided by two. Similar reactions have been recorded in the eel by von Holst (1935 a).

The question seems justified whether the rhythmic component of the response which appears on application of strong pressure to the body may be caused by a nociceptive stimulus, acting simultaneously with another stimulus which has the tendency to arrest or retard the movement. On the basis of two different sensory mechanisms underlying this response, it may be intelligible that after fatigue of one of them the response emerges in a different form. Thus if after repeated application of a strong clamp the nociceptive elements drop out, only arrest will take place due to the pressure. This type of reaction has been recorded by Gray & Sand (1936). The nature of the rhythm as it ensues in a fresh preparation after the release of the pressure (Fig. 11) is suggestive of a participation of a nociceptive stimulus, while the retarding effect may involve deep pressure receptors, or proprioceptors responding to isometric conditions. A possibly related response has been recorded from Scyllium to the denervated tail of which a string was attached. A gradual increase of longitudinal tension applied by the string produced a gradual slowing down of the rhythm (Fig. 12), and an increase in the amplitude and force, until all movement was arrested. The most likely suggestion is that in this case it is the more isometric contraction which causes the alteration in the rhythm. This is surprising in view of the fact that von Holst (1935 b) has demonstrated that there is no evidence to show that isometric conditions as such alter the normal swimming rhythm ; von Holst produced mechanical rigidity over a considerable portion of the vertebral column of an eel, without markedly affecting either the swimming movements or the co-ordination between the anterior and the posterior end.

Von Hoist’s statements about the accelerated rhythm following a nociceptive stimulus do not seem altogether satisfactory. While in one experiment (1934) he suggests that it is a special nociceptive reflex bearing no relation to swimming, in another experiment (1935 b) he releases the swimming rhythm through a strong pinch to the pectoral fins in an otherwise de-afferentated tench. Although in the dogfish, after the application of a strong stimulus, the change from regular swimming to an accelerated rhythm may be very sudden, the response merges without a perceptible break back into the normal swimming rhythm (Fig. 13); this may be taken as an indication that the same nervous elements form an integral part of the mechanism in both cases.

Von Holst also claims that the rhythmic-automatic neurones are quite distinct from the motor-neurones, and that the latter are excitable through reflex stimulation, without the participation of the former. It might therefore be expected that if a reflex be released in an actively swimming dogfish the result may be one of strict superposition of an independent automatic rhythm to which an equally independent reflex response is added. The great regularity of the swimming movements in the spinal dogfish makes this preparation particularly suitable for a study of the effects of transitory stimuli on the subsequent performance. Through an appropriate slight touch any single beat can be augmented, accelerated, suppressed or reversed at any phase of the movement (Fig. 14). When the rhythm emerges again with the next regular beat—it can be of the same or nearly the same amplitude and frequency as prior to the stimulation—but in nearly every case which has been recorded it is out of phase with the rhythm as recorded before the stimulation was applied. The resetting of the rhythm after a very short and feeble stimulation shows how susceptible the rhythmic elements are to outside influence. At the same time there appears a definite break in the response whenever the rhythm emerges again after reflex modification, which may point to a certain degree of independence.

As has been shown above, the application of a localized stimulus may cause anything from a purely local response to a rapid spreading of a reflex posture over the body. If a stimulus be applied to an actively swimming dogfish, exhibiting a dynamic reflex posture, it will depend on the site of stimulation and the phase of the movement, whether the dynamic and the induced postures will be competitive, synergic, or to some extent out of phase. The nature of the response will also depend on the strength and duration of the stimulus. Thus if the induced posture is in competition with a given phase of the movement the result can be (a) a retarding effect, (b) temporary arrest (Fig. 14E), (c) reversal (Fig. 14D). If the two phases are synergic an early application causes augmentation and acceleration, often associated with rebound (Fig. 14B) ; if, on the other hand, the stimulus is extended until the phase of movement reaches the turning point, augmentation may occur without acceleration and without rebound, and provided the stimulus extends equally in duration and strength over two phases of a swimming stroke, the frequency of the rhythm may remain unaltered (Fig. 14C). The significance of these phenomena will be referred to later. A feeble, localized stimulus does not necessarily impress a posture over the entire length of the body. If it is applied to an actively swimming dogfish, only the posture nearest to the stimulus will be modified for a short period. After that period, when the effect of the stimulus has faded out, the two parts of the body, those affected by the stimulus and the unaffected ones, reintegrate to form a new swimming wave. The break in the tracing appears to indicate that this process is rather sudden, probably because the effect of a feeble stimulus wears off rapidly (see Fig. 1).

It appears that the fin reflexes in the dogfish have not, hitherto, been subjected to anything but very casual examination. Apart from observations on the clasper reflex, it is stated that the pectoral fins act as inclined planes (Bethe, 1899; le Mare, 1936), that the dorsal fins are bent away from any stimulus applied near the base (le Mare, 1936; ten Cate, 1934). However, the responses of the fins are not quite as monotonous as one might gather from these descriptions, certainly not in Acanthias, and probably not in Scyllium. There are indications of not very well developed, but nevertheless distinct, biphasic or rhythmic responses to a single exteroceptive stimulus.

