1. The medulla of the tench brain was searched systematically by means of needle electrodes for rhythmic bursts of action potential activity coinciding with the breathing movements.

  2. The neurones which produced these rhythmic bursts of activity were located in the grey matter, mainly beneath the IXth and Xth motor nuclei and in the region round the Vllth motor nucleus. This type of activity was also found in some of the neurones forming the Vth and Vllth motor nuclei.

  3. The respiratory neurones were not arranged in a discrete and homogenous nucleus anywhere in the medulla, but were scattered through the grey matter. The distribution was not uniform, the neurones tending to occur in very small groups. There was also a relatively higher density of respiratory neurones in the central, as compared with the more anterior and posterior, parts of the respiratory region. The possibility that variations may occur in the constitution of the respiratory centre, in different individuals and in the same individual at different times, is considered.

  4. The manner in which neurones of the respiratory centre function to produce the rhythmic activity is discussed. Localized destruction of active respiratory regions, over a wide area of the medulla in different fish, was never followed by a breakdown in the rhythmic movements. This is interpreted as evidence against the existence of a pacemaker and favouring the hypothesis that the rhythm is produced by a general reciprocal interaction of large numbers of respiratory neurones.

In a previous paper (Shelton, 1959) the effects of brain transections on the breathing movements of the tench were described. These experiments showed that a region of the medulla oblongata, extending from the posterior border of the facial lobe up to the level where the Vth and Vllth cranial nerves emerge, had to be intact for normal breathing movements to be produced. In the experiments to be described in this paper, the medulla was explored with needle electrodes in order to determine the distribution of neurones whose activity could be correlated with the breathing movements of the fish. Although they did not make an exhaustive search, Woldring & Dirken (1951) recorded volleys of spike potentials in rhythm with the movements of respiration in the medulla of the carp. They attributed this activity to two strips of tissue to the right and left of the midline at the anterior border of the facial lobe, and suggested that the discharges came from the motor neurones of the Vllth, IXth and Xth cranial nerves. Similar respiratory volleys have been recorded by Hukuhara & Okada (1956) in the carp and the catfish. These volleys were apparently obtained from a single electrode site in both species of fish. Hukuhara & Okada went on to show that the rhythmic discharges produced by the neurones at the active region continued when the medulla was isolated by cutting all the cranial nerves and transecting the midbrain and spinal cord. No work described so far has involved a detailed examination of the fish medulla, and information about the arrangement of respiratory neurones is very limited.

Since the pioneering work of Gesell,Bricker & Magee (1936), many workers have used recording electrodes to explore the medulla of mammalian preparations for respiratory neurones (Dirken & Woldring, 1951 ; Amoroso, Bainbridge, Bell, Lawn & Rosenberg, 1951; von Baumgarten, von Baumgarten & Schaefer, 1957; Haber, Kohn, Ngai, Holaday & Wang, 1957; Salmoiraghi & Burns, 1960). In spite of the considerable volume of work that has been done, using stimulation and transection as well as recording techniques, there is not complete agreement about the site and arrangement of respiratory cells within the mammalian brain stem. Several of the differences are undoubtedly due to characteristics of the technique used. Some features of the recording technique (particularly vis-à-vis that of electrical stimulation) have emerged from the work on mammals and the ones relevant to the present work will be considered briefly. The recording electrode detects activity in a very localized region around the non-insulated tip and so is very selective, locating active units with considerable accuracy. Respiratory neurones are hard to find with recording electrodes and so are probably loosely scattered through the medulla. This dispersion of individual respiratory units is not revealed to the same extent by stimulation techniques and this is due, in part, to the spread of stimulus around the tip of the electrode. The spread will vary according to the stimulus parameters but inevitably a considerable number of units must be excited at any locus and the chances of affecting some part of even a dispersed system will be quite high. The indiscriminate excitation of a considerable number of neurones, probably involved in the co-ordination of different physiological activities, is an undesirable property of the technique which may complicate interpretation of the results (Amoroso et al. 1951 ; Liljestrand, 1953). Another reason for the recording technique showing a less widespread distribution of respiratory neurones is that some elements may not be detected by this method even though their stimulation may cause modification of the breathing movements. It is generally agreed that fine needle electrodes of the type usually used will detect activity from the neighbourhood of cell bodies and not from nerve fibres. Consequently, impulses in afferent and efferent pathways would not be recorded, whereas stimulation of these areas via similar electrodes would, presumably, be effective. This argument may be extended to include neurones which may operate in respiratory co-ordination only when the breathing is much modified, or even to generalized reticular units which never operate solely in respiratory coordination (Liljestrand, 1953). The stimulation method could include such units as part of the general respiratory system, although they would not be detected by recording techniques in the normally breathing animal.

