Recent Advances in Applied Acoustics*

F. Trendelenburg, Berlin

V. Hearing and Speech

In reviewing works concerned with elucidating the properties of hearing, first place must be given to Treger’s work80. He succeeded, by precise physical methods, in investigating the properties of the tympanic membrane in a living subject. To study the properties of the tympanic membrane, use was made of the reverse action of its vibrations on the sound process produced by a sound emitter in a tube inserted into the auditory canal. The properties of the tympanic membrane influenced the sound process in the tube in the same way as the properties of a terminating resistance in an electrical circuit influence the process in the circuit itself. We have already indicated above that the differential equation of the process in an acoustic conductor is analogous to the “telegraph equation.” The solution of this differential equation shows that the acoustic properties of the tympanic membrane can be determined by measuring the amplitude and phase of the pressure at certain points of the conductor.

Let us denote by \(P_{\max}\) the pressure at the measured point, chosen at a distance \(\frac{\lambda}{2}\) from the sound emitter, and by \(P_{\min}\) the pressure at the same place, but with the length of the tube (between the point at which the measurements are made and the terminating resistance) shortened by \(\frac{\lambda}{4}\); the ratio \(\frac{P_{\max}}{P_{\min}}\) depends, according to theory, on the magnitude of the mechanical

* Continued; see Uspekhi fizicheskikh nauk, XI, issue 4, 650; XII, issue 1.

resistance \((R)\), characterizing the tympanic membrane as a sound receiver (mechanical resistance is understood to mean the ratio of the pressure on the receiver to its velocity).

The determination of the amplitude and phase \(P_{\max}\) and \(P_{\min}\) was carried out with the aid of Tischer’s compensating microphone, which has already been discussed above[^81]. It should also be noted that the relation between \(R\) and

\[ \frac{P_{\max}}{P_{\min}} \]

gives the conductance characteristic of the acoustic conductor; the conductance characteristic in the acoustic case is determined in the same way as in the electrical case—from measurements under no-load conditions. In no-load investigations the tube is not introduced into the auditory canal, but is closed tightly with a metal plug, so that complete reflection occurs at the end.

Fig. 35. Apparent resistance of the tympanic membrane as a function of frequency (after Treger).

Fig. 36. Phase of the resistance of the tympanic membrane as a function of frequency.

Measurements with pure tones in the range from 200 to 3000 hertz showed that the tympanic membrane acts as a pressure receiver: the mechanical resistance has an elastic character. The absolute values of the resistance (in CGS units) as a function of frequency are shown in Fig. 35. Fig. 36 gives the phase of the resistance of the tympanic membrane. In the range between 700 and 800 hertz the resistance of the tympanic membrane reaches its minimum; its absolute value here is almost equal to the acoustic

to the resistance of the air, namely—it amounts to about 40 CGS units. Thus an interesting fact is clarified: in the region of maximum sensitivity the human ear works almost ideally[^28].

Measurements made on various persons showed that, at frequencies below 500 hertz, the resistance of the tympanic membrane is sufficiently similar for everyone (in Fig. 37a the resistance of the tympanic membrane of seven subjects is represented vectorially, at a frequency of 500 hertz), but in higher regions individual differences become noticeable (Fig. 37b, 1000 hertz).

Fig. 37. Resistance of the tympanic membrane of various subjects: a—at 500 hertz and b—at 1000 hertz (vector representation).

In similar works one can also find indications concerning the directional sensitivity of hearing when listening with one ear[^33]. At frequencies above 300 hertz the screening action of the head already becomes noticeable. At frequencies above 2000 hertz perception becomes sharply directional. Fig. 38 shows how, in this case, an optimum occurs for a direction perpendicular to the auricle.

Work was also carried out on investigating the properties of the inner ear. The basis of this question is that its complete solution by purely physical methods is impossible; in the boundary region between physics and physiology it is in general difficult to find an unambiguous solution, as in many problems of pure physics. Investigations by biologists, carried out on the living organism, allow only insignificant variations, whereas physicists, working with inanimate matter, can choose the experimental conditions more or less freely.

The most important question is how the ear can distinguish tones by their pitch. The motion

the tympanic membrane (about whose behavior under the action of sound incident upon it we can judge with sufficient accuracy from the studies of Tröger just mentioned) is transmitted to a series of auditory ossicles. The last of the auditory ossicles⁸⁴ is the stirrup (see the schematic Fig. 39). The motion of the stirrup is transmitted to the thin membrane of the oval window, situated at the beginning of the cochlea; the cochlea is filled with lymphatic fluid, in which, in this way, pressure oscillations arise corresponding to the incident sound.

Fig. 38. Diagram of the directional sensitivity of the ear (after Tröger).

The cochlea is divided into two parts by a partition consisting of a bony and a membranous part. The membranous part of the partition is a tightly stretched fibrous membrane, the basilar membrane, which has—as may quite well be supposed—a decisive significance for the ability of the ear to distinguish tones according to their pitch. The lower part of the cochlea is also filled with lymph and ends in the round window, closed by a membrane. Pressure oscillations in the upper part of the cochlea produce forced oscillations of the basilar membrane, which are transmitted to the lower part of the cochlea, where, with the aid of the round window, they can be compared with the external pressure.

Fig. 39. \(G\)—auditory canal; \(K\)—auditory ossicles; \(T\)—tympanic membrane; \(OF\)—oval window; \(S\)—cochlea with the basilar membrane \(B\) (schematic).

The endings of the auditory nerve are distributed over the basilar membrane; by means of these endings the excitations arising during the oscillation of the basilar membrane can be transmitted to the acoustic center of the brain.

The principal problem is to clarify the question of what the forced vibrations of the basilar membrane are; in particular, whether it is possible that the basilar membrane is a sound analyzer, and that, for a tone of a definite pitch, only a definite part of the membrane vibrates. The basilar membrane has an elongated form; its length is approximately 35 mm, the width of the membrane at the oval window reaches only 0.04 mm, and at the apex of the cochlea its width is about 0.5 mm. The membrane has—as has already been mentioned—a fibrous structure, and its transverse tension is considerably greater than its longitudinal tension.

