Home Search Collections Journals About Contact us My IOPscience

On the stationary-wave method of measuring sound-absorption at normal incidence

This content has been downloaded from IOPscience. Please scroll down to see the full text.

1926 Proc. Phys. Soc. 39 269

(http://iopscience.iop.org/0959-5309/39/1/325)

View the table of contents for this issue, or go to the journal homepage for more

Download details:

IP Address: 139.80.123.49
This content was downloaded on 02/10/2015 at 16:50

Please note that terms and conditions apply.

 The Measurement of Sound-A bsorption. 269

XX1.-ON THE STATIONARY-WAVE METHOD OF MEASURING SOUND-

ABSORPTION AT NORMAL INCIDENCE.*
BY E. T. PARIS, DSC., F.rmlrtst.P.
Received Janzcavy 12, 1927.

ABSTRACT.
A description is given of a paratus employed for measuring coefficients of sound-absorption
by the stationary-wave methoif The apparatus differs from that used by earlier workers in the
use of (1) a small tuned hot-wire microphone for determining relative pressure-amplitudes in the
sound-waves ; (2)the employment of a steady valve-driven source of sound with arrangements
sfor maintaining the strength at a constant value ; (3) the screening of source and experimental
pipe from disturbances due to the movements of the observer. B y the employment of a certain
procedure the relation between the response of the microphone and the amplitude of the pressure-
variation in the sound-wave is eliminated. Some examples of the employment of the apparatus
for determining the coefficients of absorption at normal incidence of acoustic plasters and hair-
felt are given.
1. INTRODUCTION.
T HE work described in this Paper arose in connexion with investigations initiated
by the Building Research Board of the Department of Scientificand Industrial
Research into the acoustic properties of certain materials. These investigations
were concerned with testing the sound-absorbing properties of various plasters and
other materials which could be used for lining the walls of rooms and auditoria for
the purpose of reducing the reverberation in cases where it was in excess of that
required for good acoustical conditions. The tests conducted by the Building
Research Board were made by Sabine’s reverberation method, and the results were
expressed as percentage reductions in the period of reverberation produced by the
introduction of 100 square feet of the material under test into a room of a certain
size and shape.?
For some purposes, however, it was desirable that a method should be available
for which considerably less material would be required than was needed for Sabine’s
method. It was, for example, desired to test a series of acoustic plasters in order
to trace the effect of varying the proportions of the materials used in their making,
and the necessity for preparing the comparatively large quantities needed for a
reverberation experiment was an obvious handicap.
For this reason experiments were begun in 1923 with the object of establishing
some method whereby a test of sound-absorption could be carried out on a specimen
having an area of not more than one or two square feet. The method first suggested
* The work described in this Paper was carried out in the Acoustical Research Section of
the Air Defence Experimental Establishment, Woolwich, with the assistance of the contribution
made t o the funds of that Section by the Department of ScientiGc and Industrial Research on
the recommendation of their Physics Research Board.
t These experiments are described in “ A n Interim Note on Acoustic Bxperiments carried
out for the Government of India ” by Messrs. Barnett and Glanville, Building Research Board
Paper No. 181, January, 1924.
U2
 270 Dr. E . T. Paris 0%

was one in which resonance boxes were made up and lined with the materials to be
tested, the sound-absorbing properties being deduced from the sharpness of resonance
exhibited by the boxes. Several experimental and theoretical difficulties arose,
however, which could not be readily overcome, and it was decided to seek for an
alternative method of procuring the desired information.
This alternative was found m what is here called the ''Stationary-wave Method "
-a method which had previously been used for the study of sound-absorption by
J. Tuma (1902), F. Weisbach (1910) and Hawley Taylor (1913). Some notes on the
work of these investigators are given in 3 3. The theory of the method is very simple
and is briefly as follows. The specimen to be tested is supplied in the form of a flat
disc, and (with suitable backing) is used to close one end of a cylindrical pipe (Fig. 1).
The other end of the pipe remains open, and opposite to it is placed a source of sound
of constant pitch and loudness. Sound-waves from the source travel down the
pipe and are reflected back from the specimen. If the latter were a perfect reflector,
the amplitudes of the

e
cELccmDI
SOURCE incident and reflected
waves would be equal
and a stationary wave
w..,oLErs ez,L~IION W T F
~ O~.IIO* 04 smm"mv
*LYE
would be formed inside
the pipe, with no pres-
sure-variation at the
s p ~ ~ / n ~ loops
~ and maximum pres-
sure-variation at the
nodes. If, however, the
specimen absorbs part
, M ~ O , + W L ~ T EM F L E X ~ O N w / m FORMAT,OION s
l l ~ l i AND
! ~ NfN'w of the incident sound-
FIG.l.-DIAGRtLVS ILLUSTRATISG T E E STATIOXARY-WAVE
energy, the amplitude of
METHOD OF MEASURING SOUND-ABSORPTION. the reflected wave will
be less than that of the
incident wave, so that instead or a stationary wave being formed inside the pipe
there will now be a series of maxima and minima.
If the ratio of the pressure-amplitude at a maximum to that at a minimum
be determined, say alb, the coefficient of absorption can be calculated by means
of the formula (see f 2 )
A
a=2+a,b+bla . . . . . . . . . . (1.1)
It may be noted here that the "absorbed " sound is taken to be that part of the
incident sound which is not reflected. Thus, if E6 is the energy-flux in the incident
waves and E, that in the reflected waves, then the rate at which energy is being
transmitted across the reflecting surface (the waves being incident normally) is
E<-E, ergslcm.2 per sec., and the coefficient of absorption is
Et -E,
a=- 5 . . i . . (1.2)
Ei
.. . . .
In the case of a porous reflecting surface, therefore, the absorbed sound includes
both that which is dissipated into heat by Viscous forces acting in the pores and that
which is intromitted as wave-motion into the material itself.
 The Measurement of Sound-A Fsorpion. 271

The results so far obtained by this method indicate that it is a satisfactory one
for testing the sound-absorptive properties of materials at frequencies below 700
vibrations per second, when only small specimens are available. The gresent Paper
contains a description of the apparatus which was employed by the author and
some results obtained by its use. The apparatus differs from that used by earlier
workers mainly in the following particulars :-
(1) The employment of a small tuned hot-wire microphone for determining
the ratio of maximum to minimum pressure-amplitude inside the pipe. This insku-
ment is very simply constructed and easy to use, and requires no accessory apparatus
beyond a battery, galvanometer and a few resistance boxes. As it is a tuned instru-
ment, measurements of absorption at a definite wavelength can be made without
the necessity of having a pure-tone source of sound.
(2) The employment of a valve-oscillator and loud-speaker as a source of sound,
together with a special arrangement described in 5 4(c) for maintainilzg the source
at a constant strength.
(3) The use of a sound-chamber t o enclose the source of sound and the open
end of the pipe as described in 5 4 (a). This is important, as the output of the source
may otherwise be influenced by the movement of the observer.
Emphasis is also laid on the use of the procedure (due t o Tuma) described in
8 6 , by which the relation between the pressure-amplitude of the sound and the
response of the microphone is eliminated.
The apparatus described was designed primarily for employment with sound
of a frequency of 512 vibrations per second, but a few measurements have also been
made at 380 and 650 vibrations per second.
The method measures the coefficient of absorption at normal incidence under
the condition that transmission is prevented as far as possible by backing the
specimen with a suitable material.

5 2. THEORYOF THE METHOD.

