... *-.<-) '-^f^. .., --f _-" ;-*---- *->-( OF THE UNIVERSITY OF PRINCIPLES OF WIRELESS TELEGRAPHY Published by the McGraw-Hill Book. Company \Succ arsons to the Book Departments of the McGraw Publishing Company Hill Publishing- Company Publishers of Books for Electrical World The Engineering and Mining Journal The Engineering Record Rjwer and The Engineer Electric Railway Journal American Machinist PRINCIPLES OF WIBELESS TELEGRAPHY BY GEORGE W. PIERCE, A.M., PH.D. ASSISTANT PROFESSOR OF PHYSICS IN HARVARD UNIVERSITY Of THE I UNIVERSITY J Of [UFORNifc* McGRAW-HILL BOOK COMPANY 239 WEST 39TH STREET, NEW YORK 6 BOUVERIE STREET, LONDON, E.G. 1910 COPYRIGHT, 1910, BY THE McGRAW-HILL BOOK COMPANY Stanbopc iPms F. H. GILSON COMPANY BOSTON. U.S.A. PREFACE THIS volume comprises the non-mathematical portions of a course of lectures, entitled " Electric Waves and their Application to Wireless Telegraphy/' which for several years have been given by the author to classes at Harvard University. In giving the lectures and in preparing this volume, the design has been : First, to present, in as elementary a form as possible, the course of reasoning and experimentation that has led to the conception of electric waves; second, to follow this with a discussion of the properties of electric waves and electric oscillations; third, to give a history of the application of electric waves to wireless telegraphy ; and fourth, to elaborate the general principles and methods of electric-wave telegraphy in sufficient detail to be of possible use to elementary students of electricity and to amateur and pro- fessional electricians engaged in operating and constructing wire- less telegraphic apparatus. The author wishes to express his sincere thanks to Commander S. S. Robison of the United States Navy, to Mr. Elliott Woods of Washington, and to Chief Inspector D. M. Mahood of the New York Navy Yard for their kindness in supplying photographs for some of the illustrations. Also, the author is grateful to the Editors of the Physical Review for the loan of Plates I and II, and to Mr. Greenleaf Whittier Pickard for the privilege of consulting his manuscript account of experiments on the effects of daylight on transmission. Finally, the author takes great pleasure in expressing his gratitude to his friend Mr. George Francis Arnold, who has kindly read the proofs and made many valuable sug- gestions. G. W. PIERCE. HARVARD UNIVERSITY, CAMBRIDGE, MASS. July, 1910. TABLE OF CONTENTS CHAPTER I Page INTRODUCTION 1 CHAPTER II ON THEORIES AS TO THE NATURE OF ELECTRICITY 6 CHAPTER III ON THE RELATION BETWEEN ELECTRICITY AND MAGNETISM 12 CHAPTER IV ON THE RESEMBLANCE OF SELF-INDUCTION TO MECHANICAL INERTIA . 20 CHAPTER V ON ELECTROSTATIC CAPACITY > 23 CHAPTER VI ON THE DISCHARGE OF A CONDENSER THROUGH AN INDUCTANCE AND RESISTANCE 28 CHAPTER VII MAXWELL'S THEORY. ELECTRIC WAVES. THE ELECTROMAGNETIC THEORY OF LIGHT 36 CHAPTER VIII THE EXPERIMENTS OF HERTZ 42 CHAPTER IX EXPERIMENTS ON THE IDENTITY OF ELECTRIC WAVES AND LIGHT 51 CHAPTER X ON THE PROPAGATION OF ELECTRIC WAVES ON WIRES 62 CHAPTER XI WIRELESS TELEGRAPHY BEFORE HERTZ 75 vii viii TABLE OF CONTENTS CHAPTER XII Page WIRELESS TELEGRAPHY BY HERTZIAN WAVES. MARCONI, 1896-1898 . . 80 CHAPTER XIII ELECTRIC WAVE TELEGRAPHY BY RESONANT CIRCUITS 92 CHAPTER XIV NATURE OF THE OSCILLATION. THE GROUNDING OF CIRCUITS 108 CHAPTER XV PROPAGATION OVER THE EARTH 122 CHAPTER XVI ON DETECTORS 140 CHAPTER XVII ON DETECTORS (Continued). CRYSTAL RECTIFIERS 157 CHAPTER XVIII ON DETECTORS (Continued) . FURTHER EXPERIMENTS ON THE CRYSTAL RECTIFIERS 175 CHAPTER XIX ON DETECTORS (Concluded). THE ELECTROLYTIC DETECTOR, AND VACUUM DETECTORS 201 CHAPTER XX ELECTRICAL RESONANCE. .WAVE METERS. RESONANCE IN SIMPLE CONDENSER CIRCUITS 215 CHAPTER XXI ON RESONANCE (Continued). ON THE ELECTRICAL OSCILLATIONS OF CONNECTED SYSTEMS OF CONDENSER CIRCUITS 228 CHAPTER XXII TUNING THE SENDING STATION 243 CHAPTER XXIII SOME RECENT METHODS OF EXCITING ELECTRIC WAVES. THE SING- ING ARC, THE SINGING SPARK, AND THE QUENCHED SPARK 253 CHAPTER XXIV RESONANCE OF RECEIVING CIRCUITS. THE POSSIBILITY OF PRE- VENTING INTERFERENCE 271 TABLE OF CONTENTS ix CHAPTER XXV Page DIRECTED WIRELESS TELEGRAPHY 296 CHAPTER XXVI WIRELESS TELEPHONY 305 CHAPTER XXVII SOME DETAILS OF CONSTRUCTION OF WIRELESS TELEGRAPHIC APPA- RATUS 312 CHAPTER XXVIII CONCLUSION v 327 APPENDIX I ELEMENTARY FACTS ABOUT ELECTRICITY, AND DEFINITIONS OF UNITS 329 APPENDIX II CONCERNING THE CALCULATION OF RESISTANCE, SELF-INDUCTANCE AND CAPACITY. . 337 OF THE UNIVERSITY OF WIRELESS TELEGEAPHY CHAPTER I INTRODUCTION ALMOST every one has seen and heard the noisy, brilliant spark produced by the discharge of a Leyden jar. The experiment, shown in elementary courses in physics, is usually performed as follows: The inner and outer coatings of the Leyden jar are connected to the terminals of a static electric machine. The machine is set in rotation and the jar is charged. After the jar has been charged, the electric machine is disconnected, and one end of a metallic rod, held by an insulated handle (see Fig. 1), is FIG. 1. Leyden jar and discharger. FIG. 2. Leyden jar with coil in discharge circuit. touched against the outer coating of the jar, while the other end of the rod is made to approach a knob connected with the inner coating. Before the conductor to the inner coating is actually touched, a discharge occurs through the metallic rod, producing a vivid spark at the gap intervening between the knob and the dis- charge rod. As a variation of the experiment, in the place of the straight or slightly curved metallic rod used in the discharge appa- ratus of Fig. 1, a coil consisting of a few turns of heavy wire may 2 WIRELESS TELEGRAPHY be employed to form a part of the circuit between the two coatings of the jar, as is shown in Fig. 2. The flow of electricity in a circuit of the form of Fig. 1 or Fig. 2 has been the subject of many interesting theoretical and ex- perimental investigations di- rectly applicable to the subject under consideration. The Experiments of Joseph Henry. Some experiments performed by Professor Joseph Henry 1 of Princeton Univer- sity in the year 1842 gave intimation that, under certain conditions, the discharge of the Leyden jar takes place in an oscillatory fashion. Let us give a brief description of Henry's experiment. A small sewing needle was placed within the coil of wire of the discharge circuit, as is shown at N, Fig. 2, so that the elec- tricity from the Leyden jar was made to flow in the coil around the needle. At the time of Henry's experiment it was already well known that a current of electricity from an ordinary galvanic battery, when caused to flow in a coil of wire encircling a steel needle, magnetizes the needle. Henry's experiment showed FIG. 3. Rotating-mirror photograph that the current of electricity of oscillatory discharge. f rom the Leyden jar also produced magnetization of the needle. It was partly in search of this fact, showing the identity of static and galvanic elec- tricity, that Henry's experiment was undertaken. In the experi- ment, however, Henry discovered the additional fact, that with 1 See Scientific Writings of Jos. Henry, Vol. I, p. 201, Washington, 1886. INTRODUCTION 3 the Leyden jar always charged in the same direction by the electric machine used to charge the jar, the needle was some- times found to be magnetized in one direction and sometimes in the opposite direction, indicating that the current that pro- duced the magnetization of the needle was flowing in the coil in the one case from the outside of the jar towards the inner coat- ing, while in the other case it was flowing from the inner coating to the outer coating. This effect could be explained by supposing that the current from the Leyden jar was oscillatory, having first one direction and then the other, and that the magnetization of the needle was reversed .at each reversal of the current, the direc- tion of the magnetization at the end of the experiment being fortuitously determined by the direction last taken by the current. Professor Henry's experiment, though not conclusive, gave strong evidence of the oscillatory character of the discharge; and the opinion that the discharge is oscillatory was repeatedly expressed and defended by Professor Henry in a number of papers and scientific addresses delivered between 1842 and 1850. Sir William Thomson's Theoretical Proof of the Oscillatory Nature of the Discharge of the Leyden Jar. In 1853 Sir William Thomson, 1 who was afterwards Lord Kelvin, proved by mathe- matical reasoning that under certain conditions the discharge of a Leyden jar occurs in an oscillatory manner. Under certain other conditions the discharge is non-oscillatory. In the case of the oscillatory discharge the electricity does not simply flow from one coating to the other until the jar is in a condition of electric neu- trality, but rushes back and forth between the two coatings a great number of times, with a frequency depending on the dimen- sions of the jar and the dimensions and form of the coil through which the discharge occurs. Feddersen's Revolving-Mirror Experiment. In 1859 Doctor Feddersen of the University of Leipzig, by a very beautiful experi- ment, proved the correctness of the surmise of Henry and the mathematical predictions of Thomson. Feddersen's experiment consisted in photographing the spark produced by the discharge of the Leyden jar. A photograph similar to that obtained by Feddersen is shown in Fig. 3. A sketch of the apparatus used in taking the picture is shown in Fig. 4. Instead of employing an ordinary camera to take the picture, the light from the spark S, produced by the discharge of the jar, was allowed to fall upon 1 Wm. Thomson: Philosophical Magazine [4], 5, p. 393, 1853. 4 WIRELESS TELEGRAPHY a rapidly revolving concave mirror M of Fig. 4, and was received upon a photographic plate P after reflection from the mirror. Just as the light of a sunbeam entering a room may be reflected upon the wall or ceiling of the room by a mirror held in the hand, and may be made to move rapidly up the wall or across the ceiling by a rotation of the mirror, so in Feddersen's experiment the motion of the mirror caused the image of the spark to trail rapidly across the photographic plate. If the spark had been steady and unidi- rectional, the image on the plate would have been simply a band of light with a length depending on the speed of the mirror and the duration of the spark. The picture (compare Fig. 3) shows, on the contrary, that the conditions at the spark reversed several times during the discharge, so that first one terminal of the spark FIG. 4. Rotating mirror apparatus. was bright and then the other, the bright terminal being evi- denced by the bright spots on the photograph of Fig. 3. The suc- cessive alterations of the bright spots from one side to the other of the photograph showed the successive reversals of the current across the spark gap during the discharge. Feddersen's photographs proved beyond doubt the correctness of Thomson's prediction of the oscillatory nature of the discharge, and gave, as we shall see later, a very beautiful method of measur- ing the periodic time of oscillation of the discharge, a time which may be only a small fraction of a millionth of a second, and which is yet subject to accurate physical measurement. It is by means of electric oscillations similar to those produced by the Ley den-jar discharge that wireless telegraph signals are produced. INTRODUCTION 5 Electric Waves. Maxwell's Theory. In a letter to C. H. Cay, Esq., dated 5th of January, 1865, James Clerk Maxwell, then Pro- fessor of Physics in the University of Edinburgh, wrote: " I have also a paper afloat with an electromagnetic theory of light, which till I am convinced to the contrary, I hold to be great guns." This paper to which Maxwell referred contained a prediction, based on careful mathematical reasoning, that electric oscillations in a circuit produce electric waves in surrounding space, that these waves travel away with the velocity of light, and that light itself is simply a train of electric waves of extremely short wave length. This prediction of Maxwell, correlating the phenomena of light and electricity, is one of the most beautiful philosophic specula- tions in the history of science, and long remained without direct experimental confirmation; but now, thanks to the brilliant experi- ments of Heinrich Hertz, the existence of electric waves with properties intimately related to those of light waves is a well- established fact of experience capable of verification in even very elementary physical laboratories. It is by means of these electric waves that the signals of wireless telegraphy and telephony are propagated through space. In the succeeding chapters, we shall take up more in detail the course of reasoning that led to Thomson's and Maxwell's pre- dictions, the course of experimenting that led to the proofs of the existence of their electric oscillations and electric waves, and the development of the very striking methods that have been employed in utilizing these electric oscillations and electric waves in the transmission of signals. The discussion will introduce some details apparently remote from commercial usefulness ; but 1 it should be borne in mind that it has been by means of persistent and labo- rious study of these details that the practical result has been attained. CHAPTER II ON THEORIES AS TO THE NATURE OF ELECTRICITY IN the preceding chapter mention has been made of the oscil- latory flow of electricity back and forth between the two coatings of a Ley den jar, when the jar is allowed to discharge through a conductor. The description there given of the " flow of elec- tricity " will probably call to the mind of the reader a picture of a motion back and forth of some kind of material substance from one reservoir to another. At the same time,, it may be difficult to imagine the flow of any kind of substance through the solid metal of which the conductor is composed. What then is this electricity that can flow through solid conductors ? This is a question that we cannot hope to answer to our complete satisfaction, but we have recently come to have so much light thrown upon the question that it is proposed to devote a few pages to the discussion of theories as to the nature of electricity. In the works of the early writers on electricity two prominent hypotheses have been made as to the nature of electricity. These have been called the two-fluid theory and the one-fluid theory. The chief facts that these theories were at first called upon to explain were: (1) The phenomenon of electrostatic attraction and repulsion;, for example, the attraction or repulsion between two charged pith balls, and (2) The fact that when electrification was produced in any way two opposite charges were always obtained; for example, when a glass rod is rubbed with silk, a certain quantity of positive elec- tricity appears on the glass rod and an equal quantity of negative electricity appears on the silk. THE TWO-FLUID THEORY According to the two-fluid theory all bodies in their unelectrified condition were supposed to contain equal quantities of two subtle fluids, one of which was called positive electricity, and the other negative electricity. On this theory the process of positively elec- 6 THEORIES AS TO THE NATURE OF ELECTRICITY 7 trifying a body consists in adding to it a quantity of the positive fluid or taking from it a quantity of the negative fluid. The state of electrification of a body is hence determined by the excess in amount of one of the fluids over the other. In order to account for the fact that the appearance of electrification of one sign is always accompanied by the appearance of an equal amount of electrification of the opposite sign, the two fluids were supposed to be uncreatable and indestructible, so that the accumulation of positive electricity in one body is always accompanied by the loss of positive electricity in some other body. This is the principal property that the electrical fluids were supposed to have in com- mon with ordinary material fluids; namely, the property of conserva- tism in amount according to which the total amount of electricity in a given system can only be changed by the transfer of electricity through the boundary of the system. The electrical fluids, on the other hand, must possess properties that do not belong to the material fluids; for example, portions of the positive fluid must be supposed to repel each other, as do also portions of the negative fluid, while the two unlike fluids attract each other. Another property of the electrical fluids still more at variance with the known properties of material fluids is found in the fact that if we add equal quantities of the two electrical fluids to the same body, the condition of the body will be unchanged, so that according to this theory we must suppose that " the mixture of the two fluids in equal proportions is something so devoid of physical properties that its existence has never been detected." THE ONE-FLUID THEORY Benjamin Franklin attempted to describe the phenomena of electricity in terms of a single fluid. According to his theory, one of the fluids, the positive, was retained and called the electric fluid, while the other, the negative fluid of the two-fluid theory, was replaced by ordinary matter. Quantities of the electric fluid were supposed to repel other quantities of the fluid according to the law of the inverse square of the distance and to attract matter according to the same law. Quantities of matter were supposed to repel each other and attract the electric fluid. According to Franklin's theory an excess of the electric fluid rendered the body positive, while a deficiency rendered it negative. 1 J. J. Thomson, Electricity and Matter, Charles Scribner's Sons, 1904. 8 WIRELESS TELEGRAPHY THE ATOMIC STRUCTURE OF ELECTRICITY Both of the theories sketched above are useful in supplying a terminology for electricity and in affording a simple mode of presen- tation of some of the phenomena, but both theories are charac- terized by indefiniteness as to the physical properties of electricity. Recently, however, a number of phenomena have been studied that have led to a somewhat bolder statement as to the nature of electricity. In accordance with data obtained chiefly from the study of the conduction of electricity by liquids and gases, elec- tricity is now generally supposed to have a structure that may be called atomic. The first evidence pointing in this direction was obtained by Faraday in the course of a research on the conduction of electricity by decomposable liquids. When an electric current is passed through water, the water is decomposed into hydrogen, given off at one electrode, and oxygen, given off at the other. A great many other liquids for example, the aqueous solutions of vari- ous salts are similarly decomposed by the action of the current. An electrically decomposable liquid is called by Faraday an Elec- trolyte. Faraday discovered the following laws of electrolytic decomposition . I. In a given electrolyte, the amount of substance decomposed by various electric currents is proportional to the quantity of electricity sent tnrough the electrolytes. II. If the same amount of electricity is sent through various electrolytes, the amount of the several decomposition products obtained from the various electrolytes is proportional to the com- bining weights of the products obtained. For example, if hydro- gen (2 H) and Oxygen (0) are obtained in one electrolytic cell, and silver (Ag) and chlorine (Cl) in another cell, the amounts of these various substances obtained, when a given electric current is sent through both cells, are in the ratio of their chemical combining weights. According to the atomic theory of matter, these two laws may be interpreted by supposing that each of the decomposition prod- ucts carries a charge that is an integral multiple of the charge carried by the hydrogen atom; so that, if the hydrogen atom, in the process of carrying a current electrolytically, is supposed to have associated with it a definite small quantity of electricity, any combination of atoms, when carrying a current, have asso- THEORIES AS TO THE NATURE OF ELECTRICITY 9 elated with them an equal small quantity of electricity or an inte- gral multiple thereof. That is, the charges we meet with are never fractional parts of the charge carried by the hydrogen atom; whence we may suppose that the latter charge is an elemental quantity of electricity. In discussing the evidence afforded by Faraday's experiments Helmholtz 1 says that " if we accept the hypothesis that the elementary substances are composed of atoms, we cannot avoid the conclusion that electricity, positive as well as negative, is divided into definite elementary portions which behave like atoms of electricity." The study of the conduction of electricity through gases gives still stronger e vide ace of the atomic character of electricity. Gases under the action of certain agencies Roentgen rays, ultra-violet light, radium, high electromotive forces, electric spark, etc. become conductive and retain their conductivity long enough to permit a study of the mechanism by which the electricity is con- ducted. As in the case of the study of conduction in liquids, we are again " led to the conception of a natural unit or atom of electricity of which all charges are integral multiples, just as the mass of a quantity of hydrogen is an integral multiple of the mass of a hydrogen atom." 2 By the study of conduction in gases definite information is obtained in regard to the magnitude of this charge. In a series of experiments performed chiefly at the Cavendish Laboratory of Cambridge University the quantity of electricity in one electrical atom is found to be 3.4 X 10~ 10 electrostatic c. g. s. units. 3 This quantity obtained from experiments on conduction in gases is the same as the quantity of electricity carried by one hydrogen atom in the electrolysis of liquids. Mass of the Carriers of Electricity. Also at the Cavendish Laboratory evidence as to the mass of the carriers of electricity has been obtained by an experimental determination of the ratio of e/m, in which e is the elemental charge and m is the mass of matter carrying the charge. The result obtained is that the mass of the carrier, when the electricity is negative, is about 1/1000 of the mass of the hydrogen atom. This mass is apparently the same 1 J. J. Thomson, Electricity and Matter, p. 73, Charles Scribner's Sons, 1904. 2 J. J. Thomson, Electricity and Matter, p. 83, Charles Scribner's Sons, 1904. 3 The electrical units are defined in Appendix I. 10 WIRELESS TELEGRAPHY whatever the nature of the gas in which the particle happens to be found. While the mass of the carrier of positive electricity is approximately the mass of the atom of ordinary matter, and apparently differs from one gas to another in the same way as the atoms of the gas differ. J. J. Thomson proposes the name " cor- puscle " for the unit of negative electricity, and sums up the corpuscular theory of electricity as follows: ' These results lead to a view of electrification which has a striking resemblance to that of Franklin's One-Fluid Theory of Electricity. Instead of taking, as Franklin did, the electric fluid to be positive we take it to be negative. The Electric Fluid of Franklin corresponds to an assemblage of corpuscles, negative electrification being a collection of these corpuscles. The trans- ference of electrification from one place to another is effected by the motion of corpuscles from the place where there is a gain of positive electrification to the place where there is a gain of nega- tive. A positively electrified body is one that has lost some of its corpuscles. We have seen that the mass and the charge of the corpuscles have been determined directly by experiment. We in fact know more about the electric fluid than we know about such fluids as air and water." In applying Thomson's Theory to the flow .of electricity in con- ductors we must suppose that these small charged bodies, with a mass equal to about 1/1000 of the mass of the hydrogen atom, are able under the action of electric forces to move through the substance of even such solid conductors as the metals, and that a stream of these small charged bodies constitutes or carries the electric current. We must, however, bear in mind that the stream of negative particles is in the opposite direction to the direction conventionally ascribed to the electric current. If we wish now to picture to ourselves the flow of electricity in the Ley den-jar discharge, we may think of a stream of these small negatively charged corpuscles passing from the outer coating of the jar through the discharge rod and across the spark gap and accumulating on the inner coating. This charges the inner coating negatively and leaves the outer coating deficient in corpuscles and therefore charged positively. The stream of corpuscles then re- verses, flows from the inner coating to the outer, and reverses the charge on the jar. This process continues, each time with a loss 1 J. J. Thomson, Electricity and Matter, p. 88, Charles Scribner's Sons, 1904. THEORIES AS TO THE NATURE OF ELECTRICITY 11 of electromotive force, until the electric tension finally becomes too small to force the corpuscles across the spark gap. This description is given merely so that the reader may picture an electric current to his mind. The question as to how and why the discharge of the Leyden jar oscillates will be discussed later, after some more of the elementary facts about electric currents have been presented. CHAPTER III ON THE RELATION BETWEEN ELECTRICITY AND MAGNETISM IN the preceding chapter I have given some of the newest specu- lations in regard to the nature of electricity. The particular views there expressed are not essential to the development of the con- ception of electric oscillations and electric waves, so that the reader may be skeptical about the atomic structure of electricity and still be able to follow the arguments for Maxwell's Theory. In the present chapter I wish to return to surer ground, and give some of the older experiments on electricity and magnetism and on the relation of electricity to magnetism. Prior to 1820 the phenomena of electricity and magnetism were not known to be related to each other. The familiar facts about magnetism were: that there is a mineral called loadstone that has the power of attracting pieces of iron; that a piece of soft iron brought near the loadstone becomes also a magnet with the power to attract iron, but only temporarily, for the piece of soft iron loses most of its magnetism when it is removed to a distance from the loadstone; while a piece of hardened steel brought near the load- stone or another magnet becomes a so-called permanent magnet, and retains a considerable part of its magnetism even when at a great distance from the loadstone. It was also known that a steel needle, magnetized by rubbing it on a loadstone or another magnet, and pivoted so as to be free to rotate in a horizontal plane, points in approximately a north and south direction. About electricity it was known that amber, glass and sealing wax were capable of being electrified by rubbing them with silk, flannel, fur, etc. ; that the electrifications so produced were of two kinds, positive and negative; that unlike charges attract each other and like charges repel; that these positive and negative charges could be stored in an apparatus of the form of a Leyden jar; that certain bodies, such as metals, carbon, water and so forth, were conductors cf electricity, so that the electricity would flow freely through such bodies. Also the galvanic cell was known, and 12 RELATION BETWEEN ELECTRICITY AND MAGNETISM 13 was employed to produce a continuous flow of electricity in wires. This continuous flow of electricity in a wire or other conductor is an electric current, and was known to produce heating of the conductor through which it flows. In 1820 a new impetus was given to a study of electricity and magnetism by the discovery by Hans Christian Oersted of Copen- hagen that magnetism and electricity are interrelated. This dis- covery and some of its consequences is described in the succeeding paragraphs. On the Production of a Magnetic Field by a Current of Elec- tricity. Oersted's discovery was nothing less than the important fact that when a pivoted magnetic needle is placed near a wire carrying a current of electricity, the magnetic needle tends to set itself at right angles to the wire which carries the electric current. If the current is reversed, the direction of the deflection of the mag- netic needle is reversed. If the wire carrying the current is moved from a position below the needle to a position above the needle, the deflection of the needle is again reversed. Oersted's discovery has been utilized in the construction of the galvanometer, which is a very delicate instrument for detecting and measuring small electric currents. The principle of the gal- vanometer is as follows: A magnetic needle pivoted as in the ordinary compass, so as to be free to move in a horizontal plane, will, if undisturbed, take up a position in the magnetic me- ridian of the earth; that is, the needle will point approximately north and south, (M, Fig. 5). Suppose, now, that a wire is passed alternately above and below the needle several times so as to form a coil ^.^.^^-^^j (C, Fig. 5), with its windings in the plane S X L - of the magnetic meridian. Let a current -piGJS. Coil and needle be passed through the coil, so as to flow of galvanometer, north above the needle and south below it; the north current above the needle and the south current below it both tend to deflect the north-seeking end of the magnetic needle to the west, so that the effect of the current on the needle is multiplied by the combined action of the several turns of the conductor around the needle. For a highly sensitive galvanometer, the magnetic needle instead of being pivoted is delicately suspended by a fine fiber of spun quartz. 14 WIRELESS TELEGRAPHY In addition to its application to the construction of the galva- nometer, Oersted's principle is utilized in the construction of almost every kind of electromagnetic device. Interpretation of Oersted's Experiment. The results of Oer- sted's experiment are now usually expressed by saying that a current of electricity in a conductor produces a field of magnetic force in the neighborhood of the con- ductor. In explanation of this statement the reader is asked to recall the familiar experi- ment in which a sheet of paper laid upon a bar magnet is cov- ered with iron filings. The filings become magnetized and arrange themselves in curved lines stretching from one pole of the magnet to the other, as shown in Fig. 6. The direc- tion of these lines traced by the filings is approximately the direction of the magnetic force about the magnet. These lines, delineated by the filings, are the lines along which a small suspended magnetic needle would orient itself if brought near the bar magnet. The region in which cuch a magnetic force exists is called a field of magnetic force. A piece of unmagnetized steel when placed in such a field becomes magnetized, and retains some of its magnetism even after it is removed from the field of magnetic force. To show the form of the field of magnetic force about a wire carrying a current, as in Oersted's experiment, iron filings may also be used with the results given in Figs. 7, 8, and 9. Figure 7 is obtained with a straight conductor running perpendicular to the plane of the paper on which the filings are disposed. Tne picture shows that when a current of electricity is sent through the straight conductor the lines of magnetic force are circles about the con- ductor. The magnetic force is stronger near the conductor and weaker at a distance from the conductor. Figure 8 is obtained with a coil of a few turns of wire. Figure 9 shows the magnetic field produced by a long helical coil called a solenoid. With the FIG. 6. Magnetic field about a bar magnet, as depicted by iron filings. RELATION BETWEEN ELECTRICITY AND MAGNETISM 15 solenoid the field of magnetic force is seen to be remarkably like that obtained with the bar magnet. It may be observed that in the case of each of the coils the lines FIG. 7. Magnetic field about FIG. 8. Magnetic field linking with a coil of a straight conductor carry- two turns carrying a current, ing an electric current. of magnetic force depicted by the filings interlink with the electric current. This conception of a field of magnetic force about a conductor carrying an elec- tric current is of funda- mental importance in the study of electric waves, in which the action in the medium rather than the ac- tion in the wires is the chief factor to be reckoned with. So long as the electric current in the conductor remains steady, the mag- netic field remains steady. With changes in the elec- tric current, the magnetic field changes. This changing magnetic field about a conductor carrying an oscillatory current will later be shown to be one of the components of the electric waves produced at the sending station of a wireless telegraph system. FIG. 9. Magnetic field produced by a solenoid. 16 WIRELESS TELEGRAPHY On the Production of an Electric Current by a Variation of the Magnetic Field. Bearing in mind that an electric current pro- duces a field of magnetic force about it, let us turn now to the question whether an electric current can be produced by the action of a magnetic field. For a period of ten years succeeding Oersted's discovery, experi- ments directed to this question gave the answer in the negative. Finally, in 1831, Faraday in England and Joseph Henry in America succeeded, almost simultaneously, in obtaining electric currents by the action of a magnetic field, and in explaining the cause of pre- vious failures. Faraday and Henry showed that an electric cur- rent in a conductor in a magnetic field is obtained as the result of a change in the magnetic field, whereas the previous experiments had sought to produce the effect by the magnetic field in a steady state. One way of producing the required change of magnetic field in the neighborhood of the electric circuit is by the motion of a per- manent magnet, with its accompanying field, toward or away from the circuit. This was done in some of Faraday's and Henry's experiments and is here described with the aid of Fig. 10. A coil of wire C is connected to a galva- FIG. 10. Apparatus for showing the pro- nometer G. When the north duction of a transient electric current i r xt. Arc by the motion of a permanent magnet. P ole of the magnet NS IS made to approach and enter the coil C, the needle of the galvanometer is deflected, showing that an electric current is produced. The current is, however, only tran- sient, and after the magnet NS has arrived at its final position and ceased to move, the needle of the galvanometer comes back to its zero position, showing that the current has subsided. Now, however long the magnet NS is left stationary within the coil C, no current is produced. But if the magnet is quickly withdrawn, the galva- nometer registers a current in the direction opposite to the current obtained by the introduction of the magnet. This current is also transient, and subsides when the magnet NS becomes stationary. If the south pole of the magnet is now introduced into the coil, the galvanometer shows a transient current opposite to that produced by the introduction of the north pole. The withdrawal of the south pole gives a transient current opposite to that caused by RELATION BETWEEN ELECTRICITY AND MAGNETISM 17 its own introduction, and in the same direction as that given by the introduction of the north pole. Another way of obtaining a similar result is to employ two coils of wire placed near each other but not electrically connected, as FIG. 11. Apparatus for showing electromagnetic induction. shown in Fig. 11. One of these coils, S, which we will call the secondary, is connected with the galvanometer (7, while the other, called the primary, P, may be connected with the ter- minals of a galvanic battery B. No current is shown in the gal- vanometer when a constant current is sent through the primary; but when the current in the primary is made, broken or reversed, transient currents are obtained in the galvanometer. That is to say, the current in the primary sets up a magnetic field linking with the secondary circuit. While the primary current is steady, this field is steady and no effect is obtained in the secondary. But variations of the current in the primary cause variations of the magnetic field and consequently currents in the secondary. The variable currents in the secondary are said to be induced by the variable currents in the primary, and the phenomenon is referred to as electromagnetic induction. It is in part by action of this kind that currents at the receiving station of a wireless tele- graph system are produced by the action of variable currents at the sending station. The extension of the effects of electromag- netic induction to the case of two circuits widely separated from each other we shall see to be the result of the use of extremely rapid electric oscillations at the sending station. On Mutual Induction. Let us examine a little more specifi- cally the case of electromagnetic induction described hi the gal- vanometer experiment cited above. This experiment shows that when the current in the primary coil is increasing, the current induced in the secondary coil is hi 18 WIRELESS TELEGRAPHY the opposite direction to the primary current; while, if the current in the primary is decreasing, the current in the secondary is in the same direction as the primary current. Perhaps it would be better to speak of the electromotive force l in the secondary rather than the current, because the electromotive force in the secondary bears a simple relation to the current in the primary. The simple relation is, that the electromotive force in the secondary is pro- portional to the time rate of change of the current in the primary. If E 2 is the electromotive force induced in the secondary, /i the current in the primary, /i the time rate of increase of the current /i, then theory and experiment show that # 2 = - Mi,, (i) in which M is a constant depending on the form and position of the two circuits. M is called the coefficient of mutual induction, or, more briefly, the mutual inductance of the two circuits. M is found to have the same value if the variable current is sent through the second- ary and the electromotive force examined in the primary. Consistent with the above equation, the Mutual Inductance of two circuits is defined as the electromotive force induced in one of the circuits when the current in the other is changing at the rate of one unit current per second. The mutual inductance between two circuits is increased by increasing the number of turns on either or both of the circuits or by bringing the circuits nearer together, or by introducing iron or other magnetizable metals within the circuits. Methods of calculating the mutual inductance of circuits of various forms are given in Appendix II. By a reference to equation (1) given above it is seen that the electromotive force induced in the secondary is increased by in- creasing the rate of change of current in the primary. That is, in order to get a large induced electromotive force at our receiving station we should have as large a current as possible at our send- ing station and change it as rapidly as possible. This result is best attained by the use of currents of high frequency at the sending station, such as are obtained by the discharge of a Ley den jar. Self-induction. In the case of the two coils placed near together in the preceding discussion, it was found that the elec- 1 This term is defined in Appendix I. RELATION BETWEEN ELECTRICITY AND MAGNETISM 19 tromotive force in the secondary is produced by a variable magnetic field from the primary interlinking with the secondary. Now, if instead of two coils we have one coil alone carrying a vari- able current, the variable current produces a variable magnetic field linking with the circuit itself, and in consequence a back electromotive force is produced in this coil tending to oppose the variation of the current in it. This action of the current on itself is called self-induction. The back electromotive force due to self-induction in the circuit is connected with the current in the circuit by the formula #1= - Li/!, (2) in which L^ is called the coefficient of self-induction, or, more briefly, the self-inductance of the circuit. 7 t is an abbrevia- tion for the time rate of change of the current. The subscripts 1 show that all the quantities refer to the same circuit. Consistent with equation (2), the self-inductance of a circuit may be defined as the back electromotive force of induction in the circuit when the current in the circuit is changing at the rate of one unit current per second. The numerical value of the self -inductance depends on the geo- metrical form of the circuit. In Appendix II formulas are given for calculating the self-inductance of some simple forms of circuit. This discussion of self-inductance is here introduced in quanti- tative terms, because this quantity is of fundamental importance in the study of oscillatory currents. I am aware that the semi- mathematical form in which the idea is presented may fail to give a clear conception of the phenomenon, so I propose to attempt in the next chapter to describe self-induction by the aid of cer- tain familiar analogies. CHAPTER IV ON THE RESEMBLANCE OF SELF-INDUCTION TO MECHANICAL INERTIA IN the previous chapter it has been pointed out that sejf -induc- tion is the action of a variable current on itself due to the produc- tion of a variable magnetic field by the current. When a current of electricity flows in a circuit of any form, a field of magnetic force is set up and links with the circuit. The* manner in which the flow of current in the wire produces magnetic effects in the surrounding medium is not completely understood, but that such effects exist is made evident by bringing a magnetic compass needle up near the circuit; the compass needle tends to set itself in certain intangible lines called lines of magnetic force. The lines of magnetic force produced by a current are closed curves linking with the wire carrying the current, as is shown by the compass needle or by the distribution of the iron filings depicted in Figs. 7, 8 and 9. Experiments similar to those cited in the previous chapter show that when a change is made in the electric current in the wire, the magnetic field surrounding the wire is changed, and that these changes in the magnetic field impress back upon the circuit an electromotive force opposing the change of current. The self- induction of an electric circuit may thus be described as a property that tends to prevent a change of the electric current in the circuit. In this respect self-induction resembles the property of inertia in matter. The inertia of a body is that property by virtue of which a body tends to persist in its state of rest or motion. If a body is at rest or is moving with a given velocity, the inertia of the body opposes a change of its state of rest or motion. In a similar manner, the self-induction in an electric circuit opposes a change of the electric current in the circuit. From this it need not be inferred that electricity itself is a form of matter possessing inertia, because in the case of the electric current we may believe that the inertia resides primarily not in the electricity but in the magnetic field set up by the current. 20 SELF-INDUCTION TO MECHANICAL INERTIA 21 The correctness of this belief is evidenced by the fact that with a fixed current flowing in a wire the self-induction may be greatly increased by bending the wire into the form of a coil. Now mak- ing the wire into a coil does not change the amount of electricity flowing in the wire, but it does change the strength of the mag- netic field about the wire. The inertia of the current, therefore, has its existence not primarily in the conductor but in the medium surrounding the conductor. The Contrast of Self-induction with Resistance and its Resemblance to Inertia. The self-induction of a circuit acts upon the current in a manner entirely different from the manner in which resistance acts. The resistance of a circuit always op- posed the flow of the current, and when a current is sent through a conductor, some of the energy of the current is used up in over- coming the resistance of the conductor; or, more properly speaking, some of the electric energy is converted into heat. This is true whether the current is increasing or diminishing or is steady ; and the heat developed is not again completely available for producing electric current, so that a continuous supply of energy is needed at the source of the electric current to keep up the current against the resistance of the circuit. Self-induction, on the other hand, does not change the electrical energy into heat. When the current is steady, self-induction has no effect. If, however, the current is increasing, some of the energy supplied to the system is employed in establishing the magnetic field. If now the current is allowed to decrease by an equal amount, the energy stored up in the magnetic field is re- stored to the conductor and helps to maintain the current. Thus, during a cyclic * change of the current as much energy may be obtained from the magnetic field as was given to it. Hence, if we have an oscillatory current in a circuit, none of the energy of the current is consumed by the action of the self-induc- tion, and the supply of energy at the source is wasted only in overcoming the resistance of the circuit. 2 It is apparent that in respect to the consumption of energy self- induction resembles inertia in matter. Energy is required in order 1 A cyclic change is a change from any value A to any other value B y and from B back to A again. 2 Later we shall see that for some forms of circuit this statement is not strictly true, because some of the energy may be radiated as electric waves. Also in the case of some media, as iron, in the field of magnetic force, some of the energy is converted into heat by hysteresis. 22 WIRELESS TELEGRAPHY to set a heavy body in motion, but this energy is recovered when the body is stopped, and the only loss of availability of energy in a cyclic process in which a body is started in motion and stopped again is that lost in overcoming friction in the machinery used for starting and stopping the body. When a body is set in motion, the energy supplied in producing the motion is stored up in the body as kinetic energy, so that analo- gously many writers refer to the energy of the magnetic field as kinetic in character. Without necessarily committing ourselves to this specific proposition as to the kinetic character of the magnetic field, it will still be useful to keep in mind that self-inductance opposes changes in the electric current in the same general manner as inertia opposes changes in the motion of bodies. Keeping this analogy in mind, we can easily foresee many of the facts about the flow of electricity; for example, suppose that a rapidly alternating electromotive force is applied to a circuit con- taining a large self-inductance; usually only a small current will flow, just as only a small motion will generally be communicated to a heavy body by a rapidly varying material force. There are. however, special cases in which the periodic force will set up a large motion of the material body. This happens when the period of the force is the same as the natural period of the body. But in order for the body to have a natural period something besides inertia is required ; namely, the body must be elastic or must be elastically attached to something. So in the case of the electric circuit it is also possible to get a large current with a rapidly varying electromotive force, provided the circuit con- tains besides its self-inductance a suitable amount of electrostatic capacity, which will be shown to supply the factor required to give periodicity to the electric circuit. Before developing further our notions in regard to self-induc- tance, it is proposed to introduce this other phenomenon that enters prominently into the discussion of electric waves; namely, the phenomenon of electrostatic capacity. CHAPTER V ON ELECTROSTATIC CAPACITY THE last two chapters have been devoted to a discussion of electric currents and the magnetic field accompanying such cur- rents. In order to arrive at a conception of the nature of electric waves it is necessary also to give some attention to the action of electric charges at rest. This is the subject of electrostatics. Here again we must look to Faraday for the fundamental discoveries. In the beginning paragraph of his most important research on this subject Faraday says: l " To those philosophers who pursue the inquiry zealously yet cautiously, combining experiment with analogy, suspicious of their preconceived notions, paying more respect to fact than to theory, not too hasty to generalize, and above all things, willing at every step to cross-examine their own opinions, both by reasoning and by experiment, no. branch of knowledge can afford so fine and ready a field for discovery as this." Influence of Intervening Medium on Electric Attraction. The result obtained by Faraday in the research referred to is that the electrostatic repulsion or attraction between two charged bodies is influenced by the medium intervening between the charged bodies. If, for example, we have two flat metallic plates placed parallel to each other, and we charge one of the plates positively and the other negatively, the electrostatic attraction between the two charges on the plates will be less when the plates are separated by glass than when they are separated by air, provided the plates are charged with the same quantity of electricity in the two cases. The attraction between the charges on the plates with glass inter- vening will be about one-sixth as much as that with the same thick- ness of air intervening; so that in order to get the same force between the charges on the plates in the two cases we must put upon the plates with glass between them six times as much elec- tricity as is required with air between. 1 Faraday: Experimental Researches in Electricity and Magnetism, Vol. I, Eleventh Series, Nov., 1837. 23 24 WIRELESS TELEGRAPHY We thus come to the result that the insulating medium between the oppositely charged metallic plates serves not merely to separate the plates and prevent them from losing their charge, but serves also to determine the charge the plates will receive for a given electromotive force; for example, a given battery connected between the plates. And since the insulating medium between the plates has other functions than merely to insulate, Faraday proposes to designate the insulating medium by the name dielec- tric, when reference is made to the force acting through it. He says, " I use the word dielectric to express that substance through or across which the forces are acting." On Condensers. The apparatus consisting of two conducting bodies separated by a dielectric is called a condenser. An ordinary Ley den jar, consisting of two metallic coatings separated by glass, is a familiar case of an electric condenser. Any two conducting bodies with a. dielectric between constitute a condenser. As an extreme case, a single conducting body isolated in space is con- sidered a condenser, with empty space as dielectric, and with the other conductor removed to an infinite distance. As another example, a charged body in the neighborhood of the earth is a condenser, with the earth for the other conductor and with air as dielectric. Capacity of Condenser. Different condensers are said to have different capacities, which term does not refer to the total amount of electricity that the condensers can contain, but to the quantity of electricity they will take under the action of a given electromotive force; namely, a unit electromotive force. In the practical system of units (see Appendix I), the unit of electromotive force is the volt, the unit of quantity of electricity is the coulomb, and the unit of capacity the farad. A farad is the capacity of a condenser that can be given a charge of one coulomb under the action of electromotive force of one volt. The farad is a very large unit of capacity; for example, the electrostatic capacity of the whole earth is only about .000708 farad. That is to say, it would take only about seven ten-thousandths of a coulomb to raise the poten- tial of the earth one volt. Since the farad as a unit is very large, the capacity of a condenser is often designated in millionths of a farad, or microfarads. The quantity of electricity, Q, on each plate of a condenser of capacity C is Q = CV, where V is the difference of potential between the plates. ELECTROSTATIC CAPACITY 25 Dielectric Constant. Returning, now, to the function of the dielectric in determining the capacity of a condenser, the term dielectric constant of a substance is used to designate the capacity of a condenser with the substance as dielectric relative to the capacity of the same condenser with empty space as dielectric. The dielectric constant of air and all the gases at ordinary pressure is approximately unity; this means that the capacity of a con- denser with a gas as dielectric is not much changed when the gas is pumped away. In the example cited above the dielectric con- stant of a particular glass is given as six; that is, the quantity of electricity that a condenser will contain under a given electro- motive force with this glass as dielectric is six times the quantity the condenser will contain under the same electromotive force when air is substituted for the glass. A table of dielectric con- stants, together with some numerical formulas for calculating the capacity of some simple forms of condenser and rules for combina- tions of condensers in series and parallel, is given in Appendix II. General Facts about Energy and Electromotive Force of Charged Condenser. In order to send a charge of electricity into a condenser, energy is required, but the energy is not converted into heat, as it is in the case of a current of electricity flowing through a resistance ; for the energy of the charge may be recovered as electric energy when the condenser is allowed to discharge. In a cyclic process in which a condenser is charged and discharged again, there is no loss of availability of energy in the processes that occur in the condenser. And when a condenser charges and dis- charges several times in an oscillatory manner, it is necessary to supply energy from without only in so far as the electric energy is radiated or is converted into heat in flowing through some resist- ance in the circuit. 1 It has undoubtedly been observed by the reader that in respect to the reception of energy from the circuit and the return of the same amount of energy to the circuit again the medium of the condenser behaves somewhat like the medium of the magnetic field. There is, however, one marked difference. In the case of the magnetic field, the opposing electromotive force called into play by self-induction is proportional to the rate at which the current is changing; while, hi the case of the condenser, the electromotive force V opposing the flow of electricity into the condenser is proportional 1 This statement is not always strictly true, because in some forms of con- denser a small part of the energy is consumed by hysteresis in the dielectric. 