Helmholtz^85 was the first to investigate theoretically the vibrations of a membrane of this type. In his calculations he assumed that the acting force was distributed uniformly over the entire membrane. Later it was established that the membrane has the form of an isosceles triangle and that the longitudinal tension of the membrane is vanishingly small in comparison with the transverse tension. The calculations show that, with such a structure of the membrane, only a small part of the fibers will vibrate; this is analogous to what we have in the case of a series of strings over which some tone is sung: the group of strings whose natural period of vibration is close to the period of the given tone begins to vibrate, while the amplitude of the remaining strings, whose period lies far from resonance, remains small.

Furthermore, calculation shows that the base of the triangle (the region of the membrane near the apex of the cochlea) responds to low tones; for high tones the resonance region lies at the beginning of the basilar membrane—near the oval window. These data of Helmholtz’s theory are entirely correct; they agree well with the results of experiments performed on animals. In these experiments the basilar membrane was damaged by excessively strong acoustic or mechanical stimuli; for low tones, lesions were observed at the apex of the cochlea, and for high tones at the oval window.^86 But against Helmholtz’s theory there is a whole series of objections.^87 One of these objections is connected with the question of how the ear senses low tones

even when they last only an extremely short interval of time (the smallest, about two vibrations). With so brief an action, the system cannot be excited sufficiently for an exact sensation of pitch to arise. Another objection is also very substantial, namely: it is extremely difficult to imagine physically that the basilar membrane, being very light, can have natural frequencies as low as the frequencies of tones at the lower limit of audibility; and then how can the basilar membrane analyze the entire range of audible frequencies, spanning about ten octaves. Lux, Roaf and Fletcher, Wegel and Lane, and others ^89^ tried to eliminate these contradictions by regarding the basilar membrane as a complex system, the mass and elasticity of which should be considered as consisting of the mass and elasticity of the lymphatic fluid and of a flexible partition; this theory better explains the anatomical relations, but does not fully remove the difficulties mentioned above ^89^.

Fig. 40. “Sound pattern” on a membrane model (after Ewald).

Ewald ^90^ attempted to verify Helmholtz’s theory on a model. For this purpose he excited vibrations of a thin rubber membrane and found, in complete contradiction to Helmholtz’s views, several zones of maximum excitation; he found that standing waves are formed on the membrane, and that at a low frequency of the exciting vibration the nodes of the standing waves lie far apart, while at higher frequencies they come closer together (Fig. 40). Hence, in contrast to the Helmholtz “single-place theory” of forced vibrations of the basilar membrane, one may speak of Ewald’s “many-place theory” ^91^, since according to Ewald each tone corresponds to a definite “sound pattern” (Schallbild) covering the entire basilar membrane.

Ewald’s theory, if considered from the point of view

the results of observations of ear injuries can hardly be regarded as valid. Injuries to the membrane, caused acoustically by means of a tone of a definite pitch, are in fact located at one definite point and are not distributed over the entire surface of the membrane. The causes of the incorrectness of the results obtained by Évald apparently consist in the fact that the model of the membrane, in its tension and damping, did not fully correspond to reality.

Fig. 41. Formation of vortices at a model of the membrane (schematic, after Békésy).

Békésy carried out interesting investigations to clarify this question ⁹². He observed the vibrations of a membrane placed in a viscous liquid. In doing so he assumed that a membrane stretched in a liquid with sufficiently great internal friction cannot have its own vibrations. Békésy found no definite nodes of vibration for the model of a membrane immersed in a glycerin solution. By stroboscopic investigations it was discovered that, from the excited regions, groups of waves run along the membrane, their amplitude gradually decreasing. Further, he established that in the liquid, near the membrane itself, vortices are formed (Figs. 41 and 42); at low frequencies the vortices lie far from the window, at high frequencies—near it. Under very strong excitations it was possible to obtain injuries at definite places of the membrane. Thus it proved possible to produce on this model injuries analogous to the injuries that had occurred in experiments with animals. Fig. 43 shows injuries of the membrane caused by a vortex of this kind. Békésy’s investigations ⁹⁴ are based on the “single-point theory,” although the mechanism of action here is entirely different from what Helmholtz assumed. The advantage of this theory in comparison with Helmholtz’s theory is that here the difficulties of explaining the oscillatory process consisting of a single vibration are less pronounced.

Not long before this, theoretical questions concerning the basilar membrane were again taken up. Koch[^94] considered the forced oscillations of a model analogous to the basilar membrane, but under somewhat different assumptions than those laid by Helmholtz at the foundation of his theory.

Fig. 42. Formation of vortices in a membrane model (after Bekesy).

Fig. 43. Destruction of a membrane model by vortex formations (after Bekesy).

Those assumptions were that the membrane is rectangular, that the longitudinal tension is small in comparison with the transverse, but not vanishingly small, as Helmholtz had believed. It was further assumed that the range of audible frequencies lies below the lowest natural frequencies of the membrane, which is quite probable in view of the smallness of the latter. This assumption is directly opposed to Helmholtz’s theory, which supposed that the membrane analyzes sound by means of its natural frequencies. Calculations carried out under these assumptions show that standing waves arise both in the transverse and

Fig. 44. Standing waves on a membrane with internal damping (according to calculations by Koch and Guildsmeister).

and in the longitudinal direction; but since, by assumption, the membrane is strongly stretched in the transverse direction, waves propagating in the transverse direction may be neglected. The longitudinal waves can be decomposed into two systems, each of which begins at the narrow edge of the membrane and ends on the other side. Fig. 44A depicts the system of standing waves arising as a result of the superposition of both groups of waves; the greatest amplitudes lie at the edges.

It is quite obvious that this system of waves is extremely similar to the “sound pattern” obtained by Ewald. Thus Koch’s calculations apparently agree with Ewald’s “multipoint theory.” But it should be noted that further calculations, carried out by Koch himself and by Guildemeister ^95, led to a solution of the original equation indicating the validity of the “single-point theory.” Such a solution is possible under the following assumptions.

1. One end of the membrane is free; then the waves coming from this edge disappear.

2. The periodic force acts especially strongly on the narrow edge, and not uniformly over the whole membrane.

3. The damping increases in the direction from one narrow part to the other. If these three conditions are fulfilled, then in practice only a single maximum is formed on the basilar membrane (Fig. 44B), situated the farther from the oval window the lower the exciting tone; this latter point agrees well with the investigations of hearing defects that have already been mentioned more than once.

