NRLF 
 
 B M E53 733 
 
REESE LIBRARY 
 
 OF THK 
 
 UNIVERSITY OF CALIFORNIA. 
 
 Accession No. 8 i/~ Ok. Clots No. 
 
' T ' T ^ ' T '"""""""" m \ ' T T T ^ T ^ 'T ' * ' T 'IT' 
 
 I** 
 
 I: 
 : 
 
 ; 
 
 THE 
 
 Sapillary 
 " ' electrometer 
 
 IN THEORY AND PRACTICE. 
 
 \] 
 
 part l. 
 
 BY 
 
 GEORGE J. BURCH, M.A., Oxon 
 
 REPRINTED FROM "THE ELECTRICIAN. 1 
 
 COPYRIGHT. 
 
 |t , 4 LONDON : 
 
 PRINTED BY GEORGE TUCKER, SALISBURY COURT, FLEET STREET, B.C. 
 
 
RESEARCHES MADE WITH THE APPARATUS 
 DESCRIBED IN THIS PAPER. 
 
 BURCH "On a Method of Determining the Value of Rapid Variations 
 of a Difference of Potential by means of the Capillary Electrometer," 
 Proc. Roy. Soc., Vol 48. 
 
 BURCH. "The Time Relations of the Capillary Electrometer," Phil. 
 Trans., Vol i83A, p. 81. 
 
 BURCH. "On the Calibration of the Capillary Electrometer," Proc. Roy. 
 Soc., Vol. 59, p. 18 
 
 BURCH AND VELEY. "The Variations of the E.M F. of certain Metals 
 in Nitric Acid," Phil. Trans., 1&2A p 319 
 
 SANDERSON AND BURCH. " On the Localization of the Effect of Injury 
 in Muscle," Proc. Physiol. Soc , June 24, 1893. 
 
 BURCH AND HILL. " On d'Arsonval's Physical Theory of the Negative 
 Variation," Journal of Physiology, Vol. XVI., p. 319. 
 
 GOTCH AND BURCH. " Electromotive Properties of Malapterurus Elec- 
 tricus," read before the Royal Society, May 7, 1896. Will appear 
 shortly. See Nature. May 28, 1896, p. 92. 
 
 J. BURDON SANDERSON. ' Electrical Response of Muscle," Journal oj 
 Physiology, Vol. XVIII., p. 117. 
 
PREFACE. 
 
 I HAVE often been asked to describe my method of 
 making a capillary electrometer, and my apparatus 
 for the measurement of rapid changes of E.M.F. 
 My aim in this series of articles, written in October, 
 1895, nas been to do so with sufficient fulness to 
 enable anyone possessed of average skill to construct 
 the instrument 
 
 I have laid most stress on the production and 
 interpretation of the photographic records as illustrating 
 the special characteristics of the capillary electrometer ; 
 but it is hardly necessary to point out that it may be 
 used by direct observation with the Wheatstone bridge, 
 or with the potentiometer, instead of a high-resistance 
 galvanometer, and that it may replace the quadrant 
 electrometer in comparing the capacities of condensers. 
 
 I^X, In response to a generally expressed wish, these 
 articles have been reprinted, as a preliminary to the 
 more complete work upon the subject which I hope 
 
 <^> to write. 
 
 GEORGE J. BURCH. 
 
 21. Norham Road, Oxford, 
 " 
 
 August, 1896. 
 
THE 
 
 CAPILLARY ELECTROMETER IN 
 THEORY AND PRACTICE. 
 
 The capillary electrometer was invented by Lippmann, and 
 described by him in a thesis published in 1875. Being specially 
 adapted for the study of electromotive changes of short dura- 
 tion, it attracted the attention of physiologists. In 1877 
 Marey employed it in investigating the functions of the 
 electrical organ of the torpedo, and succeeded in obtaining 
 photographic records of its indications. In England Prof. 
 Burdon Sanderson used it first in a research by which he 
 proved that the closure of the leaf of the Venus' Fly Trap 
 is preceded by electrical changes resembling those which 
 accompany the contraction of animal muscles. In a lecture 
 delivered before the Eoyal Institution in 1882 he exhibited 
 photographs of the excursions of the capillary electrometer 
 produced in this way. 
 
 My own connection with the subject dates from the winter 
 of 1886, when I was shown the originals of these photographs 
 and the apparatus which had been in use up to that time. As 
 none of the capillary electrometers were in working order, and 
 there was a difficulty in obtaining fresh ones in England, I 
 offered to make some, and succeeded, after a few trials, on the 
 last day of the year. Finding the whole subject to be of great 
 interest, I undertook also the installation of new photographic 
 apparatus, and since that time have devoted a good deal of my 
 leisure to perfecting the method and collaborating with Prof. 
 Burdon Sanderson in its physiological applications. 
 
( 2 ) 
 
 It is strange that an instrument so easily constructed and 
 so sensitive should have remained for nearly 20 years practi- 
 cally unknown, save to physiologists. But although the 
 principle of the electrometer had been studied up to a certain 
 point, the practical details essential from the standpoint of the 
 instrument-maker had not been worked out. My object was 
 to find out these details by a systematic investigation of the 
 mechanical and electrical constants involved and to publish 
 them. 
 
 After two or three months' work I began to see much greater 
 possibilities. There was a feeling among physiologists that 
 the indications of the capillary electrometer did not represent 
 quantitatively the value of rapid changes of E.M.F., because 
 the meniscus, although it moves very rapidly, requires an 
 appreciable time for its excursion. Accordingly such photo- 
 graphic records as had been obtained were relied on solely 
 to determine the direction of a current and the period, 
 but not the amount, of any change in its intensity. It 
 occurred to me that if the laws of the movement could be 
 determined it might be possible from the photographed 
 curve of an excursion to draw another curve representing 
 the variations of E.M.F. which caused it. I had already 
 nearly sufficient data, and after a few more experiments 
 was able, without using any hypotheses, to show that the 
 . velocity of the meniscus at any moment is directly propor- 
 tional to the distance through which it would have to move 
 before coming to rest under the influence of a constant 
 current equal in intensity to the P.D. between the terminals 
 at that instant. Furthermore, I invented a simple method of 
 measuring the velocity of the meniscus at any point on the 
 photograph of an excursion. 
 
 This discovery places the capillary electrometer in a unique 
 position. It becomes available for a class of work that no 
 other instrument can perform. A cycle of electromotive 
 changes can be plotted down within 1 per cent, of the truth, 
 at intervals of ^Vffth of a second, with the present apparatus, 
 and there is every indication that it might be driven five times 
 as fast. It seems probable that the capillary electrometer 
 may do for the dynamo what the indicator diagram does for 
 the steam engine. In the summer of 1894 I obtained by its 
 
( 3 ) 
 
 means photographs of the excursions produced by telephone 
 currents, in which the characteristic vibrations of the various 
 consonant sounds could be distinguished. 
 
 I propose in these articles to describe minutely the method 
 of making an electrometer, and to discuss the relation of its 
 physical constants to the dimensions and form of the capillary. 
 I shall then give full details of the apparatus for producing 
 the photographic records and measuring them, together with 
 a brief account of the applications of the method during the 
 seven years that it has been in use, including among the 
 illustrations the hitherto unpublished photographs of telephone 
 currents which I exhibited at the meeting of the British 
 Association in Oxford. These will be followed by the current- 
 curves of several dynamos recently photographed by myself. 
 I shall conclude with the methods of using the instrument for 
 direct observation without photography. 
 
 The principle of the capillary electrometer may be best 
 illustrated by the following experiment. Take a piece of soft 
 glass tubing of about 5mm. or 6mm. diameter, and draw it 
 out in the middle so as to form a capillary of about 0'5mm. 
 bore. Cut it in two, and bend each half into a U-shape, the 
 limb formed by the capillary being shorter than the wide limb. 
 Fill each with pure mercury and support them side by side in 
 a beaker of dilute sulphuric acid of 25 per cent., so that the 
 orifices of both capillaries may be immersed. Expel the air from 
 the capillaries by blowing down the wide limbs of the U -tubes. 
 
 The mercury will stand higher in the wide portions of the 
 two tubes than in either of the capillaries, and the difference 
 of level will be greatest in that U -tube which has the narrowest 
 capillary. Now pass a clean platinum wire into each of the 
 wide limbs, and connect these wires for a moment with the 
 derived circuit of a rheocord, or any convenient source of P.D. 
 not exceeding 0-5 volt. The mercury column in the capillary 
 by which the current enters will rise, and that by which it 
 leaves the circuit will be depressed. This altered condition 
 will persist after the circuit is broken, but on joining the two 
 platinum wires together the mercury will instantly return in 
 both tubes to its former level. Lippmann explains this pheno- 
 menon by saying that the surface-tension between the mercury 
 and the sulphuric acid is altered by the action of the current, 
 
 B2 
 
( 4 ) 
 
 and that this alteration is directly proportional between certain 
 limits to the P.D. between the terminals. In practice it is 
 unnecessary, and would be inconvenient to observe the move- 
 ment in both tubes, and accordingly one of them is made 
 larger in both limbs than the other, and serves to contain the 
 acid into which the capillary point of the second tube dips. 
 For some purposes, however, the arrangement above described 
 answers better than any other. 
 
 How to make a Capillary Electrometer. 
 
 In determining the form of the instrument the following 
 considerations must be borne in mind. The pressure of the 
 mercury against the glass must be sufficient to keep the acid 
 from creeping past it to the platinum wires, and the columns 
 should therefore occupy at least 5cm. of the vertical part of 
 the tubes. Provision must be made for washing out the acid 
 and replacing it with fresh, from time to time. 
 
 The maximum sensitiveness, and the greatest rapidity of 
 action, are only obtained together when the tube is cut off 
 at a certain part of its length. This is best done after the 
 electrometer has been put together. For these reasons I 
 much prefer those forms in which the capillary is not perma- 
 nently fixed to the U-tube. 
 
 Soft soda-glass answers best. The tubes should be 
 examined with a lens for capillary passages in the walls. 
 These, if large, are not always obliterated in the subsequent 
 manipulation, and they impair the definition under the micro- 
 scope. The selected pieces must be washed out three times 
 with aqua regia, and then rinsed very thoroughly with pure 
 distilled water. If on heating a tube so prepared over a 
 Bunsen burner and sucking air through it, the least taste 
 can be perceived, the cleansing is not sufficient. Finally, the 
 tubes are dried by aspirating air through them. The opera- 
 tion of making the electrometer should be proceeded with at 
 once. 
 
 The normal form for ordinary use is shown in Fig. 1. The 
 electrometer tube A is from 4mm. to 6mm. in diameter and 
 10cm. or 12cm. long from the top to the first bend. The 
 walls should be about 1mm. thick. To make it, cut off about 
 25cm. of the cleansed tube, and heat it in the middle before 
 
( 5 ) 
 
 the gas blow-pipe till it thickens, and the bore is reduced to 
 one-third of its original diameter. Remove it from the flame, 
 and very gently draw it out, till the centre portion is 1mm. or 
 l*5mm. in diameter. Fuse the middle of the narrow part in 
 the point of the flame till it separates, sealing up both ends 
 in the act. You have now a pair of tubes tapering at one end 
 to a strong, thick walled capillary, the bore of which is several 
 times larger than that of the finished instrument. The. next 
 step is to make the crook. Holding the tube in both hands, 
 
 FIG. 1. Normal Form of 
 Capillary Electrometer. 
 
 FIG. 2. The Pressure Apparatus. 
 
 heat it just above the tapering part for a distance of 1cm. 
 When it is soft a slight lateral displacement of the hands, the 
 two ends of the tube being kept parallel, forms both bends of 
 the crook in one operation. A little care is necessary to pre- 
 vent the tube from becoming flattened. Lastly, the upper 
 end of the tube is rounded in the flame, which must not be 
 allowed to blow into it. 
 
