ON THE TONE AND CURVES PF THE PHONOGRAPH. By JoHN G. M'KENDRICK, M.D., LL.D., F.R.S., Professor of Physiology in the University of Glasgow. (PLATES XI., XII., XIII.) 1. DURING the last twelve months my attention has been largely occupied with the phonograph. The instrument chiefly studied has been the machine used in this country,' which is so geared that the wax cylinder, 6j inches in circumference, makes two revolutions in one second, while the spiral grooves described on the cylinder are only ^ inch apart. A spiral line about 136 yards in length may be described on the cylinder, and the recording or reproducing point travels over this distance in about 6 minutes. The instrument is shown in the accompanying figure.2..~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~...... *!.,._..,,,,,,.. 2. By using conical resonators of considerable size, made of thin block tin, the tone of the phonograph can be much increased in volume, and the sounds become quite audible and agreeable 1 Supplied by the Edison-Bell Phonograph Corporation, who have kindly given me much assistance in this investigation. 2 As to the mechanism and uses of the phonograph, see X PacticZ (kid t the Use of t~e Phonograph. By-James L. Andem, Cincinnati, Ohio, 1892.
584 PROFESSOR JOHN G. M'KENDRICK. in a large room. The method of conveying the sounds directly to the ears by flexible tubes may be discarded when it is desirable to allow many persons to hear the same tones. The largest resonator I use is about 9 feet in length, with a diameter at the widest end of 3j feet.l The best results yet obtained have been with the use of tin resonators. A conical resonator about 3 feet in length and 6 inches in diameter at its wide end, made of thin aluminium, gives remarkably good tones, and encourages me to have another made of larger size.2 Wood, vulcanite, papier-mache, all have the effect of muffling the tones. By placing a Y-shaped tube over the disk of the phonograph and connecting each limb of the Y with a separate resonator, a wonderful augmentation of volume of tone may be obtained. Large resonators have also the effect of quenching many of the higher partials produced by the vibrations of the glass disk of the phonograph, and thus the quality of tone is softened and improved. This is well illustrated by placing the aluminiunm resonator in connection with the phonograph so that its wide end is opposite to the wide end of a large tin resonator. The tone then becomes louder and of better quality. 3. I have endeavoured to increase volume of tone by connecting with the phonograph (a) Mr Alfred Graham's transmitter along with (b) the loud speaking telephone also made by Mr Graham.3 By placing a large resonator in front of b, and by using a glass plate in the telephone (having the keeper of the magnet attached to the centre of the glass plate), the volume of tone is greater than that obtained by the large resonator alone, but the quality undoubtedly suffers. A certain hollowness of tone is communicated by the arrangement, and the telephone speaks better for certain tones than for others. I always place in circuit a dry cell battery of from 7 to 9 volts. A very fine effect is obtained by using the Y-shaped tube already referred to, connecting one limb with a large resonator and the- other with the telephonic arrangement. In this case the resonator gives 1 The use of resonators of considerable size has been practised by Mr Edison, and is especially common in America, but I am not aware that auyonc has used resonators of so great a size as above indicated. 2 The aluminium resonator was obtained for me by my friend Mr Paul Rotten. burg, merchant, Glasgow. 3 Supplied to me by Messrs Muirhead & Co., electrical engineers, Westminster.
