Hearing Research 152 (2001) 100^109

www.elsevier.com/locate/heares

Human middle-ear sound transfer function and cochlear input impedance

a;b
Ryuichi Aibara , Joseph T. Welsh a , Sunil Puria c , Richard L. Goode a;b;
*
a
Department of Veterans Aairs Medical Center, 3801 Miranda Ave., MC 112-B1, Palo Alto, CA 94304, USA
b
Division of Otolaryngology-Head and Neck Surgery, Stanford University School of Medicine,
300 Pasteur Drive, R135, Stanford University Medical Center, Stanford, CA 94305, USA
c
Stanford University Department of Mechanical Engineering, Stanford, CA 94305, USA
Received 24 February 2000; accepted 18 October 2000

Abstract

The middle-ear pressure gain, defined as the ear canal sound pressure to cochlear vestibule pressure gain, GME, and the ear canal
sound pressure to stapes footplate velocity transfer function, SVTF, simultaneously measured in 12 fresh human temporal bones for
the 0.05 to 10 kHz frequency range are reported. The mean GME magnitude reached 23.5 dB at 1.2 kHz with a slope of
approximately 6 dB/octave from 0.1 to 1.2 kHz and 36 dB/octave above 1.2 kHz. From 0.1 to 0.5 kHz, the mean GME phase angle
was 51, rolling off at 378/octave above this frequency. The mean SVTF magnitude reached a maximum of 0.33 mm s31 /Pa at
1.0 kHz with nearly the same shape in magnitude and phase angle as the mean GME. The ratio of GME and SVTF provide the first
direct measurements of Zc in human ears. The mean Zc was virtually flat with a value of 21.1 acoustic G6 MKS between 0.1 and
5.0 kHz. Above 5 kHz, the mean Zc increased to a maximum value of 49.9 G6 at 6.7 kHz. The mean Zc angle was near 0 from 0.5
to 5.0 kHz, decreasing below 0.5 kHz and above 5 kHz with peaks and valleys. 2001 Elsevier Science B.V. All rights reserved.

Key words: Middle-ear sound transmission; Middle-ear sound pressure gain; Cochlear input impedance ;
Scala vestibuli sound pressure; Stapes footplate velocity

1. Introduction Pec is termed the ear canal sound pressure to stapes

Several methods have been described for the mea-
surement of middle-ear sound transmission in human
temporal bones with the cochlea intact (Gundersen,
1971 ; Kringlebotn and Gundersen, 1985; Vlaming
and Feenstra, 1986; Gyo et al., 1987; Brenkman et
al., 1987; Kurokawa and Goode, 1995). The input pa-
rameter is universally the sound pressure level (SPL) at
the ear canal surface of the tympanic membrane (TM),
termed Pec . The output parameter used varies, depend-
ing on the particular study; in intact human temporal
bone measurements, stapes footplate velocity (Vs ) or
scala vestibuli sound pressure near the stapes footplate
(Pv ) are two physiological choices. The ratio of Vs to
* Corresponding author. Tel.: +1 (650) 723 6030;
Fax: +1 (650) 496 2502; E-mail: goode@leland.stanford.edu
footplate velocity transfer function, SVTF, while the
ratio of Pv to Pec is termed the middle-ear sound pres-
sure gain, GME. Human GME has also been measured
by comparing stapes displacement in response to a con-
stant SPL before and after removal of the TM, malleus
and incus (Kurokawa and Goode, 1995). Pressure gain
estimates by Bekesy (1960), Onchi (1961) and others
(Rubenstein et al., 1964; Fishler et al., 1965) using dif-
ferent methods have been reported as well. These meth-
ods have the disadvantage of requiring the middle-ear
to be removed or the cochlea drained to make the mea-
surement.
Dierences in human temporal bone SVTF and
GME results have been found by dierent investigators,
particularly at the lower and higher test frequencies
(reviewed by Goode et al., 1987, 1994 ; Puria et al.,
1997). While these dierences may be due to the wide
range in normal middle-ear function, particularly when
small numbers of ears are tested, it is also possible that
the measurement method contributes to the dierences.

0378-5955 / 01 / $ ^ see front matter 2001 Elsevier Science B.V. All rights reserved.
PII: S 0 3 7 8 - 5 9 5 5 ( 0 0 ) 0 0 2 4 0 - 9