( 6 ) 
 
 The U-tube B has next to be made ; this may be of thinner 
 glass than the other. After forming a slight lip on one end 
 and rounding the walls of the other, it is heated in the flame 
 until it thickens and begins to close up. It is then very 
 slightly stretched out and at the same time bent into the 
 shape shown in the figure. The object of making the tube 
 narrow in the bend is to prevent the mercury from flowing 
 too freely from side to side when the instrument is carried 
 about. The clip C is of thin sheet brass. It is about 25mm. 
 wide, and is lined with a strip of morocco leather of the same 
 width. The tubes A and B are gripped firmly by it, but not 
 so tightly as to require undue force to adjust them. It remains 
 to draw the capillary point of the tube A. For this purpose 
 I employ the flame of an ordinary bat's wing burner turned 
 so low as almost to go out. Holding the tube in the left 
 hand and the drawn out end between the finger and thumb 
 of the right hand, I bring the middle of the narrow part into 
 the flame, the heat of which is just sufficient to soften, but 
 not to melt it. Removing it, I suddenly and firmly draw it 
 out, taking care to keep it parallel with the large end of the 
 tube. I then cut it across the middle. For this purpose I 
 use a small three-cornered file with the teeth ground off, 
 hardened by dipping it while red-hot into cold mercury. 
 Nothing but practice will ensure the drawing of a good 
 capillary, but if on examination with the lens it does not 
 appear satisfactory a piece of glass may be fused on and a 
 new point drawn. I have even made a new point to an old 
 electrometer tube without emptying out the mercury. The 
 instrument is now ready for putting together. The tubes are 
 slipped into the clip and twisted round until the end of A 
 comes directly over the mouth of B. Exact parallelism of 
 the capillary part of A with the walls of the U-tube B is 
 secured by slightly bending the brass clip C. Mercury is then 
 poured into A from a capillary pipette freshly made from a 
 piece of perfectly clean glass tubing. The tube B is also filled 
 with mercury to a depth of about 5cm. The mercury must 
 have been recently distilled in vacuo. 
 
 It remains to attach the pressure apparatus. This consists 
 of a pair of glass bulbs, P and Q, connected by a thick- walled 
 rubber tube. P is suspended by a cord passing over a small 
 
( 7 ) 
 
 pulley and attached to a bobbin, by turning which it can be 
 raised or lowered. The bulb Q has attached to it the four-way 
 tube R, leading to the pressure gauge S, the stop-cock T, and 
 the tube V, with which the electrometer is connected. The 
 stop-cock is necessary in order that the bulb Q may be easily 
 refilled with air in case of leakage. A piece of pressure tubing, 
 D, about 40cm. long, part of which is shown in Fig. 1, is slipped 
 over the tube V (Fig. 2) and secured with wire. 
 
 The mode of completing the circuit is seen in Fig. 1. F is 
 a terminal attached to a piece of brass tube, inside which a 
 platinum wire is soldered. This wire passes down the short 
 piece of rubber tube E nearly to the bottom of the straight 
 part of the tube A, and a similar wire is inserted in the open 
 limb of the tube B. The platinum should be made red-hot 
 before putting it into position, to burn off impurities. All the 
 joints should be made with burnt rubber, save that which 
 connects E with A, for which vaseline may be used, so that it 
 can be easily taken on and off. 
 
 A screw clamp fixed to the adjustible stage of a microscope, 
 with its tube horizontal, grasps the tube A firmly just above 
 the clip C, and an insulated support must be provided to take 
 the weight of the rubber tube D so as to prevent any undue 
 strain on the capillary. 
 
 When everything is ready the bulb P, Fig. 2, is raised, and 
 mercury flows into Q, expelling some of the air from it, and 
 thus forcing the mercury in the electrometer tube through the 
 capillary. When this is the case a few drops of dilute sul- 
 phuric acid of 25 per cent, are poured into the short limb 
 of the tube B, and the pressure-bulb immediately lowered till 
 the flow ceases. The circuit is next completed by connecting 
 the platinum wire in B with the terminal F, after which the 
 pressure may be cautiously increased until the mercury column 
 nearly reaches the tip of the capillary, when the electrometer 
 is ready for testing. The circuit must always be closed before 
 increasing the pressure, otherwise the instrument acquires a 
 charge which prevents the mercury from rising to its proper 
 position in the capillary. 
 
 Another form of electrometer, illustrated in Fig. 3, is useful 
 for some kinds of work. It differs mainly in having the point 
 of the capillary directed upwards instead of downwards. The 
 
tube B is reduced to a mere collar slipped over the wide part 
 of the U-shaped electrometer tube. A platinum wire twisted 
 round the neck of the capillary passes downwards between it 
 and the tube B, which may be fused on, or secured by cement. 
 Mercury is then poured into A and B, and a few drops of 
 acid are put into B. These electrometers work well for a 
 time, but do not last long if cement is used. 
 
 Owing to the curvature of the surface through which the 
 capillary has to be observed, the top and sides of the mercury 
 
 FIG. 3. Electrometer with 
 Upward Pointing Capillary. 
 
 FIG. 4. 
 Projection Electrometer. 
 
 column are not in focus simultaneously under the microscope. 
 The- definition is greatly improved by placing a thin cover- 
 glass against the tube B, to which a drop of acid causes it to 
 adhere. The acid has the advantage of not drying up, and 
 any other liquid would be likely to creep round into the 
 electrometer and spoil it. 
 
 For the production of photographic records by projection 
 more perfect definition is needed, and I accordingly employ 
 
( 9 ) 
 
 the modification shown in Fig. 4, in which nothing but a thin 
 cover-glass comes between the capillary and the objective. 
 The tube A is somewhat longer than the corresponding part 
 of Fig. 1, in order to allow room for two clips, G and C, 
 instead of one. The tube B is considerably larger, and much 
 more strongly lipped. In the shorter limb it carries a piece of 
 very thick-walled burette tubing, H, ground in like a stopper, 
 and projecting about 25mm. The front of this tube is ground 
 away, so that the bore of it is laid open from end to end, thus 
 forming a trough just wide enough to contain the capillary with- 
 out undue risk of breakage. A piece of thin cover- glass closes the 
 trough in front, and the surface tension of the acid keeps it 
 full of liquid. The clips C and G are clamped firmly by 
 means of screws not shown in the diagram, and they are so 
 arranged that the small adjustments necessary to bring the 
 capillary exactly parallel with the front of the trough can easily 
 be made. This instrument is somewhat fragile, but the definition 
 of the projected image is greatly superior. I am at present 
 engaged upon an improved form, which will need less careful 
 handling. 
 
 The excellence of these electrometers depends on the shape 
 and dimensions of the capillary. The following explanation 
 may therefore be found useful. When a glass tube heated in 
 the middle is drawn out, the walls of the central reduced portion 
 are approximately parallel for a certain distance. Between this 
 cylindrical part and the unaltered tube at each end is a portion 
 the bore of which tapers rapidly at first and then more slowly, 
 passing insensibly into the parallel walled capillary. If a 
 tube so drawn out is fixed vertically, and some mercury poured 
 into it, the mercury will be forced by its own weight some little 
 distance down the tapering part, until stopped by the action of 
 its surface-tension. On increasing the pressure with the 
 apparatus described in Fig. 2, the top of the mercury 
 column will be forced lower, but the meniscus at the bottom 
 of the column will move through a much greater distance, 
 owing to the small diameter of the tube at that part. The 
 total increase of pressure on the lower meniscus will therefore 
 be greater than that communicated to it from the pressure 
 apparatus by the additional length of the mercury column in 
 the electrometer tube. As the less tapering parts of the 
 
capillary are approached, this effect becomes greater and 
 greater, until a point is reached beyond which the meniscus 
 cannot be pushed without danger of its running down into the 
 cylindrical portion, when the mercury will continue to flow till 
 the tube is empty. To use such a tube for the capillary of an 
 electrometer, it must be cut off at some point short of that at 
 which the mercury begins to run through without further 
 increase of pressure. If the section is made too close to the 
 neutral point, the instrument will be very sensitive, but 
 troublesome to use, on account of the difficulty of stopping 
 the flow when once it has been established. On the other 
 hand, if the capillary is cut off too short, the sensitive part 
 will be sacrificed. 
 
 But although any tube thus drawn out can be cut so that it 
 will act as an electrometer, it does not follow that it can be 
 made into a good one. To ensure definiteness of response to 
 a difference of potential it is essential that friction should be 
 avoided. The working portion of the tube must therefore be 
 as short as is consistent with a good shape. For this purpose 
 I draw the tube down to a diameter of about 1mm, keeping 
 the walls fairly thick, and just before putting the instrument 
 together I soften the middle of this part in a very small flame, 
 and draw it out 15mm. or 20mm. Thus the actual capillary 
 is not more than 5mm. to 10mm. long. As the instrument 
 is always used under the microscope, with a power of from 
 20 to 400 diameters, this length is amply sufficient. The 
 end of the tube should be cut clean across. A splintered 
 orifice impairs the action of the instrument. 
 
 Capillaries which are to be used horizontally should be less 
 tapering if equal sensitiveness is desired. Those which point 
 vertically upwards should be of equal bore throughout the 
 working portion. As they can be made of constant sensitive- 
 ness and constant capacity for all positions of the meniscus, 
 they are useful for some kinds of work ; but the time-relations 
 of the excursion are more complicated, and they are, there- 
 fore, not so well fitted for the production of photographic 
 records. 
 
 In all downward-pointing instruments, sensitiveness to 
 small differences of potential is gained both by using a 
 smaller capillary, and by making it taper less rapidly. In 
 
( 11 ) 
 
 either case the movement is rendered slower, but from a 
 different cause. With a smaller tube, the internal electrical 
 resistance is greater ; but with a capillary of wider bore, made 
 equally sensitive by being more nearly cylindrical, the electro- 
 static capacity of the instrument is increased, and it may 
 take longer to charge although the resistance is less. Small 
 inequalities in the tube produce a greaterdisturbing effect on the 
 excursions, and the meniscus is more sensitive to mechanical 
 vibrations. There is, therefore, a limit beyond which the dia- 
 meter of the tube cannot profitably be increased. On the other 
 hand, a very fine tube is liable to get clogged, and although 
 it is very little affected by jars or vibrations, it is not nearly 
 so sensitive to electromotive changes of short duration. More- 
 over, the pressure-tube has to be wired on to the capillary, 
 and sooner or later leakages occur. Practically, I prefer tubes 
 that require a working pressure of about 20cm. and do not 
 care to go above 30cm. or below 10cm. After a little practice 
 it is easy to draw capillaries of any desired diameter. 
 
 When not in use the instrument should be invariably short- 
 circuited. For class work a spring key, held open by the 
 student daring an observation, is very useful. Should all the 
 mercury run out of the capillary the electrometer is hopelessly 
 ruined, as it is impossible to get rid of the sulphuric acid 
 which immediately enters. But by having a sufficient quantity 
 of mercury in the upper tube to completely fill the lower tube 
 in case of such an accident, this risk may be obviated. The 
 mercury may then be cautiously sucked back by a negative 
 pressure until it refills the capillary, taking care to stop before 
 any acid enters. Should the supply run short at any time 
 fresh mercury may be introduced with a fine pipette, inserted 
 in the top of the upper tube after disconnecting it from the 
 pressure apparatus. 
 
 Sometimes crystals form in the capillary. If these cannot 
 be expelled they may be decomposed by electrolysis, using a 
 P.D. of about 0'5 volt, and making the capillary the cathode. 
 Inasmuch as this process tends to form crystals at the anode, 
 these must be carefully washed out of the U-tube and fresh 
 acid substituted. Old electrometers in which no crystals are 
 visible are often much improved in this way. They must be 
 subjected to the action of the current for an hour or more. 
 
( 12 ) 
 
 The pressure should be invariably taken off when the 
 instrument is left. The weight of the column of mercury in 
 the tube itself is sufficient to keep the acid from creeping up, 
 and if the meniscus is allowed to stand continually in the 
 working portions of the capillary the mercury is likely to 
 stick whenever it passes that part. 
 
 A good instrument should respond easily to ^TO^th f a 
 volt. 
 
 If the meniscus recedes when the electrometer is on short 
 circuit it is a sign of leakage in the pressure tubing. If it 
 "creeps" up or down when the circuit is open there is an 
 E.M.F. somewhere, either due to a dirty contact or to the 
 acid in the electrometer having crept past the mercury to the 
 platinum wires ; the latter defect is fatal. If when a charge 
 has been communicated to it and the key is open, the mercury 
 gradually returns to zero, the insulation is defective. Not 
 more than 5 per cent, of the charge should escape in half- an - 
 hour. If made with clean and pure materials an electrometer 
 should remain in good order for several years. 
 
 Induction shocks, strong enough to be felt, exert a very 
 deleterious effect ; the sudden jar appears to disintegrate 
 the glass where the meniscus touches it, causing the mercury 
 to catch as it passes along the tube, though no mark can be 
 seen. But a step- down transformer with a couple of cells on 
 the primary may be used with impunity. The capillary 
 electrometer is not affected by ordinary magnets, nor by 
 proximity to an arc light, but wires in circuit with it should 
 not be carried near high-tension alternating currents. 
 
 Physical Constants of the Capillary Electrometer. 
 
 Both from the point of view of the manufacture of the 
 capillary electrometer and of its application as an instrument 
 of research, Prof. Lippmann's Paper left a great deal to be 
 done. The question of the electrical capacity of the instru- 
 ment and of its resistance, and of the time relations of its 
 excursions, were as yet untouched. 1 determined to attack 
 these problems systematically by the experimental method. 
 