It is assumed that the incident and reflected sounds are plane waves travelling
in opposite directions inside the pipe. In practice this condition can be secured
by choosing a pipe of suitable diameter and length and axranging the source at an
appropriate distance from the mouth of the pipe. Moreover, a preliminary test
can always be made t o ensure that a stationary wave of the required type exists
inside the pipe when a good reflecting substance is used to close the distant end.
Let the axis of the pipe be parallel to the axis of x , and let the surface of the
specimen to be tested lie in the plane x=O. The potential of the incident wave
(supposed travelling in the direction x positive) can be represented by
'p=A COS k ( a t - ~ ) . . . a. . . . . (2.1)
After reflexion there will in general be a wave of reduced amplitude travelling
in the negative direction of x . This can be represented by
'p'=B COS k(at+x+s) . . . . . . . . (2.2)
In practice, of course, repeated reflexions take place both at the open or "source "
end of the pipe and a t the surface of the specimen ; but this does not necessitate any
modification of the above expressions, since the net effect, when a steady state is
reached, is that of a single train of incident and a single train of reflected waves.
272 DY.E. T.Paris on
The total potential within the pipe is
Q=cp+cp’ . . . . . . . . . . . (2.3)
If the reflecting surface at x=O were perfectly rigid, the condition to be satisfied
by 0 at x=O would be

expressing the condition that there is no motion in the plane x=O, so that no vibra-
tional energy can be transmitted into the reflecting substance. Under these cir-
cumstances, the amplitudes of the incident and refkcted waves are equal ( A = B )
and s=O. The result is a stationary wave
@ = 2 A cos kx cos kat . . . . . . . (2.5)
within the pipe.
If part of the energy of the incident sound is lost at reflexion we have
0 - A cos k(at -x)+B cos k(at+x+e) . . . . . . (2.6)
1 E
which, by a simple transformation (t’=t+- -, X ’ = X + ~ E ) , may be written
2a
@=A cos R(at’ -x’) +B cos k(at‘+x’)
= ( A +B) cos kx’ cos kat‘
+ ( A -B) sin kx’ sin kat’ . . . . . (2.7)
so that the motion within the pipe can be regarded as being due t o two superimposed
stationary waves, one of amplitude ( A f B )and the other of amplitude ( A -B),the
nodes and loops of the first being one quarter of a wavelength distant from the
nodes and loops of the second. The result is that there is within the pipe a series of
positions of maximum and minimum pressure-variation proportional respectively
to (A+B) and ( A -B), spaced at distances of one quarter of a wavelength. It is
the ratio of the amplitudes at these maxima and minima which is found in the
experiment,
To calculate the coefficient of absorption we note that the energy-flux in the
incident wave is proportional to AB, and the energy-flux in the reflected wave is
proportional to B2. The fraction of the incident sound-energy lost at reflexion is
thus proportional t o ( A 2-Bg)/AB, and this is, by definition, the coefficient of absorp-
tion (a). Let the observed ratio of the maximum amplitude to the minimum
amplitude be a/b. Then to find a in terms of ratio alb we have, since
a A+B
-=-
b A-B’
. . . . . . . . . . . (2.8)

-
A2-B2 4ab
a= A 2 ---- . . . . . . . . (2.81)
4
a= 2+a/b+ b/a
. . . . . . . . . (2.82)
This last expression is that given by Hawley Taylor, and is the most convenient
form for computing a from the observed ratio alb.
 TJze Measurement of Sound-Absorption. 273
Alternatively, let-
A =(l+m)B, . . . . . , . . . . (2.9)
where m is positive, since A >B. Then-
@=B COS k ( d -x’)+B COS k(~f’+x’)+mBCOS k ( ~ t ’ + x ’ )
=2B cos kx’ cos kal’+mB cos k(at’+x’). . . . . , . .. . . . (2.91)
Alternatively, therefore, the motion within the pipe may be regarded as being that
due to a stationary wave of amplitude 2B, on which is superimposed a progressive
wave of amplitude mB. The loops and nodes of the stationary wave are a t the
positions of maximum and minimum pressure-variation which occur when an
absorbent substance is placed at the end of the pipe.
Also, according to (2.91), the positions of the maxima and minima differ from the
position of the nodes and loops which would occur if a perfect reflector were used to
close the pipe, by a distance + E . No special attempt was made to measure this shift
in the experiment here described, but it may be of interest to note that it was of the
order of 1 cm. for a substance with an absorption-coefficient of about 0.3 at 612
variations per second. The shift was in the direction x positive, and thus indicated
a virtual node of the stationary vibration in (2.91) about 1cm. within the surface of
the absorbing material. It is possible that the measurement of this shift might be
of service in investigating the mechanism by which absorption takes place in some
materials. I n the case of resonant reflexion it might happen that the virtual node
would be in front of the reflecting surface.

$ 3 . PREVIOUS
EXPERIMENTS WITH THE STATIONARY-WAVE METHOD.
The first experiments with the stationary-wave method appear t o have been
made by J, Tuna,* in Vienna, in 1902. Later experiments were made by F. Weis-
bach (191O)tand Hawley Taylor (1913).$ Taylor appears to have been unaware of
the work of Tuma and Weisbach.
Tuma’s observations were entirely aural. By means of listening tubes, he
observed the loudness of the sound in the maximal and minimal positions in a pipe
closed by absorbent materials and then found positions of the same loudness in a
stationary-wave. Since the amplitude of the pressure-variation in a stationary-
wave is proportional to a simple sine or cosine function, he could in this way find
the ratio of the maximum to the minimum amplitude when absorption was taking
place. Thus suppose the positions corresponding to minimum and maximum
loudness were distant x1 and x 2 from the nearest loop in the stationary-wave, then
the ratio of maximum t o minimum pressure-amplitude would be sin kx,/sin hx1, where
k = 2 n / t . From this ratio a quantity a was calculated such that (1 -a) was the
ratio of the reflected t o the incident amplitude. The actual results obtained were
very meagre, and the interest of the Paper lies in the description of the method.
* “ Eine Methode zur Vergleichung von Schallstarkungund zur Bestimmung der Reflexions-
fahigkeit verschiedenen Materialien,” J. Tuma. Sitmngsber. d. K. Akad. d. Wissenschaften.
Vol. 111,Pt. 2A, pp. 402-410 (1902).
t “ Versuche iiber Schalldurchlassigkeit, Schallreiiexion und Schallabsorption,” F. Weis-
bach, Ann. d. Phys, Vol. 33, pp. 763-798 (1910).
$ “ A Direct Method of Finding the Value of Materials as Sound Absorbers,” Hawley
0.Taylor, Phys. Rev., Vol. 2,pp. 270-287(1913).
 274 Dr. E. T . Paris o n
The principle of comparing amplitudes by reference to a stationary-wave has been
retained in the present work, the method here described being essentially that of
Tuna, with the ear and listening tube replaced by a hot-wire microphone.
Weisbach made experiments with a wooden " resonance-pipe,'' 238 cm. long,
25 cm. diameter, and walk 4 cm. thick. The source of sound was an electrically
&ven tuning-fork, mounted on a resonance-box, and in order to ensure that the
source was steady, observation of the amplitude of the prongs was made b y means
of a microscope. The amplitude of the sound in the pipe was measured by means of
a magnetophone (mounted on a slider), this being connected with a string galvano-
meter, the deilexion of which was taken to be proportional to the amplitude of the
soma.
In the experiments of Taylor, observations were conducted with a wooden
flue," 115 cm.long, and of square cross-section (9 cm. x9 cm.), the specimen t o
be tested being held in place by means of a wooden cap. To explore the amplitude
of vibration in the flue use was made of a Rayleigh disc instrument, to which the
sound from inside the flue was lead by means of a long glass tube. As the Rayleigh
disc involves the employment of a delicate suspension, this instrument could not be
moved during an experiment, and it was therefore necessary, when comparing
amplitudes at a m a x i " and a minimum, to move the whole of the flue together
with the source of sound. T%s made the method very cumbersome, and it may be
noted that the movement of the source during an experiment may introduce errors
into the measurements, basmuch as the output of the pipe may be influenced by
its position in the room. The source of sound was a stopped organ-pipe, the note of
which was purified by means of a "tone-screen," consisting of a number of resonators
arranged at the open end of the flue, each resonator being designed to prevent the
passage of one of the h m o n i c s of the organ-pipe note. The absorption-coefficients
were calculated from the deflexionsshown by the Rayleigh disc, these being assumed
proportional t o the squwe of the amplitude of the sound. Taylor found that for
1 inch of hair-felt backed by wood, the coefficient of absorption was 0.61 at 500
vibrations per second.
0 4. DESCRIPTION OF APPARATUS.
The description of the apparatus employed can be conveniently divided into
t h e e parts dealing with-
(U) The experimental pipe and the mounting of the specimen ;
(6) The hot-wire microphone and its connexions ; and,
(c) The source of sound.
(a) The Ex$erimentul Pi$e and the Method of Mounting Specimens.-The dimen-
sions of the pipe &ally adopted after some preliminary trials were 30.5 cm. diameter
and 22Q cm. length. This pipe was made from two glazed earthenware drainpipes
cemented together, and was mounted, with its axis horizontal, inside a wooden box
supported on four steel legs set in concrete. The pipe was insulated from its sup-
ports by pieces of hair-felt, and the box was lined with the same material. The
object of the mounting was to prevent the communication of vibrations due to traffic,
db other local' causes, to the walls of the pipe and thence to the contained air and the
microphone.
The general arrangement of the pipe and its hsusing is shown in Fig. 2. One
 pi<:.3 -vll:.\v tIV l ~ . < i ~ l . I < l \ \ i ~\.I S. T l ' i l ~ l ~I.. M < l \ l
OIITSII)I< I,.\liOR.\'~OktY.