HEARES 3609 15-2-01

 R. Aibara et al. / Hearing Research 152 (2001) 100^109
The stapes footplate motion is in general a vector
quantity with translational as well as rotational compo-
nents. The piston-like translational component of the
footplate Vs is generally assumed to be the primary
determinant of stapes volume velocity and the input
to the cochlea (Dallos, 1973). Thus, the SVTF measure-
ment method assumes that stapes footplate volume ve-
locity (Vs times the area of the footplate, Afp ) to be the
appropriate input to the cochlea, so Vs is selected as the
output parameter since Afp is considered to be relatively
constant. A laser Doppler vibrometer (LDV) is usually
used to make the measurement and the center of the
footplate is chosen as the measurement site based on
the assumption that stapes movement is piston-like at
physiological sound inputs and frequencies. Recent data
from our laboratory suggests that stapes footplate
movement is not entirely piston-like but has an anteri-
or^posterior rocking motion with rising frequency (Hei-
land et al., 1999); this may make measurement of stapes
velocity at one site sub-optimal as the middle-ear out-
put measure at higher frequencies ( s 2.0 kHz).
Sound pressure measured in the scala vestibuli has
been reported as a better output parameter, because it
does not make assumptions about the piston-like mo-
tion of the stapes (Puria et al., 1997). However, the
measurement is more dicult to perform. Measurement
of round window volume velocity, assumed to be the
same as stapes volume velocity, has also been used but
is not a direct measure (Kringlebotn and Gundersen,
1985).
Until now, no comparative measurements with these
two methods have been made in the same human tem-
poral bones. In addition, simultaneous measurement of
stapes volume velocity and intra-cochlear sound pres-
sure allows the direct measurement of human cochlear
input impedance, an important measure that has only
been previously measured indirectly. It is the purpose of
this paper to describe experiments and compare results
from the two methods in the same human ears as well
as provide the direct measurement of human cochlear
input impedance.
The complex acoustic input impedance of the co-
Table 1
Denitions
Afp area of the stapes footplate
GME middle-ear sound pressure gain, ratio of Pv to Pec
UGME phase angle of middle-ear sound pressure gain
Pec ear canal sound pressure at the surface of the TM
Pv scala vestibuli sound pressure near the stapes footplate
SVTF stapes velocity transfer function, ratio of Vs to Pec
USVTF phase angle of stapes velocity transfer function
Us stapes footplate volume velocity, Vs times Afp
Vs stapes footplate velocity
Zc acoustic input impedance of the cochlea
UZc phase angle of acoustic input impedance of the cochlea
HEARES 3609 15-2-01
101
chlea, Zc , is an important variable in our understanding
of sound energy transfer through the middle-ear into
the cochlea. Zwislocki (1975) dened the acoustic input
impedance of the cochlea as ``the ratio of sound pres-
sure produced in the scala vestibuli at the stapes foot-
plate and the volume of perilymph the footplate displa-
ces per unit time''. In other words, Zc is the ratio of Pv
to stapes footplate volume velocity, Us , which is calcu-
lated by Vs times Afp (see Table 1 for denitions).
Direct measurements of the input impedance of the
cochlea have been reported in the cat (Lynch et al.,
1994), guinea pig (Dancer and Franke, 1980) and chin-
chilla (Ruggero et al., 1990). However, only estimations
of Zc for humans have been reported by Merchant et
al. (1996) and Puria et al. (1997), as well as theoretical
predictions by Zwislocki (1965, 1975) and Bekesy
(1960). Direct measurement of Zc in human ears is re-
ported here for the rst time.
2. Materials and methods
2.1. Measurement system
Measurements were performed using a SYSid 6.5
audio band measurement and analysis system (www.
sysid-labs.com, Berkeley, CA, USA), a software pro-
gram that interfaces with a DSP-16+ processing board
(Ariel, Cranbury, NJ, USA) running on an IBM com-
patible PC. Simultaneous stimulus generation and re-
sponse averaging are performed with the SYSid system,
which extracts the amplitude and phase of the response
at each frequency through fast Fourier transformation,
or FFT (Puria and Allen, 1992; Puria et al., 1993).
A schematic of the measurement system is shown in
Fig. 1. Sound was presented at the lateral end of an
articial ear canal using an 83-13A/024 earphone (Tib-
bets Industries, Camden, ME, USA). Sound pressure in
the ear canal (Pec ) was measured within 2 mm of the
TM with an ER-7C probe-tube microphone (Etymotic
Research, Elk Grove Village, IL, USA). Sound pressure
in the scala vestibuli (Pv ) was measured using an EPIL-
127*-.13 D/R strain gauge (Entran, Faireld, NJ, USA)
modied for use as a hydrophone by placing a silicone
seal on the tip of the sensor (Puria et al., 1997). An
OFV-302 and OFV-3000 LDV System (Polytec, PI,
Costa Mesa, CA, USA) was used to measure stapes
footplate velocity (Vs ). The LDV contained a 50 mm
lens (Nikon, NY, USA) to focus the helium^neon laser
on a reective target on the center of the stapes foot-
plate. An articulating arm was connected to a joystick-
operated beam-splitter, which was mounted over the
lens of an operating microscope and reected 95% of
the laser beam onto the target (see Heiland et al. (1999)
for more details on the velocity measurement methods).
102 R. Aibara et al. / Hearing Research 152 (2001) 100^109
Fig. 1. Schematic of the measurement method.
2.2. System calibration
The probe-tube microphone and the hydrophone
were calibrated as follows. A sound level calibrator
type 4230 (Bruel and Kjaer, Norcross, GA, USA) was
used to calibrate a 1/8 inch reference microphone type
4138 (Bruel and Kjaer, Norcross, GA, USA) which had
a at response ( 0.5 dB) and zero phase shift ( 1) in
the 0.05^10 kHz frequency range. This reference micro-
phone was then used to calibrate each microphone by
comparing their response to that of the reference micro-
phone at 0.05^10 kHz in a small 25 mm long cavity ; the
microphones were within 1 mm of the reference micro-
phone. Due to its high impedance, the hydrophone's
sensitivity is assumed to be the same in air and water.
To check the calibration of the two microphones, the
probe-tube microphone and hydrophone were placed
together in a dierent 25 mm long cavity, 1 mm apart,
and their responses were compared. The maximum de-
viation in magnitude was 1.0 dB at 1 kHz with a max-
imum deviation in phase of 15 at 10 kHz. The ob-
served dierences place a boundary on possible
calibration errors in the measurement system.
The LDV system was calibrated by aiming the LDV
at a target on top of an accelerometer type 4371 (Bruel
and Kjaer, Norcross, GA, USA) that was attached to a
mini shaker type 4810 (Bruel and Kjaer, Norcross, GA,
USA). The response of the LDV was then compared to
the accelerometer's, which had a known sensitivity and