26 WIRELESS TELEGRAPHY to the quantity Q of electricity in the condenser. Numerically V ~ Q, Q and in proper units V = ^ , where C is the capacity of the con- C denser. Mechanical Systems Analogous to an Electrical Condenser. I. We have a condition of things analogous to the charging of a condenser in the act of supplying water to a tall cylindrical reser- voir. The force P required to send water into the reservoir against the hydrostatic pressure of the water already in the reservoir is proportional to the height h of water in the reservoir, which is proportional to the amount of water M in the reservoir. Numeri- cally, in suitable units P = -, where S is the area of cross section >S of the reservoir. S may be looked upon as analogous to C. II. Another analogue to the action of a condenser is found in the forces called into play in the act of compressing an elastic spring. The restoring force F of the spring is proportional to the amount x by which the spring is compressed. Numerically, F = ex, where e is the stiffness of the spring. In the case of the condenser it should be borne in mind that the greater the capacity of the condenser the less the electromotive force required in order to charge it with a given amount of elec- tricity. In this respect capacity of the condenser resembles the reciprocal of the stiffness of the spring, for the greater the stiffness e of the spring the greater the force F required to compress it by a given amount. Flow of Current in a Circuit Containing a Condenser. The reader will note the following fundamental facts in regard to the action of a condenser. If a battery having a constant electromo- tive force E has its positive pole connected to one plate of a condenser and its negative pole connected to the other plate, electricity will flow into the condenser and charge it. As the con- denser charges it gives rise to a back electromotive force opposing the flow, so that the current is diminished more and more by the opposing e.m.f. of the condenser, as the condenser is charging. The e.m.f. at each instant is proportional to the quantity q of electricity in the condenser and is inversely proportional to the capacity C of the condenser. When this opposing e.m.f. becomes equal to the e.m.f. of the battery, E, the flow of electricity ceases. Then E = where Q is the final charge attained by the con- C ELECTROSTATIC CAPACITY 27 denser. After this condition is reached, no further current flows. This process of charging the condenser is described as gradual because time is required for the final condition to be established, but this time is usually very short. Work Done in Charging Condenser. During this process of charging the condenser, the average e.m.f. of the condenser was E; the work * done, which is the charge introduced multiplied by the e.m.f. of the condenser, is Q X k E', or, substituting for Q its value EC, the work W done in charging the condenser is W = J E 2 C. 1 See definitions of electrical work, in Appendix I. CHAPTER VI ON THE DISCHARGE OF A CONDENSER THROUGH AN INDUCTANCE AND RESISTANCE The Oscillatory Discharge. We are now ready to undertake a more critical examination of the proposition set down in the first chapter that, under certain conditions, the discharge of a Ley den jar is oscillatory. As a mechanical analogy, let us con- sider the motion of a heavy bob attached to an elastic spring. Let the position of rest of the bob be the position a, Fig. 12. If now the bob is pulled down to a posi- tion b and released, the spring draws it back again to a. During this process the bob acquires a ve- locity determined by the stiffness of the spring and the mass of the bob. When the bob reaches a, the spring ceases to pull, but the bob by reason of its inertia moves on up to a position c, during which process the spring is compressed. When the bob has reached c, it has lost its velocity and is now driven back by the compressed spring. In this way the vibratory motion is kept up for some time, and would be kept up indefinitely but LJLJ for the fact that the re- FIG. 12 Spring s i s tance of the air and and bob for . , . . illustrating the imperfect elasticity oscillatory o f fag spring convert motion. _ some of the energy into heat during each excursion, so that the amplitude of the motion is diminished more and more until the body finally comes to rest at a. As another illustration, suppose a body of water to be contained in a bent tube of the form of Fig. 13. Let the surface of the water in its position of rest be at a and a' in the two arms of the tube. Suppose now that the water is moved into the position W and 28 FIG. 13. Water column showing vibratory motion. DISCHARGE OF CONDENSER 29 released. The column of water will vibrate back and forth in the tube so that its level in the left-hand arm of the tube comes suc- cessively above and below the position a. During each excursion the amplitude of the motion is diminished till the water finally comes to rest in its initial position. Both of these forms of mechanical vibratory motion are easily realized in practice, and both bear a marked resemblance to the flow of electricity in the discharge of a condenser through an inductance and resistance. In order now to understand how a condenser discharge may be oscillatory in character, suppose a Ley den jar, or other form of electrical condenser, of capacity C to be initially charged, say from an electric machine, with a quantity of electricity +Qo on one plate and Qo on the other. And suppose that the condenser has in series with it a self -inductance L, and a spark gap S. (Fig. 14.) At first let the spark gap be too wide for the spark to pass. Positive electricity will be distributed over the one coating and one knob of the spark gap, and negative elec- tricity will be distributed over the other coating, the coil L and the other knob of the spark gap. Let V Q be the difference of po- tential between the plates of the condenser. Before the current starts there will be the same dif- ference of potential between the knobs of the spark gap, because all parts of a conductor in which no current is flowing are at the same potential. Let us suppose, now, that the knobs of the spark gap are made to approach each other until the gap is short enough for the poten- tial to start a spark (i.e., about 39,000 volts to the centimeter, if the terminals of the gap are balls 1 cm. in diameter). When the spark starts, the resistance of the gap suddenly drops to a very small value, in some cases to a small fraction of an ohm, 1 and the electric current begins to flow across the gap under the action of the high difference of potential between the plates. 1 We have seen in Chapter II that a spark is one of those agencies that render gases conductive. FIG. 14. Leyden jar, inductance coil, and spark gap. 30 WIRELESS TELEGRAPHY The current flowing through the circuit has a small value when the spark first begins to pass. If it were not for the self-induction of the circuit, the current would spring to a large value, because the electromotive force of the circuit is high and its resistance low. We have seen, however, that the self-induction acts in such a man- ner as to oppose rapid changes in the current. As a result the current requires time to attain its maximum. When the current reaches its maximum, the condenser is completely discharged, but there is a large current flowing. This current cannot stop at once, for the self-induction now acts in the reverse direction and opposes the decrease of the current, so that the current con- tinues to flow after the electromotive force of the condenser has become zero. This process charges the condenser oppositely to its original charge, and when the current in this direction ceases, the back electromotive force of the condenser starts the current in the reverse direction. The condenser is again charged in its original direction, the current again reverses and the process con- tinues for a number of oscillations depending on the resistance, self -inductance and capacity of the circuit. The essential factors entering into the production of the oscilla- tory discharge are the self-inductance and the capacity of the cir- cuit, characterized in their actions by the fact that they are out of phase with each other, so that when the effect of the capacity is a maximum that of the induction is a minimum, and vice versa. On account of the resistance of the circuit some of the electrical energy is converted into heat during each flow of the current, so that the maximum attained by the current at each oscillation falls lower and lower until the spark ceases. The decrease of the ampli- tude of the oscillation under the action of the resistance is referred to as damping of the oscillation by the resistance. It will be seen later that the radiation of energy- as elecbric waves acts also in a manner to damp the oscillations. Criterion. In his mathematical investigation of this problem Sir William Thomson showed that the discharge occurs in the oscillatory manner here described only when the resistance of the circuit does not exceed a certain value relative to the ratio of the self-inductance to the capacity of the circuit. The exact expres- sion of this condition under which the discharge is oscillatory is, R 2 < 4 L/C. DISCHARGE OF CONDENSER 31 Non-oscillatory Discharge. If, on the other hand, J? 2 is greater than 4 L/C, Thomson showed that the discharge is unidi- rectional ; that is, no reversal of the sign of the charge takes place. We should have an analogous condition of affairs with the elastic spring used as an illustration if the bob B (Fig. 12) should be submerged in a liquid, provided the liquid should offer sufficient resistance to the passage of the bob through it. Evidently the amount of resistance required to prevent the oscillation of the bob will increase with increase of the inertia of the bob and with increase of the stiffness of the spring. The former of these cor- responds to L, and the latter to the reciprocal of C, so that the fact that L/C will occur in the condition for the oscillation or non- oscillation of the electrical system might have been anticipated. In the case of the water column, if the connectiag tube EF between the two vertical cylinders in Fig. 13 is made sufficiently small to offer enough friction, the motion of the water will also be non-oscillatory. This is analogous to the case of the non- oscillatory discharge of the condenser. Mathematical Formulas for the Discharge of the Condenser. Thomson derived the following equations for the current i at any time t, where t is measured in seconds from the time when the dis- charge begins: Case 1. If R 2 < 4 L/C, 2V ~ Rt i in which F = the initial difference of potential, R = the resistance, L = the self-inductance, C = the capacity, and e = 2.718281 . . . (base of natural logarithms). This is the case of the oscillatory discharge. Case II. If R 2 > 4 L/C, ^O 32 WIRELESS TELEGRAPHY 2 LC in which TI = - ~ , and RC- V# 2 C 2 -4 LC T, RC This is the general case of non-oscillatory discharge. Case III. If R 2 = 4 L/C, V * if i = -e 2L (5) LI This is the critical case, in which the discharge is just non- oscillatory. Graphical Representation of Results. By the aid of the equa- tions (3), (4) and (5) the current in the condenser circuit at any time can be calculated in any case in which the constants of the circuit and the initial difference of potential of the plates of the condenser are known; of the calculated values so obtained we can construct a table, in the first column of which we may place the time in convenient fractions of a second, and in the second column we may write the different values of the current corresponding to these different values of the time. There is, however, another method of representing the results, which affords an easier comprehension. This is the graphical method, and consists in constructing a curve on a sheet of squared paper with a scale of time and a scale of current at right angles to each other. As an example of this method of showing results, let us refer to Fig. 15, which is a graphical representation of the flow of current in a condenser circuit in which the resistance is supposed to be zero. The horizontal scale through the center of the figure gives the time in millionths of a second; the vertical scale at the left of the figure gives the current. Such a diagram gives the current at any time; for example, when the time is zero, the current is zero. To get the current at one one-millionth of a second, one goes out on the horizontal line to one one-millionth second (which is halfway between and 2), and at this point one erects a vertical line which will be seen to cut the curve at a point the same height as 150 amperes on the margin. This 150 amperes is, then, the current at TI> oo OOF sec. In like manner, at TI> oooo sec., the current is seen to be about minus 130 amperes. From this description of the method of interpreting the curves DISCHARGE OF CONDENSER 33 it will be evident how the curves are drawn; namely, a table is made of current for different values of time, by the aid of formula (3), and then for each value of time plotted horizontally the cor- responding value of current is erected vertically, and through the points so obtained a smooth curve is drawn. This process resem- bles the method employed by navigators to show the route of a ship. Each day, or oftener, an observation of latitude and longi- tude is made, and a point is put on the map at the intersection of the given latitude and longitude; and through the points thus obtained at successive observations a smooth curve is drawn, which represents the course of the ship, and from which the position of !?00r-- -160 -200 1- FIG. 15. Current from a condenser of capacity .01 microfarad discharging through an inductance of .0001 henry. Initial potential 20,000 volts. Resistance zero. the ship at points intermediate between the observations may also be approximately obtained. Curves Showing Condenser Discharge. The manner in which the discharge of a condenser occurs under different conditions is represented graphically in the curves of Figs. 15, 16, 17 and 18. In these curves the time in millionths of a second is plotted hori- zontally, and the current in amperes is plotted vertically. These curves are calculated from the formulas given on page 31. In all four cases the capacity, self-inductance and initial potential are the same; namely, C = 10 ~ 8 farads, L = 10 ~ 4 henrys, V = 20,000 volts. The only difference between the conditions of the discharge in the four cases is the difference in resistance of the circuit through which the discharge occurs. In Fig. 15 the resistance is supposed to be zero, and we have 34 WIRELESS TELEGRAPHY as a result what is called an undamped oscillation. The current oscillates back and forth between a positive maximum of 200 amperes and a negative maximum of 200 amperes. In Fig. 16 the resistance of the circuit is 10 ohms, and in Fig. 17 200 150 100 S 60 <5 q -60 a 5 -100 -160 7V~A 4 C/ 8 \10 12/ 14 \16 18 / 20 Time\ in /MillionthB\ of /a SecondX / 4 \16 18/20 22 24/26 28 --\J/ \JZ--~ k/V t^ FIG. 16. Same as Fig. 15, except that the resistance is 10 ohms. this resistance is 20 ohms. These two curves show how the current is damped by the resistance of the circuit. The curves of Figs. 15, 16 and 17 all come under the conditions of Case I. If, however, the resistance be 200 ohms, we have the condition g-60 o -100 -150 -200 2- \4 0/8 \10 12 / 14 Time\ in /Millionths\ of /a Second 2 4 6 Millionths Second FIG. 18. Same as Fig. 15, except that the resist- ance is 200 ohms. FIG. 17. Same as Fig. 15, except that the resistance is 20 ohms. for the current to be just non-oscillatory, R 2 = 4 L/C, and the equation of the curve is then the equation given under Case III. This kind of discharge is shown in the curve of Fig. 18. This DISCHARGE OF CONDENSER 35 case has also the same capacity, self-inductance and initial voltage as the preceding cases, but the current is seen to rise only to about 75 amperes and then gradually to approach zero. If the resistance be made greater than 200 ohms, we have Case II, in which the discharge is also non-oscillatory. A curve representing this case is not given; the form of such a curve is somewhat like that of Fig. 18, with the exception that the curve does not rise to so great a value and does not approach zero so rapidly as does the curve in Fig. 18. The Period of Oscillation. From equation (3), p. 31, it can be shown that the period of a complete oscillation of the current, in case the discharge is oscillatory, is T = 2* / 2LC =, (6) V4 LC - R 2 C 2 in which T is the time of a complete oscillation in seconds; L, C and R are measured in the same set of units; e.g., henry s, farads and ohms respectively; TT is 3.1416 . . . , the ratio of the circum- ference to the diameter of a circle. Equation (6) is the exact expression for the period, but in most practical cases that occur in the use of electric waves it is found that the effect of the resistance is inappreciable in its effect on the period; that is, in equation (6), R 2 C 2 is small in comparison with 4 LC, so that the expression for the time of a complete oscil- lation simplifies to T = 27rV / LC. (7) This formula is usually sufficiently accurate. For example, in the case plotted in Fig. 16, the period of oscillation calculated by equa- tion (7) differs from the exact value, obtained from equation (6), by one-fourth of one per cent. The various formulas given in this chapter were first obtained mathematically by Sir William Thomson in 1855. In 1859 Fed- dersen demonstrated the oscillatory character of the discharge by a revolving mirror photograph of the spark, similar to the photo- graph shown in Fig. 3 of Chapter I. Since then all o" Thomson's equations have been submitted to careful tests and have been found to be accurate. CHAPTER VII MAXWELL'S THEORY. ELECTRIC WAVES. THE ELECTRO- MAGNETIC THEORY OF LIGHT IN the preceding chapter we have seen that when a condenser, in series with a self-inductance and resistance, is charged and allowed to discharge, the current obtained, if the resistance is not too large, will be oscillatory in character. In this arrangement of apparatus we have a mechanism that serves as the source of electric waves. In 1865 Maxwell predicted, .by mathematical reasoning based on some experiments of Faraday, that variable currents in a con- ductor produce electric waves in space, that these electric waves travel with the velocity of light, and that light itself consists of electric waves of extremely short wave lengths. While direct experimental verification of this theory by the actual discovery of electric waves did not come during Maxwell's lifetime, Max- well yet showed that his predictions were strongly supported by many of the known facts about electricity and light. Without the aid of mathematics it is difficult to follow the steps of Maxwell's reasoning, so that the discussion here given will undoubtedly appear to the reader to be inconclusive. In the next chapter we hope to remedy this defect of the theoretical discus- sion by a description of the actual experimental demonstration of the chief propositions of Maxwell's theory. In the derivation of his theory Maxwell makes use of the two facts about the relation of electricity to magnetism that we have given in Chapter III ; namely, I. An electric current in a conductor produces a magnetic field in the neighborhood of the conductor, and II. A variable magnetic field in the neighborhood of a con- ductor produces an electromotive force in the conductor. To these two well-known experimental facts Maxwell adds a third proposition in the form of an assumption, which has been called the displacement assumption. The Displacement Assumption. This assumption is an attempt on the part of Maxwell to give expression to the idea of Faraday, 36 THE ELECTROMAGNETIC THEORY OF LIGHT 37 that when a condenser is charged, the condition of things is not completely described by saying that a positive charge is given to one plate and a negative charge to the other plate of the condenser. Faraday showed that something takes place in the medium between the plates, and Maxwell makes the assumption that the action in the medium partakes somewhat of the nature of an electric current, although the medium is an insulating substance. It is difficult to determine just how Maxwell imagined this action to take place, and different writers have employed different mechanisms in the description of the current that Maxwell sup- posed to exist in the insulators. One way of representing his idea is to suppose that the insulating medium, whether a solid, liquid, or gaseous dielectric, or even empty space, is made up of small parts, and to suppose that the electricity in these small parts of the insulator may flow freely in the small parts but can- not flow from one part to the next. If we call these small parts molecules, we may describe the current in the insulating medium as the act of polarizing the molecules. That is, for ex- ample, when the left-hand plate of the condenser in Fig. 19 is charged positively, the positive electricity added to this plate attracts the nega- tive electricity and repels the positive electricity of the neigh- boring molecules, so that the part of each molecule near the plate becomes negative and the distant part becomes positive. Mole- cules in this condition are said to be polarized. The layer of molecules so polarized acts on the next layer and produces a similar polarization, so that in turn the molecules throughout the medium between the plates become polarized. It is seen that this general transfer of positive electricity to the right and negative electricity to the left in the molecules would have an effect similar to an electric current flowing from the positive plate to the negative through the insulator. Maxwell called this general transfer of electricity in the dielectric a displacement current. During the charging of the condenser, the displacement current is in the FIG. 19. Illustrating displacement current. 38 WIRELESS TELEGRAPHY same direction as the current in the conducting parts of the circuit, so that the displacement current may be said to complete the con- duction current. During the discharge of the condenser the dielec- tric loses its polarity, and according to Maxwell's view, gives rise to a displacement current in the dielectric. In this case, also, the dis- placement current completes the conduction current, which is now flowing away from the positive plate of the condenser. It has been stated above that the displacement current partakes of the nature of an electric current. The displacement current differs from the ordinary current in that there is within the molecules nothing corresponding to ordinary resistance, so that none of the energy of the displacement current is converted into heat. The displacement current also differs from the conduction current in that the displace- ment current, under a given applied electromotive force, sets up a restoring force in the dielectric which, like the reaction of a com- pressed spring, soon becomes large enough to equalize the electro- motive force and stop the current; whereas the conduction current in a circuit that is wholly conductive continues to flow as long as the electromotive force is applied to the circuit. These are the differences between the displacement current and the ordinary current. On the other hand, according to Maxwell's theory, the displacement current is exactly like an ordinary electric current in respect to its relation to the magnetic field. We may thus add to the two propositions stated on p. 36, the proposition III. In the case of a circuit not entirely closed by conducting parts, the current in the conducting parts is completed by a dis- placement current through the dielectric. This displacement' cur- rent produces a magnetic field in its neighborhood; and a variable magnetic field in a dielectric produces displacement currents in the dielectric. Electric Waves. In Maxwell's treatise the propositions I, II and III are discussed quantitatively, with the result that he obtains a number of quantitative relations about light and electricity. How- ever, without such a mathematical discussion we may be able to see how the facts assumed to be correct in proposition III lead to the idea of electric waves in the dielectric. For this purpose let us suppose that we have two conducting bodies of the form shown in Fig. 20. A and B are two metallic rods with a small spark gap between. Suppose now that A is charged with electricity of one sign, and B with electricity of the other sign, and suppose the charges are gradually increased until a spark THE ELECTROMAGNETIC THEORY OF LIGHT 39 passes between them. If the resistance is not too large, the current that flows will be oscillatory, because the rods have electrostatic capacity and self-inductance. The two metallic rods here pictured constitute an electric " oscillator." According to Maxwell's theory, the oscillatory currents in the oscillator will be completed by displacement currents in surrounding space. A part of this displacement current takes place along the black loops in the direction of the arrows from one end of the oscil- lator around to the other. The displacement loops are really sec- tions of a sheet such as would be obtained if we rotated the figure FIG. 20. Displacement current and magnetic force. about the oscillator as an axis. These displacement currents in the sheet will reverse their direction when the current in the oscillator reverses, and are accompanied by a magnetic field of which a single line is shown encircling the displacement sheet. The magnetic field produced by the displacement current in the shaded region, being oscillatory in character, will induce displacement currents in a portion of the medium farther out from the oscillator, and the latter current will lag somewhat behind the former. Thus, a sheet corresponding to the shaded region will sustain a displacement current oscillating with the period of the oscillator. The unshaded region farther out will sustain similar oscillations a little later, so that we have the condition of things that exists in a wave motion traveling with a finite velocity; namely, a series of disturbances first in one direction, then in the opposite direction, taking place all over a closed surface, and traveling outward from the source. 40 WIRELESS TELEGRAPHY Properties of the Electric Waves. A masterly mathematical treatment by Maxwell of this idea of an electric displacement in dielectric media led not only to great prog- ress in the knowledge of electromagnetism, but also to a complete revision of theories as to the nature of light, so that now all the phenomena of optics are describable in terms of Maxwell's electric waves. From his theory Maxwell deduced the following facts in regard to electric waves: 1. The electric wave in the dielectric consists of a displacement current in one direction with a magnetic force at right angles to it, both of these quantities being in the wave front; that is to say, at right Electric force E angles to the direction of propagation of Thus electric waves, like light waves, are transverse waves. 2. The velocity of propagation of the FIG. 21. and magnetic force M ,, / 91 x perpendicular to direc- the wave (se Fig. 21). tion of propagation T. \JU electric waves (in a non-magnetic insulating medium) is - , where a is the ratio of the c. g. s. electromagnetic unit of quantity to the c. g. s. electrostatic unit of quantity, 1 and k the dielectric constant of the medium. In empty space, by definition, k is unity, and the ratio a was known from older experiments to be the velocity of light (3 X 10 10 cm. per second) ; whence the velocity of the electric waves in free space is identical with the velocity of light, which is 3 X 10 10 cm., or about 186,000 miles (seven times around the earth) in one second. 3. In an insulating medium other than free space (for example, in glass or paraffin) it is seen from the preceding section that the velocity of the electric waves is in which VQ is the velocity of waves in free space, and v the velocity of the waves in a dielectric of dielectric constant k; whence, v /v = Vk (9) That is to say, the index of refraction 2 of a medium for electric waves is equal to the square root of the dielectric constant of the medium. 1 For definitions of these units see Appendix I. 2 The index of refraction is the ratio VQ/V. THE ELECTROMAGNETIC THEORY OF LIGHT 41 4. All good conductors are opaque to electric waves, all good insu- lators are transparent to electric waves, and semiconductors like wood and stone are semitransparent. Metallic surfaces are prac- tically perfect reflectors of electric waves. The Electromagnetic Theory of Light. Among these several properties of electric waves the properties stated in 1 and 2 are identically true of electric waves and light; while the properties enu- merated in 3 and 4 have also met with very useful application to light as well as to longer electric waves. Thus Maxwell came to the conclusion that light waves are electric waves of short wave length. This theory is now generally accepted. It is interesting to note, on this theory, how light can be produced. We have seen how electric waves may be produced by oscillating electric currents in a circuit of the form shown in Fig. 20. Now if we suppose the oscillator of Fig. 20 to be made smaller and smaller, the capacity and inductance will both be decreased, and the time of oscillation is thereby decreased. If then we think of the oscillator as possessing atomic dimensions, the period of oscillation approaches that of light. It is, however, not necessary to think of an actual electric discharge taking place between the atoms of our atomic oscillator, because the rapid vibratory motion of a single charged particle, or electron, back and forth would have the same effect as an electric discharge between particles, and would produce electric waves of which the period, for a particular size and velocity of the vibrating particle, would be the period of light of some particular color. Let us turn next to the experimental demonstration of the exist- ence of the electrical waves predicted by Maxwell. This did not come during Maxwell's lifetime; in fact, twenty-two years elapsed between Maxwell's remarkably clear presentation of the theory and Hertz's brilliant confirmation of it. CHAPTER VIII THE EXPERIMENTS OF HERTZ first direct experimental confirmation of Maxwell's theory of electric waves was made by Professor Heinrich Hertz 1 of Karlsruhe in 1888. At Karlsruhe, and later at Bonn, Hertz performed a great number of experiments, in which he produced and detected electric waves; measured the wave length; showed that the electric waves were transverse, polarized waves; that they were capable of reflec- tion from metallic surfaces and were freely transmitted through insulators; that they could be refracted by prisms of pitch and other dielectrics; and that as the wave length of the electric waves was shortened, the electric waves showed properties more and more analogous to the properties of light. Lodge's Resonance Experiment. Prior to the work of Hertz, Sir Oliver Lodge 2 in England had made some experiments on the inductive action between Ley den-jar circuits which were a close approach to the discovery of electric waves. A description of these experiments will aid us to understand Hertz's apparatus. Lodge employed two circuits of the form shown in Fig. 22. The Leyden jar A had its two coatings connected with an electric machine, so that when the machine was operated, the jar was charged, and when the tension of the charge reached a certain value, a discharge occurred through the loop BCD and across the spark gap S. This discharge was oscillatory and acted inductively upon a second circuit A'B'C'D' placed parallel to the first. The second circuit was provided also with a spark gap at S', which was formed by a strip of metal folded over the jar so as to touch the inner coating and come near the outer coating as S'. This circuit, which we shall call the receiving circuit, had its period of oscillation variable in that the inductance of the circuit could be changed by the movable slider at C'D'. When sparks were passing in the discharge circuit, Lodge found that there was a certain position of the slider C'D' that gave a maximum effect at the receiving circuit, as was shown by the lively passage of sparks 1 Electric Waves, translated by D. E. Jones, Macmillan & Co., 1893. 2 Lodge: Report British Association, Vol. 50, p. 567, 1888. 42 THE EXPERIMENTS OF HERTZ 43 across the spark gap at S'. The two circuits were then in resonance; that is to say, they had the same period of oscillation as determined FIG. 22. Sir Oliver Lodge's resonant Leyden jars. by the formula T = 2 TiVLC. The oscillatory current in the dis- charge circuit induced an electromotive force in the receiving circuit, and when the circuits were in resonance, this induced electromotive force was capable of forcing sparks across the gap at S f , even when the two circuits were several meters apart. According to Maxwell's theory, the inductive action between the two circuits consisted of electric waves sent out from the discharge circuit and striking the receiving circuit; but Lodge was not able to demonstrate the existence of these waves. To do this it was neces- sary to make the wave length shorter and the radiation freer than that produced by Lodge's discharge circuit. Hertz's Experiments with Electric Waves in Air. In order to produce shorter waves than those employed by Lodge, Hertz made use of a discharge system with smaller capacity and self-inductance. One form of Hertz's " oscillator " is shown in Fig. 23. It consists of two flat metallic plates, 40 cm. square, each attached to a rod 30 cm. long. The two rods were placed in the same line, and were provided at their nearer ends with balls separated by a spark gap about 7 mm. long. The oscillator was charged from the secondary of a Ruhmkorff coil J attached to the rods near the spark gap. The 44 WIRELESS TELEGRAPHY primary of the coil was fed by a battery, and contained a vibrator for interrupting the primary current so as to produce a high poten- tial in the secondary. At each interruption by the vibrator in the primary, the two halves of the oscillator became charged, and dis- 1 FIG. 23. Hertz oscillator. charged in an oscillatory manner across the spark gap of the oscil- lator. At each spark, according to Maxwell's theory, there was sent out a train of waves from the oscillator. In order to detect these waves, Hertz employed a receiving circuit, Of FIG. 24. Hertz's circular resonator. FIG. 25. Hertz's apparatus for showing the existence of electric waves in air. now generally called a " resonator/' of the form shown in Fig. 24, which is seen to consist of a circular loop of wire broken by a diminu- tive air gap at X. The radius of the loop was 35 cm., which was found by experiment to be the proper size to be in resonance with the oscillator. THE EXPERIMENTS OF HERTZ 45 To demonstrate the existence of the electric waves Hertz made use of the phenomenon of interference. The arrangement of apparatus is shown in Fig. 25. M is a metallic reflector, consisting of a sheet of zinc, 2. meters wide by 4 meters high, from which the waves sent out by the oscillator are reflected. The reflected waves superimpose upon the direct waves, producing in the region between the oscillator and the metallic reflector certain positions where the direct and the reflected waves neutralize each other and certain other positions in which their effects add. In demonstrating these effects Hertz per- formed a number of beautiful experiments. In one experiment the plane of the resonator was kept parallel to the reflector, with the spark gap at the side, as shown in Fig. 25. Then wherever the resonator may be placed along the line SNi, the electric force F and F' is the same at the two sides of the resonator. But the force F', being applied to a completely metallic part of the loop, acts to a greater advantage l than the force F, so that sparks are produced unless both F and F' are very small. With this orien- tation of the resonator, Hertz started with the resonator at N\ close to the reflector and moved it gradually away toward the oscillator. In the position Ni there were no sparks in the resonator, showing that there is a node of electric force at the reflector. This result is consistent with the fact that a large difference of potential cannot be set up in the surface of a good conductor. As the resonator is moved away from the reflector, sparking begins in the resonator, becomes more and more lively, until a maximum is reached at LI. This position LI, is called a loop of electric force. On proceed- ing further in the same direction, a second minimum of sparking is found at Nz, and so forth. Discussion of this Experiment. The occurrence of maxima and minima in the region between the reflector and the oscillator is evidence of the undulatory nature of the disturbance, and the distance NiN* or LiL 2 , is the half wave length. To make this proposition clear, reference is made to Fig. 26, which shows several drawings of the direct and the reflected wave and the resultant obtained by their superposition. The reflecting mirror is repre- sented by the heavy vertical line at the right. The undulating line, made up of dashes, represents the direct wave, which is moving toward the reflector; and the dotted wavy line is the reflected wave, moving from the reflector. The heavy line in the 1 In the same way that plucking a violin string at the middle will produce a greater motion than plucking it near the end. 46 WIRELESS TELEGRAPHY diagram is the resultant effect obtained by adding the two waves. The distance from one crest to the next similar crest (C 3 to Ci) is called the wave length, and the time for the wave to move one wave length is called the period T. The different diagrams, (a), (b), (c), (d), (e), (/), (g), (h), (i), show the conditions that exist in the region between the oscillator and the mirror at different times. At a time that we have called t = 0, as represented in diagram (a), the direct and the reflected waves are exactly opposed to each \ \ v /%' (a) t=0 d) FIG. 26. Showing superposition of direct and reflected waves. other throughout the space between the oscillator and the reflector, so that the resultant electric force is everywhere zero. In (6), t = 778, the direct wave has moved nearer to the mirror by a dis- tance equal to i NiLi(= i wave length), while the reflected wave, which moves with the same velocity, has moved from the mirror by an equal amount. It is seen that now the direct and the reflected waves do not oppose each other everywhere in the region. In some parts of the region, e.g., at _Ni, N%, N 3 , they do oppose and THE EXPERIMENTS OF HERTZ 47 neutralize each other, while at other points their intensities add. At Li, L 2 and L 3 the added intensities give a resultant about 1.4 times the maximum of either wave alone. In (c), t = 2 T/8, the direct wave has approached the mirror by another eighth of a wave length, the reflected wave has receded (f) h) FIG. 26 (Continued). from the mirror by an equal amount, and the two waves exactly superpose. The resultant intensity of electric force is still zero at Ni, Nz t N 3 and N*, while at LI, L 2 and L 3 the intensity is double that of either wave separately. In a -similar manner th? remaining drawings (d), (e), (/), (g), (h), (i) represent the progress of the direct wave toward the mirror and the recession of the reflected wave from the mirror by successive eighths of a wave length. The resultant intensity is always zero 48 WIRELESS TELEGRAPHY at Ni, NZJ N 3 and 7V 4 , while, on the other hand, if we pass down the figure from the diagram (a) to (h), we see that the intensity at LI, L 2 and L 3 begins at zero (a), rises to double the intensity of the single wave (c), falls to zero (e), then to minus the double intensity (g)j and finally, at the expiration of a time equal to T, rises again to zero (i) . To show this more clearly, all the resultants are collected in the last diagram (r) of the figure. The positions Ni, N%, N s and JV 4 are the positions in which the resonator of Hertz gave no sparks at the detecting spark gap, because the electric force at these positions is t constantly zero. The positions LI, L 2 and L 3 are places where the sparking at the resonator was a maximum, because at these places the electric force fluctuates up and down during each period of the wave. The distance from Ni to N 2 or from LI to L 2 is half the distance from C 3 to Ci, diagram (a), and is therefore equal to half the wave length. With the dimensions of apparatus used by Hertz in the experiment represented in Fig. 25, this half wave length was 4.8 meters. The set of drawings given in Fig. 26 represents the conditions that exist in a " stationary wave system," in which the direct and the reflected wave are both moving, while the interference between these two waves gives a set of maxima and minima fixed in space. The minima are positions where there is never any resultant force, while at the maxima the force fluctuates between positive and negative maxima, with a period equal to T, the period of the waves. The conditions assumed in the drawings given in Fig. 26 are somewhat simpler than the conditions actually occurring in Hertz's experiment, because the direct and the reflected waves in the case represented in the drawings are supposed to have the same ampli- tude, whereas in the actual experiments the. reflected wave is weaker than the direct wave, so that JVi, Nz, N 3 and N 4 are not positions of zero intensity, but yet have intensity small enough to enable them to be located by the experiment. Nature of the Wave. The experiment by Hertz, just described, shows that the disturbance sent out from the oscillator and detected by the resonator travels as a train of waves. To give the reader an idea of the nature of this wave motion reference is made to the diagram of Fig. 27, which shows in part the electric field about the oscillator at a particular moment. This figure is a simplification of a diagram theoretically obtained by Hertz from Maxwell's equations. The oscillator is shown in the center of the diagram, and on THE EXPERIMENTS OF HERTZ 49 either side of the oscillator are shown the lines along which Max- well's displacement currents occur. These lines are called lines of electric induction. We have seen in Chapter VII how we can imagine the displacement current in the dielectric to complete the conduction current in the oscillator. In that case the lines of elec- tric induction terminate on a positive and a negative charge at their two ends. At the instant represented in the diagram, the two halves of the oscillator have opposite charges, and some of the lines of electric induction near the oscillator terminate upon the charges on the oscillator. But a little farther out from the oscil- lator the lines in the diagram are represented as closed upon them- selves. This closing of a loop on itself occurs when the positive and the negative charges on the oscillator come together as the current in the oscillator reverses. The closed loops represented in the diagram have been produced by successive oscillations of the current on the oscilla- tor, and have been liber- ated from the oscillator and are moving freely away. The condition of things in the space around the oscillator in action may be pictured to the mind by supposing that these closed loops of electric induction move away from the oscillator, and as they move they elongate and grow less intense. Their width, however, remains constant, so that if a receiver be placed in any fixed position, say in the equatorial plane, PP, the inductive action of the loops, as they successively pass, changes continuously from one direction to the other with a period equal to that of the oscillator. This train of continuously reversing electrostatic induction is one aspect of the electric-wave train. Another aspect of the electric wave train may be discovered by examining the magnetic field about the oscillator. The lines of magnetic force about the oscillator are circles in a plane perpen- dicular to the oscillator, and these lines in a non-magnetic medium are everywhere perpendicular to the lines of electric induction, so FIG. 27. Simplified diagram of electric force about an oscillator. 50 WIRELESS TELEGRAPHY that the receiving circuit, placed, for example, in the equatorial plane, experiences also a series of continuously varying magnetic forces which tend to induce an electromotive force in the receiving circuit in the same direction as that induced by the electric induc- tion, so that both the electric and the magnetic effects act together and are called the components of the electric wave. One compo- nent is electric induction, which is in the plane of the oscillator. The other component is magnetic force, which is perpendicular to the electric induction. Both of these components are perpendicu- lar to the direction of propagation of the wave; that is to say, the wave is transverse. Attempt to Determine the Velocity of the Wave in Air. Hertz attempted to determine the velocity of the wave. We have seen that he found the wave length, X, to be 9.6 meters. This is the dis- tance traveled by the wave during the period of one oscillation of the current in the oscillator, so that, if we knew the period, T, we could calculate the velocity by the equation v = \JT. Hertz attempted to obtain the period, T, of the oscillator by calculation from such formulas as could be had for oscillators of this shape, and he ob- tained the period of complete oscillation to be 2.8 hundred-millionths of a second. This gave for the velocity of the waves the value 340,000 kilometers per second. In this calculation, as Professor H. Poincare pointed out, t Hertz made an error, and overestimated the period in the ratio of V2 : 1, so that, with this correction, he would have obtained the velocity of the waves to be 480,000 kilometers, while the velocity of light is 300,000 kilometers per second. This apparent discrepancy between the experiment and Maxwell's theoretical conclusion, that the velocity of the waves is equal to the velocity of light, was due, as Hertz suggested, to the inapplicability of the formula used in the calculation of the period of oscillation. Experiments which we shall soon come to discuss show that the velocity of the electric waves is the same as the velocity of light, and thus confirm Maxwell's predictions. CHAPTER IX EXPERIMENTS ON THE IDENTITY OF ELECTRIC WAVES AND LIGHT Hertz's Apparatus for Shorter Electric Waves. After Hertz had succeeded in proving that the action of an electric oscillation spreads out as a wave into space, he planned experiments with the object of concentrating this action and making it perceptible to greater distances, by putting the oscillator in the focal line of a large con- cave cylindrical mirror. In order to avoid the disproportion between the length of the waves and the dimensions he was able to give to the O o FIG. 28. Hertz's rec- tilinear oscillator. FIG. 29. Hertz's cylindrical mirrors. Oscillator is at left; resonator, at right. mirror, Hertz made the oscillator smaller, so that the length of the waves was less than one-tenth of those first discovered. The form of oscillator used in these experiments is shown in Fig. 28. The two halves of the oscillator were cylindrical bodies 3 cm. in diameter, terminating in spheres 4 cm. in diameter. The total length of the oscillator was 26 cm., and the spark gap was usually about 3 mm. For a receiving circuit, the circle of wire used in the previous experiments was replaced by a linear resonator, consisting of two straight pieces of wire, each 50 cm. long and 5 mm. in diameter, adjusted in a straight line so that their near ends were 5 cm. apart. 51 52 WIRELESS TELEGRAPHY From these ends two wires, 15 cm. long and 1 mm. in diameter, were carried away parallel to each other to a micrometer spark gap simi- lar to that used for indicating the waves in the previous experiments. The method of mounting the oscillator and resonator in the focal line of the cylindrical mirrors is shown in Fig. 29. The reflecting surface of the cylindrical mirrors was of thin sheet metal. The dimensions of the reflectors are shown in the diagram. With these reflectors about the oscillator and the resonator Hertz was able to get indications of waves up to a distance of 20 meters. The length of the wave, measured by the method of the last experiment, was 66 cm., and the period of oscillation, assuming that the waves travel with the velocity of light, was 2.2 thousandths of a millionth of a second. With this wave length Hertz succeeded in carrying out many of the elementary experiments that are commonly performed with light. Rays and Shadows. With the electric waves, as with light and radiant heat, shadows may be cast by objects opaque to the waves. FIG. 30. Plan of oscillator, receiver and metallic screens. Hertz found that a metallic screen interposed between the oscillator and the receiver, in the position A, Fig. 30, stopped the sparking of THE IDENTITY OF ELECTRIC WAVES AND LIGHT 53 the resonator completely, while the two screens in the position B and B' did not materially diminish the sparks at the resonator. If, how- ever, the opening between B and B' was made narrower, the sparks became weaker, and disappeared when the opening was reduced below a half meter. In experiments of this kind, although the dimensions of the screens are measured in meters, these screens are yet not large in comparison with the wave length of the waves, and the phenomena of diffraction are very marked, so that there is no sharp geometrical limit either to the rays or to the shadows. Polarization. Hertz showed that the electric waves produced by his linear oscillator are polarized waves. One way employed by him for showing this was to start with the focal lines of the two reflec- tors parallel, as in Fig. 29, so that there is lively sparking at the FIG. 31. Showing polarization by the absence of effects when the oscillator and the resonator are at right angles to each other. resonator, and turn the receiving mirror about the line joining oscil- lator and resonator. During this operation the resonator sparks become more and more feeble, and when the two focal lines are at right angles, as in Fig. 31, no sparks whatever are obtained at the resonator, even when the two mirrors are moved up close to each other. In another method of showing that the electric waves are polarized, Hertz made use of a grating of wires. The wires of the grating were 1 mm. in diameter and 3 cm. apart, and were mounted in an octagonal wood frame 2 meters high and 2 meters long. When the grating was interposed between the oscillator and the resonator so that the direction of the wires of the grating was perpendicular to the oscillator and the resonator, as shown in Position 1, Fig. 32, the screen practically did not interfere at all with the sparks at the resonator. But if the screen was set up in such a way that its wires 54 WIRELESS TELEGRAPHY were parallel to the oscillator and the resonator (Position 2, Fig. 32) it stopped the rays completely. With regard, then, to the transmis- sion of energy the screen behaves toward the electric waves as a tourmaline plate behaves toward a plane polarized ray of light. Another way of showing polarization of the electric waves was also devised by Hertz. The receiver was again placed so that its Position .1 Transparent FIG. 32. Position 2 Opaque Polarization proved by the interposition of a grating of wires. focal Jine was perpendicular to that of the oscillator, as in Fig. 31. Under these circumstances, as already mentioned, no sparks appeared. Nor were any sparks produced when the screen was interposed in the path of the waves, so long as the wires of the screen were either horizontal or vertical. But if the frame was set up in such a position that the wires were inclined at 45 to the horizontal on either side (see Fig. 33), then the interposition of the screen immediately pro- duced sparks at the resonator spark gap. Clearly the screen resolves the electric force of the advancing wave into two components, and transmits only that component which is perpendicular to the direc- THE IDENTITY OF ELECTRIC WAVES AND LIGHT 55 tion of its wires. This component is inclined at 45 to the axis of the receiver, and so has a component along the direction of the resonator. From these experiments it is evident that the interposition of the screen stops the waves when the wires of the screen are parallel to FIG. 33. Rotation of plane of polarization by a wire grating at 45. the electric component of the waves. It is in this position that the electric force would produce currents in the wires. The changing magnetic force at right angles to the wires would also produce cur- rents in the wires, so that both the components, that is to say, the whole electric wave, would be absorbed or reflected. Hertz showed that the action was one of reflection rather than of absorption; in this the wire screen differs from the action of the tourmaline crystal on light, for the extinguished component in that case is absorbed rather than reflected. Refraction. Hertz also performed some experiments on the refraction of electric waves, employing for the purpose a large prism 30 C FIG. 34. Showing refraction of electric waves by prism. of pitch cast in -a wooden box. The base of the prism was an isos- celes triangle 1.2 meters on the side, and with a refracting angle of 56 WIRELESS TELEGRAPHY nearly 30. The height of the prism was 1.5 meters, and its weight was 1200 pounds. With the arrangement of apparatus as shown in Fig. 34 the rays were refracted by the prism through an angle of 22. From this value Hertz calculated the index of refraction of the pitch to be 1.69, while the refractive index of pitch-like materials for light is given as being between 1.5 and 1.6. In concluding this series of experiments Hertz says: " We have applied the term rays of electric force to the phenomena which we have investigated. We may perhaps further designate them as rays of light of very great wave length. The experiments described appear to me, at any rate, eminently adapted to remove any doubt as to the identity of light, radiant heat, and electromagnetic wave motion. I believe that from now on we shall have greater confidence in making use of the advantages which this identity enables us to derive both in the study of optics and of electricity." Experiments of Righi. Immediately following the discovery of electric waves by Hertz, a great number of experiments were made M FIG. 35. Professor Righi's oscillator for short electric waves. FIG. 36. Righi's resonator. FIG. 37. Mounting of Righi's resonator. by various investigators in repetition of Hertz's expermients and in the effort to extend his results, particularly in the direction of the study of the properties of short electric waves, so as to obtain a further comparison of their properties with the properties of light. In order to obtain electric waves shorter than those of Hertz, Pro- fessor Righi 1 of the University of Bologna devised an oscillator consisting of two spheres (B, C, Fig. 35) separated by a small spark gap in oil. A and D are the terminals of an induction coil or electric 1 Augusto Righi: L' ottica delle Oscillazioni Elettriche, Bologna, 1897. THE IDENTITY OF ELECTRIC WAVES AND LIGHT 57 machine used to charge the oscillator. These terminals are provided with the spheres A and D, which are separated from the spheres B and C of the oscillator by spark gaps in air, so that the oscillator BC is without metallic connection with the other parts of the circuit. The spheres B and C were fastened with shellac into the truncated cones of glass EF and GH, which were supported in an ebonite frame. The lower funnel-shaped glass vessel served to contain the oil. The spark length in oil between B and C could be regulated by the screw V. The advantage of having the spark between the spheres take place in oil instead of in air, as had already been pointed out by MM. Sarasin and De la Rive, arises from the fact that it takes a greater difference of potential to start a given length of spark and therefore gives a more energetic discharge. When the spark is once started, the oil is carbonized and becomes conducting, so that the succeed- ing oscillations pass with comparatively little damping. Also the oil obviates the necessity of repeatedly polishing the terminals, as Hertz found he had to do when he attempted to get short waves with the spark in air. Righi found that vaseline oil is especially well adapted for use with his oscillator. For a receiving apparatus Righi made use of a resonator consisting of a strip of silver AB deposited on glass and interrupted by a diamond scratch C across the middle of the strip. This provided an extremely short spark gap between the two parts of the resonator, as shown in Fig. 36. Also the spark across this small gap will occur more easily than a spark of equal length in free air. 1 Righi's reso- nator is thus seen to be an extremely sensitive modification of the rectilinear resonator used by Hertz. In most of Righi's experiments the oscillator and the resonator were mounted in cylindrical reflectors. The mounting of the resonator is shown in section in Fig. 37. The resonator is at A, and is fastened upon a strip of ebonite BC. The observer looks through the con- verging lens at H, which serves to magnify the minute sparks between the two halves of the resonator. The apparatus could be used quan- titatively by observing the angle through which it was necessary to turn the resonator and its reflector in the support LM in order to extinguish the sparks. The angle of turning was indicated by -the pointer N moving over a graduated circle OP. 1 The author has shown that the potential required to start a spark along a surface of glass is about .44 of the potential to start a spark of equal length in free air. (Pierce: Physical Review, Vol. 2, p. 99, 1894.) 58 WIRELESS TELEGRAPHY The following table gives the dimensions of Righi's apparatus and the corresponding wave lengths obtained : Denomination of the appa- ratus. Oscillators. Resonators. Wave length in cm. Diameter of the spheres. Length in cm. Width in cm. I. .8 0.9 .1 2.6 II. 3.75 3.6 .2 10.6 III. 8.0 10. .2 20. 3.6 .6 11.8 10. .6 21.4 In a test of the sensitiveness of various combinations of this appa- ratus, Righi found that with the resonator III and the oscillator II both armed with their respective cylindrical reflectors, sparks appeared across the minute diamond scratch of the resonator when it was at a distance of 25 meters from the oscillator. This is a distance of 125 times the wave length for this apparatus. With the oscillator III, the sparks were evident at a greater distance. With resonator II and oscillator II, the greatest distance to which indica- tions of the waves could be obtained was 20 meters, which is 190 times the wave length. While with the minute apparatus, resonator I and oscillator I, the maximum distance was about 80 centimeters, which is 31 wave lengths. With this smaller apparatus, in spite of the comparative feebleness of the waves, many experiments that are commonly performed with light waves could be successfully carried out with the electric waves. For example, a small coin (10 centes- imi) can be used to reflect the waves. The coin does not need to be polished as with experiments on the reflection of light, because irregu- larities of the surface of the coin are too small to have any effect on -the reflection of the electric waves. Refraction and total internal reflection of these short waves could be shown with prisms of sulphur or paraffin that were very little larger than the glass prisms used in optics. Righi also succeeded in demonstrating the double refraction and elliptic polarization of the waves by slabs of the wood of the fir tree. The Use of a Thermal Junction for Measuring Electric Waves. In the experiments of Hertz and Righi the presence of the electric waves was manifested by the production of sparks across a minute spark gap between two parts of the receiving conductor. In 1892 THE IDENTITY OF ELECTRIC WAVES AND LIGHT 59 Ignaz Klemencic l showed that a thermal junction could be employed to detect and measure the waves. Klemencic's device, Fig. 38, con- sists of two thin sheets of brass MM, 10 cm. broad and 30 cm. long, placed 3 cm. apart, and having soldered to them respectively a very fine platinum and a very fine platinum-nickel wire, which were crossed at k and were thence conveyed off at right angles and soldered at their other ends to the leads I, I of a sensitive galvanometer. This resonating system was fixed at the focal line of a suitable cylindrical metallic reflector. When electric waves, with the electric force parallel to MM , fall on this receiver, electric oscillations between M and M produce heating of the knot k, which is the FIG. 38. Resonator employing thermal junction. FIG. 39. Oscillator for very short electric waves. point of contact of two dissimilar metals, and in consequence the heat developed gives rise to a thermoelectromotive force at the knot and consequently to a current in the galvanometer. By the use of this instrument and a Righi oscillator, Klemencic has studied the reflection of electric waves from metals and insulators. Various investigators have made use of the Klemencic thermal junction in quantitative experiments on electric waves. By reducing the size of the metal vanes MM, Professor A. D. Cole 2 has applied the apparatus to measurements with waves with a wave length of 4 cm. Professor Lebedew, 3 employing a slightly different form of 1 Ignaz Klemenctf: Wied. Ann., 45, p. 62, 1892. 