Hearing in Terrestrial Urodeles: A Vibration-sensitive Mechanism in the Ear

It has often been suggested that urodele amphibians are able to ‘hear’ vibrations in the ground (Kingsbury & Reed, 1909; Noble, 1931; Monath, 1965). Unlike the anurans, they possess no external eardrum. Their middle ear consists of a movable cartilaginous operculum, inserted in an opening in the ear capsule, and an opercular muscle, which connects the operculum to the supra-scapula. These structures are thought to allow vibrations, received from the ground via the forelimbs, to reach the inner ear (see references above). Although both the anatomy of the middle ear, and its variations in different families, have been thoroughly investigated (Hilton, 1950; Olson, 1966; also references above), experimental evidence for a vibration-detection system is lacking. The present study considers the function of the inner ear and its surrounding structures in such a system.

It has been suggested that in both urodeles and anurans two small regions of the labyrinth are important in the detection of vibratory stimuli ; these are the amphibian papilla and the basilar papilla (Harrison, 1902; de Bur let, 1935; van Bergeijk & Witschi, 1957). The receptor cells in these regions are of the type found throughout the vertebrate acoustico-lateralis systems, namely hair cells. The effective stimulus to hair cells is believed to be a displacement of the ‘hair’ relative to the cell body (Harris & van Bergeijk, 1962). Thus if the urodele ear is important in the detection of vibratory stimuli, such stimuli must cause displacements of parts of the ear relative to other parts. These displacements must in turn be capable of stimulating hair cells. The present study examines displacements of the fluid in the inner ear.

Periodic mechanical disturbances transmitted through the substrate will be referred to as substrate vibrations, rather than ‘sound’. Little is known about the substrate vibrations to which these animals are normally exposed. In an extended solid medium vibrations are transmitted by both longitudinal compression waves and transverse shearing waves. Their properties will be affected by the presence of a free surface and the heterogenous composition of the natural substrate, including variable amounts of water. Thus in the present study it has been necessary to consider the effects of vibrations perpendicular to the surface of the substrate (vertical vibrations) and parallel to the surface (horizontal vibrations). The term hearing is used to mean the detection (without necessarily the location) of a vibrating source of mechanical energy, not in actual contact with the animal, by means of receptors in the inner ear.

Adults of two salamandrid species were used. These were Salamandra salamandra, the European fire salamandra, and Triturus cristatus, the great crested newt. Animals were kept in terraria with access to water. Both species spent the majority of the time on land. The anatomy of the ear region is very similar in the two species, and since Triturus was the more easily obtainable, it was mainly used for initial experiments ; these were later repeated with Salamandra.

After ether anaesthesia half of the head was removed by a median vertical longitudinal cut from the snout to the first cervical vertebra, and by a transverse vertical cut through the neck to the mid-line, joining the first cut. The brain tissue was carefully removed from the remaining half of the cranium, exposing the cranial wall around the ear capsule. The perilymph foramen was exposed by removing the overlying membranes. It was important to minimize bleeding and the escape of white granules from the endolymph sac in order to prevent blockage of the perilymph foramen and to ensure visual access to its opening.

The ear capsule was opened by drilling and chipping away the bone forming its dorsal wall. The labyrinth was detached from the capsule wall by severing connective tissue, endolymph and perilymph ducts, and the branches of the auditory nerve, with a fine tungsten needle. Removal of the labyrinth from the capsule then exposed the capsular opening of the perilymph foramen. The perilymph duct was fairly easily blocked by the retraction of its severed end into the foramen, by blood clots and by the accumulation of cellular debris, bone dust or small air bubbles. Preparations were judged satisfactory when slight movements of the head-neck joint caused large displacements of fluid through the perilymph foramen.

Suspended carmine particles or red blood cells were used to indicate displacements of the fluid in the ear. These particles were observed through a binocular dissecting microscope equipped with an eyepiece micrometer, usually at a magnification of × 60. Light from a small high-intensity stroboscopic lamp was focused on to the preparation through a biconvex lens. A difference of about 1 Hz between the discharge frequency of the lamp and the frequency of the vibrations under test enabled displacement amplitudes to be measured under the microscope. The displacement amplitudes of the platform and of the skull of the animal were measured by the same technique.

