]>HEARES3476S0378-5955(00)00098-810.1016/S0378-5955(00)00098-8Elsevier Science B.V.Fig. 1ABR responses to various intensities of 2 kHz logon bone conducted stimuli delivered to the intact skull, to the skull in the presence of a craniotomy (bone) and directly on the exposed brain. The amplitude calibration bars represent 0.5 μV. Thr.: threshold.Fig. 2The instantaneous acceleration response of skull bone and of brain (craniotomy) to the 2 kHz logon bone conducted stimuli (at 120 dB pe SPL) applied to the skull in the presence of a craniotomy (bone) and directly to the brain (craniotomy). Note that no acceleration could be picked up from the skull when the stimulus was applied on the brain (craniotomy), while clear vibrations of the brain (craniotomy) were recorded in response to bone vibrator-induced vibrations of the skull.Table 1The mean (±S.D.) thresholds, latencies and amplitudes of wave 1 of the ABR to bone vibrator stimuli (logons at 2 and 8 kHz) applied to the intact skull, to skull bone after a small craniotomy had been made and directly to the exposed brain, through the craniotomyLogon 2 kHzIntact skullCraniotomyBoneBrainΔ(bone−brain)Threshold X7084.693.2−8.64±S.D.510.49.08.09P<0.01Latency W1 X1.241.521.380.14±S.D.0.050.080.060.07P<0.001Amplitude W1 X4.962.552.200.35±S.D.1.641.311.590.93NSLogon 8kHzThreshold X7082.786.8−4.09±S.D.4.56.17.27.01NSLatency W1 X1.131.261.190.07±S.D.0.050.110.070.10P<0.05Amplitude W1 X5.023.283.080.21±S.D.1.251.060.740.56NSAlso shown are the differences between these values on skull bone and on brain, along with their statistical (paired t-test) evaluation. ABR wave 1 latencies and amplitudes are to a stimulus intensity (setting on the EP system) of 120 dB pe SPL. There was no difference in ABR threshold to air conducted (earphone) stimulation between pre- and post-craniotomy states.Bone conduction experiments in animals – evidence for a non-osseous mechanismSharonFreemanaJean-YvesSichelbHaimSohmera*sohmer@md2.huji.ac.ilaDepartment of Physiology, Hebrew University, Hadassah Medical School, P.O. Box 12272, Jerusalem 91120, IsraelbDepartment of Otolaryngology/Head and Neck Surgery, Hadassah University Hospital, Jerusalem, Israel*Corresponding author. Tel.: +972 (2) 6758385; Fax: +972 (2) 6439736AbstractBone conducted stimuli are used to differentiate between conductive and sensori-neural hearing loss. It has been thought that the main route for the transfer of vibratory energy from the point of application of the bone vibrator on the skull to the inner ear is completely osseous. An additional mechanism may play a prominent role. In rats, a bone vibrator was applied to the skull and also directly on the brain, after removing bone (a craniotomy), exposing the brain. Auditory nerve-brainstem evoked response (ABR) could be elicited not only with the vibrator on bone, but also with the vibrator directly on the brain. Similar results were obtained in guinea-pigs and fat sand rats. Noise masked this ABR. Extensive removal of skull bone did not alter the ABR to bone-conducted stimuli delivered to the exposed brain. Experimental elimination of the ossicular chain inertial mechanism and of the occlusion effect did not greatly alter the bone conduction response. A reduction in the fluid volume of the cranial cavity induced threshold elevations of the bone conducted ABR but not of the air conducted ABR. These findings can be interpreted as evidence that the ‘classical’ bone conduction mechanisms should be modified to include a major pathway for cochlear excitation which is non-osseous: when a bone vibrator is applied to the skull, the bone vibrations may induce audio-frequency sound pressures in the skull contents (brain and cerebro-spinal fluid) which are then communicated by fluid channels to the fluids of the inner ear.KeywordsBone conductionAuditory nerve-brainstem evoked responseVibratorSkullCerebro-spinal fluidCochlea1IntroductionBone conduction stimulation of the cochlea is a widely used clinical test to differentiate between a conductive hearing loss (where air conducted thresholds are elevated while bone conduction thresholds are normal) and a sensori-neural (inner ear-auditory nerve) hearing loss (both air- and bone-conducted thresholds are elevated to the same extent). Recent animal experiments conducted in this laboratory, along with a review of the literature have led us to the conclusion that even though determination of bone conduction thresholds is an integral part of the clinical test battery for the successful differentiation between conductive and sensori-neural hearing losses, the mechanism of bone conduction is quite complex and requires further clarification. This is particularly true with respect to the exact pathway of vibratory energy transmission from the point of application of the bone vibrator on the skull to the inner ear, which then leads to neural excitation.Major contributions to the mechanism of bone conduction have been made by von Bekesy (1932,1960), Barany (1938), Wever and Lawrence (1954), Kirikae (1959) and Tonndorf (1966,1968), Tonndorf and Tabor (1962). Bekesy and Wever and Lawrence maintain that the primary vibratory energy pathway from the point of contact of the bone vibrator on the skull to the ear is completely osseous; i.e. as transverse waves conducted along the skull bones. Tonndorf has added the possibility that a small amount of energy may be conducted as surface waves along the skin and soft tissues of the head and, in addition, that some energy may pass through the interior of the skull as pressure and/or translation waves and act directly on the cochlea through the so-called ‘third window’, e.g. the cochlear aqueduct (Tonndorf, 1966). Nevertheless Barany (1938) and Schneider (1959) concluded that the skull contents are not involved in bone conduction. These latter pathways have not been evaluated and are considered to be of minor importance.It is thought that the response of the inner ear following the skull and cochlear shell vibrations is initiated by three basic mechanisms, all based on osseous pathways. 