2 A. D. Cole: Wied. Ann., 57, p. 290, 1896, and Phys. Review, 7, Nov., 1898. 3 Peter Lebedew: Wied. Ann., 56, p. 1, 1895. 60 WIRELESS TELEGRAPHY thermal junction, worked with waves of wave length of only 6 mm., and succeeded in showing the double refraction of electric waves by crystals. A form of oscillator similar to that used by Cole and by Lebedew for producing their short electric waves is shown at o, o, Fig. 39. Professor Lampa and Professor Bose have also succeeded in mak- ing measurements with electric waves of only 6 mm. wave length. Wave Length of Electric Waves and Light. The following table contains in round numbers the value of wave length and number of vibrations per second of some electric waves and waves of radiant heat and light: Electric waves produced by Wave length in cm. Number of vibrations per second. Commercial Alternating Current 200,000,000 150 Leyden Jar Discharge, Feddersen .... Hertz's First Oscillator 300,000 1,000 100,000 30,000,000 Hertz's Rectilinear Oscillator 60 500,000,000 Righi's Oscillator 2.6 11,000,000,000 Lebedew, Lampa, and Bose's El. Waves Longest Radiant Heat .6 .01 50,000,000,000 3,000,000,000,000 Orange-colored Light .00006 500,000,000,000,000 Shortest Ultra-violet, Schumann, Lyman .00001 3,000,000,000,000,000 Physicists have long been accustomed to recognize that the dif- ference between radiant heat, visible light, and the actinic ultra-violet radiation is merely difference in wave length, and that our greater familiarity with the visible portion of the spectrum arises merely from the fact that we have a particular set of nerves sensitive to these rays. The visible part of the spectrum lies between wave lengths .000040 and .000076 centimeter. By the aid of the thermopile and the photographic plate the spectrum has been extended to include all the radiation with wave length between .00001 (extreme ultra-violet) and .01 centimeter (extreme infra-red). This upper limit is about 1000 times the lower limit. It is interesting to note that the Hertzian waves measured by Lebedew, Lampa, and Bose have a wave length only about 60 times the wave length of the limit attained in the infra-red. That is to say, the shortest Hertzian waves that have been measured are nearer in wave length to the longest measured heat waves than these are to the shortest measured ultra-violet. Also in properties the Hertzian waves are nearer to the long heat THE IDENTITY OF ELECTRIC WAVES AND LIGHT 61 radiations than these are to the ultra-violet or even to the visible. For example, some of the long heat waves, like the Hertzian waves, pass readily through vulcanite and other insulators opaque to visible light. Space is lacking to consider further the experimental evidence in favor of Maxwell's proposition that electric waves are of the same nature as light waves, and that the light waves are in fact simply electric waves of those particular wave lengths that possess the prop- erty of being capable of affecting the retina of the eye. CHAPTER X ON THE PROPAGATION OF ELECTRIC WAVES ON WIRES Wheatstone's Experiments. Early in the history of the electric telegraph the question arose as to the velocity of propagation of elec- tric disturbances along wires. The first attempt to measure this velo- city was made by Wheatstone 1 in 1834. Wheatstone attempted to measure the velocity of electricity in a circuit consisting of a copper wire about half a mile long, and extended back and forward so as to form twenty parallel lines, 15 cm. apart. Three spark gaps were inserted in this line, one at each end and one at the center. These were arranged horizontally, side by side, in front of a mirror mounted on a horizontal axis and capable of being revolved at the rate of 800 revolutions per second. Upon discharging a condenser through the two end spark gaps into the circuit, the image of all three of the sparks could be seen in the revolving mirror, and the image of the central spark was found to be displaced with reference to the other two, showing that the central spark occurred later than the two end sparks. The amount of the displacement of the* central spark, together with the speed of the mirror, furnished the data for computing the speed of propagation of the electric current. Wheatstone had difficulty in determining the amount of the displacement, which he could obtain only by eye observations. Computations from Wheatstone's observations seemed to show that an electric discharge traversed the copper wire at a speed of 288,000 miles (463,000 kilometers) per second, which is greater than the velocity of light; and this was long accepted as the true " velocity of electricity." While the numerical result obtained by Wheatstone is now known to be incorrect, the experiment is yet interesting in that it showed that time was required for the electrical disturbance to traverse the wire. The revolving mirror employed in this experiment has now become a classical apparatus in physical investigation. Other Early Experiments. In 1850 Fizeau and Gounelle like- wise made a series of experiments on the velocity of the electric 1 Wheatstone: Phil. Trans., Part II, p. 583, 1834; Pogg. Ann., 34, p. 464. 62 THE PROPAGATION OF ELECTRIC WAVES ON WIRES 63 current, and for this purpose availed themselves of the telegraph lines between Paris and Amiens (314 kilometers) and between Paris and Rouen (288 km.). Their measurements gave a velocity of 101,700 km. per second for iron wires, and 172,000 km. per second for copper wires. In other similar measurements of the apparent velocity of the electric current various results have been obtained in practice which are much lower than those of Wheatstone, and Fizeau and Gounelle, being in some cases 2240 kilometers per second, and in others 4800, 28,000, 96,000 and so on. What, then, is the explanation of this great variability in the experimental results ? Theoretical Discussion. In 1855, in discussing the feasibility of an Atlantic cable, Sir William Thomson gave a mathematical treat- ment of a case of the propagation of electric disturbances in con- ductors. In 1857 Kirchhoff, and in 1876, Heaviside, developed extended theoretical treatments of the problem. The results ob- tained by these mathematical physicists show that the velocity of propagation of electrical disturbances in conductors depends on the nature of the disturbance and the A relative values of the capacity, self -inductance and resistance of the conductor. If we have two long parallel FlG " 40 > J w . Parallel wires with & applied electromotive force, wires (Fig. 40) as in the case of land telegraph and telephone lines, or one wire in an insulating sheath submerged in a conducting body, as in the submarine cable, three important cases arise in practice. Case I. Telegraphy. If the self-induction of the line is negli- gible in comparison with its resistance and we have an electromotive force impressed on one end of the line, the current in the conductor grows in a manner described as "diffusion." Fig. 41 gives a set of curves 1 showing the difference of potential between the two conduc- tors at various positions along the line, at different times after the application of the electromotive force. In this case there is no proper velocity of the electricity; for at the instant the battery is applied some electricity appears all along the line, and the charge at a short distance from the origin grows faster than the charge at a greater distance. ' This is approximately the case that occurs in submarine 1 Redrawn from Professor A. G. Webster's Electricity and Magnetism; Macmillan, 1897. 64 WIRELESS TELEGRAPHY cabling, and Sir William Thomson showed that in the case of the proposed Atlantic cable, the time required for each signal would be sixteen times as long as the time for'a cable of the same cross section Distance FIG. 41. Diffusion of electric current in parallel wires with negligible inductance. with one-quarter of the length, such as then existed in the French submarine telegraph to Sardinia and Africa. The condition assumed by Sir William Thomson is only approxi- mately realized in practice, for in no line is the action of self-induction FIG. 42. Distance Modified diffusion. completely negligible. Especially is the action of the self-induction not negligible at the instant of applying the battery at A, Fig. 40, because this application of the battery is sudden, and for a sudden charging of the conductor the effect of the self-induction is greater THE PROPAGATION OF ELECTRIC WAVES ON WIRES 65 than for a slow application of the charge. For this reason the prop- agation of the disturbance is more accurately represented by the set of curves given in Fig. 42. In this diagram it is seen that the disturbance has a nearly square wave front, which, according to the theory, travels with the velocity of light, while succeeding parts of the impulse lag more and more behind the wave front. The square wave front itself becomes also more and more attenuated as the disturbance progresses along the wires. This same condition of things exists to some extent in the case of land telegraph lines, and accounts for the indefiniteness of the results that have been obtained in the attempt to measure the velocity of propagation. If for a particular length of line the apparatus used by the experimenter for detecting the wave is sufficiently sensitive to respond on the arrival of the wave front, the value obtained for the velocity is the velocity of light ; while with a greater length of line the wave front is too feeble to affect the instrument, which then responds to a more intense part of the wave arriving later, and hence gives a smaller value for the velocity. Case II. Telephoning. Suppose, now, that instead of simply ap- plying a battery to the line, as in telegraphing, we apply a telephonic electromotive force to the parallel wires of Fig. 40 or to the submarine cable. This telephonic electromotive force is an alter- nating electromotive force. Although the self-inductance and resist- ance of the circuit may be the same as before, the effect of the self-induction is larger in the telephonic case, because of the rapidity of the alternations of the electromotive force at the source. Under this condition Heaviside finds that the different waves generated by the sounds of different pitch travel with different velocities, and that this results in a distortion of the wave and puts a limit to the dis- tance to which the telephone can be used. This distortion is caused by the resistance and capacity of the line, and is partially eliminated by self-induction. Heaviside says that this " self-induction is the telephonist's best friend," for it tends to preserve the sharpness of the wave and to eliminate the part of the disturbance lagging behind the wave front. Heaviside pointed out that the addition of properly distributed self-induction was beneficial to prevent distortion in telephony; and in actual practice, by adding inductance coils at intervals along telephone lines, Professor Pupin has considerably increased the distance to which distinct speech may be transmitted. In the case of the submarine cable, on account of the relatively small value of the self-inductance, submarine telephony is not at OF THE UNIVERSITY 66 WIRELESS TELEGRAPHY present practicable to a greater distance than about twenty miles (32 kilometers). Case III. Electric Waves of High Frequency. As a third case, which is the one in which we are here chiefly interested, let us suppose that the electromotive force applied to the end of the two parallel wires oscillates with very great frequency. The effect of the resist- ance then becomes negligible in comparison with the effect of the self- indue tion. The theoretical treatment of this case shows that such a disturbance travels with the velocity of light, and except by a decrease of amplitude, the wave is not distorted during its progress along the wires. In what follows we shall see how experiments have confirmed this deduction from the theory. Hertz's Experiments with Waves on Wires. In 1888, a short time before his experiments with electric waves in air, which have been described in chapters VIII and IX, Hertz performed a series of FIG. 43. Hertz apparatus for waves on wires. experiments with electric waves on wires. The form of apparatus employed is shown in Fig. 43. At a short distance behind one of the plates A of the oscillator, a second plate P was placed. From P a copper wire was bent through the arc mn and thence led off horizon- tally. When the plate A is charged positively, negative electricity is attracted to the nearer side of the plate P, and an equivalent posi- tive charge is sent away along the wire. When the charge on A becomes negative, a similar negative charge moves away along the wire, so that during the oscillations between A and A', in which the charge on A changes continuously back and forth from positive to negative values, a train of positive and negative impulses, constitut- ing a tram of waves, travels out along the wire. The train of waves on the wire will be reflected from the end of the wire, as may be seen from the following reasoning. The current < THE PROPAGATION OF ELECTRIC WAVES ON WIRES 67 cannot flow past the end of the wire, nor does the electricity con- stituting the current merely flow out to the end of the wire and stop in a state of equilibrium. Two forces are acting on the current: (1) the accumulation of electricity near the end of the wire raises the potential of the wire and provides a force opposing the current; (2) the slowing down of the current causes change in the magnetic field surrounding the wire, and this tends to prevent the cessation of the current. These two forces do not act together, when one is a maximum, the other is a minimum. As a result first one and then the other of these forces will predominate, so that the charge will first be sent into the parts near the end of the wire by the magnetic field (self-induction) and will then be sent out again by the electrostatic rise of potential (reciprocal of capacity) . The effect of this is that the periodically arriving impulses will be sent back again with the same period, and we shall have, therefore, a direct and a reflected train of waves. The direct and the reflected waves will interfere with each other, so as to form a stationary system of waves like that obtained in the experiment with waves in air reflected from a sheet of metal (Chapter VIII). In this case, however, the end of the wire will be a loop of potential; whereas the metal reflector of the waves in air is a node of potential. There is also another difference; for in the case of the wire, the returning wave will again be reflected at P, and a simple stationary wave system can only be realized provided the horizontal wire has a proper length, which may be determined by experiment. Hertz studied the waves produced in the wire, with the aid of his circular resonator, shown in the figure. With the resonator in the vertical position C, Hertz was able to locate the nodes and loops of current in the wire by the absence or presence of sparks at the reso- nator. When, however, the resonator was placed in the horizontal position B, the effect obtained was due partly to the waves in the wires and partly to a linking with the resonator of magnetic lines directly from the oscillator. The compound effect obtained in the latter case was utilized by Hertz in a study of the interference between the waves in the wire and the waves in the air. He came to the con- clusion that the wave length, and consequently the velocity of prop- agation, was different in the two cases. This was in contradiction of Maxwell's theory. Later, by the use of a smaller oscillator at A A' , he found that the difference between the velocities of the waves on wires and in air very nearly disappeared. 68 WIRELESS TELEGRAPHY Experiments of Sarasin and De la Rive. While Hertz was puzzling over this problem, and attempting to explain the dis- crepancy between his experiment with the long waves, which did not agree with Maxwell's theory, and his experiment with the shorter waves, which did agree with the theory, MM. Sarasin and De la Rive at Geneva repeated the experiment with the longer waves in a room larger than that available to Hertz, and obtained from this case also approximately the same velocity for the waves on the wire and the waves in air. Hertz's difficulty probably arose from the disturbing influence of electric waves reflected from ob- jects in the room. Maxwell's proposition of the equality of the two velocities is strictly true only provided the waves on wires are produced on two parallel wires close together, a positive impulse being started along one of the wires and at the same time an equal negative impulse being started along the other wire. Introducing this precaution, numerous subsequent experimenters have con- firmed Maxwell's conclusion that the velocity of the electric waves in a pair of nonmagnetic, conducting wires is the same as the velocity of these waves in the dielectric surrounding the wires. Direct Determination of the Velocity of the Waves on Wires. - Blondlot, 1 Trowbridge and Duane, 2 and Saunders 3 have made direct experimental determinations of the velocity of electric waves on wires. In all of these experiments the method consisted in determining the wave" length X of the waves on the wires, and in determining independently the time of the oscillation T that produced the waves. The quotient obtained by dividing the wave length by the time of oscillation gives the velocity rs"* 0)j for the wave length is the distance traveled in the time of one oscillation, and dividing the distance traveled by the time re- quired to travel it gives the velocity. In all of the experiments the wave length X was determined by exploring the stationary- wave system on the wires by a method like that devised by Lecher (p. 70). Trowbridge and Duane and Saunders determined the period of oscillation T by spark photographs taken with the aid of the revolving mirror, while Blondlot determined the period by 1 Blondlot: Comptes Rendus, Vol. 117, p. 543, 1893. 2 Trowbridge and Duane: American Journal of Science, Vol. 49, p. 297, 1895. 3 Saunders: Physical Review, Vol. 4, p. 81, 1896. THE PROPAGATION OF ELECTRIC WAVES ON WIRES 69 a resonance method,* like that at the present day used in getting the wave length in a wireless telegraph antenna. The following results were obtained for the velocity of electric waves on wires: Observer. Velocity in kilo- meters per second. Blondlot ( 293,000 \ 298,000 Trowbridge and Duane. . . . j 298,800 I 300,300 {295,400 299,400 Saunders 299,800 299,800 299,500 299,900 The average of the best determination of the velocity of light is about 299,900 kilometers per second, with which the above determinations of the velocity of the electric waves on copper wires is in good agreement. Velocity of Electric Waves in Air. Although the velocity of the electric waves in air has not been determined by a direct method, the experiment of Sarasin and De la Rive showed that the velocity of the waves in air is the same as their velocity in copper wires surrounded by air, and therefore the same as that of light. Waves on Iron Wires. On account of the magnetic properties of iron, the velocity of the waves on small iron wires has been found to be slightly less than the velocity of waves of the same period on a nonmagnetic metal like copper. With wires J milli- meter hi diameter and with 115,000,000 oscillations per second, St. John found that the velocity on the iron wire was 4 to 5% less than the velocity on the copper wires. This result showed that the magnetization of the iron is able to follow extremely rapid reversals of the magnetizing current. On Surface Travel. In addition to this slight change in veloc- ity due to the magnetic property of the iron, the damping effect of the resistance of the iron is very large. In attempting to esti- mate the effect of resistance on the damping of oscillations of high frequency, it should be remembered that these rapid currents travel in a very thin film on the outside of the conductor. By TO WIRELESS TELEGRAPHY electrolytically coating an iron resonator with copper and a copper resonator with iron, Bjerknes found that when this coating was greater than a hundredth of a millimeter, the coated iron resonator acts like one of copper and the coated copper resonator like one of iron. This showed, in the case of electric oscillations of very high frequency, that the currents are confined to a shell whose thickness is of the order of a hundredth of a millimeter. The thickness of this shell depends, however, on the frequency of the oscillations, and on the radius and material of the conductor. (See Appendix II.) Waves on Wires Studied with a Vacuum Tube Detector. A form of apparatus devised by Professor Lecher for showing the existence of stationary waves on wires is shown in Fig. 44, which is T' FIG. 44. Lecher apparatus. a view of the apparatus from above. A' FA is an ordinary Hertz oscillator. Parallel to the plates A A' of the oscillator are placed two equal plates BB' connected to a pair of parallel horizontal wires. A bridge of wire, shown in a separate drawing at the right, and having an insulated handle, may be placed across the horizontal wires. A Geissler tube gg', which is pumped to a sensitive vacuum, is placed across the wires near their outer end, so that the glass of the tube rests on the wire. When the oscillator is in action, the Geissler tube will glow. Let us now put the bridge across the wires near the Geissler tube, the glow will cease, because it is short-circuited by the bridge. If now we move the bridge toward the oscillator, a position will be found, X X', for which the Geissler tube at the ends of the wires will again light up into a lively glow. A slight motion of the bridge in either direction from this position causes the glow to diminish. In explanation of this phenomenon we must think of the wires as divided into two circuits by the bridge. One of these circuits, which we will call the " oscillator circuit," is FABXX'B'A'F, comprising the two condensers AB THE PROPAGATION OF ELECTRIC WAVES ON WIRES 71 and A'B' and the spark gap F. This circuit has its own definite period of oscillation. The other circuit, which we will call the " resonator circuit," consists of the conductors gXX'g'. When the bridge XX' is in the position that causes the Geissler tube to glow, the oscillator circuit and the resonator circuit are in reso- nance, and during one complete oscillation the electric wave goes from the bridge out to g', back across the bridge, out to g, and back again to the bridge. Whence it is seen that the length of the conductor from g' across the bridge to g is the half wave length of the oscillator. If now the bridge is moved from XX' toward the oscillator, a second position SS f of the bridge is found for which the tube is caused to glow. During this displacement of the bridge, the self-inductance, and therefore the period, of the oscillator circuit is diminished, while the length of the wire to the right of the bridge is increased. Therefore, the wire to the right of the bridge cannot be in resonance, as a whole, with the oscillator circuit. We can show this experimentally, for if we leave the first bridge at SS' and place a second bridge across the wires, a position TT' can be found for which the presence of the second bridge does not affect the glow of the tube. A slight motion of the second bridge to the right or to the left diminishes the glow. The two positions SS' and TT' are called nodes of electric potential. In a similar way with longer parallel wires several nodes may be located. The free end of the wires is always a loop of potential, and other loops of potential exist halfway between the nodes. The presence of these nodes and loops at equal intervals along the parallel wires shows the existence of a stationary wave system similar to that discovered by Hertz in his experiments with electric waves in air. Blondlot's Apparatus. A modification of Lecher's apparatus made by Professor Blondlot is shown in Fig. 45. The two halves of the oscillator are here bent into semicircles, while the parallel wires lead out from a secondary circuit placed immediately be- neath the oscillator. The oscillator and the circular portion of the secondary are submerged in a glass vessel containing oil. Leads from the induction coil are brought into the oil and connected to the two sides of the spark gap, one connection being made directly at a and the other connection being through a small spark gap at b. In this form of apparatus the waves on the wires are produced by electromagnetic induction from the oscillator. 72 WIRELESS TELEGRAPHY Arons' Tube. A very beautiful method of demonstrating the presence of waves on parallel wires was devised by Professor Arons. The two parallel wires for the greater part of their length were inclosed in a glass tube from which the air could be pumped. When the proper degree of exhaustion is attained, and the wires W w FIG. 45. Blondlot apparatus. in the tube are made to take the place of the parallel wires WW in air in Blondlot 's apparatus (Fig. 45), a bright glow, as represented in Fig. 46, appears at intervals along the wires indicating the pres- ence of a large fluctua- r, tion of potential (loop) < ^ m > > a j^ >_ m ^ >^> - at the positions of glow. This beautiful apparatus ' of Professor Arons exhibits to the observer at a glance the whole character of the potential distribution in the system. Exploration by the Bolometer. In the place of the vacuum tube in experiments with waves on wires, Paalzow and Rubens 1 l|l have shown how to adapt the bolom- eter to this purpose, and to obtain with it striking quantitative results. The bolometer (Fig. 47) consists of an accurately balanced Wheatstone bridge, so arranged that the oscilla- tory current to be measured is made to pass through a fine wire EF con- stituting one arm of the bridge. FIG. 47. Bolometer. & M1 . . , This oscillating current heats the fine wire, thereby changing its resistance, which throws the bridge out of balance and produces a deflection of the galvanometer G. 1 Paalzow and Rubens, Wied. Ann., Vol. 37, p. 529. THE PROPAGATION OF ELECTRIC WAVES ON WIRES 73 In Paalzow and Rubens's arrangement of apparatus (Fig. 48), in order to avoid disturbing the waves on the wires PQRS, the leads to the bolometer were not connected directly to the wires under examination, but were connected inductively by a single turn around capillary glass tubes TT, sliding on these wires. The glass tubes TT act as diminutive Ley den jars with the horizontal wires inside the tube for one coating, and the turn of wire on the outside of each tube for the other coating. Variations of electric potential at a point inside the little tubes induce (by electrostatic action) alternating potential in the turns of wire outside and produce alternating currents through one arm of the bolometer bridge. Figure 48 shows a form of apparatus suitable for experiments with this method. This is the form of apparatus used by Professor FIG. 48. Exploration of waves on wires by bolometer. St. John. As has been before mentioned, in order to get a simple stationary wave system in the parallel wires, these wires must have a proper length in comparison with the wave length of the waves. In St. John's experiment the proper length of the wires was determined by trial. The exploring terminals of the bolom- eter were put at the ends P and S of the wires of Fig. 48. The oscillator was set in activity, and a reading of the bolometer was taken for this length of wire. A few centimeters of wire were cut off, and the reading again taken. This process was repeated until a maximum point was passed. A sharp and unmistakable maxi- mum was found when PQ had a certain length (859 centimeters). The effect fell off rapidly when the wires were shortened or length- ened from this point. The result is shown graphically in Fig. 49, Deflection of Bolometer -^ S g g ^ WIRELESS TELE A \ I \ 1 \ / \ * ' \ \ I N where distances from Q are plotted horizontally, and deflections of the galvanom- meter are plotted verti- cally. To determine the char- acter of the vibration along the wire, the lengths of QP and RS were fixed at 859 centimeters, the exploring terminals were then moved along the wires, and the bolometer readings taken for each position of the exploring terminals. A dia- grammatic representation of the result is shown in Fig. 50. The curve is seen to be simple in form, with maxima and minima at approximately equal inter- 800 900 Crn. Distance from Q R FIG. 49. Showing adjustment of parallel wires to resonance (Professor St. John). s 1234 5678 Meters from Q R FIG. 50. Curve of distribution of potential on parallel wires (St. John). vals along the wires, and with a maximum at the ends of the wires. The discussion of these experiments has been given at some length, because a wave system resembling that here described is produced in the antennae used in wireless telegraphy, and the study of the resonance conditions in wireless telegraphy circuits will be seen to be closely related with the study of stationary waves in wires. CHAPTER XI WIRELESS TELEGRAPHY BEFORE HERTZ By Conduction through Water. The first successful attempt at electric telegraphy l between stations not connected by wires seems to have been made by S. F. B. Morse in 1842. Morse describes his experiments in a letter to the Secretary of the Treasury of the United States, which was laid before the House of Representatives on Decem- ber 23, 1844. He says: " In the Autumn of 1842, at the request of the American Institute, I undertook to give the public in New York a demonstration of the practicability of my telegraph, by connecting Governor's Island with Castle Garden, a distance of a mile; and for this purpose I laid my wires properly insulated beneath the water. I had scarcely begun to operate, and had received but two or three characters, when my intentions were frustrated by the accidental destruction of a part of my conductor by a vessel, which drew them up on her anchor, and cut them off. In the moments of mortification I immediately devised a plan for avoiding such an accident in the future, by so arranging my wires along the banks of the river as to cause the water itself to conduct the electricity across. The experiments, however, were deferred till I arrived in Washington; and on December 16, 1842, I tested my arrangement across the canal, and with success. The simple fact was then ascertained that electricity could be made to cross the river without other conductors than the water itself; but it was not until the last Autumn that I had the leisure to make a series of experiments to ascertain the law of its passage. The follow- ing diagram will serve to explain the experiment : " A, B, C, D (Fig. 51) are the banks of the river; N, P, is the battery; G is the galvanometer; ww, are the wires along the banks connected with copper plates, /, g, h, i, which are placed in the water. When this arrangement is complete, the electricity, generated by the battery, passes from the positive pole P, to the plate h, across the 1 A large part of the historical information contained in this chapter was obtained from Mr. J. J. Fahie's excellent History of Wireless Telegraphy, Dodd, Mead & Co., 1902. 75 76 WIRELESS TELEGRAPHY river through the water to the plate i, and thence around the coil of the galvanometer to plate /, across the river again to plate g, and thence to the other pole of the battery, N. " The distance across the canal is 80 feet " In these experiments Morse found that it was necessary to make the wires along each shore three times as great as the distance from shore to shore across the stream. Later, under Morse's direction, his assistants, Messrs. Vail and Rogers, established communication in the same way across the Sus- quehanna River, a dis- tance of nearly a mile. Similar attempts to send signals through water by utilizing the water were made by ^M^^^^^^^ James Bowman Lindsay ~^ between 1854 and 1860. By gradually increasing the power of his plant and the length of his conductors, Lindsay succeeded, with an appa- ratus like that of Morse, in signaling across the Tay where the river is more than a mile wide. In 1880 Professor John Trowbridge of Harvard University sug- gested the use of circuits resembling those of Morse, modified by the employment of an interrupted current in the sending circuit and a telephone receiver in the receiving circuit. This modification takes advantage of the high sensitiveness, portability and rapidity of action of the telephone as a current indicator. About 1882 Professor Graham Bell made some successful experiments with the method suggested by Professor Trowbridge. The following is an extract from Prbfessor Bell's description: " Urged by Professor Trowbridge, I made some experiments which are of very great value and suggestiveness. The first was made on the Potomac River. " I had two boats. In one boat we had a Leclanche battery of six elements and an interrupter for interrupting the current very rapidly. Over the bow of the boat we made water connection by a metallic plate, and behind the boat we trailed an insulated wire, with a float at the end carrying a metallic plate, so as to bring these two terminals about 100 feet apart. I then took another boat and sailed off. In this boat we had the same arrangement, but with a WIRELESS TELEGRAPHY BEFORE HERTZ 77 telephone in the circuit. In the first boat, which was moored, I kept a man making signals; and when my boat was near his I would hear those signals very well a musical tone, something of this kind; turn, turn, turn. I then rowed my boat down the river, and at a distance of a mile and a quarter, which was the farthest distance I tried, I could still distinguish those signals." In these experiments of Morse, Lindsay, Trowbridge and Bell the signals were carried from one station to the other by conduction through the water. The current in ^- -^^ flowing from one submerged plate to /' '^ the other at the sending station spreads / out through the water in curves like / those of Fig. 52. If, now, the termi- ! nals of the receiving circuit dip down \ into the conducting area, the current ^ divides, part going through the water x and part through the receiving circuit, / / in the inverse ratio of their resistances. / This method of signaling, though at- 1 tempted with improved apparatus * by Messrs. Rathenau, Rubens, and FlG - 52 - Lines of flow - Strecker, and by the latter carried to a distance of 14 kilometers (8.7 miles), has not contributed to the art of wireless telegraphy, as it is now practiced. Dolbear 's Apparatus. A somewhat more suggestive apparatus was invented by the late Professor Dolbear of Tufts College, Massachu- setts, and was awarded a United States patent in March, 1882. Figure 53, taken from the patent specifications, shows a diagram of the apparatus. The transmitting station, shown at the left, consisted of a condenser H f connected to one terminal of the secondary t>f an induction coil G, of which the other terminal of the secondary was grounded at C. The primary of the induction coil contained a bat- tery /' and microphone transmitter T. The receiving apparatus, shown at the right, consisted of a telephone receiver R with one terminal connected to ground at D, and the other terminal connected to a condenser H, which was in turn connected through a battery l B with a second condenser H 2 . Professor Dolbear, in his patent specifications, describes the action of the apparatus as follows : " Now if words be spoken in proximity to transmitter T, the vibra- 1 The function of this battery is not evident. 78 WIRELESS TELEGRAPHY tion of its diaphragm will disturb the electric condition of the coil C, and thereby vary the potential of the ground at A, and the variations of the potential at A will cause corresponding variations of the poten- tial of the ground at B, and the receiver R will reproduce the words spoken in proximity to the transmitter, as if the wires CD were in FIG. 53. Dolbear's apparatus for wireless telegraphy. contact, or connected by a third wire. Electrical communications may be thus established between points certainly more than half a mile apart; but how much farther I cannot now say." In some other of Professor Dolbear's writings he speaks of using an automatic break and a Morse key in the primary of his coil instead of the microphone transmitter, and he also speaks of using a gilt kite carrying a fine wire from the secondary of the Ruhmkorff coil. Professor Dolbear thus made an approach to the method that was subsequently, in the hands of Marconi, to be crowned with success. The difficulty with the Dolbear apparatus was that the elevated con- ductor had to discharge through the secondary of the induction coil, and thus (as we see now) had a very slow frequency, so that the inductive action of the waves emitted was very feeble. Dolbear, therefore, did not have in his sending station the one essential that makes the Marconi sender a success; namely, electrical oscillations of high frequency. Also the detector used in Professor Dolbear's apparatus (his electrostatic telephone receiver) was not of sufficient sensitiveness. Sir William Preece's Method. Another serious attack on the problem of wireless telegraphy, before the work of Hertz and WIRELESS TELEGRAPHY BEFORE HERTZ 79 Marconi had made the way clear, was made by Sir William Preece, engineer-in-chief of the postal telegraph system of Eng- land. Preece attempted to utilize the electromagnetic induction between two long horizontal wires, one at the sending station and the other at the receiving station. These horizontal wires were supported parallel to each other on telegraph poles, and were grounded at their two ends. The sending wire contained a battery and an interrupter, or else an alternating current generator, so that the line was traversed by an interrupted or an alternating current ; while the receiving circuit contained an ordinary telephone re- ceiver. The surging current in the sending wire produced a vari- able magnetic field surrounding it. This variable magnetic field produced by the sending circuit cut or linked with the receiving circuit, and induced a periodic electromotive force in it, which was evidenced by sounds in the receiver. After several years of experimenting, Sir William Preece was able to utilize this apparatus for signaling to some of the islands a short distance off the coast of England, and in 1898 a regular installation was established at Lavernock Point on the mainland and at Flatholm in the Bristol Channel, 3.3 miles (5.2 kilometers) apart. Preece's experiments can be said to have availed only to show the futility of the attempt to get inductive action at long distance without the use of oscillations of high frequency. CHAPTER XII WIRELESS TELEGRAPHY BY HERTZIAN WAVES. MARCONI, 1896-1898 WE come now to the application to wireless telegraphy of the principles discovered by Maxwell and Hertz. For this application we are chiefly indebted to the genius, skill, and forceful initiative of Signor Guglielmo Marconi. We are to see, however, that the achievement did not come as a scientific revolution, but as a steady development to which many other investigators also contributed. The Coherer. In the extension of the effects of the Hertzian waves to great distances the first need was a detector of high sensitiveness. Such a detector was already at hand in a crude form; for as long ago as 1866 S. A. Varley had discovered that metallic filings in a loose condition have a high resistance and that this resistance is decreased to a small value under the action of an electric discharge sent through the filings. This fact was utilized by Varley in the construction of a " lightning bridge," or lightning arrester, used to protect electrical apparatus from lightning. Varley had also noticed that the resistance of the filings, when lowered by the discharge, could be brought back to its high value by tapping or shaking the vessel containing them. In 1884 Calzecchi-Onesti also made and published some inde- pendent experiments on this interesting phenomenon, verifying and extending the results of Varley. This work of Varley and Onesti remained unnoticed until 1890, when Professor E. Branly, of the Catholic University of Paris, rediscovered the phenomenon. Professor Branly studied the conductivity of metallic filings placed in a glass tube between two metallic plugs, by which the filings could be put into an electric circuit. He found that the filings were rendered conductive by electric discharges in the neighborhood of the tube, even when the discharges did not actually pass through it. He also observed that- tapping the tube restored the filings to their high resistance. He gave the name " radio-conductor " to the apparatus, and made 80 WIRELESS BY HERTZIAN WA VES MARCONI, 1896-1898 81 many interesting experiments in the effort to obtain an explana- tion of its action. Branly's radioconductor is now familiarly known as the " coherer," - a name invented by Sir Oliver Lodge. Coherer Applied to Study of Electric Waves. In 1893 and 1894 Sir Oliver Lodge applied the coherer to the study of electric waves by putting it in the place of the micrometer spark gap in a Hertz resonator, as is shown in Fig. 54. Under the action of the electric waves sent out from a properly placed Hertz oscillator, the resistance of the metallic filings in the coherer fell to a low value, so that the galvanometer G connected in series with a battery B in a local circuit through the coherer gave a deflection. After the waves ceased the resistance of the coherer remained low, so that the galvanometer remained deflected. In order to prepare FIG. 54. Sir Oliver Lodge's apparatus for detecting electric waves. for another reading it was necessary to restore the filings to high resistance by tapping the tube. Lodge effected this restoration either by a tapping mechanism driven by clockwork, or by an electric trembler (like an electric bell) mounted on the same base as the coherer. With this apparatus Professor Lodge succeeded in detecting Hertz waves at a distance of about 55 yards from the source. Experiments of Popoff. A still nearer approach to an operable form of receiving apparatus for wireless telegraphy was made in 1895 by Professor Popoff of Kronstadt, and a description of the apparatus was communicated by him to the Physico-Chemical Society of St. Petersburg in April of that year. Popoff's apparatus, which was designed for use in the study of atmospheric electricity, is shown in the diagram of Fig. 55. The left-hand terminal of the coherer was connected to a metallic rod extending above the house-top ; the right-hand terminal of the coherer was connected to earth; so that electric currents produced by the atmospheric elec- 82 WIRELESS TELEGRAPHY tricity were conducted to earth through the coherer. Through the coherer there was also a local circuit containing a battery B and a relay R. Under the action of the atmospheric electrical disturbances, the filings in the coherer became conductive, so that a current from the battery flowed through the coherer and around the electromagnet of the relay. This current magnetized FIG. 55. Professor PopofTs apparatus for studying atrnospheric electricity. the core of the relay and attracted the armature so as to close the contact at D, which put the battery in circuit with an electric bell. The bell was so placed that its hammer while in vibration struck the bell, and also struck the coherer, causing it to decohere. Thus the atmospheric discharges caused the bell to sound, a^id after the cessation of the discharge the filings were decohered so that the bell ceased to sound and awaited another discharge. In addition to the bell, Popoff used also a telegraphic registering apparatus in shunt with the bell, so as to get a written record of the duration of each atmospheric electric disturbance. In a note, dated December, 1895, he says: " I entertain the hope that when my apparatus is perfected it will be applicable to the transmission of signals to a distance by means of rapid electric vibrations when in fact a sufficiently powerful generator of these vibrations is discovered." 1 1 This quotation is from Fahie's History of Wireless Telegraphy, 1892. WIRELESS BY HERTZIAN WAVES MARCONI, 1896-1898 83 Marconi's 1896 Apparatus. We come now to the early work of Marconi. After having made some preliminary experiments on his father's estate near Bologna in Italy, Signor Marconi went to Eng- land, and on June 2, 1896, filed in the Patent Office of Great Britain a part of his first application for a patent for " improvements in trans- mitting electrical impulses and signals, and in apparatus therefor." The part of the application filed at this date is without diagrams, and contains only provisional specifications. A complete speci- fication covering the same subject matter, amply illustrated with drawings and full of details as to the invention, was filed March 2, 1897. This patent application of Mr. Marconi contains the first published account of a completed apparatus for successful wireless telegraphy by electric waves, and is, therefore, a document of con- siderable interest. It would seem to be not unprofitable to give careful attention to Marconi's description of his invention. In the description that follows, the quotations are taken from the Marconi patent specifications; and after some of the paragraphs of quoted or paraphrased description I have added a brief paragraph in the form of a summary. Hertz or Righi Oscillator and Receiver. At the transmitting station he employs " a Ruhmkorff coil having in its primary circuit a Morse key for starting or interrupting the current." The secondary of the coil he connects to " pole appliances " for producing the desired oscillations. Under " pole appliances " he mentions " insulated balls separated by small air spaces or high vacuum spaces, or compressed air or gas, or insulating liquids kept in place by a suitable insulating material, or tubes separated by similar spaces and carrying sliding discs." This form of the transmitting apparatus, as may also be seen by reference to the original drawings, is an ordinary Hertz or Righi Oscillator, actuated by a Ruhmkorff coil with a Morse key in its primary circuit. There is, however, also the suggestion of the use of a high vacuum or compressed air or gas about the spark gap. " At the receiving instrument there is a local battery circuit con- taining an ordinary receiving telegraphic or signaling instrument and an appliance for closing the circuit." The appliance for closing the circuit " consists of a tube containing conductive powder, or grains, or conductors in imperfect contact, each end of the column of powder or the terminals of the imperfect contact or conductor being connected to a metallic plate of suitable length so as to cause 84 WIRELESS TELEGRAPHY the system to resonate electrically in unison with the electrical oscil- lations transmitted to it." This part of the apparatus is, therefore, essentially the receiving apparatus of Lodge shown in Fig. 54, with a telegraphic relay or sounder substituted for the galvanometer G. That the substitution could be made had already been shown in detail by Popoff. The specifications say further: " When transmitting through the air, and it is desired that the signal or electrical action should only be sent in one direction, or when it is necessary to transmit electrical effects to the greatest possible distance without wires, I place the oscillation producer at the focus or focal line of a reflector directed to the receiving station, and I place the tube or imperfect contact at the receiving instrument in a similar reflector directed towards the transmitting instrument." This part of the specification provides for the use of reflectors like those of Hertz. However, on account of the difficulty of construct- ing reflectors sufficiently large, the reflectors were soon abandoned by Marconi in his practical work, and have not been subsequently revived. Grounded Circuits. He says further: "When transmitting through the earth or water I connect one end of the tube or contact FIG. 56. Mr. Marconi's 1896 transmitter. FIG. 57. Marconi's 1896 receiver. to earth and the other end to conductors or plates, preferably similar to each other, in the air and insulated from the earth." A diagram of this receiving apparatus is shown in Fig. 57. The corresponding earthed sending apparatus is shown in Fig. 56. The earthing of the circuits was for a time considered to be the strongest feature of the Marconi apparatus; but recent experiments WIRELESS BY HERTZIAN WAVES MARCONI, 1896-1898 85 have shown that the earthing of the circuits, though a convenience in construction, is not essential. 1 It was with these earthed circuits that Marconi made his first great gains in the distance of transmission; but as we now look back over the experiments, we see that the gain in distance came about primarily through the fact that with this apparatus his circuits were placed vertical rather than horizontal, and also through the use of longer waves and more energy and larger radiating and receiving antennae, rather than through the use of the mere earth connections. To this subject we return in Chapter XIV. Marconi's Coherer. In addition to the practical introduction of the vertically placed radiator with ground connection Signer Mar- coni also made tremendous progress over other early investigators in his skill in constructing and using the coherer. A sketch of the coherer, drawn natural size from Marconi's specifications, is shown in Fig. 58. The metal plugs PP are of silver slightly amalgamated with mercury, but no excess of /^== / r FJ== mercury in the form of globules i f , , i rri i r, FIG. 58. Marconi coherer is left on them. The plugs fit (natural size). accurately into a glass tube, and are within jV inch of each other. The filings in the space between the plugs are preferably 96% nickel and 4% silver, and should not be fine, but rather coarse. They should be dry and free from grease and dirt, and should be uniform in size. The tube containing the filings is preferably exhausted of air and sealed up. In sealing up the tube care should be taken not to oxidize the filings. In order that the coherer may not be injured by the current through it, not more than T^OU of an ampere of current should be used in the local circuit. Marconi's Decohering Device. One of the greatest difficulties to be overcome in operating a delicate coherer arises from the fact that the signal causes the coherer to become conductive, and if left alone, the coherer perseveres in this conductive condition. In order to restore it to its high resistance so as to be ready for the next signal, it is necessary to employ an automatic tapper, or trembler, which is started into action by the incoming impulse, and which stops the signal and itself when the incoming impulse ceases. Signor Marconi brought the decohering device to a high state of perfection, and as a result changed the capricious tube of filings into a reliable instrument for practical use. 1 See Chapter XIV. WIRELESS TELEGRAPHY A diagram drawn from Marconi's descriptions is shown in Fig. 59. The continuous lines of the drawing show the coherer Co, the relay R, the trembler T, and the sounder S, and their circuits. These had been used in almost the same form by Popoff. The dotted lines p, pi, q, q f , and h show the further improvements, introduced by Marconi, and consisting of resistances to prevent the inductive kick of the instruments from acting on the coherer. The action of the decohering device and the protective resistances FIG. 59. Receiving circuit showing coherer and protective devices. may be best understood by following the process as an actual message is sent from one station and received by another. This is done in the next paragraph. Action of the Apparatus in Sending and Receiving. Let us suppose the Morse key in the primary of the Ruhmkorff coil at the sending station to be operated as in ordinary line telegraphy. While the key is closed, the interrupter in the primary of the coil makes and breaks the circuit at the rate of, say, 100 a second. At each break of the primary the potential of the secondary rises to a high value and charges the oscillator sufficiently to produce a WIRELESS BY HERTZIAN WAVES MARCONI, 1896-1898 87 spark. We have thus in our supposed case 100 sparks a second. Each spark occurs with oscillations of very high frequency and produces a train of electric waves. With such a sending station we should have arriving at the receiving station a train consisting of a few l of these extremely rapid waves, followed T IF of a second later by a second similar train; and thus at intervals of T <5i> second there would arrive successive short trains of waves while the send- ing key is depressed. Under the action of these trains of incoming waves the filings in the tube cohere, so that a current flows from the battery Bi (Fig. 59) through the coherer Co and the relay R. This battery current pulls the armature of the relay so as to close the gap at A. When the gap is closed a second battery B sends a current through the coils of the sounder S t and also through the coils of the trembler T. The trembler is like an electric bell (with a somewhat shorter striking arm), and makes a series of strokes against the tube of the coherer. This decoheres the filings, but so long as the key at the sending station is closed, the waves continue to arrive and cause a repetition of the coherence, thus putting the coherer in a state of repeated coherence and decoher- ence during the arrival of the waves. The armature of the relay is adjusted so that the relay is somewhat sluggish and does not open at each decoherence. Therefore, the contact of the relay remains closed, and consequently the sounder armature stays down as long as the trains of waves continue to arrive. When, however, the sending key is released, and the waves cease to arrive, the decoherence due to the tapper perseveres, the relay contact opens, the sounder arm is released, and at the same time the trembler stops. Thus each closing and opening of the key at the sending station produces a corresponding down and up stroke of the sounder, making a dash or a dot, according as the sending key is depressed for a long or a short interval of time. Instead of the sounder for translating the message an ordinary Morse registering tape-machine may be used to give a written record of the dashes and dots. Marconi's Protective Resistances and Inductances. Return- ing now to diagram Fig. 59, let us examine into the purpose of the resistances p, p { , q, q r , and h, represented by the dotted lines. 1 I have taken revolving-mirror photographs of the spark of a Marconi Oscillator with a period of TtftfiUffff second, and found that there are about 12 waves in a train. 88 WIRELESS TELEGRAPHY One of these resistances is shunted about each of the electromagnets of the circuit and one about each make-and-break contact in the circuit. These resistances are for the purpose of preventing a sudden rise of electromotive force in any of the circuits due to the action of the self-inductances of the electromagnet. We have learned in Chapter III that when a current is flowing in a large self-inductance, for example, through the coherer and relay, and the circuit is suddenly interrupted, e.g., by a stroke of the tapper against the coherer, the self-inductance in the circuit produces a large electromotive force tending to cause the current to continue. This large e.m.f. of inductance in the coherer cir- cuit is equivalent to a discharge through the coherer, and if not prevented would affect the coherer and cause it to cohere again, when in fact the tapping was designed to cause it to decohere. This action is prevented by the resistance q shunted about the relay. In a similar manner the other resistances absorb the energy sent into the circuit by the inductive kicks from the other electro- magnets, which would otherwise act either directly or inductively on the coherer. Mr. Marconi gives the following appropriate values for these resistances: p and p v ought each to be of resistance four times the resistance of the trembler; the resistance of q should be three or four times that of the relay; q' should be about 20,000 ohms, or should be a water resistance offering polarization voltage equal to the battery E\\ h should have a resistance equal to four times the resistance of the telegraph instrument S. These resistances should all be non-inductive. The resistance of the relay itself should be more than 1000 ohms, and the working current should be less than TITO o ampere. Two coils of a few turns of wire CC wound inductively on iron cores are inserted in the relay circuit to prevent the electric oscil- lations due to the incoming waves from escaping the coherer by going into the relay circuit. However, the inductance of the relay itself effects this purpose to a large extent. The decohering and protective devices described in Marconi's patent specifications are still a model for the proper construction of these important accessories to the coherer. Recently, however, the coherer has been almost completely replaced by other forms of detectors operating on other principles and that do not need to be decohered. These newer detectors are the subject matter of a later chapter. WIRELESS BY HERTZIAN WAVES MARCONI, 1896-1898 89 Balloons or Kites. Another important suggestion contained in the 1897 specifications is the suggestion that " the larger the plates of the receiver and the transmitter, and the higher from the earth the plates are suspended, the greater is the distance at which it is possible to communicate at parity with other conditions." " Balloons can also be used instead of plates on poles, provided they carry up a plate or are themselves made conductive by being /^ r EH FIG. 60. Simple Marconi circuits with antenna sustained by a kite. Switch for "cutting over" from sending to receiving. covered with tinfoil. As the height to which they may be sent is great, the distance at which communication is possible becomes greatly multiplied. Kites may also be successfully employed if made conductive by means of tinfoil." This sentence, therefore, provides for the use of antennae of great height. It should be noted here that the plates or tinfoil covering on the balloons or kites, which the inventor makes a necessary provision of the apparatus, are really nonessential. A diagram of circuits in which kites are used for suspending the vertical wires is shown hi Fig. 60. 