THE TONE AND CURVES OF THE PHONOGRAPH. precision and definiteness to the sound, say, of a military band, while the telephonic arrangement increases the volume of tone. It is easy to fit up several telephonic arrangements, all linked together; and by placing these in different parts of the room, the effect becomes next to marvellous. 4. Apart from the pleasure of working with the instrument, one cannot help having a special interest in the question of the forms of the curves described on the wax cylinder. Many attempts have been made to obtain tracings of the vibrations of membranes and of glass or metallic disks. In 1856 LUon Scott invented the well-known phonautograph, which may be regarded as the precursor of the phonograph, and by which vibrations were recorded. Donders applied this instrument to the investigation of vowel sounds. Next came the logograph of W. H. Barlow, by which curves were obtained by the vibrations of a thin membrane of gold-beater's skin. About 1873 Koenig introduced his manometric flames, and flame-pictures of vibrations were thus obtained. In 1876 Stein succeeded in photographing the vibrations of strings.' All these instruments made it possible to record vibrations, but the sound could not be reproduced from the tracings thus obtained. This was accomplished by Edison, whose first phonograph was patented in January 1877. In 1878 Fleeming Jenkin and J. A. Ewing2 succeeded in obtaining tracings of the record of vowel sounds on the tin-foil phonograph, and the curves were submitted to harmonic analysis. Tracings were. taken, or at all events the marks on the tin-foil were examined, by Griitzner, Mayer, Graham-Bell, Preece, and Lahr.3 The imperfections of the tin-foil phonograph made progress impossible for a good many years (from 1877 till 1888), during which time, however, the ingenuity of Edison was engaged in working out the mechanical details of the wax-cylinder phonograph as we now have it, one of the most beautiful and (almost) perfect of instruments. The subject was then taken up by L. 1 Die Photographie der Tone. Von Dr Med. S. Th. Stein Poggendorif's Annalen., 1876, p. 142. 2 Tranm. Roy. Soc. Edin., "On the Harmonic Analysis of certain Vowe Sounds," vol xxviii. p. 745. The Telephone, The H&Mihone, and the Phonograph. By Count du Moncel, London, 1884. Also The Speakling Telephone and Talking Phonograph. By George B. Prescott, New York, 1878. VOL. XXIX. (N.S. VOL. IX.) 2 P 585
586 PROFESSOR JOHN G. M'KENDRICK. Hermann of Kbnigsberg, and he succeeded in obtaining photographs of the vibrations produced by the vowel sounds, a beam of light reflected from a small mirror attached to the vibrating disk of the phonograph being allowed to fall on a sensitive photograph film while the phonograph was slowly travelling.' The curves thus obtained were very beautiful, and present a striking resemblance to some of Koenig's flame-pictures. Finally, Boeke of Alkmaar,2 in a laborious microscopical research, measured the transverse diameters of the depressions on the wax cylinder at different depths, and from these measurements calculated the depths of the curves. He thus, in a manner, reconstructed the marks on a large scale. These are all the attempts to reproduce -the curves that have come under my notice. It has always been a matter of surprise to me that the question has not been further investigated by physicists. The recording and analysis of these curves are somewhat out of the beat of a physiologist, although their investigation is of much interest when we view it as bearing on the theory of vowel tones, the appreciation of quality of tone, and the mode of action of the membrana tympani and chain of ossicles. 5. I have endeavoured to study the marks on the wax cylinder in three different ways:-(a) taking a cast of the surface of the cylinder; (b) taking a microphotograph of a portion of the surface of the cylinder; and (c) recording the curves on a slowly moving surface, by a method to be afterwards described. (a) As regards the first method, taking casts, which was also attempted by Hermann, the results were not very satisfactory. The most efficient method followed by me was to paint on the cylinder, with a camel-hair brush, a layer of celloidin dissolved in ether. This soon hardened, and it could then be peeled off. The thin film thus obtained was then inverted on the stage of a microscope, and the marks were seen in relief. Microphotographs were also taken of the films. (See Plate XI. fig. 10.) This method had the disadvantage of flattening the elevations in the curves, but beautiful pictures of the depressions were obtained. 1 Ueber das Verhaluen dcer Vocale am neuen EdisoScheL Phonographen. Pifger's Archiv., vol. 47, 1890, p. 42; also PhonophotographisU Untermuchungen, ii. p. 44; also Phonophotographische Unuersuchungen, iii. p. 347. 2 MlikroskopischePkonogrammstudien. Pfluger's Archiv., vol. 50, 1891, p. 297.