1.According to the translatory (or inertial) mode of bone conduction, as the cochlear bony shell is vibrated, there is an inertial lag between cochlear shell vibration and that of the ossicular chain. Thus relative motion is set up between the cochlear shell and the ossicular chain. This mechanism has been emphasized by Barany (1938), by Kirikae (1959) and by Wever and Lawrence (1954). It is thought that this pathway is more important at lower frequencies (Kirikae, 1959; Tonndorf, 1966). An inertial component between the cochlear shell and the cochlear fluids has also been suggested. This induces alternating pressure changes across the basilar membrane, exciting the cochlea.2.In the compressional mode of bone conduction (modified by Tonndorf, 1968 – to the ‘distortional’ mode of bone conduction), the skull vibrations are propagated to the temporal bone and cause distortion of the bony cochlear shell. This causes fluid displacements in and out of the cochlear windows, with basilar membrane displacement and excitation. Kirikae (1959) and Tonndorf (1966) maintain that this is a major pathway at higher frequencies.3.In addition, particularly when the external auditory meatus is blocked, an additional, occlusion mechanism is present, whereby the vibrations of the skull radiate into the occluded meatus, producing aerial waves which then act on the tympanic membrane like any other aerial stimulus.Upon reaching the ear, these waves (surface, transverse and pressure) interact with each other. The final stimulus transferred to the cochlea is a vectorial integration of these pathways of wave energy transfer across and along the skull and its excitation will depend on the vectorial summation of all the modes of bone conduction, depending on their relative magnitudes and phases. It is possible that each of these mechanisms is more effective at different frequencies. The inner ear response is finally initiated by the same transduction mechanism as in air conduction (von Bekesy, 1932,1960). Thus, experimental modification of any one of the vibratory pathways from the point of contact of the stimulating vibrator on the skull to the ear will lead to alterations in cochlear excitation, either increase or decrease, depending on the relative phase and magnitude of the deleted component and hence in the final bone conduction response of the ear.A series of animal experiments conducted in this laboratory has provided evidence that an additional major pathway involved in the transmission of skull-induced vibrations to the inner ear is by a completely non-osseous route. In this mechanism, the skull vibrations (distortions) induce audio-frequency pressure variations in the brain/cerebro-spinal fluid (CSF) which are then transmitted by the fluid communications between the CSF and the inner ear fluids directly to the inner ear. A similar conclusion has been reached from the results of experiments in humans and these are reported separately (Sohmer et al., 2000).2Materials and methodsThese experiments were conducted on rats, guinea pigs and fat sand rats (Psammomys obesus), anesthetized with intraperitoneal injections of pentobarbital (rats and sand rats: 60 mg/kg; guinea pigs: 45 mg/kg). Body temperature was maintained (36.5–37.5°C) throughout the experiment. All experiments were conducted in accordance with the guidelines published by the Hebrew University–Hadassah Medical School Animal Care and Use Committee.The auditory stimuli used consisted of either air conducted (earphone or the bone vibrator suspended in the air over the skull, but not in contact with it) or bone conducted, alternating polarity clicks, or alternating polarity logons at frequencies of 2 and 8 kHz. The different frequency stimuli were used in order to study the possible frequency dependence of bone conduction mechanisms, as suggested by Tonndorf (1966). The bone-conducted stimuli were delivered by means of a standard clinical bone vibrator (Radioear B-71) to which a flat-headed screw was attached with acrylic dental cement in order to reduce the area of contact between the bone vibrator and the small head of the animal. The skin of the scalp was excised, and the scalp muscles were retracted. Thus there was no possibility of transmission of vibrations as surface waves along the skin and soft tissues of the head. In the first three experiments, a 90 g weight was attached to the bone vibrator and it was applied directly to the skull, suspended so that its pressure on the skull was that of its own weight (26 g) and the 90 g weight (116 g). This weight was too large for the experiments involving placement on exposed brain (craniotomies), so that in all the remaining experiments, the bone vibrator was applied to all sites, suspended with its own weight (There was no difference in threshold when the bone vibrator was applied to the skull with a force of 26 or 116 g). Thus in all experiments, the effective transmission of the vibrations to the head of the animals was uniform across animals and the transmission to the exposed brain through the craniotomy was also uniform across animals, though probably somewhat different from that to the skull. In most experiments, the external auditory canals were filled with plasticine in order to induce a conductive hearing loss so that air-conducted stimulation by the bone vibrator was severely reduced.The response of the cochlea to these stimuli under the various experimental conditions was evaluated by recording the auditory nerve-brainstem evoked response (ABR) with needle (Grass) electrodes in the skin, somewhat caudal to the vertex, with respect to the chin. The ground electrode was in a forepaw. Special attention was paid to the threshold of the ABR, defined as the lowest intensity stimulus, from a maximum setting of 120 dB pe SPL to which an ABR response could be recorded (in 5 dB steps). A setting of 120 dB pe SPL on the evoked response system used delivers a bone-conducted click stimulus 40 dB above threshold in rats (Geal-Dor et al., 1993). The latency and amplitude of the first wave of the ABR at 120 dB pe SPL were also evaluated. This wave represents the compound action potential of those auditory nerve fibers synchronously activated by the stimulus. The auditory stimuli and the ABR responses were generated and analyzed by a Microshev 4000 evoked response system.Experiment A: comparison between ABR thresholds elicited with bone vibrator on intact skull, on the skull in the presence of a small craniotomy and directly on the exposed brain in the region of the craniotomy. These experiments were conducted on 11 rats. In each, ABR threshold and the latency and amplitude of the first ABR wave (120 dB pe SPL) were first determined in response to the bone vibrator (using 2 and 8 kHz logon stimuli) applied directly on the intact skull, slightly rostral to the bregma (the intersection of the four main sutures, perpendicular to each other, on the rat skull). A small craniotomy (approximately 1 cm in diameter) was then made caudal to the bregma. Usually the dura was torn during this procedure and there was a loss of some CSF. The ABR threshold, wave 1 amplitude and latency to the bone vibrator on the skull, in the same region as before (caudal to the craniotomy), were again determined, in the presence of this craniotomy. Finally, the bone vibrator was applied directly to the exposed brain and the ABR measurements were repeated. Thresholds were alternatively and repeatedly measured with the bone vibrator on the skull and on the brain. Differences in ABR threshold, amplitude and latency across these experimental conditions were then statistically evaluated (paired t-tests). Similar experiments were conducted on two guinea pigs and on three fat sand rats. In one rat, a Bruel and Kjaer accelerometer (type 4393) was applied to the skull and used to measure the amplitude of the skull vibrations induced by the bone vibrator when it was applied to the skull and when it was applied to the exposed brain (craniotomy). This was repeated in a guinea pig and in addition, the accelerometer was also applied to the exposed brain (craniotomy) and the bone vibrator applied to the skull in order to measure the vibrations of the brain induced by vibration of the skull.Experiment B: the effect of the size of the craniotomy on ABR threshold, amplitude and latency to bone vibrator stimulation directly on the skull. These experiments were conducted on four rats and two guinea pigs. The bone vibrator was always applied to the skull – initially to the intact skull and then a small (2 mm) craniotomy was produced and ABR measurements were made with the bone vibrator applied to the skull. Then the craniotomy was enlarged further in one or two stages, and ABR repeated each time. ABR threshold, amplitude and latency were evaluated at each stage. In several rats, the bone vibrator was applied to the brain also in the presence of craniotomies of various sizes, including maximal removal of skull bone (calvarium).Experiment C: the effect of the presence of a craniotomy on ABR air-conducted thresholds. In eight rats, air-conducted ABR thresholds (in one rat using an earphone and in seven, using the bone vibrator suspended over the head of the rat, but not in contact with the skull) were determined in the absence of a craniotomy (intact skull) and in the presence of craniotomies. In the experiment with the earphone, the external canals were not blocked with plasticine. ABR threshold across these experimental conditions (presence and absence of a craniotomy) was evaluated.Experiment D: the effect of eliminating the ossicular chain inertial component and of the external meatus occlusion mechanism on ABR threshold, amplitude and latency. These experiments were conducted on six fat sand rats (P. obesus) since they have a