 The point to be first determined was whether any leakage 
 of current could be detected through the instrument itself 
 after the meniscus had ceased to move under the action of a 
 
( 13 ) 
 
 difference of potential. I joined up a high-resistance galvano- 
 meter of great sensitiveness, and the electrometer, in series in 
 the derived circuit of a potentiometer. The galvanometer 
 being short-circuited, I made the battery connection, and 
 waited till the meniscus of the electrometer had come to rest 
 in the new position corresponding to the P.D. employed. I 
 then opened the short circuit of the galvanometer. There 
 was no deflection. But on increasing or diminishing the 
 E.M.F. of the derived circuit, there was a temporary current 
 one way or the other while the meniscus was in move- 
 ment.* 
 
 The corresponding problem was this : Does an electrometer 
 which has been charged to a certain P.D. lose its charge when 
 disconnected from the source of electricity ? I found that it 
 does not, if the external insulation is good, but that owing to 
 the necessary presence of the dilute sulphuric acid, perfect 
 insulation is not easy to attain. Still, a displacement of the 
 meniscus through 50 divisions of the eye-piece micrometer 
 was maintained for more than an hour with a loss of less 
 than one division. We are led, therefore, to the conclusion 
 that the capillary electrometer, although its circuit is com- 
 posed entirely of conductors, does not transmit a current, but 
 merely receives a charge as though it were a condenser. But 
 there is no visible dielectric between the mercury and the 
 acid. I have left an instrument for a whole day under the 
 strain of an E.M.F. of half a volt, and not a trace of hydrogen 
 could be detected in the capillary. 
 
 The next point was to determine the capacity of the electro- 
 meter and its relation to the form of the tube. The following 
 method was adopted. A standard condenser was charged to a 
 known P.D. by means of a potentiometer. It was then sud- 
 denly connected with the uncharged electrometer, causing the 
 meniscus to move from zero to a position corresponding to the 
 E.M.F. of the charge communicated to it. The actual value 
 of this E.M.F. was found by switching the potentiometer on 
 to the electrometer circuit, and shifting the rider till the 
 meniscus was adjusted to the observed position. From these 
 data the capacity of . the electrometer was calculated by the 
 usual formula. Double break keys were used, and they and 
 * I found afterwards that Fleischl had tried this in 1879. 
 
the commutators were arranged like the keys of a piano, so 
 that the operations could be conducted with great rapidity. 
 
 The results of a large number of experiments with about 40 
 instruments may be summed up as follows : 
 
 1. The narrower the tube the smaller the capacity. The 
 greater the sensitiveness the greater the capacity. 
 
 2. As in most cases the capillary tapers less rapidly towards 
 the tip, and is consequently more sensitive in that part, so 
 also the capacity of most electrometers is somewhat greater 
 when the meniscus is near the tip than when it is farther 
 from it. 
 
 3. The capacities of the instruments examined ranged from 
 0*1 microfarad to 3OO microfarads from O5 to 2-0 giving 
 the best results. 
 
 4. The electrical capacity as thus measured is that of the 
 part of the capillary at which the meniscus comes to rest 
 under the influence of the charge, and may be greater or less 
 than that of the point from which it started to move. 
 
 5. With this proviso, the capacity is entirely independent 
 of the potential difference of the condenser charge, which may 
 be as great as can be safely employed, or as small as can be 
 accurately measured. 
 
 6. The capacity is the same for any given part of the tube, 
 whether the zero be above it, so that the condenser charge 
 causes the mercury to fall ; or below it, so that a similar 
 charge in the reverse direction causes the mercury to rise, so 
 long as it comes to rest when charged at the same part of the 
 tube in both cases. 
 
 7. The capacity can be measured equally well by causing 
 the charged electrometer to share its electricity with an 
 
 uncharged condenser of known capacity. 
 
 8. Some capillaries are equally sensitive through a consider- 
 able portion of their length. In such cases the capacity is 
 constant throughout the same part. 
 
 9. The capacity of a newly made electrometer usually, 
 though not always, increases during the first few days. After 
 about a week it becomes constant. I am inclined to ascribe 
 this to the presence of air or gas condensed on the surface of 
 the glass. 
 
( 16 ) 
 
 It is evident from these experiments that the capillary 
 electrometer may be regarded for all practical purposes as a 
 condenser possessing a definite electrical capacity for each 
 position of the meniscus, and that this capacity is independent 
 of the potential difference of the charge, which is a function of 
 the change of pressure necessary to keep the mercury stationary 
 in that position. 
 
 The change of level of the end of the mercury column pro- 
 duced by putting it in connection with a constant current is 
 due to the flow of a measurable quantity of electricity. It also 
 involves the performance of a definite amount of work, 
 namely, in lifting the column of mercury from one position to 
 the other, in altering the shape and area of the meniscus, and 
 in overcoming friction. Evidence of the effect of friction can 
 be observed in electrometers of which the tube is not perfectly 
 clean, but otherwise I was unable to detect it.* The shape of 
 the meniscus while at rest in any given portion of the tube is 
 not visibly different under the action of a difference of potential 
 in either direction, its outline projected on a screen with a 
 magnification of 250 diameters remaining unaltered. The 
 difference of section area in the working portion of the 
 capillary of a sensitive electrometer is also small compared 
 with the variation of the length of the mercury column. 
 Putting P for the total pressure, I for the distance through 
 which the meniscus is moved, and r for the radius of the 
 tube, we may say that, approximately, 
 
 where V is a very small difference of potential and C is the 
 capacity of the tube at the point where the meniscus stops, 
 and a is a constant depending on the instrument used. But 
 the working pressure varies inversely as the radius of the 
 
 capillary. Writing therefore -~ = r, and omitting constants, 
 
 I 
 we find that is proportional to the work done, and propor- 
 
 tional also to the quantity of the charge C V. I determined 
 
 this ratio for a large number of instruments, and found it to 
 
 be fairly constant, in spite of the great differences in sensi- 
 
 * Fluid friction affects the velocity, but not the extent of an excursion. 
 
( 16 ) 
 
 tiveness, diameter and electrical capacity of the capillaries 
 employed. 
 
 ^Resistance. 
 
 The resistance of the circuit has not the same effect upon 
 the capillary electrometer as upon a galvanometer. It simply 
 diminishes the rapidity of the movement without in the least 
 reducing its extent unless, of course, the instrument is partly 
 short-circuited by defective insulation. This being so, it is 
 impossible to measure the resistance of the electrometer by 
 the ordinary methods. 
 
 When a difference of potential is communicated to the 
 instrument, the meniscus moves with a gradually- diminishing 
 rapidity to its new position of rest, which is attained only 
 after the lapse of an appreciable, but indefinite, time. The 
 time required to fully charge the instrument cannot be 
 measured. But the time of half-charge i.e., when the P.D. 
 between the terminals of the electrometer is half that of the 
 source, is well defined, and easy to measure. I effected it, in 
 the first place, by means of a rheotome arranged so as to close 
 the circuit of a potentiometer, and break it again after a 
 known fraction of a second. The duration of the closure 
 was varied until the excursion produced by it was equivalent 
 to that given by a constant P.D. of half the strength. I then 
 added a resistance of 100,000 ohms to the electrometer circuit 
 and found that a longer closure was necessary to produce the 
 same excursion, namely, that of the half-charge. The total 
 quantity of electricity being the same in each case, it was 
 possible from these data to calculate the " equivalent resist- 
 ance" of the electrometer. This varied, in the different 
 instruments, from 90,000 ohms to 700,000 ohms. It is con- 
 venient in practice to know the " equivalent resistance" of an 
 electrometer, especially when high resistances have to be used 
 or a shunt introduced. But the actual resistance varies from 
 point to point of the tube, and necessarily so, since it is due, 
 to a considerable extent, to the low conductivity of the dilute 
 acid contained in the portion of the capillary not occupied by 
 the mercury. But it is not wholly due to this. I found by a 
 series of careful measurements of the time of half- charge, with 
 the zero position of the mercury at different distances from the 
 tip of the tube, that the " equivalent resistance " does not vary 
 
( 17 ) 
 
 so rapidly as it should do it' it depended solely on the resist- 
 ance of the sulphuric acid and the mercury. Some other 
 cause probably mechanical conditions the velocity of the 
 movement, so that the equivalent resistance at any given part 
 of a capillary would be expressed by the formula 
 
 where I = the distance of the meniscus at any moment from 
 the tip of the tube, and L is a constant of the particular 
 instrument employed. 
 
 With an electrometer adapted for the production of photo- 
 graphic records, the determination of the equivalent resistance 
 for any given position of the capillary is extremely simple. Two 
 normal excursions are photographed, one with say 100,000 
 ohms in circuit, and the other with no resistance, or if the 
 electrometer is affected by overshooting, with a resistance of 
 10,000 ohms. The angle of inclination of each curve at 
 corresponding points is then ascertained in a way that will be 
 subsequently described, by measuring their respective sub- 
 normals. Let the subnormal to the first curve be N, and let 
 that to the same part of the second be N', 
 then N - N' ; N' : ; 100,000 ohms : R, 
 
 where R = the equivalent resistance of the electrometer at 
 that point. This method has been used by me in the Physio- 
 logical Laboratory, Oxford, ever since July, 1890. 
 
 Production of Photographic Records. 
 
 The capillary electrometer is essentially the poor man's 
 instrument, costing, even including the price of a microscope, 
 less than any other form of electrical measuring instrument of 
 equal delicacy. It is also by far the easiest to make. But it 
 has certain disadvantages. It is fragile and easily spoilt by 
 careless usage, and unless employed for zero methods, requires 
 careful calibration against a standard instrument. It has, 
 however, one property which outweighs all these disadvan- 
 tages, and in virtue of which it stands alone among 
 the implements of research. The rapidity with which 
 it responds to all changes of potential is so great 
 that nothing but photography can render all its move- 
 ments visible. It is accordingly this branch of the 
 
( 18 ) 
 
 subject to which I have devoted a great deal of niy time. 
 The form of the instrument is such as to render the production 
 of photographic records of its excursions extremely easy. 
 The magnified image of the mercury column is projected on 
 to a narrow vertical slit, behind which a sensitive plate travels 
 with a uniform horizontal motion. It thus forms a continuous 
 band of shadow, varying in height according as the position of 
 the meniscus changes. But however sharply the top of the 
 mercury column may be focussed on the screen, a sudden to- 
 and-fro excursion of the meniscus will produce a blur instead 
 of a sharp-edged shadow, unless the plate moves through a 
 space greater than the width of the slit during its occur- 
 rence. And if the slit is made very narrow, not only is the 
 illumination feeble, but diffraction comes into play and spoils 
 
 FIG. 5. Hydraulic Mottr. 
 
 the definition. My first improvement consisted in the use of 
 a cylindrical lens of 25mm. focus placed about 25cm. behind 
 the slit. This had two advantages. It concentrated the light 
 from a much wider slit into a mere line, and it increased the 
 working distance between the sensitive plate and the appa- 
 ratus. But it necessitated the adoption of a more perfect 
 machine for propelling the plate. Driving clocks were found 
 to be useless, and I therefore designed the hydraulic motor 
 illustrated in Fig. 5. 
 
 A is a brass tube 3cm. in diameter and 110cm. long. It 
 has a slot about 1cm. wide and 45cm. long cut in the middle 
 portion. It is flanged at each end, and provided with taps, 
 P and G. The piston B is tubular, and has a cup leather 
 packing at each end, and a tap C in the middle. To this 
 
OF THK ' IT ^\ 
 
 UNIVERSITY!! 
 
 ( 19 ) 
 
 piston B is attached the plate holder D, which is guided at 
 its upper part by the brass rod E held parallel to the tube A by 
 the wooden framework of the apparatus L, L, M. The tap G 
 is connected with the water supply, and a waste-pipe is led 
 from F to the nearest sink. The lower end of the waste-pipe 
 must be kept constantly under water to prevent the entrance 
 of air. For the same reason the supply pipe must be fitted 
 with an air trap if the water is taken direct from the main. 
 The tap H serves to empty the central portion of the tube A 
 of any water that may leak past the piston heads and of the 
 surplus oil used for lubricating it. 
 
 To use the instrument, close the taps F and G and open the 
 middle tap C. The plate-holder D can then be drawn as far 
 as it will go to the left hand, the water passing from one end 
 of the tube to the other through the hollow piston rod. C 
 having been closed the apparatus is ready for action. The 
 tap F on the waste pipe is opened more or less according to 
 the speed at which the plate is require to travel. When all 
 is ready, the tap G is suddenly turned full on. The pressure 
 of the water upon one end of the piston is resisted by the 
 friction of the flow through the tap F, and the result is a very 
 steady motion of the plate-carrier from left to right. The 
 power of the instrument is considerable, and the rate is con- 
 stant after the first 2cms. The speed may be varied from 
 0-3cm. to 5cm. per second, by adjusting the tap F, the long 
 handle of which comes np against a set screw. 
 