The responses of the two dorsal fins in Acanthias are essentially similar, and have been recorded by attaching a recording lever to the posterior margin of the fin. The nature of the response is determined by the site of stimulation; the receptive fields extend over a large part of the body, and partly overlap. If the stimulus is feeble the response is restricted to one or both dorsal fins; if the strength is increased, movements of the trunk may be associated with it. Stimulation along the base of either dorsal fin causes two distinct types of reaction, depending on the level of application. Anteriorly to the fin a slight touch with a camel-hair brush or straw causes a bending of the fin away from the side of stimulation; this response may be sustained. Posteriorly it causes a very rapid flick towards the side of stimulation, after which the fin returns to its initial position (Fig. 15). If the latter. stimulus is more sustained, the response may be either repeated or reversed, and followed by a beat of the body musculature. The response of the fin itself may, however, be more sustained, and in such cases a quivering motion can be noticed on the posterior margin of the fin (Fig. 15 B). This quick response is always more sustained when the fin engages the straw or brush which causes the stimulus. It seems reasonable to suppose that this is due to reinforcement produced by more isometric conditions, analogous with similar reactions in the tetrapod limb. The quivering of the fin margin can also be observed when the stimulus be applied anteriorly, causing a bending to the contralateral side.

It has been suggested that the movements of the fins during active locomotion may be governed by the proprioceptors of the trunk. An alternative theory suggests that ‘their movement may be governed by a similar or the same internal automatic rhythm which regulates the swimming rhythm of the trunk’ (le Mare, 1936). In either case the influence of the proprioceptors, or of the supposed internal rhythm can be certainly overruled by local reflexes, because the reflexes described here can be evoked in any phase of the swimming movements, and then appear superimposed on the swimming rhythm (Fig. 15 C–F).

It does not appear difficult to assess the probable biological significance of these two reflexes. If a light object is placed posteriorly within the action radius of a dorsal fin, it is removed by a quick beat. The receptive field of this reflex, however, extends considerably beyond the action radius.

If, on the other hand, an object touches the back of an actively swimming dogfish anteriorly, and towards one side of the dorsal fin, it is likely to be swept backwards and engage the fin ; consequently the fin is bent to the contralateral side.

The range of movement of the pectoral fins in the dogfish is undoubtedly limited ; the fins can be elevated and depressed, abducted and adducted, and to a small extent ‘they are capable of rotation. Touch near the base of the anterior border of a pectoral fin causes elevation and adduction (retraction) of the anterior margin and depression of the posterior lobe (rotatory movement). In fresh preparations this response may be repeated up to five times. From a receptive field extending from the posterior border of the pectoral fin to the level just beyond the first dorsal fin, an elevation of the posterior lobe can be produced. The responses of the two pectoral fins can be identical or opposed according to the site of stimulation. Thus both pectoral fins are elevated posteriorly when the back of the animal is touched; they are both depressed in response to ventral touch. This response also occurs when one of the fins and the adjoining segments are de-afferentated. If the ventral stimulus is moved towards the posterior border of one of the fins, depression changes to ipsilateral elevation, while the contralateral fin is still being depressed. All these responses can probably be interpreted as leading to an avoiding reaction through changing the course of swimming of the animal. On the other hand, if in Scyllium a finger is placed near the posterior angle of insertion between the body and the fin, the fin is first elevated and then adducted, the placoid scales scraping noticeably against the finger; this response may be repeated several times. It seems legitimate to suppose that this reaction is a primitive precursor of the scratch or cleaning reflex.

The numerous recent endeavours which try to explain swimming movements of fish, and locomotory movements in general, on the basis of an essentially central mechanism, have led to a complicated interpretation, which does not always appear plausible. It is difficult to visualize the reasons why the medulla, the rhythmic activities of which in the isolated state have been related to the swimming rhythm, should exert a steady and continuous effect on other elements in the spinal cord which are termed the ‘automatic-rhythmic cells’, and which in turn are said to affect rhythmically the ‘motor-cells’ (von Holst, 1934–9). No conclusive evidence appears to have been produced to show the existence of an independent automatic-rhythmic process on to which reflex responses may be superimposed through direct excitation of the motor-cells. As long as the animal is at rest neither the strength of any locomotory reflex nor the reflex time have been demonstrated to alter rhythmically in such a way that it might appear reasonable to ascribe such changes to a concealed, continuous, inherent rhythm in the central nervous system playing a ‘predetermined score’ (Weiss, 1941).

It is difficult not to ascribe to reflexes a very important role in swimming, although there is overwhelming evidence to show that the undulatory posture evoked through stimulation does not spread over the body of a fish as a reflex chain from segment to segment. The formation of the undulatory posture with primary and secondary regions of contraction must be considered essentially as one reflex; it is determined by the stimulated receptive field, and the central and peripheral pathways. In this process the spinal cord acts as an entity.

All this, however, does not exclude the possibility that the normal passage of the undulatory posture (swimming) is brought about by a temporal chain of processes, which also involves proprioceptive excitation, and may therefore be considered as the integration of two distinct types of reflex. Little is known of the exact mode of interaction, but the work of Fessard & Sand (1937) indicates the probable occurrence of proprioceptive reflexes during active locomotion in fish. Once the undulatory posture has been established it is, therefore, most likely that the pattern of the spatial arrangement of these impulses corresponds in some way to the pattern of the existing primary and secondary contractions which form the undulatory posture.

I wish to express my thanks to Prof. J. Gray, F.R.S., for the interest he has taken throughout the course of this work. The experiments were performed while holding a Cambridge University Table at the Laboratory of the Marine Biological Association in Plymouth, and the Marine Station, Millport, of the Scottish Marine Biological Association. I am indebted to the Directors and their staffs for hospitality and for considerable assistance.