phase response between 0.05 and 10 kHz.
HEARES 3609 15-2-01
2.3. Postmortem materials
Twelve fresh human temporal bones, consisting of
9 males and 3 females ranging from 36 to 81 years
with a mean age of 67.9 years, were extracted from
human cadavers within 48 h of death using a Schu-
knecht bone saw at the time of autopsy. The cored
specimens were wrapped in gauze, placed in a 100 ppm
merthiolate in normal saline solution and stored at 5C.
Measurements on individual bones were performed in
1 day within 6 days postmortem. With the exception of
one temporal bone, all were considered normal upon
visual inspection using an operating microscope. One
exception was made for a temporal bone where palpa-
tion of the umbo showed a stier malleus than normal.
Since the footplate did not appear otosclerotic, only the
cochlear input impedance measurement was used.
2.4. Temporal bone preparation and baseline
measurements
Upon removal of attached connective tissue, the
bony wall of the external ear canal was drilled down
to 2 mm from the TM annulus, keeping the posterior
wall intact. A simple mastoidectomy and posterior hy-
potympanotomy were performed. The horizontal seg-
ment of the facial nerve and surrounding bone were
removed to provide a good view of the stapes footplate,
leaving the chorda tympani and stapedius muscle intact.
After enlarging the internal auditory canal (IAC) to
 R. Aibara et al. / Hearing Research 152 (2001) 100^109
approximately 15U10 mm, the bone was drilled until a
blue line over the vestibule was seen at a site lateral to
the transverse crest and just superior to the singular
foramen. The bone was then placed in a latex nger
cot which kept the wet bone moist and thus prevented
it from drying.
A plastic tube (8.5 mm internal diameter, 25 mm
length) was attached around the bony ear canal opening
using modeling clay so that the axis of the tube was
perpendicular to the TM. The Tibbets earphone was
located at the lateral end of the tube for sound delivery.
The probe-tube microphone was inserted into the tube
for the sound pressure measurements. The tip of the
probe-tube microphone was typically 1^2 mm from
the TM.
Except for the articial ear canal and an opening in
the mastoid for the laser beam, the remainder of the
temporal bone was embedded in Hydrock dental ce-
ment (Kerr Co., Romulus, MI, USA), creating a solid
specimen block. The cement edge of the mastoid open-
ing was covered with Jeltrate (Dentsply Caulk Co., Mil-
ford, DE, USA) ; the opening provided a window that
allowed the stapes footplate to be seen. The cemented
bone was then secured in a temporal bone holder. Two
0.5 mm square pieces of reective tape (3M, St. Paul,
MN, USA), weighing 0.04 mg, were placed; one on the
lateral surface of the TM over the umbo and the other
on the center of the stapes footplate. Adhesive on the
back of the targets kept them adhered to the umbo and
footplate. Glass cover slips were placed over the mas-
toid opening and the lateral end of the ear canal sound
assembly and held in place by silicone grease to prevent
drying and create a good sound seal.
The earphone presented a stimulus consisting of
200 pure tones logarithmically spaced from 0.024 to
25 kHz and ranging from 60 to 120 dB SPL at the
TM, however, only the 0.05 to 10 kHz frequency range
is reported here. The lower frequency limit is due to a
decreased signal-to-noise ratio and the upper frequency
limit is due to microphone calibration limitations.
The stimulus was amplied using a D-75 power am-
plier (Crown, Elkhart, IN, USA), fed through a 200 6
resistor (to increase the high frequency response), and
then through the earphone. The LDV output was am-
plied by 20 dB using the D-75 power amplier. Sound-
source attenuation and amplier gains were chosen to
minimize harmonic distortion. Although the noise oor
varied with frequency, all data reported here had a
signal-to-noise ratio greater than 10 dB near the lowest
and highest frequency points and greater than 20 dB in
the middle frequency region. Responses from the ER-
7C and LDV were then collected concurrently and the
Pec and related Vs data saved.
For the purpose of the experiment, the actual stapes
footplate movement was considered to be predomi-
HEARES 3609 15-2-01
103
nantly piston-like and parallel to the motion of the
umbo. Therefore, because the laser beam passed
through the mastoid opening and was not perpendicu-
lar to the plane of the stapes footplate vibration, only a
component of the actual movement of the stapes foot-
plate was considered measured. The angle between the
laser beam aimed through the ear canal perpendicular
to the umbo target and the laser beam aimed at the
stapes footplate target through the mastoid was mea-
sured using a protractor. The output of the LDV was
divided by the cosine of that measured angle to deter-
mine the actual stapes footplate velocity, Vs . SVTF was
then determined from the ratio Vs /Pec , while the phase
angle, USVTF, was determined from the dierence in
phase between the outputs of the LDV and probe-tube
microphone.
2.5. Hydrophone insertion and measurement
After baseline measurements of SVTF, the temporal
bone was orientated so the IAC was superior. After
removing cement and latex to expose the IAC, the canal
was lled with saline solution. The saline prevented the
introduction of air bubbles while a 1.4 mm diameter
opening was made just anterior and superior to the
singular canal into the scala vestibuli. The saccule was
moved aside using a ne pick so that the stapes foot-
plate could be directly seen through this vestibular
opening. The bone was placed in a temporal bone hold-
er xed to a vibration-reducing table. The hydrophone
had a diameter of 1.4 mm and was encased in a brass
tube with an outer diameter of 1.8 mm. Using a micro-
manipulator type M3301 (World Precision Instru-
ments, Sarasota, FL, USA) attached to the vibration-
reducing table, the hydrophone was centered and
placed perpendicular and ush to the vestibular open-
ing. This allowed the hydrophone to lie against the
bony edge of the opening without entering the vestibule
(Puria et al., 1997). Jeltrate was used to ll the IAC
to provide a static pressure seal of the hydrophone
in the vestibular opening. Previous measurements indi-
cate that this seal has an eect only at frequencies be-
low approximately 200 Hz. The hydrophone response
was amplied by 60 dB using a SR-560 low noise am-
plier (Stanford Research Systems, Sunnyvale, CA,
USA).
The responses of the LDV and hydrophone were si-
multaneously collected three times; techniques men-
tioned in Section 2.4 were used to derive Vs from the
LDV response. Zc was determined for each measure-
ment from the ratio Pv /(Vs Afp ); the phase angle, UZc
was determined from the dierence in phase between
the outputs of the hydrophone and LDV. Afp was de-
termined from a calibrated digitized image of the foot-
plate taken through the opening made in the vestibule.
104 R. Aibara et al. / Hearing Research 152 (2001) 100^109