 This machine answered admirably. Electrical phenomena 
 lasting from 4sec. to 2min. were recorded by its means, and 
 the resulting photographs were perfectly sharp and even. 
 But we had not realised with what great rapidity the capillary 
 electrometer can respond to changes of potential. 
 
 While examining with a lens one of the physiological 
 photographs which will be described later on, I noticed a well- 
 marked notch in the commencement of the curve, showing 
 that two impulses had been given to the mercury in rapid 
 succession, and that it had responded first to one and then to 
 the other. It was known that two changes of potential 
 difference had actually taken place in the circuit ; but the 
 interval between them could not possibly be greater than 
 0-Olsec. This indicated that the instrument was capable of 
 
 c2 
 
acting much more rapidly than we supposed. I accordingly 
 designed a simple experimental arrangement for trying higher 
 velocities. It was in the form of a balanced pendulum, and 
 carried the plate in a circular arc instead of a straight line. 
 This was at first considered a disadvantage ; but shortly after- 
 wards I discovered that, owing to a peculiar property of polar 
 curves, it was in reality far better for our purpose, as will be 
 seen from the description of the mode of measuring the 
 photographs. The principle was, therefore, retained, and 
 
 FIG. 6. Pendulum Motor. Plan. 
 
 FIG. 7. Pendulum Motor. Side Elevation. 
 
 subsequent alterations have been simply improvements of the 
 details of the apparatus. 
 
 The Pendulum Apparatus. 
 
 The present arrangement of the photographic apparatus is 
 shown diacrrammatically in plan in Fig. 6 and in elevation in 
 Fig. 7. A is an ordinary 5-inch condenser, in the focus of 
 which is placed an arc lamp on an adjustable stand. A tank, 
 b, filled with water or alum solution, serves to intercept some of 
 the heat rays. Sometimes a second tank containing ammonio- 
 
( 21 ) 
 
 sulphate of copper is also used. This effectually stops the 
 heat, and does not appreciably diminish the actinic power of 
 the light, but makes focussing rather difficult. Next comes 
 the microscope, m, with its condenser, and the electrometer. 
 This part of the apparatus is situated in a window in the wall 
 of the dark room. The microscope, which is fixed to the 
 solid stone, occupies half the window, and the other half is 
 closed by a door made perfectly light-tight. The electrometer 
 is supported on a stand provided with vertical and horizontal 
 rack work motion and two tilting adjustments for setting the 
 capillary vertical and at right angles to the optic axis. The 
 tube of the microscope is fitted with a light-tight joint where 
 it passes through the partition W, which it does not touch. 
 
 Inside the dark room the apparatus is attached to the floor, 
 and is only connected with the window by a creased paper band, 
 which keeps out the light without transmitting mechanical 
 vibrations. The rays pass first into a narrow box 50cm. 
 long, with a sliding lid, /, and a pane of ruby glass, g, at the 
 farther end. By opening the lid, and placing a piece of paper 
 in the path of the light, the movements of the mercury can be 
 observed, and the ruby glass allows the operator to watch the 
 meniscus up to the very moment of the exposure. At the end 
 of the box c are the two false slits d, and between them is 
 the shutter e. This shutter works on a pivot underneath the 
 apparatus, and is connected with the same handle that releases 
 the pendulum I. Immediately beyond the false slits are the 
 tuning fork and the signal, which will be separately described, 
 and close behind them is the true slit h. This is contained 
 in a separate frame, so that it can be removed entirely during 
 focussing and replaced exactly in the centre of the image of 
 the mercury column. The width of the slit can be adjusted 
 to suit different velocities of the plate. Behind the slit comes 
 the cylindrical lens j. This is of 25mm. focus and is placed 
 about 25cm. from the slit. It is provided with an adjust- 
 ment for focussing, and is fixed to a slide so that it can be 
 drawn up quite out of the path of the light. The plate carrier, 
 and focussing screen k, are fixed to^the top of the pendulum I- 
 
 Fig. 8 is a diagram of the apparatus for propelling the plate, 
 seen from behind. It consists of two parts, viz., the pendulum 
 and the automatic key for determining the instant at which 
 
( 22 ) 
 
 an experiment shall commence. The pendulum A is a wooden 
 bar 125cm. long by 12cm. wide, which swings by an axis 
 at its centre on bearings fixed to the table. To avoid friction 
 the shaft is very small, but it is supported on each side of both 
 bearings by a solid bracket bolted to A. The bearing nearest 
 the pendulum has no cap. The shaft rests upon it, and the 
 weight is so arranged that the other end of the shaft, in which 
 a V-shaped groove is turned, presses upwards into a semi-cir- 
 cular notch in the edge of a brass plate. Thus the pendulum 
 can be easily unshipped, and at the same time the plane of its 
 movement is always at a fixed distance from the cylindrical 
 lens. C is the plate carrier. It is made to hold an ordinary 
 dark slide. S is a ground- glass focussing-screen. At the 
 lower end of the pendulum is a movable weight, B. This has 
 to be fixed so that when the dark slide is in place with a plate 
 in it and the shutter drawn up, the pendulum will rest 
 indifferently in any position. The arm D carries a weight, E, 
 attached to it by a cord. In the actual instrument the weight 
 is replaced by a board, hinged to the table at one end and 
 loaded more or less heavily according as a slow or high speed 
 is required. 
 
 The action is as follows : the pendulum is held in the 
 position shown in the figure by the latch G. When this is 
 raised the weight E causes the pendulum to swing over with 
 increasing velocity until it rests on the table, just before the 
 sensitive plate comes into the line of light. The acceleration 
 now ceases, and the pendulum being in neutral equilibrium 
 continues to move with a constant angular velocity until 
 caught and held by the catch H. To counteract the slight 
 retardation due to friction, a small weight, fixed to the bar D, 
 is adjusted until the vibrations of a tuning fork, recorded on a 
 long negative, measure the same at each end of it. For fast 
 rates, up to 150cm. per second, a weight of 2 kilos, is 
 employed. For slow speeds the bar A is loaded with equal 
 weights at both ends, and the weight E is reduced until it is 
 only just sufficient to act with certainty. 
 
 The apparatus is capable of two classes of work, viz., it can 
 record a short portion of a series of repeated movements of the 
 image, as in investigating dynamo currents, or the vibrations 
 of a spring ; or it can register the effect of a single excitation 
 
( 23 ) 
 
 of a muscle, or of an induced current. For the latter class of 
 work it is necessary to have means of determining with great 
 accuracy the exact instant at which the experiment com- 
 mences. This is done by the arrangement shown to the left 
 hand in Fig. 8. M is a bar attached to the pendulum A. It 
 
 H 
 
 FIG. 8. Pendulum Motor. End Elevation. 
 
 carries a pin on which rests the light arm J of the automatic 
 key. This is furnished with a contact screw, K, dipping into 
 a cup of mercury, L. Terminals are provided, by which this 
 key is connected with a battery, and with the signal key shown 
 
on a larger scale in Fig. 9. It is essential that the screw K 
 should be connected with the terminal by a flexible wire, and 
 not through any rubbing contact. By adjusting K, and then 
 moving the pendulum slowly across by hand, this key may be 
 made to break contact when the plate reaches any desired 
 position, with an error of not more than 1mm. It only serves, 
 however, to actuate the signal key, which records the exact 
 instant of the stimulus in the act of giving it. In Fig. 9, A 
 and B are the coils of a small electromagnet, the terminals of 
 which, M lf M 2 , are in circuit with the automatic key. The 
 armature C is connected by a piece of watch spriog with a 
 brass pillar on which is the terminal P 2 . It rests upon the 
 pole of A by a knife edge, and is just kept from touching B by 
 the point of the set- screw D, connected with the terminal P r 
 The end of C, which is very light, but stiff, projects somewhat 
 
 FIG. 9. Signal Key. 
 
 beyond the rest of the instrument. A stop, not shown in the 
 figure, prevents the spring from moving the armature more 
 than a couple of millimetres away f i-ora the pole. The signal 
 key is fixed in the position indicated in Figs. 6 and 7, between 
 h, and d, so that the end of the armature crosses the slit near 
 the upper part. As long as the automatic key is closed, the 
 magnets A and B of the signal key hold the armature C 
 against the screw D, but directly the plate comes into the 
 requisite position, the circuit M x M 2 is broken and the 
 armature springs back, breaking the circuit P x P 2 and record- 
 ing the exact instant of so doing by the sudden drop of its 
 shadow upon the sensitive plate. 
 
 It will be noticed that in this key also there are no rubbing 
 contacts, and only one place at which the circuit can be broken. 
 I sometimes use a modification in which a spring key is struck 
 
( 25 ) 
 
 open by a small hammer connected with an electromagnet. 
 But the principle is the same the shadow of the key itself is 
 recorded on the plate. 
 
 It remains to describe the time-marker. This is simply a 
 standard fork making 500 doable vibrations per second. It is 
 fixed horizontally so that the lower prong projects across the 
 slit at its upper part, leaving a mere line of light 1mm. long 
 above it. The prongs are forced apart by a cam so as to allow 
 a block attached to one end of a pivotted lever to be placed 
 between them. On pulling the lever the fork is released, and 
 vibrates with sufficient amplitude for 15 or 20 seconds. 
 
 To illustrate the working of the machine I will describe the 
 process of photographing a " normal excursion," i.e., the 
 excursion produced by placing the uncharged electrometer 
 suddenly in connection with a constant source of E.M.F. of 
 known intensity, e.g., 0-01 volts. 
 
 The derived circuit of a potentiometer is put in series 
 with the electrometer terminals, and a short-circuit across 
 them is connected with the terminals Pj and P 2 of the 
 signal key. The arc lamp having been switched on, the lid 
 f (Fig. 7) is slid back and a piece of paper placed in the path 
 of the light. The pressure-bulb is then raised until the 
 meniscus appears in the capillary in the proper position. The 
 light-tight door in the window of the dark room is kept open 
 while the electrometer stand is adjusted so that the image of 
 the mercury column appears on the shutter in the middle of 
 the false slit. The shutter e is then opened, the true slit h 
 removed, the cylindrical lens j drawn up out of the way, and 
 the pendulum placed so that the magnified image falls on the 
 focussing screen S (Fig. 8). The top of the meniscus is then 
 focussed as sharply as possible, with the aid of a lens. A 
 long rod, not shown in the figure, passes through the partition 
 W to the fine adjustment of the microscope. Two pins on 
 the end of this rod engage in two large holes in the head of 
 the screw, which can thus be adjusted with great nicety. 
 When the focussing is finished the rod is drawn back clear of 
 the screw. The next step is to replace the slit h exactly in 
 the centre line of the mercury column. Then the cylindrical 
 lens is pushed down into its place, and focussed until the 
 image of the slit appears upon the screen as a mere line of light. 
 
( 26 ) 
 
 It remains to test the circuit. The rider of the potentio- 
 meter is first lifted off the wire so as to break the electrometer 
 circuit. Placing his finger and thumb on the terminals of the 
 electrometer, the operator rubs first one and then the other. 
 The meniscus ought to move up and down freely, remaining 
 up or down if the fingers are removed, and returning to zero 
 sharply when the key is closed. But when the rider is on the 
 potentiometer, it ought not to be possible to cause the mercury 
 to move by rubbing the terminals, because the circuit is no 
 longer open. 
 
 A dark slide containing a plate is now put into the plate- 
 holder and the pendulum pushed back into the position shown 
 in Fig. 8. This action closes the automatic key, so that on 
 raising the armature C of the signal key (Fig 9) it will be 
 held by the magnet. The short-circuit of the potentiometer 
 being now closed, no charge should enter the electrometer, 
 and accordingly there should be no movement of the meniscus 
 on opening the electrometer key. It remains to set the tuning 
 fork by placing the block between its prongs. The light-tight 
 door is closed, the sliding lid shut to, and the shutter of the 
 dark slide drawn up. The tuning fork is made to sound, and 
 a second later the handle of the latch G (Fig. 8) is raised. 
 This opens the exposing shutter and at the same time releases 
 the pendulum, which swings over. The automatic key releases 
 the signal key, and this in its turn breaks the short circuit of 
 the potentiometer, the current from which flows into the 
 electrometer and charges it, the shadow of the moving 
 meniscus forming a curved band upon the sensitive plate. 
 
 As soon as the operator hears the clash of the pendulum 
 against the catch H, he lowers the handle, thus closing the 
 exposing shutter, and removes the sensitive plate. 
 