That region of the resting animal normally in contact with the substrate will be referred to as the contact area. On land this consists of the abdomen posterior to the pectoral girdle, the cloacal region, the tail, the fore and hind feet, and sometimes includes the ventral part of the pectoral girdle. The head is only rarely in contact with the substrate. To simulate natural conditions as far as possible, vibrations were presented only to the contact area by means of a suitable platform (see Figs. 1).

To avoid undue distortion of the input waveform, the mass of the platform was kept at a minimum. It was constructed from Perspex, and had approximately the shape of the animal’s contact area. Its mass was further reduced by partially drilling through from the undersurface at closely spaced intervals. It was supported by light coil springs, and could be coupled to an electro-magnetic vibrator (Goodman’s Model V 47) for either vertical or horizontal vibrations. The animal’s head rested on a heavy steel block to which it could be fixed by tightening a clamping screw. The ventral part of the pectoral girdle and the neck lay over the gap between the platform and the steel block. The vibrator was activated by a sine-wave oscillator and power amplifier; transistorized battery-powered apparatus was used in order to avoid line-frequency hum, since this lay in the frequency range under test.

In the design of the apparatus the assumption was made that for fairly low frequencies vibrations in all parts of the substrate in contact with the animal are in phase. Although the velocity of propagation of vibrations in the natural substrate is not known, the wavelengths at low frequencies are likely to be long compared to the dimensions of the animal (the contact area of a large Salamandra is approximately 10 cm in length). The effect of small phase differences over the contact area was not considered important in the present study.

The general structure of the ear region of Salamandra has been described by Francis (1934). The adult ear capsules are composed of bone, and lie at the posterior margin of the skull on either side of the cranial cavity. The mesial wall of the capsule, separating the cavity of the capsule from the cranial cavity, contains five openings; three of these contain the branches of the auditory nerve, one the perilymph duct, and one the endolymph duct. The capsule has a sixth opening, the fenestra vestibuli, in its postero-lateral wall. The operculum, a roughly oval piece of cartilage, lies in this opening, attached to its edges by an elastic membrane. The operculum is connected to the supra-scapula of the corresponding half of the pectoral girdle by a muscle, the opercularis (Francis, 1934; this muscle is also known as the levator scapulae, e.g. Monath, 1965).

The membranous labyrinths of both Triturus and Salamandra have been described by Retzius (1881), and that of Salamandra by Francis (1934). Since the anatomy is very similar in the two species, the following description will apply to both.^*

The labyrinth consists of three semicircular canals, a utriculus, a sacculus and a lagena, and two smaller parts, the amphibian papilla and the basilar papilla (Figs. 2). In each of these eight parts there is a sensory macula containing hair cells; each macula is supplied by a branch of the auditory nerve. The labyrinth contains a watery fluid, the endolymph. External to the labyrinth, but also inside the ear capsule, is connective tissue and another fluid-filled system, the perilymphatic system. The largest part of this system, the perilymph space, lies lateral to the sacculus; it is bounded mesially by the thin lateral wall of the sacculus, and postero-laterally by the operculum. A duct, the perilymph duct, arises from the dorso-lateral region of the perilymph space and follows a complex course around the posterior end of the labyrinth until it passes out of the capsule through the perilymph foramen (Figs. 3).

Both the endolymphatic and perilymphatic systems extend into the cranial cavity. The endolymph duct arises from the dorso-mesial wall of the sacculus and runs dorso-medially until it passes through the endolymph foramen, which perforates the wall between the cranial and ear cavities. Inside the cranial cavity it expands into the endolymph sac which lies over the dorsal and lateral surfaces of the brain. The endolymph sac and the endolymph duct are packed with white granules identical in appearance with the otolith granules of the sacculus. The perilymph duct, after passing through the perilymph foramen, expands into a small perilymph sac in the ventro-lateral corner of the cranial cavity close to the perilymph foramen.

In cranial preparations large fluid movements were observed in the perilymph foramen when the operculum was moved with a pair of fine forceps. Movements of fluid in a direction out of the ear capsule were seen when the operculum was displaced in a direction into the capsule. The displacement amplitude of the fluid seemed in all cases considerably greater than that of the operculum. Movements of fluid through the endolymph foramen and the acoustic foramina were very small in comparison with the movement through the perilymph foramen.

Movements of the fluid in the perilymph foramen were also observed in capsule preparations; these accompanied slight movements of the head about the neck joints, or slight pressure on the membrane overlying the foramen magnum. A downward flexing of the head or an inward pressure on the membrane caused a movement of fluid from the cranial cavity into the ear capsule. As in the cranial preparation, little movement was observed through the other foramina. In a preparation in which the capsule had been opened, but the labyrinth left in situ, a displacement of bone particles (formed by the drilling procedure) showed that there was a movement of fluid into the perilymph space via the perilymph duct when a positive pressure was applied to the membrane overlying the foramen magnum.