From the standpoint of the purity of sound transmission, the question of how distortions affect the subjective perception of sound is especially interesting. Earlier, work dealt chiefly with investigations of the influence of the frequency characteristics of instruments on the purity of a tone; now the effects of nonlinear distortions on sound perception are also being investigated. Nonlinear distortions are characterized by “combination tones”—components not contained in the original sound. Suppose that two tones, with

frequencies \(\omega_1\) and \(\omega_2\); if the system is nonlinear, then, in addition to these fundamental tones, combination tones are also transmitted, whose frequencies are determined by the following law:

\[ \omega_n=m\omega_1\pm n\omega_2 \qquad m,n=1,\,2,\,3\ldots \]

Usually the first difference tone \(\omega_1-\omega_2\) and the sum tone \(\omega_1+\omega_2\) are especially strongly noticeable.

If a simple sinusoidal tone \(p=p_1\sin\omega t\) falls upon the system, then, together with the original tone, higher harmonics also arise, with frequencies \(2\omega,\,3\omega\ldots\), etc., formed according to the law mentioned above. At Klopmüller’s proposal,⁹⁷ the so-called clirrfactor is adopted as a measure of nonlinear distortion:

\[ k=\sqrt{\frac{P_2^2+P_3^2+\ldots}{P_1^2}}, \]

where \(P_1\) is the amplitude of the fundamental vibration—the only one acting on the system—and \(P_2, P_3\ldots\), etc., are the amplitudes of the overtones arising as a consequence of distortion.

Fig. 45. Dependence of limiting distortion on frequency (after Yanovsky).

Yanovsky⁹⁸ investigated the audibility of nonlinear distortions. The distortions were produced in a vacuum-tube circuit (the operating point on the curvilinear part of the characteristic). From the course of the operating characteristic it was possible to calculate the clirrfactor. The dependence of the audibility threshold of distortions on the frequency of the tone acting upon the distorting system (“threshold distortion”) is shown for various loudnesses in Fig. 45. So long as the pitch of the fundamental tone lies below 600 hertz, the first overtone (the only one noticeable for a small clirrfactor), owing to the greater sensitivity of both the ear and the telephone to high frequencies, is heard better than the fundamental tone; correspondingly, distortions at low frequencies are more noticeable, even at

in a small clipping factor. If the fundamental tone lies in those frequency regions to which the ear is more sensitive, then the overtone caused by nonlinear distortion is heard less well, and therefore the threshold of audibility of the distortions rises.

When two tones act simultaneously, the audibility of nonlinear distortions may be different; it depends both on the relative position of the two tones and on their position in the audible region. The following phenomena may occur here.

1. Change of timbre. One or several new tones arise which, taken separately, are not audible, but which cause a change in timbre (threshold value of the clipping factor 1–6%).

2. A separate tone becomes audible. The newly arising tone is heard in the distorted sound (threshold value from 0.3 to 1.4%).

3. The tone becomes hoarse. The new tones caused by distortion form beats with the original tone or with one another, as a result of which the distorted tone seems rough (threshold value from 1.4 to 2%).

Practical measurements of limiting distortion are especially important in the transmission of music. The following table gives the results of investigations carried out on five persons.

Among works on the study of the voice, it is necessary to mention investigations of the mechanism of the voice itself, as well as investigations of the sound field of the human voice (as, for example, the directivity of the voice).

In the main, the correct theory of the production of vowels in singing was given by Helmholtz. The stream of air goes from the lungs through the glottis into the larynx; the excitation of the vocal cords is in this case analogous to the excitation of a flute. The mass and elasticity of the vocal cords determine the pitch of the fundamental tone. The sound is rich in overtones, some of which are amplified by the resonance of the oral cavity. These amplified overtones form the formants. Proof of the correctness of Helmholtz’s views lies in the strict periodicity of the oscillatory process for vowels sung at a definite pitch.

The mechanism of phonation is represented schematically in Fig. 45. When the pressure in the respiratory tract increases, the vocal cords move apart; the increase in the cross-section of the glottis causes a decrease in the pressure difference between the respiratory tract and the larynx, as a result of which the vocal cords come together again, the pressure difference again increases, and so on.

Fig. 46. Self-excited generator as an equivalent of the human voice.

The mechanism by which sound arises is analogous to the mechanism of self-excitation of a tube generator. In a tube generator a direct current (from the anode battery), as a result of self-excitation of the oscillatory system, is transformed into an alternating current.

The production of vowels can be examined visually by means of an electrical equivalent. In the right-hand part of Fig. 46 a circuit is given for a self-excited tube generator operating into a circuit connected with an antenna. Circuit I is the electrical equivalent of the system of vocal cords; the anode battery corresponds to the pressure reservoir—the lungs. The arrow → denotes feedback (in the acoustic case, the action of the oscillations of the vocal cords on the air stream). It must be emphasized that the connection must not be under-

be regarded as simple. In the acoustic case there is a combined coupling (coupling through friction and mass in the glottis, elastic coupling in the vocal cords), in contrast to the electrically most frequent case—purely inductive coupling. The process in circuit $I$ affects the process in circuit $II$ (in the acoustic case, the resonating cavity of the mouth, taken for simplicity as a simple system[^98])—and the antenna (the mouth opening). Here, too, the coupling must be regarded as combined.

The theory of the production of vowels in the general form briefly set out above has not yet been developed. On the electrical equivalent only the oscillations of the vocal cords have been analyzed—the work of Wegel[^99]. In this work the conditions of sounding were also determined: the conditions of damped sounding and the conditions of the stationary oscillatory state of the vocal cords. It is extremely important to note the fact that a change in the amplitude of the oscillations of the vocal cords causes a certain change in frequency; the term of the differential equation that determines the friction depends strongly both on the acoustic self-induction and capacitance and on the magnitude and shape of the glottal opening[^100].

The dependence of the just-mentioned quantity and of the resistance of the (air) jet on amplitude accounts—something that has not hitherto been pointed out—for the fact that the sound of the larynx is rich in overtones; here we have an analogy with a vacuum-tube generator with strong one-sided excitation. A large number of overtones in the laryngeal sound is the basic condition of Helmholtz’s theory—as was indicated above. In this respect the processes in acoustic systems are analogous to processes in electrical ones; in the latter, when the rectilinear part of the characteristic is exceeded, higher harmonics arise, which may serve as an indication that the final stationary state has been reached[^101].

The questions of the directional action of the voice are treated in Trendelenburg’s works[^102]. The directionality of the human voice is determined, on the one hand, by the shielding action of the head, and, on the other hand, by the horn-like action of the mouth and in some cases, for individual sou—

as, for example, hissing ones with a planar distribution of sound. To embrace theoretically the influence of these various features of the organs of speech is extremely difficult. Experimental investigations were carried out as follows.