 After the first adjusment very little attention is needed, 
 save an occasional glance at the focussing screen, and plates 
 can be exposed almost as quickly as in an ordinary camera. 
 
 It should be pointed out that the actual form of the 
 apparatus was determined by the position in which it had to 
 be erected. I do not confine myself to the pendulum motion. 
 I have also made a portable camera in which the sensitive 
 plate is attached to a disc which for very high speeds is made 
 to revolve continuously. It is provided with an arrangement 
 
( 27 ) 
 
 for giving the exposure as soon as the requisite velocity is 
 attained. Details of this machine will be communicated 
 hereafter. 
 
 Measurement of tlw Potential Difference indicated by Rapid 
 
 Movements of the Mercury of the Capillary Electrometer. 
 In all other electrical instruments measurements are effected 
 by noting the amount of a completed deflection. It was a new 
 thing to attempt the estimation of an E.M.F. by determining 
 the initial velocity of the excursion caused by it. I had long 
 believed that this would be possible in the case of the ballistic 
 galvanometer, and accordingly set myself to determine the 
 time relations of the capillary electrometer with a view of 
 putting the method into practice. While the photographic 
 apparatus was being made I had studied the physical constants 
 of the instrument, and was now ready to attack the more 
 important problem. I knew the conditions to be complex. 
 A mechanical effect was produced by an electric cause, and the 
 rate of movement might be determined by either or by both. 
 In the former case the flow of liquid might be governed by 
 Torricelli's law or by Poiseuille's, or it might take place 
 according to some intermediate formula. But if electrical 
 causes predominated, then there were the possible variations 
 of sensitiveness and capacity, and the inevitable alteration 
 of resistance, as the mercury approached the end of the tube, 
 to take into account. It would have been futile to seek 
 hypotheses. I therefore made experiments. 
 
 (1) Dilution of the acid beyond a certain point lessened 
 the rapidity of an excursion without diminishing the extent of 
 it. This afforded additional evidence of the effect of electrical 
 resistance. 
 
 (2) Shortening the capillary, so as to reduce the length of 
 the acid column above the mercury, made the instrument act 
 more quickly. But here both frictional and electrical resist- 
 ance were lessened. 
 
 (3) The shape of the tapering part of the tube as well as 
 of the orifice, had a marked influence on the rapidity of the 
 excursions. This was clearly a mechanical effect. 
 
 (4) The next step was to ascertain whether there was 
 any delay between the communication of a P.D. to the 
 
( 28 ) 
 
 electrometer, and the commencement of the movement of the 
 meniscus. 
 
 This was done by photographing a normal excursion in the 
 manner already described, although not with that apparatus, 
 which had not then been constructed. An examination of the 
 photograph showed that the fall of the signal was synchronous 
 with the commencement of the excursion. 
 
 The signal key was then arranged to actuate a rheotome, by 
 which short currents of known duration were thrown into the 
 electrometer. The photographs obtained showed that with 
 such differences of potential as would be employed in practice, 
 the mercury ceases to move directly the source of electro-motive 
 force is withdrawn, i.e , the instrument is practically dead-beat. 
 
 It has been generally stated that it is perfectly dead beat, 
 but this is only true of " slow" instruments. A strong con- 
 denser charge communicated to a quick electrometer such as 
 I employ, will in some cases cause the meniscus to overshoot 
 its final position of rest by as much as 10 per cent, of the 
 excursion, if there is no resistance in circuit. But the 
 insertion of 50,000 ohms makes it dead-beat, without seriously 
 reducing the rapidity of the action. Overshooting is more 
 marked with the condenser charge than with rheotome 
 experiments. The legitimate conclusion to which these ob- 
 servations tend is that in practice the meniscus commences to 
 move the instant a potential difference is communicated to it, and 
 stops directly it is withdrawn. Moreover, the movement com- 
 mences quite suddenly. No trace can be detected, even with a 
 lens, of gradual acceleration. The inference follows that " the 
 velocity of the meniscus at any moment must be some function of 
 the accelerating force at that moment" i.e., it is independent of 
 any previous motion and contributes nothing to the velocity 
 with which it moves during the next interval. 
 
 It remained to determine the form of this function. This I 
 did in two ways, with electrometers differing as widely as 
 possible. 
 
 For "slow" instruments I employed the method of direct 
 observation with the rheotome already referred to. The 
 potentiometer was arranged so as to give with 150mm. of 
 wire a total excursion of 126 divisions of the eyepiece micro- 
 meter. The rheotome was then switched on to the circuit, 
 
ex 
 
 ( 29 ) 
 
 and the period of closure adjusted until it produced an 
 excursion of exactly half the previous extent namely, 63 
 divisions. This required a closure of O414sec. The rider of 
 the potentiometer was then shifted to 50mm. This gave, 
 with the constant current, an excursion of 42 scale divisions, 
 and with the rheotome 21 divisions. With the rider at 
 25mm. the total excursion was 21 divisions, and that given 
 by the rheotome 10-5 divisions. The experiments were 
 repeated with the direction of the currents reversed and with 
 other electrometers, and the results were similar. 
 
 There is one simple function that fulfils these conditions. 
 
 Let y = the distance through which the meniscus has to 
 move before reaching its position of rest under the action of a 
 constant current. 
 
 Let x = indefinite time required to complete the excursion. 
 Then, if x and y be so related that 
 
 log2/= -ex, 
 we shall have 
 
 log (I) = log 2- 
 
 that is to say, the time occupied in tracing the first half of a 
 curve having x and y for its co-ordinates will be independent 
 of the initial values of both x and y. 
 
 I had not expected to find so simple an expression for the 
 time relations of the excursion, and it still remained to be 
 seen whether in the more rapid electrometers used for photo- 
 graphy the same relation would hold good. I accordingly 
 took a number of photographs of normal excursions, using 
 first the hydraulic motor, and afterwards the more perfect 
 pendulum machine. 
 
 The results are embodied in my Paper, " On the Time 
 Relations of the Capillary Electrometer," published in the 
 Philosophical Transactions, Vol. 183A, p. 81. 
 
 Briefly, it may be stated that although the complete 
 expression for the time relations of the excursion is very 
 complicated, it is easy to select instruments in which the two 
 principal sources of error neutralise each other, and that these 
 instruments fulfil the following laws* : 
 
 * Since my results were published, Prof. Einthoven, of Leyden, has 
 arrived independently at the same conclusions, and I desire to acknowledge 
 his courteous recognition of my claims to priority in the matter, G. J. B, 
 
( 30 ) 
 
 1. The distance of the meniscus from its zero position at 
 any instanfc is proportional to the P.D. of the charge contained 
 in it at that instant. 
 
 2. The velocity of the meniscus at any instant is propor- 
 tional to the difference between the E.M.F. of the external 
 circuit at that instant, and the E.M.F. of the charge already 
 contained in the electrometer. 
 
 3. In all cases in which the action is dead beat, the algebraic 
 sum of the P.D. represented by the distance through which 
 the meniscus has already moved, and that indicated by the 
 velocity with which it is still moving, equals the total E.M.F. 
 acting upon the electrometer at that instant. 
 
 This law holds good during the most rapid changes in the 
 direction and intensity of a current. 
 
 FIG. 10. Diagram to Explain the Theory of the Subnormal. 
 
 I now come to the most important part of my discovery, 
 viz., the method of analysing the photographic records and 
 expressing in volts the variations of E.M.F. indicated by the 
 electrometer. 
 
 I shall here deal only with the method as it relates to the 
 curves obtained with the pendulum machine, because with 
 these the process is more simple and more accurate. The 
 principle of the operation is shown in Fig. 10, and the measur- 
 ing machine itself will be shown in Figs. 12, 13 and 14. 
 
 Since the plate is carried through a circular arc, and the 
 capillary coincides with a line passing through the centre of 
 motion, the curve given by the excursion is most naturally 
 expressed in polar co-ordinates. In Fig. 10, o is the pole and 
 
( 31 ) 
 
 o r the radius vector, the plate A B being supposed to move 
 from right to left. The position of a point p upon the curve is 
 given by the length o p and the angle q o p = 0, through which 
 the plate has moved since the arbitrarily selected initial point 
 q passed the fixed line o r. The dotted line c c is called the 
 reference circle. It corresponds to the upper limit of the slit, 
 and being sharply defined on the negative is used to measure 
 from. The dotted line z z is the circle of zero-potential on 
 which the point g is situated. 
 
 At the point p the excursion has already commenced. It is 
 required to find the E.M.F. acting on the electrometer at that 
 instant, and two operations are necessary. 
 
 (a) Measure the distance through which the meniscus has 
 already risen. This is done by turning the plate about the 
 centre o till the initial point q comes under the line o r. Let 
 the distance between q and the reference circle be c q milli- 
 metres. Measure cp in the same way. Then 
 
 c q - c p = A r, 
 
 where A r is the distance through which the meniscus has 
 already risen in millimetres. 
 
 But each millimetre of rise or fall of the meniscus corre- 
 sponds to a certain P.D. easily ascertained once for all by 
 calibrating the electrometer. Let k be the factor which turns 
 scale-readings into volts. Then the portion of the E.M.F. 
 indicated by the rise that has taken place, 
 
 v = k A r volt. 
 
 (b) The remainder of the E.M.F. is indicated by the velocity 
 with which the meniscus was moving at the point p. As the 
 angular velocity of the plate is constant, the time t is directly 
 proportional to 6, and the velocity of the movement of the 
 meniscus is the differential d r/d 6. But this, by a well- 
 known property of polar curves, can be easily ascertained. 
 Draw p t tangent to the curve at the point p, and drawjo??, 
 the normal, at right angles to the line p t. Through the pole 
 o draw a line at right angles to o r, cutting the normal at n> 
 The line o n is called the polar subnormal, and 
 
 In other words the length of the subnormal N is directly pro- 
 
( 32 ) 
 
 portional to that portion of the E.M.F. indicated by the 
 velocity with which the meniscus is moving at the point p. 
 And if / be the factor which turns subnormal scale-readings 
 into volts, we have for the total E.M.F. at the point p 
 
 V-ftAr+IN. 
 
 It is necessary to observe the signs of both terms of this 
 expression. At the apex of the curve, where the tangent is 
 horizontal, the sub-normal is reduced to zero, and then only 
 is the total E.M.F. directly proportional to the extent of the 
 excursion. But on the descending portion of the curve the 
 sub-normal passes to the left side of the pole, and the formula 
 becomes 
 
 V-fcAr-lN, 
 
 and before the meniscus has descended imm. the opposite 
 P.D. indicated by its downward velocity may be greatly in 
 excess of that corresponding to its position. Should this state 
 of the circuit continue the meniscus would descend below the 
 zero line, when the formula would be V = - K A r - I N. 
 
 It remains to describe in what way the factor I, which turns 
 the scale readings of the sub-normal into volts, is ascertained. 
 A " normal excursion," i.e., that given by the action of a 
 constant P.D. of known value, is photographed. The method 
 of doing this has been described. The resistance must be 
 approximately the same as that in circuit when the curves to 
 be analysed were taken. But as 1,000 ohms more or less 
 makes very little difference, one photograph will serve for a 
 large number of experiments. The curve of such an excursion 
 is shown in Fig 11. 
 
 The most accurate measurements of the sub-normal are 
 obtained when it makes an angle of about 45deg. with the 
 radius vector o r. It is also easier to find the tangent to the 
 curve at a point some little distance from its commencement. 
 Selecting a convenient point, I measure its vertical distance 
 from the zero line. Deducting the known value of this from 
 the E.M.F. given by the potentiometer, the remainder is the 
 E.M.F. corresponding to the velocity of the meniscus at that 
 point, and therefore also to the length of the polar sub- 
 normal N, V-feAr 
 
 whence N 
 
 * A r is measured in millimetres N in centimetres. 
 
( 38 ) 
 
 In the electrometer at present in use 15cm. on the sub- 
 normal =0-02 volt, so that Z = 0-13. And an excursion of 
 10mm. is given by 0-013 volt, so that & = 0-0013. 
 