The average footplate area was 3.21 mm2 , with a stan-

dard deviation (S.D.) of 0.34 mm2 .
Using Pec from the baseline measurements, the mag-
nitude of GME was then determined from the ratio
MPv M/MPec M; the phase angle, UGME, was determined
from the dierence in phase between the outputs of
the hydrophone and probe-tube microphone. Measure-
ments show that SVTF was the same before and after
hydrophone insertion. For each ear, Zc , SVTF and
GME were derived using the mean of the three hydro-
phone and LDV measurements of Pv and Vs . The time
required for the entire preparation and all measure-
ments was less than 8 h for each temporal bone.

3. Results

3.1. Middle-ear sound pressure gain, GME

The ear canal sound pressure to vestibule sound pres-

sure transfer function, GME, from 11 ears is shown in
Fig. 2. The mean GME magnitude has a maximum of
23.5 dB at 1.2 kHz, with a slope approximately 6 dB/
octave from 0.1 to 1.2 kHz and around 36 dB/octave
above 1.2 kHz. From 0.1 to 0.8 kHz, the mean GME
phase angle is relatively at with an average value of
Fig. 3. Ear canal sound pressure to stapes footplate velocity transfer
function, SVTF, in 11 human temporal bone ears. (A) shows the
magnitude and (B) shows the phase angle. The curve for each ear
represents the mean of three consecutive measurements of SVTF.
Also shown is the mean magnitude and phase (thick line) of the
11 ears.

48. Above 0.8 kHz, the mean GME phase angle de-
creases with a slope of 387/octave.

3.2. Stapes velocity transfer function, SVTF

The ear canal sound pressure to stapes footplate ve-

locity transfer function, SVTF, in the same 11 ears is
shown in Fig. 3. The mean SVTF magnitude has a
maximum of 0.33 mm s31 /Pa at 1.0 kHz, with a mean
slope of approximately 6 dB/octave from 0.1 to 1.0 kHz
and around 37 dB/octave above 1.0 kHz. The mean
SVTF phase angle is nearly the same as the mean
GME phase angle, except for small dierences below
0.2 kHz and above 5.0 kHz.

3.3. Cochlear input impedance, Zc

Fig. 4 shows Zc for each bone, including the mean.

Fig. 2. Ear canal sound pressure to scala vestibuli sound pressure
transfer function, GME, in 11 human temporal bone ears. (A)
shows the magnitude and (B) shows the phase angle. The curve for
each ear represents the mean of three consecutive measurements of
GME made within 1 h of hydrophone insertion. Also shown is the
mean magnitude and phase angle (thick line) of the 11 ears.
From 0.1 to 5.0 kHz, the mean Zc magnitude is rela-
tively at, with a value of 21.1 G6. In the 0.5^5.0 kHz
region, the Zc phase angle is approximately 0 ( 10)
and nearly at, indicating that Zc is resistance dominat-
ed. At 0.1 kHz, the mean Zc phase angle is 333 and
increases to 0 at 0.5 kHz. The mean Zc minimum is
13.4 acoustic G6 MKS at 0.5 kHz, with a maximum of
49.9 G6 at 6.7 kHz. Above 5.0 kHz, the mean Zc
magnitude increases with a peak at 6.7 kHz and then