unique middle-inner ear anatomy, including a large bulla cavity with a thin-walled inner ear so the cochlea and the three semicircular canals bulge into the middle ear cavity, facilitating experimental manipulations. Initially, ABR threshold to bone conduction stimulation was measured in the intact animal. The cochlea on one side was surgically destroyed and the experiment was conducted on the remaining intact ear. Cyanoacrylate glue was then applied to the middle ear cavity, immobilizing the stapes footplate, and cementing (immobilizing) the ossicular chain to the walls of the bulla cavity of the temporal bone. The external canal was filled with plasticine. ABR threshold, amplitude and latency to bone vibrator stimulation (click) on the skull in the intact ear were then compared to that in the same ear following the experimental manipulations of blocking the ossicular chain inertia and filling the external auditory meatus. Finally, a craniotomy was made and ABR responses were recorded with the bone vibrator on the skull and repeated with the bone vibrator directly on the brain. In four of the animals the middle ear fixation procedure was carried out prior to the craniotomy, while in the remaining two animals, it was performed after the craniotomy.Experiment E: the effect of attempts to reduce intracranial fluid volume on bone conduction thresholds. In seven rats with intact skulls (no craniotomy), ABR thresholds to air conducted and bone-conducted 2 kHz logon stimuli were recorded. The air conduction thresholds were recorded at the beginning (control) and at the end of the experiment. The bone conduction thresholds were recorded before (control) and several times after injecting i.p. a bolus of about 2 cc of a 25% (hypertonic) solution (2 g/kg) of mannitol. This induces a reduction of intracranial water content (of the brain and of the CSF) and is therefore used therapeutically to reduce intracranial pressure (Donato et al., 1994; Treib et al., 1998). If there was no effect on bone-conducted thresholds or if the effect was temporary (in most cases), additional boluses were administered, with 30 min between injections. In this way, there were 13 injections of mannitol in these seven rats.3ResultsExperiment A: Fig. 1 shows the ABR recordings in a typical rat in response to several intensities of bone-conducted stimuli (2 kHz logon), from a maximum instrument setting of 120 dB pe SPL, down to threshold. In the left column, the bone vibrator was placed on the (still) intact skull; in the middle column, the vibrator was on skull bone in the presence of a craniotomy and in the right column, the bone vibrator was placed directly on the exposed brain, through the craniotomy. As can be seen, clear ABR responses could be recorded when the vibrator was applied directly to the brain, through the craniotomy. These ABR responses could be masked by broad band noise presented by a speaker (air-conducted). The lowest threshold was obtained when the vibrator was on the intact skull.Table 1 shows the mean results obtained in the rats studied in this way. The lowest ABR thresholds, shortest latencies and largest amplitudes were obtained in rats when the bone vibrator was applied to the intact skull. This was true both for 2 and 8 kHz (low and high frequencies) logon stimuli. In the presence of a craniotomy, ABR thresholds were consistently elevated, latencies increased and amplitudes decreased. For the 2 kHz logon, the threshold was significantly higher when the bone vibrator was applied to the exposed brain compared to when it was on the bone. With respect to the 8 kHz logon, there was no significant difference between the thresholds (vibrator on bone or vibrator on brain, both in the presence of the craniotomy).In the guinea pigs and in the fat sand rats, clear ABR responses were also obtained in response to bone vibrator stimulation directly to the brain, through a craniotomy.An accelerometer placed on the skull was clearly able to record the vibrations induced by a bone vibrator applied to the skull about 15 mm distant both in the intact skull and in the presence of a craniotomy. However when the bone vibrator was applied directly to the exposed brain through the craniotomy, the accelerometer, at maximum sensitivity, was unable to record any vibrations of the skull. Identical results were obtained in a guinea pig and furthermore the accelerometer on the exposed brain recorded large vibrations that were induced by the application of the bone vibrator on the skull (see Fig. 2).Experiment B: the size of the craniotomy had a consistent effect on the ABR threshold to bone vibrator stimuli applied to the bone. As the craniotomy was successively enlarged, the bone vibrator on bone ABR thresholds became more and more elevated, reaching a maximum elevation of about 15 dB. However, the bone vibrator on brain ABR thresholds was not further elevated by the size of the craniotomy (starting with the minimal size of craniotomy required to place the vibrator on the brain without making contact with the surrounding bone – a diameter of about 1 cm).Experiment C: on the other hand, the size of the craniotomy did not affect the ABR threshold to air-delivered stimuli (earphone or bone vibrator suspended above the skull) with plasticine in the external