THE TONE AND CURVES OF THE PHONOGRAPH. 587 (b) I took numerous photographs, with the aid of microscope and camera, of portions of surface of the cylinder on which were records made by many -instruments and by the voice. Specimens of such photographs are shown in Plate XI. Each figure represents -th of an inch on the surface of the wax cylinder, magnified about 14 diameters. The vertical grooves seen in each figure are -gith inch apart, and the length of each figure represents in time Path second, that is to say, when each tracing was recorded, the sapphire point of the marker connected with the vibrating plate of the phonograph travelled over the distance represented in magnified proportions in L1ath part of a second. (Fig. 8 is slightly shorter.) By counting the number of indentations or marks, which in a photograph have a curious appearance of being in relief, one can at once determine approximately the pitch of the tone, the vibrations of which made the impressions. Tracing 6 represents the picture of tones produced by the violin, and it will be seen that they vary much in character. Sometimes the marks are a little apart from each other, and at other times they blend into each other, the mark widening out as the sapphire point cut into the wax and then contracting as it receded. It is to be borne in mind that even when the glass disk of the phonograph is not vibrating, the sapphire-pointed marker connected with it is ploughing a groove in the cylinder; and when the glass disk vibrates, each vibration cuts deeper into the groove. The moniliform marking (seen also in 5, the tones of a tenor voice, and in 7, the tones of a flute) shows that the disk may not in some instances return to its position of rest for a short time. In other cases, the sapphire point cuts into the bottom of the groove as the glass disk is suddenly pressed in, and then the disk flies back suddenly to its position of rest. It is interesting to find that during many individual vibrations there may be a plus pressure on the outer surface of the glass disk, and that each individual vibration adds to and takes from this pressure. The first four examples (Plate XI., 1, 2, 3, and 4) are given to show the accuracy of the phonograph. I took four records from four of Koenig's tuning-forks, the vibration numbers of which were 64, 128, 256,512. When the record was taken the cylinder of the phonograph was making as nearly as possible two revolutions
588 PROFESSOR JOHN G. M'KENDRICK. per second, the rate at which all records should be taken. Photographs were then taken of portions of the cylinder representing Path of a second; these are shown in figs. 1, 2, 3, 4, in Plate XI. It will be noticed that in 1 (64) there is one lozenge-shaped mark, in 2 (128) there are two, in 3 (256) there are four, and in 4 (512) there are eight marks. Interesting, however, as this method of investigation is, it does not give the form of the curves represented by the bottoms of the depressions made by each vibration of the disk of the phonograph, and it was abandoned for the next method of mechanically tracing these curves. 6. A mechanical representation of the curves presents many difficulties. These were so far overcome by the device of Jenkin and Ewing with the tin-foil phonograph. The method followed by these observers, which was entirely mechanical, was to cause the disk of the phonograph to record its movements. As already mentioned, Hermann photographed the oscillations of a beam of light reflected from a small mirror connected with the disk of the phonograph, the whole apparatus moving slowly. My notion was to adapt a light lever to a marker connected with the phonograph itself, and so to arrange that it (the point of the marker) would travel over all the ups and downs of the phonographic curve on the wax cylinder at an extremely slow rate. The obvious objection to this method is that the inertia of the lever might cause extraneous vibrations, while at the same time the smaller marks on the wax cylinder might be missed, an objection, however, that may be removed by (a) reducing friction to a minimum, and (b) moving the phonograph cylinder so slowly as to make the movement almost invisible to the naked eye. After various attempts with simple appliances, the apparatus shown in Plate XII. was fitted up, and by means of it the taking of curves became comparatively easy. In fig. 1, Plate XII., a diagram of the arrangement is shown as simply as possible.' The apparatus was not specially made for the experiment, but consisted of fittings in my own laboratory. It might be simplified still further, and made even more convenient. The 1 For this drawing I am indebted to Mr Mackinnon, assistant to the Professor of Engineering.