J:il;. 4 . - .Sl*I,.(I\II,.S Ill: .\s .\<Yl( \'rl<' l'l,\.,rl I<

hfouxrisu IUH T I ~ S T I S C ((liarllc.1c.r of Pl;istcr
Disc = 12 ins.).
[ T O/(r<-e p q e 275
 The Measurement of Sound-Absorfition. 275
end of the pipe passed through the wall of the laboratory into a sound-chamber, in
which was situated the source of sound, the pipe being insulated from the wall b y
means of felt. The object of the sound-chamber {approximately 7 ft. by 9 ft. by
7 ft.) was to prevent the output of the source from being affected by the movement
of the observer in the laboratory. At the other end of the pipe, the specimen, in
its mounting, was held firmly in place between the end of the pipe and the detach-
able cover a t the end of the box (see Fig. 2). The detachable cover was held in
place by four bolts and wing-nuts, and could easily be removed and replaced by hand.
The mountings for the specimens were made of discs (37 cm. in diameter), to
which were screwed rings measuring 37 cm. externally, and 30.5 em. internally.
Both disc and ring were made from three layers of five-ply wood glued together.
The specimens were provided as discs about 30.5cm. in diameter, and these were
fitted inside the rings, the surface of the backing provided by the disc being first
covered with enough cement to fill up any inequalities in the back of the spechen.
A. Scale Read Here. '
B. Sound-Chamber
E
C. Loud-Speaker.
D. Slider.
S . Wall of Laboratory.
F. Felt-Lined Box Exlosing
Pipe.
G. Earthenware Pipe.
H. Microphone.
I. Specimen.
J . Thin Washer of Rubber.
IC. Backing to Specimen
(3 layers of 5 . ~ 1 ~ ) .
L. Packing (1-inch Wood).
M. Cover (held by Bolts and
Wing Nuts.)
N. Felt.
0. Steel Supporting Legs set
FIG. 8.-ARRANGEMEST OF EXPERIMESTAL P I P E , S O V R C E in Concrete.
AND SPECIMEF (BOT TO SCSLE). P. Ground-Level.
Q. Concrete.
Any crevices between the specimen and the ring were filled in with. plaster of Paris
which was allowed to set before a test was made.
The mountings, made as described, were found to keep their shape very well,
and were in every way satisfactory.
When fixing a specimen in place for a test a washer of thin leather or rubber was
placed between the end of the pipe and the ring of themounting t o ensure an airtight
fit.
A view of the pipe in its housing as seen from the outside of the laboratory is
shown in Fig. 3, and a specimen of an acoustic plaster in its mounting ready for
testing is shown in Fig. 4.
(b) The Hot-wire Micro$hone.-The microphones used in these experiments
did not differ in any essential detail from the ordinary selective hot-wire microphone,*
but, in order that they should offer as little obstruction as possible to the passage of
the sound-waves along the experimental pipe, they were made to smaller dimensions
* V i & Tucker and Paris, Phil.Trans., A, Vol. 221, pp. 389-430 (1921).
276 Dr. E. T . Paris on
than is customary for instruments tuned to the frequencies employed in the experi-
ments. The microphone used for measurements at 512 vibrations is shown in Fig. 5.
Its overall length is 6 cm., and diameter of the holder 3.4 cm. The orifice is 1cm.
long and 045cm. in diameter (internal), the conductance being 0.114cm. The
container was made in two parts, which could be slid over one another, so that the
microphone was tunable over a moderate range (approximately 460 to 560 vibrations
per second). The damping factor at 512 vibrations per second was found to be
81 seconds-1 (see Appendix I).
The microphone was carried in the experimental pipe at the end of a long wooden
slider (see Fig. 2), which passed through a slot in the wall of the sound chamber. At
the end which protruded through the wall of the sound chamber, the slider was pro-
vided with a millimetre-scale, so that the movement of the microphone in the experi-
mental pipe could be recorded by an observer from outside the sound-chamber.
The orifice of the microphone was on the axis of the pipe.
The electrical connexions of the microphone are shown in Fig. 6. The response
was measured by a bridge method, the bridge being of the “ battery ” type with
compensating microphone. This type of bridge is preferable to the ‘ r Wheatstone ”
bridge on account of its greater sensitivity, while the use of a compensating micro-
phone makes the balance very steady. The compensating and active microphones
were of the same type, but the former was, of course, shielded from the sound.
RI A. Slider.
B. Loud-Speaker.
C. Active Microphone.
D. Control Grid.
E. Experimental Pipe.
F. Specimen.
G. To Oscillator.
H. Balancing Microphone.
I. Milliammeter.
FIG. B.-ELECTRICA& COS3~XIONS. J. Micro-Ammeter.

The grid of the microphone carried a heating current of about 2 8 milliamperes,

and had a “ hot ” resistance of about 320 ohms. When exposed to the sound the
fall in the resistance was anything up t o 50 ohms, depending on the position of the
microphone within the tube and the output of the source of sound. This fall in
resistance was measured by introducing resistance in the three-dial resistance box
RI, reading from 0.1 ohm up to 100 ohms (Fig. 6), until the balance of the bridge
was restored. The bridge was initially balanced by means of the sliding resistance R,.
(c) The Sozcrce of Sound-The provision of a satisfactory source of sound is a
matter of importance, as the source must be capable of maintaining an output of
constant amplitude and pitch for considerable periods. The oscillator and loud-
speaker used in the present experiments were found to be satisfactory in this respect
for frequencies between 300 and 1,000 cycles per second.
During the course of an experiment it was necessary t o turn the source of sound
on or off several times. This was done by means of a switch, which could be used
to change over the output terminals of the oscillator from the loud-speaker to an
equivalent resistance.
It was found, however, that the amplitude of the sound affecting the microphone
in the experimental pipe was often different before and after the source had been
 6. --HOT- Wini:, MICROPHONR
Fit:.
TLWI?U TO 612 VIBRNS. 1'1IK SEC.