auditory canal, showing that cochlear function was not compromised by the induction and presence of these craniotomies. Furthermore the mean air-conducted ABR threshold (±S.D.) in seven rats using the bone vibrator as an air-conducted sound source (suspended above the skull, with plasticine in the external auditory canal) was 115.0±7.1 dB pe SPL before the craniotomy and 113.9±8.1 dB pe SPL after the craniotomy (difference not significant) and significantly higher than those recorded with the bone vibrator on the skull or brain. This provides confirmation that the ABR thresholds obtained with the bone vibrator on bone or on brain were not in response to air-conducted sounds.Experiment D: in three sand rats, following contralateral destruction of the cochlea, the ossicles, the stapes footplate in the oval window and the bulla cavity of the opposite temporal bone were fixed into one solid mass (cyanoacrylate), without obstructing the round window. In these animals, the ABR threshold to stimulation of the intact skull was 5–10 dB better compared to that prior to fixation. The persistence of these responses in the presence of cyanoacrylate shows that this glue in the middle ear is not toxic to the ear, at least for the duration of the experiment. After the craniotomy, ABR could still be recorded in these animals in response to application of the bone vibrator to the skull and also directly to the brain, with thresholds on the brain 5–10 dB better than on the skull. Post mortem observation of the middle ears confirmed that the ossicles and middle ear cavity were one solid mass and the round window was not obstructed. In the remaining three animals, the round window was also obstructed by the glue, limiting its vibration. In these cases, ABR threshold with the bone vibrator on the brain was elevated by 25–30 dB compared with the threshold prior to middle ear ossicular and window immobilization.Experiment E: following the administration of mannitol to seven rats with intact skulls, the ABR threshold to air-conducted 2 kHz logon stimuli was temporarily lowered (improved) by 3.93±3.4 dB compared to the pre-mannitol state. This threshold improvement was statistically significant (P<0.025; paired t-test). At the same time, when the bone vibrator was applied to the skull, ABR thresholds became elevated following the mannitol injection compared to the baseline threshold in the same animals. In most cases, the bone conduction thresholds recovered spontaneously about 30 min after the mannitol bolus injection, so that multiple boluses could be given to the same animal. In no case did the bone conduction threshold improve following mannitol administration. The mean bone conduction threshold elevation was 7.5±3.82 dB compared to the baseline bone conduction threshold and this elevation was significant (P<0.005; paired t-test).4DiscussionThe major result of these experiments is that ABR responses to vibratory (bone conduction) stimuli can be elicited in rats, in guinea pigs and in fat sand rats when the bone vibrator is applied directly to the exposed brain, and not only with the bone vibrator on the skull. The waveform of the ABR was similar in these two conditions. The ABR recorded with the bone vibrator on the brain could be masked by white noise, presented by an earphone, confirming the cochlear origin of the responses.Since the accelerometer on the skull could not detect bone vibrations when the vibrator was directly on the brain in the craniotomy, it is not likely that direct vibration of the brain and its fluid induced vibrations of the skull. Furthermore, consideration of the impedances of the tissues makes it unlikely that brain-fluid vibrations would be able to induce vibrations of overlying skull, or of the bony wall of the cochlear shell. On the other hand, vibration stimuli delivered to the skull induced clear vibrations of the underlying brain.In these experiments (for example, Experiment A) vibrations were induced by placing the bone vibrator on skull bone and directly on the brain through a craniotomy. However the load impedances of these tissues (bone and brain) are not identical and it is possible that the accelerations of brain induced by the vibrator on the brain could be greater than those of bone induced by the vibrator on bone. These cannot be simply measured since the source impedance of these two tissues with respect to the measuring accelerometer also differ. In fact the acceleration measured on brain in response to vibration of brain was smaller than that of skull bone in response to vibration of skull bone (detailed results not presented). In any case, the magnitude of the skull bone vibrations induced by the bone vibrator on the skull was sufficient to give rise to large vibrations of the underlying brain. On the other hand, the magnitude of the vibrations induced in the overlying bone by the bone vibrator on the brain were below the sensitivity of the accelerometer (see Experiment A).In addition, the presence of a craniotomy and its size did not cause threshold elevations in the ABR to air-delivered stimulation, showing that the cochlea was not affected by the craniotomies. Also the extent of the craniotomy did not affect the ABR threshold when the bone vibrator was applied directly to the brain.Finally, ABR