In cranial preparations the movements of suspended particles showed that 100 Hz sinusoidal vibrations delivered to the contact area were accompanied by 100 Hz sinusoidal displacements of the fluid in the perilymph foramen. In different preparations the amplitude of these displacements varied from 40 μ for a platform amplitude of 30 μ to more than 75 μ for a platform amplitude of less than 10 μ. The vibratory amplitude of the skull varied between one-half and one-third that of the platform. Similar results were obtained for vertical vibrations and antero-posterior horizontal vibrations. No obvious displacements were seen in the other foramina. The effect of lateral horizontal vibrations was not examined.

The accuracy of measurements was limited by several factors. At high amplitude the observed particle disappeared inside the foramen over part of the displacement cycle. At low amplitudes the microscope did not provide adequate magnification. In different preparations, and in the same preparation with time, there was a variation in the position of the observed particle relative to the walls of the foramen. However, in all cases the displacement amplitudes of suspended particles were greater than those of both the skull and the platform. A comparison of the displacements at different frequencies was also subject to the above difficulties, but displacements of the perilymph in the foramen were found to have greater amplitudes than those of the platform from 50 to 200 Hz, and greater amplitudes than those of the skull from 50 to 400 Hz. At frequencies above 400 Hz, displacements were too small to be studied by this technique ; frequencies below 50 Hz were not studied because of inadequacies of the signal generating equipment.

In capsule preparations suspended particles were observed from the ear side of the perilymph foramen. Their movements showed that vibrations delivered to the contact area resulted in high-amplitude displacements of the fluid in the foramen, as in the cranial preparation. At an input frequency of 100 Hz and amplitude of 30 μ displacements of suspended particles varied between 75 and 100 μ in many preparations; corresponding skull amplitudes were approximately 20 μ. No obvious displacements could be detected in the other foramina. Measurements at different frequencies showed a decrease in the displacement amplitude of the fluid from between 120 and 160 μ at 50 Hz to between 15 and 40 μ at 250 Hz for a constant input amplitude of 30μ. Over this frequency range the amplitude of skull vibrations also decreased, from 20μ at 50 Hz to less than 7μ at 250 Hz for a platform amplitude of 30μ.

It has been suggested by Pumphrey (1950) that the pars neglecta of the labyrinth, a sensory region without otoliths, may be stimulated by fluid displacements resulting from the movement of the saccular otolith in response to vibrations of the head. That the fluid displacements observed in cranial preparations of these two urodeles were not caused by movements of the mass of otolith granules in the sacculus was shown by the following experiment.

Cranial preparations which showed high-amplitude fluid displacements in response to low-amplitude inputs were selected. The remaining half of the head was removed from the body by cutting transversely at the level of the first vertebra, leaving the ear region intact. The vibrator was then attached to the skull with a small clamp, and the skull was vibrated in the three principal axes at a frequency of 100 Hz and an amplitude of 30 μ. No displacements of the perilymph relative to the walls of the foramen could be detected, although any slight mechanical pressure in the region of the operculum caused violent displacements of the fluid through the foramen, showing that the system remained operational.

Further evidence was produced by clamping the half-head of a normal preparation to the steel platform. Although this reduced the amplitude of skull vibrations sufficiently that they could no longer be detected visually, there was no obvious change in the amplitude of the fluid displacements.

Similarly, it was found that the fluid displacements originating in the cranial cavity, and observed in the ear side of the foramen, were independent of movements of the head. During the experiments described previously for this preparation observations were made when the head of the animal rested on the non-vibrating platform, when it was suspended by a limp rubber band, and when it was clamped rigidly to the steel block. In the three situations the observed fluid displacements were of similar value, and greater than those of the input vibration.

Whatever the function of the operculum may be, it is capable of movement in and out of the fenestra vestibuli. Even though movements of the operculum will result in pressure waves propagated in the fluid of the ear and in displacements of that fluid, only the displacements will be considered. The dimensions of a salamandrid ear are very small compared to the wavelengths of fairly low-frequency sound waves in an aqueous medium. At distances from a sound-source much less than one wavelength (the near-field), displacement effects are dominant over pressure effects (van Bergeijk, 1964). Moreover, labyrinthine receptor cells are displacement detectors rather than pressure detectors (Harris & van Bergeijk, 1962).