Though all the techniques used in locating the neurones of the respiratory centre give valuable information, the detection of active sites by means of recording electrodes seems potentially more accurate and selective than other methods. It probably gives little or no information about activity in fibre pathways but this is hardly a serious limitation. A more significant fault of the method is that only those neurones which produce bursts of activity in rhythm with the breathing movements can be defined as participating in the co-ordination of these movements. It is possible that neurones producing different types of activity, such as a continuous discharge, may have a significant part to play in the generation and regulation of the respiratory rhythm. Though it is possible to think of ways in which the respiratory function of this hypothetical type of neurone could be detected, it is clearly not possible to test every active neurone found in the medulla for such a function. In the present paper, therefore, only those neurones which are active in rhythmic bursts are considered to be part of the respiratory centre.

Tench from 45 to 70 g. in weight were deeply anaesthetized in 0·5–1·0% urethane solution, and a hole was cut in the top of the skull over the region of the medulla in which the search was to be made. The fish were fixed in the holder described previously (Shelton, 1959) and then were allowed to recover to a lighter level of anaesthesia which was maintained for the duration of the experiment. The urethane concentration for this was about 0·2 %, although there was considerable individual variation and the level was arrived at by trial and error. With very light anaesthesia, rhythmic swimming movements were produced by the fish and occasionally the trunk and breathing rhythms were synchronized. This synchrony has been described in the goldfish by von Holst, and attributed to the rhythmical activity of a single automatic system in the medulla or spinal cord (von Holst, 1934a, b). In the present work complications due to the spread and synchronization of rhythmic activity were undesirable and so the concentration of urethane was increased to the point where the swimming movements became irregular or ceased altogether. During an experiment the fish were allowed to breath normally in about 1000 ml. of water which was continuously aerated. The water-level in the experimental tank was adjusted so that it came just below the hole in the skull, thus avoiding any problems of protecting the brain from osmotic and ionic stresses. In general, no physiological saline was necessary as the brain was bathed in a pool of body fluid which accumulated in the well formed by the cranium. Survival of the animals during the experiments was good and, after successful preliminary operations with little haemorrhage, there were no signs of deterioration. The experiments were done at room temperature.

The medulla was searched systematically using fine needle electrodes held in a manipulator which permitted calibrated movement in three planes. The electrodes were made from 25 μ diameter platinum wire which was insulated except at the tip by means of varnish or glass. The glass insulation gave a more durable electrode and was most frequently used, the over-all diameter of a unipolar electrode of this type being 40− 60 μ. Bipolar electrodes were tried in some experiments, but localization of activity, as determined by changes in the discharge pattern produced by slight movements of the electrode, appeared to be the same with both types. The majority of the experiments were done with the simpler unipolar electrode. The signals were amplified in a Grass P4 pre-amplifier and displayed on one beam of a double-beam Cossor oscillograph. A loudspeaker unit was used to monitor the activity picked up. The breathing movements of the fish were displayed on the other beam of the oscillograph by means of a simple mechano-electric transducer attached to the operculum.