HEARES 3609 15-2-01

 R. Aibara et al. / Hearing Research 152 (2001) 100^109
Fig. 4. Human cochlear input impedance, Zc , in 12 human temporal
bone ears. (A) shows the magnitude and (B) shows the phase angle.
The curve for each ear represents the mean of three consecutive
measurements of vestibule pressure and stapes velocity made within
1 h after hydrophone insertion. The mean magnitude and phase
(thick line) of all ears is also shown. The phase from one ear was
excluded because of a phase measurement error.
Fig. 5. The curves shown are the mean dierence in SVTF after in-
sertion of the hydrophone in 11 ears. (A) shows the magnitude and
(B) shows the phase angle. This comparison shows that hydrophone
insertion has a negligible eect on stapes velocity, Vs .
HEARES 3609 15-2-01
105
decreases. Above 5.0 kHz, the mean phase angle of Zc
decreases 59/octave to a minimum of 349 at 8.6 kHz,
then increases to 340 at 10 kHz.
3.4. Eect of opening the scala vestibuli
SVTF was measured before and after inserting the
hydrophone to assess the eect of opening the scala
vestibuli. As shown in Fig. 5, the average change in
magnitude is about 1.0 dB, with a maximum of 3.7 dB
at 5.7 kHz. The average change in phase angle is 5
with a maximum of 30 at 9.4 kHz. The eect of open-
ing the scala vestibuli on SVTF measurement is small.
3.5. Eect of time on stapes velocity
In a temporal bone not used for cochlear impedance
measurements, Vs was measured at 0.5, 1, 2 and 20 h
after preparation. The eect due to drying was less than
1.5 dB, therefore, we conclude that Vs remained consis-
tent over the measurement time of the experiments.
4. Discussion
4.1. Previous measurements of middle-ear pressure gain
In Fig. 6, our mean GME is compared to the GME
reported by others. There is relatively good agreement
between our results and those obtained by Puria et al.
(1997), at most, disagreeing by 6 dB in magnitude and
40 in phase. However, there are dierences in the
shapes of the average curves. Puria et al. found a at
mean pressure gain of 20 dB between 0.5 and 2.0 kHz,
decreasing to 18 dB at 4.0 kHz. From 0.1 Hz to 0.5
kHz, the magnitude increased by around 6 dB/octave
while above 4.0 kHz the magnitude rolled o by 38 dB/
octave. In contrast, our data show a 23.5 dB peak gain
at 1.2 kHz with a roll o of 36 dB/octave above this
frequency ; the gain is 20 dB at 2.0 kHz and 12 dB at
4.0 kHz. However, the dierences between the present
measurements and those of Puria et al. are not statisti-
cally signicant since their GME is within 1 S.D. of
the present measurements. The GME measurement by
Huttenbrink and Hudde (1994) is overall greater than
our mean result up to near 3.0 kHz while above this
frequency the gain rolls o precipitously possibly due to
air in the cochlea (Puria et al., 1997).
4.2. Previous measurements of stapes velocity
In Fig. 7, we compare the mean SVTF obtained in
this study with a previous SVTF measurement from our
laboratory using a similar measurement method (Goode
et al., 1994) and model calculations by Kringlebotn
106 R. Aibara et al. / Hearing Research 152 (2001) 100^109

(1988) based on temporal bone data obtained by

Kringlebotn and Gundersen (1985) using round win-
dow velocity as the output measure. In our previous
measurements peak gain was at 0.7 kHz rather than
1.0 kHz. The slopes above and below the resonant fre-
quency of our previous measurements are essentially
identical to the mean SVTF. However, because of the
lower resonant frequency in the previous measurements,
velocity magnitude at 0.5 kHz is about 5 dB greater
while above 1.0 kHz it is about 4 dB less reaching a
maximum dierence of 7 dB at 4.4 kHz. No phase
measurements were made in the 1994 experiments. In
these previous measurements, a thin coat of oil was
used on the TM to decrease any drying eects; this
could add a slight mass to the TM and decrease sti-
ness; the eect of both being a decrease in the reso-
nant frequency. Comparing our current data with the
Kringlebotn model shows only small dierences in mag-
nitude and phase up to 4.0 kHz; the slopes are essen-
tially identical. Above 4.0 kHz there are increasing
dierences with rising frequency reaching a peak at
10 kHz where the magnitude diers by 7 dB and the Fig. 7. Comparison of mean SVTF obtained in this study in 11 ears
phase angle by 90. Although the mean values of SVTF with the mean SVTF reported by Goode et al. (1994) and the
of the three studies dier, as discussed, the dierences SVTF calculated by Kringlebotn (1988). No phase measurements
are not statistically signicant. This is evident by the were obtained in the earlier Goode study. (A) shows magnitude and
observation that both the Goode et al. (1994) measure- (B) shows phase angle.
ments and the Kringlebotn (1988) results are well
within 1 S.D. of the current measurements. Of the many available studies, these two were se-
lected for comparison because the Goode et al. (1994)
SVTF data was obtained in the same laboratory using a
similar measurement method while the Kringlebotn and
Gundersen (1985) data is based on the largest number
of temporal bones (68) in the literature and was used
for the 1988 Kringlebotn model that Puria et al. (1997)
based their estimate of Zc on. As mentioned earlier, a
variety of measurement methods have been used to de-
termine human middle-ear sound transmission with
varying results. We have previously described some of
the possible reasons for this variation (Goode et al.,
1987, 1994; Puria et al., 1997).
Fig. 8 shows the mean data of Figs. 2 and 3 on one
graph for comparison purposes. SVTF has been con-
verted to magnitude in dB with 0 dB assumed to be
2.1U1032 mm s31 /Pa at 0.05 kHz. Such a scale factor
results in magnitude and zero degree angle overlap at
1 kHz. The two magnitude curves are quite similar,
separating slightly at higher frequencies. The two phase
curves are also similar but are slightly dierent at fre-
quencies below about 0.5 kHz.