The subject was placed on a rotating platform on which a microphone was fixed, recording the amplitude of the pressure on the axis and in the middle of the mouth during motion. In the room there was also another microphone, receiving the sound radiated “to the side” as the platform rotated. The microphone currents were recorded oscillographically. Comparison of the oscillograms

Fig. 47. Diagram of investigations of the directivity of the human voice.

Fig. 48a. Directed action of the voice for \(U\).

gave a diagram of directed action (Fig. 48 a, b, c). The directivity of the action is small for \(U\) (frequency region around 200 hertz), somewhat greater for \(I\) (3000 hertz) and especially *

markedly noticeable for \(S\) (formants in the region of 5000–6000 hertz). From Fig. 48c it is evident that for \(S\) the amplitude of the pressure behind the head amounts to only 10% of the pressure along the normal to the middle of the mouth. Thus only about 1% of the sound energy goes backward. Further, it is evident from the diagram that a more or less uniform region extends to 45% from the middle normal. If the receiver of sound is outside this angle, then the intelligibility of speech decreases, and the high formants of hissing sounds disappear especially.

\(I.\ 3300\ (220)\) hertz

Fig. 48b. Directional action of the voice for \(I\).

\(S.\ 5\text{–}6000\) hertz

Fig. 48c. Directional action of the voice for \(S\).

Lüder \(^{104}\) investigated, by the method of octave analysis, the distribution of the mean and maximum values of speech sounds over the regions of the various octaves of the acoustic spectrum. Some results of his investigations are presented in Figs. 49–50. Of interest is the fact that low frequencies in speech are contained only to an insignificant extent and are especially strongly expressed in the region of 800 hertz. The most essential component of German speech is the formant \(A\). Observations

other researchers, in particular the studies of Stumpf, Wagner, Crandall, Trendelenburg, and others, confirm these results. In the highest octave (from 6400 to 12,000 hertz) lie, for the most part, the components \(S\), \(Z\), \(f\), and \(ch\).

Analogous studies were carried out by Sivian[^105]. Interesting results were obtained. With an increase in voice intensity, the principal mass of the various components shifts toward high frequencies. This is of special importance for unnaturally loud artificial transmissions; in such transmissions the timbre of the sound becomes dull, which does not occur in the natural voice in view of the above-mentioned objective changes.

Fig. 49a. Distribution of the sounds of the human voice according to Luder (male voices of four persons with different pitch of voice). Spectrum of maximum values.

Fig. 49b. Distribution of the sounds of the human voice according to Luder (male voices of four persons with different pitch of voice). Spectrum of mean values.

VI. ARCHITECTURAL ACOUSTICS AND ROOM ACOUSTICS

I. Reflection

Studies of the reflection of sound waves from the walls of rooms were formerly carried out, for the most part, on models

reduced scale or graphically on the plan of the building. The development of objective methods of recording sound by means of a microphone and oscillograph made it possible to carry out similar investigations by other methods. Schindelen[^107] investigated various methods of such measurements; in some cases the reflected sound of a revolver shot was recorded oscillographically (testing with a short impulse), in others—the reflection of a periodically repeated short tone produced by a loudspeaker (testing with a tone). Scharstein investigated by these methods the vestibule of the Higher Technical School in Munich—a room approximately 5.90 m high and with a radius of curvature of the ceiling equal to 10.72 m. In Fig. 51 the path of the rays in this

Fig. 50a. Distribution of the sounds of the human voice according to Loder (three female voices of different pitch). Spectrum of maximum values.

Fig. 50b. Distribution of the sounds of the human voice according to Loder (three female voices of different pitch). Spectrum of mean values.

room, obtained by construction, is given. In Fig. 52 the corresponding oscillogram is given. In this case the sound rays, after the first reflection, are distributed over the whole com-

nate, then are reflected from the floor and, having been reflected once more, converge on the ceiling at a point symmetrical with respect to the initial point. In the same way they return back to the initial point. The total path of the rays is about 46 m. Fig. 52 shows the echo very clearly (marked in the figure by point 4); point 8 marks the echo occurring after traversing this path twice, i.e. after covering a distance of about 92 m; point 2 corresponds to the echo caused by side rays.

Fig. 51. Section of the hall of the Munich Higher Technical School.

Investigations carried out in the large physics auditorium of the Higher Technical School in Munich showed—

Fig. 52. Echo phenomena (after Sharschtein).

that, when a short impulse is excited, “reflection tones” (Reflexiontöne) arise: the sound reflected from the first row of benches \(d\) and the following reflections from other

Fig. 53. Section of the new physics auditorium of the Munich Higher Technical School.

rows were recorded by a microphone located at point \(a\) (Fig. 53). Calculating the pitch from the distance between the benches, we obtain 212 hertz, which agrees well with the oscillo-

by a graphic recording (Fig. 54, points 1, 2, 3, 4). The investigations carried out by Sharstein and Schindelin[^109] in the assembly hall of the University of Freiburg, which is distinguished by especially poor acoustics, are extremely illustrative. These investigations give an example of what distortions can arise in a hall whose surfaces possess foci.

Fig. 54. Formation of tones reflected from a row of auditorium benches (after Sharstein).

Fig. 55. Longitudinal section of the assembly hall of the University of Freiburg.

In Fig. 55 a longitudinal section of the assembly hall is given, in Fig. 56 the plan of its base, and in Fig. 57 an oblique section having the form of an ellipse. The speaker’s desk is placed approximately at the focus of the ellipse; sound rays issuing from the desk in the plane of the section gather at the other focus and, after double reflection, gather again at the original point. The oscillograms of Fig. 58 clearly show how the echo returns to the speaker’s desk with an intensity only very slightly weakened in comparison with the initial one.

Fig. 56. Plan of the base of the assembly hall of the University of Freiburg in Breslau.

Fig. 57. Oblique section of the hall.

In the cases considered above, reflections from the walls of a room are quite noticeable to subjective observation. The echo returned to the original point after a comparatively long interval of time and with great strength; the time interval was so large that the echo arrived after a new syllable had already begun, as a result of which individual syllables were mixed together in the most inconceivable manner \(^{110}\).