Since the walls of the ear capsule are composed of rigid bone, movements of the operculum will be accompanied by a displacement of fluid through at least one of the openings in the capsule. The operculum is slightly greater in diameter than the fenestra vestibuli, and its area is considerably greater than that of the elastic membrane connecting it to the capsule. A ‘leaking’ of fluid round the edges of the operculum (or a ‘bulging’ of the elastic membrane) is therefore unlikely to account for a significant proportion of this displacement. Thus the displacement must occur between the ear capsule and the cranial cavity.

Results described above indicated that fluid displaced between these two cavities passes almost entirely through the perilymph foramen. This conclusion is supported by certain anatomical features of the ear. The perilymph foramen has approximately six times the cross-sectional area of the endolymph foramen, and contains only the perilymph duct. The endolymph foramen contains connective tissue and small blood vessels as well as the endolymph duct. The endolymph duct is much narrower than the perilymph duct (compare Pl. 1 and 2A). In addition, the endolymph duct is packed in vivo with otolith-like granules, which might increase frictional forces opposing volume displacements of the endolymph. Displacements of fluid through the acoustic foramina must be largely prevented by the branches of the auditory nerve, which fill the openings. For these reasons, it seems likely that the least resistance to displacements of fluid between the ear capsule and the cranial cavity is presented by the perilymph system.

Results of experiments using both gross mechanical stimuli and sinusoidal vibrations showed that from 50 to 200 Hz the displacements of fluid in the perilymph foramen have an appreciably greater amplitude than those of the original stimulus. A partial explanation for this result was found by further study of the structures involved.

The operculum may be considered as a piston A capable of displacing fluid in a hydraulic system (Figs. 4). Movements of this piston will cause volume changes in the ear capsule, chamber E; these volume changes will be accompanied by the displacement of fluid through the duct BC (the perilymph foramen). Chamber F represents the cranial cavity, and piston D represents openings in the cranial cavity other than those into the ear capsule. The ratio of the displacement amplitudes of the fluid in BC to those of piston A (the gain of the system) equals the ratio of the cross-sectional area of A to that of C.

The movement of the operculum differs in one respect from that of the piston in the model. It was found to have a far greater range of movement at its ventro-lateral edge than at its dorso-mesial edge; it moves as if it were hinged. The greatest displacements of the operculum will therefore occur at the ventro-lateral edge. If the operculum moved as suggested in the model, the resultant volume change in the ear capsule would be da, where d is the displacement and a is the cross-sectional area of the operculum. Since one edge is effectively fixed, the actual volume displaced is approximately $12$da, when the maximum displacement d occurs at the ‘free’ edge. Thus the effective piston area is one half the actual area.

In Triturus the operculum is roughly oval, and typically measures 1·0 by 0·75 mm. Its effective area is thus somewhat less than $12$×1·0×0·75 mm.², that is approximately 0·35 mm.². The dimensions of the oval perilymph foramen are typically 0·2 by 0·18 mm., giving a cross-sectional area of about 0·035 mm.². Thus the system possesses a hydraulic gain of approximately 10. A similar calculation using the dimensions of the ear of Salamandra showed a gain of approximately 12.

In the proposed hydraulic system (Figs. 4) the duct BC opens into a second chamber F, the cranial cavity. Although piston D, unlike piston A, does not represent a discrete structure, it might be expected that pressure changes in the vicinity of C would also cause high-amplitude displacements in BC. It is of interest therefore to note that in capsule preparations high-amplitude displacements were observed in the perilymph foramen.

The forces exerted by the fluid in the perilymph duct (BC in Figs. 4) must be considerably less than those exerted by the operculum or its equivalent in the cranial cavity. This was shown by the ease with which the system was blocked by blood clots, cellular debris and air bubbles. It is therefore necessary to consider the effects of opening one of the chambers and removing the tissues adjacent to the perilymph foramen.