1'0 face page 2761

 T h e Measurement of Sound-A bsor@tion. 277
switched off, and that adjusting the current in the oscillating circuit was not a satis-
factory method of correcting for this variation. To overcome this difficulty and to
ensure that the strength of the source remained constant throughout the period
during which a set of observations was being made, the following device was made
use of. A hot-wire grid of the usual pattern was mounted in the narrow end of the
loud-speaker just above the diaphragm. This grid was connected to a battery-
bridge of the same pattern as that used for the grid in the active microphone. At
the commencement of a set of observations, the oscillator was adjusted so that this
control grid showed a resistance change of about 10 ohms, and this change of resis-
tance was maintained throughout the experiment by making small adjustments to
the oscillator. This could be done without difficulty, to within about 0.05 ohm,
and since experiment shows that when the resistance change in a grid is about 10
ohms, a further change of 1 ohm corresponds to a change in the amplitude of the
sound of about 1.9 per cent., it follows that the amplitude of vibration of the source
could be maintained constant to within about 1 part in 1,000.
$5. METHOD OF OBSERVATION.
The procedure adopted'when taking observations was as follows. The micro-
phone was first of all tuned as nearly as possible t o the frequency of the note given
by the oscillator by altering the volume of the container, a final tuning being made
by a slight adjustment of the note of the oscillator. The specimen t o be tes'ted was
next mounted in its place, the microphone balanced up, and the source of sound
turned on. The microphone was then moved along the pipe by means of the slider
shown in Figs. 2 and 6 until the position of the minimum nearest to the specimen was
found. The resistance-change suffered by the microphone-grid in this position was
then noted. The microphone was then moved along (away from the specimen)
until a maximum was reached. The resistance-change in this position was also
noted, and the specimen was then moved.
Let pl be the resistance-change at the minimum, and pz that at the maximum.
It is required to find the ratio of the pressure-amplitude at a maximum t o that at a
minimum (Le,, a/b in the notation of § 2) from the observed resistance-changes p,
and pB.
To accomplish this the experimental pipe was closed with a disc of a good
reflecting substance. I n the earlier experiments a piece of polished teak l i n c h
thick was employed, but later this was replaced by a disc of brass + in. thick, backed
by three layers of five-ply wood. The position of the loop nearest to the reflector
was found by sliding the microphone along the pipe. I n this position, if the reflexion
were perfect, there should be no change in the resistance of the microphone when the
source is switched on or off. In practice the reflexion is, of course, never quite
perfect, and sometimes a small change was discerned but this never exceeded
0.005 ohm a t 512 vibrations per second. This residual effect at the loop appeared
t o be too small to be taken into consideration and it was ignored.
One method of finding a / b is to displace the microphone from the loop until it
suffers a resistance change pl, the distance through which has been moved (say,yl)
being noted It is then further displaced until the resistance change is pz, and the
distance through which has been moved (say, yz) again noted. The pressure-
amplitude in a stationary-wave being proportional to sin ky, where y is the distance
 278 DY.E. T . Paris on
from a loop and k=%nlA, the pressure-amplitudes producing the resistance-changes
p1 and pz must be proportional to sin ky, and sin ky2, so that

- sin kyp
a=-
b sin kyl’ - . . . . . . . . (5.1)
and Q can be calculated from (1.1).
I n practice, however, it is easier to adjust the strength of the sound, so that
ky2=900, and hence, writing y instead of y1
a
,-=(sin ky)-l, . , . , . . . . (5.2)

4
and a= (5.3)
2+cosec kyfsin ky‘ .
To ensure that ky2=900,the microphone was first placed a t a node in the stationary
wave, and the strength of the source adjusted until the resistance change suffered
by the grid was p2. The microphone was then moved into a loop position and the
displacement (y)determined which was necessary to obtain a resistance-change pl.
Since y is in practice often less than 1 cm., the microphone was first moved towards
the specimen until the resistance change p1 occurred, and then away from it until
the same change was repeated, the distance between the two positions being 2y.
The half-wavelength of the sound was found by observing the position of the
second loop from the specimen.

5 6. CORRECTION FOR THE EFFECT OF THE PRESENCE OF THE hfICROPHONE IN THE

EXPERIMENTAL PIPE.
It has so far been assumed that the effective pressure-variations actuating the
microphone are equal to the pressure-variations which would exist in the pipe if the
microphone were absent. In general, however, this will not be the case. If the
pipe be narrow, the hot-wire microphone and the pipe together form a well-known
type of double resonator. On the other hand, it is obvious that if the pipe be wide
and the microphone very small, the coupling between pipe and microphone will be
very loose, and the microphone serves merely to indicate what goes on within the
pipe.
The theory of the double resonator can be utilised to calculate the effect of the
presence of the microphone in the experimental pipe. Details of the calculation
are given in Appendix 11, where the particular case of the pipe and microphone used
in these experiments for observations at 512 vibrations per second is worked out in
full. The results show that for all practical purposes the effects of reaction may be
ignored. The greatest effect is at a node, where the amplitude recorded by the
microphone is 2 per cent. less than it would be if there were no reaction. Midway
between a node and a loop the effect is about 0.9 per cent., and this decreases to zero
at the loop.
It may be noted that the effects of reaction, small as they are, are largely
eliminated by means of the procedure described a t the end of the last section, in
which, after the resistance changes pI and pz have been determined with the speci-
 The Measurement of Sound-Absorption. 279
men in position, the microphone is placed a t a node in the stationary wave and the
strength of the source adjusted until the change pz again occurs.
In order t o see how this procedure eliminates the effect of reaction, we may
refer to equation (2.91)of 9 2, which shows that, when the specimen is in position, the
potential within the pipe may be regarded as being due t o a stationary wave of
amplitude 2B, and a progressive wave of amplitude mB. The effect of reaction in
the case of a pure stationary wave is discussed in Appendix 11, and, if the small
difference between x and x' in (2.5) and (2.91) be disregarded, the same theory is
applicable to the stationary wave of amplitude 2B occurring in (2.91). The case of a
progressive wave is dealt with in Appendix 111, where it is shown that the effective
amplitude of the potential operating on the microphone, corresponding to an nndis-
turbed amplitude mB,due to a progressive wave is

where a is the velocity of souqd in air, c is the conductance of the orifice of the micro-
phone, h is the damping-factor of the microphone, and o is the cross-sectional area of
the pipe.
I n the first part of the experiment, with the absorbing material in place, the
resistance changes p1 and pz are observed corresponding to actzlal amplitudes of
oscillation in the neck of the microphone of, say, bo and a,. The corrected amplitude
ac
corresponding to p1 would be Ob,, where 0=1+ -. Now a, is equal to the s u m of
4ho
the amplitudes of the stationary wave and the progressive wave, say, ao'+b,. Hence
the corrected value of a, is 'pa,'+Ob,, where 'p is the ratio of the corrected to the
uncorrected amplitude a t a node-that is, 1.019 (see Appendix 11).
I n the second part of the experiment the sound is adjusted until the resistance
change p1 occurs a t a node in the stationary wave. This therefore again corresponds
to an actual amplitude of ao=ao'+60, but the corrected value is now cp(a,'+b,).
We wish t o find the ratio of the corrected amplitudes at maximum and minimum;
that is, ('pa;+8b,)/0b0, by reference to a stationary wave. By taking the reading
for pz a t a node we make the correction for reaction nearly the same as it was at the
maximum in the first part of the experiment. For the corrected amplitude corre-
sponding to pz is 'pa:+'pb,, which is to be compared with 'pa,'+Ob,, that is, we
have to compare a,'+b, with a,'+-
ebo. Now rp=l.O19, approximately, and 8=1.016
'p
1
(see Appendix 111),so that, remembering that bo is seldom as great as -U,,', and
5
frequently less than -101 a,', we see that the e€fectof reaction a t the node is eliminated
for all practical purposes.
It remains to consider the error in measuring the amplitude corresponding to pl.
As shown by the figures given in Appendix 11, the error involved in matching a n
amplitude in the stationary wave when ky is small (as it is for the amplitnde corre-
sponding to el) is quite negligible, while Appendix 111 shows that the effect of
 280 Dr. E. T.Paris on
reaction on the measurement of the amplitude at a minimum can be corrected for by
ac
means of the factor r3=1+= A more exact formula for the absorption-coefficient
than (8.3) is therefore

where 8, for the apparatus used in the present experiment at 512 vibrations per
second, is equal t o 1.016. Here again, however, the effect is negligible for most
practical purposes, as shown by the figures in the following table :-
I kY I c( calculated from (5.3). 1 m calculated from (6.1).