The electrodes were inserted perpendicularly into the brain, the distance of any recording site being measured from fixed reference points on the surface of the brain by means of the calibrated slides of the manipulator. These measurements on the three co-ordinates of the manipulator gave an approximate idea of the position of the electrode tip in the medulla and the loci from which respiratory activity was obtained were confirmed later when the brain was sectioned. The brains were fixed in 10 % formol in Young’s teleost Ringer; sections were cut at 12− 15 μ and stained in Heidenhain’s iron haematoxylin. The electrode tracks were usually discernible in these sections and allowed identification of the active region. In many experiments the electrode was lowered further into the brain after an active site had been found, and in these cases the depth measurement on the manipulator had to be used for location. Measurements of the brain before fixation and after sectioning were used to correct the depth readings for the shrinkage which occurred. An electro-cautery device was also used in experiments on fifteen of the fish examined. The cauterizing current was applied through the electrode when this was in an active region and so produced a small burn in the brain tissue. A difficulty with this method was that the size of the burn varied in different electrode sites even though the output of the cautery was kept constant. In some cases there was spread of the burn back up the stem of the electrode. The breathing movements of the fish used in the cautery experiments were examined carefully for any effects caused by the destruction of active respiratory regions. In general, the methods described above made it possible to assign activity to particular regions within the medulla, although it is a little difficult to decide with what accuracy. Errors of 0·2–0·3 mm. may have been made in the location of some sites but in the majority of cases, where electrode tracks could be followed, the error was less than this.

Some anatomical studies were made on sections of the tench brain which had been sectioned transversely or longitudinally and stained by the silver techniques of Holmes (1947) or Romanes (1950).

The experiments were performed on forty-four tench, exploration of the brain being carried out from the level of the obex to the front of the cerebellum. In all the experiments, a total of 470 insertions of the electrodes was made from the dorsal surface of the brain. Several different types of activity were found, the most common being discharges which were not obviously related to any activity of the animal or to any stimulus that it was receiving. Spike discharges were found frequently in the vestibular nuclei and the crista cerebellis when vibrational stimuli of various kinds were given. Injury discharges were recognizable by their short duration and high frequency and great care had to be exercised in lowering the electrode through the nervous tissue if this type of discharge was to be avoided. In eighty-nine of the 470 electrode insertions made, it was possible to detect potential discharges, in rhythm with the breathing movements, from some point on the track of the electrode through the brain. In the fish, where the principal musculature involved in the breathing movements is that of the head region, there is clearly a danger that rhythmic activity of this sort detected in the brain may be due to movement artifacts. Movements of the brain with respect to the electrode tip could cause a periodic injury discharge in an otherwise inactive neurone or could bring a continuously discharging neurone within range of the recording system during a certain phase of the breathing cycle. Every precaution was taken to avoid such artifacts ; the skull was held immobile in the clamp and the discharges were examined for characteristics, such as a gradual increase in a unit’s size or the eventual loss of a unit in a typical injury discharge, which might be expected if they were due to a movement of the brain.

Localization of the respiratory neurones

The regions on the dorsal surface of the medulla in which successful insertions of the electrodes were made are plotted in Fig. 1. On the diagram there is clearly a higher density of active sites around the posterior border of the cerebellum and the front of the facial lobe than there is elsewhere. This is due in part to a relatively greater number of electrode insertions being made in this region during the searches, but this is not the complete explanation. If the successful insertions are expressed as percentages of the total number of insertions made in particular regions then differences are still found. The dotted line in Fig. 1 delimits the proposed respiratory area of the medulla within which the rhythmic activity was found. The area was divided transversely into eleven regions at the levels shown by the lines A to L. In the regions between lines D to F, over 30 % of the electrode insertions were successful in detecting respiratory activity, whereas anterior and posterior to these regions the successes fell to below 20 %. It seems likely therefore that the respiratory neurones are more densely arranged in this central area. This is not to suggest that even here large groups of respiratory neurones occur, forming a more or less discrete nucleus. At no point was it possible to make an electrode insertion with the certain knowledge that respiratory activity would be detected.

Fig. 1.

Dorsal view of the tench medulla showing the positions of successful electrode insertions. The broken line delimits the proposed respiratory area and the percentages of electrode insertions which were successful at various levels within this area are indicated.

Fig. 1.

Dorsal view of the tench medulla showing the positions of successful electrode insertions. The broken line delimits the proposed respiratory area and the percentages of electrode insertions which were successful at various levels within this area are indicated.