 The measuring machine shown in plan and elevation in 
 Figs. 12 and 13 enables these operations to be rapidly and 
 accurately performed. The negative is clamped to a carrier 
 B, pivoted at upon the high table A, in the top of which a 
 hole is cut so as to allow the photograph to be examined by 
 transmitted light. Two screws, C and D, serve to adjust it to 
 a position corresponding exactly to that which it occupied on 
 the pendulum machine. The staff E is two metres long, and 
 corresponds to the fixed line ton of the diagram (Fig. 10). It 
 is graduated in centimetres, and the values of the subnormal 
 
 FIG. 11. A Normal Excursion of the Capillary Electrometer given by 
 communicating to it a known constant P.D. 
 
 are read off upon it. F is a brass plate, projecting sufficiently 
 high above the front edge of the table to support a fine thread, 
 attached to the pivot 0, and stretched by a weight, just clear 
 of the surface of the negative. This thread corresponds to the 
 radius vector o r of Fig. 10. But it is only used in adjusting 
 the apparatus. The line which actually represents the radius 
 vector is the image of a very fine copper wire G, stretched 
 across an adjustable frame about a foot from the floor. There 
 is a mirror underneath the wire G, and the lens H an ordinary 
 lantern objective serves to focus the wire upon the negative. 
 In using the machine, the thread is laid in a notch in F, 
 exactly at right angles to the centre of the staff E. The wire 
 
 D 
 
FIG. 12. The Measuring Apparatus. Plan. 
 
 E nn 
 
 B 
 
 FIG. 13. The Measuring Apparatus. Elevation. 
 
( 35 ) 
 
 G is set so that its image on the plate, as seen through a lens, 
 coincides with the thread, which is then removed. The plate 
 has now to be centred. This is done by adjusting the screws 
 C and D, till, on traversing the plate carrier from side to side, 
 the reference circle on the negative runs truly against the 
 image of a pin fixed by the side of the wire G. 
 
 To find the zeros of time and of E.M.F. that part of the 
 record of the signal key at which it drops suddenly is brought up 
 to the radius vector, the image of G, and a mark is made with a 
 needle point where it cuts the serrated record of the tuning fork. 
 The distance c q along the radius vector between the reference 
 circle and the electrometer curve is measured with a glass 
 millimetre scale. It is entered in the note book under the 
 
 M 
 
 FIG. 14. Plan and Elevation of Mirror-Block. 
 
 column K, against the entry t Q = Q of the first column. The 
 carrier is then shifted until the earliest point at which any 
 change in the level of the meniscus can be detected comes 
 under the radius vector. The time ^ at which this took place 
 is noted. From this point the analysis of the curve com- 
 mences. The carrier is moved on through a suitable time 
 interval, usually less than OOOlsec., so as to make the next 
 value of t a whole number. The distance cp is entered in the 
 column R - r, and the difference cq cp in a third column A r. 
 Then the subnormal is measured by the following method. 
 
 The block L, shown in plan and elevation in Fig. 14, has 
 upon its under surface a glass plate, k, on which are ruled two 
 lines at right angles to each other. One of these, the tangent, 
 is continuous, but the other, the normal, is broken where it 
 
( 36 ) 
 
 crosses the tangent. On the side of the block, set exactly 
 parallel with the tangent line, is the mirror M, and on the 
 mirror, in a line with the end of the normal, is a pointed 
 index. To measure a subnormal, the block L is placed upon 
 the negative and adjusted so that the three lines, tangent, 
 normal and radius vector, may intersect at the point where the 
 radius vector line crosses and the tangent touches the curve. 
 Then, passing to the back of the staff E, the observer moves the 
 slider J upon it until the pfn of the slider and its image in the 
 mirror M coincide simultaneously with the point of the index 
 on the mirror. In this way the length of the subnormal can 
 be measured within about 1 per cent. The values so obtained 
 are entered under the column N. 
 
 The process is repeated for a number of points on the 
 curve. Where the changes of P.D. are very sudden the time 
 intervals may be less than 0-OOlsec., and any changes of 
 direction separately noted. Finally, the results are calculated 
 out by the formula already given, and plotted in graphic 
 form. All measurements are made with a lens magnifying 
 10 or 12 diameters.* 
 
 The relative importance of the two components of the 
 indicated E.M.F. varies according to the character of the 
 experiment. Slow changes, //., from to 0-1 volt in a tenth 
 of a second, are shown mainly by the change of level of the 
 meniscus, but the same P.D.. established in the course of the 
 hundredth part of a second would produce a comparatively 
 small rise with large values of the subnormal. That the 
 method is reliable was proved by applying it to the estimation 
 of known P.D.'s, which were caused to act for 0-Olsec. on 
 the electrometer. 
 
 It must be borne in mind that the value in volts of the 
 subnormal scale-readings depends on the velocity of the plate 
 during the observation. It is, therefore, essential that the 
 rate of the pendulum should be the same as when the normal 
 excursion was taken from which this value was calculated. 
 The velocity of the plate, with a given load on the pendulum, 
 is so nearly constant that the error resulting from it may be 
 neglected ; but for different loads, or for greater accuracy, the 
 
 * Detailed examples of the method of analysis will be given at the 
 conclusion of the series. 
 
( 37 ) 
 
 values of the subnormal may be either calculated or ascer- 
 tained by measuring a specially photographed normal excursion. 
 
 Each photographic record represents the integration of the 
 charges received by the electrometer during its production. 
 And as the electrometer throughout that portion of its range 
 is generally of constant capacity, the curve represents also the 
 integration of the quantities of current which have flowed 
 through the circuit, if there have been no changes of resistance- 
 The derived curve, obtained by this method of analysis, shows 
 the changes of E.M.F. also on the supposition that there have 
 been no material changes of resistance. But as the equivalent 
 resistance of the electrometer circuit is seldom under 100,000 
 ohms, such variations must be considerable to affect the 
 result. 
 
 It may be asked, How does this method distinguish between 
 the natural return of the meniscus to its zero position by 
 discharging itself through the experimental circuit, and the 
 reversal of any given movement by a difference of potential in 
 the opposite direction ? In such a case, if the E.M.P. indi- 
 cated by the subnormals is exactly equal and opposite in sign 
 to that corresponding to the distance of the meniscus from 
 the position of zero potential, the electrometer is simply dis- 
 charging through the circuit, and the E.M.F. calculated by 
 the formula is zero. If there is an excess in either direction, 
 that excess corresponds to the E.M.F. acting at the time. 
 The curve of discharge and the curve of charge have precisely 
 the same time-relations. 
 
 I have already explained the method of finding the value of 
 the subnormal when the resistance in circuit is not the same 
 as that used in photographing the normal excursion. 
 
 Application of the Capillary Electrometer to tlw Study of 
 Telephone Currents and Dynamos. 
 
 In the spring of 1887 I photographed the result of blowing 
 a whistle near a telephone connected with the electrometer. 
 It was found that the currents so produced were of sufficient 
 intensity to electrolyse the acid in the capillary, and the 
 shadows of the gas bubbles were very evident upon the plate. 
 Electrolysis is also brought about by a single induction shock 
 whenever the resulting E.M.F. exceeds a certain limit, and in 
 
( 38 ) 
 
 such cases, of which Fig. 15, photographed June 21, 1892, is 
 an example, no measurable interval can be detected between 
 the establishment of the P.D. and the evolution of the gas. I 
 am investigating the bearing of this fact on Kohlrausch's 
 method of measuring resistance. It had been long known 
 that a good electrometer would respond visibly to certain 
 sounds spoken into the telephone, but no attempt had been 
 made to photograph these excursions. 
 
 In the summer of 1893 I made one of the assistants sing a 
 high falsetto note at a distance of about two feet from a tele- 
 phone in circuit with the electrometer. The result is shown 
 in Fig. 16, which was the last of several photographs taken 
 in succession. His voice was beginning to be fatigued, and 
 the sound wavered. It will be noticed that there is a periodic 
 change in the inclination of the serrations.* This is absent 
 in the earlier negatives taken when the voice was steady. The 
 pitch of the note, as judged by comparing it with the sound 
 of the tuning fork, corresponded exactly with the number of 
 vibrations recorded, viz., 650 double changes of E.M.F. per 
 second. Had the photograph showed simply a serrated band, 
 it might have been ascribed to a mere vibration set up in the 
 meniscus under the action of an alternating current ; but this 
 evidence that the electrometer could record changes in the 
 quality of a tone determined me to make further experiments 
 on the subject. 
 
 No opportunity offered till the spring of 1894. I then 
 found that an ordinary telephone did not give currents of 
 sufficient intensity to record more than the pitch of the voice, 
 unless the mouth was placed quite close to the diaphragm, 
 and in that case the disturbing influence of air currents made 
 the results irregular. I therefore tried the microphone, using 
 a contact between two carbon buttons resting on a diaphragm 
 of thin vulcanite. The microphone circuit was connected 
 with the primary of an induction coil, the secondary of which 
 
 * These details were too small for reproduction by the method used in 
 Fig. 15. An enlarged image of the curve was thrown on a sheet of Bristol- 
 board, carefully outlined, and reduced by photography to 1 times the 
 size of the original. I have measured the E.M.F. of these serrations, and 
 found it to vary between O'Ol and 0'02 volt. In another experiment of 
 the same kind the maximum E.M.F. was 06 volt. This is much greater 
 than is commonly supposed in the case of telephone currents. 
 
was in circuit with the electrometer. This arrangement was 
 so very sensitive that I could do nothing with it. I succeeded 
 once in photographing the sound of the letter "r"; but in 
 most cases there was an irregular vibration running through 
 the records. The cause was discovered by placing a telephone 
 in the circuit, when the sound of the engine and dynamo, 
 which supplied the arc lamp, was plainly heard. The 
 apparatus was fixed to the wall of the room, and this trans- 
 mitted the sound. Another reason induced me to abandon 
 the microphone. The slightest change of the adjustment 
 would completely alter the clang-tint of a sound reproduced 
 
 FIG. 15. Effect of a Single Induction Shock strong enough to cause 
 electrolysis. The upper serrated line is produced by a tuning fork making 
 500 double vibrations per second. The narrow band beneath it is the 
 gnomon of the signal key, the fall of which breaks the primary circuit of 
 the induction coil. The curve at the lower part of the plate is the 
 shadow of the meniscus, and the faint streaks above it are produced by 
 bubbles of hydrogen. This photograph shows that the interval between 
 the fall of the signal and the evolution of gas is too small to be measured 
 even with a plate moving 80cm. per second. 
 
 by it. For instance, the ticking of a watch placed near the 
 microphone could be made to resemble that of a clock by 
 merely adjusting the weight upon the carbons. And although 
 the sound was always recognisable and even characteristic, it 
 was impossible to doubt that in one case the deeper vibrations 
 were reinforced and the upper tones suppressed, while in the 
 
- 1 fco 
 o S 
 
 LO ' 
 vO 0) 
 
 i 
 
 v o 
 
 8 8 
 
 other case the lower tones were absent. 
 Then I tried a similar experiment with 
 a number of telephones, and found that 
 each one reinforced certain notes more 
 powerfully than others, and that the 
 effect depended on the shape and size of 
 the diaphragm, and was constant for 
 each instrument. This accounts for the 
 fact that electrolysis is produced in the 
 capillary of an electrometer when certain 
 notes are sounded near a telephone con- 
 nected with it, while other notes, although 
 much louder, produce no such effect. I 
 decided, therefore, in order to obtain more 
 faithful records of the vibrations of the 
 voice in speech, to join up all my tele- 
 phones in series. This plan had also 
 another advantage, namely, that of 
 increasing the E.M.F. of the currents. 
 For the capillary electrometer differs 
 from the high-resistance galvanometer in 
 requiring much less current but a higher 
 E.M.F. to give good readings. I used 
 five telephones of three several types- 
 each one, as it chanced, sensitive to a 
 different fundamental tone. These were 
 placed close together in a convenient 
 position for receiving the sound of the 
 voice. To shield the electrometer from 
 the effect of accidental noises two 
 mercury keys were attached to the 
 pendulum machine, one to short-circuit 
 the electrometer until the plate had 
 begun to move, and the other to 
 perform a similar function directly it 
 had passed the slit. 
 
 Vowels and sibillants are easy to 
 photograph, because they can be sounded 
 continuously from the moment the 
 handle is raised until the click of the 
 
pendulum against the catch warns the operator that the 
 plate has passed the slit. But consonants are more difficult 
 to catch. I was obliged to use a slower rate, and to 
 repeat the sounds as fast as possible. Thus the charac- 
 teristic vibrations of the letter "p" are given by the 
 syllables " Pop-op-op-op," and those of "d" and "k" by 
 " Dod-od-od " and "Kok-ok-ok" respectively. It will be 
 observed that the vowel is the same in all three cases. This 
 was done purposely for the sake of uniformity, although the 
 plate did not move fast enough to bring out the vibrations due 
 to the vowel. The accompanying illustrations, Fig. 17 and 
 Fig. 18, show a marked difference between the labial and 
 
 FIG. 17. The Syllables "pop-op-op" Spoken near a Battery of Five 
 Telephones in series. Plate moving more slowly. (Enlarged).* 
 
 FIG. 18. The Syllables " dod-od-od " spoken in the same way. (Enlarged).* 
 
 dental consonants. The guttural sound gave a curve different 
 from either, but unfortunately the two examples I obtained 
 were not well adapted for reproduction. The vowel "e" 
 in " me " was easily distinguishable from the short sound 
 of "o" used in the experiments on the consonants, and 
 
 * Figs. 17 and 13 are simplified drawings, more or less diagrammatic, of 
 the original negatives, showing only the principal excursions and the 
 general shape of the curves. The peculiar flat-topped groups in Fig. 17 
 are characteristic of the labials, and the more or less lenticular series of fine 
 oscillations in Fig. 18 always result from the dental checks. The intervals 
 between the fine lines correspond to the pitch of the voice. The excursions 
 extend a good deal farther both upwards and downwards than is here 
 represented, and many of them have a hazy outline, indicating that they 
 are in reality groups of excursions, which on a more rapidly moving plate 
 would be separately visible. 
 
it is here reproduced (Fig. 19). It will be observed 
 that the plate was driven much faster than in the two 
 previous cases, and that the negative has not been enlarged 
 as those were. While it was still wet, fine serrations 
 could be distinguished with a lens, and, I have no doubt, that 
 by using a still higher velocity these could be rendered visible 
 
 FIG. 19. The vowel " e " as iu " mo." 
 