THE TONE AND CURVES OF THE PHONOGRAPH. 589 motive power for driving the apparatus is a small water motor acting on a wooden wheel. From this wheel two trains of wheels or pulleys pass, the one set, namely, 2, 3, 4, and 5', being gradually geared down so as to drive the cylinder of the phonograph, 6,1 at an extremely slow rate, and the other set, namely, 2', 3', and 4', to drive at a slow rate the recording drum a. By this arrangement the rate of rotation of the cylinder of the phonograph is about once in five or six minutes, instead of being once in one-half second, the usual speed in recording. The drum, a, also moves very slowly, but a little faster (not twice as fast) than the cylinder of the phonograph, the object being to open out the curves somewhat in a linear direction. By attaching an electric arrangement to the axle of the mandril carrying the phonograph cylinder, 6, the time of each revolution of the phonograph cylinder was registered on the drum a. Thus the amount of surface on a, representing one-half second on cylinder 6, could always be measured. As a rule, it was found that 3 inches of the drum a represented one-tenth of a second. Occasionally it was one-seventh of a second, but it could be so timed as to represent one-tenth. In the next place, a light lever of hard wood, braced like the mast of a ship, was fixed firmly into a socket bored into the lead weight, i k, seen in fig. 2, Plate XII. This square leaden weight is hinged to the frame carrying the recording and reproducing part of the phonograph by the hinge, h, a slot is cut in the under surface of the lead weight, as seen in fig. 3, PI. XII, and the marker, a n (also m n in fig. 2), moves on an axle delicately pivoted to the sides of the slot. In fig. 2, at the end of wm, is seen the wire, w, passing up to the glass disk, d, of the phonograph. It will be seen that the point of m n touches the surface of the wax cylinder, o, o, o, o. The lead weight accurately follows the movements of m n, the marker, and the lever, s, records these movements, as seen in fig. 1, on the drum, a. The slowness of movement does away with any movement of s, except that which is communicated to it from the lead weight and from the marker, m n, and to prevent all extraneous vibra- 1 The cord passes over (b), the wheel seen on the left end of the phonograph spindle in the figure showing the instrument. The wheel will be recoguised by a tape band passing vertically over it.
590 PROFESSOR JOHN G. M'KENDRICK. tions, the phonograph was seated on a solid stone pillar erected for a galvanometer. The point of the lever s is the point of a very fine hard needle, and a was covered with very smooth paper, carefully smoked. As the marker of the phonograph is always travelling to the one side so as to describe a spiral groove, that is, towards the observer, when one looks at fig. 1, it is evident that the point of the lever x would soon leave the surface of a. To get rid of this difficulty, the drum a is mounted on a kind of movable table moving in slots, and controlled by the fine screw c (one thousand to the inch thread). And thus it was easy, by a turn of c, to keep the point of x in contact with the surface of a. Finally, the movable table rested on a plate of metal moving vertically in slots by turning the fine screw b, and thus, after a tracing had been taken once round a, it was easy to move a a little up or down without disturbing the lever x (fig. 1). Thus I had three motions of a, vertical, rotatory, and horizontal, and there were two movements of the cylinder of the phonograph b, rotatory and horizontal. Everything was as steady as it could be made, and the apparatus worked almost automatically, the point n (fig. 2) slowly crawling over the surface of the wax cylinder o, abbut 600 times slower than when the record was made on the wax cylinder. After tracings had been taken, the paper on a was varnished and it was cut into longitudinal strips, each three inches in length, and the strips were mounted between two ordinary microscope slides (English make) in the way that slides are prepared for the lantern. Such slips can then be examined by the microscope and the curves may be drawn by Abbe's camera lucida, or any portion of the tracing may be photographed by a convenient microphotographic apparatus. Thus the size of the curves may be much increased, and so made available for purposes of harmonic analysis. (See Plate XIII. and description of plate.) In the present paper, I desire only to describe the method and to show portions of a few curves (to be examined with the aid of the description of the Plate), leaving such a discussion of these as may be necessary to a communication to be made at a future time. Meanwhile, I would point out how similar these curves are to those obtained long ago with the phonauto-