90 WIRELESS TELEGRAPHY Shifting from Sending to Receiving. In practice there are both a sending and a receiving apparatus at each station. Only one antenna is needed, and this is shifted from the spark balls to the coherer, in changing from sending to receiving. This shifting is done by means of a switch, as is shown in Fig. 60, or by means of the key itself, as is shown in Fig. 61. In Fig. 61 the key, which is also a switch, is in the position for sending, and the coherer, which is at R, is disconnected from the antenna. When the message is finished, or, if desired, between words of the message, the key is released, and the arm b, which is insulated from 6 1 , is allowed to u^ FIG. 61. Marconi's key for changing from sending to receiving. descend so that the contact point b 2 rests on the point b 3 . This connects the antenna with the coherer and puts the station in a condition to receive. The " Claims " of Marconi's 1896 Patent. The " claims " of a patent are a series of succinct statements at the end of the descrip- tive matter and are supposed to embody the invention in its most general form. Signor Marconi's English patent of 1896 contains nineteen claims. The three most comprehensive of these claims are the following: 15. A receiver consisting of a sensitive tube or other imperfect contact inserted in a circuit, one end of the sensitive tube or other imperfect contact being put to earth whilst the other end is connected to an insulated conductor. 16. The combination of a transmitter having one end of its sparking appli- ance or poles connected to earth, and the other to an insulated conductor, with a receiver as is mentioned in claim 15. 17. A receiver consisting of a sensitive tube or other imperfect contact inserted in a circuit, and earth connections to each end of the sensitive con- tact or tube through condensers or their equivalent. WIRELESS BY HERTZIAN WAVES - MARCONI, 1896-1898 91 Marconi's Achievements between 1896 and 1898. In July, 1896, soon after arriving in England, Mr. Marconi submitted his plans to Sir William Preece, director of the postal-telegraph system of Eng- land. Preece, of whose activity in connection with attempts at wireless telegraphy we have already learned, entered eagerly into the new experiments. The first messages were sent from a room in the General Post Office to an impromptu station 100 yards distant. Soon afterwards, at Salisbury Plain, with parabolic reflectors about the instruments, communication was established at a distance of two miles. In May, 1897, discarding the reflectors and using grounded circuits, Mr. Marconi covered a distance of 8.7 miles between Lavernock Point and Brean Down. Kites were employed in this experiment to sup- port the vertical wires. In July, 1897, important trials were made at Spezia, Italy, at the request of the Italian Government, and communication was estab- lished at a distance of 12 miles between a warship and a shore station. In July, 1898, the Marconi apparatus was used to report the yacht races at the Kingston Regatta, and a large number of correct mes- sages were ^exchanged between a press boat and the shore at dis- tances extending up to 20 miles. These various experiments constituted a complete demonstration of the utility of the invention. CHAPTER XIII ELECTRIC WAVE TELEGRAPHY BY RESONANT CIRCUITS A SIMPLE radiating circuit, 1 like that shown in Fig. 60 of the pre- ceding chapter, consisting of a spark gap with one side grounded and the other attached to an antenna, has a definite period of electric oscillation. The receiving circuit has also a characteristic period of oscillation. A maximum effect will be obtained when the two have the same fundamental period; that is to say, when the two circuits are in resonance. Marconi recognized this fact, and in his specifica- tions, particularly with reference to the use of a receiving resonator of two metallic vanes, he provides a method of experimentally de- termining the size of the vanes that will give resonance with the transmitter. In the course of further experiments it was soon found, however, that the type of simple receiving conductor in which the coherer is inserted directly in the antenna circuit is not a very discriminating resonator. Also it is usually not practicable to change the size of the vanes or the length of the vertical wires in order to make changes in the period of the circuit. When several wireless telegraph stations are to be operated at once, it is highly desirable, in order to be able to avoid confusion, to have a method of readily adjusting the receiving circuit to resonance with the wave lengths it is desired to receive and out of resonance with undesired signals of a different wave length. Several methods have been devised for accomplishing this result, which though attain- ing only limited success, have yet been of great advantage in wireless telegraphy. Circuits capable of being attuned, or adjusted for resonance, are called syntonic circuits, or resonant circuits. In the present chapter it is proposed to discuss some of the general types of resonant circuits. Quantitative experiments in regard to resonance, and some of the practical details of construction of appara- tus, will be given later. 1 We shall refer to a rectilinear oscillator of this character as a circuit, since according to the work of Maxwell a circuit need not be conductively closed. 92 ELECTRIC WAVE TELEGRAPHY BY RESONANT CIRCUITS 93 A Simple Variable Circuit. A simple method of easily varying the period of a receiving circuit consists in the use of a variable in- ductance L (Fig. 62), inserted between the detec- tor and the antenna or between the detector and the ground at the receiving station. Such a vari- able inductance, or tuning coil, is made of a single layer of wire wound on an insulating tube of glass or ebonite, and is varied by a contact sliding along the coil so as to put more or fewer turns of induct- ance into the circuit. A similar tuning coil, though usually of larger wire, may be used at the sending station also. At the sending station the coil is inserted between the spark gap and the an- tenna or between the spark gap and the ground connection. Increase of inductance in either circuit increases the time of vibrations, which brings a correspond- ing increase of wave length. The use of adjustable inductances in both the sending and the receiving circuits was apparently first suggested by Sir Oliver Lodge in a patent application of 1887, which is reviewed later in the present chapter. Coupled Circuits. Certain other methods, employed for adjust- ing both the sending station and the receiving station, and found to produce better results both for transmitting with large quantities of energy and for receiving with comparatively sharp resonance, make use of coupled circuits. The resonance relations in these coupled cir- cuits has been the subject of much theoretical and experimental research. As introductory to the description of the coupled circuits, I shall recall to the reader the familiar and interesting experiments of Mr. Tesla and of Professor Thomson on the production of electric oscillations of high frequency and high potential. High-frequency Transformers of Thomson and Tesla. The high-frequency transformer that was apparently independently developed by Mr. Nikola Tesla and Professor Elihu Thomson about 1890 is shown in sketch in Fig. 63. A primary coil P, consisting of one or two turns of heavy wire, is connected in series with a bank of Ley den jars C and a spark gap G. A secondary coil S, consisting of three or four hundred turns of wire wound in a single layer on a paper or vulcanite tube, is inserted axially within the primary. When the :G. 62. Simple antenna circuit having a variable inductance for tuning. 94 WIRELESS TELEGRAPHY bank of jars is connected by means of the leads W, W with the second- ary of a Ruhmkorff coil, or better with the secondary of an alternating current step-up transformer, the Ley den jars are repeatedly charged by the Ruhmkorff coil or transformer at intervals of, say, TTJTF or v^ of a second. When the potential at each charge of the jars reaches a value high enough to spark across the gap G, the jars dis- charge with oscillations of extremely high frequency. A group of these oscillations occurs during each spark at the gap G. These FIG. 63. Tesia or Thomson coil. high-frequency oscillations in the primary coil P act inductively on the secondary coil $, and on account of the extreme rapidity of change of current in the primary, the electromotive force induced in the secondary is very high, and produces a series of sparks between the terminals T and T 2 of the secondary. It should be noted that the primary has a period of its own, and that the coil of wire S used as a secondary has also a period of its own; and in order to get the greatest spark at the secondary terminals, it is necessary to adjust the number of jars C or else the number of turns of wire on either P or S, so that the condenser ELECTRIC WAVE TELEGRAPHY BY RESONANT CIRCUITS 95 circuit and the secondary coil shall be in resonance with each other. By the use of apparatus of this character Mr. Tesla has pro- duced enormous sparks twenty-three feet long and of great volume graphically described as being accompanied by a roar like Niagara. The transformer PS is called a high-frequency transformer, an oscillation transformer, or an air-core transformer, to distinguish it from an ordinary iron-core transformer, such as is used with commercial alternating currents of slow frequency. Oscillation transformers, built on somewhat different lines from the one above described, have met with application to both the send- ing and the receiving circuits of electric- wave telegraphy, and by the use of these transformers a considerable advance has been made, both in the greater distances attained and in the diminished con- fusion of signals of different wave lengths. Two Systems of Coupled Circuits. The form given to these coupled circuits is considerably varied in practice. There are, however, two important general types. These are represented in the accompanying figures (Fig. 64 and 65) and are called re- spectively the inductively coupled and the direct coupled types. The Inductively Coupled Type. This type is shown in Fig. 64. In this system the sending station, shown on the left, is seen to consist of a Tesla high-frequency apparatus, with one secondary terminal connected to an antenna and the other secondary ter- minal connected to the ground. Power is supplied to the circuit by an alternating current transformer or a Ruhmkorff coil to which the wires W , W lead. The receiving station of this system, shown at the right in Fig. 64, has also an oscillation transformer P' S', and is in prin- ciple like the sending station, except that the detector D f with its accessories is usually put in place of the spark gap of the sending apparatus. The coils P' and S' and condenser C" used with the receiving apparatus generally have different inductances and capacity from those of the sending apparatus, and not being traversed by high-potential currents they are usually made more compact. In this inductively coupled system of circuits, oscillatory cur- rents in the sending antenna are produced inductively by the oscillatory discharge of the condenser C through the primary coil P. These oscillatory currents in the sending antenna produce 96 WIRELESS TELEGRAPHY electric waves, which travel away in all directions with the velo- city of light (186,000 miles per second). Arriving at the receiving station these electric waves set up oscillations in the antenna cir- cuit A'P'E'. These currents through P r act inductively on S', and produce oscillatory currents in the detector circuit. \ FIG. 64. Inductively coupled transmitting and receiving circuits. \ FIG. 65. Direct coupled transmitting and receiving circuits. The Direct Coupled Type. Figure 65 shows the other form of coupled apparatus, constituting the direct coupled system. This ELECTRIC WAVE TELEGRAPHY BY RESONANT CIRCUITS 97 system employs auto-transformers; that is to say, instead of hav- ing separate primary and secondary coils in the high-frequency transformer, the primary coil (P or P') at either station is a part of the secondary coil. At the sending station (at the left) the condenser discharges through some of the turns of the secondary, and the discharge acts inductively on the whole of the secondary. Likewise, at the receiving station the oscillations in the antenna pass through a part P' of the secondary S' and act inductively on the whole of S'. Theory and experiment show that in principle the direct coupled circuits differ very little from the inductively coupled system. Introduction of Coupled Circuits into Practice. Postponing for a time the direct discussion of the principles involved in the use of the coupled circuits, let us take up historically the matter of the introduction of these circuits into wireless telegraph practice. The examination of the question as to the priority of the different claimants to this improvement is fraught with considerable diffi- culty. Lodge, and Marconi in England, and Braun in Germany, have clearly established dates of publication by patent applica- tions. While examining the question of priority, I shall also give a brief description of the apparatus of these several inventors, so far as pertains to the form of circuits used. Sir Oliver Lodge's Apparatus. On May 10, 1897, Professor Lodge filed a patent application in England for improvements in FIG. 66. Lodge's transmitter and receiver. wireless telegraphy. The corresponding application in the United States was filed Feb. 1, 1898. What he claims to be the most promi- 98 WIRELESS TELEGRAPHY nent feature of his invention is represented in Fig. 66. The emitter and receiver consist preferably of two large conical con- ductors, called capacity areas, supported by poles; but horizontally placed conductors may also serve as his capacity areas, and one of these capacity areas, he says, may be the earth. At the sending station (left) he joins the two capacity areas h and h l to polished knobs h 2 and h 3 . Between either capacity area and its knob be places a syntonizing self-inductance coil. " The object of this coil," Lodge says, is " to prolong the electric oscilla- tions occurring in the radiator, so as to constitute it a radiator of definite frequency or pitch and obtain a succession of tone waves emitted, and thereby to render syntony in a receiver possible, because exactitude of response depends on the fact that the total number of oscillations in a suitably arranged circuit is very great." He provides also for varying the number of active turns of these coils for the purpose of varying the period of the circuits. After having described the ordinary way of charging the vanes of the emitter by connecting, them directly or through small spark gaps to the secondary of a Ruhmkorff coil, as Hertz and Righi had FIG. 67. Lodge's apparatus for exciting an antenna by means of a Leyden jar discharge. done, Lodge proposes also the use of Leyden jars in the manner shown in Fig. 67. In this diagram the jars are shown at jj. The inner coatings of the jars are connected to the secondary of a Ruhmkorff coil, the outer coatings of the jars are joined by an inductance coil of thin wire k. This coil is necessary in order that the jars may charge. When the jars discharge across the gap h 10 h n , sparks also appear across the gaps h 6 h 7 and h?h 3 . This apparatus, therefore, shows the use of a Leyden jar circuit dis- ELECTRIC WAVE TELEGRAPHY BY RESONANT CIRCUITS 99 charging into the antenna circuit, and if the coil k has a large inductance, as it seems to have from the fact that it is made of " fairly thin wire," this sending arrangement may be looked upon as a special and very imperfect form of the direct-coupled type of sending circuit of Fig. 65. It is imperfect in the use of the mul- tiplicity of spark gaps, for if all the gaps except h w h n had been closed, the coil k, which was put in as a charging bridge across the unnecessary gaps, could then have been omitted, and the apparatus would have been a very useful form of direct-coupled emitter. While we are accustomed to the use of multiple gaps in replace- ment of a single gap, and while the multiple gap is in some construc- tions a distinct advantage over the single gap, still the introduction of one of the multiplicity of gaps directly into the antenna circuit FIG. 68. Lodge's inductively coupled receiving transformer. is certainly an annulment of the chief advantage accruing from the coupled circuits. In the receiving apparatus Lodge shows the use of an oscillation transformer. Reference is made to Fig. 68. His conical capacity areas or their equivalent are connected to the primary coil h 4 . About this a secondary coil u is placed, and is connected with the coherer e, a battery /, and the telegraphic receiving instrument g. The purpose of connecting the detector in a secondary circuit instead of directly in the antenna, is, according to the patentee, to " leave the resonator freer to vibrate electrically without disturbance from attached wires." This is an excellent reason, but the receiving apparatus, as shown in this diagram, which was taken from Lodge's patent specifications, has the fatal defect that no condenser is shown in the secondary circuit, and that the high-frequency oscillations have to go through the telegraph instrument. Hence, apart from the suggestion of the 100 WIRELESS TELEGRAPHY use of a transformer in the receiving apparatus, we cannot consider this circuit of Lodge to be a clear disclosure of the inductively coupled receiving system. Later Lodge and Muirhead, in a British specification filed Dec. 8, 1897, partially remedied the above defect, by the arrangement of circuits shown in Fig. 69. Between the two winged-shaped capacity areas A A, they inserted an induc- tance coil d, and a large condenser C. About this condenser a battery B and a telegraph instrument / are connected; and about both induc- tance and condenser is connected a coherer Co. With proper values of the inductance and capacity, this arrangement is a special case of the direct coupled receiving apparatus. These contrivances of Lodge and of Lodge and Muirhead seem not to have been developed by them experimentally to their natural com- pletion, which would perhaps have led to one or the other of the types given in Figs. 64 and 65. They did come to one of these types much later, but only after Mr. Marconi and Professor Ferdinand Braun had published their descriptions of the coupled circuits, and by various public demonstrations and polemics had shown the great advantage of the coupled circuits. Just a word as to Lodge's wing-shaped vanes at the transmitter and receiver. These are an enlarged oscillator and resonator pre- serving the symmetry of Hertz's original apparatus. In practice they have not been much used in this symmetric form, probably on account of the difficulty of giving to the apparatus sufficiently large dimensions when both vanes have to be supported vertically in an elevated position. In practice, Lodge and Muirhead early replaced the bottom vane and sometimes also the top one by the alternative, horizontally-placed conductor, consisting of a sheet of wire netting. The spark gap, instead of being halfway between these conductors, is usually nearer the lower conductor, as shown in Fig. 70, with an FIG. 69. A method employed by Lodge for connecting the receiving apparatus to the antenna. ELECTRIC WAVE TELEGRAPHY BY RESONANT CIRCUITS 101 inductance L added below the gap, for preserving approximate electrical symmetry and for tuning. In addition to these various sugges- tions by Lodge in regard to the use of tuning coils and transformers in the circuits, and the maintenance by him of the possibilities of the ungrounded circuits, Professor Lodge, together with Messrs. Muirhead and Robinson, has also devised a new form of coherer. This is described in Chapter XVI. The Coupled Circuits of Ferdinand Braun. Let us return to the matter of the coupled circuits. In a German patent, No. 111,578, applied for October 14, 1898, Professor Ferdinand Braun of Strassburg in Germany describes "a (Lodge). M M FIG. 71a, 716, 71c, 7ld. Professor Ferdinand Braun's methods of coupling a condenser circuit to an antenna. 102 WIRELESS TELEGRAPHY form of connection for an oscillator coupled with an air-wire for spark telegraphy." He proposes to use in this invention, which is a sending apparatus, the waves which are produced by the 11 discharge of Ley den jars in the presence of induction coils." The accompanying Figures, 71 a , 71 6 , 71 C , 71 d , are taken from Braun's patent specifications. In 71 a " F is a Ley den jar, a the spark gap, P an inductance coil and M the emitting wire." In 71 & " the form of connection is so changed that a primary and a secondary coil are used, so that in the circuit of the emitter M a coil only is present." Fig. 71 C " shows two Ley den jars connected one behind the other in so-called cascade connection, and " Fig. 71 d " shows the same connection for the use of a transformed current." In examining these figures it should be borne in mind that the two coils P and S, which for simplicity of drawing are shown side by side, are really wound one around the other so as to form an oscillation transformer. It should also be borne in mind, while reading the specifications, that Professor Braun means primarily to describe a method of con- necting the Ley den-jar circuit to what he calls the air-wire circuit, and that the document does not purport to give a description of a complete sending apparatus. This is clear from his single claim at the end of his description. His claim is a " form of connection of an oscillator coupled with an air wire for spark telegraphy, characterized by an oscillation circuit containing a Ley den jar and a spark gap, to which the air wire for sending out the waves is connected either directly or by means of a transformer, for the purpose of bringing by this means greater quantities of energy into action." His clear understanding of how this energy may be increased is evident from a paragraph of the specification which says: "Above all, the slower oscillations have the advantage that their energy may be increased by increasing their potential amplitude (by transforma- tion) as well as by increase of capacity and by the use of powerful sources of electricity." The gain of energy by increasing the poten- tial could not be attained with an emitter having the spark gap directly in the antenna, because there is a certain " active " spark length that cannot be exceeded. This is pointed out in the specifi- cations. However, in neither the German nor the corresponding American ELECTRIC WAVE TELEGRAPHY BY RESONANT CIRCUITS 103 patent, filed Feb: 6, 1899, does Professor Braun speak of the neces- sity of properly attuning the secondary circuit to the period of the condenser circuit, which is a prerequisite for attaining the high poten- tial in the antenna circuit, and without this attuning of the secondary to the primary circuit, the large capacity of the primary condenser and the use of powerful sources of electricity would not give any advantage over the simple Marconi antenna. The first mention by Braun of the required tuning, so far as I have been able to find, is in a publication of the 5th of March, 1901, in the Physikalische Zeitschrift, Vol. 2, p. 373, and in a book by Braun entitled Drahtlose Telegraphie durch Wasser und Luft, published in 1901. In examining Braun's patent drawings one may wish to know whether the antenna circuit is to be grounded or otherwise bal- anced by a capacity at the other end of the secondary. Nothing is said on this subject in the German patent, but in the correspond- ing American patent he says, with reference to Fig. 7 Id, that " one end of the secondary coil of the transformer S is connected with the transmitting wire M, and the other end is shown prolonged and ending in an arrow to indicate that it may be prolonged by adding a suitable length of insulated wire or connected to some other capacity area/' In his book of 1901 a corresponding prolongation or addition of capacity is indicated in his draw- ing of the direct coupled circuit, as is shown in Fig. 72. There is nothing in these early patents of Professor Braun relating to coupled circuits at the receiving station. The coupled receiving circuits were undoubtedly invented by Mar- coni, and also his description of the induc- tively coupled sending station, though of pub- lished date a little later than that of Braun, is a much fuller and a more complete disclosure of the invention. The work of Marconi in developing the coupled circuits will now be discussed. Marconi's Coupled Circuits. Mr. Marconi, in an English patent applied for June 1, 1898, clearly sets forth the transformer arrangement for a receiving station. This is shown in Fig. 73, in which A leads to the antenna, and E to earth. The coils JV 2 , oo FIG. 72. Form of direct coupled transmitter de- vised by Fer- dinand Braun. 104 WIRELESS TELEGRAPHY which are represented as side by side, are the oscillation trans- former and are really wound one around the other. The primary J l is connected to the antenna and the earth; while the secondary is in circuit with the coherer T and the condenser K 1 . A relay R and battery B are connected about the coherer through the choking coils c l c 2 . This is, therefore, a clear presentation of the inductively connected receiving station. FIG. 73. Marconi's inductively coupled receiving circuit. FIG. 74. Marconi's inductively coupled transmitter. In 1900 Marconi was granted an English patent for an induc- tively coupled sending station also. This is shown in Fig. 74, and is of the typical form of our Fig. 64, with, however, an added variable inductance g in the antenna. In the 1900 description of this apparatus Mr. Marconi clearly points out the necessity of having the primary and the secondary circuits at the sending station and the corresponding circuits at the receiving station adjusted to resonance with one another. He says: " The capacity and self-induction of the four circuits i.e., the primary and secondary circuits at the transmitting station and the primary and secondary circuits at any one of the receiving stations in a communicating system are each and all to be so independently adjusted as to make the product of the self-induc- tion multiplied by the capacity the same in each case or multiples ELECTRIC WAVE TELEGRAPHY BY RESONANT CIRCUITS 105 of each other that is to say, the electrical time periods of the four circuits are to be the same or octaves of each other." The advantages of the coupled circuit at the sending station, according to Signor Marconi, arises in " the approximately closed circuit of the primary being a good conserver and the open cir- cuit of the secondary being a good radiator of wave energy." The variable inductance g in Fig. 74 placed in the antenna of the sending circuit and a corresponding coil at the receiving station were used to aid in this process of tuning. K FIG. 75. Apparatus used by Marconi for sending two messages at once. A sending and a receiving station devised by Mr. Marconi for sending or receiving two messages at once with the use of a single antenna are shown in Fig. 75 and Fig. 76. This was successfully employed and exhibited by Marconi in the autumn of 1900. Two operators at the two keys K and K of the sending station made the signals. The two condenser circuits having different values of capacity and self-inductance were independently charged, and discharged with different periods of oscillation. These two periods were both impressed on the antenna through circuits which by means of the antenna inductances d and d' were made to resonate with the respective condenser periods. At the receiving 106 WIRELESS TELEGRAPHY station, Fig. 76, the waves constituting the double message acted on the receiving antenna. The waves of the shorter period in- duced oscillations through the antenna and through the right- hand primary to the earth; while the oscillations of longer period passed through the circuit to the left, which contained the greater inductance in the antenna circuit. The two periodic disturbances, thus separated, act inductively on their properly tuned coherer circuits, and give up the two messages without confusion to the two receiving instruments. This duplex wireless telegraphy can FIG. 76. Marconi duplex receiving apparatus. be carried on only provided the wave lengths of the two sending stations are not two nearly equal. Some very notable achievements were made by Mr. Marconi with these resonant circuits in 1901 and 1902. A sending station of great power was completed at Poldu in Cornwall, England, in 1901. In December of that year signals were reported to have been sent across the Atlantic Ocean from Poldu to Cape Race near St. Johns in Newfoundland. The signals, which consisted of the letter "S " repeated at stated intervals, were said to have been clearly received at Cape Race, by means of a receiving antenna consisting of a copper wire 400 feet long ELECTRIC WAVE TELEGRAPHY BY RESONANT CIRCUITS 107 supported by a kite. The detector employed in the transatlantic experiments was an instrument known as the " Italian Navy Coherer," and consisted of a globule of mercury between iron terminals in a glass tube. This form of detector is self-restoring, and with it a telephone receiver in series with a battery is used in the local circuit in the place of the ordinary telegraph relay. In March, 1902, messages sent out from Poldu were received by the .Marconi apparatus on board the Steamer Philadelphia when the steamer was 1550 miles (2400 kilometers) from the send- ing station. In December of the same year Marconi announced the transmission of three entire messages from Glace Bay, Nova Scotia, to Poldu in England, a distance of 2300 miles. On January 19, 1903, the powerful Marconi station at Well- fleet, Cape Cod, Massachusetts, transmitted to Poldu, England, the following message from the President of the United States to the King of England : "His MAJESTY, EDWARD VII, London, England. In taking advantage of the wonderful triumph of scientific research and ingenuity which has been achieved in perfecting a system of wireless teleg- raphy, I extend on behalf of the American people most cordial greetings and good wishes to you and to all the people of the British Empire. THEODORE ROOSEVELT." WELLFLEET, MASS., January 19, 1903. This message, though intended to be relayed at Cape Race, was received, according to reports issued by the Marconi Company, direct at the Poldu station in England. CHAPTER XIV NATURE OF THE OSCILLATION. THE GROUNDING OF CIRCUITS IN the two preceding chapters, devoted to a period of invention and rapid development of wireless telegraphy, several important facts have been introduced with only casual examination. Among the questions there raised the most interesting is perhaps the ques- tion of the role played by the earth. This question has two aspects. First, it has been seen that both grounded and ungrounded oscil- lators have been employed. What is,the relation between these two forms of oscillator, and what effect has the ground connection on the nature of the vibration? Second, it has been apparent from the great distances attained, in the transmission of messages entirely across the Atlantic Ocean that the electric waves are not lost to the receiver by reason of the curvature of the earth, even when the two stations are separated by a distance that is a considerable fraction of the earth's whole circumference. How is it .that the electric waves are propagated from one station to the other, and how does the earth contribute to the process? These two questions, dealing with the nature of the vibration and the manner of the propagation of the waves, will be considered respec- tively in this chapter and in the next chapter. As introductory, we shall need first to consider the oscillations occurring in a simple ungrounded Hertz oscillator. Current and Potential in a Hertz Oscillator. Suppose the Hertz oscillator to consist of two metallic rods or wires with a spark gap between, and suppose the two halves of the oscillator to be charged, the one positive and the other negative, as shown in the first diagram, (a), of Fig. 77, in which the shaded area attached to the upper half of the oscillator is taken to indicate positive electricity, while the unshaded area attached to the lower rod indicates negative electric- ity. These areas are made rectangular to show that there is at the beginning a uniform distribution of the two electricities respectively on the two rods. 108 NATURE OF THE OSCILLATION 109 If the electrostatic capacity per unit of length of the rods is uni- form throughout both rods, which is approximately true, when the rods are not too short the potential of the conductor at any point of its length will be proportional to the charge, so that the shaded area representing a distribution of positive charge may also be looked upon as showing the distribution of positive potential, while the unshaded area represents negative potential. Thus, in the condition depicted in diagram (a), there is a uniform positive potential over t = o W Charged top + Neutral Neutral Charged top -4- (O }urre] ent o DQ Current Max .down CO Current FlG. 77. Current Current =O Max.up =o Potential and current distribution. the top rod, and a corresponding negative potential over the bottom rod. This is before the spark begins. Suppose, now, the spark to start between the rods; the gap between the rods becomes conductive, and a current begins to flow between the rods. There is a flow of positive electricity from the top rod and a flow of negative electricity from the bottom rod. The elec- tricity to flow first across the gap is that in the neighborhood of the spark gap, because it is there that the potential gradient is greatest. After a short time one-fourth the period of a complete oscillation the condition of the charge, and likewise the potential, of the rod WIRELESS TELEGRAPHY will be that condition represented in diagram (6), in which one-half of the positive electricity has gone into the lower rod, and half the negative electricity has gone into the upper rod, giving both rods equal quantities of positive and negative electricity, so that both rods are neutral. Why, then, does not the action cease? In order to be able to see why the action does not cease when the charge has become neutral throughout the oscillator, we must take into consideration the current in the conductor as well as the charges. The second row of diagrams of Fig. 77 represents the current at the epochs corresponding to the potential representations of the first row. At the beginning the potential, or charge distribution, is shown by diagram (a) of the top row. At the same time no current is flowing, which fact is represented by the inactive oscillator shown at (a') of the second row. The current is now supposed to begin. It cannot spring to its final value at once, because the increase of the current builds up a magnetic field surrounding the oscillator, and this grow- ing magnetic field produces an electromotive force in the conductor opposing the growth of the current. Time is thus required for the current to become established. At a time equal to one-fourth the period of complete oscillation, t = T r /4, the current has grown to its maximum value, which is represented at (&') The shading to the right of the conductor is meant to represent the magnitude of the current at each point of the conductor, though the current is along the conductor and not perpendicular to it, as the shading is. It is interesting to note that the current is not uniform throughout the length of the conductor. The current is greatest near the middle of the conductor, and is zero at each end. The reason that it is greatest at the middle is that the current flowing out front the center toward either end decreases by reason of the charge that it leaves along the conductor en route. At the very end of the conductor the current is zero, because no electricity flows out beyond the end and none flows in from beyond the end. We have thus in the conductor a distribution of current like that of (6') large in the middle and zero at both ends. Thus, at the time t = T/, the conductor is in a neutral condition with respect to charge, but is being traversed by a current in a down- ward direction. This current is a maximum with respect to time, for the next instant the positive electricity in the lower rod begins to be in excess. This calls into play an opposing electromotive force, and diminishes the current, which, however, cannot cease at once, NATURE OF THE OSCILLATION 111 because any diminution of the current diminishes the surrounding magnetic field, and gives an electromotive force tending to preserve the current. The current thus continues to pile up a positive charge on the lower rod, in spite of the fact that this piled-up charge is exerting a restoring force. Presently, however, this restoring force, which has gone on increas- ing, brings the current to a stop. Then when there is no current, there is no magnetic field, and the accumulated positive electricity on the lower rod starts the current upward. This reversal of the current occurs at a time t = T/2; and the condition of the charge and current is represented at (c) and (c 7 ). The upward current continues to flow, and produces successively the conditions (d) and (d'), at t = 3774, and (e) and (e f ) at t = T. In the last named state the upper rod is entirely positive, while the lower rod is entirely negative. This resembles the initial state of the rod, but is not identical with it, because the initial state was brought about by an extraneous slow charging source (Holtz machine or Ruhmkorff coil) instead of by the very rapid surging that is going on in the oscillator when it is oscillating with its own natural period. From the condition of initial uniform distribution we have followed the charge and current, by rather large stages of a quarter of a period each, through a single oscillation. The charge on the conductor will continue to oscillate, going through the successive steps several times the accumulation of electricity becoming less and less at each oscillation until the spark extinguishes. Nodes and Loops of Potential and Current. From the pre- ceding discussion it is apparent that the two ends of the Hertz oscillator undergo maximum fluctuations of potential, and are, therefore, loops of potential. The middle of the conductor during the oscillation has no accumulation of charge on it; the potential of the middle, therefore, never rises above zero (after the start), and is a node of potential. On the contrary, the nodes of current are at the ends of the oscil- lator, while a loop of current is at the middle of the oscillator. EXPERIMENTS ON THE DISTRIBUTION OF CURRENT IN AN UNGROUNDED HERTZ OSCILLATOR In the preceding dicussion there was given a theoretical exami- nation of the nature of the potential and the current distribution occurring in a Hertz oscillator. I have recently made a simple 112 WIRELESS TELEGRAPHY experiment that approximately confirms the deductions there given in regard to the current. The Principle of the Experiment and a Description of the Oscil- lator. The principle of the experiment is illustrated in Fig. 78. Instead of making a breach at various points in the oscillator and inserting therein an instrument for determining the current, this current at different points in the oscillator was studied by means of its inductive action on a small neighboring circuit WM placed successively at different positions along the oscillator, as indicated by the dotted squares in Fig. 78. The spark gap of the oscillator is shown at G. The two conductors of the oscillator 00, were each a wire 1 mm. in diameter and 9 meters long, supported hori- zontally 1 meter above the wooden floor of a long room in the third story of the laboratory. The oscillating system was thus at a height of about 10 meters above the surface of the earth, and was probably very little disturbed by the capacity of the earth. The oscillator was supported by three insulating stands, one at the spark gap and one at each end of the wire. The central stand for supporting the spark gap carried also a storage battery and a small Ruhmkorff coil for charging the oscillator. ] FIG. 78. Plan of apparatus for explor- ing current distribution. FIG. 79. Detail of exploring circuit. The Exploring Circuit. An enlarged view of the apparatus, showing details of the exploring circuit by which the measurements were made, is given in Fig. 79. This circuit, shown at the left of the oscillator, consists of a square loop of heavy copper wire, NATURE OF THE OSCILLATION 113 L, 30 cm. on a side, and having in series with it a variable con- denser C and a high-frequency current-reading instrument at 7. I shall now describe the instrument / and the condenser C. Description of the Instrument. The instrument at / as is shown in Fig. 80 consists of a disc of silver, suspended by a fine fiber of spun quartz so as to hang near a small coil of a few turns of wire, with which the disc made an angle of 45 degrees. The disc is at M , and the coil, which in this experiment consisted of five turns of wire wound on a vulcanite tube, is shown at C, Fig. 80. The two ends of the coil are connected to binding posts, by which the coil is put into the circuit. The front of the disc carries a small mirror, enabling the deflections of the disc to be measured by means of a telescope and scale such as is used with delicate galvanometers. The mounting of the instrument is also shown in Fig. 80. The disc is sus- pended in the vertical vulcan- ite tube, which is mounted on leveling screws; the support of the coil is in- FIG. 80. High-frequency dynamometer. Mounting shown serted in the Side at left, suspension at right. o f the vertical tube, and is arranged to be moved in and out by a micrometer screw. This delicate motion of the coil in or out brings the coil nearer to or farther from the suspended silver disc so as to vary the sensitiveness of the instrument, to make it suitable for measuring small or large oscillating currents. The action of the instrument, which we shall call a " high- frequency dynamometer," is as follows: oscillations in the coil induce oscillations in the disc. Between these two sets of oscilla- tions there is a force which causes the disc to tend to set itself at right angles to the coil. 1 The deflections of the dynamometer are proportional to the square of the current through it. 2 1 The principle of this instrument was independently discovered by Dr. Elihu Thomson and by Professor Fleming. The instrument was first shown 114 WIRELESS TELEGRAPHY The Variable Condenser. Returning to Fig. 79, the loop L contains, besides the high-frequency dynamometer /, a variable condenser C. This condenser is of a form much used in wireless telegraphic apparatus, and is described by Korda in German Patent No. 72447, issued Dec. 13, 1893. It consists of two sets of semicircular plates, one set F being connected together and fixed in position, and the other set H being also connected together FIG. 81. Korda air condenser. and capable of rotation about a central axis. By rotating the plates H so as to bring a greater or less area of the two sets of plates into interlapping position, the capacity of the condenser can be varied. The position of the plates H with respect to F can be read on a scale attached to the top plate of H and passing under a fixed pointer. A photograph of a condenser of to be applicable to the measurement of oscillating currents of high frequency by Messrs. Northrup, Pierce and Reichmann, and has been used in the present improved form by the author in a large number of resonance experiments, some of which are later to be described in this volume. 2 A theoretical and experimental proof of this proposition is given by the author in Phys. Review, Vol. 20, p. 226, 1905. NATURE OF THE OSCILLATION 115 this character, with capacity somewhat larger than that of the condenser employed in these experiments, is shown in Fig. 81. Large Current at Resonance. Variations of the capacity of C varies the natural period of oscillation of the condenser circuit, and when this period is made equal to that of the Hertz oscillator OGO, a maximum deflection of the instrument / is obtained, under the action of the oscillation. The resonant condenser circuit when calibrated in terms of wave length is a form of " wave meter." How this calibration is effected will be shown later. Exploration of Current Distribution. Since the wave meter in this form, on account of the instrument /, is not conveniently movable, it was necessary to move the oscil- lator in order to explore the distribution of current in the oscillator. The oscillator, with its exciting induction coil and storage bat- tery, was moved lengthwise, keeping it al- ways the same distance from the wave meter, by means of the vulcanite guides DD of Fig. 79. Readings of the dynamometer were taken for various positions of the oscillator with respect to the wave meter. This was equivalent to moving the wave meter along the oscillator, and the readings of the dynamometer were proportional to the square of the current in the wave meter, and therefore proportional to the square of the current at different points of the oscil- lator; because the induced current, keeping everything else the same, is proportional to the inducing current. The results obtained for the distribution 2 e 8 10 of the current in the oscillator are plotted Relative current in Fig. 82. The curve of Fig. 82 shows that FIG 82. Distribution of current along a y 8 7 6 ! I 4 02 3 2 gi 1 I 1 2 S3 4 5 6 7 8 *, , I N \. \ \ \ V \ \l ~] 2 i / 1 1 / / / ,,' /' 1 Hertz oscillator, as determined by ex- periment. the current in the oscillator is greatest near the gap and falls off to zero at the ends of the oscillator in a manner not very different from that shown in the theoretical drawings of Fig. 77. There is a loop of current in the middle and a node at each end of the oscillator. 116 WIRELESS TELEGRAPHY EXPERIMENTS ON THE WAVE LENGTH OF THE UNGROUNDED HERTZ OSCILLATOR Wave Length of the Hertz Oscillator. Having investigated the distribution of current and potential in an ungrounded os- cillator, let us next inquire what is the wave length of the electric wave emitted by such an oscillator. With an oscillator of two parallel wires near together, as shown in Fig. 83, the length of the wave is very approximately four times the length of one of the wires, GBC. If now we take the two parallel wires, separate them, and extend them out oppositely to each other from the gap so as to form the Hertz oscillator, does the wave length remain the same; namely, X = 4 I, where X is the wave length and I is the length of the half-oscillator? Some theoretical writers (for example, Abraham x ) say that it does remain equal to 4 Z; while Macdonald 2 has computed X in this case to be 5.06 X I. Recently Messrs. Webb and Woodman, 3 for very short oscilla- tors, with a half length I between 1 and 5 cm., have obtained experimentally the relation X = 4.8 I. For oscillators of half length between 1 and 3 meters, F. Con- rad 4 has obtained the values presented in the accompanying table, with the average relation X = 4.24 I. CONRAD'S TABL'E FOR RELATION OF X TO I I Half length of oscillator in meters. X Wave length in meters. X j 1.00 4.20 4.20 1.92 8.00 4.17 2.00 8.40 4.20 2.75 12.0 4.37 3.15 13.4 4.25 Average 4 . 24 The experiments of Conrad and those of Messrs. Webb and Woodman both give evidence of being very careful experiments, 1 M. Abraham, Wied. Ann., Vol. 66, p. 435, 1898. 2 Macdonald, Electric Waves, p. 111. 3 Webb and Woodman, Phys. Review, Vol. 29, p. 89, 1909. 4 F. Conrad, Drude's Ann., Vol. 22, p. 670, 1907. NATURE OF THE OSCILLATION 117 and we must therefore conclude that the ratio of \/l for very short oscillators is greater than for the long oscillator. We are primarily interested in the long oscillators, and in order to extend the experimental records to the case of longer oscillators than those studied by Conrad, I have made a series of measurements with the apparatus of Figs. 78, 79, 80. Calibration of the Wave Meter. The wave meter was calibrated for various adjustments of the condenser C by setting it to resonance with various lengths of the two parallel wires of Fig. 83, as had been previously done by Drude. With the wave meter calibrated to read directly in wave lengths, the parallel calibrating wires were removed, and the Hertz oscillator, consisting of two oppositely extending wires FIG. 83. Parallel- of various lengths (1 mm. in diameter), was brought up near the wave meter, and the wave length produced by the oscillator was deter- mined. The results obtained are given in the following table: AUTHOR'S TABLE OF RESULTS FOR RELATION OF X TO I wire oscillator for calibrating wave- meter for short wave-lengths. I Half length of oscillator in meters. X Wave length in meters. X I 4.0 16.9 4.22 4.5 18.9 4.20 5. 21.2 4.23 5.5 23.2 4.22 6. 24.9 4.15 7. 29.5 4.21 8. 33.6 4.20 9. 38.7 4.23 10. 41.6 4.16 11. 46.1 4.22 12. 49.5 4.13 13. 53.9 4.14 14. 57.5 4.11 15. 63.0 4.19 Average 4.19 The average of the results obtained by the author for the ratio of X to I, namely, \/l = 4.19, for wave lengths between 17 and 63 meters, is a little less than the corresponding ratio, 4.24, ob- 118 WIRELESS TELEGRAPHY tained by Conrad for wave lengths between 4 and 13 meters. The difference is only 1%. From these results we may conclude that the wave length produced by a Hertz rectilinear oscillator is very approximately 4.20 times the length of one limb of the oscillator, provided this limb is greater than 1 meter long and of comparatively small diameter. Let us next see how the vibration of a conductor is modified when one end is connected to earth. ON GROUNDED CIRCUITS Grounded Circuits. Image Theory. Suppose now the lower limb of a vertical Hertz oscillator to be cut away close up to the spark gap and be replaced by a connection to earth. According to electrical theories, if the earth were a perfect conductor, the electrical wave length of the earthed system would be the same as that of the Hertz oscillator, the earth merely taking the place of the other half of the Hertz oscillator. The earthed system, which is a simple Marconi emitter, would have the same distri- bution of current and potential in the antenna as the upper half of the Hertz oscillater had before removing the lower limb. In order to examine this theory, let us confine our attention in the beginning to a receiving station, and suppose that we have there simply an ungrounded rectilinear conductor isolated in space, and placed parallel to the electric force of the incoming waves. Let the length of this straight-line conductor be so chosen that its natural period of electric oscillation is equal to the period of the waves. The distribution of current in the conductor would re- semble that shown in Fig. 82. It we could introduce a current reading detector into the circuit without disturbing the conditions, the instrument would give a maximum reading when placed at the center of the receiving conductor; this is the point at which the fluctuation of potential is zero. Suppose with such an instrument in the circuit we should cut away the lower half of the conductor; the reading would become zero, because there would be no capacity out beyond the instru- ment into which the current could flow. If now a capacity is attached to the instrument in the place of the removed conductor, some current would flow between the straight wire and the capacity and register in the instrument. NATURE OF THE OSCILLATION 119 If the capacity attached were very large (e.g. the earth), the point of zero fluctuation of potential would again be brought near the instrument, because a large fluctuation of potential cannot occur in a very large capacity under the action of the currents with which we are concerned. We should, therefore, have the same current as when the conductor was made up of two parts symmetrical about the instrument. In actual systems, the grounding may be imperfect. In that case the symmetrical image would give only approximately an equivalent system. I have made some experiments to test this image theory of the action of the ground connection. The experiments consisted in comparing resonance curves taken with various forms of grounded circuits with the corresponding resonance curves taken with an image circuit in the place of the ground. Two of these experi- ments are here briefly described. EXPERIMENTS TO TEST IMAGE THEORY OF THE GROUND Experiment I. The Aerial Circuit and its Image Tuned by Variable Inductances. In testing the image theory of the action of the ground at the receiving sta- tion the form of circuit shown in t -in mm FIG. 84. Circuit employed in study of the image theory of the ground. 1234567 Inductance x!0' 5 Henry FIG. 85. Resonance curves in study of the image theory of the ground. Curve H was obtained with horizontal duplicate of antenna; curve G, with ground. Fig. 84 was employed. The high-frequency dynamometer de- scribed on p. 113 was used for detecting and measuring the minute oscillating currents at the receiving station, and was placed at / 120 WIRELESS TELEGRAPHY in series with a variable inductance and a vertical antenna 23.2 meters long. This aerial system, by means of a switch at S, could be connected to the ground G, or the ground could be thrown off and replaced by metallic parts duplicating the aerial system. The duplicate was, however, not an exact theoretical image of the aerial system, because the second antenna had to run off horizontally instead of straight down. The horizontal wire was made equal in length to the vertical antenna, 23.2 meters, and was supported about 1 meter from the ground by cords attached to posts. In series with the horizontal wire was a variable inductance L f duplicating the tuning coil L, and a small coil /' of fine wire duplicating the coil of the receiving instrument. Curves giving the results of the experiment are shown in Fig. 85. For curve G the grounded circuit was used, and deflections of the receiving instrument were taken for various values of the inductance of the tuning coil L; the deflections are plotted against values of L. The switch S was then thrown so as to connect the receiving circuit to the horizontal system instead of to the ground. With this arrangement the curve H was obtained. In taking this curve, the tuning coil L and its image L' were kept identical and varied together. The curve H, therefore, shows the deflections of the re- ceiving instrument plotted against the common values of L and L'. Discussion of Results 'in Experiment I. The two curves G and H of Fig. 85 are seen to have their maxima for the same value of inductance. That is, a given value of inductance, 2.1 X 10 ~ 5 henries, gives a maximum deflection in the case of the grounded circuit. To obtain a maximum with the duplicated system the same inductance 2.1 X 10 ~ 5 henries must be used in both the vertical circuit and in its horizontal duplicate. The result is a confirmation of the image theory of the grounded circuit. The earthing of the circuit gives it the same period of vibration as the duplication of the aerial system gives. It is interesting to note that the deflection (current square) is about 20% larger with the duplicated system than with the grounded system a fact that may be accounted for by supposing a higher resist- ance with the grounded system than in the wholly metallic system. Some other facts in regard to the experiment are discussed in the original publication. 1 1 G. W. Pierce: Resonance in Wireless Telegraph Circuits. Part IV, Physi- cal Review, Vol. 22, p. 174, 1906. NATURE OF THE OSCILLATION 121 The curve F, with which we are not here concerned, was obtained with the duplicate antenna wound around the house of the receiv- ing station. Experiment II. Quarter- Wave Ground. What was perhaps a more interesting experiment confirmatory of the image theory of the ground was made by replacing the ground by a horizontal wire of which the length could be varied. The relative amounts of energy received (deflections) for different lengths of the horizontal wire are shown in the curve A of Fig. 86. Resonance was obtained when this wire had the length of 38 meters, which was very close to one- fourth the wave length (155 meters). The ground gives the system the same period as an added quarter- wave wire gives the system. Curve B obtained with different conditions leads to the same results. Conclusion from the Experiments. These experiments I and II show that the effect of the ground, so far as concerns the vibration in the an- tenna, is to introduce into the circuit at the ground a point of zero fluctu- ation of potential, an effect that can also be obtained with an arti- ficial ground consisting of a sym- metrical duplicate of the aerial system or consisting of a horizontal wire not far from the earth and of length equal to one-quarter of the wave length to be received. Professor Ferdinand Braun at a date earlier than that of my experiments has suggested the use of horizontal wire in replace- ment of the ground and also the use of a capacity consisting of a large cylindrical conductor in the place of the ground. He has not, however, so far as I know published any quantitative results on the subject. 