O0 0 0
50 0.295 0.299
loo 504 510
15' 653 659
20a 760 766
250 835 841
30° 888 887
-
1:1<;, 8.
~ ~ l ~ , ~ ' ' I '' l ~ l, lSi ~ o l ~ ~
( ~I cl lI l l ' s ' r l c'l*ll'i;.
: x 12.

[TO face Page 280

The Measuremennt of Souutd-Absovption. 28r

m
ILL
Q
e

-I I I I
 282 . Dr, E. T. Paris on
in a plane perpendicular to the surface which would normally receive the sound.
A photograph of a section in a plane parallel to the sound-receiving surface revealed
a similar structure, the average proportion of the area occupied by the pores being
between 30 and 36 per cent. in each case. The tile is one inch thick.
The details of four determinations of the coefficient of absorption of a sample
of this tile are given in the table on page 281.
The figures given in the second column of the table are the ohmic changes which
were observed in the resistance of the control grid mounted in the throat of the loud-
speaker. By removing this grid and placing it in the holder of a microphone tuned
to 511 vibrations per second and calibrating it in a stationary wave, the relative
strength (c.c. per sec.) of the source employed in each experiment could be discovered.
The figures are given in the third column.
The figures in the seventh, eighth, tenth and eleventh columns are the scale-
readings of the slider when the position of the microphone was such that the resistance-
change in the grid was equal to pl. Settings were made in the vicinity of the loop
nearest the specimen (readings A and B), and then, as a check, at the second loop,
half a wave-length nearer the open end of the pipe (readings C and D). Generally
two or three settings were made in each position, the scale being read t o the nearest
half millimetre, and the mean of all readings being used to find 2y. The magnitude
of the error in a likely to arise from inaccuracies in the measurement of 2y are dis-
cussed in Appendix IV.
A number of other determinations, some made with a different sample, gave
the same mean value (0-26) for the coefficient of absorption a t normal incidence
of this type of tile a t or near 512 vibrations per second.
This value for the absorption-coefficientis considerably smaller than that quoted
for a similar kind of tile by F. R. Watson*-namely, 0.36 at 612 vibrations per
second. It is to be noted, however, that the coefficient quoted by Watson was
obtained b y a reverberation method.
(b) Acoustic Plasters.-The following table contains the results of tests made
on specimens of acoustic plaster submitted by the Building Research Board in 1924.
Determinations were made of the absorption coefficients of each specimen a t three
frequencies-viz., 380, 512 and 650 vibrations per second. Each value given in the
table is the mean of four or five separate determinations.

Description of specimen.
I Coefficient of absorption.

Acoustic Tile 2A ............ 0.13 0.26 0.31

Acoustic plaster M ......... 0.10 0.20 ...
Acoustic plaster 5 A (3f parts slag t o 1 part
magnesite) ............ 0.18 0.25 0.27
Acoustic plaster BA (44parts slag t o 1 part
magnesite) ............ 0.18 0.27 0.30
Acoustic plaster 7 A (&Is parts slag to 1
part magnesite) ......... 0.21 1 0.31 0.36

All specimens tested were 1 inch thick, and were backed by a thin layer of
cement, and either 1 inch of teak wood or three layers of five-ply.
* "Acoustics of Buildings," p. 25.
 The Measwemint of Sound-Absorption. 283
The acoustic tile 2A was the same as that just described, and its coefficients
are reproduced in the table for comparison with those of the acoustic plasters.
The plaster 4A was made at the Building Research Station to a specification
provided by the Sabine Laboratories in America. The constituents are granulated
blast furnace slag and magnesium oxychloride cement. The slag and cement are
beaten up with water to a frothy consistency, the included bubbles giving the material
its porous and sound-absorbent character. The specification to which this par-
ticular specimen of plaster was made has since been considerably modified and
improved by the inventor.
The plasters 5 A , 6A and 7 A are modified forms of the Sabine plaster, and are
the subjects of a patent held by the Building Research Board. The constituents are
magnesium chloride, granulated slag, magnesium oxide, lime, powdered aluminium
and glue. These constituents are beaten up t o a froth, the formation of which is
assisted by the evolution of gaseous bubbles by the reaction between the lime and
the powdered aluminium.
The samples tested show the increase in absorption-coefficient caused by
increasing the proportion of slag.
(c) Effect of Varying the Thickness of Acoustic Plasters.-Three specimens of
the acoustic plaster 6A were provided b y the Building Research Board. These
specimens were identical as regards composition, but had different thicknesses-
namely, 1, 13 and 2 inches. The results of tests at 512 vibrations per second were
interesting as showing that thickness has a marked influence on the absorption-
coefficient. The figures obtained were as follows :-
Thickness of specimen. Absorption-coefficient.
Inches.

1 0.28

I
14
2
! 0.59
0.67

It is clear, therefore, that in making comparative tests it is of the utmost

importance to see that the thicknesses of the specimens and their backings are
precisely alike.
(a)Effect of Distempeying an Acoustic Plaster.-It is clear that, since acoustic
plasters owe their sound-absorbent properties t o their porous character, any treat-
ment with paint, varnish or distemper which tends to close up the pores will reduce
the amount of sound absorbed. To verify this conclusion, specimens which had
been treated with distemper were provided by the Building Research Board. Tests
were made a t two frequencies (n=256 and %=512), with the results shown below.

14A
15A
Normal
Specimen,

... ...
One coat of distemper
...
... I 0.16
Coefficient of absorption.

n=P56

0*06*
I %=512

0.29
0.13
I-
16A Twocoats of distemper ... 0*05* 0.11

* These figures are t o be regarded as upper limits t o the coefficients, since, on account of
the small amount of sound absorbed, the measurements became very difficult to take.
VOL. 39 X
28 4 Dr. E . T. Paris on
'
(e) Absorption of Sozllzd by Hairfelt.-The amount of sound absorbed by various
thicknesses of hairfelt has been the subject of measurement by Sabine and others,
and observations made on two samples of this material are included here. It is
obvious, however, that a material such as this is liable t o vary considerably in
structure and composition, and that no very marked agreement between the
observations made by different workers is to be expected.
The felts employed in the tests were cut into circular pieces 12 in. in diameter.
A mounting was used similar t o that employed with the acoustic plasters, and a
preliminary test was made with the mounting empty to ensure that there was no
leak of sound. One disc of hairfelt was then fitted in to the mounting, and thn
absorption-coefficient measured. The measurement was then repeated with two,
three or four 1ayers.of felt.
The two samples of hairfelt, which will be referred to as "A " and " B," differed
somewhat in texture and thickness, though both were nominally &in. thick. The
following particulars were noted.
Sample A.-Soft, light-coloured felt. Average thickness 0.46 in. (mean of six
measurements). Weight per square foot, 84 oz.
Sample B.-Brown felt, rather coarser texture, and harder than A. Average
thickness 0.41 in. (mean of six measurements). Weight per square foot, 72 oz.
The~sampleA was tested at 512 vibrations per second, and sample B at 380
and 512 vibrations per second. The results of the tests are given in the following
table :-
Hairfelt.

Number
of
layers.
-
approximate
thickness.
I
Sample A.

ofCoefficient
absorption
(W=512).
1 Approximate
thickness.
1 -Sample B.