In Fig. 2 the loci within the medulla from which the respiratory discharges were obtained are plotted on a series of twelve cross-sections. These sections are taken at the levels A to L in Fig. 1. In the more posterior regions (sections A to C) the respiratory activity was detected very largely in the scattered reticular cells below the Xth motor nucleus. It is thought that none of the rhythmic discharges was obtained from neurones actually within the motor nuclei of the IXth and Xth cranial nerves, though it is difficult to be absolutely certain of this. The majority of active sites are certainly too deep to be associated with these nuclei, but one or two borderline sites occur where the motor nuclei are merging into the undifferentiated parts of the grey matter. At a somewhat higher level (section D) respiratory activity was quite definitely picked up from motor neurones as well as from reticular cells. The IXth and Xth motor nuclei, together with the posterior motor nucleus of the Vllth cranial nerve, form a continuous visceral efferent column extending from behind the obex almost to the point of emergence of the IXth cranial nerve (Kappers, Huber & Crosby, 1936). Rhythmic activity was detected only in those motor neurones situated at the front end of this column, which were therefore almost certainly part of the posterior facial nucleus. Kappers et al. suggest that this part of the facial motor nucleus is concerned with the innervation of some gill apparatus muscles such as the levator and adductor operculi. These muscles play an important part in the breathing movements of the tench (Shelton, unpublished). Respiratory discharges also occurred in the anterior facial motor nucleus (sections E, F and G) as well as in the reticular cells which in this region forms a fairly discrete nucleus situated ventrolaterally to the fasciculus longitudinalis posterior. Active units were found scattered in the grey matter at still higher levels, though the frequency of their occurrence decreased. The most anterior region from which respiratory discharges were obtained was the motor nucleus of the Vth cranial nerve (sections K and L).

Fig. 2.

For legend see opposite. Transverse sections through the tench medulla at the levels A to L shown on Fig. 2. The regions in which bursts of activity coincided with the breathing movements are plotted ; squares indicate activity coinciding with the opening (expansion) phase of the breathing cycle, circles activity coinciding with the closing (contraction) phase. Key to lettering:

a.c.n. auditory (VIIIth) cranial nerve

a.f.n. anterior facial (VIIth) motor nucleus

a.l.l. anterior component of the lateral line nerve

cer. cerebellum

c.c. crista cerebellis

d.t. descending root of trigeminal (Vth) nerve

f.l. facial lobe

f.l.p. fasciculous longitudinalis posterior

g.c.n. glossopharyngeal (IXth) cranial nerve

g-l- glossopharyngeal (IXth) lobe.

g.n. glossopharyngeal (IXth) motor nucleus

g.t. secondary gustatory tract.

m.f.c.n. motor component of the facial (VIIth) cranial nerve

o.l. optic lobe

p.f.n. posterior facial (VIIth) motor nucleus

p.l.l. posterior component of the lateral line nerve

r.n. reticular nucleus

s.f.c.n. sensory component of the facial (VIIth) cranial nerve

t.n. trigeminal (Vth) motor nucleus

v.c.n. vagal (Xth) cranial nerve

ves.n. vestibular nuclei

v.l. vagal lobe

v.n. vagal (Xth) motor nucleus

Fig. 2.

For legend see opposite. Transverse sections through the tench medulla at the levels A to L shown on Fig. 2. The regions in which bursts of activity coincided with the breathing movements are plotted ; squares indicate activity coinciding with the opening (expansion) phase of the breathing cycle, circles activity coinciding with the closing (contraction) phase. Key to lettering:

a.c.n. auditory (VIIIth) cranial nerve

a.f.n. anterior facial (VIIth) motor nucleus

a.l.l. anterior component of the lateral line nerve

cer. cerebellum

c.c. crista cerebellis

d.t. descending root of trigeminal (Vth) nerve

f.l. facial lobe

f.l.p. fasciculous longitudinalis posterior

g.c.n. glossopharyngeal (IXth) cranial nerve

g-l- glossopharyngeal (IXth) lobe.