30 40 50 Length of Horizontal Meters FIG. 86. Showing that the ground may be replaced by a quarter-wave wire. CHAPTER XV PROPAGATION OVER THE EARTH The Propagation of the Waves over the Surface of the Earth. In the preceding chapter it was shown that so far as concerns the wave length and the distribution of current and potential, the grounding of an antenna was equivalent to attaching it to its image. Let us discuss further this idea with reference now to the man- ner of propagation of electric waves over the surface of the earth. Does the earth contribute to the propagation of the waves in an advantageous or only in a detrimental way ? We shall begin this discussion with the assumption that the earth is a perfect conductor. With this assumption we can apply to the problem further reasoning based on Sir William Thomson's theory of electrical images. Theory of Images. Suppose we have two small bodies A and B with equal charges of electricity of opposite signs. The direc- tion of the electric force between A and B is represented by the curved lines in Fig. 87. A plane P drawn every- where equally distant from A and B will be a surface of zero potential. The proof that P is a surface of zero potential is as follows: The potential of a point is the work re- quired to be done in order to bring a unit positive charge up to the point from an infinite distance. Now a unit charge of positive electricity can be brought up to any point of the plane P without doing any work; because the force at any point of the plane, being made up of an attraction F due to B and an equal repulsion F f due to A, exerts a force perpendicular to the plane, but no force along the plane. The force is therefore per- 122 FIG. 87. Lines of electric force between two oppositely charged bodies A and B. The plane P is at zero potential. PROPAGATION OVER THE EARTH 123 pendicular to the direction of motion when the charge is brought up along the plane, and the work done is therefore zero. For further details in regard to work and potential see Appendix I. Having shown that the plane P is everywhere at zero potential, let us next introduce the idea well established in treatises on elec- tricity, that so long as we keep the potential of the plane P equal to zero the electric force in the region between A and the plane P is completely fixed, no matter what changes we may introduce below the plane. If, then, the lower half of the diagram is removed and the plane is in some other way kept at zero potential, the electric force between A and the plane will be the same as before; namely, that represented in Fig. 88, which is the upper half of zot. o FIG. 88. Lines of electric force between a charged body A and an infinite conducting plane kept at zero potential. Fig. 87. We may keep the plane at zero potential by grounding it so that it comes into coincidence with the surface of the earth; or the surface of the earth itself may take the place of the plane, provided the earth for a considerable area around the charged body A is a good conductor. That is to say, if the earth's surface is a good conducting plane for a considerable extent, and a charged body A be placed above the surface of the earth, the field of electric force between A and the plane surface of the earth will be the same as the upper half of the field between A and a body B, which has a charge equal to A and opposite in sign, B being at the distance below the plane that A is above it. This equal opposite charge symmetrically placed in regard to the plane is called the electrical image of A in the plane. Similar Theory Applied to the Oscillator. If we next consider the case of the electric oscillator, the field of electric force for the symmetrical oscillator, as we have seen in Chapter VIII, is roughly that represented in Fig. 89. The ideal, nonmaterial plane PP through the figure is at zero potential, so that the lower half of the diagram could be replaced by the surface of the earth, if it were plane and perfectly conductive, without disturbing the upper 124 WIRELESS TELEGRAPHY half of the figure. Whence it follows that the oscillation and radia- tion from an oscillator grounded to an infinite, plane, perfect conductor is the same as the oscillation and radiation of the upper half of a symmetrical Hertz oscillator. The nature of the wave sent out from an oscillator so grounded is represented in Fig. 90. FIG. 89. Lines of electric force about a Hertz oscillator. FIG. 90. Lines of electric force about a half-oscillator discharging to a perfectly conductive ground. Guided Electric Waves. Figure 90 shows approximately the theoretical mode of propagation of the electric wave over those parts of the surface of the earth where the earth is a good conduc- tor. The loops there shown receding from the oscillator are lines of electric force. Now a line of electric force must be either a closed line, as in Fig. 89, or must terminate at one end on a positive charge and at the other end on an equal negative charge. There is, therefore, a series of successive positive and negative charges induced in the surface of the earth, and these positive and negative charges move with the wave with the velocity of light. The earth thus serves as a guiding conductor and causes the loops of electric force terminating on it to follow the surface of the earth. This accounts for the fact that communication is possible between sta- tions which on account of the intervening curvature of the earth PROPAGATION OVER THE EARTH 125 are not visible from each other. We have here a simple view of the matter, obtained on the assumption that the earth is a perfect conductor. The Earth not a Perfect Conductor. The surface of the earth is, however, not everywhere a good conductor of electricity. The sea and moist soil are better conductors than dry stone. In some places the surface materials of the earth are in fact good insulators. The attenuation of the electric wave is on this account very different over different parts of the surface of the earth, condi- tioned on the fact that there is a greater or less penetration into the insulating portions and a greater or less absorption of energy at the poorly conducting portions. This subject has been sub- mitted to a very remarkable mathematical treatment by Dr. Zen- neck. The mathematical reader is referred to Dr. Zenneck's paper 1 or to Professor Fleming's 2 translation and " free para- phrase " of it, for a beautiful discussion of this interesting question. I shall attempt to give here a brief statement of some of Dr. Zenneck's results without attempting to present his reasoning. In doing this I wish to acknowledge the assistance afforded by Pro- fessor Fleming's excellent commentary on Zenneck's paper. In order to simplify the matter, Dr. Zenneck at' first considers only the case of a plane electric wave traveling without divergence over a flat surface. He is thus at first leaving out of account the spreading out of the wave and the consequent diminution of ampli- tude by mere distance; and he is also omitting the attenuation of the wave due to the curvature of the surface. Instead of considering the earth to be a perfect conductor, as has usually been done before, Zenneck looks upon the boundary between the earth and the air as the boundary between two media of different conductivities and different dielectric constants; and he transforms Maxwell's equations so as to take account of the two media. He arrives at the conclusion that where the earth is a good con- ductor (for example, sea water), the electric force (at the surface) is perpendicular to the surface. For waves of wave length 600 meters, which is the wave length used in most of the calculations, sea water acts as a good conductor, and the electric force at the surface of the sea is perpendicular to the surface, as is shown in 1 J. Zenneck: Annalen der Physik, Vol. 23, 1907. 2 Fleming: Engineering (London), June 4 and 11, 1909. 126 WIRELESS TELEGRAPHY diagram (a), Fig. 91. This figure represents merely how one side of one loop of electric force of our Fig. 90 comes down to the sur- face. The other side of the loop would likewise be perpendicular to the surface of the sea, but would have an opposite sign. There would thus arrive at a station at sea a train of electric waves, and the electric force would be vertical, and would go through a series of continuous oscillations between positive and negative values, with the frequency of the waves; that is, a train Air FIG. 91. Diagrams taken from Professor Fleming's paper in the Electrician, illustrating Dr. Zenneck's Theory. FIG. 92. Diagram of the electric force in a wave train. like that of Fig. 92 would come along near the surface in the medium above the surface, and would affect the antenna first in one direction and then in the opposite direction. In the sea water itself beneath the surface, the forces would be zero. Still confining our attention to a wave of wave length 600 meters, let us next suppose the wave to be traveling along a surface of dry rock of resistance 100,000 ohms for a meter cube and of dielec- tric constant k = 2 to 3. Zenneck finds for this case that the electric force in the air above the earth is by no means perpendicu- lar, but leans forward in the direction of travel; and that not only the magnitude of the force changes, but the inclination also changes as the wave progresses. There is a similar force, although differently inclined and of different magnitude, below the surface in the rock itself. This condition Zenneck finds to be represented by the semiellipses of (6), Fig. 91. The electric force is obtained in magnitude and direction from this diagram (6) by PROPAGATION OVER THE EARTH 127 considering a radius drawn from the center of the ellipse to a particle moving around the ellipse with the frequency of the wave. The length and the direction of the radius so drawn would repre- sent the changing magnitude and direction of the electric force. Such an electric wave, oscillating both in magnitude and direction is equivalent to two waves, one tending to produce vertical cur- rents and the other tending to produce horizontal currents (the two effects being also out of phase with each other). The hori- zontal oscillating force induces currents in the earth's surface, and diminishes the energy of the progressing wave, so that in this case the distance to which signals can be sent is less than in the case of the good conductor. In the case of propagation over very dry soil, which is not so good an insulator as the rock (r = 10,000 ohms per meter cube, k = 1 to 3) Zenneck finds the result represented in diagram (c), Fig. 91. X=Distance in Kilometeres at which the "Wave .Amplitu is reduced to 0.367 =*y e of .Amplitudejlt Origi _p H- o 8 o S < \ / M 10 01 g g g BielectEic Constant \ ~7 \ / / X \ ^^ ^s ^^ ^ k 'x A a- ya+ 5 s 1 - 00 33 S'S.so a ^ ^== ^^ ^x ^ ^ & 2i:30 5: 7:30 10: 12.:30 3: 5:30 8: 10:30 Hour A..M. Hour P.M. FIG. 98. Rate of dissipation of electricity at different hours of day and night (Zoelss). a were obtained when the body was charged positive or negative respectively. Curves of this character, although they differ at different places on the earth, usually show a minimum dissipation of electric charge at sunrise and a little after sunset, a maximum near noon, and a second maximum near midnight. This would correspond to good transmission at sunrise and sunset and poor transmission at noon and at midnight, which does not accord with the facts. Effect of lonization of Upper Strata. The action of the sun's light in ionizing the air ought to be much greater in the upper PROPAGATION OVER THE EARTH 139 regions of the atmosphere than at the surface of the earth, because the chief ionizing rays of light are those of very short wave length (the ultraviolet), and these short waves of light are strongly absorbed by the air, and therefore do not penetrate to a very great depth in the earth's atmosphere. The stratum of upper atmosphere, rendered conductive by the sunlight, may serve to some extent as a reflector of the electric waves so as to assist in confining the waves to the surface of the earth. If this effect were appreciable, the waves would be more strongly confined to the surface of the earth in the daytime than in the night, and trans- mission would be easier in the daytime than at night, except for a possible interference between the direct and the reflected wave. This interference, if it should exist, would intensify waves of some wave lengths and partially annul waves of a different wave length, so that by changing the wave length through a range correspond- ing to a half period it ought to be possible to turn the interference to advantage. No such effects have been found, and the increase of the conductivity of the upper air by ionization in daylight when looked upon as a reflector does not act in the proper direction to be the determining factor in explaining the inequality of transmis- sion of electric waves by day and by night. Professor A. E. Ken- nelly has called my attention to the fact, however, that there may exist in the upper strata, as we pass upward, a gradual change from insulating to good conducting strata, which, coupled with irregularly distributed conducting areas, might result in a general deflection upward of the waves, and a consequent loss of received energy, and that this effect might be greater in daylight than at night. This theory has not yet been given exact mathematical expression, so that up to the present we seem not to have found an adequate explanation of the difficulties of daytime transmission in comparison with night transmission of electric waves to great distances. The question is one of great importance from a theo- retical standpoint, and if the discovery of the explanation of the phenomenon should bring with it the discovery of a means for bringing the distance of communication by daytime up to that by night, it would remove a very exasperating limitation to electric wave telegraphy. Experiments with the use of very long electric waves are under way by the National Electric Signaling Company and by the Marconi Company, and it is reported that some approach toward uniformity of day and night transmission has been made. CHAPTER XVI ON DETECTORS HAVING examined at some length various problems in connection with the propagation of electric waves to great distances over the surface of the earth, let us take up next a description and exami- nation of some of the instruments used in receiving the oscillations of wireless telegraphy and translating them into audible or visible signals. The instruments employed are the indicating instrument (relay, galvanometer, telephone, etc.), by which the signals are read, and the detector, by which the high-frequency oscillations are put into a condition to affect the indicating instrument. INDICATING INSTRUMENTS Classification of Indicating Instruments. The indicating in- strument now usually employed is (1) a sensitive telephone receiver, but (2) a relay, in connection with a sounder or ordi- nary telegraphic recording instrument, (3) a galvanometer, or (4) an electrometer, may serve as indicating instrument. Sensitiveness of Relay. The most sensitive relay will trip with about one one-thousandth of a volt e.m.f . applied to its ter- minals. With the restoring spring of the instrument set under sufficient tension to act reliably and rapidly enough to receive messages, a relay (even when constructed to have high sensitiveness) would require perhaps one two-hundredth of a volt to operate it. Sensitiveness of Telephone Receiver. Dr. L. W. Austin l has recently made some experiments on the volt sensitiveness of a pair of 800-ohm Schmidt- Wilkes head telephone receivers, such as have been very much employed in recent electric- wave telegraphy. In stating the sensitiveness of a telephone receiver it is necessary to specify the frequency, because the sensitiveness depends very markedly on the frequency of the e.m.f. applied to the circuit. This is no doubt largely due to the possession by the diaphragm 1 Bulletin of the Bureau of Standards, Vol. 5, p. 149, 1908. 140 ON DETECTORS 141 of a natural period of vibration. The following table (Table II) taken from Dr. Austin's paper gives the number of volts required to produce just audible sounds in the pair of telephone receivers under the application of sinusoidal electromotive forces of various numbers of complete cycles per second. * TABLE II. VOLT SENSITIVENESS OF A PAIR OF SCHMIDT-WILKES 800-OHM TELEPHONES. No. of cycles Volts to produce audible per second. sound. 60 620 millionths of a volt. 120 290 180 170 300 60 420 17 540 8 660 3 780 1.1 900 0.6 Sensitiveness of Galvanometers. A very sensitive galvano- meter of ordinary construction and of about 1000-ohms resistance will give a visible deflection with less than one ten-millionth of a volt, but such an instrument has too slow a period (ten seconds) to use in indicating wireless telegraph messages. In 1903 Profes- sor Einthoven 1 designed a new form of galvanometer that has a very rapid period and at the same time a high sensitiveness. Einthoven's instrument consists of a very fine silvered or platinized quartz fiber hung between the poles of a strong magnet. The current to be measured is sent through the silver or the platinum coating on the fiber, and the fiber tends to move out of the mag- netic field. The deflections of this fiber may be observed with a microscope, or may be photographed on a rotating drum carrying a photographic film. The direction of the deflection of this galva- nometer, like that of the ordinary galvanometers, reverses with reversal of the current. In one one-hundredth of a second Ein- thoven's instrument will give a deflection sufficiently large to be registered on the photographic plate, under application of an e.m.f. of one ten-thousandth of a volt. Used in connection with a suitable 1 Annalen der Physik, Vol. 12, p. 1059, 1903. 142 WIRELESS TELEGRAPHY detector it is, therefore, adapted to the photographic registration of wireless telegraph messages, and has been employed for this purpose. Sensitiveness of the Capillary Electrometer. A very minute column of mercury in a capillary glass tube and in contact with sulphuric acid is employed in the construction of a capillary elec- trometer. Under the action of a current, the electrolytic polariza- tion of the contact causes a change of the surface tension of the mercury and causes the column of mercury to rise or fall in the glass tube. This minute motion of the mercury column is observed with a low-power microscope. A delicate capillary electrometer will give a readable deflection with an applied electromotive force of one ten-thousandth of a volt, and is capable of use as an indicating instrument. In what follows I shall describe the method of employing some of these indicators in connection with detectors for rapid oscillations. Why a Detector in Addition to the Indicating Instruments Must be Employed. Some misconception exists as to why a detector must be employed with these various indicating instru- ments in order to receive and read the messages. The misconcep- tion is that the detectors are more sensitive to electrical energy than the telephone receiver or galvanometer is. This is not the case. But in the reception of the electric waves the electrical energy received, being in the form of rapid oscillations, cannot affect the telephone or the galvanometer. These rapid oscillations cannot affect the galvanometer because the deflections of the galvanometer reverse with reversals of the current, so that the deflecting impulses, if applied directly to the galvanometer, would be first in one direc- tion and then in the other, with a frequency of the order of a mil- lionth of a second, and motion of a mass as light even as the fiber of the Einthoven galvanometer could not result from these rapidly reversing impulses. Likewise, a telephone diaphragm could not be made to move with such rapidity. In the case of the telephone, on account of the large self -inductance of the instrument, the high- frequency e.m.f . generated by the waves would produce in a circuit containing a telephone receiver only extremely weak currents. The use of the detectors is to transform these rapid oscillations into effects that can be manifested by the indicating instruments. How this transformation is accomplished will be explained in the subsequent discussion. ON DETECTORS 143 CLASSIFICATION OF DETECTORS We shall describe the detectors under the following more or less arbitrary titles: Coherers. Magnetic Detectors. Thermal Detectors. Crystal Rectifiers. Electrolytic Detectors. Vacuum Detectors. In illustrating the manner of introducing these various detectors into the receiving system a diagram of only a simple form of receiv- ing circuit will be exhibited with the descriptions. It is to be understood, however, that all the detectors can also be used in various forms of direct and inductively connected circuits as well as in the simple circuits. COHERERS As coherers, we shall include only those detectors which employ a loose contact and require to be shaken, tapped, or otherwise moved to restore the contact to its sensitive condition after the receipt of a signal. We have already described the filings-tube coherer of Branly and Marconi. A great many modifications of this instrument have been made, including the use of a single contact or a few contacts in series or parallel, between metallic balls or points, to take the place of the filings. Also a great many variations in the method of decohering the contacts have been made. These will not be described here. These various forms of coherer have their importance in the fact that, on the receipt of electric waves, a sufficiently large cur- rent is started in the local circuit to operate a relay, ring a bell, or give other form of alarm that can be heard at a distance from the operator's desk. Also the current permitted to flow in the local circuit of the coherers during the receipt of electric waves is sufficiently large to start machinery and control a mechanism (for example, a torpedo or dirigible craft) at a distance. This kind of result is not easily attained with the other form of detectors listed above, which do not permit of the use of sufficiently large currents in the local circuit to sound an alarm or start electrical machinery. 144 WIRELESS TELEGRAPHY Thus the coherer, though lacking in sensitiveness to feeble waves and not now generally employed in the receipt of messages, has still a field of usefulness. Besides the filings coherer described in Chapter XII, we shall describe here another interesting form of coherer, that devised in 1902 by Lodge, Muirhead and Robinson. The Lodge-Muirhead Coherer. This instrument consists of a small steel disc A (Fig. 99), rotated by a clockwork, so that the disc is just separated from a column of mercury B by a thin film of mineral oil on the surface of the mercury. One electrical con- nection is made to the wheel through a brush E, the other con- nection is made to the mercury well through the binding post H. The impulse of the electric oscillations breaks down the oil film and establishes momentary cohesion between the steel disc and the mercury. A current from a local battery passes through the disc and mercury contact, and operates a siphon recorder, which is used in series with the battery and the coherer. After the impulse ceases the motion of the disc brings continuously a fresh oil film into the contact and causes de- coherence. The siphon recorder gives a written record of the dots and dashes of the message. A .felt brush at K serves to keep the rotating disc free from dust before and after contact with the FIG. 99. The Lodge-Muir- head-Robinson coherer. mercury. Concerning the Theoretical Explanation of the Action of the Coherers. A generally accepted theory as to the reason for the coherence of the filings, or other form of imperfect contact used in the coherers, has not been established. I shall state briefly some of the views presented in explanation of the phenomenon. Before the arrival of the waves, the high resistance of the contact is generally supposed to be due to the presence of some kind of poorly conductive film at the contact. In the case of the Lodge- Muirhead coherer, the insulating film is evidently present in the form of a film of oil. In many of the other coherers a poorly conductive film is present in the form of an oxide of the metal. ON DETECTORS % 145 This is evident from the fact that in some cases the metallic par- ticles (e.g., iron or steel) are artificially prepared by oxidizing them in order to make of them a good coherer. The poorly conductive film may also be present in some cases in the form of a sulphide of the metal. On account of the readiness with which many metals (called the " baser metals ") enter into combination with the oxygen or sulphur dioxide of the air, a thin film of oxide or sulphide is always present on the surface of most of the baser metals, unless special care is taken to remoVe it. Apart, however, from the existence of such films of foreign matter at the contact, it seems not impossible that the high resistance before the arrival of the waves may be a property of the surfaces of even pure metals when these surfaces touch only very lightly. If we assume the presence of the poorly conductive film at the contacts of the coherer, we may suppose that, on the arrival of the electric waves, the poorly conductive film is removed by the heat developed by the oscillatory currents. This starts the local current, which, developing further heat, still further improves the contact and permits the passage of further current. Instead of heat being the chief agency in removing the oxide or other poorly conductive film, or in bringing together the loose contacts, it may be that this is done by the electric attraction between the filings, which before the current starts will be charged with opposite signs of electricity, and which under the added e.m.f. produced by the electric oscillations may attract each other strongly enough to pull the contacts together. We shall learn more about the electrical properties of high resist- ance contacts when we come to the study of crystal rectifiers. It is therefore proposed to omit further discussion of the specific action of the coherers, because of the more general character of the information to be presented later. In the meanwhile some of the other detectors which do not depend on the properties of a loose contact are discussed. MAGNETIC DETECTORS Rutherford's Magnetic Detector. In 1895 and 1896 Pro- fessor E. Rutherford 1 discovered a sensitive method of detecting electric waves by causing the electric oscillations set up by the 1 E. Rutherford, "A Magnetic Detector of Electrical Waves and Some of Its Applications." Phil. Trans. Roy. Soc. London, 1897, Vol. 189, A., p.l; also Proc. Roy. Soc. London, 1896, Vol. 60, p. 184. 146 WIRELESS TELEGRAPHY waves to demagnetize a bundle of fine steel wires. This bundle of steel wires consisted of about twenty pieces, each 1 cm. long and .007 cm. in diameter. The individual wires were insulated from one another by shellac varnish, and the bundle was placed within a small coil of about 80 turns of insulated copper wire. The bundle of steel wires was magnetized by the use of a magnet, and was then brought up near a magnetometer, consisting of a small compass needle suspended by a fine fiber and carrying a small mirror by which its deflections could be read. The needle of the magnetometer was deflected by the magnetized bundle of steel wires. If now electric oscillations were passed through the coil surrounding the bundle of steel wires, these wires lost some of their magnetism, which was shown by a diminished deflection of the neighboring magnetometer. Rutherford found that by connecting the coil around the wire bundle to a resonator, electric waves from a small Hertz oscillator placed at a distance of a half mile across the city (Cambridge, England) could be detected. With this instrument Rutherford performed many interesting experiments and carried out an important research on the damping of electric oscillations. Marconi's Continuous Band Magnetic Detector. In 1902 Marconi devised two other forms of magnetic detector, one of which has met with extensive use in practical wireless telegraphy, and is here described. Reference is made to Fig. 100. A band made up of a bundle of fine, hard-drawn iron wires, insulated from one another to prevent eddy currents, is carried on the periphery of two wooden discs, one of which is turned by a clockwork or a motor, so that the band moves at the rate of 7 or 8 cm. per second. This endless band of iron wire passes axially through a small glass tube g, around which two coils are wound. One of these coils, b, is connected into the oscillation circuit. In the example shown, the receiving circuit is of the simple type consisting of antenna, detector and ground. In this case the coil b is put directly into the antenna circuit, so that electric oscillations from the antenna, A, pass through this coil of the detector. We shall call the coil b the oscillation coil of the detector. Around the oscillation coil is a second coil, (7, connected in series with a telephone receiver. To produce a state of magnetization in the moving band, two permanent horseshoe magnets are placed near it. Two like poles, NN, of the magnets are placed above the center of the oscillation coil, and the other two poles, SS, are placed near the ON DETECTORS 147 band where it approaches and leaves the coils. These magnets induce magnetic poles in the moving band. One of these induced poles, say the South pole, is within the coils, and the two other FIG. 100. Marconi magnetic detector. consequent poles (North poles in our illustration) are near the point where the band enters and leaves the coils. General Facts in Regard to the Explanation of the Action of the Marconi Magnetic Detector. If we confine our attention to a point on the moving band, it is seen that, as the band moves forward, this point becomes a North pole outside the coils, changes to a South pole within the coils, and becomes again a North pole after issuing from the coils. There is, however, -within the coils, a steady state of magnetization, for although the band is in motion, every particle of the band, as it passes a particular point within the coils, comes to a particular state of magnetization, so that the magnetic condition is fixed with respect to the magnetizing mag- nets. This gives a steady state of magnetization within the coils and produces no inductive effect in the form of currents in the telephone circuit. If now a train of electric oscillations passes through the oscilla- tion coil b, the magnetization of the part of the band within the 148 WIRELESS TELEGRAPHY coil is changed, and this change of the magnetization produces a transient current in the coil C, and the telephone gives a click. A whole series of trains of electric oscillations gives a series of clicks, producing a musical note with a pitch depending on the frequency of arrival of the trains; and this is the frequency of the sparks at the sending station. So that one hears, when listening into the telephone attached to the magnetic detector, a sound like that produced by the spark at the sending station. The pitch of this sound is determined by the period of the vibrator of the sending induction coil; or, in case an alternating current transformer is used to charge the sending antenna, the fundamental pitch of the spark, and consequently the note that one hears at the receiving station, is determined by the number of reversals per second of the alternating current supply at the sending station, although other notes may be superposed on this fundamental note, due to the fact that with some adjustments more than one spark at the sending station occurs at each reversal of the alternating source. We shall now discuss the nature of the change occurring in the magnetization of the iron band of the detector under the action of the oscillations set up by the incoming waves. The very rapid oscillations produced by the electric waves used in wireless teleg- raphy cannot produce a sound in a telephone either when applied to it directly or inductively, because, on account of the self- inductance that is necessary to the telephone, these very rapid oscillatory currents cannot traverse its circuit. If they could tra- verse its circuit, the diaphragm of the telephone could not take up such rapid vibrations, and if it did we could not hear them, for the highest note audible to the human ear makes only 35,000 vibrations per second. Our wireless telegraph detectors must be so constructed that the rapid oscillations of a train of waves act integratively upon it, so that the train produces a single response in the telephone; 1 and a series of trains produce a series of responses. This series of responses we can hear in the telephone, because the series of trains of waves follow each other with a periodicity that is audible. In regard to the manner in which a train of oscillations act integratively upon the magnetized moving iron band of Marconi's form of the magnetic detector, I shall present a few paragraphs of explanation. Explanation Assuming a Suppression of Hysteresis by the Oscillations. Many experiments have been made in the effort ON DETECTORS 149 to discover just what is the effect produced on the magnetization of the bundle of iron wires by the oscillations within the coil surrounding the bundle. A steady current in the coil would magnetize the iron wires of the bundle. An oscillatory current, according to the experiments of C. Maurain, l produces a suppres- sion of hysteresis in the iron. In explanation of the term " hysteresis," reference is made to Fig. 101, in which magnetizing force is plotted horizontally and the magnetization produced by it is plotted 'vertically. This curve represents the hysteresis in a specimen of hard-drawn iron wire such as is used in the magnetic detectors. If we start with the magnetizing force equal zero, and increase it to OL, the magnetiza- tion follows the curve OA. If now we reduce the magnetizing force gradually to zero, the magnetization follows the curve AC. That is, the state of magnetization produced by the magnetizing force when it is decreasing is not the same as the state of magneti- zation produced by the force when it is increasing, and after the force is removed, some magnetization represented by OC is left in the specimen. In order to reduce this magnetization to zero, it is necessary to apply a reversed magnetizing force OD. If we go on increasing the reversed magnetizing force to OM, the mag- netization follows the branch DE of the curve. On decreasing and again reversing the force, the magnetization traces out the branch EFGA. The complete diagram is called a hysteresis cycle. Hysteresis is the property of iron, steel and other magnetizable metals characterized by the fact that the change in magnetization due to the application of a magnetizing force depends on the pre- vious state of magnetization of the specimen. The state of mag- netization assumed by a specimen when the magnetizing force is gradually removed is not the same as the state of magnetization assumed by the specimen when the force is gradually applied. The magnetization produced by a given magnetizing force is not completely annulled by withdrawing the magnetizing force. The hysteresis effect is small in very soft iron, is increased by harden- ing the iron, and is very great in glass-hard steel. According to the experiments of C. Maurain, which we are now discussing in their application to the magnetic detector, the super- position of a sufficiently strong oscillatory magnetizing force upon a slowly varying magnetizing force causes a suppression of the hysteresis in the specimen. If the oscillatory force is weak, the 1 C. Maurain, Comptes Rendus, Vol. 137, p. 914-916, 1903. 150 WIRELESS TELEGRAPHY suppression is only partial, giving for the specimen characterized in Fig. 101 a diminished hysteresis, such as is represented in Fig. 102. M 7 rnetising Force agnetising Force FIG. 101. Hysteresis curve. FIG. 102. Hysteresis curve. In terms of this result we have a possible explanation of the magnetic detector. Reference is made to Fig. 103. With the poles FIG. 103. Diagram in explanation of Marconi magnetic detector. of the permanent magnet in the positions SNNS, the mag- netizing force acting on the band will be positive under the ON DETECTORS 151 South poles and negative under the North poles; and following our usual method of plotting, the magnetizing force can be repre- sented approximately by the dotted wavy curve H of Fig. 103. Now if we suppose the band to be moving in the direction of the arrows, the North magnetization under the first South pole will not follow the curve of force, but will persist, and follow approxi- mately the continuous curve B. If now oscillations produced by the electric waves are allowed to flow around the oscillation coil, the hysteresis in the band is suppressed, so that the curve of magnetization B falls back into the position B', which is nearer the curve of magnetizing force H of Fig. 103. This change from the condition B to B 1 is equivalent to a motion toward the left of the magnetic distribution in the coil, and therefore induces a current in the coil containing the telephone in circuit. When the waves cease, the state of magnetization returns to that repre- sented by the curve B. We have thus with each train of waves a back and forth shift of magnetization of the band, and conse- quently a to and fro motion of the telephone diaphragm. While this description of the process seems a very reasonable explanation of the action of the detector, yet, for the benefit of those readers who may wish a little more insight into the processes occurring in iron or steel submitted to an oscillatory field, I beg leave to present a brief account of some experiments by E. Made- lung, in which he made direct observations of the effect of electric oscillation on the magnetization of iron and steel. Experiments of E. Madelung. A very comprehensive and beautiful series of experiments On Magnetization by Rapid Oscilla- tions, and on the Operation of the Rutherford-Marconi Magnetic Detector has been made by E. Madelung, and described in his Gottingen Dissertation. 1 By means of a very ingeniously devised application of Braun's cathode tube, Madelung was able to obtain on a fluorescent screen the hysteresis cycle produced by a slowly varying magnetic force, and to obtain also the effect produced on this hysteresis cycle by superposing the rapidly oscillating magnetic force pro- duced by sending a condenser discharge through the magnetizing coils. , Reference is made to Fig. 104. I. With a slowly varying magnetizing force the hysteresis cycle EAKFGE was described. II. Upon slowly applying and withdrawing a magnetizing force 1 E. Madelung: Drude's Annalen, 1905, Vol. 17, p. 861. 152 WIRELESS TELEGRAPHY OM the curve AKC was obtained. III. On applying in the coil surrounding the specimen a rapidly oscillating electric current, giving a magnetizing force of initially the same amplitude OM and falling off in amplitude by damping, the spiral curve AD was described. IV. Applying a second oscillation gave a similar spiral starting with the arc DJ. The complete spiral for this case is not drawn; it is like that of AD, but is somewhat lower down. V. Applying more of these oscillations brought the spiral down into the position L, after which further oscillations simply caused the magnetization to describe over and over the closed spiral FIG. 104. Dr. Madelung'e curve showing effect of rapid oscillations on magnetic hysteresis. path L. The path L is thus the limit of the condition attained by the specimen when several oscillations are applied. Thus a series of oscillations applied to the specimen originally in the state A reduced its magnetization to the state L. The jump from A to L is the demagnetization effect of the oscillation, which was first utilized in the construction of a detector for electric waves by Rutherford. Suppose now that these oscillations be applied to the specimen when it is in various different states of magnetization; Madelung found the effect shown in Fig. 105. Applied at A, the effect was a change from A to B] applied at C, the specimen, after the oscil- lation, was left almost in the state C unchanged; applied at D, the effect was a change from D to E. The effect of the oscillating field is thus a hastening of the progress of the cycle in the direction it was already going under the action of the slowly varying field. ON DETECTORS 153 A suppression of hysteresis would attain the same end results, but instead of being contented with calling the effect " suppression of hysteresis," which is a purely negative account of the phenomenon, Madelung, by his delineation of the spiral course taken by the magnetization during the application of the oscillating magnetic force, has given us a very distinct picture of the active processes occurring in the specimen. He has shown that the magnetic state of the iron has been violently agitated by the oscillating mag- netic force, and in this way the sluggishness of the specimen in following the slowly changing magnetic force has been overcome. FIG. 105. High frequency oscillations superposed on different parts of cycle (Madelung). Applying this process to our Fig. 103, we must think of the curve B as going through a set of vibratory tremors back and. forth horizontally as it settles down toward the curve H. These tre- mors are of too high frequency to act on the telephone, which therefore responds only to the general displacement of the mag- netization from the curve B toward the curve H. Sensitiveness of the Magnetic Detectors. The magnetic detectors are more sensitive than the coherer, but seem to be less sensitive than the electrolytic detector and some of the solid con- tact detectors (the crystal detectors). THERMAL DETECTORS There are two general classes of detectors in which the heat developed by the electric waves is made to manifest itself at the receiving station. In one of these classes, including the bolometer 154 WIRELESS TELEGRAPHY and the barretter, a change of electrical resistance under the heat developed is observed; and in the other class of thermal detectors, the thermoelectric detectors, it is the thermoelectromotive force called into play by heating the junction of two dissimilar metals that is observed. Bolometer. The Bolometer, which was applied by Paalzow and Rubens to measurements with electric waves, has been de- scribed in Chapter X (see Fig. 47). Briefly, the bolometer consists of an accurately balanced Wheatstone-s bridge of which one arm is composed of a very fine wire. When electric waves are passed through this fine wire, it is heated. The heat developed changes the resistance of the fine wire and throws the bridge out of balance, so that the galvanometer in circuit with the bridge gives a deflec- tion. This apparatus has been applied by Tisot to measurements of the energy received in a wireless telegraphic receiving station. The action of the bolometer is not sufficiently rapid for use in practical wireless telegraphy, unless one should use with it the newly developed Einthoven galvanometer, which has a very small period. Barretter. In a United States patent, for which application was filed June 6, 1902, Professor R. A. Fessenden has described a detector operating on the same prin- ciple of change of resistance with heat, but capable of being used with a tele- phone receiver. He calls the appar- In the construction of the barretter Pro- fessor Fessenden made use of a Wol- laston wire, which is obtained by casting an ingot of silver with a platinum core and drawing down the ingot. This pro- duces a wire with a silver exterior and a fine platinum thread running through the center. Then by etching off the silver with nitric acid from a short length of this wire the very fine platinum core is left. Fessenden's barretter consists of a small loop of atus a barretter. FIG. 106. Profes- sor Fessenden 's barretter. FIG. 107. Diagram of circuit with barretter as detector. ON DETECTORS 155 this fine platinum wire, which may be as small as one or two ten-thousandths of an inch in diameter. In the finished instru- ment this fine loop of wire is inclosed in a glass or metal bulb, as shown in Fig. 106. The method of using the detector is shown in Fig. 107, which contains the detector D in series with the antenna A and ground G of a receiving station. In the local circuit about the detector is a battery B and a telephone re- ceiver T. Oscillations in the antenna circuit passing through the detector heat the fine loop of wire. This changes the resist- ance of the little loop, and consequently modifies the current in the local circuit, and produces a sound in the telephone receiver. When the waves cease the little loop rapidly cools, restoring the current to its original value. The adaptability of the instrument to the receipt of signals is due to the very small heat capacity of the fine wire, by reason of which it heats and cools with sufficient rapidity to respond with the train-frequency of waves. The diffi- culty with the use of this instrument arises in its liability to be burned out when the signals become too strong. In sensitiveness the barretter falls far below the sensitiveness of the electrolytic and crystal detectors to be described later, and its use, except for the purposes of laboratory measurements, has been practically discontinued. Thermoelectric Detectors. We have already described two thermoelectric detectors: Klemencic's thermal junction (Chapter IX) and DuddelFs thermogalvanometer (Chapter XV). These instruments change the energy of the electric waves into heat localized in a small amount of metal. The heat developed, in the case of Klemencic's thermal junction, is developed at the thermal junction itself; while in Duddell's instrument the heat developed in the " heater " is conveyed by radiation and convection to the thermal junction. The heating of the thermal junction produces an electromotive force, which gives rise to a unidirectional electric current in the local circuit and produces a galvanometer deflection. We have in these instruments, first, a change of the energy of the electric oscillation into heat, and then a change of this heat energy again into electric energy. The instruments of Klemencic and Duddell, though very useful for the purposes of measurements, are not sufficiently rapid or sufficiently sensitive for use in the recep- tion of actual messages. It has been found, however, that a high resistance contact between a common metal and certain crystalline substances, or 156 WIRELESS TELEGRAPHY between two crystal substances, when connected with a telephone receiver, are highly sensitive to electric waves, and are among the most sensitive detectors known. These have been described in many cases by the patentees or by writers on the subject as ther- moelectric detectors. It has been found, however, that in a great many cases, at least, the thermoelectric explanation of the phe- nomenon is not the correct explanation; and these detectors are described and discussed in the next chapter, under the head of Crystal Rectifiers. CHAPTER XVII ON DETECTORS (Continued) . CRYSTAL RECTIFIERS WE come now to a very sensitive and interesting class of de- tectors for receiving the signals of wireless telegraphy and wireless telephony. These are the detectors consisting of a self-restoring high-resistance contact between solid bodies, and since one of the bodies is usually crystalline in character, I have given to this class of detectors the, name Crystal Rectifiers. The crystal rectifiers are self-restoring, and are usually employed with a telephone receiver; but a capillary electrometer or galva- nometer can be used in the place of the telephone receiver. Many of the detectors of this type will give a very strong response without a battery in the local circuit, but most of them require the battery of small e.m.f. for the best sensitiveness. Fig. 108 shows the connections for use of a self-restoring de- tector with a battery B in the local circuit. Fig. 109 shows the U^fi pi FIG. 108. Crystal contact detector with battery in local circuit. FIG. 109. Contact detector without battery. detector without a battery. The detector D is shown attached to the antenna and ground in a very simple form of receiving circuit. 157 158 WIRELESS TELEGRAPHY Electric oscillations in the antenna pass through the detector, and a response is obtained in the telephone. In the case of the use of the battery in the local circuit, we might explain the action by supposing that the resistance of the detector is changed by the electric oscillations through it (perhaps by the heat developed), and that in consequence of the change of resist- ance a larger or smaller amount of current is sent through the telephone receiver T. In the case where no battery is employed (Fig. 109), we might explain the response in the telephone by supposing that the oscil- lations heat the contact D, and that the heat developed at the contact (which consists of two dissimilar bodies E and F) gives rise to thermoelectric currents in the circuit containing the telephone. These are explanations that are apparently simple, and that apparently accord with many of* the known facts about thermo- electricity. We shall see, however, in what follows, that a careful experimental study of the subject has led to the rejection of the thermoelectric explanation, and has brought us to regard the action of these detectors, as a case of the easier passage of elec- tricity in one direction than in the other through the contact; that is to say, we are dealing with a newly discovered case of rectifica- tion of an alternating electric current at a contact between solid bodies, and in this process of rectification heat plays only a negligible role. Before entering into a discussion of this view, let us describe some of these detectors of the self-restoring contact type. We shall begin with a rather poor representative, the carbon microphone. Microphonic Detector. In 1879, Professor D. E. Hughes, the inventor of the microphone, accidentally found that the contact of a piece of carbon with bright steel, when used with a telephone receiver, was responsive to the inductive effect produced by the make and break of the primary current of an induction coil. Hughes did not publish his results until interviewed on the subject by Mr. Fahie in 1899. He then wrote Mr. Fahie a letter, which was published in the London Electrician, May 5, 1899. In look- ing over the description of Professor Hughes's experiments we now see that in 1879 he was producing and receiving electric waves, and had discovered in the microphone a self-restoring contact detector. A diagram of one form of Hughes's microphonic detector is shown in Fig. 110, which is redrawn from a sketch in DETECTORS CRYSTAL RECTIFIERS 159 Mr. Fahie's History. 1 In this diagram. C is a carbon pencil touching a steel needle N; S is a brass spring by which the pres- sure of the contact can be regulated. The adjustment of the spring is regulated by means of the disc D. Professor Hughes used the mi- crophone with or without a bat- tery in the local circuit ; and when the battery was omitted, he attributed the sound in the telephone to the thermoelectro- motive force developed at the carbon-steel junction. The de- tector was more sensitive with a battery in the local circuit than without it. Various modifications of this microphonic detector of Hughes have been employed in practical wireless telegraphy. One modi- fication, which had a considerable application a few years ago, FIG. 110. Hughes's microphonic steel-carbon detector. FIG. 111. Steel-carbon detector. FIG. 112. Detector of carbon gran- ule between metallic plugs. is obtained by placing a steel needle across two blocks of carbon, as shown in Fig. 111. Another is made by placing a granule of carbon between metallic plugs in a tube, as shown in Fig. 112. The microphone is more sensitive than the filings coherers. It is, however, somewhat troublesome on account of sensitiveness to mechanical vibrations and on account of liability to cohere under strong signals, and it is surpassed in sensitiveness to electric 1 Fahie, History of Wireless Telegraphy, 1902, Dodd, Mead & Co. 160 WIRELESS TELEGRAPHY waves by the crystal detectors, in which the carbon of Hughes's microphone is replaced by certain crystalline mineral substances. Dunwoody 's Carborundum Detector. In 1906 General H. H. C. Dun woody x of the United States Army (retired) discovered that a fragment of carborundum, when provided with suitable electrodes for connecting it into the circuit, will act as a receiver for electric waves. Carborundum is a carbide of silicon, manu- factured in the electric furnaces at Niagara; it is a comparatively poor conductor of electricity, is crystalline in character, and is next to the diamond in hardness. In Dunwoody's description of the detector the connection of the carborundum into the circuit was made by twisting wires around the carborundum or by holding it in a clamp between metallic jaws supported on an insulating base. General Dunwoody found that when the carborundum detector was placed in a wireless telegraph receiving circuit, and a telephone was connected about the detector, responses were obtained in the telephone with or without a battery on the local circuit. The detector was, however, more sensitive with the battery than without it. A number of experiments on this form of detector have been described by the author in publications in the Physical Review, and are abstracted later in the present chapter. The carborundum detector is not highly sensitive. Austin's Tellurium-Aluminium and Tellurium-Silicon De- tectors. In 1906 Dr. L*. W. Austin found that tellurium in contact with aluminium or in contact with silicon is a sensitive detector for electric waves with or without a battery in the local circuit. He attributed the action to thermoelectricity in his patent applications and early writings 2 on the subject, but afterwards found that this was not the true explanation of the phenomenon. 3 1 Dunwoody: U.S. Patent, No. 837,616, filed March 23, 1906, issued Dec. 4, 1906. 2 L. W. Austin: Letter to the Electrical World, 1906, Vol. 48, p. 924; U. S. Patent, No. 846,081, filed Oct. 27, 1906, issued March 5, 1907; "The High Resistance Contact Thermo-Electric Detector for Electrical Waves/' Physical Review, 1907, Vol. 24, p. 508. 3 After the publication of the author's research on the Carborundum Detector (See Pierce: "Crystal Rectifiers for Electric Currents and Electric Oscillations, Part I, Carborundum." Physical Review, 1907, Vol. 25, p. 31), in which it was pointed out that thermo-electricity could not explain the phe- nomenon, Austin came to the opinion that the action in the case of his tellurium detector and other detectors of a similar type was also not the rmoelect ric. DETECTORS CRYSTAL RECTIFIERS 161 Pickard's Crystal Detectors. Mr. Greenleaf W. Pickard has been very prolific in the discovery of materials of a crystalline character that can be used as a member of contact detectors. Among the substances used and patented by him in this connec- tion are silicon, 1 zincite, 2 chalcopyrite, 3 bornite and molybdenite. 4 The mounting of Mr. Pickard's silicon detector, which is repre- sentative of a favorable method of .constructing the detectors of this class, is shown in Fig. 113. A rod of brass A is pressed down by a spring S into contact with a mass of polished silicon B, H i D- FIG. 113. Pickard's silicon detector. embedded in an easily fusible solder of Wood's metal, M . The solder in which the silicon is embedded is contained in a metallic cup P, which rests upon a metallic plate K. Connection to the rod A is made by means of the binding post E. Connection to 1 G. W. Pickard: Electrical World, Vol. 48, p. 1003, 1906; U.S. Patent, No. 836,531, filed Aug. 30, 1906, issued Nov. 20, 1906; U. S. Patent, No. 888,119, filed Nov. 9, 1907, issued May 19, 1908. 2 G. W. Pickard: U.S. Patent, No. 886,154, filed Sept. 30, 1907, issued April 28, 1908. 3 G. W. Pickard : U. S. Patent, No. 912,726, filed Oct. 15, 1908, issued Feb. 16, 1909. 4 G. W. Pickard: U.S. Patent, No. 904,222, filed Mch. 11, 1907, issued Nov. 17, 1908. 162 WIRELESS TELEGRAPHY the silicon is made by means of a binding post not shown, which connects with the plate K. The ability to move the cup contain- ing the embedded silicon is an advantage, because not all parts of the silicon surface are equally sensitive, and this motion permits the selection of a sensitive place on the silicon as the point of con- tact. Mr. Pickard sometimes uses two of these active materials in the same detector. For example, a contact of zincite with bornite is one of the most sensitive electric wave detectors known. The action of these detectors was at first attributed by Mr. Pickard to thermoelectric effects, but after I had published the opinion that the action was not thermoelectric, Mr. Pickard amended many of his patents to comply with this latter view. 1 EXPERIMENTS CONCERNING THE ACTION OF THE CARBORUNDUM DETECTOR AND THE OTHER CRYSTAL-CONTACT DETECTORS Soon after the discovery by General Dunwoody that a crystal- line mass of carborundum when supplied with a contact electrode acts as a detector for electric waves, I began a series of experiments to determine, if possible, the nature of the phenomenon. The experiments were extended to other crystal detectors. The results of these experiments have been published in the Physical Review in a series of papers entitled " Crystal Rectifiers for Electric Cur- rents and Electric Oscillations." 