 FIG. 20. The Letter " r " Trilled with the Tongue, about 2ft. from a 
 Microphone. Plate moving very rapidly, shows about two-thirds of the 
 complete cycle of vibration, as ascertained from other photographs with 
 slowly-moving plates. 
 
 to the eye. The letter "r," strongly trilled, was taken on the 
 slow rate with the battery of telephones, and on the fast rate 
 with the microphone, the result of the latter experiment being 
 given in Fig. 20. The tuning fork was not sounded in this 
 
( 43 ) 
 
 case, as it would have affected the microphone, but the 
 velocity of the plate was between two and three metres per 
 second. By comparing it with the telephone records, it was 
 evident that the microphone photograph represents about two- 
 thirds of a very complex series of vibrations, the repetition of 
 which constitutes the characteristic sound of the trilled 
 lingual "r." 
 
 1 propose with a special apparatus to continue the in- 
 vestigation. Meanwhile, these experiments bid fair to lead to 
 results of practical as well as scientific importance. Now that 
 the familiar use of the telephone inclines us to regard its 
 shortcomings in a more critical spirit, few persons would deny 
 that it caricatures rather than reproduces the tones of the 
 voice. The reason is not far to seek. We are dealing with 
 an instrument capable of responding to a large number, but 
 not all, of the vibrations concerned in speech-sounds. More- 
 over it reproduces them, not with the same, but with 
 different relative intensity. As in the Jew's harp we reinforce 
 by resonance certain harmonics of a fundamental tone, so in 
 the telephone and also in the phonograph we bring out the 
 harmonics of the diaphragm. These experiments suggest the 
 possibility of improving the telephone by the use of several 
 diaphragms of different dimensions. They have an interest 
 also to the student of phonetics. The curves obtained by 
 my method differ from those of Prof. McKendrick and Prof. 
 Hermann in the relative amplitude of the slow and rapid 
 vibrations. In the present stage of the investigation it is not 
 possible to say which is nearer the truth, but, unquestionably, 
 in all such researches it will be necessary to take into account 
 the characteristics of the recording apparatus to a greater 
 degree than has been done hitherto. 
 
 But the main interest to myself is the proof they afford of 
 the extraordinary rapidity of the response of the capillary elec- 
 trometer to alternating currents. Evidently the limit of its 
 capabilities has not yet been reached, and these results, even 
 though they should have no further application, imply that 
 confidence may be placed in the photographic records of less 
 rapid changes. 
 
 The success of the experiments with the telephone induced 
 me to photograph some dynamo currents. The facilities for 
 
such work are not great in a laboratory devoted to physiology, 
 but I was able to do sufficient to convince myself of the possi- 
 bility of using the capillary electrometer for this purpose. 
 
 In Fig. 21 are shown the curves given by a small inductor- 
 alternator with a frequency of 400 per second. Such curves 
 are, of course, too small to be analysed individually, but the 
 maximum E.M.F. of each could be easily measured. But the 
 electrometer would stand currents a good deal stronger ; and 
 even without putting any greater pressure upon it, by increas- 
 ing the velocity of the plate, which in this particular instance 
 was only 60cm. per second, a curve could easily be produced 
 on which eight or ten determinations of P.D. could be made 
 for each period, as in the case of the photograph of micro- 
 phone currents reproduced in Fig. 20. 
 
 FIG. 21. Curves given by a small Inductor Alternator making 400 
 complete periods per second. 
 
 The next example, Fig. 22, is taken from a very poor 
 photograph, but an exceedingly interesting one. It was taken 
 after altering the adjustments of the inductor-alternator so as 
 to throw it a little out of truth. The change was very slight, 
 but it is recorded with unmistakable plainness upon the plate, 
 a periodic rise and fall of potential, corresponding to a com- 
 plete revolution of the armature-shaft, being superposed upon 
 the more rapid alternations due to the passage of the pole- 
 pieces past the magnets. The importance of this result can 
 hardly be over-estimated. It confirms the evidence afforded 
 by the experiments with the telephone, particularly that in 
 which the wavering of the voice when fatigued was recorded 
 
( 45 
 
 OQ 
 
 O 
 
 01 
 
 o 
 
 o - 
 
 ^ s 
 
 n 
 ec 
 
 OF THE 
 
 UNIVERS 
 CALIFOP 
 
( 46 ) 
 
 (Fig. 16). But whereas in that case the degree of correspon- 
 dence between the variation of the sound and of the photo- 
 graphed curve could only be judged by memory, in this 
 photograph of alternating currents it can be tested by observ- 
 ing that the number of serrations in each period corresponds 
 precisely with the number of field magnets in the cycle.* I 
 am continuing my experiments in this direction, and will for 
 the present confine myself to the expression of my conviction 
 that this use of the capillary electrometer will be of great 
 service in dynamo investigations. 
 
 Fig. 23 shows the current curves of the same inductor- 
 alternator driven at a slower speed, and giving about 30 
 
 FIG. 23. Inductor Alternator running slowly and making 30 periods 
 per second. 
 
 double alternations per second. Were it not for the imper- 
 fection of the photograph, due to the spitting of the arc lamp, 
 which was a very poor one, these curves could easily be 
 measured and the E.M.F. determined for each two-thousandth 
 part of a second. The electrometer was kept on short circuit 
 until after the plate had reached the slit, and the interval 
 between the fall of the signal and the commencement of the 
 curve is due to the delay of the auxiliary magnet, by which 
 
 * Fig. 16 illustrates the interference of two vibrations of nearly equal 
 period, but Fig. 22 is the result of compounding two cycles of P.D., one 
 slow and the other rapid. 
 
( 47 ) 
 
 the key was opened and the current thrown in. It should be 
 noticed that the average level of the mercury in tJie capillary is 
 not altered by the action of an alternating current not too 
 powerful for the instrument to bear with safety. 
 
 This is not the case with the next photograph (Fig. 24), 
 taken with an ordinary lecture-table dynamo fitted with a 
 Siemens shuttle armature, and driven by a water motor. 
 The current was sent through a non-inductive resistance, 
 and a derivation led to the electrometer, which was kept on 
 short circuit until the plate had reached the slit. The result 
 shows the effect of suddenly throwing in a pulsating direct 
 
 FIG 24. Curves given by a Shunt wound Dynamo with Siemens Shuttle 
 Armature, making about 10 revolutions per second. 
 
 current. The curves look very much like those of the alter- 
 nator, but a closer inspection shows that the descents are not 
 so steep as the ascents for the first few periods, and that 
 consequently the meniscus mounts. 
 
 In Fig. 25, in which the plate was driven more slowly, this 
 is shown better. The mercury column rises steadily through 
 a distance depending on the average P.D. between the ends of 
 the derived circuit, and continues oscillating about this new 
 level. At first sight this might seem to indicate that the 
 E.M.F. never falls to zero, but on analysing the curves by 
 
( 48 ) 
 
 the method I have described it is found that at a certain point 
 on the descent of each the downward velocity of the meniscus 
 indicates an E.M.F. exactly equal and opposite to the P.D. 
 corresponding to the height of the mercury above the zero 
 line, showing that the current actually falls to zero twice 
 during each revolution of the armature. 
 
 It remains to describe briefly a method of observation 
 which I had in view, but which Prof. Wedensky had quite 
 independently discovered and perfected before I put my own 
 idea into execution. 
 
 It is not always convenient, or, indeed, possible to use 
 the photographic method. Prof. Wedensky has supplied us 
 
 FIG. 25. Curves given by a Shunt- wound Dynamo with Siemens Shuttle 
 Armature, making about 10 revolutions per second, taken on a slowly 
 moving plate. Note the rise of the meniscus, which does not occur when 
 an alternating current is used. 
 
 with a means of observing the phases of a rapidly recurring 
 cycle of electromotive changes without recording them. His 
 apparatus, which was designed for physiological work, is 
 equally applicable to the study of dynamo currents. It is an 
 adaptation of the familiar principle of the zoetrope. The 
 microscope used for observing the capillary electrometer is 
 illuminated by a series of flashes, nearly, but not quite 
 synchronous with the cycle of electromotive changes under 
 observation. Thus, although the meniscus is in rapid, 
 
( 49 ) 
 
 motion, the observer sees it only during a definite small period 
 of each excursion. The stroboscopic disc, by which the illu- 
 mination is interrupted, may be either driven at a slightly 
 slower speed, or fitted with an arrangement by which it can 
 be given a definite amount of lag, variable at will. In this 
 way, by a series of separate observations, the electrometer 
 curve for the entire cycle can be plotted down. But it must 
 not be forgotten that in this case, as with the photographic 
 records, the distance through which the mercury is observed 
 to have moved above or below its zero-point is not pro- 
 portional to the actual P.D. between the terminals at the 
 instant of each flash. To ascertain the true variation of 
 P.D. with Prof. Wedensky's apparatus, it wou]d be necessary 
 to plot a curve from a large number of observations, and then 
 analyse it in the manner described in this Paper. And the 
 analysis of a plotted curve is neither easy nor satisfactory. 
 Still, I have no doubt that the stroboscopic method will prove 
 of service in cases where photography may not be practicable. 
 In conclusion, I would recommend the capillary electro- 
 meter to the notice of electricians for two reasons. It is the 
 cheapest form of electrometer that can be made. When it has 
 received the attention it merits, I have no doubt it will become 
 as familiar an instrument as the thermometer. The construc- 
 tion is so simple that even if it should be broken a new one 
 can be made, unless the operator is very unfortunate, in less 
 than an hour. It is essentially the instrument for those who 
 have to think twice before spending a shilling. But it has quali- 
 ties of its own distinct from those of any other form of electro- 
 meter or galvanometer, and these will make it a valuable addi- 
 tion to the most perfectly appointed laboratory. I have 
 described the method by w r hich these special properties may be 
 utilised, and some of the apparatus I have designed for the 
 purpose during the nine years I have worked with the 
 instrument, in order if possible to bring it into general use. 
 For the same reason I have refrained from patenting any part 
 of it, preferring to present my discovery and invention to the 
 scientific public. The same method may be applied not only to 
 the capillary electrometer, but to any form of electrometer or 
 galvanometer, or thermometer, or other measuring instrument, 
 the movement of which is either dead beat or governed by a 
 
( 50 ) 
 
 known law. The motion of the recording surface may be 
 rectilinear or circular. The surface itself may be flat or 
 developable. It may be propelled with uniform velocity or with 
 uniform acceleration by any of the devices used for such 
 purposes and known to scientists. The motion of the plate 
 may be continuous, and the exposure may be given by open- 
 ing and closing the shutter during the passage of the whole 
 or a part of the sensitive surface. Several methods are avail- 
 able for determining with the requisite accuracy the exact 
 moment of the experiment. For slow speeds diffused daylight 
 or a lamp will serve, while for higher velocities the limelight, 
 or magnesium, or the electric arc or sunlight must be 
 employed; in short, the methods and apparatus by which 
 labour is lessened and improved results obtained in other 
 branches of photography are applicable to this. From the 
 records so obtained the derived curve may be calculated or 
 drawn either in the way I have described, or by means of 
 any known device for differentiating curves, and for certain 
 purposes the same process may be repeated on the derived 
 curve itself. The apparatus has already done good service in 
 physiology, and it now remains to extend its application to 
 other departments of science. 
 
 Note added July 2, 1896. 
 
 The following detailed analyses of the photographed curves 
 of two dynamos, one giving an alternating and the other a 
 pulsating current, will illustrate both the practice of the 
 method and its applicability to the study of motors and 
 generators. By far the greater number of curves I have 
 measured relate to the physiology of muscle and nerve. Most 
 of them are as yet unpublished, but several are discussed by 
 Prof. Burdon Sanderson in the Journal of Physiology, Vol. 
 XVIIL, pp. 117-159. 
 