responses to the bone conduction stimulation could still be obtained even when the relevant classical mechanisms thought to be involved in transmission of vibratory energy from the point of application of the vibrator to the final cochlear ABR response had been eliminated by experimental manipulations:1.The bone vibrator was applied directly to the skull following retraction of skin and muscle from over the skull. In this way, transmission as surface waves along the soft tissues of the skull was eliminated.2.In addition, the external auditory meatus was completely filled with plasticine, so that the occlusion affect was greatly reduced, since the effective external ear cavity was then very small.3.Also the ossicular chain was fixed to the middle ear cavity of the temporal bone, preventing relative motion between the cochlear shell and the ossicular chain. This would eliminate the ossicular chain inertial component of bone conduction (Experiment D), which has been thought by Barany (1938), Wever and Lawrence (1954) and Kirikae (1959) to be the major mechanism of bone conduction. Stenfelt et al. (2000), based on studies on a dry human skull, also suggest that ossicular chain inertia has a major influence on bone conduction. It was not possible in the present study to completely eliminate the cochlear shell distortional component but it is probable that this mechanism would not be activated by direct stimulation of the brain. Stenfelt et al. (2000) also conclude that compression-distortion of the cochlear shell is of minor importance.In spite of the successive experimental removal of the classical bone conduction mechanisms, ‘bone conduction’ responses could still be obtained. The ABR responses were much larger in amplitude than those obtained in response to the air conduction stimulus generated by the bone vibrator. Similar results to those obtained here in experiment D had previously been seen in rats (Geal-Dor et al., 1993) and now confirmed in fat sand rats. In this situation, only the cochlear shell distortional–compressional mechanism of bone conduction remained and even this was largely removed when a relatively wide craniotomy was made and the bone vibrator was applied directly to the brain. Although ABR thresholds to bone vibrator on bone were then elevated by 12–14 dB, the ABR thresholds to bone vibration on brain were not affected by even maximal removal of calvarium. This latter result thus shows that skull bone may not even be necessary for the induction of ABR responses to ‘bone-conducted’ stimuli, as long as the bone vibrator is applied directly to the brain. Following removal of the ossicular chain inertia mechanism, bone conduction ABR thresholds improved by 5–10 dB compared to those before immobilization. This could be due to the presence of a small inertial bone conduction component of opposite phase with respect to the major fluid pathway component. Then ossicular immobilization, removing this inertial component, could give rise to a larger response, i.e. threshold improvement.Thus, it is likely that when the bone vibrator is applied directly to the brain, it induces audio-frequency pressure vibrations in the contents of the skull, probably in the CSF, which are then communicated via fluid channels directly to the cochlea, leading to cochlear excitation. The results of the experiments involving the apparent reduction in CSF volume by mannitol administration in rats with intact skulls (Experiment E) are highly relevant here. The small air conduction ABR threshold improvements (3.93 dB) show that cochlear function was slightly improved by this manipulation, probably related to the increase in cochlear blood flow shown following mannitol injection in animals (Larsen et al., 1982; Goodwin et al., 1984; Baldwin et al., 1992). At the same time, the bone conduction thresholds in the same animals became elevated following the presumed reduction in intracranial fluid (including CSF) volume. Thus if one considers the air conduction threshold improvements, it is likely that the simultaneous bone conduction threshold elevations were even slightly greater than those actually recorded. This provides strong support for the suggestion that the skull content components which are involved in the transfer of a major part of the vibratory energy from the skull to the inner ear fluids, are the fluid contents of the cranial cavity, especially CSF. This is further confirmation of the involvement of a fluid pathway in bone conduction stimulation in the intact animal, and that the results in rats with craniotomies are not artifacts due to the craniotomy. It is therefore suggested that in bone conduction stimulation with an intact skull, the audio-frequency vibrations of the skull induce audio-frequency pressure alternations of the skull contents (CSF), which are conveyed by fluid pathways directly to the inner ear fluid spaces. In addition, one or more of the classical bone conduction mechanisms vectorially sum with this major fluid mechanism, exciting the cochlea.It had been thought that the final cochlear response to a bone conduction stimulus represents the vectorial summation of the energy pathways between the point of vibrator application on the skull and the cochlea (surface, transverse and pressure waves), and