4.3. Previous measurements of cochlear input impedance

Fig. 6. Comparison of mean GME obtained in this study in 11 ears
with mean GME reported by Puria et al. (1997) in four ears and Prior to this report, only indirect estimations of Zc
Huttenbrink and Hudde (1994) in one ear. (A) shows magnitude have been reported. Fig. 9 compares our measurement
and (B) shows the phase angle. of Zc to estimates made by Puria et al. (1997), Mer-

HEARES 3609 15-2-01

 R. Aibara et al. / Hearing Research 152 (2001) 100^109 107

Kringlebotn that Puria et al. used is greater than that

Fig. 8. Comparison of the mean GME and SVTF in the 11 ears.
(A) shows the magnitude and (B) shows the phase angle. The solid
line represents GME and the dashed line represents SVTF. Zero dB
for the SVTF curve was set at 2.1U1032 mm s31 /Pa at 0.05 kHz.
found in the present experiments (Fig. 7B).
Also shown in Fig. 9 is an estimate of Zc by Mer-
chant et al. (1996) from one ear. In that experiment,
they rst measured the acoustic input impedance of the
stapes and cochlea, Zsc , from the ratio of the sound
pressure measured near the stapes head and the stapes
head velocity. They then drained the cochlea and re-
peated the measurement. Zc was calculated from the
dierence in Zsc before and after draining the cochlea.
The shapes of the Zc curves are similar, however the
magnitude of Merchant et al. is higher than ours by as
much as a factor of ve. Their phase angle and the
present measurements are in relative agreement up to
5 kHz, above which, theirs decreases rapidly to 3180
between 9 and 10 kHz. One possibility is that their
measurement is simply from an ear that is a statistical
outlier. Another possibility that can contribute to the
dierences, although not completely account for it, is
that they measured stapes head velocity and we mea-
sured stapes footplate velocity.
Model predictions by Zwislocki (1965) for living hu-
mans predicts that Zc has a magnitude of 35.2 acoustic
G6 and is purely resistive (0 phase) in the 0.2 to 4 kHz

chant et al. (1996) and a theoretical prediction by Zwis-

locki (1965).
Puria et al. (1997) measured vestibule pressure but
used stapes velocity from the Kringlebotn (1988) model
(Fig. 7A) to estimate Zc . Their mean magnitude and
phase angle for Zc is similar to ours for frequencies
below 1 kHz. Above 1 kHz, their magnitude increases
more rapidly than ours and is higher by a factor of four
(12 dB) at 10 kHz. Between 1 and 10 kHz, their phase
steadily decreases to below 390 at 10 kHz, whereas in
the current phase measurements, UZc is nearly constant
at 0 to approximately 5 kHz. Above this frequency,
UZc decreases, leveling o near 345.
Figs. 6 and 7 provide an explanation for the dier-
ences between the present measurements of Zc at higher
frequencies and those reported by Puria et al. (1997).
First, Fig. 6 shows that above 2 kHz, the Puria et al.
GME magnitude is larger than the present measure-
ments. Second, Fig. 7 shows that above 1 kHz, the
stapes velocity of Kringlebotn that Puria et al. used is
lower in magnitude than our measurements. The com- Fig. 9. Comparison of measured cochlear input impedance, Zc , with
bination of these dierences leads Puria et al. to esti- previous estimates. (A) shows magnitude while (B) shows phase.
mate a higher impedance magnitude, above 1.0 kHz, Merchant et al. (1996) estimated Zc for one ear by measuring the
dierence in stapes cochlea acoustic input impedance after draining
than we do from the present data. Similar arguments
the cochlea. Puria et al. (1997) estimated Zc based on mean mea-
can be made for the dierences in UZc . The primary sured middle-ear pressure gain and the Kringlebotn (1988) model
reason why the UZc estimated by Puria et al. continues calculation of SVTF. Zwislocki's (1965) Zc is based on theoretical
to decrease is that the phase of the stapes velocity of parameters for cochlear input impedance.