The works considered above dealt exclusively with questions of the distorting action of reflection, but there exist a number of works concerned with the use of reflection to improve the acoustics of buildings. Noyens and Philippi \(^{111}\) calculated a sound reflector reflecting the sound rays issuing from a point source in such a way that they were radiated cylindrically symmetrically. The reflector was a fourth-order surface with the equation

Fig. 58. Course of the sound process in an assembly hall (after Sharstein and Shpindelin).

\[ \sqrt{x^{2}+y^{2}+z^{2}}=\sqrt{(x+a)^{2}+y^{2}}-b. \]

The sound excitation lies predominantly in the surface passing through the sound source and perpendicular to the axis of the paraboloid. If the various listeners are in one plane, then this change in the sound field is extremely advantageous \(^{112}\).

The construction of such a sound reflector was described—

by Fokker. Fig. 59 shows the finished model of such a reflector. Practical investigations with reflectors were carried out in the cathedral of St. Bavo in Haarlem; in these experiments good intelligibility of speech was obtained at considerable distances from the reflector[^114].

Fig. 59. Sound reflector (after Fokker).

In Fig. 60 the zones of sufficient speech intelligibility with sound reflectors are indicated (by dotted lines), the reflectors being installed: one at the entrance to the choir, the other on the church pulpit, located in the fourth bay (in the figure, point \(P\)). From Fig. 60 it is clear how advantageous it is, from the point of view of improving speech intelligibility, to use a radiating system that would concentrate the sound in the plane in which the listeners are located. In this way it is possible to improve speech intelligibility in halls considerably; but it must be noted that, in order to obtain directivity, whether by means of reflectors

Fig. 60. Speech intelligibility in the cathedral of St. Bavo in Haarlem (the dotted lines indicate at what distance speech is still intelligible).

or by means of loudspeakers of corresponding design, it is necessary to use surfaces whose dimensions are large in relation to

by comparison with the wavelength of sound. Such structures in halls are very conspicuous and therefore are often rejected for purely architectural reasons. Thus architectural and acoustical requirements here remain in a still unresolved contradiction.

2. Absorption.

Extremely important methods for determining sound-absorption coefficients were given by Sabine: the sound-absorption coefficient of various materials is established from the change in reverberation in a room whose absorption is known. This procedure is especially advantageous in that the value of the coefficient is determined under conditions that occur in ordinary practice. In these measurements the sound waves travel, as is the case in almost all practical situations, in various directions determined statistically. Since sound absorption, generally speaking, depends on the angle of incidence of the sound wave on the surface (the angle is taken with respect to the normal to the surface), this method has a fundamental advantage over other methods in which the sound waves propagate in a definite direction.

Recently, with the development of oscillographic methods for measuring reverberation, especially with the introduction of instruments that automatically measure reverberation, Sabine’s method has come into wider use. In Germany this method has been adopted chiefly in the “echo chamber” of the Hertz Institute^[117]; in America this method is used by the Bureau of Standards^[118].

Before discussing the results obtained in various works, it is necessary to make several further remarks concerning the fundamental dependence between sound absorption, reverberation time, and room volume; the relation among these quantities is of great importance both for determining the desired value of the sound-absorption coefficient and for the problem of reverberation.

If the conditions of complete randomness of the motion of sound in a room are fulfilled, i.e., an identical distribution of sound in all directions and the absence of concentrations

of sound caused by floors, concave surfaces, or anything analogous to this, then, according to Sabine, the following relations are valid:

\[ E=E_0 e^{-2\delta t}=E_0 e^{-\frac{ca_mF}{4v}t}, \]

\[ E_0=\frac{4L}{ca_mF},\quad \delta=\frac{ca_mF}{8V},\quad T=0.163\,\frac{V}{a_mF}, \]

where \(L\) is the power of the sound source, \(E\) is the instantaneous energy density, \(E_0\) is the energy in the stationary state, \(a_m\) is the mean absorption coefficient, \(F\) is the area of the surface bounding the room, \(V\) is the volume of the room, \(T\) is the reverberation time, and \(c\) is the speed of sound.

K. Schuster and E. Waetzmann gave an exact calculation of reverberation in a room of a definite shape; the calculation of the reverberation time was formulated as a boundary-value problem for the propagation of a sound wave. For a cubic space the damping was obtained as

\[ \delta_{\text{cubic}}=\frac{ca'_mF}{4\sqrt{3V}}, \]

for a cylindrical space whose diameter is equal to its height,

\[ \delta_{\text{cylind.}}=\frac{ca'_mF}{6\sqrt{2V}}, \]

and for a sphere

\[ \delta_{\text{spher.}}=\frac{ca'_mF}{12V}, \]

from which it is clear that the value given above,

\[ \delta=\frac{ca'_mF}{8V}, \]

lies between the values for the cube and the cylinder.

In analogous works, questions of the optimal reverberation time were discussed. S. Lifshitz arrived at an empirical formula for the optimal reverberation time, assuming that in a room of any dimensions \(T_0\lg E_0\) is constant (\(T\) denotes the time of actual reverberation, i.e. the time between switching off the excitation and the decrease-

…of the energy density to the value of the audibility threshold); in this connection it turned out that the time of optimum reverberation increases with the volume of the room. Schuster and Wetzmann[^120] critically discussed this question, and it proved that one cannot assume a sound power independent of the size of the room, since in practice a greater sound power is used in large rooms. Assuming that \(L\) increases with \(\sqrt[3]{V}\), it turns out that, contrary to Lifshitz’s conditions[^121], the optimum reverberation time increases only very slightly with the volume.

Fig. 61. Dependence of the optimum reverberation time on the volume of the room, under different assumptions (after Schuster and Wetzmann).

In what follows we shall discuss the question of whether, according to Lifshitz’s assumption, \(T_0 \lg E_0\) is indeed constant, or whether \(E_0\) varies as \(\sqrt[3]{V}\) while \(T_0\) is constant; in the latter case the optimum reverberation depends only very little on the volume. Finally, it will be clarified how the relations are formed in the presence of an interfering background, which reduces the effective reverberation. In the case where, owing to the presence of background noise, the final level corresponds to an intensity 10 times greater than in a quiet room, the optimum reverberation again increases somewhat with increasing room volume. In Fig. 61 the values of optimum reverberation are presented according to Lifshitz’s assumption (curve 1) and according to three other assumptions given below (curves 2–4). In general, it is extremely difficult to judge the correctness of one or another representation, and this is especially difficult to do for a concept so dependent on subjective factors as reverberation.