the vectorial summation of the inner ear modes of bone conduction response (inertial, occlusion and distortional-compression). If this were the case, then the successive removal of each these components in the experimental manipulation in these experiments, should have been accompanied by major changes in the ABR waveform and threshold. This was not the case, providing further evidence that the major (dominant) mechanism in the initiation of the bone conduction response is the fluid pathway described, with smaller contributions from the other classical mechanisms.Why are the ABR thresholds to bone vibrator on bone stimuli elevated (12–14 dB; in addition, longer latencies and lower amplitudes) in the presence of a craniotomy? This is not due to the craniotomy (fluid leak) adversely affecting the cochlea since air conduction thresholds were not changed following the craniotomy. It is possible that the craniotomy, reducing the amount of bone on the skull, may have attenuated, for example, a smaller contribution from the transverse wave (propagation along skull bone) mechanism of bone conduction to the final response. If so, in the intact skull, the contribution from the fluid mechanism and the contribution from the transverse wave mechanism would have had to be in phase and therefore summating, so that the removal of the distortional mechanism would induce a reduction in response amplitude, that is a threshold elevation. However it is more likely that the craniotomy causes a dissipation (damping) of the CSF pressure waves induced by the skull vibrations and the larger the craniotomy, the greater the dissipation. This is similar to the explanation for the finding that the magnitude of the CSF pressure alternations related and time locked to the heart (pulse) and the respiration are dependent on the intracranial pressure such that as this pressure becomes elevated, the pulse and respiratory waves become larger in amplitude (Bering, 1955; Dunbar et al., 1966). The loss of some CSF through the craniotomy reducing CSF pressure, may also be responsible for part of this threshold elevation. Furthermore these CSF pulse pressures measured in a patient in a lateral ventricle by Bering (1955) with an electronic manometer were larger when the valve connecting to an open bore manometer in the opposite lateral ventricle was closed. This was ascribed to the damping effect of the open bore manometer.In similar experiments in humans (Sohmer et al., 2000), the bone vibrator was placed on the fontanelle in neonates and on a previous craniotomy in patients. The auditory thresholds obtained (ABR to clicks in neonates and audiometric to pure tones in patients) were similar to those obtained in response to bone vibrator placements over intact skull in the same patients. In these cases the brain and CSF were covered by dura and skin and it is likely that these overlying tissues led to less dissipation of the vibratory energy which had been induced by the bone vibrator in the intracranial contents (brain and CSF). On the other hand, in the animal experiments reported here, the experimental craniotomy was not covered so that when the bone vibrator was placed on the brain in the craniotomy, the dissipation was in the immediate surrounding of the bone vibrator and already maximal. However, when the bone vibrator was placed in the same animals on nearby intact bone, the region of dissipation was further away, with intact bone lying between the site of the bone vibrator and the craniotomy. This could lead to less dissipation in the latter situation, explaining the better ABR thresholds obtained with the vibrator on bone compared to those with vibrator on brain. Increasing the size of the craniotomy would then be accompanied by further dissipation and more elevated thresholds to stimulation of bone.The fluid communications which are probably involved in the transfer of pressure changes from the CSF to the inner ear could include the cochlear aqueduct which communicates between the subarachnoid space of the posterior cranial fossa and scala tympani of the basal turn of the cochlea and perhaps also perivascular (e.g. inferior cochlear vein) and perineural spaces (e.g. in the internal auditory meatus) (Sando et al., 1971; Palva et al., 1979; Schuknecht and Reisser, 1988). The vestibular aqueduct may also be involved (Marchbanks and Reid, 1990; Kitahara et al., 1994; Konradsson et al., 1994; Yoshida and Uemura, 1991). In fact in man the diameters of vestibular aqueduct apertures communicating between the endolymph in the saccule and the subarachnoid space are larger than the corresponding apertures of the cochlear aqueduct (Anson et al., 1965). These fluid channels need not be involved in bulk flow of CSF or perilymph or endolymph across these fluid spaces. They need only transmit audio-frequency pressures from one fluid space to another. This is similar to the cardiac and respiratory pressure pulses present in the CSF which could also be recorded in the inner ear and whose origin is intracranial (Martinez, 1968; Carlborg et al., 1982; Bohmer, 1993). Further experimentation is required in order to determine which fluid channel(s) is (are) involved.In conclusion, this study in small experimental animals has contributed to the elucidation of the complex