2 The method of experimenting consisted 1 It is not safe to take the date of application for a patent as the date of the discovery of all the facts contained in the patent, because there is nothing in the published patent to show whether the matter of the specifications and claims was introduced at the date of the application or much later, as amend- ments. The actual date of the amendments can be obtained from the file records in the patent office. Sometimes a patentee, without any intention of obtaining undue credit for priority, but in accordance with ordinary United States Patent Office prac- tice, and by reason of interference with another inventor, has had discoveries put into his patent application that were not there when the application was made. 2 G. W. Pierce: Crystal Rectifiers, etc. Part I. Carborundum, Physical Review, Vol. 25, p. 31, 1907. Part II. Carborundum, Molybdenite, Anatase, Brookite, Physical Review, Vol. 28, p. 153, 1909; and Proc. Am. Acad. of Arts and Sciences, Vol. 45, p. 317, 1909. Part III. Iron Pyrites, Physical Review, Vol. 29, 1909. See also G. W. Pierce: A Simple Method of Measuring the Intensity of Sound, Proc. Am. Acad. of Arts and Sciences, Vol. 43, p. 377, February, 1908, DETECTORS CRYSTAL RECTIFIERS 163 (1) In determining what currents would flow through the detec- tor under a given steady electromotive force; (2) In an oscillographic study of the instantaneous values of the current through the detector under the action of an alter- nating e.m.f.; (3) In measuring the thermoelectric properties of some of the specimens and comparing the thermoelectromotive force with the rectified current. Some of the facts obtained in these experiments are presented in this and the next chapter. Apparatus for Current-voltage Measurements. Figure 114 shows a sketch of a form of circuit employed in studying the con- ductivity of crystal contact under various conditions, by means of FIG. 114. Circuit for studying current-voltage characteristic of crystal rectifiers. current and voltage measurements. The crystal, held in a clamp, is shown at Cr; B is a storage battery; XYZ is a potentiometer consisting of two fixed plates of zinc X and Z, and one movable plate Y, immersed in a zinc sulphate solution. By means of the voltmeter V the difference of potential between the plates Y and Z could be read, and the resulting current through the crystal was given by a galvanometer or milliammeter at A. The resistance of the galvanometer was so small in comparison with the resistance of the crystal that the reading of the voltmeter was practically the drop of voltage in the crystal. The switch S s enables the observer to reverse the current in the crystal under examination without reversing the galvanometer. A known resistance at R could be thrown into circuit with the galvanometer for the purpose of calibrating it. 164 WIRELESS TELEGRAPHY Current-voltage Curve for the Carborundum Contact. A curve obtained by plotting the current against voltage in an experi- ment with carborundum is shown in Fig. 115. It is seen that the current through the carborundum is not proportional to the vol- tage impressed upon it; the apparent resistance of the carborun- dum or its contact diminishes with increasing current. Experiments were made by the writer on a great many speci- mens of carborundum and other crystal detectors, and curves of approximately the shape shown in Fig. 115 were obtained in all the cases. On reversing the electromotive force so as to send the current in the opposite direction through the contacts a most interesting 32 7 Volts FIG. 115. Current- voltage curve of a carborundum contact. property was discovered; namely, the property of unilateral con- ductivity. Unilateral Conductivity of the Carborundum Contact. The current through the crystal in one direction under a given electro- motive force was found to be different from the current in the opposite direction under the same electromotive force; that is to say, the heterogeneous conductor formed of the crystal and its contacts is unilaterally conductive. This effect may be seen by a reference to Fig. 116. The branch I of the curve shows the cur- rent, plotted against voltage, when the current is in one direction; DETECTORS CRYSTAL RECTIFIERS 165 branch // the corresponding values of the current obtained when the voltage is reversed. The accompanying table, Table III, contains the numerical values from which these curves were plotted. In the experiment whose result is shown in Fig. 116 and Table III, the specimen of carborundum was held in a clamp under a pressure of about 500 grams, and it is seen from the table that the current in one direction is 100 times as great as the current in the opposite 2000 1600 1200 800 400 20 10 II 10 20 Volts FIG. 116. Curve showing the carborundum contact to be unilaterally conductive. direction when an electromotive force of 10 volts is applied in the two cases. With increase of current through the specimen, the ratio of the current in the two opposite directions diminishes. At 27.5 volts Ci is only 17 times Ci. In this particular experiment the piece of carborundum was sub- merged in an oil bath designed to keep the temperature of the specimen constant. The piece of carborundum was held in a clamp, the jaws of which served to lead the current to the speci- 166 WIRELESS TELEGRAPHY men. The oil, of which the temperature was 64 C., came freely into contact with the crystal. TABLE III RELATION OF CURRENT TO VOLTAGE, SHOWING UNILATERAL CONDUCTIVITY OF A CARBORUNDUM CONTACT Volts. Current in Microamperes. eye. Pi Commutator, Left. C 2 Commutator, Right. 2.2 1 2.8 2 4.0 5 4.7 10 5.9 20 6.5 30 7.3 40 8.0 50 10.0 100 1 100 12.1 150 12.8 200 14.5 300 5 60- 16.0 400 16.8 500 10 50 17.7 600 19.4 700 20.0 800 20 40 21.0 900 21.9 . 1,000 30 33 23.2 1,200 50 24 25.0 1,500 27.5 2,000 120 17 Similar effects were obtained at various temperatures between 10 C. and 100 C., both with and without the use of oil as a bath. A like result was had with different specimens and under different pressures. The relative values of the positive and negative currents, how r ever, varied from piece to piece, and also was different under different conditions of temperature and pressure. Effects of Pressure. Figure 117 shows a series of current-vol- tage measurements with a specimen of carborundum held in a clamp under various pressures. Several experiments were made with other specimens of carborundum with considerable disparity in the results, and the DETECTORS CRYSTAL RECTIFIERS 167 curves of Fig. 117 cannot be taken to represent a general oc- currence. For more details on the effect of pressure reference is made to the original publications in the Physical Review. FIG. 117. Current-voltage curves of carborundum under different pressures. Experiments with Platinized Specimens of Carborundum. In the effort to ascertain what part the form of contact plays in the phenomenon of unilateral conductivity in crystals, a number of specimens of carborundum were selected with opposite faces plane and very approximately parallel, and some of the parallel-faced crystals were platinized on one or both of their smooth surfaces by the cathode discharge so that they could be put into good conducting contact with the electrodes. The metallic surfaces thus obtained were in many cases optically plane. Platinized on One Face only. Some of the specimens, plati- nized on one face only, gave very remarkable unilateral con- ductivity. Table IV shows results obtained with one of these specimens, designated 11 6 , when submitted to a pressure of 1 kilo- gram. This specimen was .6 mm. thick, with area of about 1 sq. mm. One of the faces, which was optically true, was heavily platinized. The other face was somewhat rough and was without platinum. The specimen was held in a clamp with silver jaws. 168 WIRELESS TELEGRAPHY TABLE IV CRYSTAL, 11&. THICKNESS, .6 MM; AREA, 1 SQ. MM. PLATINIZED ON ONE SIDE. PRESSURE, 1 KG. Volts. Cj , Current toward Platinum in Microamperes. C 2 , Current from Platinum in Microamperes. C,/C 2 , 4.5 3.92 g , 6 7.84 3 Frances G. Wick 1 Matthiessen 2 }> >t Bismuth Antimony Tellurium Selenium 1 Phys. Rev., 25, 390. 2 Everett, Units and Physical Constants. It may be said in passing that the specimens D and E, with soldered connections, still showed the phenomenon of rectification when used with alternating currents, even when the two junctions of the copper with the molybdenite were in oil baths at the same 194 WIRELESS TELEGRAPHY temperature as the room and the oil in the baths was vigorously stirred with motor-driven stirrers. The rectification in this case was, however, very imperfect. Temperature Coefficient of Resistance. Another interesting thermal property of the molybdenite is its temperature coefficient of resistance. A brief report of this coefficient is here given. Two specimens of the molybdenite were made into the form of resistance thermometers by depositing heavy copper-plated areas near the two ends of thin pieces of the molybdenite and soldering thin copper strips to the copper plate. For insulation a thin strip of mica was placed over the molybdenite, and one of the copper leads was bent back over the mica so that both leads ran away parallel with the mica insulation between. The whole conductor was then placed between two mica strips and inserted in a flat- tened brass tube. The tube was then mashed tight together so as to clamp securely the molybdenite and its leads. The end of the tube adjacent to the molybdenite was soldered up. The leads were brought out at the other end of the tube and connected to binding posts insulated by a hard-rubber head from the tube. The two molybdenite resistances thus mounted are called No. 50 and No. 51. The molybdenite in No. 51 was .65 cm. wide by .7 cm. long; the thickness was about .3 mm. The resistances of these, two conductors were measured at vari- ous temperatures with the aid of a Wheatstone bridge. They showed no evidence of rectification. The values plotted in Fig. 132 were obtained. The curves marked " 50 " and " 51 " give the resistances of No. 50 and No. 51 respectively. The ordinates for these curves are at the left margin of the diagram, and are in ohms. The curves " C 50 " and " C 51 "are for the reciprocals of the resistance of No. 50 and No. 51 respectively. The ordinates for these curves are at the right-hand margin of the diagram. Each of the specimens has a large negative temperature coeffi- cient of resistance. With No. 50, for example, the resistance at 93.1 C. is 229 ohms; at C. the resistance is 561 ohms; at - 76 the resistance is 3051 ohms; and at the temperature of liquid air the resistance of this specimen was found to be over 6,000,000 ohms. This last value is not plotted on the curves. It is interesting to note that between 15 and 93 the temperature- conductance curve of each of the specimens is a straight line. At C. the resistance of each of the specimens decreases about 1.53 percent per degree centigrade increase of temperature; at DETECTORS CRYSTAL RECTIFIERS 195 20 the decrease of resistance per degree increase of temperature is 1.19 percent. Plausibility of Thermoelectric Explanation. The large thermo- electromotive force of the molybdenite against the common metals, together with its large negative temperature coefficient of resist- ance, lends plausibility to the hypothesis that the rectification is due to thermoelectricity. For if we pass an electric current through the rectifier and the current begins to make its way -70 -60 -50 -40 -30 -20 -10 10 20 Temperature 40 50 60 70 80 90 FIG. 132. Resistance and conductance of molybdenite as a function of the temperature. through a small area at the contact, this small area is heated and decreases in resistance, so that the greater part of the current flows through this particular small area, heating it still more, while the portions of the contact through which the current has not started remain cool and continue to offer a high resistance. The effect of this action is to confine the heating to an extremely small area, which is the condition necessary for the extremely rapid and efficient action of the rectifier, on the hypothesis of a thermo- electric explanation. That there is, however, an insuperable ob- 196 WIRELESS TELEGRAPHY jection to this explanation of the phenomenon is, I think, made clear in the succeeding experiments, in which it is shown that the thermoelectric effect is often opposite to the rectification, and that the amount of heat associated with the rectification accounts for less than sWoo^ of the rectified current as thermo- electric. EXPERIMENTAL FACTS ADVERSE TO THE THERMOELECTRIC EXPLA- NATION OF THE PHENOMENON OF RECTIFICATION Thermoelectric Effect Opposite to the Rectification. A num- ber of experiments with different specimens of molybdenite were made in which the rectification and the thermoelectric effect could be simultaneously studied. A diagram of the arrangement of apparatus is given in Figure 133. The specimen of motybdenite is shown at M, and was held down upon a wooden base by a spring clip. One end of each of the specimens, which were easily in- terchangeable in the apparatus, was electroplated with copper at S. To this copper-plated area a copper lead was soldered. A copper rod C, supported as in Figure 124, was brought into con- tact with the part of the molyb- denite distant from the soldered junction. The molybdenite and FIG. 133 Apparatus for compar- the con tact were put in an elec- mg rectified current with thermo- ... . . electric effect. trie circuit containing a galva- nometer at A and a source of variable alternating potential at V. The alternating potential V could be applied or omitted by closing or opening the switch at T. A small heating coil was wound on the rod C, and another similar heating coil was wound on a second copper rod E placed im- mediately below the contact of C with M . An auxiliary thermal junction formed by a small constantan wire attached to the lower end of the copper rod C was connected to a second galvanometer shown at G, for use in a later experiment. The copper rods C or E could be heated by the surrounding coils, and the thermal current in the circuit through the molyb- DETECTORS CRYSTAL RECTIFIERS 197 denite or the circuit through the constantan could be read on the galvanometers A or G. Also the rectified current obtained by applying the alternating voltage V could be read on the galva- nometer A. When the thermal current or the rectified current through A is in the direction of the arrow B, the molybdenite, following the usage in thermoelectricity, is said to be positive. When the current in A is in the direction "opposite to the arrow B, the molybdenite is said to be negative. The results obtained with a number of specimens of molybdenite when heat was applied above, and when heat was applied below, and when the alternating voltage was applied, are contained in Table X. TABLE X SIGN OF MOLYBDENITE WHEN HEATED ABOVE OR BELOW AND WHEN SUBJECTED TO ALTERNATING VOLTAGE Specimen No. Heated Above. Heated Below. Under Alternat- ing Voltage. 75 + 81 _j_ Turned over + _ _ 93 _j_ _|_ Another point + ^ f Turned over _ _ 78 _|_ _j_ _j_ Another point + - - 94 _ 1 + Another point - + + From this table it appears that the thermoelectric voltage when the junction is heated by heat conducted from above, in twelve out of the thirteen cases tried, is opposite to the direct voltage ob- tained when an alternating current is passed through the junction. When the heat is conducted to the junction from below, through the molybdenite, the thermoelectromotive force in four cases is opposite to the rectified voltage, and in nine cases is in the same direction as the rectified voltage. In only one case, one point of No. 78, is the rectified voltage in the same direction as the thermal voltage, both when the junction is heated from above and when it is heated from below. In all of these cases the heat was applied in the neighborhood of the same junction, and there is no opportunity for heat to get to 198 WIRELESS TELEGRAPHY the other junction (copper-plated and soldered) by conduction, on account of the great distance of the other junction from the source of heat. To make this absolutely certain this distant junction was in some cases submerged in an oil bath. So far as I have been able to learn, this phenomenon of the reversal of the thermoelectro motive force at a thermal junction, conditioned on whether the heat is conducted to the junction through one element of the junction or the other element of the junction, is novel. It may be explained by the assumption of another thermal junction of opposite sign in the molybdenite itself below and in the immediate neighborhood of the copper- molybdenite junction. This assumption is plausible because it has been shown above that the molybdenite with which these experiments were performed is thermoelectrically an extremely heterogeneous substance. However, whatever the explanation of the dependence of the sign of the thermoelectromotive force on the manner of applying the heat, it is seen that the thermoelectric effect is usually opposite in sign 1 to the rectified effect. By applying heat from, above and at the same time applying the alternating voltage, one can make the thermal current and the rectified current neutralize each other. This opposition of sign of the rectified current and the thermal current renders the correct- ness of the thermoelectric explanation of the phenomenon of rectification extremely improbable. Insufficient Heating of the Contact to Account for Rectification. The most convincing experiment on the subject is the follow- ing: With the aid of the auxiliary thermal junction of copper-con- stantan placed at the contact of the copper with the molybdenite, as shown in Fig. 133, it was possible to look for a rise of temperature of the copper molybdenite junction by the alternating current which was being rectified. If any appreciable heat were developed at the molybdenite copper junction, the copper-constantan junc- tion ought to show it. The following result was obtained: When the rectified current was 118 microamperes, the heating shown by the copper-constantan junction did not exceed .01 C. 1 In the case of silicon-steel, carbon-steel, and tellurium-aluminum, L. W. Austin has found that the rectified current generally flows in opposite direc- tion to that produced by heating the junction. In his experiments (Bulletin of the Bureau of Standards, 5, No. 1, August, 1908) the heat was applied by conduction from above only. DETECTORS CRYSTAL RECTIFIERS 199 When, on the other hand, as a control experiment, heat was applied to the copper-molybdenite junction from below so that it had to be conducted through the molybdenite and through the copper-molyb- denite junction to get to the copper-constantan junction, the heating shown by the auxiliary copper-constantan junction was 11.4C., while the thermal current from the copper-molybdenite junction was only .2 microamperes. In both the case of the recti- fied current and the case of the application of heat from below the heat had to be conducted from the point of rectification to the auxiliary junction. Therefore, with a rise of temperature of the auxiliary junction 1100 times as great as the rise shown during the rectification, the thermal current in the copper-molybdenite circuit was 5 D 202 WIRELESS TELEGRAPHY the sensitive " point " may be as small as one or two ten-thou- sandths of an inch in diameter. For a less sensitive detector, which is not so likely to be destroyed by strong signals, wire as large as one-thousandth of an inch or even larger may be used. Only a very short length of the fine wire comes into contact with the electrolyte; for the fine wire is either sealed into a glass tube so as to protect all but the mere end of the wire from contact with the electrolyte, or the fine silver-coated platinum wire, as at W, FIG. 135. Professor Pupin's electrolytic rectifier. FIG. 136. Electrolytic detector with adjust- able contact. Fig. 136, is carried up or down by a micrometer adjustment so as to bring it into contact with the electrolyte. The silver is removed from about T V of an inch of length of the wire by submerging it in the electrolyte, which is in this case nitric acid, and sending a current from the local battery through it for a few minutes, with the small wire as anode. This takes off the silver, leaving the bare platinum. The point, so formed, is then withdrawn by a motion of the screw B, until only a very minute area of the bare platinum wire is left in contact with the electrolyte. The detector in this form is used with an adjustable source of e.m.f. in its local circuit. The fine platinum electrode may be connected either to the positive or the negative terminal of the battery, but the detector is usually more sensitive when this fine platinum electrode is positive (i.e., anode). The voltage in the local circuit required in this case is about 1.5 volts. Variation in which Source of Polarizing Voltage is Located in the Detector Itself. Instead of employing an external voltage E (Fig. 134) to polarize the detector, a similar effect can be ob- tained by constituting the electrodes in the detector of different metals, one of which (zinc, say) is attacked by the electrolyte, and the other of which, the fine platinum wire, is inert to the action of the electrolyte. ELECTROLYTIC AND VACUUM DETECTORS 203 This makes the detector itself a primary battery. This arrangement for which a United States patent has been issued to Schloemilch, 1 and also to Shoemaker, 2 would seem to be incapable of the high sensitiveness attained by the form in which the accurately adjustable external voltage, as in Fig. 134, is employed. Regarding the Theory of the Electrolytic Detector. Con- siderable diversity of opinion has been expressed by various writers as to the manner in which the electrolytic detector acts as a receiver for electric waves. Professor Fessenden in his original patent attributes the action to heat, and he calls this form of detector a " liquid barretter." Pro- fessor Armagnat, 3 who has made an experimental study of the sub- ject, attributes the action to a rectifying effect resulting from polarization. Armagnat obtained a curve of the form of Fig. 137 for the current-voltage character- istic of the electrolytic detector. Dr. L. W. Austin 4 also found that the electrolytic detector acted as a rectifier for small alternating currents, but came to the opinion that heat, chemical action, rectifi- cation, and electrostatic .attraction FlG< across the gas film might have a part in the explanation of the phenomenon when the detector was used with electric waves. A doubt that arose in the minds of some investigators of the subject as to a possible explanation of the phenomenon in terms of rectification alone came, it seems, from the idea that there could not be energy enough in the electric waves received at great distances to produce the effects in any other way than by a triggering action, by which the local energy of the battery was 1 Wilhelm Schloemilch, U. S. Patent, No. 936,258, filed Oct. 3, 1903, issued Oct. 5, 1909. 2 Harry Shoemaker, U. S. Patent, No. 795,312, filed Feb. 13, 1905, issued July 25, 1905. 3 Armagnat, Bui. soc. franchise, session of April, 1906, p. 205; Journal de Physique, Vol. 5, p. 748, 1906. 4 Austin, Bui. Bureau of Standards, Vol. 2, p. 261, 1906. 204 WIRELESS TELEGRAPHY brought prominently into play. Now, however, since some of the crystal detectors, that act entirely without any local source of energy, are as sensitive as the electrolytic detector, we see that the energy to produce the sounds in the telephone is really present in the incoming waves, and does produce the sounds when the incoming energy is applied to the telephone receiver with the aid of a suitable rectifier. The Electrolytic Detector as a Rectifier. That an electrolytic cell with one of the electrodes small, when suitably polarized with a direct current, is a rectifier for alternating currents was first shown by Professor M. I. Pupin 1 before such a cell came into commercial use as a detector for electric waves. The following account of Pupin's rectifier is translated from an article published in the " Jahrbuch der Elektrochemie," Vol. 6, p. 35, 1899: " In Fig. 3 " (here reproduced as Fig. 135) " A is a battery, B an electrolytic cell with the platinum electrodes a and b and acidulated water. If the polarization of the cell B is as great as the e.m.f. of A, no current flows in the circuit. If one allows an alternating current to act upon the circuit ABC, the circuit contains resistance, self-inductance, and a capacity localized in the plates a and b. The cell B acts, however, as a condenser only so long as the potential difference of the plates a and b is smaller than the decomposition voltage. If this value is exceeded, a current goes through the circuit. If the alternating current, for example, has an amplitude that is twice as great as the e.m.f. of A, in case the phase has the same direction as A a current flows in the circuit, e.g., in the direction BC; when the phase is oppositely directed, the condenser B sends a current in the opposite direction. This last can be diminished by making the capacity of B very small. If, for example, the area of one of the electrodes is only one square millimeter, one may easily rectify alternating currents with a frequency of 1000 per second; with greater frequency the electrode must naturally be made still smaller. It is best to em- ploy a platinum wire sealed into glass the wire being cut off immediately at the end of the glass. The author (Pupin) suc- ceeded in rectifying electric oscillations of Hertzian frequency and producing electrolytic effects with them; the wire for this purpose was .025 mm. in diameter." 1 Pupin, Electrical World, Vol. 34, p. 743, 1899; Zeitsch. f. Elektrochemie, Vol. 6, p. 349, 1899; Jahrbuch d. Elektrochemie, Vol. 6, p. 35, 1899; Bui. Am. Phys. Soc., Vol. 1, p. 21, 1900. ELECTROLYTIC AND VACUUM DETECTORS 205 This quotation shows that Pupin had employed the electrolytic detector in 1899 as a rectifier for electric waves of Hertzian fre- quency, and that he had a well-defined explanation of the processes occurring in the rectifier. I have made some experiments that fall into close agreement with Pupin' s explanation of the phenome- non. These are described in the succeeding paragraphs. OSCILLOGRAPHIC STUDY OF THE ELECTROLYTIC DETECTOR 1 In these experiments the current through the detector under the action of an alternating e.m.f., superposed on a polarizing current, is determined by means of an oscillograph. The application of the oscillograph to the problem gives the instantaneous values of the current through the detector, and permits an examination of the wave form of the rectified cycle. The oscillographic apparatus was the Braun's tube described in Chapter XVIII. Circuits Employed with the Detector in Taking the Oscillo- grams. The electrolytic detector used in these experiments made FIG. 138. Oscillographic apparatus and circuits for study of electrolytic detector. use of a platinum point, .0002 inch in diameter, dipping into 20 per cent nitric acid, and was adjusted to high sensitiveness as an electric wave detector immediately before taking the oscillograms. A diagram of the circuits employed in the experiment, together with a sketch of the oscillographic apparatus, is shown in Fig. 138. 1 This account is an abridgment of an article by the author on "The Electrolytic Detector, Studied with the Aid of an Oscillograph." Physical Review, 1909, Vol. 28, p. 56. 206 WIRELESS TELEGRAPHY The detector is at Z>, and is connected in series with the deflecting coils MM of the oscillograph and with the variable sources of voltage V and E. The voltage V is taken from a potentiometer connected with the 60-cycle alternating mains of the laboratory. E is an adjustable steady voltage taken from a battery. The voltage at E could be reversed. By opening the switch near D the electrolytic detector could be disconnected from the circuit, and by throwing this switch downward an ohmic resistance R could be substituted for the detector. In taking the oscillograms of Plate II the following steps were employed: The drum carrying the film was set rotating. The high potential current was started in the tube. The chosen value of the polarizing current was applied to the circuit and was read on a direct-current milliammeter. The alternating current was superposed on the circuit, and by adjustment of the potentiometer at V the voltage of this alternating current was given any desired value. The Exposures. After the preliminary adjustment of the direct and alternating currents through the detector, four expo- sures were made on each picture, while the film was being carried around continuously by the synchronously driven drum. Axis of Zero Current. This is the lower- straight line across the pictures, and was obtained by an exposure of 20 seconds taken with the circuit open. Axis of Polarizing Current. This is the upper straight line across the picture, and was obtained with the detector in circuit and traversed by the polarizing current. The exposure was 20 seconds. In oscillogram No. 1 this axis is not apparent, because on account of the small value of the polarized current employed it falls into coincidence with the axis of zero current. The Rectified Cycle. This cycle may be identified in the oscil- lograms as a positive l loop for a half-period, followed by a nearly straight portion lying along the axis of zero current for a part of a half-cycle, and going over into the positive loop through an inter- mediate " building up " segment. This cycle (exposure of 60 sec.) was taken with the detector in circuit, with the alternating e.m.f. applied to the circuit, and with the polarizing current also flowing. The Voltage-Phase Cycle. This is the sine curve of the pic- tures, and was taken in order to obtain the e.m.f. immediately 1 In describing the oscillograms, values above the axis of zero current are called positive; values below this axis are called negative. ELECTROLYTIC AND VACUUM DETECTORS 207 about the detector. 1 A similar curve was made use of in the experiments of the preceding chapter and is there discussed. In the present experiments, because of the employment of the polar- izing current with the rectifier, a question arises as to the appro- priate method of taking this cycle. Two different methods were tried, either of which, by proper elimination of the constants of the oscillogr^phic apparatus, will give the desired result. The method yielding simplest results for the voltage-phase cycle is the following : After the exposure for the rectified cycle had been made, the alter- nating voltage was left unchanged, and a resistance was substituted for the rectifier. A double adjustment of the substituted resist- ance and the direct voltage was made by successive approxima- tions until the result was attained that (1) the direct voltage alone gave through the substituted resistance a current equal to that used in polarizing the rectifier and (2) the alternating voltage superposed on this direct current gave a deflection of the lumi- nescent spot to a point coincident with the maximum point attained with the rectifier in the circuit. This means that the voltage- phase cycle was taken with the axis of polarizing current as axis, and with amplitude equal to the maximum amplitude of the recti- fied cycle. This method was employed in oscillograms 1, 2, and 5. The second method of taking the voltage-phase cycle was as follows: The polarizing voltage was reduced to zero, the detector was short-circuited, and an alternating voltage equal to that used with the detector was applied to the circuit. This method was employed in oscillograms 3 and 4. Coordinates of the Oscillographic Curves. In taking all of the curves of the oscillograms, the motion of the light spot over the film is from left to right; the time coordinate is, therefore, the horizontal scale of the curves and is drawn as usual from left to right. The current coordinate is given in the scale drawn in ink at the left-hand margin of each picture one division being one milliampere. DISCUSSION OF THE OSCILLOGRAMS OF PLATE II The oscillograms shown in Plate II are reproductions of positives printed from the films carried by the rotating drum. They were taken with a 60-cycle alternating current applied to the circuit 1 The ordinary method, which would be to take the leads from the two sides of the detector through a high resistance to the oscillograph, could not be used because the oscillograph was working at the limit of its sensitiveness on the full voltage without the added resistance. (208) PLATE II. G. W. Pierce, The Electrolytic Detector. ELECTROLYTIC AND VACUUM DETECTORS 209 containing the electrolytic detector. The reproduction is one- third the size of the original. The several curves shown in the plate were obtained with different polarizing currents superposed on the circuit. Table XI contains a tabulation of the polarizing current and voltage, the applied alternating voltage, the maximum current through the detector, and the substituted resistance employed in taking the voltage curve. e TABLE XI TABULAR DESCRIPTION OF THE OSCILLOGRAPHIC RECORDS Maximum No. Polarizing Direct Current in Milli- Polarizing E.M.F. in Volts R.M.S. Volts A.C. Positive Cur- rent through Equivalent Resistance araperes. Detector in in Ohms. Milliamperes. I 1 .1 1.45 2.09 2.37 440 1 2 1.0 5.5 4.00 9.6 70 3 2 1.2 5.5 4.00 9.6 OO 2 42 1.4 Not measured 5.00 10.0 OO 2 5 2.2 5.00 11.0 150 1 It should be noticed that the sensitiveness of the oscillograph when No. 1 was taken was three times as great as when the other oscillograms of the plate were taken. 2 The voltage-phase cycle of oscillograms 3 and 4 was taken with the polar- izing current omitted, so that they have the axis of no current as axis of the cycle. Point Anode or Cathode the Large Loop in the Direction of the Polarizing Current. Some of the oscillograms were taken with the polarizing current from the point to the electrolyte and some with the polarizing current in the opposite direction. Al- though the values of the polarizing voltage required to produce a given polarizing current were different in the two cases the general characteristics of the cycle were the same. A reversal of the polar- izing current reversed the rectified current, and whether the polar- izing current was from the point to electrolyte or in the opposite direction the large loop of the rectified cycle (always oscillographed positively) was obtained when the alternating current was flowing in the same direction as the polarizing current. The Form of the Rectified Cycle. The cycle obtained with the rectifier in the circuit has the same general form in all the pictures. When the current, having traversed the positive loop, comes to the axis of zero current, it follows along this axis for a short way, 210 WIRELESS TELEGRAPHY then takes a small negative dip, becomes positive again, follows along just above the axis of zero current for a short time, and then rises along a transition curve to the positive loop. Calculations Concerning the Form of the Cycle. The rectified cycle, when examined by comparison with the voltage-phase cycle, makes a misleading impression unless one takes carefully into account the condition under which the curves are obtained. One must bear in mind that the form of the current through any recti- fier is not determined by the rectifier alone, but is a function also of the constants of the circuits employed with the rectifier. In the present experiments the deflecting coils of the oscillographic apparatus possessed appreciable self -inductance and resistance, and these factors must be taken into account. Taking these factors into account, by a mathematical investi- gation not here given I have obtained the following results for !' I' .1 / f *5 V zs x / / w \> \ NV / 1 1 A Y/ 2 \ & 1 \ \ - \ \ .___ 4 --** # / .___ \ A A / ^ 1 \ \ Y 3 1 / \ \ \ \ \ V {/ \ \ \ / FIG. 139. Computed curves. oscillogram No. 2. The applied alternating e.m.f. is represented by the dotted curve of Fig. 139, with volts at right-hand margin. The resulting voltage-phase curve calculated from this e.m.f. and the constants of the circuit is displaced 38 to the right from the e.m.f. curve and is given by the continuous-line sine curve. The rectified cycle, calculated approximately from the current-voltage characteristic of the detector and the applied vol- tage gives a curve of the form of the heavy line in Fig. 4. This calculation accounts for the general form of the rectified cycle and its relation to the voltage-phase cycle. The electrolytic detector has an important peculiarity ; which is shown by the oscillogram, and to which attention is now directed. Evidence of Polarization Capacity. On oscillograms 1, 2 and 3 there is a small positive rise of the photographic curves in the ELECTROLYTIC AND VACUUM DETECTORS 211 region to the immediate right of the negative maximum. This rise is more striking in the original photographs than in the repro- ductions; and, though small, it deserves attention, because the occurrence of this small positive maximum is evidence of the existence for about 15 V^ of a second of a positive e.m.f. greater than the e.m.f. immediately following. Now in this part of the cycle the externally applied e.m.f. is greater follow- ing the rise than during the rise; therefore the rise indicates the existence of a positive e.m.f. in the circuit itself. This is capable of the following explanation in terms of the theory of polarization. After the prevalent external e.m.f. has been in a negative direc- tion and has returned to zero, the polarization tension which has been opposing the negative current at the electrode continues to exist for a short time and produces a positive current. This action, resembling that of a capacity, is familiarly known as the polarization capacity of the electrode. By the existence of the small positive maximum near the axis of the cycle, the oscillogram shows that the polarization capacity of the electrode is not entirely negligible. Evidence of the existence of this polarization capacity is clearly given by the oscillograms 1, 2, and 3. The oscillograms 4 and 5, while not having a positive maximum near the axis, show also a striking tendency toward a maximum at this point, which is, however, masked by the rapid rise of the building-up curve in this part of the cycle. CONCLUSIONS IN REGARD TO THE ELECTROLYTIC DETECTOR L The whole phenomenon of the rectification of small alter- nating currents by the electrolytic detector seems to be explicable in terms of the theory of electrolytic polarization. 2. The polarization capacity of the small platinum electrode is not entirely negligible, even with currents making only 60 cycles per second. The polarization capacity may, however, aid in pro- ducing a rectified current as well as in opposing this effect, and apart from the effect of this capacity on the tuning of the circuit, does not detract from the utility of the rectifier as a detector for electric waves. 3. The present conclusions in regard to the action of the detector are entirely in accord with Pupin's original brief description of the phenomenon as quoted above. 212 WIRELESS TELEGRAPHY COMPARISON OF THE ELECTROLYTIC DETECTOR WITH THE CRYSTAL RECTIFIERS The resemblance of the oscillograms with the electrolytic de- tector to those with the crystal rectifiers l is close, in so far as de- pends on the fact that both classes of rectifiers are nearly perfect 2 rectifiers when employed under their best conditions. The electro- lytic rectifier, in order to approximate perfection 3 as a rectifier, must be polarized by the superposition of a direct current; while the use of the direct current with the crystal rectifier, does not always materially improve the rectification. Also the two rectifiers are different, in that the electrolytic rectifier shows evidence of elec- trolytic polarization capacity, which, so far as may be judged from the oscillograms, is absent with the crystal rectifier. The experiment with the electrolytic detector, since it shows in the matter of polar- ization capacity the integrative action of this detector, which was sought for and not found with the crystal rectifier, is thus an interesting "control " experiment. In the matter of sensitiveness the best crystal rectifiers are about equal to the electrolytic detector. VACUUM DETECTOR Another highly sensitive detector, by which the electrical oscilla- tions at a wireless telegraph receiving station are rectified and detected, makes use of the unilateral conductivity of a vacuous space containing electrons produced by an incandescent body. A rectifier for electric oscillations making use of this principle, in- vented by Professor J. A. Fleming, 4 and called by him an "oscilla- tion valve " is represented in Fig. 140. In this figure a is a glass bulb, a little smaller than an ordinary incandescent lamp bulb; 6 is a carbon filament, like that of an incandescent lamp, which is heated to incandescence by connection through the leads / and e with a battery h. Surrounding the filament but not touching it is a metallic cylinder c. The bulb is pumped to a high degree of exhaustion. When the filament is raised to incandescence by a 1 Pierce, Part II., L c. 2 A rectifier is called "nearly perfect " when the ratio of the current in one direction to that in the opposite direction is large. 3 The current through the electrolytic rectifier is slightly asymmetric when no polarizing current is employed. 4 Proc. Roy. Soc. London, 1905, Vol. 74, p. 476; also U. S. Patent, No. 803,684, filed April 19, 1905, issued Nov. 7, 1905. ELECTROLYTIC AND VACUUM DETECTORS 213 current through it, negative electrons are sent off from it and render the space between the filament and the cylinder conductive for an electric current, provided the e.m.f. producing this current is directed from the cylinder to the hot filament. In case the e.m.f. is applied in the opposite direction, no current, or a much smaller current, flows. An oscillating e.m.f. applied to the cylin- der and filament produces more current in one direction than in the opposite direction. One method of connecting the valve into a wireless telegraph receiving circuit is shown in the diagram, which is taken from FIG. 140. Professor Fleming's vacuum tube rectifier. FIG. 141. Circuit employed by Dr. DeForest with vacuum detector. Professor Fleming's U. S. Patent Specifications. Here the valve is in a circuit connected inductively with a wireless telegraph antenna. Electrical oscillations in the antenna induce an oscil- lating electromotive force in the coil k, and this oscillating e.m.f. sends more current in one direction than in the opposite direction through the valve and through the current-indicating instrument I. A modification of the method of connecting the indicating instrument to the oscillation valve has been made by DeForest so as to permit the use of a telephone as indicator. A diagram of 214 WIRELESS TELEGRAPHY a circuit of this form taken from DeForest's U. S. Patent Specifi- cations 1 is shown in Fig. 141, in which F represents a telephone receiver and H a battery connected in the local circuit through the vacuum rectifier B (which Dr. DeForest calls an audion). 1 See U. S. Patent, No. 836,070, filed Jan. 18, 1906, divided May 19, 1906, issued Nov. 13, 1905. CHAPTER XX ELECTRICAL RESONANCE r WAVE METERS. RESONANCE IN SIMPLE CONDENSER CIRCUITS ON account of the multiplicity of facts requiring presentation in an elementary discussion of electric wave phenomena, it is often difficult to decide what is the most direct course to follow. For a part of the way, in the earlier chapters, we were able to proceed almost in the historic order. Up to about the year 1900, the growth of knowledge of electric waves, so far as pertains to wireless telegraphy, occurred as a fairly direct sequence of important events, which have been sketched in Chapters I to XIII. About the year 1900 the literature of the subject began to multiply enormously and practical progress began to develop in many directions. Two main branches of this development we have already pursued, in a discussion of the propagation of the electric waves to long distances over the surface of the earth and in a discussion of some of the detectors used in receiving the signals. We shall now begin the study of a third main branch of the sub- ject; namely, Electrical Resonance. Introduction to a Study of Electrical Resonance. In previous chapters attention has been called to the importance of bringing different parts of the sending and receiving circuits into resonance with one another. By this means the strength of the signals is increased, and the interference arising when several stations are operated simultaneously is partially eliminated. The main elements of variation in attuning circuits one to another are inductance and capacity. Preparatory to the study of more complex cases of resonance, let us recall the experiments of Sir Oliver Lodge, described in Chapter VIII, in which two Ley- den-jar circuits were attuned to each other. One of the Ley den- jar circuits, which I shall call the oscillating circuit, was provided with a spark gap, and was charged by an electric machine and allowed to discharge. The other Leyden-jar circuit (compare Fig. 142) was at a distance of perhaps a meter or two from the oscil- lating circuit, and could be adjusted as to period of vibration by a 215 216 WIRELESS TELEGRAPHY variation of its self -inductance by means of the slider C'D'. When brought into resonance by adjusting its period to that of the oscil- lation circuit, this second circuit was thrown into a violent state of electrical oscillation which might even break through the glass of the receiving Ley den jar unless provision for preventing this were made by providing it with a spark gap in shunt with the jar. FIG. 142. .Lodge's resonant circuits. The presence of a maximum sparking across this gap served to indicate that the jars were in resonance. Drude 's Use of Lodge's Resonant Receiving Circuit for Deter- mining the Period of the Oscillating Circuit. In 1902 Professor Paul Drude L published a description of a resonant method for determining the period of an oscillatory condenser discharge. Drude used an apparatus in every way similar to Lodge's receiving circuit, with, however, capacity and inductance of such shape as to be easily calculable, and with a scale attached to the inductance, so that the period of the receiving circuit was known for any par- ticular adjustment of the variable inductance. Such a calibrated receiving circuit is a frequency meter or a wave meter. Reference is made to Fig. 143. Suppose that it is required to determine the period of the oscillating circuit, shown as Circuit I. The fre- quency meter, shown as Circuit II, is brought up near the oscilla- 1 Annalen der Physik, Vol. 9, p. 611, 1902; see also Vol. 60, p. 17, 1897. ELECTRICAL RESONANCE 217 tory circuit, and by adjusting the slider S of Circuit II this circuit may be brought into resonance with the Circuit I of unknown period. The condition of resonance is indicated by the maximum glow in a sensitive vacuum tube in contact with one of the plates of the condenser of the frequency meter. When this resonant ()sciHation Circuit Scale ^ Drude's NX Frequency Meter Vacuum Tube Indicator FIG. 143. Drude's resonant method of measuring wave-length and frequency. adjustment has been made, the position of the pointer P on the scale is read, and from this reading the period of the frequency meter is known, for by calculation Drude has calibrated the fre- quency meter in terms of the period corresponding to any par- ticular adjustment of the pointer on the scale. The period of the frequency meter at resonance is the same as that of the oscillating circuit; which is, therefore, also known. Likewise, the wave length in air that Circuit I emits is known, for this wave length is the velocity of light times the period. 1 In terms of units, Wave length in meters = 3 X 10 8 X period in seconds. By means of this apparatus Drude was able to determine wave lengths between 2 and 445 meters. Doenitz's Wave Meter. Dr. Johann Doenitz 2 of Berlin, Ger- many, has constructed a wave meter that is in a very compact and convenient form for measuring the wave lengths of wireless teleg- raphy. Instead of a gradually variable inductance, as in Drude's apparatus, Doenitz's instrument has a gradually variable condenser 1 See Chapter X. 2 Elektrotechnische Zeitschrift, Vol. 24, pp. 920-925, 1903. German Patent, No. 149,350, from April 4, 1903. U. S. Patent, No. 763,164, filed Sept. 15, 1903, issued June 21, 1904. 218 WIRELESS TELEGRAPHY fb (Fig. 144). This variable condenser consists of two sets of semi- circular plates, one fixed and the other movable by rotation on a vertical axis. The capacity of the condenser is thus changed by bringing a larger or smaller area of the two sets of plates into interlapping position. This is the condenser of Korda x described FIG. 144. Doenitz wave meter. in Chapter XIV. The condenser is provided with a pointer passing over a scale t. This scale on the wave meter is calibrated directly in wave lengths. In series with the variable condenser is a loop of wire s, and another smaller loop i. When the instrument is brought up near an oscillating circuit so that the oscillations act inductively on the loop s, currents are induced in the wave-meter circuit, and these currents are the larger the nearer the period of the wave meter 1 Korda, German Patent, No. 72,447, issued Dec. 13, 1893. ELECTRICAL RESONANCE 219 approaches resonance with the oscillating circuit whose wave length is to be measured. Resonance is determined by noting the amount of current in the wave-meter circuit. This is done by means of a Harris or Riess hot-wire air thermometer h, which is, however, not connected directly into the wave meter circuit, but is coupled with it by means of the oscillation transformer Hi. The action of this transformer and thermometer is as follows: The primary i of the transformer is in the wave-meter circuit; the secondary ii of the transformer is in series with a resistance w, designed to be heated by the current through it. This heating of the resistance heats a quantity of air in a glass bulb surrounding the resistance, causing this air to expand, and to push up a column of mercury in the bent tube h. As the wave meter approaches resonance with the oscillation circuit, the rise of the column of mercury in the bent tube increases. By reading this indicator, not only can one determine the reso- nant adjustment of the wave-meter circuit, but one can also form some idea of the sharpness of resonance by noting whether small or large variations of the condenser are required for a given rise of the indicator. The range of wave lengths measurable by Doenitz 's wave meter is changed by substituting various coils of different numbers of turns for the receiving loop s. For each of the coils there is a corresponding calibration of the scale. Sample of Observations Made with a Doenitz Wave Meter. - The curves of Fig. 145 were obtained l by a Doenitz wave meter. The curves show the scale reading of the air thermometer for vari- ous settings of the wave meter. Curve I was obtained by tuning the wave meter to an oscillating antenna circuit; Curve II was obtained by tuning the wave meter to an oscillating condenser circuit. The condenser circuit and the antenna circuit are seen to have the same wave length, 320 meters, indicated by the fact that this value, 320 meters, is the reading of the wave-meter scale when the thermometer scale reading is a maximum. Now when the two circuits of curves I and II were coupled together, and the wave meter applied to a study of the oscillations occurring in the coupled system, the results plotted in Curve III were obtained. The resonance curve in this case has two maxima. To this subject we shall return. 1 Figure 145 is copied with some slight modifications from Lieutenant- Commander S. S. Robison's Manual of Wireless Telegraphy, 1906. THE UNIVERSITY OF 220 WIRELESS TELEGRAPHY Fleming's Wave Meter. A wave meter devised by Professor Fleming, 1 and called by him a cymometer, is adjustable to reso- nance by gradual variation together of both the capacity and the inductance. A photograph of the instrument is shown in Fig. 146. The condenser consists of two concentric brass tubes and I, separated by a vulcanite dielectric V. The variable in- ductance is a coil of wire LM wound on a tube of vulcanite, and is varied by a clip K, sliding over the bared wire of the coil. The 250 300 350 Wave-length in Meters FIG. 145. Curves obtained with the Doenitz wave meter. electric oscillations to be measured act inductively on the receiv- ing loop, consisting of the condenser 01, the inductance LM, and the wire PQ, which are in series. The clip K of the inductance, the tube of the condenser, and the pointer passing over the scale S are moved together by the handle H, so that the capacity and the inductance of the instrument are increased or decreased to- gether and almost uniformly along with the motion of the pointer. The condition of resonance is indicated by a maximum glow of a Geissler tube G attached to the terminals of the tubular condenser. 1 Fleming: The Principles of Electric Wave Telegraphy, p. 404. Long- mans, 1906. ELECTRICAL RESONANCE 221 An advantage of Fleming's cymometer over other forms of wave meter arises in the fact that the scale readings are nearly proportional to the wave length (giving a nearly uniform scale when calibrated in wave lengths), whereas with instruments of the Doenitz type the wave length is nearly proportional to the square root x of the capacity of the adjustable condenser, so that the divisions on the scale become wider apart as the wave lengtH increases. Fleming's instrument has, however, the disadvantage of lack of compactness, for the inductance and condenser of this instru- ment are from one to two meters long. Pierce Wave Meter. I have designed a wave meter that has met with some use in practical application to wireless telegraphy. It consists of a Korda semicircular plate condenser C (Fig. 147), in series with a loop L for receiving the inductive action, and in FIG. 146. Fleming cymometer. series with a specially constructed high-frequency telephone receiver T. A pointer carried by the axle of the movable plates of the condenser passes over a scale, which is calibrated directly in wave lengths. At resonance, a maximum sound is produced in the high-fre- quency telephone receiver. On account of the high sensitiveness of the telephone receiver the wave length of currents in which the oscillations are extremely feeble may be determined, and also, on account of this high sensitiveness, the condenser can be made very compact and light, so that the whole instrument in the standard form weighs only 14 pounds. 1 This the reader may verify by examining the formula X = 2 TTU x/LC. 222 WIRELESS TELEGRAPHY For extending the range of the instrument to long wave lengths, an inductance (to right of condenser) is included in the instrument and can be thrown in or out of circuit by the switch S. Also the receptor loop L has a double rotation. One rotation is about an axis running from right to left in the picture, so that the loop can be placed parallel to any oscillating circuit; and the other rotation is about an axis perpendicular to the figure, so that the loop may be folded back over the pointer and inclosed in the cover of the instrument, as is shown in the lower cut of Fig. 147. FIG. 147. Pierce wave meter. Calibration of Wave Meters. One method of calibrating a wave meter is by tuning it to resonance with various lengths of two parallel wires. This method, which we have already de- scribed, is applicable up to about 200 meters. For greater wave lengths the parallel-wire method is cumbersome, and, according to Diesselhorst, shows a systematic error, increasing with increase of wave length. With wave lengths above 200 meters I have employed the device of photographing the spark of a discharge circuit with the aid of a revolving mirror. The revolving mirror apparatus was like that of Fig. 4, with, however, the addition of ELECTRICAL RESONANCE 223 an accurate device, called a " stroboscope," for determining the period of revolution of the mirror. Having the period of revolution of the mirror and the distance between spark-terminal images on photographs like those of Fig. 3, one has a direct measurement of the period T of the discharge of a given oscillating circuit. By constructing a large number of such oscillatory discharge circuits giving various periods of discharge, or, better, by using a discharge circuit whose period could be varied at will, one may obtain accurate values of various periods by the use of the revolving mirror; and from the various periods T one can obtain the wave length X in air of the emitted wave by the f rmula X = v X T, where v = 3 X 10 8 meters per second, T is the time in seconds of one complete oscillation of the circuit, and X is wave length in meters. The wave meter to be calibrated is now set to resonance with each of these known wave lengths and the wave length is written at its appropriate position on the scale of the instrument. Another method of calibrating a wave meter is by tuning it to resonance with circuits of which the period is known by calcu- lation from a knowledge of capacity and inductance. Method of Using a Wave Meter. Let it be required to deter- mine the wave length in air emitted by the oscillation circuit S, FIG. 148. Position of wave meter for determining the wave length or frequency of the circuit S. Fig. 148. The wave meter must be placed in such a position that the magnetic force from S links with the loop L of the wave meter; the oscillations in S then act inductively on the wave meter. This action is a maximum when the loop L is close up to S and in a plane parallel with it. It is, however, not advisable to have the two circuits too close together, because in this case the oscillations 224 WIRELESS TELEGRAPHY induced in the wave-meter circuit react on the oscillating circuit and change its period. With the wave meter in inductive relation to the discharge circuit, by adjusting the condenser of the wave meter, a maximum deflection is obtained in the hot-wire air thermometer in the case of the Doenitz wave meter. This deflection is a maximum when the wave meter is adjusted to resonance with the discharge circuit; and when this adjustment has been made the required wave length is read off directly on the calibrated scale. With the use of the Fleming wave meter a maximum glow is obtained in the Geissler tube, at resonance, and the corresponding wave length is directly read. With the Pierce wave meter a maximum sound is obtained in the high-frequency telephone, and the corresponding wave length is directly read. Use of the Wave Meter in the Determination of the Capacity of the Discharge Condenser. Professor Fleming has pointed out the utility of the wave meter in the determination of the capacity of a condenser. His method consists of discharging the condenser across a spark gap through a known inductance and measuring the wave length produced. He then calculates the capacity C by use of the formula X = v . 