g.n. glossopharyngeal (IXth) motor nucleus

g.t. secondary gustatory tract.

m.f.c.n. motor component of the facial (VIIth) cranial nerve

o.l. optic lobe

p.f.n. posterior facial (VIIth) motor nucleus

p.l.l. posterior component of the lateral line nerve

r.n. reticular nucleus

s.f.c.n. sensory component of the facial (VIIth) cranial nerve

t.n. trigeminal (Vth) motor nucleus

v.c.n. vagal (Xth) cranial nerve

ves.n. vestibular nuclei

v.l. vagal lobe

v.n. vagal (Xth) motor nucleus

Destruction of active sites

In experiments on fifteen tench, a number of active sites were destroyed by cautery after they had been detected by the recording apparatus. The average number of sites destroyed was just over three in each fish, and the highest number in any individual was eight. The destruction was on a very small scale therefore, and the conclusions which can be drawn are limited. The searches during these cautery experiments were arranged so that the regions destroyed were distributed over the whole respiratory area. In none of the fish were the breathing movements stopped completely as a result of the destruction of either a single or several sites. However, it was quite common for the movements to stop immediately a region was destroyed and for 2 or 3 min. to be taken for recovery. In many cases the pattern of breathing was different upon resumption although the changes were small ones such as slight variations in the breathing frequency. On other occasions, the effect was on some component of the pumping mechanism rather than on the breathing rhythm as a whole. Such things as the failure of one opercular flap to close effectively and distortion of the mouth during the complete breathing cycle were observed. As might be expected, some of these local failures were due to destruction of parts of the Vth or Vllth motor nuclei, although similar effects were obtained by destruction of other regions. On two occasions, for example, it was found that damage in the grey matter beneath the Xth motor nucleus caused some malfunction in the operculum of the same side.

Characteristics of the activity from respiratory neurones

In spite of the difficulty experienced in locating them, the respiratory neurones were not usually dispersed to the extent of occurring singly in the nervous tissue of the medulla. When an active site was located it was common to find that the discharges consisted of more than one unit (Figs. 3, 4) and that slight movements of the electrode at the active site would bring in different rhythmically firing units. In Fig. 3,a there are records obtained from points 0 ·05 mm. apart in nervous tissue below the anterior facial motor nucleus, between it and the reticular nucleus. Lowering the electrode through the brain by this distance brought in a new unit towards the end of the original discharge which was then detected less definitely. A similar effect is seen in the records of Fig. 3 b, obtained from the trigeminal motor nucleus. In this case the unit introduced by lowering the electrode by 0·08 mm. fired approximately in the intervals between bursts from the original unit.

Fig. 3.

Respiratory activity. The effect of small movements of the electrode on the discharge pattern, (a) Activity from the reticular formation near the Vllth motor nucleus. The second record was obtained from a point 0·05 mm. lower than the first. (b) Activity from the Vth motor nucleus. The second and third records were obtained from points 0·03 and 0·08 mm. lower than the first (down on movement trace = operculum closing).

Fig. 3.

Respiratory activity. The effect of small movements of the electrode on the discharge pattern, (a) Activity from the reticular formation near the Vllth motor nucleus. The second record was obtained from a point 0·05 mm. lower than the first. (b) Activity from the Vth motor nucleus. The second and third records were obtained from points 0·03 and 0·08 mm. lower than the first (down on movement trace = operculum closing).

Fig. 4.

Respiratory activity. Discharges showing phase, frequency and duration differences obtained from similar regions of the reticular formation near the anterior facial motor nucleus in different animals. Up on movement traces = operculum closing.

Fig. 4.

Respiratory activity. Discharges showing phase, frequency and duration differences obtained from similar regions of the reticular formation near the anterior facial motor nucleus in different animals. Up on movement traces = operculum closing.

Though the rhythmic pattern of activity from any one respiratory neurone remained fairly constant for long periods, there were considerable differences in the patterns obtained from different neurones. The duration of the rhythmic bursts in relation to a complete respiratory cycle, the frequency of the spike discharges within the bursts, and the phase of the respiratory cycle in which the burst occurred were all quite variable from neurone to neurone. Differences occurred in these features of activity obtained when the electrodes were in similar regions of the medulla in different animals or even in regions very close to one another in the same animal. The two records of Fig. 4 were obtained from similar regions in the reticular formation of different animals and show duration, frequency and phase differences.