Coefficient of absorption.

w=380. j
-
n=512. I

1
Ins.
0.5 0.21
0.52
1 Ins.
0.4
0.8
1 0.12
0.28
C.17
C.38

1 3
0.9
1.4 0-69 I/ 1.2 0.41 0.60
1.8 0.69 11 - - -
It may be worth noting that the figures obtained for the felt A are very similar
to those found for a +in. felt by Sabine* by the reverberation method. This is shown
in Fig. 9, in which the coefficients for A are plotted against number of layers
employed, and Sabine's results are plotted for comparison. One point of difference
may be noted. In the case of the felt A, no appreciable increase in absorption was
recorded when four layers were substituted for three layers. The curve plotted from
Sabine's results, however, shows that he found a definite increase in absorption.
( j ) Cotton Waste.-The most absorbent surface which has so far been tested
consisted of a thickness of 4in. of loosely packed cotton-waste, held in place by
wires stretched across it. This material has been used by Mallett and Dutton to
line a sound-chamber in which acoustical experiments were being performed, and
they claimed that it was "sufficiently sound-absorbent to prevent any trouble
* "Collected Papers," p. 213, Fig. 4.
0
structed which would be sufficiently
SAMPLC ‘A‘
sensitive for employment up to 1,000
- ___--- . vibrations per second, and yet have
dimensions small enough compared
05
with the wavelength for the effect
of external reflexion, or scattering,
t o be ignored. Above this frequency,
n
however, the difficulty of making a
286 Dr. E. T . Paris 0%

Since that time the work remained practically in abeyance (owing to the demands
of other experiments) until the end of 1926, when the above account was written
up, and a few additional observations made. At the time of writing I had not
seen a Paper by E. A. Eckhardt and V. L. Chrisler, which appeared in April, 1926
(“The Transmission and Absorption of Sound by Some Building Materials,” Scien-
tific Papers of the Bureau of Standards, No. 526). In this Paper are given the results
of measurements of sound-absorption made with a stationary-wave apparatus
resembling that employed by Hawley Taylor (Zoc. cit. sztpr.). The source of sound
was a loud-speaker, and the sound-measuring instrument a telephone, to which
the sound was led by a long, narrow exploring tube. The relative amplitudes of the
sounds observed were found by comparing the E.M.F.s generated in the telephone
by Karcher’s method (Sci. Papers of the Bureau of Standards, No. 473, 1923). An
advantage of this form of sound-receiver appears to be that it can be used with
very high frequencies, Eckhardt and Chrisler quoting absorption-coefficients up to
2,890 vibrations per second, and, in one case, up to 3,210 vibrations per second.
The only materials for which they give absorption-coefficients which can be com-
pared with those obtained with the apparatus described above are “Akoustolith”
(i.e., the acoustic tile 2A) and 1in. of hairfelt. For Akoustolith they gave a=0*301
at 569 vibrations per second, which lies between the values 0.26 at 512 vibrations per
second, and 0-31 at 650 vibrations per second, given in 5 7 (b). It is, however, rather
greater than would have been expected, and this may be accounted for by the fact
that the specimen appears not to have been backed by a reflecting substance. For
hairfelt (1in. backed by a reflector) they give 0.33 a t 297 vibrations per second.
This again is rather higher than would be expected, but a comparison is not of much
value in the case of this substance.
There is one respect in which the procedure adopted by Eckhardt and Chrisler
in making observations differed notably from that described above. A different
experimental pipe was used for every frequency employed in the tests, the length of
the pipes being such that there was resonance to sound of the particular frequency
with which the observations were made. This procedure appears to me to be quite
unnecessary and to arise from an incomplete understanding of the theory of the
method. The reason given by Eckhardt and Chrisler is, in effect, the following :
They say that sound which enters from the source, after passing down the pipe and
being reflected from the specimen, returns t o the source end, where it may be again
reflected with some loss of energy, and that this process may be repeated many
times. This,they appear t o believe, will vitiate the measurements except in two
special cases : (1) When the pipe is in resonance with the sound, and (2) when there
is practically no reflexion at the input end. The second case being difficult to
realise experimentally, they adopted the first alternative, and constructed pipes
to resound to the sound at every frequency at which measurements were made.
It should be obvious, however, that when a steady state is reached, the direct and
reflected waves coming from the input end will all be reflected by the specimen in the
same way and that the net effect is the same as that of a single train of incident
and a single train of reflected waves. The form of the potential within the pipe is
thus determined solely by the reflecting properties of the specimen under test. The
am9litudes of the pressures and motions, however, are dependent on the length of
the pipe and the amount of energy lost at reflexion at the input end. A formal
demonstration is not difficult.
 The Measurement of Sownd-Absor$tion. 28.7
Eckhardt and Chrisler also deduce form* for calculating the coefficient
of absorption when there is loss of energy from the sound during its propagation in
the pipe itself. The necessity for taking this loss of energy into consideration
appears to arise from the fact that somewhat narrow pipes were used. No dimen-
sions whatever are given in their Paper, but they state (p. 58) : I n order to get a
tube t o resonate satisfactorily its diameter must be relatively small.” In the
apparatus described above, no appreciable absorption was observed in the pipe
itself (see, for example, the observations recorded in the table in 4 7).
APPENDIX I.
THEDETERMINATION
OF THE DAMPING
FACTOR MICROPHONE.
OF A HOT-WIRE
The stationary-wave apparatus provides an easy qeans for determining approxi-
mately the damping factor of a hot-wire microphone. The procedure is as follows :-
The microphone is placed inside the sound-chamber with the source working
steadily a t the frequency at which it is desired to measure the damping factor. A
series of observations is then made of the ohmic changes in the grid corresponding
to different settings of the sliding part of the container. Let Q1,Qz,etc., be the
internal volumes of the container corresponding to the ohmic changes pl, p,, etc.-
the volume being determined in the usual way by weighing the container empty and
full of water. Also let Qm denote the volume for which p is a maximum, say, pm.
The microphone is next placed in a stationary wave in the experimental pipe, and
the positions found a t which the ohmic changes are pl, p2 . . .‘ pm, etc. If these
positions are distant yl, yz . . . ym, etc., from the nearest loop, then the amplitudes
of the oscillatory air-currents in the neck of the microphone producing the ohmic
changes pl, p,, . . . pm, etc., are proportional in sin kyl, sin ky, . . . sin kym,etc.
From the theory of the Helmholtz resonator we know that the amplitude of the
air-current in the neck, corresponding to a primary potential (po, is
C
40=(n2+*h2)& I yo 1 ’
where c is the conductance of the neck, h is the damping factor, and

no being the resonance frequency of the resonator and n the frequency of the sound.
Hence, if qa and q b are the amplitudes corresponding to two different values of n, we
have
qa2-sin sky nb2+4hz
.- -a-
q b 2 Sin 2kyb-na2+4h2

so that

If we pUtAa=O (the case for exact tuning) then pa= pm, and ya=ym, and we have

wheren and y correspond to any other tuning of the microphone, the upper or lower
sign being closer according as is positive or negative.
 288 Dr. E. T. Paris on
In the expression f o r n it is to be. remembered that n is constant, and equal to
'a -
--v/c/Qm, by Rayleighs formula, a being the velooity of sound in air ; while no is
27L
varied according to the setting of the sliding part of the container, and is equal to

Thus
2n

It is obvious that when Q is near t o Qm,the errors in findinga may be large.

The following figures were obtained by this method with the 512-frequency
microphone used for the absorption-tests.
Volume of container kY 4A I A
C.C. Degrees. sec.-l. scc.-l,
11.39 6.45 337.6 79
11.28 10.5 244.5 86
11.575 12.1 2C4.5 73
11.93 15.05 155 83
12.325 20.15 104 84
12.60 27.7 69
12.96 324 22 5
13.16 33.3 ( s k y m ) 0
13.25 32.1 13
13.565
13.86
27.3
20.7 1 49.5
84.5
14.49
15.33
13.95
1C.5 1 188
246 84

When is small, the results are somewhat erratic, as would be expected.