2 TT VZC7 where X = wave length in meters measured by the wave meter, v = 3 X 10 8 (the velocity of light in meters per second), and L = the known value of the inductance through which the discharge occurs. A sample set of observations that I have taken in this way is shown in Table XII. The values of the inductance in the discharge circuit (see first column) were obtained by a bridge method. TABLE XII DETERMINATION OF CAPACITY BY THE WAVE METER. LEYDEN JAR NO. 45 Inductance in Discharge Circuit in Henrys. Wave Length in Meters. Capacity in Farads Computed by Thomson's Formula. 3.10X It)" 5 690 . 00432 X 10~ 6 4.90 865 .00432 6.61 1005 .00430 8.35 1130 .00432 10.0 1235 .00430 12.0 1345 .00427 14.05 1450 .00418 16.1 1560 .00423 Mean, . 00428 X 10~ 6 1 per cent. ELECTRICAL RESONANCE 225 The last column contains eight independent determinations of the capacity with an average error of only 1%. This is one of the best methods of determining the capacity of a condenser under conditions of actual use. Effect of Resistance on the Sharpness of Resonance. In tun- ing a condenser circuit with adjustable capacity or inductance to resonance with an oscillating circuit, as .was done in the wave metrical experiments above described, we have a simple case of the kind of tuning that is made use of at a receiving station when it is desired to receive signals of one wave length and exclude signals of a different wave length. One of the main difficulties in completely excluding undesired signals arises from the fact that the detectors used in receiving the signals have a high resistance. Let us see how the sharpness of resonance is affected by resist- ance of the receiving circuit, in the simple case in which a con- denser circuit (e.g., the wave-meter circuit) is attuned to a given wave length. As an example, I shall take a case in which the constants of the receiving circuit are within the range employed in wireless teleg- raphy. In Fig. 149 suppose that L is an inductance of .0001 henry, / an instrument for measuring the oscillatory current (root of mean square cur- rent) produced by an incoming electric wave, which is supposed to have a wave length \i = 300 meters; C is a variable ca- pacity, and this capacity is supposed to be calibrated directly in wave lengths, \2. Let the receiving circuit be set at various wave lengths and let the corresponding current be read on the instrument 7. FIG. 149. Simple oscil- By a calculation that is not here repro- duced, it can be shown that the results plotted in Fig. 150 will be obtained. The relative current is plotted vertically, while the settings of the wave length of the receiving circuit divided by the wave length of the incident wave (\2/Xi) are plotted horizontally. The different curves in the diagram show the effects of putting different values of the resistance R into the receiving circuit. A maximum current is received in each case when X 2 = Xi, but the sharpness or flatness of the curves depends on the value of R. When R = 628 ohms the top curve is obtained. This curve is nearly 226 WIRELESS TELEGRAPHY flat, so that large changes in, the period of the receiving circuit pro- duce only small diminutions of the received current. Going suc- cessively to the values R = 314, 207, 125, 63, and 6.3 ohms, we get sharper and sharper resonance, shown by the curves corre- sponding to these values of R. By a further examination of this set of curves we can see how well the receiving circuit with various values of resistance can discriminate between signals of different wave lengths coming at the same time. Suppose, for example, with the 300 meter wave coming, we try to get a message of wave length 1.20 X 300 = 360 628 Ohms 6.3 Ohms .70 .80 .90 1.00 1.10 1.20 1.30 1.40 1.50 1.60 1.70 1.80 1-90 Wave-length Relative to Resonant Wave-length FIG. 150. Effect of resistance on sharpness of resonance, assuming a constant inductance of .0001 henry. Wave length varied by varying capacity. meters. If we have only 6.3 ohms in the receiving circuit, and set for 360 meters (1.2 in horizontal scale), we should get about 3% of current from the 300 meter wave along with the full value of the 360 meter wave. This would usually not cause any difficulty. If; on the other hand, our receiving circuit has a resistance of 63 ohms we should receive 33% of the 300 meter wave along with the 360 meter wave ; and with a resistance of 628 ohms in the receiving circuit, we should receive 95% of the full current of the undesired 300 meter wave when we were in tune for the 360 meter wave. ELECTRICAL RESONANCE 227 In this problem I have supposed ;that the waves which are arriving are themselves undamped. If they also have strong damping, the interference would be a little greater than that described, but the main imperfections of tuning are due to the resistance of the receiving station and not to the lack of purity of the wave from the sending station. The illustration shows that we cannot get very sharp resonance so long as we have to use a high resistance (the detectors) in the particular receiving circuit here employed. This difficulty is, however, considerably reduced by the use of coupled circuits at the sending and receiving stations, in the place of the simple condenser circuit of this computation. In the next chapter some facts in regard to resonance with coupled circuits will be presented. CHAPTER XXI ON RESONANCE (Continued) ON THE ELECTRICAL OSCILLATIONS OF CONNECTED SYSTEMS OF CONDENSER CIRCUITS .HAVING briefly examined the conditions of resonance in simple condenser circuits, let us next consider the case of the coupled circuits involving principles that are now generally employed in sending and receiving the signals of wireless telegraphy and wireless telephony. Reason for Using Coupled Sending Circuits. The chief reason for the use of coupled circuits at the sending station is as follows : A closed condenser circuit is not a good radiator of electric energy, hence an antenna is employed for the purpose of radiating the energy. But on account of the comparatively small capacity of the antenna we cannot easily apply large amounts of power l directly to the antenna without using a very long spark gap in the antenna, so as to get the necessary high potential. Now the use of a long spark gap carries with it disadvantages; it does not pro- duce good oscillations. To avoid this disadvantage, the high potential in the antenna is obtained, not by the use of a long spark gap, but by the induc- tive action of a discharge occurring in a condenser circuit con- nected with the antenna and put into resonant relation with it, as shown in Figs. 151 and 152. The large amount of power in the condenser circuit is attained by the largeness of the capacity instead of by the length of the spark gap. By the use of a suitably large capacity in the condenser circuit we can obtain tremendous current in this circuit, which will induce very large potential in the antenna, if the antenna is in resonance with the condenser circuit. We thus get a large amount of radiation. It is proposed to describe some experiments on the oscillations of connected systems of circuits. Attention is here chiefly confined to the sending station. The receiving station will be examined later, j p _ No. of charges per second X Capacity X (Maximum Potential) 2 228 RESONANCE OSCILLATIONS OF COUPLED SYSTEMS 229 Simplified Form of Circuits. In order to simplify the con- ditions somewhat, in the present experiments, instead of employ- ing the wireless telegraph circuits with the antenna constituting U>-1^ Key FIG. 151. Inductively coupled transmitting station. FIG. 152. Direct coupled transmit- ting station. FIG. 153. Inductively coupled condenser FIG. 154. Direct coupled con- circuits, with the antenna and ground denser circuits, of Fig. 151 replaced by a condenser. the capacity of the secondary circuit (such an antenna being in the form of a capacity distributed along a wire also possessing inductance), this antenna, for the purposes of these experiments, is replaced by a condenser, so as to have a localized capacity in 230 WIRELESS TELEGRAPHY each of the circuits. While this change does not simplify the experiments, it enables us to apply certain fairly simple theoretical FIG. 155. Sketch of inductively connected condenser circuits. formulas to the examination of the result. Without these for- mulas, we should have difficulty in seeing any interrelation among the results and the constants of the circuits used in the experiments. FIG. 156. Sketch of direct coupled condenser circuits. Our simplified circuits are of the forms shown in diagram in Figs. 153 and 154 and in sketch in Figs. 155 and 156. RESONANCE OSCILLATIONS OF COUPLED SYSTEMS 231 In Fig. 155, which represents the inductively connected system, two condensers C\ and C 2 are connected to two coils L\ and L 2 , which are inductively related but insulated from each other. The number of active turns of wire on each of the coils may be varied ; Z/2 is varied by the clip contacts, and L\ is varied by a wheel contact that may be moved along the inner spiral by a rotation of the drum on which the inner spiral is "vound. Each of the condenser circuits is provided with a spark gap, so that either circuit, when connected to a step-up transformer, may be used as the discharge circuit. The other circuit may then be looked upon as a secondary circuit. When the spark gap of the secondary is opened too wide to permit the passage of a spark, or, what is the same thing, when the secondary is removed, the period of oscillation is the period of the primary alone. When, on the other hand, the secondary is left in place and the spark gap of the secondary is closed (compare Fig. 153), the oscillations of the discharge circuit C\ L\ induce oscillations in the secondary circuit C 2 L 2 , and we have a periodic flow of current in both circuits. It is proposed to give an account of some measurements of the wave length produced in the circuits when uncoupled and then when coupled with each other, . and to compare the measured values with values computed from certain useful formulas. In the Direct Coupled System, represented in Fig. 156, which was also studied, the transformer of the inductive coupling is replaced by an auto-transformer; that is, the two condensers C\ and C 2 are made to discharge through parts of the same coil. In this case, also, both the inductances LI and L 2 can be varied independently by the motion of the contacts W and S. Also, both the condenser circuits are provided with spark gaps, so that either circuit may be caused to oscillate alone or to constitute the discharge circuit in a connected system with closed secondary. These two forms of circuits, Figs. 155 and 156, are derived from the ordinary wireless telegraph circuits by replacing the antenna and ground of the wireless telegraph station by the two coatings of a condenser respectively. The circuits in these simplified forms will yield results that will aid in understanding the actual wireless telegraph circuits, which are to be examined in subsequent chapters. Dimensions of the Inductances. The coils employed in the apparatus shown in Figs. 155 and 156 had the following dimensions: 232 WIRELESS TELEGRAPHY Coil. No. Turns. Diam. Wire. Coil Diam. Pitch. Outer of Fig. 155 24 .208 cm. 18 cm. .81 cm. Inner of Fig. 155 51.5 .208 13 .42 Coil of Fig. 156 51.5 .208 13 .42 The inductances of various numbers of turns of these several coils were measured on a Rayleigh's bridge, and these values are recorded in the subsequent tables which contain the wave-length measurements. General Statement of the Results. Because of the difficulty of following the details of the experiments, I shall make at the outset a general statement as to the results. When two circuits are coupled together, either inductively or directly, the primary circuit will have two periods of oscillation, and the secondary will have the same two periods of oscillation; and this is true, with a few exceptions, even when both of the cir- cuits have been attuned to the same period before being coupled together. This double periodicity of the oscillation of the coupled circuits produces two distinct wave lengths, so that a coupled system emits two waves. Also, the energy of the t oscillation is at first all in the primary circuit, and gradually passes over into the secondary circuit, during which process the current in the primary becomes less and less with each vibration, while the current in the secondary becomes more and more with each vibration. After the energy has all gone into the secondary, the current in the primary becomes zero. Then the energy gradually comes back into the primary and the current in the secondary be- comes zero. This process may repeat itself many times. Experiment with Sympathetic Pendulums. As a digression, in order to see just how this takes place, the reader is asked to set up a simple apparatus like that shown in Fig. 157, which consists of two pendulums hung from a loosely suspended transverse cord about 3 feet long. This supporting cord may be tied to any two FIG. 157. Coupled pendulum. RESONANCE OSCILLATIONS OF COUPLED SYSTEMS 233 convenient objects; for example, the backs of two chairs. The pendulum bobs may be any two small bodies of about the same weight two heavy nails will do. At first make the lengths of the threads supporting the two pendulum bobs the same. Now leave one of the bobs at rest, pull back the other in a direction at right angles to the plane of the strings, and then release it. Note what happens. Try the effect of making the cross cord tighter or looser, and also the effect of making the two pendulums of unequal length. The vibratory motion of the pendulums represents very well the electrical vibratory motion that takes place with the coupled condenser circuits. Oscillograms of the Pendulum Motion. In order to show graphically the nature of the pendulum motion, I have elaborated the pendulum apparatus a little, and taken a moving picture FIG. 158. Coupled pendulum with arrangement for photo- graphing the motion. (oscillogram) of the motion of each of the pendulum bobs. To do this, a camera was placed in the position shown at C in Fig. 158. At the back of the camera is a small horizontal slit A, and back of this slit is a sheet of bromide paper F carried by a rotating drum D. In order to have a bright object upon which to make the exposure, a small Nernst glower G was hung just above one of the pendulum bobs. This Nernst filament was put into an electric circuit by means of the small wife W, which also served as the suspension for the pendulum, and by means of the return wire R, which was carried up in such a manner as not to interfere with the freedom of motion of the pendulum. The current was started in the glower by heating it with a match while the current was on. As the pendulum swung, the image of the Nernst glower moved back and forth along the slit A . A small horizontally moving point of light thus entered the slit and fell upon the film. If now the sensitive 234 WIRELESS TELEGRAPHY paper is given a uniform motion by the rotation of the drum, the paper moves vertically past the slit, while the image of the swing- K A f\ A \ r\ i FIG. 159. Photographs of motion of the coupled pendulum. ing light moves horizontally along the slit. The combined effect of these two motions is a wavy line on the photographic paper. Four curves thus obtained are shown in Fig. 159. These curves RESONANCE OSCILLATIONS OF COUPLED SYSTEMS 235 show the displacement of the bob plotted vertically, against time plotted horizontally. The first curve P, of Fig. 159, was obtained by leaving the ball M initially at rest, and pulling aside and releasing ball L (Fig. 158). The motion here corresponds to the primary current in the coupled condenser circuits. The second curve S was obtained by leaving the ball L initially at rest and releasing M. This curve corresponds to the secondary current of the coupled condenser circuit. The two cords supporting L and M were of the same length in the case of these two experiments. As another experiment, the two cords were both equally short- ened, and the transverse supporting cord was loosened; the curves P' and S' were obtained for the motion of the ball L initially dis- placed (primary) and initially at rest (secondary) respectively. The curves P and S or P' and S' represent very well the electrical vibratory motion of the coupled condenser circuits, if we think of the displacement of the bob in the two curves as representing the current in the primary and secondary circuits of the coupled- condenser oscillation. How the Curves Show the Existence of Two Periods. Each of the curves of Fig. 159 shows the existence of two periods, in the motion of the pendulum, by the presence of "beats." If two vibra- tions of different periods coexist in the same system, the slower of these vibrations will fall more and more behind the other in phase until the two vibrations become just opposite to each other and neutralize each other; then the slower vibration will again fall more and more behind till it is a whole vibration behind the faster, and the two vibrations will then add and intensify each other. This is what has happened in the experiment with the pendulums. The same thing happens with the electrical vibrations of the con- denser circuits that are coupled together. Theoretical Values of Wave Lengths in the Coupled Circuits. - Let us now return to the experiments with the condenser circuits. By the use of the wave meter we can pick out and measure each of the periods or the corresponding wave lengths of the connected system of condenser circuits. When this has been done, we shall find that the wave lengths obtained satisfy the following theoretical relations l : 1 Lord Rayleigh, Theory of Sound; J. v. Geitler, Sitz. d. k. Akad. d. Wiss. z. Wien, February and October, 1905; B. Galitzine, Petersb. Ber., May and June, 1895; V. Bjerknes, Ann. der Physik, Vol. 55, p. 120, 1895; Oberbeck, Ann. der 236 WIRELESS TELEGRAPHY x/= X 2 ' = X 2 2 + V(X 1 2 -X 2 2 ) 2 H-4r 2 X 1 2 X 2 2 - X 2 2 )+ (1) (2) In these equations Xi= the natural wave length of the primary alone, X 2 = the natural wave length of the secondary alone, r = the coefficient of coupling, defined by the equation, , (3) where M = mutual inductance between the two circuits, LI = self-inductance of the primary, L 2 = self-inductance of the secondary, and X/ and X/ are the resultant wave lengths in both primary and second- ary when the circuits are coupled together. I shall give 1 a few mea- surements of the wave lengths obtained with the coupled condenser circuits, together with a compari- son with values computed from formulas (1) and (2). Experiment with Induc- tively Connected Circuits. Lz = 24 Turns of Outer Coil. (Fig. 155) X 2 = 1060 Meters. The results ob- FIG. 160. Curves showing observed and cal- tained in this experiment culated values of the two wave lengths , , . _ produced by two condenser circuits indue- are plotted in rig. loU, tively coupled. an( j a complete record of the observations is given in Table XIII. Physik, Vol. 55, p. 623, 1895; Domalip and Kolacek, Ann. der Physik, Vol. 57, p. 731, 1896; M. Wien, Ann. der Physik, Vol. 61, p. 151, 1897, and Ann. der Physik, Vol. 8, p. 686, 1902; compare also Webster, Theory of Electricity and Magnetism, p. 499, 1897, and Fleming, The Principles of Electric Wave Teleg- raphy, p. 209, 1906; also Cohen, Bui. Bu. of Standards, Vol. 5, p. 511, 1909. 1 For a fuller description of these experiments and several others of a similar character see Pierce: Physical Review, Vol. 24, p. 152, 1907. 1600 1400 1200 10CO 800 600 400 200 ^ E.^I.C. Outer 2' T > s / c Oj- C 2 = 'X-2- 2 R -.0043. =.0048 -. 1C50 _ * 2m.f. 2 M s / z ^^ >? / / z t ~&r* ^' I "" Xo ^ s* / 2CO 400 600 800 1000 1200 1400 16( Xi Meters " " * x Observed RESONANCE OSCILLATIONS OF COUPLED SYSTEMS 237 TABLE XIII INDUCTIVELY CONNECTED SYSTEM Primary capacity .00432 microfarad. Primary inductance varied. Secondary capacity .00482. Secondary inductance 24 turns outer coil,!/., = 6.60 X 10 ~ 5 henry. Wave length of secondary X 2 = 1060 meters." Turns Primary. LI Primary Inductance. Henry. M Henry. T 2 50 15.85X 10~ 5 6.52X 10~ 5 .412 45 13.9 6.14 .421 40 11.8 5.80 .430 35 10.0 5.12 .397 30 8.20 4.45 .360 25 6.50 3.56 .295 20 4.82 2.70 .228 15 3.15 1.95 .183 10 1.72 1.20 .128 5 .69 .47 .048 3 .32 .23 .0277 Calculated. Observed. Turnc \ 1 Urllb Primary. A l Meters. V V V X 2 ' Meters. Meters. Meters. Meters. 50 1560 1740 727 1750 710 45 1460 1670 712 1650 685 40 1350 1567 686 1570 465 35 1230 1462 680 1480 660 30 1130 1390 660 1370 660 25 1000 1273 685 1280 660 20 870 1185 680 1185 630 15 700 1127 595 1125 565 10 510 1080 467 1090 460 5 300 1060 292 1040 285 3 210 1062 193 1075 210 The method of taking the observations is as follows : First, the condenser Cz ( = .00482 mf.) was connected in series with 24 turns of the outer coil (Fig. 155) and was provided with a spark gap. In this position, with the inner coil thrown out of circuit by dis- connecting both plates of its condenser, the wave length X 2 was found to be 1060 meters. Next, with the secondary condenser disconnected, the wave length of the primary (inner) circuit was determined with its condenser Ci ( = .00432 mf.) connected in series with 50 turns of the inner coil. This wave length \i was 1560 meters. Next, with the primary left unaltered, the second- ary was closed by attaching its condensers without spark gap to the 24 turns of the outer coil. This is the case of the closed second- 238 WIRELESS TELEGRAPHY ary, and when the discharge was established in the primary, the wave lengths were found to be X 2 ' = 710 meters and X/ 1750 meters. The value of X 2 ' and X' 2 were plotted against Xi = 1560, Fig. 160. Now decreasing the primary inductance to 45 turns, the values Xi = 1460, X/ = 1650 and X 2 ' = 685 were obtained, and the last two values were plotted against the value of Xi, and so forth. The complete curves of Fig. 160 were obtained in this way. In the curves of Fig. 160 the crosses are the observed values and the circles are the corresponding calculated values. The 45 line between the two curves may be looked upon as Xi plotted against itself, while a horizontal line across the figure at 1060 meters (not shown) would represent the graph of X 2 . With this in mind it will be seen that the two derived wave lengths X/ and X 2 ' approach X 2 and Xi respectively toward the origin. The observed and the calculated values are in satisfactory agreement. The formulas for the calculation of X/ and X 2 ' are the formulas (1) and (2) of page 236, which involve merely the independent periods of the two circuits and their coefficient of coupling. The latter quantity was obtained by the measurement on a Rayleigh's bridge of Lj, L 2 and M for each setting of the oscillation circuit. The values of these inductances and the values of r calculated from them is also included in Table I. The intensity of the various periods of the circuits under the different conditions of the experiment varies greatly. No attempt was made to determine these intensities and the experiments are designed merely to show the wave-length relations. Experiments with the Inductively Connected System in the Special Case Where X 2 = Xj. A case of especial interest is the case in which the primary and secondary have the same independ- ent periods. This is the case of so-called " resonance " between the two circuits. In this case the wave-length formulas (1) and (2) become greatly simplified, as may be seen by substituting X 2 = Xi in these equations, which under this condition become ( X /)2 = Xl 2 (l+r), (4) (X 2 ') 2 = X! 2 (1 -r). (5) In the present experiment the two independent wave lengths Xi and X 2 were made equal, and the wave lengths produced by the com- pound system were then measured and compared with calculations from the formulas (4) and (5). Two wave lengths X/ and X 2 ' were RESONANCE OSCILLATIONS OF COUPLED SYSTEMS 239 obtained both by measurement and by calculation. The observed and calculated results are plotted in the curves of Fig. 161. In this case also the agreement is fairly satisfactory. These two experiments with the inductively connected system of circuits give an experimental verification of the formulas (1), (2), (3), (4) and (5), and serve to show how the wave lengths obtained with the connected system depend on the constants of 1200 10CO 800 600 400 200 E.^I.C. C, = 00432 m C 2 =00482 X 2 200 400 600 800 1000 12(K \ a Meters 4 +++ Observed oooo Calculated FIG. 161. Curves of wave lengths obtained with inductively coupled condenser circuits having individually the same period. the two circuits. We shall return to this subject after giving briefly the results of an experiment with the direct coupled system of circuits. Experiment with the Direct Coupled Circuit. C 2 = .00178 Microfarads, L 2 = 25.5 Turns = 6.7 X io~ 5 Henrys, X 2 = 645 Meters. The apparatus for this experiment with the direct circuit is shown in Fig. 156. The steps of the experiment are similar to those with the other system of circuits. The observed and calculated values of the wave lengths in the compound oscil- lating system are plotted in Fig. 162. The formulas of calculation are the formulas (1) and (2), and the agreement between the observed and calculated results (crosses and circles) is seen to be satisfactory. 240 WIRELESS TELEGRAPHY Xi D.C. SEC at 25.5 One interesting result shown by this experiment is the fact that the curve X 2 ' comes down to the horizontal axis in the neighbor- hood of Xi = 1010 me- ters. This point is the point of so-called per- fect coupling, and was obtained when the pri- mary and secondary condensers were both connected through the same inductance, 25.5 turns of the coil, as shown in Fig. 163. 1600 1400 1200 1000 800 600 400 200 .00432 00178 645-: m.f. 200 400 600 800 1000 1200 1400 1600 \, Meters -H-+ + Observed Xl K cooo calculated FIG. 162. Curves showing observed and calcu- lated values of the two wave lengths produced by two condenser circuits directly coupled. r lid FIG. 163. Diagram of a case of coupling that gives but a single wave length. In this case we may explain the result in two ways: (I.) The two condensers may be looked upon as discharging in parallel through the same inductance, and producing, therefore, only one wave of wave length Xi' = 2 TT v VLi(Ci+ C 2 ) = vV+ \ 2 2. (6) II. This result is also obtainable from the theoretical equations (1) and f (2) = 4 T - X 2 2 ) 2 + X 2 ' v- X 2 2 - (Xi 2 - X 2 2 ) 2 + 4r 2 X 1 2 X 2 2 (2) For when the primary and secondary condensers are connected about the same inductance, Ll = L 2 = M, therefore M 2 = 1. RESONANCE OSCILLATIONS OF COUPLED SYSTEMS 241 When r is equal to unity the coupling is said to be perfect and the equations (1) and (2) become Xi' = vV + X, 2 ; and X 2 ' = 0. That is to say, the oscillation, as shown also by method I, becomes single- valued. The case of perfect coupling was not observed in the experiments with the inductively coupled system, because for perfect coupling the primary and secondary coils must have the same number of windings and the two coils must be so close together as to be practically coincident, conditions that could not be realized with the inductive coupling. Close Coupling and Loose Coupling. One of the most interest- ing facts derivable from an examination of the equations (1) and (2), which are verified by the experiments, is the influence of the coefficient of coupling (T) on the wave lengths produced by the coupled circuits. In general, two wave lengths are obtained when a coupled system of circuits is set into oscillation. This duplicity of the wave length is often an inconvenience in wireless telegraphy, because, to avoid interference when a neighbor is sending a mes- sage we do not wish to hear, it is necessary to tune to avoid, not one undesired wave, but two. The influence of the coefficient of coupling on the wave length is very easy to investigate in case the primary and secondary of the coupled system are attuned to the same wave length X, as they generally are in practice. In this case, the formulas for the com- pound wave lengths X/ and X 2 ' become the simple forms of equation (4) and (5); namely, (Xi') 2 = X 2 (1 + r), (4) and (X 2 ') 2 - X 2 (1 - r). (5) Dividing each of these equations by X 2 , and extracting the square root, we have, -= VT+~r CO A *L = VT=~r (8) A Now, putting in various values of r = (.1, .2, .3, etc., up to 1.0), we obtain the relative values of X/ and X 2 ', shown in the curves of Fig. 164. 242 WIRELESS TELEGRAPHY These are all of the possible values of the derived wave lengths X/ and X 2 ' for the given wave length X, because r cannot be greater than unity. Circuits coupled together with a large value of r are called dose-coupled circuits; those coupled with a small value of r are called loose-coupled. The looser the coupling, the nearer the two derived wave lengths approach the wave length of each condenser 1.0 .2 .4' .6 ,8 1.0 1.2 1.4 1.6 Wa^e-lengths Cbupled -=- Wave-lengths Uncoupled FIG. 164. Effect of coefficient of coupling on the resultant wave lengths of a coupled circuit. circuit alone. For sharp resonance, then, the coupling ought to be loose; while for strength of signals the coupling ought not to be too loose. i Similar considerations apply more or less directly to the receiv- ing circuits also. This subject, of the closeness or looseness of the coupling, will come up again in connection with the actual wireless telegraph sending and receiving circuits, comprising an antenna circuit coupled with a condenser circuit, which are discussed in the next chapter. CHAPTER XXII TUNING THE SENDING STATION HAVING investigated, in the preceding chapter, the condi- tions of resonance and the manner of vibration of two condenser circuits connected together, it is proposed now to consider the actual wireless telegraph sending circuits. For this purpose let us examine the method of adjusting the direct coupled or the inductively coupled sending station to resonance. 0A diagram of a direct coupled sending station is shown in Fig. 165. The condenser C, repeatedly and periodically charged from a transformer Tr, discharges through a spark gap G and a few turns P of a " helix." The oscillations in this circuit act inductively and produce oscillations in the an- tenna circuit consisting of the antenna, the coils S of the helix, and the ground E. A maximum effect is produced when these two circuits are properly adjusted to each other. A photograph, Fig. FIG. 165. Direct^coupled transmitting 166> is gj ven to ghow the CQn _ struction of the sending helix (right) and a method of inclosing the spark gap for reducing the noise of the spark. A diagram of the inductively coupled sending circuit is shown in Fig. 167. Here the primary and secondary inductances are parts P and S of two separate helices. These two helices may be one above the other, as represented in the diagram, or may be one 243 244 WIRELESS TELEGRAPHY FIG. 166. Showing construction of helix and spark gap for a direct coupled transmitter. inside of the other, as shown in the photograph of Fig. 168. They must, however, be separated by sufficient distance to prevent sparking between them when the station is in operation. In this system of circuits also, the condenser circuit and the antenna circuit must be adjusted to reso- nance, in order to get strong oscil- lations in the antenna. We shall show the details of the process of attuning the primary and secondary of the coupled circuits to resonance (1) by the use of a wave meter, and (2) by the use of a hot-wire ammeter. Wave-Metrical Method of At- tuning a Direct Coupled Sending FIG. 167. Inductively coupled Station. To adjust the station transmitting station. ^o resonance one first disconnects the condenser circuit, as shown in Fig. 169, and places the wave TUNING THE SENDING STATION 245 meter WM up near the helix. The lower end of this helix is con- nected through a spark gap to the ground. The secondary of the station's transformer is connected about the spark gap. The FIG. 168. Showing construction of the helices of an inductively coupled transmitting station. antenna is connected by means of a clip contact K to some particu- lar number of turns of the helix. The transformer is set into opera- tion so as to produce a spark at the gap. 1 This sets up oscillations in the antenna circuit, and the wave meter is adjusted to resonance with these oscillations. The wave length is read, and this reading 1 In this case, where the spark gap is in the antenna circuit, there is a tend- ency for the spark to go over into an arc and not produce good oscillations. This may be obviated by playing a small blast of air on the spark. 246 WIRELESS TELEGRAPHY of wave length, together with the number of turns of helix, is entered in a table. The clip contact is now moved to another point on the helix, thus putting in more or less inductance in the circuit, and the wave length is again determined and entered with the number of turns in the table. A table is thus formed for the wave length corresponding to various numbers of turns of the helix. Condenser Disconnected FIG. 169. Method of determining wave length of antenna circuit. These results are then plotted, and give a curve like that marked " Antenna Circuit " in Fig. 170. A similar operation is performed with the condenser circuit. In this case the antenna and ground, see Fig. 171, are disconnected; and the condenser circuit, with the spark gap in series, is con- nected with various numbers of turns of the helix; and the wave length for each case is determined, and a curve of wave lengths against turns is plotted. The curve for this case is put on the same chart with the antenna observations, and marked " Condenser Circuit," Fig. 170. By a reference to the curves we can now obtain the number of turns required either in the condenser circuit or in the antenna circuit to produce TUNING THE SENDING STATION 247 1000 Turnsof Helix FIG. 170. Curves showing wave lengths of antenna circuit and condenser circuit with different numbers of turns of the helix. FIG. 171. Method of measuring wave length of the condenser circuit. 248 WIRELESS TELEGRAPHY a given wave length. For example, let it be required to have both the condenser circuit and the antenna circuit produce a wave length of 420 meters. One sees that to get this wave length in the condenser circuit one must use 1.4 turns of the helix, and to have the same wave length in the antenna circuit when alone, one must use in this circuit 5.1 turns. Hence, if we connect the condenser about 1.4 turns of the helix, and the antenna and ground about 5.1 turns of the helix, we shall have the two circuits in resonance, 1 and shall get powerful oscilla- tions induced in the antenna circuit under the action of the dis- charge in the condenser circuit. The method of making the required connections is shown in Fig. 165. Although the primary and secondary are now connected in resonance, the electrical vibration of the system is not a simple vibration giving 420 meters wave length. If we bring the wave meter up near the coupled system in operation, two positions of resonance are found on the wave meter corresponding to two wave lengths. In the actual case, from which the above numerical values were taken, these two wave lengths obtained were X 2 ' = 358 meters and A/ = 462 meters. This duplicity of resultant wave length exists in the antenna circuit and also in the condenser circuit and therefore gives rise to a series of beats like those obtained with the coupled pendulum experiments, described in the preceding chapter. Photograph of the Double Oscillation in the Antenna Circuit. In a wireless tele- graph station attuned to resonance, as just described, I inserted a small spark gap in the lead to ground just below the helix, and took a revolving-mirror photograph, a negative of which is shown in Fig. 172. mirror spark of the Although this photograph had to be made with a vei T brief exposure and is therefore faint, the beats are clearly visible, and at about every fourth oscillation the beats reduce the antenna current to zero. FIG. 172. Rotating inductively coupled on ' Compare Paragraph on "Detuning" on p. 251. TUNING THE SENDING STATION 249 Adjustment of Direct Coupled Sending Station to Resonance with' the Aid of a Hot-wire Ammeter. Another method of adjusting the condenser circuit and the antenna circuit to resonance makes use of a hot-wire ammeter, inserted in the antenna circuit as represented at A, Fig. 165. This instrument contains a fine wire through which the oscillations pass, producing heat. The heated wire expands, and by means of a delicate gearing attachment, the sagging of the expanding wire acts upon a hand passing over a dial. The movement of the hand over the dial is thus an indica- tion of the amount of current passing through the sensitive wire. The instrument may be calibrated directly in amperes, but this calibration (chiefly on account of the shunts that have to be employed) is without much absolute value, when the hot-wire ammeter is used with the very rapid oscillations of wireless teleg- raphy. Nevertheless, a maximum deflection of the instrument indicates a maximum of current in the antenna, and this is all that is required of the hot-wire ammeter in order to decide when the antenna and condenser circuits are in resonance. Instead of inserting the hot-wire ammeter in the antenna above the helix, it may just as well be placed in the lead from the helix to the ground. In either case oscillations in the antenna circuit pass through the instrument. To tune up a station with a hot-wire ammeter, let the station be coupled up as shown in Fig. 165. Set the transformer in action, and read the hot-wire ammeter. Now keeping the spark gap con- stant, and leaving the antenna clip K unchanged, move the clip K' of the condenser circuit to a different number of turns of the helix, and again read the current. Make a table containing the number of turns of helix in primary circuit and corresponding hot- wire ammeter readings. Then plot a curve of readings against turns in the form shown in Fig. 173. From this figure it is seen that the maximum reading of the ammeter was obtained when the primary was discharging through 1.3 turns of the helix. This is, therefore, the adjustment that must be given to the primary in- ductance in order to bring the condenser circuit into resonance with the antenna circuit, for the fixed value of the secondary induc- tance employed throughout the adjustment. Since the readings of the hot-wire ammeter depend on the values of the mean square current through it, one can, by a process like that described, find out just what conditions of the two circuits give the greatest mean square current in the antenna, and if 250 WIRELESS TELEGRAPHY everything is kept constant in the experiment except the induct- ance variation, one can determine the resonance adjustment by the maximum reading of the ammeter. But in extending the use of this instrument to other conditions it is necessary to keep in mind that the readings of the ammeter do not give any information of the current amplitude of an individual oscillation; it always gives s 4 d fed 1 10 20 40 50 60 70 80 90 100 110 120 130 140 150 160 Receiving Capacity FIG. 199. Relation of resonant receiving capacity to height of receiving antenna. as is shown by the following comparison of observed values, with values calculated from this equation (Table XIV) : TABLE XIV RELATION BETWEEN HEIGHT OF RECEIVING ANTENNA AND RESONANT CAPACITY. FOUR WIRES RECEIVING Curve No., Fig. 196. Meters Antenna Above Coil, H a . Maximum Deflection, cm. Resonant Capacity Ob- served, C4 Resonant Capacity Cal- culated. 1 23.8 64 92 91.9 2 20.8 47 94 94.4 3 17.8 43 100 99.3 4 15.8 29.5 106 106.6 5 14.8 21 115 113.9 6 13.8 13 130 128.6 7 12.8 7.5 165 172.6 The only large difference between the observed and the calcu- lated value of resonant capacity is in the case of Curve 7, where 284 WIRELESS TELEGRAPHY on account of the obtuseness of the experimental curve its maxi- mum could not be accurately determined. An examination of equation (a) shows that if we make H a (height of receiving antenna) = 11.8 meters, C 4 would become infinite. This is the interesting fact that, with the particular fixed inductance coils employed in this experiment, if our equation is exact, no adjustment of the side condenser would enable us to receive any appreciable amount of current of the particular wave length arriving from the sending station. The experiment showed this to be approximately true, notwithstanding the fact that the receiving antenna, 11.8 meters of 4 wires, was not very different from the sending antenna, 15.8 meters of 4 wires. Let us look at equation (a) again and suppose H a to be less than 11.8. Let H a be 10, then equation (a) shows that - 1.8 (C 4 - 84.6) = 88, C 4 - 84.6 = - 49, C 4 = 35.6; that is, having lost resonance at H a = 11 .8, if we decrease H a to 10 meters we ought to find the resonance again, but instead of finding it out near where we lost it (beyond C 4 = 170) the reso- nance ought to reappear at a comparatively small value of C 4 . This was tried with the following result: Search for the Other Branch of the Curve. As the height of the four wires of the receiving antenna was decreased by small intervals below the values that gave Curve 7, Fig. 198, the deflec- tions in the region of capacity between 90 and 180 became smaller and smaller, as if the resonant point were going away to infinity, and the deflections in the neighborhood of 50 began to grow, until when the height of the antenna was made 10.5 meters, a maximum became evident for about 50 cm. of the receiving con- denser. The readings by which this maximum was obtained are plotted as Curve 8 in Fig. 198, along with Curves 1 to 7. Decreas- ing the height still further, Curves 9, 10, 11 and 12 were obtained with respectively 10, 9, 8 and 7 meters as the height of the antenna. These curves increase in intensity up to Curve 10 and fall off in 11 and 12. The five Curves 8 to 12 were taken with sensitiveness of the receiving instrument about five times as great as the sensitive- ness used in taking Curves 1 to 7. Curves 8 to 12 in the left- hand group of Fig. 198 are plotted thus magnified five times in comparison with the group to the right, numbered 1 to 7. RESONANCE OF RECEIVING CIRCUITS 285 The two groups when plotted with resonant receiving capacity against height of antenna form a curve of two branches A, A', Fig. 199. Values calculated from the equation (a) are plotted as the dotted lines in Fig. 199. The heavy curves are the observed values. From a comparison of the observed values with the com- puted values, we see that our equation, although it led us to look in the right direction for the resonance, is yet an imperfect equa- tion. There are other terms in it beyond those here set down. \7 FIG. 200. Various types of inductively coupled receiving circuits. Applicability of these Experimental Results to Practice. One may ask, what is the use of this experiment in which the receiving transformer is kept constant and the length of antenna and the 286 WIRELESS TELEGRAPHY capacity of condenser in the receiving side circuit are varied, since we are not going to vary the length of antenna in actual practice? The answer is, that if we set up an antenna at random and depend upon variations of C* alone to get our resonance, we may have our antenna of a length (capacity) that bears to the waves we wish to receive the same relations that 11.8 meters of four wires bear to the waves of my experiment. In that case our tuning curve would correspond to Curve 7 of Fig. 198, and would give us very little current and very dull resonance. The remedy is: Tune the antenna circuit as well as the side condenser. This can be done by having (1) a variable primary of the receiving transformer or (2) a variable inductance in series with the primary, or (3) a variable condenser shunted about the primary, or (4) a variable condenser in series with the primary. The several methods are shown at (1), (2), (3), and (4) of Fig. 200, respectively. The methods (1), (2) and (3) permit an increase of the wave length of the antenna and adapt it to longer waves. The method (4) permits a decrease of the wave length of the antenna and adapts it to shorter waves. A very de- sirable arrangement is to combine all of these variables in one appa- ratus, as shown in Fig. 201. Then we can make such adjustments as are necessary for obtaining best resonance. In order to see further the applicability of the experimental curves of Fig. 198, let us ex- press in somewhat more general form the relation which we have given in the experimental equa- tion (a). Approximate Theoretical Equation for Resonance Relation at Inductively Coupled Receiving Station. If we have a wave of wave length X arriving at an inductively coupled receiving station of which the antenna circuit is adjusted to wave length X a , and the condenser circuit adjusted to wave length X c; then theory FIG. 201. An inductively coupled receiving station with several variable elements. RESONANCE OF RECEIVING CIRCUITS 287 shows that the following is approximately 1 the relation between the several wave lengths in order to produce a maximum current in the condenser circuit: in which r is the coefficient of coupling at the receiving station. By a maximum current in the condenser circuit one or another of the maxima of the twelve different curves of Fig. 198 is meant. Not all of these maxima are equally strong, nor is the resonance for all of the maxima equally sharp. But for nearly any value of X a we can get a valve of X c that will give resonance of a more or less pronounced character. Let us try a few numerical examples that will make this clear. Let r = .20; and suppose waves are arriving of wave length X = 400 meters. Suppose that our antenna wave length is set at X a = 300 meters. Then we have r = .20, X = 400, X a = 300, to determine X c . With these numerical values equation (1) becomes (1 1 M 1 1 ) _ (0.20) 2 K 2 (400)'} M300) 2 (400) 2 i Multiplying by (400) 4 we get (f -MS- ')= Whence X c = 390 meters. This 390 meters is the wave length at which we must set our receiving condenser (in a coupled circuit) in order to receive a 400-meter wave, provided our antenna is set for a 300-meter wave. Carrying through similar computations for other values of the wave length of the incident waves we obtain the results recorded in Table XV. 1 In the derivation of this formula the small effect of resistance on the wave length was neglected; also the capacity of the antenna was considered localized instead of distributed. The formula (of which our equation (a) is a special case) is, therefore, inexact, but will serve to illustrate some interest- ing facts about the tuning of a receiving station. 288 WIRELESS TELEGRAPHY TABLE XV RESONANT WAVE LENGTH ADJUSTMENT OF THE CONDENSER SIDE-CIRCUIT WHEN THE ANTENNA IS KEPT FIXED AT WAVE LENGTH \a = 300 METERS Wave Length of Incident Waves X. Resonant Value of Re- ceiving Condenser in Wave Length, X c . 100 103 200 210 250 267 280 330 290 430 300 310 207 330 302 350 332 400 390 500 493 600 598 The formula does not apply to the case of X = 300 meters, so this value is omitted from the calculations. 500 400 =300 ^100 100 200 300 Wave-length of Incident Waves 400 500 FIG. 202. Curves showing resonant adjustment of wave length of the condenser circuit for different values of incident waves, the antenna wave length being fixed at 300 meters. The results recorded in Table XV are shown graphically in Fig. 202. RESONANCE OF RECEIVING CIRCUITS 289 This curve shows several facts of interest. It shows, for example, that when we have been receiving a wave length slightly shorter than our antenna wave length, and a wave comes in slightly longer than our antenna wave, we must actually decrease our receiving capacity to bring tho longer wave into resonance. It shows also that any particular adjustment of our receiving capacity is reso- nant for two different waves. For example, with our antenna set at wave length 300 meters, and our condenser circuit set for 400 meters, we are really in tune for either a 290-meter wave or a 410- meter wave, not in the best tune, it is true, but sufficiently in tune to be disturbed if the interfering signals are strong. Advantage of Varying Coefficient of Coupling in Tuning. There are times when we wish to be in tune for two wave lengths at once, because the station we are receiving usually sends out two waves at once. If we set our receiving condenser at 300 meters, we are in tune for a 270-meter and a 330-meter wave, and these might well be sent out by the same station. They will in fact be sent out by the same station, if it has the same coefficient of coupling as our receiving station, r = .20, and has its condenser circuit and antenna circuit tuned to 300 meters. This suggests an important improvement in our tuning mech- anism at the receiving station; namely, a device by which we can change the coefficient of coupling at the receiving station and thus make the receiving coefficient of coupling identical with the coeffi- cient of coupling of any particular station we wish to receive. This device 1 is employed in many of the recent receiving sets, and consists of an adjustment by which the primary coil of the receiving transformer may be either moved away from or rotated with respect to the secondary coil. The same result can be at- tained by cutting out inductance in the primary of the transformer and putting it in series where it will not be in inductive relation with the secondary coil. Effect of Variation of the Coefficient of Coupling on Sharpness of Resonance and on Received Energy. Theory shows that diminution of the coefficient of coupling increases the sharpness of resonance. At the same time this diminution of coefficient of coupling brings with it a decrease of energy. I tried some experi- ments to see what improvement in sharpness of resonance we might 1 On account of the high resistance of the detectors the proper adjustment of the coefficient of coupling is not one of exact equality with the coefficient of coupling of the sending station, but must be determined by trial. 290 WIRELESS TELEGRAPHY attain by this method. With the very low-resistance dynamo- meter as a measuring instrument, the curves of Fig. 203 were obtained, with coefficients of coupling at the receiving station equal to .30 and .07 respectively. The energy received in the former case was twenty times as great as in the latter case; but to compare sharpness of resonance the two curves are both plotted with amplitude 100. With r = .30 the deflection falls to half for a change of con- denser capacity of 5%, while with r = .07 the deflection falls to 100 9C 80 70 fl60 40 30 20 10 V 20 16 12 8 4* 40 i Change of Capacity 12 16 20 24 FIG. 203. Sharpness of resonance for two different values of r, the coefficient of coupling. half for a change of capacity of 2.5%. The deflection is propor- tional to the energy, and the capacity is proportional to the square of the wave length, so we may say that the energy received falls to half for a variation of 2.5% and 1.25% of the wave length in the two cases. The experiment thus confirms the theoretical deduction that with a decrease of the coefficient of coupling the sharpness of resonance is increased. The gain in sharpness of resonance is, however, paid for in loss of energy, the energy received with r .07 being ^V of the energy received with r = .30. RESONANCE OF RECEIVING CIRCUITS 291 EFFECT OF RESISTANCE OF DETECTOR ON RESONANCE IN COUPLED WIRELESS TELEGRAPH CIRCUITS Although the coefficient of coupling of the coupled circuits influences somewhat the sharpness of resonance, a far greater influence in the case of the practical stations is exercised by the resistance of the detectors which are used in the reception of the signals. These detectors, when sufficiently sensitive to respond to weak signals, have a very high re- sistance. We have seen in Fig. 150 (p. 226) how a high resistance inserted in a simple circuit consisting o'f a condenser in series with an in- ductance renders the resonance dull. With the coupled circuits the effects are somewhat more difficult to present, and it is necessary to examine the resonance curves obtained by varying both the antenna wave length and the condenser- circuit wave length in order to ascertain the influence of resistance on the sharpness of resonance. -c, on/l ~. FIG. 204. Diagram I have submitted the problem to a mathe- of circuit provid- primary and sec- ondary by vari- able condensers. matical examination, and without giving the steps of the reasoning, I take the liberty of pre- senting some of the results. The form of re- ceiving circuits to which the discussion applies is shown in Fig. 204. The following constants of the circuits were assumed in the computations: L 3 = Self-inductance of the antenna circuit = .3 X 10~ 3 henry, L 4 = Self-inductance of the condenser circuit = .5 X 10~ 3 henry, M = Mutual Inductance = .2 X 10" 3 henry. r 2 = Square of coefficient of coupling =.267, X = wave length of incoming waves = 472 meters. The antenna circuit was given various resistances, J? 3 , and the condenser circuit various resistances, R*. The resistance R resides chiefly in the detector, and the resistance R s includes the apparent resistance due to distributed capacity in the antenna. The incoming waves were supposed to be a persistent train of undamped waves. Computations were made for two cases: I, When we fix the antenna adjustments at their best values, and tune with C 4 ; 292 WIRELESS TELEGRAPHY II, When we fix C 4 at its best value, and tune by adjustments of the antenna circuit. The appropriate adjustments for the two cases and the sharpness of resonance obtained depend in an inti- mate way upon the values of R s and R\. The results for the two cases are here briefly presented. Case I. R 3 = 10 Ohms, R 4 = 64,000 Ohms. Reference is made to the curve marked " # 4 = 64,000 " in Fig. 205. This curve is entirely flat on top, and shows that, with a detector of resistance 64,000 ohms used in a circuit with the constants we 100 200 300 Wave Length of Condenser Circuit, . .These lines show the value of X 3 H for resistances R 4 respectiveley * iff FIG. 205. Curves showing effect of resistance # 4 on resonance with a coupled system of circuits tuned by adjusting C 4 . Wave length of incident waves = 472 meters. have assumed, there is no possibility of discriminating between different wave lengths by any adjustment of the condenser C 4 . In this case we may as well leave C* set at any value above that which with the inductance L 4 gives 200 meters. It is then equally ready to detect all wave lengths. In this case the calculations show that the antenna circuit must be adjusted to the wave length to be received; namely, 472 meters in the numerical example under consideration. I have attempted to indicate this fact in the diagram by drawing a line across the wave-length scale at 472 meters and marking it 64,000 ohms. RESONANCE OF RECEIVING CIRCUITS 293 Case I (Continued). R 3 = io Ohms, R* = 10,000 Ohms. - Suppose, now, that the detector should have 10,000 ohms resistance instead of 64,000 ohms. With this reduced resistance the curve marked " fl 4 = 10,000 " is obtained. With this value of R 4 , tun- ing by the condenser (\ is possible, but the resonance is dull as is indicated by the obtuseness of the curve. Appropriate adjustment of the antenna in this case is at the line marked "10,000" on the bottom margin; namely, X 3 = 470 meters. Case I (Continued). RS = 10 Ohms, R^ = 1000 Ohms. The curve marked " # 4 = 1000 " is obtained; and the antenna must be shifted to the line on the bottom margin marked "1000"; that is, the antenna wave length must be set at 460 meters for best resonance. The resonance curve " -R 4 = 1000 " is much sharper than those obtainable with the higher resistances. Case I (Continued). RS = 10 Ohms, Ri = 100 Ohms. Reference is made to the curve marked " # 4 = 100," and to the line at the bottom margin marked " 100." The resonance is sharper than with the higher resistances, and the appropriate adjustment of antenna wave length has shifted to X 3 = 430 meters. Case I (Concluded). RS = 10 Ohms, R* = 10 Ohms. Two resonance positions appear in this case: one at 400 meters (wave length of the condenser circuit), with appropriate adjustment of antenna at 360 meters; and the other at 610 meters (condenser circuit), with antenna adjustment at 810 meters. The resonance here is extremely sharp, especially for the adjustment of condenser C 4 in the neighborhood of 400 meters. Case II. Let us now suppose a detector circuit of resistance 10,000 ohms, and let us set the condenser C 4 of this detector circuit at its resonant value in the neighborhood of 135 meters (see the diagram for Case I), and then tune with the antenna circuit; for example, by varying the condenser C 3 . The results are given in Fig. 206, the different curves corresponding to different values of R 3 in the antenna circuit. From these curves it will be seen that even with a high-resistance detector (jR 4 = 10,000 ohms) the tuning in the antenna circuit is sharp, provided the antenna effective resistance is low (curve marked " R 3 = 10 "). With increase of antenna resistance the resonance becomes less sharp. In practice with a system of coupled circuits like that under discussion and with the high-resistance detectors in use, it is difficult to realize sharper resonance than that shown in the curve 294 WIRELESS TELEGRAPHY marked " R z = 50." This is obtained by tuning with the adjust- ment of the antenna circuit. These curves shift almost uniformly with wave length, so that if a number of stations are sending ,8 ti .6 100 200 300 400 500 600 700 A-ntenna Wa.ve Length, in Meters 800 900 FIG. 206. Curves showing effect of resistance R 3 on resonance with a coupled system tuned by adjusting C 3 . Incident Wave-length 100 2CO 300 400 500 600 700 Wave-length in Antenna Circuit in Meters 800 900 FIG. 207. Curves showing the extent 0f interference in a computed case of coupled circuits. simultaneously, the series of resonance curves obtained would be like that of Fig. 207. These curves are computed with what seems to be about the conditions obtaining in good practice. It is seen by a reference RESONANCE OF RECEIVING CIRCUITS 295 to Fig. 207 that if a receiving station is attuned for a 500-meter wave, it will receive also about 7% as much energy from a 400- meter or a 600-meter wave as it does from the 500-meter wave. From a station emitting a 300-meter or a 700-meter wave the disturbing energy will amount to about 2% of the energy received from the 500-meter wave; while from a sending station emitting a 200-meter or a 800-meter wave the disturbing energy will be below 1%. These statements are on the assumption that all of the stations would give the same received energy if the receiving station were in tune for them. These computations, although not claiming to be highly accu- rate, will give a crude idea of about the extent to which inter- ference can be prevented by the use of the coupled circuits consisting of a condenser circuit containing the receiving instru- ment inductively or directly coupled to an antenna circuit. There are other methods of coupling receiving circuits to pre- vent interference which will attain better discrimination between desired and undesired signals, but these almost always greatly reduce the intensity of signals, and cannot be employed for the reception of signals from stations at a great distance from the receiving station. CHAPTER XXV DIRECTED WIRELESS TELEGRAPHY FOR some purposes it is desirable to send electric waves away in all horizontal directions from the sending station, and to receive electric waves coming in from any direction. This is the general mode of propagation of the electric waves, and permits, for example, the establishment of communication with a vessel in an unknown location at sea. Such a general diffusion of waves is, on the other hand, often very undesirable for the following reasons: (1) It is wasteful of transmitted energy; (2) the message may be received by an enemy or an unfriendly neighbor who could generally be prevented from receiving it if we could direct the waves; (3) when we wish to communicate in one direction we may unnecessarily disturb or be disturbed by an operating station in another direction; (4) if the receiving apparatus can be made to respond selectively to electric waves from different directions, a vessel at sea can get its bearings and position by finding its direction from two different known stations. For these and other reasons, several inventors have given attention to the problem of emitting or receiving electric waves directively and have made some progress toward a solution. Hertz's Parabolic Metallic Reflectors. As was pointed out in Chapter XII, Marconi in his early experiments tried to use para- bolic metallic reflectors about his oscillator and receiver, for the purpose of transmitting or receiving in a given direction. On account of the difficulty of constructing and sustaining mirrors sufficiently large to have proper proportions to the wave lengths required, this device has not been successfully used in practice. Braun's Parabolic Oscillator. In 1902, Ferdinand Braun 1 proposed the use of an oscillator consisting of several elements which were arranged to compose a parabolic surface. A diagram of this form of oscillator is shown in Fig. 208. Several vertical metallic strips A\, A%, A 3 . . . were arranged to lie in a para- 1 U. S. Patent, No. 744,897, filed Feb. 19, 1902, issued Nov. 24, 1903. 