HEARES 3609 15-2-01

108 R. Aibara et al. / Hearing Research 152 (2001) 100^109
Fig. 10. Eect of stapes footplate rocking ratio on measurement of
Zc . The rocking ratio is dened as the maximum dierence in dis-
placement between the anterior and posterior footplate to displace-
ment at the center of the stapes footplate. The peak in Zc magni-
tude is at 6.7 kHz near the frequency of the peak in the rocking
motion, and the two may be related.
frequency range. The mean of our Zc magnitude is
about half as large as Zwislocki's and the average dif-
ference from the mean phase angle is 4.7, with a max-
imum dierence of 17. Zwislocki (1975) contends that
postmortem changes in the basilar membrane, observed
in cats by Kohlloel (1971), decreases Zc by a factor of
two. Although our mean Zc corroborates Zwislocki's
contention, the question of postmortem changes in Zc
in the human is unanswerable at this time because of
our inability to make the measurement in the living
human ear.
4.4. Eect of stapes rocking motion on Zc
Fig. 10 shows that the peak in Zc seen at 6.7 kHz
correlates with a peak in the stapes footplate rocking
ratio measured by Heiland et al. (1999). The rocking
ratio of the stapes is dened as the ratio of the max-
imum dierence in displacement between the anterior
and posterior sites of the footplate compared with dis-
placement at the center of the footplate. The correlation
in the rocking motion and Zc suggests that measuring
Zc with a single point laser may not be reasonable at
higher frequencies because the footplate no longer
moves as a piston. Thus the peak in Zc at 6.7 kHz
may be due to contributions of non-piston motion in
SVTF.
4.5. Subject variability at threshold of hearing
Physiological measurements presented here show var-
iability across ears. Fig. 6 shows that the average S.D.
in GME is 5.5 dB for frequencies between 0.1 and
5 kHz, while between 5 and 10 kHz, the S.D. is 6.4 dB.
It was previously shown that the cochlear pressure is
HEARES 3609 15-2-01
the relevant input variable to the cochlea (Puria et al.,
1997). Subject variability in GME results in cochlear
pressure variability across subjects given the same ear
canal pressure.
Variability across subjects is also evident in hearing
threshold measurements. Fletcher (1995, p. 127) re-
ported a mean deviation of 5^9 dB from the average
monaural minimum audible eld measurements. More
recent data by Davis (1995, pp. 31^33) suggests that the
S.D. in threshold increases with both age and with fre-
quency. His measurements show that, in the clinical
environment and when grouped according to the better
ear, the S.D. for the 18^30 year old age group is 7^8 dB
for the 0.5^2 kHz range and 8^11 dB at 4 kHz. As age
increases S.D.s increase and could be as large as 20 dB
for the 71^80 year age groups.
The middle-ear variability reported here is less than
the variability in hearing threshold across subjects re-
ported by Fletcher and Davis. This is not surprising
given the variability in inner-ear sensitivity and to a
lesser extent noise due to more central mechanisms
adds to the variability in middle-ears across subjects
in determining thresholds.
5. Conclusions
Fig. 8 shows that either GME or SVTF produces
essentially the same curve when performed in the
same ears. Therefore, either measure, if carefully per-
formed, provides equivalent sound transmission fre-
quency shaping. Some dierences exist between our
GME data and that reported by others (Fig. 6). We
suspect that the dierences, which are not large, are
due to the high S.D.s present in middle-ear measure-
ments in human temporal bones as well as dierences in
measurement methods (Goode et al., 1987). The GME
data presented in this paper is based on the largest
number of human temporal bones reported to date,
which should minimize the eect of individual varia-
tions in middle-ear function. The GME magnitude
curve has a band pass characteristic of a resonance at
1.2 kHz and around a 6 dB/octave slope from 0.1 to
1.2 kHz and a 37 dB/octave slope above this frequency.
The SVTF curves are similar to those reported by
others using similar methods in intact ears (Gundersen,
1971 ; Vlaming and Feenstra, 1986 ; Gyo et al., 1987;
Kringlebotn, 1988 ; Goode et al., 1994). Measures of
middle-ear acoustic function made by comparing stapes
displacement before and after removal of the middle-ear
structures (Onchi, 1961; Kurokawa and Goode, 1995)
produce a dierent result. Direct stimulation of the
stapes with sound may produce a dierent type of vi-
bration than in the intact middle-ear. Studies where the
cochlea has been drained in order to make the middle-
 R. Aibara et al. / Hearing Research 152 (2001) 100^109
ear measurements are not considered reliable because of
the lack of a cochlear load.
Direct measurement of Zc , performed for the rst
time, reveals a at, resistive cochlear input impedance
of about 21.1 G6 from 0.1 to 5.0 kHz. The impedance
increases between 5.0 and 10.0 kHz with a maximum of
49.9 G6 at 6.7 kHz. At higher frequencies above 5.0
kHz and at lower frequencies below 0.2 kHz, the phase
angle becomes increasingly negative. The peak in Zc at
6.7 kHz corresponds to a peak in the same frequency