Meyer and Just recorded, with the aid of a microphone, an amplifier, and a rapidly settling torsion galvanometer, the process of reverberation produced by a loudspeaker

Fig. 62. Reverberation curves of a sound mixture (according to Meyer).

Fig. 63. Measurements of sound absorption by various people (according to Meyer).

in a room with a volume of \(120\ \text{m}^3\). To avoid interference, the loudspeaker was excited by howling tones. Plotting the obtained curves on a grid where pressures are laid off on a logarithmic scale, we obtain—

as is required by the theory of reverberation—straight lines; Fig. 62 gives “reverberation straight lines” for beating tones in various frequency regions. The reverberation straight lines intersect the parallel to the abscissa axis, constructed for a pressure equal to \(10^{-3}\) of the initial pressure, at points corresponding to Sabine’s “reverberation time”; thus the reverberation time can be determined graphically from the drawing. It is seen from the figure that in the present case the absorption of sound has a maximum at 4800 hertz and a minimum at 150 hertz.

Fig. 63 gives the reverberation straight lines of two different rooms before the entrance of three subjects and after their entrance; curve \(a\) is the reverberation straight line of room \(I\) before the subject enters it, curve \(b\) is the same, but after the subject enters; curves \(c\) and \(d\) are respectively the same for room \(II\). Between the reverberation time \(T_1\) before the introduction of a sound-absorbing substance, the volume of the room \(V\), the sound absorption \(a\), the reverberation time after the introduction of the sound-absorbing substance \(T_2\), and the resulting sound absorption \(a_1\), the following relations exist:

\[ aT_1 = k - V, \]

\[ (a + a_1)T_2 = kV, \]

whence, calculating the absorption produced by the subjects in the above-mentioned experiments, one obtains 0.70 for room \(I\) and 0.68 for \(II\)—two values in good agreement. It was further found that the absorption of sound by people depends strongly on the frequency; some values are given in the following table.

Sound absorption is strongly influenced by whether the sound-absorbing substance is located directly on the walls of the room or at some distance from them. In the first case the sound-absorbing substance is located at a velocity node, and the absorption is small; in the second it is shifted closer to a velocity antinode and the absorption increases. The results of such investigations, relating to porous substances, are presented in the following table.

The dependence of sound absorption on the angle of incidence of the sound wave (which has already been mentioned above) was investigated in the works of Rayleigh, Kreissler, and Schneider \(^{124}\). The results of the investigations are given in the following table.

Exhaustive investigations of the influence of sound-absorbing material in acoustic pipelines were carried out by Tischer \(^{125}\) with the aid of the compensation-microphone method, which has already been discussed above *.

* See Uspekhi fizicheskikh nauk, 11, p. 661, 1931.

Arranging sound-absorbing material in various ways along the tube, it turned out that the damping effect depends to a considerable degree on the place in which the material is located. The maximum effect occurs when the substance is situated at the velocity antinode of the standing wave. The position of the substance relative to the cross section does not play a substantial role. The damping effect increases strongly with frequency. The action of a substance placed across the axis of the tube is determined by two quantities, the values of which are given in the following table: the inertial resistance of the mass \(\omega M\) and the active resistance \(R\).

Thus it turns out that for fabrics the active resistance is approximately proportional to the weight per unit surface; the same is also true for the effective mass.

3. Sound Transmission

Davis and Littler \(^{126}\) carried out measurements of the sound transmission of various substances and building materials as a function of frequency. The sound beam from a loudspeaker was directed from one room (damped), through an opening, into another (also damped). In the latter there was a microphone, by which the intensity \(E_1\) was determined with the opening open and the intensity \(E_2\) when the opening was closed by a partition made of the substance under test. If

denote the ratio \(\dfrac{E_2}{E_1}\) as the sound transmission (Engl. transmission ratio), then the reciprocal value \(\dfrac{E_1}{E_2}\) will be the sound reduction factor (Engl. reduction factor). It is extremely important to note the fact that sound rays can also be transmitted through the transverse vibrations of the intermediate partition. The values of the sound reduction factor as a function of frequency are given in the following table (with the base of the logarithms equal to 10).

The table shows that sound insulation is the better, the greater the mass of the separating wall; further, it shows that double walls with an air space insulate sound especially well. At the Bureau of Standards, measurements of sound insulation were carried out by several other methods: for the measurement two highly reflecting rooms were taken, connected by an opening. In one of the rooms a certain definite energy density \(E_s\) was produced by a loudspeaker; in the other, the energy density \(E_h\) was observed by means of a microphone; by closing the opening with a partition made of the material under study,

rial, the energies became \(E'_s\) and \(E'_h\). The measure of the attenuation of sound is the ratio:

\[ \frac{E'_h}{E'_s}\cdot \frac{E_s}{E_h}. \]

The investigations of Kreissler and Snyder\({}^{127}\), carried out by this method, showed that the logarithmic coefficient of sound attenuation increases with the mass of the separating partition (provided that it is homogeneous). This was confirmed for every kind of partition, from thin cardboard to thick brick walls.

Meyer and Just\({}^{128}\) investigated sound transmission by one of the methods described above (see UFN, 1931, issue 4, p. 667). The sound was received by one microphone in the room where the sound was excited, and by another in the room into which the sound penetrated from the first room through the partition; the loudnesses given by the two microphones were equalized by subjective comparison with the aid of an attenuator. The magnitude of the active resistance served as the measure of the attenuation of sound.

Denoting the attenuation in nepers by \(b\), we have

\(e^{-b}=\dfrac{p_n}{p_s}\) (equal to the ratio of the mean pressure in the room where the sound is produced to the mean pressure in the room into which the sound penetrates through the partition).

The determination of the coefficient of sound attenuation is not unambiguous, as these investigations show. For \(b\) the following values are obtained.

Meyer\({}^{129}\) concluded from this that the ratio of the energy radiated by the partition under test into the room where the tests are carried out, to the total energy incident on the parti-

a partition, can be defined as sound permeability. The numerical values of sound insulation must give the natural logarithm of the reciprocal value of the sound permeability. In order to compare sound permeability determined in this way with sound permeability obtained by one of the above-mentioned methods, it is necessary to introduce a correction for the absorption of the room in which the investigations are carried out.