In the mammal the expiration and inspiration phases of the respiratory cycle are easily distinguished, and there is no difficulty in deciding whether activity in a respiratory neurone coincides with one or the other phase. The movements of the fish which correspond to the alternate expirations and inspirations of the mammal are the contraction and expansion of the whole pumping mechanism. A complete separation of these antagonistic phases of the respiratory cycle is not possible in the fish because the pump is a dual mechanism with the action of the buccal component slightly preceding that of the opercular one during normal activity (Hughes & Shelton, 1958). Consequently there are periods, which can occupy up to two-fifths of a complete respiratory cycle in the tench, when one part of the pumping apparatus is in one phase while the other part is in the opposite phase. Despite this complication it has been possible to make a fairly satisfactory distinction between opening and closing phase neurones in the majority of cases, and the regions in which these occurred have been plotted in Fig. 2 as squares and circles respectively. Inevitably there were a few phase-spanning discharges which were difficult to classify, and the two general types which occurred can be seen in Fig. 4. The large unit in Fig. 4,a was active for a small proportion of a complete cycle and occurred at a time when opercular expansion and buccal contraction were going on simultaneously. The discharge in Fig. 4,b was produced over a much larger proportion of the cycle so that it covered the change of phase (in this case from contraction to expansion) of both buccal and opercular components. These intermediate types of activity were classified in all cases with reference to the buccal component of the pumping mechanism. As Fig. 2 shows, a higher proportion of the respiratory neurones found were active during the closing phase. No segregation of neurones associated with the two antagonistic phases of the respiratory cycle could be detected.

The experiments described show that rhythmic bursts of action potential activity, at the same frequency as the breathing movements, can be detected by means of needle electrodes over a wide area of the tench medulla. In agreement with the conclusions of previous workers who have used electrodes with fairly large tip diameters (e.g. Salmoiraghi & Burns, 1960; Cohen & Wang, 1959), it seems likely that the activity was detected from nerve cell bodies and not from fibres. In many cases the electrode could be moved for distances of more than 50 μ (Fig. 3) without losing a particular unit completely. Activity coming from a nerve fibre would probably be much more sharply localized. Moreover, cell bodies could always be seen on histological examination of regions where activity was picked up. The area of the medulla occupied by the respiratory neurones corresponds with that which has been shown by transection experiments to be necessary for the co-ordination of normal breathing movements (Shelton, 1959). The rhythmic bursts of activity came from cells situated beneath the IXth and Xth motor nuclei and in the neighbourhood of the Vllth motor nucleus. Some of the motor neurones forming the Vth and Vllth nuclei were also found to be rhythmically active, though this was to be expected as the main muscles of the pumping mechanism are innervated by the Vth and Vllth cranial nerves. These motor neurones forming the final common path to respiratory effectors, like the spinal motor neurones innervating the thoracic musculature of the mammal, may be regarded as being outside the integrating respiratory centre in the strict sense, but there seems to be little advantage in making such a distinction. Respiratory integration may go on right up to the level of the motor neurone ; and schemes, such as most of those proposed for the mammal, which divide off the motor neurone and surrounding regions from the rest of the integrating apparatus, may be over-simplifications (Liljestrand, 1953).

The failure to detect respiratory neurones with many of the electrode insertions made in the respiratory area suggests that the fish respiratory centre, like that of the mammal, consists of a system of neurones dispersed through the grey matter of the medulla. The absence of any dense nuclei of respiratory neurones has been confirmed by searching the areas around the sites of successful electrode insertions. When several insertions were made within a radius of about 300 μ of a successful site it was quite common to find no respiratory discharges at all, even though multiunit discharges occurred at the active region. It seems likely from this that the neurones are not evenly scattered through the respiratory area, but are arranged in very small groups ; in this way it is possible to reconcile the high degree of localization of the active sites with the multi-unit discharges usually obtained from them. There is also lack of uniformity in the respiratory neurone distribution in a more general sense. The experiments have shown there to be a higher density of active sites in the region beneath the posterior border of the cerebellum and the anterior border of the facial lobe than there is elsewhere in the respiratory area.