If in vivo the perilymph sac has no openings other than that into the ear capsule, displacements of fluid through the foramen will be accompanied by changes in the sac volume. Such changes may be accomplished either by a change in shape of the sac, or by the stretching of its walls. An accurate estimate of the extent of the sac could not be obtained by histological techniques, because of distortion during fixation and dehydration. However, during exposure of the foramen as described for cranial preparations blood cells were often released into the tissue adjacent to the sac. The blood cells could be moved by gentle squashing of this tissue, but did not appear to enter the sac. The limits of their movement indicated that the sac had a diameter at least three times that of the foramen, and therefore a cross-sectional area at least nine times as great. Thus the elasticity of its walls might be expected to offer comparatively little resistance to perilymph displacements. Similarly, the cross-sectional area of the openings into the cranial cavity, represented by piston F (Figs. 4), is much greater than that of the foramen. Opening the cranial cavity will therefore have little effect on the response of the system to opercular movements. A similar argument applies to opening the capsule in the capsule preparation.

Vibratory displacements of the perilymph may be of importance to the animal if they stimulate a sensory receptor. Vertebrate labyrinthine receptors are of two types : those in which the gelatinous material surrounding the sensory ‘hairs’ contains dense calcareous otoliths, and those which merely possess a non-loaded cupula or membrane. Receptors equipped with otoliths are capable of responding to gravitational forces, linear and angular accelerations, as well as to displacements of the fluid in their vicinity (Pumphrey, 1950). Labyrinthine receptors without otoliths can only respond to displacements of the fluid in their vicinity. It would seem likely that a receptor, functional in the detection of fluid displacements, would not be loaded with otolith granules, since these would increase the inertia of the system.

There are two sensory maculae with associated gelatinous structures, but without otoliths, close to the perilymph foramen in these urodeles. They are the amphibian papilla and basilar papilla. The basilar papilla of the frog is thought to be the detector of vibrations from the external ear transmitted through the fluid of the inner ear (van Bergeijk & Witschi, 1957). In the newt and salamander this region is very small. A comparison of the basilar and amphibian papillae (Pl. 1 and 2b) suggests that the amphibian papilla is of much greater importance in these urodeles ; it alone is considered here.

A section through the amphibian papilla, together with the perilymph foramen, is seen in Pl. 1. It is a tubular evagination of the dorso-mesial wall of the sacculus projecting mesially and slightly ventrally. Its opening into the sacculus is elliptical, the long axis lying in the antero-posterior axis of the animal. In the ventro-mesial wall is an extremely thin membrane which separates the lumen of the papilla from a short wide diverticulum of the perilymph duct. This membrane therefore separates the endolymph in the papilla from the perilymph in the duct. The whole region is surrounded by dense connective tissue. A diagrammatic reconstruction of this region is shown in Figs. 5.

In the dorsal wall of the papilla is the sensory macula, containing numerous hair cells, and innervated by a small branch of the auditory nerve. The hairs of the hair cells are imbedded in a gelatinous structure, which has been described as a ‘tectorial membrane’ (Patterson, 1960; Cordier, 1964). Pending further information about its structure and function the term ‘gelatinous cupula’ will be used in order to avoid the functional implications of the former term.

Two striking features of the amphibian papilla are its close proximity to the perilymph foramen, and the thinness of the membrane separating the endolymph from the perilymph. This membrane is one of the three regions of the labyrinth described by Harrison (1902) as ‘tympanal areas’. These regions occur in the lateral wall of the sacculus between the sacculus and the perilymph space adjacent to the inner surface of the operculum (Figs. 3), and in the walls of the basilar papilla and the amphibian papilla. It is through these areas that Harrison suggests that vibrations of the perilymph are transmitted to the endolymph. It is felt that the use of the word ‘tympanal’ is not justified, since it suggests a stretched membrane specifically for receiving airborne sound, similar to the tympanal membrane of other vertebrates. These regions will therefore be referred to merely as the membranes of the respective areas.

Harrison describes the membranes of the amphibian papilla as a thin membrane ‘stretched… across a rigid frame’. This view is not supported by the following observation. Tearing the lateral wall of the sacculus in a freshly removed labyrinth enabled the otolith granules to be washed out with a fine jet of Ringer’s solution. This exposed the aperture of the amphibian papilla. The labyrinth was placed mesial side down in a watchglass containing Ringer’s solution. The fluid under the mesial wall of the sacculus could be displaced by slight pressure with a fine needle; under this treatment the membrane of the papilla was seen to move violently in and out, although no other part of the labyrinth was appreciably affected. It was concluded that this membrane is, as its very thin cross-section would suggest, a highly flexible structure presenting the minimum resistance to small movements of the fluid between the papilla and the perilymph duct, and not, as Harrison implies, a membrane under tension.