 The portion given includes a complete period. The photo- 
 graph is similar in appearance to Fig. 23, which was taken on 
 the same day, viz., August 15, 1895. Owing to the spitting 
 of the arc lamp; the negative is ribbed, making accurate 
 measurement impossible. In this, as in the next example, 
 A K is measured from the position of zero-potential. With 
 the alternating current A R changes sign, as does also N, 
 although not simultaneously. 
 
No. 52, B. Analysis of a Portion of the Curve given by the Inductor 
 Alternator, while making 21f periods per second, with the arma- 
 ture-shaft set as in Fig. 22 S9 as not to run true. The values of 
 N + & A R in the last column are directly proportional to the 
 EM.F. 
 
 T. 
 
 AR. 
 
 &AR. 
 
 jr. 
 
 N + &AR. 
 
 
 
 
 
 
 
 
 
 
 
 68 
 
 +10-0 
 
 + 2-0 
 
 - i-o 
 
 + i-o 
 
 70 
 
 + 8-5 
 
 + 1-7 
 
 -12-0 . 
 
 -10-3 
 
 72 
 
 + 7-5 
 
 + 1-5 
 
 -18-0 
 
 -16-5 
 
 74 
 
 + 5-5 
 
 + ri 
 
 -19-8 
 
 -18-7 
 
 76 
 
 + 3-5 
 
 + 0-7 
 
 -19-2 
 
 -18-5 
 
 78 
 
 + 1-5 
 
 + 0-3 
 
 -18-7 
 
 -184 
 
 80 
 
 - 0-5 
 
 -o-i 
 
 -17-5 
 
 - 17-6 
 
 82 
 
 - 2-0 
 
 -0-4 
 
 -15-5 
 
 -15-9 
 
 84 
 
 - 3-0 
 
 -0-6 
 
 -11-5 
 
 -121 
 
 86 
 
 - 3-8 
 
 -0-8 
 
 - 8-5 
 
 - 9'3 
 
 88 
 
 - 4-5 
 
 -0-9 
 
 - 3-0 
 
 - 3-9 
 
 90 
 
 - 4-5 
 
 -0-9 
 
 + 2-0 
 
 + I'l 
 
 92 
 
 - 3-8 
 
 -0-8 
 
 + 15-0 
 
 + 14-2 
 
 94 
 
 - 2-5 
 
 -0-5 
 
 + 15-4 
 
 + 14-9 
 
 96 
 
 ro 
 
 -0-2 
 
 + 16-0 
 
 + 15-8 
 
 98 
 
 + 0-3 
 
 +0-1 
 
 + 16-0 
 
 + 161 
 
 100 
 
 + 1-5 
 
 + 0-3 
 
 + 16'5 
 
 + 16-8 
 
 102 
 
 + 3-5 
 
 + 0-7 
 
 + 16-0 
 
 + 16-7 
 
 104 
 
 + 4-8 
 
 + 1-0 
 
 + 15-5 
 
 + 16-5 
 
 106 
 
 + 6-0 
 
 + 1-2 
 
 + 14-5 
 
 + 15-7 
 
 108 
 
 + 7-5 
 
 + 1-5 
 
 + 7-5 
 
 + 9-0 
 
 110 
 
 + 8-0 
 
 + 1-6 
 
 + 2-0 
 
 + 3-6 
 
 112 
 
 + 7-5 
 
 + T5 
 
 -12-0 
 
 -10-5 
 
 The curve of which the analysis is given on page 52, was 
 a duplicate of that reproduced in Fig. 24, and was photo- 
 graphed on August 15, 1895. The electrometer employed for 
 these dynamo experiments was not the same as that referred to 
 in the earlier part of the Paper, so that different constants are 
 used in the calculation. The nominal E.M.F. of the dynamo 
 was about 4 volts. The current was sent through a non- 
 inductive resistance, from which a derivation led to the 
 electrometer. As the exact value of this derivation was not ' 
 ascertained, I have not reduced the subnormal readings to 
 volts, but roughly 1cm. on the subnormal =0'25 volt at the 
 brushes. 
 
 In both cases the first column shows the time, T, in 
 thousandths of a second, from an arbitrary zero. 
 
No. 54, B. Analysis of Curves given by a Shunt-wound Dynamo 
 with Siemens Shuttle Armature, making about 700 revolutions 
 per minute, and generating a pulsating current. 
 
 T. 
 
 AR. 
 
 &AR. 
 
 N. 
 
 N + &AR. 
 
 
 
 35'0 
 
 7-00 
 
 
 
 + 7-00 
 
 2 
 
 35-0 
 
 7-00 
 
 + 2-3 
 
 + 9-30 
 
 4 
 
 35-4 
 
 7-08 
 
 + 3-7 
 
 + 10-78 
 
 6 
 
 35-9 
 
 718 
 
 + 6-7 
 
 + 13-88 
 
 8 
 
 36-8 
 
 7-36 
 
 + 10-5 
 
 + 17-86 
 
 10 
 
 37-8 
 
 7-56 
 
 + 14-0 
 
 + 21-56 
 
 12 
 
 39-3 
 
 7-86 
 
 + 14-7 
 
 + 22-56 
 
 14 
 
 40-6 
 
 8-12 
 
 + 13-5 
 
 + 21-62 
 
 16 
 
 41-4 
 
 8-28 
 
 + 9-5 
 
 + 17-78 
 
 18 
 
 42-3 
 
 8-46 
 
 + 7-0 
 
 + 15-46 
 
 20 
 
 42-8 
 
 8-56 
 
 + 4-7 
 
 + 13-25 
 
 22 
 
 43-2 
 
 8-64 
 
 + 2-5 
 
 + 11-14 
 
 24 
 
 43-4 
 
 8-68 
 
 o-o 
 
 + 8-68 
 
 26 
 
 42-9 
 
 8-58 
 
 - 2-5 
 
 + 6-08 
 
 28 
 
 42-3 
 
 8-46 
 
 - 7-0 
 
 + 1-46 
 
 30 
 
 41-0 
 
 8-20 
 
 - 8-2 
 
 + o-o 
 
 32 
 
 40-3 
 
 8-06 
 
 - 8-0 
 
 + 0-06 
 
 34 
 
 39-9 
 
 7-98 
 
 - 7-5 
 
 + 0-48 
 
 36 
 
 38-9 
 
 7-78 
 
 - 6-6 
 
 + 1-18 
 
 38 
 
 38-4 
 
 7-68 
 
 - 5-0 
 
 + 2-68 
 
 40 
 
 38-0 
 
 7-60 
 
 - 4-0 
 
 + 3-60 
 
 4'<4 
 
 37-9 
 
 7-58 
 
 - 2-0 
 
 + 5-58 
 
 44 
 
 37-9 
 
 7-58 
 
 o-o 
 
 + 7-58 
 
 46 
 
 38-0 
 
 7-60 
 
 + 3-0 
 
 + 10-60 
 
 48 
 
 38-5 
 
 7-70 
 
 + 5-6 
 
 + 13-30 
 
 50 
 
 39-2 
 
 7-84 
 
 + 12-0 
 
 + 19-84 
 
 52 
 
 40-4 
 
 803 
 
 + J5-4 
 
 + 23-48 
 
 54 
 
 41-9 
 
 8-38 
 
 + 16-5 +24-88 
 
 56 
 
 43-5 
 
 8-70 
 
 + 12-6 +21-30 
 
 58 
 
 43-9 
 
 8-78 
 
 + 8-5 
 
 + 17-28 
 
 60 
 
 44-6 
 
 8-92 
 
 + 3-5 
 
 + 12-42 
 
 62 
 
 45-0 
 
 9-00 
 
 o-o 
 
 + 9-00 
 
 64 
 
 44-9 
 
 8'98 
 
 - 1-5 
 
 + 7-48 
 
 66 
 
 44-7 
 
 8-94 
 
 - 2-7 
 
 + 6-24 
 
 68 
 
 44-4 
 
 8-88 
 
 - 4-0 
 
 + 4-88 
 
 70 
 
 43-9 
 
 8-78 
 
 - 6-3 
 
 + 2-48 
 
 72 
 
 43-3 
 
 8-66 
 
 - 8-0 
 
 + 0-66 
 
 74 
 
 42-5 
 
 8-50 
 
 , - 8-5 
 
 + O'O 
 
 76 
 
 41-6 
 
 8-32 
 
 - 75 
 
 + 0-82 
 
 78 
 
 40-9 
 
 818 
 
 - 6-7 
 
 + 1-48 
 
 80 
 
 40-4 
 
 8-08 
 
 - 5-5 
 
 + 2-58 
 
 82 
 
 39-9 ' 
 
 7-98 
 
 - 4-5 ' + 3-48 
 
 84 
 
 39-4 
 
 7-88 
 
 - 2-2 + 5-68 
 
 86 
 
 391 
 
 7-82 
 
 - 0-2 + 7-62 
 
 88 
 
 39-4 
 
 7-88 
 
 + 30 +10-88 
 
 TS".B. Here AR (and k AR) remaius continually positive, but fluctuates 
 in value, while N not only fluctuates, but changes sign. 
 
( 58 ) 
 
 In the second column, A R, the distance of the meniscus 
 from the position of zero-potential, is given in units of a scale 
 of which 50 divisions =14- 1mm. on the photograph. This 
 scale, which was ruled on glass, was fixed close to the wire G 
 of the measuring apparatus (Fig. 13). The zero of its image 
 was made to coincide with the position of zero-potential on 
 the plate, and the value of A R was read off upon it as each 
 successive point on the curve was brought over the wire for 
 the determination of the subnormal. I find this method more 
 accurate and more convenient than that previously described. 
 
 The third column shows the value of A R in terms of the 
 subnormal. The constant multiplier (& = 0-2) is found by 
 taking the ratio of the difference between any two values of 
 A R and the difference between the corresponding values of N 
 upon the normal excursion (Fig. 11). This ratio is constant 
 if the electrometer is a good one. 
 
 The fourth column gives the subnormals N at the points 
 measured, reckoned as positive if the meniscus is rising and 
 negative if it is falling. 
 
 The last column is directly proportional to the E.M.F., and 
 shows how it falls to zero twice in each revolution of the 
 armature, although the photographed curve remains con- 
 tinually above zero. It is interesting to note in No. 54 B, the 
 way in which the negative value of the subnormal balances the 
 positive E.M.F. indicated by the position of the meniscus on 
 both occasions, and how the actual zero lags behind the zero 
 of the subnormal. On examining the machine I found that 
 the armature was not centrally situated between the field- 
 magnets, and that its poles were not symmetrical, so that 
 once in each revolution it came very near to one of the horns. 
 Something of this kind is indicated by the curve of the 
 E.M.F., which is more peak-shaped in one half than in the 
 other. On plotting this curve I find that the areas of the two 
 halves corresponding to the two faces of the armature coils 
 are in the proportion of 195 to 206. This divergence from 
 equality may be due to the fact that the brushes are not 
 exactly ISOdeg. apart, and the two halves of the commutator 
 are not quite equal. The smaller area belongs to the steeper 
 curve. 
 
 I am at present engaged upon a modification of this 
 method. The sensitive plate is mounted as a disc upon the 
 
( 54 ) 
 
 armature shaft, or a shaft revolving synchronously with it, 
 and the image of the capillary is projected upon it, as in my 
 pendulum machine, the whole being enclosed in a camera 
 with an automatic arrangement for exposing the plate at any 
 given time during a single complete revolution. I propose to 
 employ various well-known ways of connecting the whole or 
 part of the armature coils with the electrometer circuit, with 
 a view of obtaining polar diagrams of E.M.F. by analysing 
 the photographic records. 
 
 NOTE. If it is intended to analyse the curves, the develop- 
 ment should be stopped as soon as the high lights show on 
 the back of the film. The definition is then perfectly sharp 
 under a magnification of 10 diameters, but the negative is too 
 " thin " for reproduction. 
 
RETURN TO the circulation desk of any 
 University of California Library 
 
 or to the 
 
 NORTHERN REGIONAL LIBRARY FACILITY 
 
 Bldg. 400, Richmond Field Station 
 
 University of California 
 
 Richmond, CA 94804-4698 
 
 ALL BOOKS MAY BE RECALLED AFTER 7 DAYS 
 
 2-month loans may be renewed by calling 
 (510)642-6753 
 
 1-year loans may be recharged by bringing 
 books to NRLF 
 
 Renewals and recharges may be made 
 4 days prior to due date 
 
 DUE AS STAMPED BELOW 
 
 DD20 15M 4-02 
 
U. C. BERKELEY LIBRARIES