Let us also point out that the definition of sound permeability, according to the proposal of the AEF, has changed somewhat for the better; according to this proposal the loudness scale must be constructed using common logarithms. If \(E_1\) and \(E_2\) are the sound intensities measured by physical instruments, then the audible loudnesses will be \(10 \lg_{10} \dfrac{E_1}{E_2}\) phons\(^{130}\), or, referring to the pressure amplitudes \(P_1\) and \(P_2\), we have:

\[ 20 \lg \frac{P_1}{P_2} \]

“phons”*.

In order to express the values of the table on p. 284 in these units, they must be multiplied by 10; the values of the table on p. 285 must be multiplied by 8.67.

Let us touch briefly once more on the question of the sound permeability of light-reflecting screens used in sound motion-picture theaters, where the loudspeakers are placed behind the screen. When thin sound-permeable fabrics are used for screens, a considerable loss of brightness results. Hopkins\(^{131}\) found that screens covered with small perforations, whose total area does not exceed 5% of the screen surface, are quite suitable for sound cinema; in this case no significant losses of loudness are obtained, while the brightness changes very insignificantly.

* In American and in our literature the term “phon” corresponds to the logarithmic unit of loudness level, defined in the same way—the decibel. Translator’s note.

LITERATURE

1. J. Tröger, Phys. Z., 31, 26, 1930. See also West, “Measurements of the acoustic impedance of Humanrars,” Past. Off. Electr. Eng. J. 21, 293, 1929.

2. See Uspekhi Fizicheskikh Nauk, 11, 661, 1931.

3. The strong change in the sensitivity of the ear with frequency cannot be explained solely by the dependence of the resistance of the tympanic membrane on frequency; evidently, processes in the sound-conducting part of the ear play a role here, and perhaps also processes in the cochlea and in the nervous system.

4. On questions concerning the directional characteristics of hearing due to the screening action of the head, see F. A. Firestone and D. L. Rich, Phys. Rev. (2), 33, 634, 1929, and the measurements of B. Langenbeck in the article “Experimentelles und theoretisches zur Hörschwellenbestimmung,” Pflügers Archiv, 226, 11, 1930. It should be noted that, for the directionality of hearing, the fact that sound arriving at an oblique angle reaches the ears with some difference in time is of great importance. On this see Z. Hochfrequ. 28, 88, 1926, E. M. v. Hornbostel and M. Wertheimer. See also H. v. Békésy, Phys. Z., 31, 824, 57, 1930.

5. In this schematic review we shall not deal with questions of the transformation of pressure by the auditory ossicles.

6. H. v. Helmholtz, Die Lehre von den Tonempfindungen, 6th ed., p. 639, Braunschweig 1913.

7. See H. Held and F. Kleinknecht, Pflügers Archiv, 216, 1, 1927; detailed references to the literature are given there.

8. On these questions and further matters see Gildemeister, “Probleme und Ergebnisse der neueren Akustik,” Z. Hals-, Nasen- u. Ohrenheilkunde, 27, 299, 1930.

9. See E. Meyer, Handbuch d. Physik, by H. Geiger and K. Scheel, Vol. VIII, p. 527 ff., Berlin 1927.

10. Recently H. Fletcher calculated the vibrations of the basilar membrane in the lymphatic fluid, Journ. Acoust. Soc. of Amer., 18, 311, 1930.

11. J. R. Ewald, Pflügers Archiv, 76, 147, 1899; 93, 485, 1903; 131, 188, 1910.

12. See the cited work of M. Gildemeister (note 88), p. 319.

13. G. v. Bekesy, Phys. Z., 23, 793, 1928.

14. We shall briefly indicate the works of G. v. Békésy confirming his results and giving further explanations. In the cited work the vibrations of the basilar membrane were studied on an anatomical preparation (see the work cited above—note 93, p. 806 ff.). Studies were also carried out on the influence of fatigue of the ear and of a statically acting pressure on the tympanic membrane upon the ability of the ear to distinguish the loudness of sound (Phys. Z., 30, 115, 1929); there were works dealing with the theory of vibrations and with amplitude and frequency changes of tones.

Phys. Z., 30, 721, 1929). The results of the investigations confirm what was set out above.

1. H. Koch, Z. f. Sinnesphysiol., 59, 15, 1928. References to the theoretical work of H. Fletcher were given above (note 90).

2. See Gildemeister (note 87), p. 297.

3. Kv. Kupfmüller, Fachber. der 31. Jahresvers. V. D. E., S. 87, 1926.

4. W. Janovsky, E. N. T., 6, 421, 1929.

5. On the fact that the larynx and the oral cavity constitute coupled systems, see J. B. Grandall, Bell. Syst. Techn. Journ. 6, 100, 1927; also Z. Hochfr., 32, S. 204, 1928.

6. R. L. Wegel, Bell. Syst. Techn. Journ., 9, 207, 1930.

7. In singing any note by an artist, a change in intensity is compensated by a change in the tension of the vocal cords. In speech sounds this dependence conditions a peculiar process of equalizing the pitch between individual sounds.

8. This dependence is most noticeable in the lower register, where for a certain interval of time the vocal cords are closed; the oscillations in this case are especially rich in overtones. If the vocal cords part only for a short interval of time—small in comparison with the duration of the period—then during each period the oral cavity is excited by an impulse. Electrically, this process is conveniently reproduced by using, to obtain oscillations, instead of an electron tube, a tube with a glow discharge; on this see F. Trautwein, Elektrische Musik, Berlin 1930, Bd. 1, der Veröff. der Rundfunkversuchsstelle bei der Staatl. Akad. Hochschule f. Musik.

9. F. Trendelenburg, Z. f. techn. Phys., 10, 558, 1929.

10. See the calculations of G. W. Stewart, Phys. Rev., 93, 467, 1911.

11. H. Luedеr, Wiss. Veröff. a. d. Siemens-Konzern, Band. IX/2, 167, 1930; see also the references in U. F. H.

12. L. J. Sivian, Bell. Syst. Techn. Journ. 8, 646, 1929.

13. В. H. Baskhaus und F. Trendelenburg, Wiss. Veröff. aus d. Siemens-Konzern, Bd. IV/I, 205, 1925.

14. W. Schindelin, Ann. d. Phys. (5), 2, 129, 1929.

15. E. Scharstein, Ann. d. Phys. (5), 2, 163, 1929.

16. E. Scharstein u. W. Schindelin, Ann. d. Phys. (5), 2, 194, 1929.

17. Further examples of distortions arising as a consequence of the action of echo are given in the following works: E. Michel, Zentralbl. f. Bauverwaltg., 48, 486, 1928; F. R. Watson, The Architectural Forum, März 1929, S. 441; W. Linck, Ann. d. Phys. (5), 4, 1017, 1930; W. Kuntze, Ann. d. Phys. (5), 4, 1059, 1930.