mechanisms involved in bone conduction stimulation of the ear. It seems that a major mechanism involves skull vibrations which give rise to alternating CSF pressure. These are communicated by fluid channels from the CSF directly to the inner ear fluids. Contributions from the ‘classical’ bone conduction mechanisms (surface waves, ossicular chain inertia and skull distortions) vectorially sum with these CSF pressure waves, exciting the cochlea. Since each of these mechanisms, including the fluid one suggested here, bypass the outer and middle ears, bone conduction stimulation is still a valid means for the differentiation between conductive and sensori-neural hearing losses. Even though the basic physiological mechanism of bone conduction is better understood as a result of these experiments, further clarification is required.ReferencesAnson et al., 1965B.J.AnsonJ.A.DonaldsonR.L.WarpehaT.R.WinchThe vestibular and cochlear aqueducts: their variational anatomy in the adult human earLaryngoscope75196512031223Baldwin et al., 1992D.L.BaldwinK.A.OhlsenJ.M.MillerA.L.NuttallCochlear blood flow and microvascular resistance changes in response to hypertonic glycerol, urea, and mannitol infusionsAnn. Otol. Rhinol. Laryngol.1011992168175Barany, 1938E.BaranyA contribution to the physiology of bone conductionActa. Oto-Laryngol.26Suppl.19381223von Bekesy, 1932G.von BekesyZur Theorie des Horens bei der Schallaufnahme durch KnochenleitungAnn. Phys.131932111136von Bekesy, 1960von Bekesy, G., 1960. In: Wever, E.G. (Ed.), Experiments in Hearing. McGraw-Hill, New York.Bering, 1955E.A.BeringChoroid plexus and arterial pulsation of cerebrospinal fluidArch. Neurol. Psychiatry731955165172Bohmer, 1993A.BohmerHydrostatic pressure in the inner ear fluid compartments and its effects on inner ear functionActa Oto-Laryngol. (Stockh.)507Suppl.1993324Carlborg et al., 1982B.CarlborgB.DensertO.DensertFunctional patency of the cochlear aqueductAnn. Otol. Rhinol. Laryngol.911982209215Donato et al., 1994T.DonatoY.ShapiraA.ArtruK.PowersEffect of mannitol on cerebrospinal fluid dynamics and brain tissue edemaAnesth. Analg.7819945866Dunbar et al., 1966H.S.DunbarT.C.GuthrieB.KarpellA study of the cerebrospinal fluid pulse waveArch. Neurol.141966624630Geal-Dor et al., 1993M.Geal-DorS.FreemanG.LiH.SohmerDevelopment of hearing in neonatal rats: air and bone conducted ABR thresholdsHear. Res.691993236242Goodwin et al., 1984P.C.GoodwinJ.M.MillerH.A.DengerinkJ.W.WrightA.AxelssonThe laser Doppler: a non-invasive measure of cochlear blood flowActa Oto-Laryngol. (Stockh.)981984403412Kirikae, 1959I.KirikaeAn experimental study on the fundamental mechanism of bone conductionActa Oto-Laryngol.145Suppl.19591111Kitahara et al., 1994M.KitaharaM.SuzukiA.KodamaEquilibrium of inner and middle ear pressureActa Oto-Laryngol. (Stockh.)510Suppl.1994113115Konradsson et al., 1994K.S.KonradssonA.H.CarlborgJ.C.FarmerJr.B.I.CarlborgPerilymph pressure during hypobaric conditions – cochlear aqueduct obstructedActa. Oto-Laryngol. (Stockh.)11419942429Larsen et al., 1982H.C.LarsenC.AngelborgE.HultcrantzCochlear blood flow related to hyperosmotic solutionArch. Otorhinolaryngol.2341982145150Marchbanks and Reid, 1990R.J.MarchbanksA.ReidCochlear and cerebrospinal fluid pressure: their inter-relationship and control mechanismsBr. J. Audiol.241990179187Martinez, 1968D.M.MartinezSimultaneous measurements of endolymphatic and perilymphatic fluid pressures before and during anaphylaxis and associated changes in cerebrospinal fluid, venous and arterial pressuresActa Oto-Laryngol. (Stockh.)238Suppl.1968553Palva et al., 1979T.PalvaV.RaunioP.KarmaJ.YlikoskiFluid pathways in temporal bonesActa Oto-Laryngol. (Stockh.)871979310316Sando et al., 1971I.SandoY.MasudaR.P.Wood2dW.G.HemenwayPerilymphatic communication routes in guinea pig cochleaAnn. Otol. Rhinol. Laryngol.801971826834Schneider, 1959W.SchneiderGegenbeweis gegen Knochenleitung mittels Druckwellen uber den Kanal des nervus acusticusZ. Laryngol.381959723734Schuknecht and Reisser, 1988H.F.SchuknechtC.ReisserThe morphologic basis for perilymphatic gushers and oozersAdv. Otorhinolaryngol.391988112Sohmer et al., 2000H.SohmerS.FreemanM.Geal-DorC.AdelmanI.SavionBone conduction experiments in humans – a fluid pathway from bone to earHear. Res.14620008188Stenfelt et al., 2000S.StenfeltB.HakanssonA.TjellstromVibration characteristics of bone conducted sound in vitroJ. Acoust. Soc. Am.1072000422431Tonndorf, 1966J.TonndorfBone conduction. Studies in experimental animalsActa Otolaryngol. (Stockh.)213Suppl.19661132Tonndorf, 1968J.TonndorfA new concept of bone conductionArch. Otolaryngol.871968595600Tonndorf and Tabor, 1962J.TonndorfJ.R.TaborClosure of the cochlear windows: its effect upon air- and bone-conductionAnn. Otol. Rhinol. Laryngol.711962529Treib et al., 1998J.TreibS.C.BeckerM.GrauerA.HaassTranscranial Doppler monitoring of intracranial pressure therapy with mannitol, sorbitol and glycerol in patients with acute strokeEur. Neurol.401998212219Wever and Lawrence, 1954Wever, E.G., Lawrence, M., 1954. Physiological Acoustics. Princeton University Press, Princeton, NJ.Yoshida and Uemura, 1991M.YoshidaT.UemuraTransmission of cerebrospinal fluid pressure changes to the inner ear and its effect on cochlear microphonicsEur. Arch. Otorhinolaryngol.2481991139143