If these values of h for which *a
is less than 100 be excluded, the mean of the remain-
ing seven determinations is 81 sec.-1.

APPENDIX 11.
THEREACTION
OF ATUNEDMICROPHONEON A STATIONARY SOUND-WAVE IN A
PIPE.
The theory of the reaction of the tuned hot-wire microphone on the stationary
sound-waves within the experimental pipe can be dealt with as a special case of the
theory of the "Boys " type of double resonator. This theory* shows that if qoeLkat
represents the oscillatory air-current (c.c. per sec.) in the neck of a Helmholtz
resonator of the type used in hot-wire microphones, placed at a distance L within a
stopped pipe of length L,due to a primary potential 2Felkata t the mouth of the pipe,
then

* Phil. Mag., Vol. 11.. pp. 751-769 (1926).

 The Measurement of Sownd-Absorption. 289
where c=the conductance of the orifice of the resonator ;
h=the damping-factor of the resonator ;
=2nlzo(lzo/n-n/lzO);
n,=the resonance-frequency of the resonator ;
n=the frequency of the primary potential ;
p=the (complex) coefficient of reflexion of sound-waves, coming from
within the pipe, at the open end ;
k=2n/2=2nn/a ;
a=the velocity of sound in air.
The factor ,LL occurring in (i) represents the "reaction " of the Helmholtz
resonator on the potential of the stationary waves within the pipe. For the theory
of the Helmholtz resonator shows that the value of qo corresponding to a primary
potential (poelkat is

so that (i) shows that the effective potential in the pipe is

It can be shown that, in the absence of the resonator, the potential at thesame
place in the pipe would be

t
-p e2(8 pI'- - e k L
-tkL cos k(Z -L). . . . , . . . , . (iv)
Hence, comparing (iv) with (iii), we see that the deviation of ,U from unity represents
the reaction of the resonator on the potential within the pipe.
The value of ,Uis-
ac
cos k(Z -L) .eLHz-L)
'+a(Zh -/A) . . . . . . (v)
ac
1- cos k(Z-L) .e-rKz-L)
a(2k -LA)
where Q is the cross-sectional area of the pipe.
In applying these expressions to the case of the tuned microphone in the experi-
mental pipe, we note first of all that since the Helmholtz resonator which forms the
tuned part of the hot-wire microphone is in tune with the sound we have n=O.
Also, it is more convenient to measure distances from the stopped end of the pipe,
and we therefore put I -L= -x, x being the distance from the stopped end. When
these alterations have been made, the expression for qo and ,U become

ac
l+5-cos
l + E c o s kx .e-rkz
Y= ac
. . . . . . . . . (vii)
1 -- cos kx . elb
2ha -,.-
290 Dr. E . T . Paris 09%

The case of no reaction is arrived at by making c,+O, in which case p + l , and

Now, (vi) can be written

Hence, comparing (viii) with (ix), we see that, to estimate the effects of reaction
on the amplitude of the air current in the node of the resonator, we must compare
the modulus of

with cos kx.

It has been assumed in the above calculation that the primary potential
2 F is constant and independent of the position of the resonator within the
pipe. I n the case of the apparatus as arranged for absorption measurements,
however, this is not strictly true, unless special precautions are taken. The
source of sound (loud-speaker) being close to the open end of the pipe is easily
influenced by changes in the disposition of the apparatus within the pipe. Thus, if
the source of electrical supply to the loud-speaker remained untouched, it was found,
by observation of the resistance of the control grid in the throat of the loud-speaker,
that the strength of the source of sound underwent changes as the microphone was
moved along the axis of the pipe. These changes were quite small (of the order of
0.01 ohm), and could easily be compensated for by adjusting the electrical supply
so that the resistance of the control grid was always the same when a reading was
being taken.
For a numerical calculation of the effects of reaction in the case of the apparatus
employed for absorption measurements at 512 vibrations per second, we have the
following values for the various quantities in the expressions given above :-
c=0.114 cm.
h=81 sec.-l.
a=33,760 cm./sec.
a=)n x (30.5)2cm.2.
L=226 cm.
K=Snlf. =2n x 412 133,760 cm.-l.
y is first calculated by means of (v), for kx=O', 15", 30', etc., up t o 90'.
I
kx P
-
00 1,0724
15' 1*0625+1XO.00052
30' 1*0500+~ X0.000723
45" 1-0331+~XO*OOO549
60O 1.0 163+ I X 0*000232
75" 1-0045+6 X 0.000036
90 1-0000
 The Measurement of Sound-Absorption. 291
If the calculation is continued between kx=90° and kx=180" by steps of 15",
it is found that the numerical values of the real and imaginary parts of ,U are the same
for 105", 120°,etc., as for 75O, 60°,etc., but the sign of the imaginary part is reversed.
Thus, when kx=105O, p=1.0045 - 1 ~0.000036.The result of this is that the effects
are not quite symmetrical about a loop position (kx=90"), but the difference is very
small, as will be seen from the figures given below.
The next step is to calculate +(e-Lkz+peLkz)and its modulus, use being made of

'
the value of p given in the above table.

kx I +(e-rk2: +perks). I Modulus.

1.0362
0.9961+tl X 0.0083 1 1.0362
0.9961
0.8874+, XO.0128 0.8875
0.7176+~XO.0119 0.7187
0.5040+' XO.0071 0.5040
0.2594+~XO.0021 0.2594
0 I 0
.___ --

It will be seen that the effect of the imaginary part is quite negligible. If the
calculation is continued between kx=90" and kx=180°, the sign of the re+ part
becomes reversed.
The remainder of the calculation of reaction effects consists {see (x)) in evaluating
Be-1kL -elkL
Pe-ckh -pe~J:L* , . . . . . . . . (4
,or, (@-1) COS k L -t(P+1) sin k L
(@-,U) cos k L - t ( p + p ) sin k L ' ' ' (XW

In order t o do this it is necessary to decide on a numerical value for @-that is

for the coefficient of reflexion of sound-waves from the open end of the pipe next
t o the source.
I n the case of a pipe the diameter of which is small compared with the wave-
length of the sound the coefficient of reflexion from a flanged open end is*

where c,, (the " conductance " of the open end) is approximately equal to 3.8 x t h e
Tadius of the pipe.
The diameter of the pipe, however, in the present experiments was of the order
of half a wavelength (actually 0.46 of a wavelength at 512 vibrations per second),
and the above expression for can no longer be regarded as strictly applicable.
* cf. Phil Mag., Vol. 11, p. 756 (1926).
292 Dr. E. T . Paris on
I n the absence, however, of a more satisfactory theoretical expression for B it was
assumed that (xiii) remained of the same form for wider pipes and that the effect
of the increased wideness could be allowed for merely by choosing a different value
for c,,.
I n order to find a n appropriate value for co the following procedure was resorted
to. The proportion of sound which escaped from the open end of the experimental
pipe (away from the source) was determined experimentally by the method used for
measuring absorption. The results of the observation showed that the amplitude of
1
the wave reflected from the open end was about - of that of the incident wave, that
5
1
is, that the modulus of j3 was about - It was then found by trial that the value of
5'

__ .___- -~ _- .___

I
Modulus of (sii). I
kx
c,=3R. cO=5R. c, =IR.
--
0" 0.952 0.947 0.944
15' 0.958 0.954 0.951
30 0.966 0.963 0.961
45a 0.978 0.915 0.974
60' 0.989 0.988 0.987
75O
goo
0.997
1.000
0.997
1.000 :- 0.996
1.000
 The Measurement of Sound-Absorption. 293
case when c,=5R ; but it can easily be seen, by reference to the preceding table,
what the effect would be of making c,, equal to 3R or 7R.
kx I C(W I I coskx 1
- . -
00 C.9809 1~0000
15' 0.9501 0.9659
500 0-8544 0.8663
45O 0.7006 0.7071
80" 0.4977 0.5000
75O 0.2585 0.2588
90 0~0000 o*oooo
105' 0.2585 0.2588
1200 . 0.4976 0.5000
135' 0.7004 0.7071
150' 0*853,9 0.8660
165" 0.9495 0-9659
1800 0.9809 1~0000