The structure and location of the amphibian papilla strongly suggest that it is the detector for the fluid displacements resulting from substrate vibrations. Except for the perilymph duct and a pathway through the amphibian papilla and its membrane the region around the perilymph foramen is occupied by dense connective tissue; thus displacements of fluid between the ear capsule and the cranial cavity via the perilymph foramen must pass mainly along these two routes (Figs. 6). The membrane of the papilla will limit the volume of fluid that can move through the papilla, but will provide little resistance to small displacements. The perilymph duct, on the other hand, would be capable of dealing with large volumes of fluid, but, by virtue of its length (Figs. 3), would present a fairly high resistance to high-velocity displacements.

Since the cross-sectional dimensions of the papilla and the perilymph duct are in the same order of magnitude as (although somewhat smaller than) those of the foramen, the amplitude of fluid displacements in these two pathways must be roughly equal to those in the foramen. These two regions are in the high-amplitude part of the system, represented by duct BC in the model (Figs. 4). In the perilymph space and the sacculus displacements would be of small amplitude, similar to those of the operculum; these parts of the system are in the low-amplitude, high-pressure part of the system, represented in the model (Figs. 4) by the chamber E. Thus the thin membrane of the sacculus and the contents of the sacculus would offer a negligible resistance to such displacements. Furthermore, the high-amplitude pathway through the amphibian papilla to the perilymph foramen is very much shorter than the pathway through the perilymph duct, although it has similar cross-sectional dimensions. It may therefore be expected to present less total resistance to fluid displacements, and thus the higher the frequency of such displacements the greater will be the proportion of the volume of fluid displaced that passes through this pathway.

De Burlet (1935), like Harrison (1902), suggests that ‘vibrations’ of the perilymph reach the amphibian papilla from the operculum; however, these ‘vibrations’ are thought to arrive via the perilymph duct rather than through the sacculus. For the reasons outlined above, this latter pathway seems unlikely. The displacements of fluid through the perilymph duct must be in phase with those through the sacculus and papilla, progressing (for an inward movement of the operculum) toward the perilymph foramen (Figs. 7), rather than 180°out of phase as would be the case in de Burlet’s system.

Displacements in the papilla would occur in a direction along its axis from the aperture to the membrane, resulting in a bending of the gelatinous cupula. Hair cells in other parts of the vertebrate acoustico-lateralis system have been shown to respond to displacements of their ‘hairs’ in a direction parallel to the free surface of the hair cells (Dijkgraaf, 1963), and if all hair cells are activated by such stimuli the displacements described above should be adequate stimuli for the macula of the papilla. Evidence drawn from the morphological orientation of the hair bundles in this macula will be presented elsewhere (Mullinger and Smith) ; these studies have shown that the hair bundles are oriented so that their maximum sensitivity undoubtedly lies in this axis of the papilla.

The biological advantage of such a system is apparent. Any substrate vibrations producing vibrations of the head could be detected by an otolith system such as is present in the sacculus. The sensitivity of this mechanism would be limited by the displacement amplitudes of the head. On the other hand, the amphibian papilla is in a position to detect much larger displacements of fluid resulting from the same substrate vibrations, and would therefore provide the animal with a vibration-detection system of a much greater sensitivity.

Unlike the mammalian auditory system, which embodies many features that isolate the cochlear receptors from vibrations in the body (von Bekesy, 1962), the present system appears to be highly sensitive to such vibrations. In many capsule preparations the fluid in the perilymph foramen underwent displacements coinciding with the heart beat; these were presumably caused by pulse waves in nearby cranial blood vessels. It is possible that such displacements occur in the intact animal; in addition, normal activities such as feeding and locomotion must produce movements of the operculum and resultant displacements of fluid between the ear and cranium. All high-amplitude components of these displacements are likely to have frequencies considerably lower than 50 Hz, and it is possible that the perilymph duct provides a pathway for them. Thus the perilymph duct may act as a shunt pathway, enabling gross low-frequency displacements of perilymph to bypass the amphibian papilla. This may serve a dual purpose of protecting the papilla and of helping to separate low-frequency vibrations resulting from activity from higher-frequency vibrations of possible biological significance originating in the substrate.

I should like to thank Dr H. W. Lissmann for suggesting the problem, Dr D. A. Parry for his encouragement during the work, Mr D. M. Unwin for his invaluable technical advice, Dr D. G. Butler for much helpful criticism of the present manuscript, and the Science Research Council for financial support.