These proposals on the arrangement and distribution of the neurones, which together make up the respiratory centre, are the outcome of results obtained from experiments on a large number of fish. The points plotted in Figs. 1 and 2 represent active sites found in all these experiments and give no indication of any individual variation which may exist. Because of the low probability of finding active units in a general search, and the time-consuming nature of the searching process which prevented an examination of large areas of medulla in any one animal, no data on individual variations are available from the experiments. It is possible that the boundaries of the respiratory centre, and the arrangement of the respiratory neurones within them, are the same in every individual and that the whole system is anatomically quite stable at all times. There is nothing in the results to suggest that this must be so; it is equally possible that the respiratory centre is labile and that variations in the distribution and density of active units occur in different animals, or in the same animal at different times. There is some evidence to support the view that the centre is not a rigidly fixed system of inter-connected neurones in a single individual. Von Baumgarten (1956) has shown that, in the mammal, the breathing of gas mixtures containing high tensions of oxygen causes an increase in activity in existing expiratory neurones and the recruitment of many new neurones of this type. Eventually the expiratory neurones become dominant in the respiratory centre, the reverse of the normal situation when air is breathed. Similarly, Burns & Salmoiraghi (1960) have found that an increase in the tension of carbon dioxide in the respired air causes the recruitment of more rhythmically active units, in addition to increasing the discharge frequency in all the respiratory cells. It seems likely that the balance of respiratory neurones may be constantly changing by the recruitment of some units and the dropping out of others, as different conditions are encountered. The recording and transection experiments done so far on the fish centre have been under conditions of complete aeration of the water in which the animals were breathing. It would now be interesting to know what effects are produced in the general organization of the centre by changes in the gas tensions of the medium.

The way in which activity of a rhythmic nature is generated within the neurones of the respiratory centre is not clear. Adrian & Buytendijk (1931) suggested that the respiratory centre of the goldfish was capable of producing normal rhythmic activity when the brain was completely isolated. This proposal has since received the support of Hukuhara & Okada (1956) following their work on the isolated brain of the catfish and carp. The property of inherent rhythmicity, persisting in complete isolation, has been interpreted as evidence for the existence of a pacemaker within the centre. A pacemaker system is one in which rhythmic activity invades the whole centre from one or a few intrinsically rhythmic neurones. A great deal of the work on the mammalian centre has led many workers to reject the pacemaker hypothesis. The more generally accepted view is that the respiratory rhythm is generated by a widespread interaction of large numbers of neurones, none of which is intrinsically rhythmic in isolation (Wyss, 1954). The effects of local destruction of neurones within the tench centre suggest that here, too, the rhythm is not due to the action of a pacemaker, for in no case did the destruction of an active site stop rhythmic breathing. On the other hand, the transection experiments show that removal of the posterior part only of the proposed respiratory centre is effective in stopping the breathing movements. It is possible to reconcile this fact with the hypothesis of a rhythm due to reciprocal interaction between large numbers of neurones, if it is supposed that interaction breaks down when a given proportion of the population is removed. The evidence for either hypothesis is not compelling and a more intensive study is required of the inter-relationships of the neurones which have been described here as constituting the respiratory centre.

It is a pleasure to thank Dr G. M. Hughes for discussion and encouragement. I also wish to thank the Department of Scientific and Industrial Research for financial support during the period when most of the work was done.