It will be seen that the greatest deviation of I C(kx)I from 1 cos kx I is at the
nodes (kx=O" and l8Oo),where it amounts to just under 2 per cent. When kx=45'
it is less than 1 per cent., and falls to zero when kx=O". The influence of this devia-
tion on the measurement of absorption-coefficientsis quite unimportant, and fpr all
practical purposes it is eliminated by the procedure described in 5 6.
I n calculating IC(kx) I, four figures have been retained in order to show the
slight asymmetry which exists about the loop position (kx=90"), due to the reversal
of the sign of the imaginary part of ,U which occurs when kx passes through the
value 90'.
APPENDIX III.
REACTION OF A TUNED MICROPHONEON A P R O G R E S S I V E WAVE IN A PIPE.
Let the microphone lie in the plane x=O, and let the potential of the incident
wave be rp=AeLkcat-x). The presence of the microphone will give use t o a reflected
wave of small amplitude ~ ' = A ' e ~ ~ ( ' "while
l + ~ ) the
, amplitude of the wave trans-
mitted across the plane x=O will be less than I A I. Let the transmitted wave be
y =B&at -W.
On the assumption that the disturbance caused by the microphone causes a
departure of the waves from their plane form only for a very short distance on either
side of the plane x=O, the conditions to be satisfied by Q , 'p' and y when x=O are
Q+ 'p'= y . . . . . . . . . . . (a)

9
at
is the current, directed inwards, through the orifice of the microphone. For exact
tuning, (c) becomes
dq-2
dt -2h
dY
'di
. . . . . . . . . . . (4
 294 Dr. E . T , Paris on

Hence, substituting for y, cp' and the values given above, and eliminating d9
-
dt'
we find
A+A'=B . . . . . . . . . . . . . (4

Thus 2A =(2+$:)B,

and B=A/(~+;;). . . . . . . . . k)
Therefore the effect of the reaction of the microphone is to reduce the original poten-
tial A to an effective potential A / ( l + Z f
ac 4ha
). From the figures given in Appendix 11
we find that 1+-=1.016.
4hcr

APPENDIX IV.
ERROR DUETO INACCURATE SETTING OF THE SLIDER.
As stated in 8 7, the position of the microphone in the experimental pipe could
be read on the slider to the nearest half-millimetre. There is, therefore, a possible
error in 2y of h 0 . 0 5 cm., and in y of j 0 . 0 2 5 cm. To find the effect of this error
on the coefficient of absorption we differentiate the expression (5.3) of § 5 and obtain

'a-' ka2cos ky .cot2ky,

'Y 4
and hence, for an increment Sy=0.025 cm., approximately,
1
6a=- ka2 COS ky .cot2 ky ~0.025.
4
Some values of 6a for various values of a are shown in the following table. It
is assumed that the frequency of the sound is 512 vibrations per second.

6%
kY K
(6y=0.025cm.)

1
10 0.070 0.379 0.0095
20 0.130 0,330 0.0083

;:
80
0.244
0.343
0.429
0.287
0.251
0.219
0.0072
0,0063
0.0055
100 0.504 0.191 0.0048
-
Thus, when a=0.25, the possible error in a due t o an error -&0.025 cm. in y is &0-007.
The error is, of course, reduced by taking the mean of a number of determinations.
The arrangements for setting and reading the slider in these experiments could
readily be improved and more accurate values of y , and hence of a, could be obtained
than have hitherto been attempted.
 The Measwement of Sound-Absorption. 295
It was ascertained, for example, that when y=0-75 cm. a perceptible change in
resistance could be produced by moving the microphone a distance of only 0.01 cm.,
a movement which could not be read on scale on the slider.
Errors in the readings arise principally from : (1) The difficulty of setting the
position of the microphone by operating the slider by hand ; (2) the unevenness
of the bottom of the experimental pipe, which tended to tilt the microphone slightly
in some positions, and thus gave rise to false scale-readings o n the slider ; (3) slight
lateral movements of the slider.
The table given above indicates that the errors in a due to inaccuracies in
measuring y may be considerable when a is 0.1 or under. On the other hand, when
a is above 0.5 small errors in y are not of much importance.
DISCUSSION.
Dr. A. H. DAVIESinquired what type of loud-speaker had been used ? The hot-wire microphone
had the disadvantage that it could not be used for the higher speech frequencies, and at the
National Physical Laboratory it had been found convenient t o use as detector a magnetic
telephone with electric circuits whose resonance could be varied at will.
Mr. R. S.MAXWELLsaid that Sabine had found that the coefficient of absorption varies with
the intensity of the sound employed, and had taken as standard a n intensity one million times
as great as that for the threshold of audibility. He had an impression that Sabine’s results for
felt were taken as 256 cycles, whereas those of the author and others were taken a t 512. If so,
the quantities plotted in Fig. 9 were not comparable, since some materials have a resonance of
their own and the absorption consequently varies with frequency.
Dr. E. G. RICHARDSON urged the value of non-resonant measuring apparatus. A simple
hot-wire would avoid resonance and also enable the size.of the pipe t o be reduced. He had used.
the stationary-wave method with the hot-wire anemometer for measuring absorption coefficients
(Proc. Roy. Soc., A, 112, 522, 1926).
Major W. S.TUCKER said that the size of the apparatus was determined by the size of t h e
specimens t o be tested. He would like t o join in congratulating the author : he had been
impressed by the speed with which a variety of specimens could be tested in succession.
Prof, E. N. DA C. ANDRADE : An interesting method bf measuring sound intensities is used.
by Messrs. Siemens Rr Halske, of Berlin. When sound issues from a small opening there is a
partial separation of the two opposite phases of the vibration, so that at certain places there
is on the whole a pressure, a t others a suction effect. Just in front of the opening there is formed.
a narrow jet of air, whose energy depends on the energy of the sound vibration, and this jet is
powerful enough to blow aside a light suspended system (of no particular sensitiveness) so as
to produce deflections giving a quantitative measure of the sound intensity. The jet may also
be used in conjunction with a bolometer or hot-wire microphone.
Mr. L. E. C. HUGHES said that, when working with Dr. Mallett, he had measured the
attenuation of sound along the axis of a loud-speaker in a region screened with cotton waste,
assuming an inverse square law for the geometrical attenuation. An energy loss of 12 per cent.
on reflection had been found a t a frequency of 500 cycles, the loss increasing with frequency.
Dr. A. B.WOODasked whether the specimen discs vibrate like a diaphragm’ Such behaviour
would have a considerable effect on absorption: the point could be tested by varying the
diameter as well as the thickness of the discs.
The AUTHOR said that a loud-speaker of the electromagnetic type was used in the experi-
ments. In reply to Mr. Maxwell, the intensity of the sound could only be varied over a Limited.
range, and no variation of the coefficient of absorption had been detected within this range. The
quantities plotted in Fig. 9 were taken from Sabine’s results for felt a t 512 vibrations per second..
With regard t o the point raised by Dr. Wood, care had been taken throughout the experiments
t o test the specimen-holders before the materials were inserted. No appreciable loss of energy
could be observed with the holders alone, and it was inferred that any vibration of the discs as
diaphragms was quite unimportant. A device of t h e type mentioned by Prof. Andrade was used
some years ago by Mr. E. S. Player, of the Air Defence Experimental Establishment, for exploring
the resonance characteristics of pipe resonators. It consisted of a fine hole in the wall of the?
resonator, the issuing jet impinging on a hot-wire microphone grid.