THE TONE AND CURVES OF THE PHONOGRAPH. graph. It is interesting, also, to compare them with the curves shown by Professor Hermann in the papers already quoted. The curves obtained by Hermann's method are more pointed at the apex, closer together, and have generally a greater resemblance to Koenig's flame-pictures than those obtained by me.' I have endeavoured to state the time relations of the curves. By counting the number of waves in a given linear distance, an approximation is at once got to the pitch of the tone under consideration. It will also be noticed that the pitch, as one would expect, in a complicated succession of sounds, quickly alters, and it appears to me that the method now illustrated may enable us to answer the question as to the number of vibrations required to appreciate the pitch of any given tone. The varying forms of the curves is also striking, pendular in character for the piccolo, less so for the flute, more complicated in the bassoon, still more so in the cornet, and most of all in the record of a military band. The analysis of these curves will test the soundness of the theoretical views now held as to the kind of movements made by vibrating disks, such as the drum of the ear. The whole matter is thus of great interest both to physiologists and to physicists. Possibly, also, the subject may have a practical outcome.2 Could we only obtain accurately the curves of syllables, by putting these together in a conceivable machine, we might " set up" speech curves instead of type, and have a machine which, without having been spoken to, would appeal to the ears by sound and speech. 1 See Pftfger's Archiv., 1890. Tafel viii. p. 390. 2 Dr John Macintyre of Glasgow has put the phonograph to extensive use in his practice as a laryngologist, and he has shown that the instrument may be adapted to medical purposes. DESCRIPTION OF PLATES XI., XII., XIII. 591 PLATE XI. (In Figs. 1, 2, 3, 4, 5, 6, 7, and 9, the vertical length of the tracing represents 1/64 sec., and the cylinder is rotating as from the bottom to the top of each tracing.) Fig. 1. Record of tuning-fork, 64 vibs. per sec. (one long mark). Fig. 2. Record of tuning-fork, 128 vibs. per sec. (two marks). Fig. 3. Record of tuning-fork, 256 vibs. per sec. (four marks)
592 THE TONE AND CURVES OF THE PHONOGRAPH. Fig. 4. Record of tuning-fork, 512 vibs. per sec. (eight marks). Fig. 5. Powerful tenor voice, about 640 vibs. per sec. Fig. 6. Violin-Scotch reel "Tullochgorum," about 640 vibs. per sec. Fig. 7. Flute-portion of " II Trovatore," about 1000 vibs. per sec. near do5. Fig. 8. Vowel 0, sung by human voice. An imperfect photograph. Fig. 9. Full organ-taken 40 feet from organ-portion of Mendelsohn's "Wedding March." Remarkable for full, powerful chords. Piano record is similar. Fig. 10. Portion of celloidin cast of cylinder on which was recorded a march by a military band. Note the varying forms of the marks. PLATE XII. Fig. 1. Shows diagram of recording apparatus described in the text. Fig. 2. Shows on a larger scale than in Fig. 1 the recording arrangement. See text. Fig. 3. Under-surface of leaden weight to which recording lever was attached. PLATE XIII. Fig. 1. Facsimile of curves obtained from record of old English coach horn. Fig. 2. Facsimile of curves obtained from record of a military band. Note 1.-These curves (3 inches of the tracing representing onetenth sec.,-that is to say, the curves in 3 inches were recorded on the cylinder of the phonograph in one-tenth sec.) should be examined with a magnifying glass. In the two upper lines of 1, note the uniform character of the curves. In the third line the character of the curve alters. These are tracings of the marks produced by the long-drawn, clear, piercing tones of the horn. In 2 the upper and lower lines show great variety of curve form. The middle line shows no curves. Note 2.- Figs. 3 to 12 are from microphotographs of curves similar (though varying in amplitude) to those in 1 and 2, but magnified from 5 to 10 diameters. Only short portions of the curves are given, to illustrate the great variety in form. Fig. 3. Vibrations of large tuning-fork, 32 vibs. per sec. Fig. 4. Cornet. Fig..5. Bassoon (two curves). Fig. 6. Cornet. Fig. 7. Flute. Fig. 8. Old English coach horn (two curves). Portions of 1. Fig. 9. Military band. Three curves, varying much in shape. Fig. 10. Piccolo. Three curves. The upper is the best.
Joau.omAnxad.Phys.J y189s vou IXlTM, Y.S. ow.x/ Ị1 "I ei' V?. :i 4.I V 1. 2. 3. 5. 6. 7. 8 9. 10. TONE & CURVES OF PHONOGRAPH4 F Huth-LitY' Edie.
:: 0 IL 0z LI. 0 0x AL Ln Ul Nll
bivoarn. of'za'1d&p~s.jazz~y1&96. Ys Vol, 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. TONE & CURVES OF PHONOGRAPH. F Huth, LithT Edine