Adrian
,
E. D.
&
Butendijk
,
F. J. J.
(
1931
).
Potential changes in the isolated brainstem of the goldfish
.
J. Physiol
,
71
,
121
35
.
Amoroso
,
E. C.
,
Bainbridge
,
J. C.
,
Bell
,
F. R.
,
Lawn
,
A. M.
&
Rosenberg
,
H.
(
1951
).
Central respiratory spike potentials
.
Nature, Lond
.,
167
,
603
4
.
Von Baumgarten
,
R.
(
1956
).
Koordinationsformen einzelner Ganglienzellen der rhombencephalen Atemzentren
.
Pflüg. Arch. ges. Physiol
.
262
,
573
94
.
Von Baumgarten
,
R.
,
Von Baumgarten
,
A.
&
Schaefer
,
K.-P.
(
1957
).
Beitrag zur Lokalisations-frage bulboreticulärer respiratorischer Neurone der Katze
.
Pflüg. Arch. ges. Physiol
.
264
,
217
27
.
Burns
,
B. D.
&
Salmoiraghi
,
G. C.
(
1960
).
Repetitive firing of respiratory neurones during their burst activity
.
J. Neurophysiol
.
23
,
27
46
.
Cohen
,
M. I.
&
Wang
,
S. C.
(
1959
).
Respiratory neuronal activity in pons of cat
.
J. Neurophysiol
.
22
,
33
50
.
Dirken
,
M. N. J.
&
Woldring
,
S.
(
1951
).
Unit activity in bulbar respiratory centre
.
J. Neurophysiol
.
14
,
211
26
.
Gesell
,
R.
,
Bricker
,
J.
&
Magee
,
C.
(
1936
).
Structural and functional organization of the central mechanism controlling breathing
.
Amer. J. Physiol
.
117
,
423
52
.
Haber
,
E.
,
Kohn
,
K. W.
,
Ngai
,
S. H.
,
Holaday
,
D. A.
&
Wang
,
S. C.
(
1957
).
Localisation of spontaneous respiratory neuronal activities in the medulla oblongata of the cat ; a new location of the expiratory centre
.
Amer. J. Physiol
.
190
,
350
5
.
Holmes
,
W.
(
1947
).
In Recent Advances in Clinical Pathology
, ed.
Dyke
,
C. S.
London
:
Churchill
.
Von Holst
,
E.
(
1934a
).
Reflex und Rhythmus im Goldfischriickenmark
.
Zool. Anz. (Suppl.)
,
7
,
93
5
.
Von Holst
,
E.
(
1934b
).
Studien über Reflexe und Rhythmen beim Goldfisch
.
Z. vergl. Physiol
.
20
,
582
99
.
Hughes
,
G. M.
&
Shelton
,
G.
(
1958
).
The mechanism of gill ventilation in three freshwater teleosts
.
J. Exp. Biol
.
35
,
807
23
.
Hukuhara
,
T.
&
Okada
,
H.
(
1956
).
On the automaticity of the respiratory centres of the cat fish and the crucian carp
.
Jap. J. Physiol
.
6
,
313
20
.
Kappers
,
C. U. A.
,
Huber
,
G. C.
&
Crosby
,
E. C.
(
1936
).
The Comparative Anatomy of the Nervous System of Vertebrates, including Man
.
New York
:
Macmillan
.
Liljestrand
,
A.
(
1953
).
Respiratory reactions elicited from the medulla oblongata of the cat
.
Acta physiol, scand
.
29
(
Suppl. 106
),
321
93
.
Romanes
,
G. J.
(
1950
).
The staining of nerve fibres in paraffin sections with silver
.
J. Anat., Lond
.,
84
,
104
10
.
Salmoiraghi
,
G. C.
, &
Burns
,
B. D.
(
1960
).
Localisation and patterns of discharge of respiratory neurones in brain-stem of cat
.
J. Neurophysiol
.
23
,
2
13
.
Shelton
,
G.
(
1959
).
The respiratory centre in the tench (Tinca tinca L.). I. The effects of brain transection on respiration
.
J. Exp. Biol
.
36
,
191
202
.
Woldring
,
S.
&
Dirken
,
M. N. J.
(
1951
).
Unit activity in the medulla oblongata of fishes
.
J. Exp. Biol
.
28
,
218
20
.
Wyss
,
O. A. M.
(
1954
).
Respiratory centre and reflex control of breathing. I. The mode of functioning of the respiratory centre
.
Helv. physiol, acta
, (
Suppl. 10
),
5
25
.