18. M. Nuyens u. G. t. Phillippi, Physik 10, 18, 1930. On sound reflectors see also F. R. Watson (note 111). On acoustic-

...experiments in halls having a parabolic form, and where, owing to this, the action of reflection may prove favorable; see F. M. Osswald, Schweizer Bauhütte, 95, 3, 1930.

1. See, for example: E. Trendelenburg, E. T. Z., 48, 1685, 1927. The advantages of using directional action are set forth exhaustively in C. Zwikker, Der Ingenieur E. Elektrotechniek, 5, 39, 1929.

2. A. D. Fokker, Archives du Musée Teyler Haarlem, 7, 73, 1930.

3. A. D. Fokker und M. I. O. Strutt, Archives du Musée Teyles Haarlem, 7, 77, 1930.

4. In this case the pulpit was covered with a sound screen in order to reduce the undesirable radiation of sound into the upper part of the congregation.

5. W. C. Sabine, Collected Papers on Acoustics, p. 13, 1923. See also the new work by P. E. Sabine, Journ. Franklin Inst., 207, 341, 1929. Tables of absorption coefficients for such materials as acoustolith, celotex, sabinite, etc.

6. E. Meyer, Z. d. V. d. J., 74, 273, 1930.

7. See, for example: V. L. Chrissler und W. f. Snyder, Bureau of Stand. Journ. of Research, 5, 957, 1930; V. L. Chrisler, Jour. of acoust. Soc. Amer., 1, 418, 1930.

8. It is necessary to mention that a further condition for the validity of the present equation is that \(a_m \ll 1\). If this relation is not fulfilled, then \(a_m\) must be put equal to \(a_m=\ln(1-a_m)\). K. Schuster und E. Waetzmann\(^{120}\).

9. K. Schuster und E. Waetzmann, Ann. d. Phys. (5), 1, 671, 1929; K. Schuster, ibid., p. 696. See also the calculations of C. E. Eyring, Journ. Acoust. Soc. Amer., 1, 217, 1930. On questions of damping see the following works: M. J. O. Strutt, Mag. 8, 236, 1929, and the article by C. F. Eyring, Journ. of the Soc. of Motion Picture Eng., 15, 528, 1930.

10. C. Lifshits, Phys. Rev., 25, 391, 1925; 27, 618, 1926.

11. We shall also point to the work of W. A. Mac Nair, concerning questions of optimal reverberation (Bell. Syst. Techn. Journ., 9, 390, 1930). In this work the optimal reverberation time is considered under the assumption that the echo still perceived after the sound source is switched off is the same for all frequencies; then, in the calculation, it is obtained that for 80 hertz the reverberation is twice as great as for 1000 hertz. Further, we shall point to the interesting statistical investigations of Vern O. Knudsen (Journ. Acoust. Soc. Amer., 1, 56, 1929), concerning the relation between intelligibility of speech, reverberation time, and room volume.

G. v. Békésy points out, in a short work (Ann. d. Phys. (5), 8, 851, 1931), the importance of one phenomenon, known from psychology, for the evaluation of reverberation processes: processes can be regarded as single only during a definite, extremely short interval of time—the presence time (Presenszeit). The duration of this interval of time reaches approximately 0.8 sec. Substituting this value into the reverberation formula, as well as the value of the mean power of the sound radiat...

…produced by orchestral instruments and voices, it is possible to determine the value of the optimal reverberation as a function of the size of the room.

1. Meyer und P. Just, L. N. T. 5, 293, 1928.

2. P. R. Reyl, V. L. Chrisler und W. F. Snyder. Bur. of Stand. Journ. of Research, 4, 289, 1930. P. R. Heyl points out in one of the notes (Nature, 126, 350, 1930) that the results of the theoretical calculations of E. T. Paris [Proc. Roy. Soc. London (A) 115, 407, 1927] and J. Larmor [Proc. Cambr. Phys. Soc. (2), 27, 231, 1930] lead to opposite results. Larmor found that, at grazing incidence, the absorption is infinitely large; according to Paris it should be equal to zero. P. R. Heyl considered not entirely justified the assumption that, in absorption and reflection, only potential flow takes place; in the layer of air immediately adjacent to the absorbing and reflecting substance, turbulent flows may arise. E. T. Paris explains the indicated contradiction (Nature, 12, 880, 1930) by the fact that inadmissible simplifications were made in J. Larmor’s work.

3. H. Tischner, E.N.T., 7, 236, 1930. Measurements of sound absorption by observing acoustic processes (in tubes) were also carried out by A. H. Davis und E. J. Evans, Proc. Roy. Inst., 127, 89, 1930. See also the measurements by L. Casper und G. Sommer, to appear shortly in Wiss. Veröff. a. d. Siemens-Konzern.

4. A. H. Davis und T. I. Littler, Phil. Mag. (7), 7, 1050, 1929. Further see the work of P. E. Sabine, Journ. Acoust. Soc. Amer., 1, 203, 1930, and also the investigations of A. E. Knowler, Phil. Mag. 10, 342, 1930.

5. V. L. Chrisler und W. F. Snyder, Bur. of Stand. Journ. of Res. 2, 541, 1929.

6. E. Meyer und P. Just, die Schalltechnik, 2, 33, 1929, Heft 3; E. Meyer, Z. d. V. D. I, 74, 273, 1930.

7. E. Meyer, Die Schalltechnik, 3, 23, 1930, Heft 2. In a short paper (Berl. Ber. 1931, S. 166) E. Meyer gives further considerations on questions of sound insulation. In the course of these investigations the amplitudes of the forced vibrations of the separating partition were measured both directly and by the condenser-probe method (see note 44). Meyer’s investigations established that sound is transmitted by means of vibrations of the separating partition. The numerical values of the sound insulation and the logarithm of the weight of the wall are in a linear dependence. It is necessary to note that the principal work on questions of sound insulation, a dissertation, has appeared in a new edition: R. Berger (München 1911) Berlin, Verlag v. J. Birkenfeld.

8. Not to be confused with the previously used unit with base 10 or with the Barkhausen “phon” (with base 2).

9. H. F. Hopkins, ref. Kinotechnik, 12, 534, 1930.