range as the stapes footplate rocking ratio (Heiland et
al., 1999). This suggests that in this frequency range,
footplate motion is particularly complex and measure-
ment of stapes velocity at one site, such as the center of
the footplate, may not be adequate to determine the
volume velocity. Below 5.0 kHz, however, as seen in
Fig. 8, velocity measured at the center of the footplate
is representative of footplate volume velocity, with only
a small dierence above 1.0 kHz.
While the mean data reported here provides us with a
general scheme of how the human middle-ear acousti-
cally interacts with the cochlea, the relationship of Zc
and middle-ear sound transfer function in individual
ears is not as clear. For example, since Zc is at and
resistive at 0.5^5.0 kHz, Pv should be related to Vs ,
which should be related to the umbo velocity produced
by a given TM. It would therefore be suspected that the
highest GME and SVTF would occur in ears with the
highest umbo velocities. This is true in some but not all
ears and is the subject of a future study.
Acknowledgements
This work was supported in part by grants from the
VA (GDE 0010ARG) and the National Institute of
Deafness and Other Communication Disorders of the
NIH (DC03085).
References
Bekesy, G., 1960. In: E.G. Wever (Ed.), Experiments in Hearing.
McGraw-Hill, New York, p. 745.
Brenkman, C.J., Grote, J.J., Rutten, W.L., 1987. Acoustic transfer
characteristics in human middle ears studied by a SQUID magne-
tometer method. J. Acoust. Soc. Am. 82, 1646^1654.
Dallos, P., 1973. The Auditory Periphery: Biophysics and Physiology.
Academic Press, New York, pp. 98^99.
Dancer, A., Franke, R., 1980. Intracochlear sound pressure measure-
ments in guinea pigs. Hear. Res. 2, 191^205.
Davis, A., 1995. Hearing in Adults. Whurr, London.
Fishler, H., Frei, E.H., Spira, D., 1965. Dynamic response of middle-
ear structures. J. Acoust. Soc. Am. 41, 1220^1231.
HEARES 3609 15-2-01
109
Fletcher, H., 1995. In: Jont B. Allen (Ed.), Speech and Hearing in
Communication, The Acoust. Soc. Am. edn. AIP Press, New
York.
Goode, R.L., Nakamura, K., Gyo, K., Aritomo, H., 1987. Comments
on `acoustic transfer characteristics in human middle ears studied
by a SQUID magnetometer method'. J. Acoust. Soc. Am. 82,
1646^1654.
Goode, R.L., Killion, M., Nakamura, K., Nishihara, S., 1994. New
knowledge about the function of the human middle ear: develop-
ment of an improved analog model. Am. J. Otol. 15, 145^154.
Gundersen, T., 1971. Prostheses in the Ossicular Chain. University
Park, Baltimore, MD.
Gyo, K., Aritomo, H., Goode, R.L., 1987. Measurement of the os-
sicular vibration ratio in human temporal bones by use of a video
measuring system. Acta Otolaryngol. 103, 87^95.
Heiland, K.E., Goode, R.L., Asai, M., Huber, A.M., 1999. A human
temporal bone study of stapes footplate movement. Am. J. Otol.
20, 81^86.
Huttenbrink, K.B., Hudde, H., 1994. Untersuchungen zur Schallei-
tung durch das rekonstruierte Mittelohr mit einem Hydrophon.
HNO 42, 49^57.
Kohlloel, L.U., 1971. Studies of the distribution of cochlear poten-
tials along the basilar membrane. Acta Otolaryngol. (Stockh.) 288
(Suppl), 1^66.
Kringlebotn, M., Gundersen, T., 1985. Frequency characteristics of
the middle ear. J. Acoust. Soc. Am. 77, 159^164.
Kringlebotn, M., 1988. Network model for the human middle ear.
Scand. Audiol. 17, 75^85.
Kurokawa, H., Goode, R.L., 1995. Sound pressure gain produced by
the human middle ear. Otolaryngol. Head Neck Surg. 113, 349^
355.
Lynch, T.J., Peake, W.T., Rosowski, J.J., 1994. Measurements of the
acoustic input impedance of cat ears: 10 Hz to 20 kHz. J. Acoust.
Soc. Am. 96, 2184^2209.
Merchant, S.N., Ravicz, M.E., Rosowski, J.J., 1996. Acoustic input
impedance of the stapes and cochlea in human temporal bones.
Hear. Res. 97, 30^45.
Onchi, Y., 1961. Mechanism of the middle ear. J. Acoust. Soc. Am.
33, 794^805.
Puria, S., Allen, J.B., 1992. SYSid: audio-band measurement and
analysis system. J. Acoust. Soc. Am. 92, 2469.
Puria, S., Allen, J.B., Elko, G.W., Jeng, P.S., Kirkegaard, D.L., 1993.
Measurements and analysis with SYSid. J. Acoust. Soc. Am. 93,
2556.
Puria, S., Peake, W.T., Rosowski, J.J., 1997. Sound-pressure measure-
ments in the cochlea vestibule of human cadaver ears. J. Acoust.
Soc. Am. 101, 2754^2770.
Rubenstein, M., Feldman, B., Frei, E.H., Spria, D., 1964. Measure-
ment of stapedial-footplate displacements during transmission of
sound through the middle ear. J. Acoust. Soc. Am. 40, 1420^1426.
Ruggero, M.A., Rich, N.C., Robles, L., Shivapuja, B.G., 1990. Mid-
dle-ear response in the chinchilla and its relationship to mechanics
at the base of the cochlea. J. Acoust. Soc. Am. 87, 1612^1629.
Vlaming, M.S.M., Feenstra, L., 1986. Studies of the mechanics of the
normal human middle ear. Clin. Otolaryngol. 11, 353^363.
Zwislocki, J.J., 1965. Analysis of some auditory characteristics. In:
Luce, Bush, Galanter (Ed.), Handbook of Mathematical Physiol-
ogy. Wiley, New York, pp. 3^46.
Zwislocki, J.J., 1975. The role of the external and middle ear in sound
transmission. In: D.B. Tower (Ed.), The Nervous System, Vol. 3,
Human Communication and its Disorders. Raven Press, New
York, pp. 44^55.