296 DIRECTED WIRELESS TELEGRAPHY 297 bolic cylindrical surface and were connected to a spark terminal Si. Another similar set of strips BI, B 2 , B 3 . . . below the first set were also provided with a spark terminal S 2 . The oscillations are produced by a discharge across the spark gap SiS*. This arrange- ment, which, according to the inventor, would send out electric waves in one direction, does not seem to have met with practical success. Braun's Phase-difference Oscillator. Another method pro- posed by Ferdinand Braun 1 makes use of two or more vertical oscillators at certain distances apart provided with means of FIG. 208. Braun's parabolic oscillator. FIG. 209. Braun's phase-difference oscilla- tor for directed wireless telegraphy. exciting in the oscillators waves suitably differing in phase. For example, if the two antennae A and B, Fig. 209, are one half wave length apart, and if the oscillations in the two antennae are opposite in phase, the two sets of waves sent out will add in directions in the plane of the two antennae and will neutralize each other in a direction at right angles to this plane. Suitable phase difference in the antennae may be partially attained by the use of a condenser circuit coupled with the an- tennae, as shown in Fig. 209. With this arrangement the problem is, however, complicated by the occurrence of oscillations of double periodicity. This difficulty has been removed in a very U. S. Patent, No. 776,380, filed July 26, 1904, issued Nov. 29, 1904. 298 WIRELESS TELEGRAPHY interesting method of excitation devised, at Professor Braun's suggestion, by Messrs. Mandelstam and Papalexi, and is described in Physikalische Zeitschrift, Vol. 7, p. 302, 1906, to which the reader is referred. With three or more antennae suitably diffejing in their phase of excitation and situated at the vertices of a triangle or of a polygon, any one of several directions may be selected as the direction of strongest transmission. In a similar way, by employing receiving stations provided with a multiplicity of antennas separated by suitable fractions of a wave length, and by using proper means of combining the impulses in a secondary detector circuit, some selec- tivity of direction from which the waves are received can be attained. Marconi's Directive Antenna. In 1906 Mr. Marconi pre- sented to the Royal Society an account of some experiments which showed that an antenna having A *O a short vertical part and then extending away to a considerable distance in a horizontal direction, as shown in Fig. 210, emitted electric waves most strongly in the direction D away from which the free end of the antenna points. Marconi's experiments showed for a given distance between the receiving station and the transmitting station the relative intensities in different directions which, plotted in polar coordi- nates, give a curve of the form of Fig. 211. In this figure the relative intensities in different directions are the lengths of the FIG ' MlrcSrf dTe^ Jd n ltn y na b Ut dii drawn from the origin to the curve. In like manner a receiving antenna consisting of a short verti- cal part and a long horizontal part receives more strongly waves arriving from the direction away from which the open end of the antenna points. Mr. Marconi has utilized this principle in the construction of his powerful stations at Wellfleet and at Poldu. "9- YE W/////M/////////^^ FIG. 210. Marconi's directive antenna. 270 300 120 60 DIRECTED WIRELESS TELEGRAPHY 299 Explanation of Directive Radiation from Marconi's Bent Antenna. Professor Fleming, 1 Dr. Uller, 2 Dr. Zenneck, 3 and others, have given explanations of the cause of the directive radia- tion from the Marconi horizontal antenna. All of these writers employ the theory of images as a starting point, by which means the antenna and ground connection of Fig. 210 is replaceable by the equivalent system of Fig. 212. Fleming's Explanation. In further explanation, Professor Fleming takes a rectangular circuit of the form shown in Fig. 213, and imagines a current flowing around the rectangle in the direc- A' B' FIG. 212. Marconi directed antenna and its image. B h | Out h H oo- In Out FIG. 213. Diagram used by Professor Fleming in explanation of the directive action of the Marconi bent antenna. tion of the arrows. This current creates a magnetic field, the direction of which along the surface of the earth is at right angles to the plane of the paper; and at equal distances from the center, the magnetic force represented by H is toward the spectator on both sides. Now, suppose a wire EF equal in length to one side of the rectangle to be placed contiguous to one vertical side, and to carry a current opposite in direction to that in the side of the rectangle (left hand) to which it is in proximity; then the magnetic field of this straight current is h' from the spectator on the left- hand and h toward the spectator on the right-hand side. Accord- ingly, the total field H + h on the right is greater than the total field H h' on the left, because, according to Professor Fleming, the individual fields are added on one side and subtracted on the other. Now, since the two oppositely directed currents in the 1 J. Fleming: Phil. Mag., Vol. 12, p. 588-604, 1906. 2 Carl Uller: Phys. Zeitsch., Vol. 8, p. 193, 1907. 3 J. Zenneck: Phys. Zeitsch., Vol. 9, p. 553, 1908. 300 WIRELESS TELEGRAPHY adjacent wires may be imagined to come so close together as to annul each other, the effect is the same as if a circuit of parts A BCD were used with the parts AD and EF omitted. Objections to Professor Fleming's Explanation. It appears that serious objection can be raised to Professor Fleming's expla- nation as follows: He does not take into account the mode of vibration of the oscillator, nor does he take account of the time required for the magnetic field to travel from the radiating system to the point under consideration. In the case of the field HH produced by the closed rectangular circuit, the time to travel to the right and to the left to the points under examination will be the same, and the two H's will be equal and in the same direction, as Professor Fleming explains, only provided the same current flows in every part of the loop. No such uniform flow of current occurs in the case of the actual oscillation. Also, in the case of the forces h and In! the distances from EF are unequal, and therefore the times to travel to the points under examination are not the same, and whether the fields h and h' will be opposite to each other or not depends on the mode of vibration of the two oscillators and the time for the waves to travel to the points under examination. The whole question of the relative strength of waves emitted in the two opposite directions is avoided by Professor Fleming because of his substitution of a system that can never represent the actual system; and after we have examined Professor Fleming's reasoning the solution of the actual problem is still completely in doubt. Explanation of Dr. Uller. Professor Fleming l had earlier employed a different method of attacking the problem directly by imagining the form of the electric field of force about the direc- tive antenna. This method was revived in 1907 by Dr. Uller, who pictured the field of electric force about the Marconi oscillator in the form given in Fig. 214. The upper half of this diagram would represent the mode of propagation of the waves over the surface of a good conducting plane. Where the surface of the earth is not a good conductor, the electric force would be inclined near the surface, as has been shown in Chapter XV, and would give a field of force slightly different from that here represented. Zenneck's Explanation. Zenneck has modified the theory of Uller so as to take further into account the effect of im- perfect conductivity of the earth's surface. As we have seen in 1 Fleming: Electric Wave Telegraphy, p. 627, 1906. DIRECTED WIRELESS TELEGRAPHY 301 Chapter XV, the electric force at the surface of the earth, where - ever it is not a good conductor, leans forward, so that we can ascribe to the electric force in a particular case a mean direction, FIG. 214. Dr. Uller's diagram of field of electric force about the bent antenna. E, Fig. 215. Now the direction of propagation is perpendicular to E; i.e., in the direction S, whence there is penetration of the energy into the earth's surface and a consequent absorption, so FIG. 215. Diagram used by Dr. Zen- neck in explaining directed wireless telegraphy. Q FIG. 216. Zenneck's diagram show- ing the course of the radiation from A to R. FIG. 217. Diagram applying to Zenneck's explanation. that the distant receiving station is reached by the energy that started in the direction AX, Fig. 216, and not by the energy that started along the surface of the earth. By examination of Fig. 302 WIRELESS TELEGRAPHY 214 it will be seen that a bent antenna of the form of ABC, Fig. 217, radiates more energy in the direction CP than in the direc- tion CQ, and therefore attains a greater distance in the direction CD than in the direction CE. This explanation of Zenneck would indicate that the directive effect of the bent antenna is much greater over land than over sea. I do not know of any experimental confirmation of this deduction. These explanations are lacking in quantitativeness, but taken together serve to give a tentative reconciliation of some of the experiments with theory. Bellini and TosPs Directive Apparatus. A very ingenious method of directively transmitting and receiving electric-wave signals has been devised by Messrs. Bellini and Tosi. 1 A diagram FIG. 218. Bellini and Tosi's directive apparatus. of a receiving station embodying their invention is shown in Fig. 218. The directive aerial system consists of two closed or nearly closed oscillation circuits of triangular shape ABBA and A 1 B 1 B 1 A 1 arranged respectively in two perpendicular planes. These two antenna circuits contain respectively the coils m and n, which may be circular coils, and are perpendicular to each other with their windings in the planes of the antenna circuits respectively. A third coil s connected to a wave detector and a condenser c is 1 U. S. Patent, No. 945,440, filed Oct. 1, 1907, issued Jan. 4, 1910. DIRECTED WIRELESS TELEGRAPHY 303 placed within the two coils m and n and is capable of rotation about an axis through o. Electric waves coming from any particular direction produce oscillation in the two antenna circuits with intensities respectively dependent on the direction from which the waves come. The oscillations thus set up, passing through the coils m and n, com- pound to form a single magnetic field with a direction perpendicular to that from which the waves come. The strength of the induced current in the movable coil s will depend on its orientation with respect to the resultant magnetic field, and will be a maximum when the coil s is in a position to embrace as many as possible of the lines of magnetic force. This optimum direction is perpen- dicular to the field, and therefore parallel to the direction from which the waves are coming. It is therefore possible to determine the direction from which the waves are arriving by merely providing the rotating coil s with a pointer in its own plane. When a maximum strength of signals is received the pointer is directed either toward or away from the signaling station. The final ambiguity as to whether the signaling station is in the direction of the pointer or in the opposite direction would have to be removed by some additional general knowledge of the probable location. A sending station, devised also by Bellini and Tosi, and capable of directively transmitting signals, consists of a similar aerial system and a similarly rotatable interior coil. t The latter is, how- ever, connected with a discharge condenser instead of with the receiving mechanism. The processes involved are, then, the reverse of those entering into the receiving apparatus. Limitations of Directive Wireless Telegraphy. The several directive devices above described act directively only in a general way; that is, some more energy is sent in one direction than in other directions, but there is still a considerable diffusion of energy in all directions. The economy effected in the energy of transmission does not seem to be very great, particularly because the closed loops, or nearly closed loops, are not such good radiators or receiv- ers as the straight vertical antenna. However, whenever the bent antenna is installed in land stations the orientation to effect maxi- mum transmission in the most useful direction is generally chosen. Also, it has been proved to be entirely possible with each of the principal systems to determine the direction of the receiving station from the sending station. This achievement does not seem to have 304 WIRELESS TELEGRAPHY been of sufficient importance up to the present to warrant special installations for the purpose. It is, however, entirely possible that greater attention will be given to this subject when the art of wireless telegraphy, which is now embarrassed by novelty in so many directions, shall have become a little more standardized in its fundamental requirements. CHAPTER XXVI WIRELESS TELEPHONY Sketch of the Method of Wireless Telephony by Electric Waves. - The circuits employed in wireless telephony by electric waves resemble very closely those used in wireless telegraphy. The transmitting apparatus for wireless telephony makes use of a persistent train of electric waves of high frequency sent out from an antenna. Instead of interrupting these electric waves by a key, as in telegraphy, modifications by the voice, correspond- ing to spoken words, are impressed upon them. These modifica- tions by the voice are applied to the electric waves by means of a carbon transmitter, or similar instrument, placed in the sending circuit or connected with it. The receiving apparatus is indentical with that employed in wireless telegraphy, and makes use of a receiving antenna coupled with a circuit containing some type of rectifying detector; e.g., an electrolytic detector, a crystal-contact detector, or a vacuum-tube rectifier. About the detector is shunted a sensitive telephone receiver. The action is as follows : If an unmodified train of electric waves having a frequency higher than the limit of human audibility (35,000 vibrations per second) arrives at the receiving station, the receiving circuit, if properly tuned, will sustain electric oscillations which, passing through the detector, will be rectified and will give a series of rectified impulses to the receiving telephone circuit. These impulses, being all in one direction, will act as a continuous pull on the telephone diaphragm, a continuous pull for the reason that the diaphragm cannot follow the rapid successive impulses, and because also, on account of the inductance of the telephone circuit, these impulses are modified electrically into a practically continuous current through the receiver. Having in mind that a continuous train of high-frequency waves produces a continuous pull on the receiving telephone diaphragm, let us now suppose that words are spoken into a carbon transmitter at the sending station in such a manner as to modify the emitted 305 306 WIRELESS TELEGRAPHY train of waves. These modifications of the emitted waves will produce corresponding modifications in the pull on the telephone diaphragm at the receiving station, so that the receiving dia- phragm will execute vibrations similar to those of the transmitting diaphragm, as in ordinary telephony over wires. Methods of Producing the Persistent Train of Waves. Some of the details of the process will now be presented. To produce the persistent train of oscillations several methods are available, of which three will be mentioned, to wit: 1. The Singing-arc Method; 2. The High-frequency Alternator Method; 3. The Mer- cury-arc Method. The first of these methods has been described in the preceding chapter. A brief description of the other two methods follows. The High-frequency Alternator Method of Producing Sus- tained Oscillations. In 1901, Professor R. A. Fessenden 1 applied for a patent for " improvements in apparatus for the wireless trans- mission of electromagnetic waves, said improvements relating more especially to the transmission and reproduction of words or other audible signals. " A diagram of the simple apparatus described in this application is shown as Fig. 219. In the diagram, which represents the transmitting station, D is an alternating-current generator of high periodicity; for example, 50,000 per second. A carbon transmitter is shown at T. The diaphragm of the transmitter is marked P. A is the sending antenna. One of Professor Fessenden's claims is as follows: " In a system of signaling by electromagnetic waves, the com- bination of means for the practically continuous generation of electromagnetic waves or impulses, means for modifying or chang- ing the character of such waves or impulses without interruption of their continuity, and an indicating means or mechanism at the receiving station operative by the electromagnetic waves or im- pulses, substantially as set forth." In carrying out the invention Professor Fessenden, in 1908, con- structed a high-frequency alternator, with an output of 2.5 kilo- watts at 225 volts, and with a frequency of 75,000 cycles per second. This is a frequency well above the limit of audibility, and in fact a frequency sufficiently high to give, when the generator is connected directly or inductively to an antenna in resonance 1 U. S. Patent, No. 706,747, applied for Sept. 28, 1901, divided July 22, 1902, issued August 12, 1902. WIRELESS TELEPHONY 307 with it, a wave length suitable for wireless telephony, namely, 3 X 10 8 /75,000 = 4000 meters. With this apparatus, Professor Fessenden reports that he has carried on telephonic communica- tion between Brant Rock, Massachusetts, using an antenna 440 feet high, and New York, using an antenna 200 feet high. The distance between these two stations is about 200 miles. Recently Professor Fessenden also reports successful wireless telephonic communication between Brant Rock, Massachusetts, and Wash- ington, D. C., a distance of about 600 miles. T P FIG. 219. Professor Fessenden's ap- paratus for wireless telephony ^ using high-frequency generator D and a microphone transmitter T. FIG. 220. Diagram of Vreeland's mercury-arc oscillator. The Mercury-arc Method of Producing Sustained Oscillations. In 1906 Mr. Frederick Vreeland 1 described a very interesting method of getting practically pure sinusoidal undamped oscilla- tions from a direct-current supply. One form of Mr. Vreeland's apparatus is shown in Fig. 220. T 7 is a glass vessel, exhausted to a high vacuum, and containing a mercury cathode K and two carbon anodes A and B. E is a small auxiliary electrode used in starting an arc in the chamber. The arc, when established, being fed from the direct-current source D, is divided into two branches Physical Review, Vol. 27, p. 286, 1908. BOS WIRELESS TELEGRAPHY one between the anode A and the cathode K, the other between the anode B and the cathode K. A resistance R and two choke coils L and L' serve to steady and maintain the two arcs. Now an oscillation circuit consisting of a condenser C and two coils F and F' is connected between the two anodes A and B. The coils F and F' in the oscillation circuit serve as field coils to deflect the arc inside the vacuum bulb, and to cause the cathode stream of this arc to oscillate in a plane perpendicular to the axis of the coils in such a manner that this oscillating cathode stream impinges first on one and then on the other of the anodes A and B. The manner in which this deflection of the cathode beam is produced is as follows: At the start, the current tends to divide equally between the two arcs in the bulb, but there are always some variable inequalities in the conductivities of the two paths. These irregular fluctua- tions are usually sufficient to start the oscillations, after which they give place to the periodic fluctuations controlled by the alternating field. The action of the magnetic field is such as to produce a deflection of the cathode beam, and when this beam is deflected, say from the anode B to the anode A, there is a tendency for the current to pass wholly or largely from the anode A to the cathode K, due to the fact that the path from B is interrupted or increased in resistance. As the choke coils L and L' oppose any change in the current passing through them, this results in the current in the branch L' flowing through the oscillating circuit F'CF from right to left, thus traversing the field coils in such direction as to increase the deflection of the cathode beam toward the left, thereby augmenting still further the inequality of the two paths through the tube and increasing the current through the oscillating circuit. This continues until the condenser C charges to a certain point, when it begins to discharge, reversing the field, and causing the arc to be deflected in the other direction, so as to force the current through the oscillating circuit from A to B. This process, being repeated indefinitely, results in feeding the energy into the oscillating circuit in synchronism with the oscillations, which are thus maintained at constant amplitude and at a frequency determined by the self-inductance and capacity of the circuit. A photograph of the completed apparatus is shown in Fig. 221. I am not able to say whether or not Mr. Vreeland has, up to the present, been able to get the frequency of his oscillation producer WIRELESS TELEPHONY 309 up to the pitch required for wireless telephony. His apparatus is, however, very ingenious and full of promise. Method of Applying the Microphone to Modify the Oscilla- tions. Having described methods of producing sustained or FIG. 221. View of Mr. Vreeland's apparatus. persistent oscillations I wish next to show briefly diagrams of con- nections by which the carbon microphone may be applied to modify these oscillations in accordance with the vibrations of the voice. In most of these diagrams I have represented the source of the persistent oscillations as a singing arc, such as has been devised by Simon, Duddell, and Poulsen. It will easily be seen how these 310 WIRELESS TELEGRAPHY diagrams should be modified to permit of the use of Fessenden's high-frequency generator or Vreeland's mercury-arc oscillator. Figure 222 shows the microphone in series with the direct-current dynamo of the feeding system. To be used in this manner, the microphone must have high current-carrying capacity, and for FIG. 222. One method of applying the FIG. 223. Showing the microphone microphone M to wireless telephone M in a circuit inductively coupled circuit. with the feeding circuit of an arc used in wireless telephony. FIG. 224. Microphone M be- tween secondary helix and ground. FIG. 225. Microphone T inductively cou- pled with secondary helix for wireless telephony. this purpose some inventors have proposed to use several micro- phones in parallel, all of the diaphragms facing upon a common air chamber into which the words are spoken. Figure 223 shows the microphone transmitter connected in cir- cuit with the primary of a transformer S, the secondary of which, L, is in series with the dynamo and the arc. In this case the heavy current of the arc does not go through the microphone. In common with the method of Fig. 222 there is the disadvantage WIRELESS TELEPHONY 311 that the microphonic modifications of current have to traverse the generator circuit, and hence meet with high impedance. Figure 224 shows the microphone connected in series with the antenna circuit, between the secondary of the oscillation trans- former PS and the ground connection. Figure 225 shows a method proposed by Mr. Vreeland and others in which the microphone circuit Is inductively connected with the secondary S of the oscillation transformer. Other methods of connecting the microphonic transmitter to the oscillating circuit are also employed. Practical Results in Wireless Telephony. I have briefly pointed out in the preceding paragraphs the general processes employed in wireless telephony. The small amount of space here devoted to the subject is not to be taken as evidence that wireless telephony is a simple or unimportant branch of the science of electric-wave transmission of intelligence. To be able to modulate a train of electric waves by waves of sound existent in the air between the mouth of the speaker and a transmitting diaphragm, and to be able to receive these modulated electric waves at a distance and reconvert them into sound waves, is a very remarkable achievement of scientific ingenuity, even when the sending and receiving stations are close together. Wire- less telephony has, however, gone far beyond this stage; and Fessenclen in America, Poulsen in Denmark, Majorano in Italy, and Messrs. Colin, Jeance and Mercier in France, have severally reported successful wireless telephonic transmission of speech to distances ranging from 40 to 600 miles. Even if these experi- ments have been lacking in some details of perfection, we cannot doubt that practical wireless telephony, especially between ships at sea at a considerable distance apart, is a possibility of the present time or of the immediate future. CHAPTER XXVII SOME DETAILS OF CONSTRUCTION OF WIRELESS TELEGRAPHIC APPARATUS IT is beyond the scope of an elementary treatise to erxter exten- sively into a discussion of the engineering details of a wireless telegraph installation. In fact, much of the wireless telegraphic engineering of the present time is done by methods of construction and trial rather than by scientific prognosis. There are, however, certain elementary facts that may be of service to amateurs engaged in constructing or operating wireless telegraphic appara- tus, and that at the same time may be not without interest for the I Straight A Flat-top Fan-shaped FIG. 226. Types of antenna. Umbrella general reader. Some of these elementary facts regarding con- struction are here presented. Antenna. The character of the equipment that may be employed in a given instance depends on the facilities that exist for the erection of an antenna. A few simple types of antenna are represented in the diagrams of Fig. 226. The Flat-topped Antenna. Of these types, the flat-topped antenna usually gives the best results for a small installation. The flat-topped antenna consists of the nearly vertical portion A B and the nearly horizontal portion BC. The horizontal portion does not contribute much as a useful radiating member, because waves emitted from this portion have their electric force parallel to the earth's surface, so that the part of this radiation that 312 CONSTRUCTION OF WIRELESS APPARATUS 313 travels out along the surface of the earth induces currents in the earth and is rapidly absorbed. The remainder of the energy radiated from this horizontal portion travels prevalently upward and, save for contributing to the directiveness of transmission as has been pointed out in Chapter XXV, does not have much effect at the receiving station unless it is desired to transmit to a balloon, when this upward-traveling component is most useful. The horizontal portion of the flat-topped antenna is, therefore, chiefly serviceable as a capacity at the top of the vertical part, which latter is the chief radiating member. As to the amount of the capacity it is interesting to note that a single wire 100 feet long and J inch in diameter when alone in space has as much capacity as an isolated flat metallic disc 16 feet in diameter. (See formulas for calculation in Appendix II.) From this it will be seen that the horizontal top to the antenna is a far more economical elevated capacity than any kind of a metallic sheet such as was employed in Marconi's early experiments. Comparison of Flat-topped with Straight Antenna. In order to illustrate some of the principles involved, let us next compare the radiation from a single vertical wire 100 feet long and say J inch in diameter with that from a flat-topped antenna consisting of a vertical wire 100 feet long having at the top a horizontal extension of the same length. For the purpose of this comparison we shall employ the experimental curve of current distribution found in Chapter XIV (Fig. 82). In the first place the flat-topped antenna, because of its greater length of wire, has approximately twice as much capacity as the simple vertical antenna. This means that if we charge the two antennae to the same potential, about twice as much electricity will flow during one oscillation of the flat-topped antenna as during one oscillation of the simple vertical antenna; but the time of the oscillation in the former case will be about twice as long; therefore the maximum current flowing to the ground will be about the same in the two cases. Let us now plot the approximate current-distribution curves for the two cases, assuming the same current at the base; and in doing this we shall make the further assumption that the distribution in the bent antenna is approximately the same as it would be for a straight antenna of the same length. The curves obtained are given in Fig. 227. In these curves the value of the current at any point of the length of the antenna is plotted as a distance between 314 WIRELESS TELEGRAPHY the antenna and the curve. Careful plotting and measurement of these curves show that the average current in the vertical portion of the flat-topped antenna is .88 of the maximum current at the base; whereas the average current for the vertical antenna is only .62 of the current at the base. Dividing .88 by .62, we find that the average current in the vertical portion of the flat-topped antenna is 1.41 times the average for the simple vertical antenna. From these considerations it appears that we have gained 41% in effective current by the use of the flat-topped extension. We could gain approximately the same by extending the simple antenna about 41 feet upwards. From this we may conclude that two poles of 100 feet in height and 100 feet apart supporting a FIG. 227. Comparison of current distribution on a flat-topped antenna (left) with that on a straight antenna (right) of the same height. flat-topped antenna would give approximately the same service as a single pole 141 feet high supporting a single vertical antenna. On account of difference in damping and on account of the effects on radiation introduced by the difference in wave length in the two cases, and also on account of the directive emission from the flat-topped antenna, the problem is not so simple as is here repre- sented; and the numerical deductions are indeed only very rough approximations, which serve merely to show wherein consists the efficacy of the flat-topped antenna; namely, in the increased average current in the vertical part due to the capacity of the horizontal part. Antenna of Several Wires, Instead of employing a single wire in the antenna, as in the illustrative example here given, several wires are usually employed. It should be noted, however, that n wires placed side by side have not anything like n times the capacity of a single wire; because the charge on one wire repels the charge on the other wires, and therefore the charge that the system will take under a given electromotive force applied at the CONSTRUCTION OF WIRELESS APPARATUS 315 base is not multiplied in the ratio that the number of wires is multiplied. For an economical installation from four to six wires may well be employed in the antenna, and by the use of light bamboo spreaders they can easily be supported three feet or more apart. Marconi Antenna at Clifden. An example of the use of the flat-topped antenna on a large scale is afforded by the Marconi high-power station at Clifden, Ireland. The horizontal part of the antenna of this station consists of 200 wires 1000 feet long supported 180 feet above the earth The wave length is about 4000 meters. The Umbrella Antenna. When only one supporting pole is available, either the straight type or the umbrella type of antenna FIG. 228. Umbrella type of antenna. is usually employed. The umbrella type meets with frequent use in small amateur stations and in the portable stations employed by armies. In this type the aerial system consists of a vertical portion terminating above in a system of wires inclining downward. These inclining wires are usually the guy wires, while the vertical part may be either a wire leading to the top of the pole, or the pole may itself be of metal and serve as the vertical conductor. A diagram of an umbrella type of antenna with a metallic pole serving as the vertical conductor is shown in Fig. 228. The 316 WIRELESS TELEGRAPHY metallic pole used in a small installation may be two or three sections of ordinary tinned gutter pipe or of one-inch iron water pipe. To keep such an antenna straight, a separate set of guy wires must be used for every section of pipe employed. The bottom of the metallic pole is supported on an insulating base B, which is protected from rain by a shelter S placed -above it and connected to the pole. The wire W leading from the operating room is connected directly to the pole near its base. The lower guy wires AAAA are preferably insulated from the pole and from the housetop. The upper parts of the upper guys CCCC are con- nected to the top of the pole, and these serve as a capacity exten- sion to the antenna. At a suitable distance from the top of the pole high-tension insulators are inserted so as to terminate the antenna. With this form of antenna it will be observed that the oscillation in the vertical pole and that in the inclined extensions CCCC are partially opposite to each other, and therefore partially neutralize each other with respect to radiation. The length of the guy-wire extension that can thus be used with advantage will depend upon the number of the guys and their inclination. The Fessenden Tower at Brant Rock. A very striking example of a station making use of the supporting structure as antenna is the powerful station of the National Electric Signaling Company at Brant Rock, Massachusetts. For the antenna of this station there is provided a cylindrical steel tower 440 feet high, carefully insulated at the base and provided above with extension capacity in the form of four horizontal arms each 80 feet long. These arms being horizontal do not offer the disadvantage of partially neutralizing the radiation from the tower. When it is remembered that this very tall and very heavy steel tube must be sufficiently insulated from the earth to withstand the enormous potential developed in a very high-power wireless telegraph sending station, it will be seen that the design and erection of such a plant, which was accomplished by Professor Fessenden, is a very considerable feat of mechanical and electrical engineering. It is interesting to compare the capacity of this large tube with that of a small wire. With the aid of Formula VII of Appendix II, it can be shown by calculation that a tube 440 feet (13,510 cm.) high and 3 feet in diameter (46 cm. in radius) has a capacity, when alone in space, that is only about twice as great as the capac- CONSTRUCTION OF WIRELESS APPARATUS 317 ity of a wire the same length and f of an inch (1 cm.) in diameter. Therefore, so far as concerns capacity, a few small wires five or six feet apart would be the equivalent of this large steel tube. The Ground. The theory of the action of the ground has been discussed in Chapter XIV. In practice, for a small station a satisfactory ground can be obtained by a connection to the pipes of a water supply. Where this is lacking, a good arrangement is to bury a netting or network of wires at a short depth below the surface of the earth. This may be supplemented by metallic pipes driven to considerable depths into the earth, and also by wire netting spread out on the surface of the earth. When the station is located near the sea or other body of water, the wire netting or wires provided with terminal plates may be led into the body of water. On board ship, the grounding is usually effected by a heavy wire attached to the metallic hull of the ship. In the high- power land stations, netting and wires are made to ramify the surface of the earth for many acres. We have seen in Chapter XIV that a properly resonant artifi- cial conductor supported without contact with the earth serves as a very good ground. The difficulty about the artificial ground is the fact that the artificial ground should be tuned along with the aerial system in order to get resonance with different wave lengths. Sending Condensers for a Coupled Transmitting Station. - The details of construction of the simple Marconi apparatus of 1896 need not be given. When a sending station of the inductively coupled or direct coupled type is to be employed, the sending con- densers must be electrically strong in order to permit the storage of the large quantities of electricity used in producing the waves. Among the types of condenser employed for this purpose the bank of Leyden jars or of flat glass plates provided with metallic coat- ings are most familiar. The use of tinfoil, for the coating of Leyden jars or flat-plate condensers for use in wireless telegraphy, has been largely discontinued. In the case of the flat-plate con- densers copper or brass sheets between the plates in the place of the tinfoil that was formerly much used gives a much smaller loss of energy, and consequently much smaller heating of the con- denser. Ordinary window glass, when selected free from flaws, is electrically stronger than plate glass for making glass-plate con- densers. When high power is to be used, the flat-plate condensers should be submerged in castor oil to prevent brush discharge. In the case of the Leyden jars, when used in stations of large 318 WIRELESS TELEGRAPHY power, glass of especially high electric breaking strength is em- ployed and the tinfoil of other days is now usually replaced by a coating of silver or copper electrolytically deposited on the inner and outer surface of the jars. A photograph, lent me by Mr. Pickard, of some jars coafced m this way by the Wireless Specialty Apparatus Company of New York, is shown in Fig. 229. Air Condensers, formed of metallic plates with air between as dielectric, are said to be employed as sending condensers in Mr. Marconi's high-power stations at Poldu, Clifden and Wellfleet. Condensers employing compressed air as dielectric have been employed by Mr. Fessenden in his Brant Rock station and in FIG. 229. Copper-plated Ley den jars. some of the ship installations supplied by The National Electric Signaling Company to the United States Navy. The dielectric constant of the compressed air is about the same as that of air at atmospheric pressure. The purpose in compressing the air is to increase its disruptive strength so as to enable the condenser to stand higher potentials. The disruptive strength is approxi- mately proportional to the gas pressure. Amount of Capacity to be Used at a Given Station. The amount of capacity to be used at a given coupled-type of sending station depends upon, (1) the amount of power to be supplied to the condenser; (2) the number of sparks per second, and (3) the CONSTRUCTION OF WIRELESS APPARATUS 319 voltage at which the discharge occurs. As a specific example, let us suppose that the power is to be supplied by an alternating current source of n cycles per second. By means of a transformer with its primary connected to the source of power and its secondary attached to the condenser, we may step up the potential to the value required to produce the required spark. Let us suppose the transformer to supply P kilowatts of power to the condenser, and let us choose the condenser and the spark gap to be such that the condenser charges to a sparking potential only once during each half-cycle ; that is, 2 n times per second. Now to charge a condenser once to a potential of V volts requires an amount of energy, W = \QV joules, (1) where Q is the number of coulombs of electricity required and | V is the average potential during the charge. (See Appendix I.) And, from the definition of capacity, Q = CV, (2) where C is the capacity of the condenser in farads. Substituting the value of Q from equation (2) in equation (1), we have W = J CV 2 joules, (3) V being the potential in volts to which the condenser is charged. In our supposed case the condenser is charged 2 n times per second; therefore the energy expended per second, which is the power supplied, is W= 2 n X \ CV 2 = nCV 2 joules per second. (4) But 1 joule per second is 1 watt, and 1000 watts make a kilowatt; therefore if P is the power in kilowatts, P. ^kilowatts. (5) In interpreting this formula, it must be remembered that V is the potential to which the condenser is charged at the time that the spark begins. The formula (5) is very useful in practical computations. By a simple transposition of terms, equation (5) may be put in the form 1000 X Power in Kilowatts C = - T79 Co; nV 2 320 WIRELESS TELEGRAPHY From this we can calculate the capacity required in a given case, provided we know the power to be employed, the number of cycles, and the voltage to which the condenser is to be charged. For a given source of power we can employ either a large condenser charged to a low potential or a smaller capacity charged to a higher potential. A simple computation, which is not here given, shows that approximately the same volume of dielectric (e.g., glass) will have to be used in the condenser in either case. In estimating the amount of capacity to be employed to con- sume a given amount of power, according to formula (6), it is well to estimate about 15,000 volts to the centimeter of spark length; for this is about the value of the potential when the spark gap is heated and ionized by continuous sending. On the other hand, in estimating the amount of dielectric to use for sufficient strength to stand the charge without breaking, it is well to estimate about 39,000 volts to the centimeter; for the voltage will rise to this value when the station is first started up. The Charging Transformer. After the dimensions and capac- ity of the condenser for the sending station have been settled upon, the transformer must be designed to be in resonance with the condenser. The proper proportioning of the primary and second- ary inductance and the mutual inductance of the charging trans- former of the sending station is one of the most troublesome factors arising in connection with wireless telegraphy design and con- struction, and cannot be adequately discussed in an elementary treatise. The fact to be kept in mind is that the transformer for this purpose must have entirely different properties from those pos- sessed by an ordinary closed iron-core lighting transformer, because the lighting transformer is designed to supply more and more power as the load is made of lower and lower resistance; while with the wireless telegraph transformer the load is a condenser, which will attain a maximum charge for a certain resonant relation of the constants of the transformer to the capacity of the con- denser. A spark will then pass. This spark amounts to a short- circuit of the secondary of the transformer. Under this condition an ordinary closed-core transformer would supply a maximum amount of power right across the short-circuited gap, so that this gap would sustain an arc, and the condenser would then not charge up again. This is not desired. What is desired is, that when the discharge of the condenser occurs and short-circuits the secondary of the transformer, the transformer should be so designed that it CONSTRUCTION OF WIRELESS APPARATUS 321 will draw a very small amount of power, and allow the spark to extinguish promptly after the discharge of the condenser. A mathematical examination of this problem shows that this result can be obtained with a proper adjustable resistance placed in the primary circuit of the transformer, if a common closed-core transformer is used. The same result can be more economically obtained by the use of an adjustable inductance in series with the D i Ground Secondary Condenser a Sending Helix r^ I Cut-over r i i i i^ ,* i i i i o Jl C \ Switch r ==i3 t-^ _JSa d = the diameter of the wire in centimeters, n = the number of complete oscillations per second, p = the specific resistance of the material of the wire in terms of absolute c.g.s. electromagnetic units, TT = 3.1416. . . , and H = the magnetic permeability of the material of the wire, and is unity for nonmagnetic wires. The formula (A) is convenient for computation when k is less than 1. When k is greater than 5 or 6, the following formula, also derived by Lord Rayleigh, is more accurate and convenient for calcu- lation: The succeeding table contains some computed values of resistance R' for 1,000,000 oscillations per second in terms of R, for different diameters of copper and German-silver wire. TABLE FOR RATIO OF -' . H R' = Resistance for.1,000,000 oscillations per second, R = Steady-current resistance. Diameter in Cm. Copper p = 1600. German Silver p = 20,900. .01 1.008 1.000 .02 1.117 1.000 .03 1.32 1.000 .05 1.95 1.000 .1 3.88 1.005 .2 7.85 1.09 .3 11.8 .4 15.7 4.30 .5 19.7 5.38 .6 23.6 6.5 .7 27.5 7.5 .8 31. 8.6 .9 35. 9.7 1.0 39. 10.7 1.5 59. 16. 2.0 79. 21.5 By reference to this table it will be seen that the resistance of a copper wire 2 centimeters in diameter is 79 times as great with the APPENDIX II 339 rapidly oscillating current as it is with a steady current. With decrease in the diameter of the wire the effect of the high frequency in diminishing the resistance decreases, and with a wire of copper 1 millimeter in diameter the resistance for current making one million oscillations per second is only 3.88 times as great as the resistance for steady current. For diameters below one-tenth of a millimeter:, the high-frequency resistance of copper does not differ from the steady-current resistance. For radii greater than .5 millimeter the resistance of a circular copper wire is very nearly inversely proportional to the radius of the wire, while the steady resistance is inversely proportional to the square of the radius. If we pass now from the case of copper to that of German silver, which has a specific resistance about 14 times as great as copper, it is seen that the departure between high-frequency resistance and steady-current resistance is not so great as for copper. For Ger- man-silver wires less than 1 millimeter in diameter the high-fre- quency resistance differs by not more than | of 1% from the steady resistance. Above one millimeter in diameter the ratio of R f to R for German silver increases progressively with increase of diameter. The formulas here given apply only to approximately straight conductors, and should not be used to apply to wires wound into coils. Formulas for Calculation of Capacity. The following formulas serve for the calculation of capacity in some simple cases. The linear dimensions are to be measured in centimeters and k is the dielectric constant of the dielectric between the plates. The dielec- tric constant of air or other gas at ordinary atmospheric pressure is approximately 1. Approximate values of k for some other dielectrics are given in the table on page 341. I. Capacity of a condenser of two parallel flat plates oppositely charged. kA C = c.g.s. electrostatic units, 4-rra in which A is the area of one of the plates overlapped by the other plate, d is the distance of the plates apart in centimeters. This formula holds accurately only when the distance apart of the plates is small in comparison with the length and breadth of the plates. 340 WIRELESS TELEGRAPHY II. Capacity of two concentric cylinders oppositely charged. kl C = - c.g.s. electrostatic units, in which I = the overlapping length of the cylinders, R 2 the radius of the outer cylinder, RI = the radius of the inner cylinder. III. Capacity of a length I of two practically infinite parallel wires of the same radius, the wires being oppositely charged. kl C = -- - c.g.s. electrostatic units, 41og.J d = distance apart of the wires, R = radius of either wire in centimeters. IV. Capacity of two concentric spheres, oppositely charged. 7? 7? C = -r-2 c.g.s. electrostatic units, d in which R 2 = the radius of the outer sphere, RI = the radius of the inner sphere, d = RZ RI. V. Capacity of a single sphere alone in space. C = R c.g.s. electrostatic units, in which R = radius of the sphere. VI. Capacity of a circular disc, or thin plate. 9 7? C = -- c.g.s. electrostatic units, 7T in which R = the radius of the disc. VII. Capacity of a single cylindrical wire alone in space. C = - c.g.s. electrostatic units, in which I = the length of the wire in centimeters, R = its radius. Rule for Several Condensers in Parallel. If several con- densers of capacities Ci, C 2 , C 3 , . . . are connected in parallel, the combined capacity C is C = d + C 2 + C 3 + . APPENDIX II 341 Rule for Several Condensers connected in Series. - =+-+- + . C Ci Cz C$ TABLE OF DIELECTRIC CONSTANTS Substance. Dielectric Constants. Empty space 1 IAir or other gas 1 under atmospherics pressure 1 1 approx Glass 6 to 10 Mica Hard Rubber Kerosene Oil Castor Oil 6.6 2.7 2.0 4.78 Water 80 Formulas for Calculating Inductance. The lengths are to be measured in centimeters, and the results are in c.g.s. electro- magnetic units. The medium surrounding the conductors is supposed to be nonmagnetic. I. The mutual inductance between a long single-layer solenoid and a lumped secondary wound about it. M =.4:TrniNzA c.g.s. electromagnetic units, in which wi = the number of turns per cm. length on primary coil, _/V 2 = the total number of turns on secondary coil, A = the area of cross section included within the primary coil. II. Self-inductance of a single-layer solenoid. 2 a4 + a *V 8 L = 4w 2 n 2 c.g.s., electromagnetic units, in which a = the mean radius of the solenoid, n = the number of turns per cm. of length, I = the length in centimeters. This is accurate to better than } of 1% when I is not less than 4 a. 1 III. Self-inductance of a length I of two practically infinite parallel wires in which the current is flowing in opposite direc- tions (i.e., a return circuit). L = 4 I loge - c.g.s. electromagnetic units, R 1 Cohen, Bulletin of the Bureau of Standards, Vol. 4, p. 385, 1907-08. 342 WIRELESS TELEGRAPHY in which d = the distance between centers of the two wires, R = the radius of each wire, supposed equal, I = the length of the pair of wires. This formula assumes that the current is flowing only on the out- side surfaces of the two wires, as is the case with oscillations of high frequency. (See next formula.) IV. Self-inductance of a return circuit like that of III, with, however, a uniform distribution of current throughout the wires instead of merely on the surfaces. L = 4 1 ] log e - + - [ c.g.s. electromagnetic units, \. ) in which I = the length of the pair of wires, d = the distance between centers of the two wires, R = the radius of the wires, supposed equal, M = the magnetic permeability of the material of the wires, the permeability of the medium between the wires being assumed unity (i.e., nonmagnetic). V. Self-inductance of a length Z of two concentric tubes. D L = 2 \og e 17 c.g.s. electromagnetic units, RI in which R% = the inner radius of the outer tube, #1 = the outer radius of the inner tube. In this case the distribution of current is supposed to be only on the adjacent surfaces of the tubes. VI. Self -inductance of a single wire of length I. ( 27 ) L = 21} log e - 1 > c.g.s. electromagnetic units, ( R } in which I = the length of the wire, R = its radius. In this case the wire is supposed to be of small diameter in com- parison with its length. INDEX Abraham, M., theoretical value of wave length, 116. Absorption of electric waves. By soil, 127, 131; by ionized air, 137. Air. Absorption by, 137; Conduc- tivity of, 137. Air condenser, 318; of Korda, 114. Alternator, high-frequency, 306. Amesbury, Mass., experiments at, 134. Analogy. Of self-inductance to in- ertia, 22, 28; of capacity to me- chanical quantities, 26, 28. Anatase, 177. Antenna. Dependence on height of, 271; resonance with various lengths of, 281; theory of directive, 299; types of, 312, 313. Antenna circuit, determination of wave length of, 246. Apparatus, construction of, 312. Arc. Mercury, 307; singing, 253; talking, 254; pulsating, 255; in steam, 259; period of singing, 260. Armagnat, characteristic of electro- lytic detector, 203. Arons tube, 72. Atlantic cable, 63. Atomic structure of electricity, 8. Attenuation of electric waves, By ab- sorption, 127; by divergence, 129. Attraction, electrostatic, 329. Audibility, limit of, 148. Audion, 214. Austin, L. W. Sensitiveness of tele- phone receiver, 140; detector, 160, 198; electrolytic detector a recti- fier, 203. Balloons, 89. Barretter, 154; liquid, 203. Bell, Graham, telegraphy by conduc- tion through water, 77. Bellini, directive wireless telegra- phy, 302. Bjerkness, waves on wires, 70. Blondlot, waves on wires, 68, 71, 72. Bolometer, 72, 153. Bornite, 134, 161. Bose, short waves, 60. Boys, C. V., radiomicrometer, 129. Brandes, H., characteristics of de- tectors, 171. Branly, E., coherer, 80, 143. Brant Rock, Mass., tower at. 316. Braun, Ferdinand. Coupled cir- cuits, 101; artificial ground, 121; cathode tube, 151, 181; directed wireless telegraphy, 296, 297. Break key, 90. British stations, 326. Brookite, 177, 187. Calibration of wave meter, 22, 117. Calzecchi-Onesti, coherer, 80. Capacity. Electrostatic, 22; of con- denser, 24; of earth, 24; analogy, 26; measured by wave meter, 224; amount at sending station, 318; formulas for, 339, 340. Cape Race, 106, 107. Capillary electrometer, 142. Carbon microphone. As detector, 158; applied to wireless teleph- ony, 309. Carbon-steel detector, 158, 198. Carborundum. Detector, 160; ex- periments with, 162; unilateral conductivity, 164; current- voltage curves of, 164, 165, 167, 169, 171; oscillograms of, 187. Cathode tube, 151, 181. 343 344 INDEX Cavendish Laboratory, 9. Cavendish, laws of repulsion, 330. Cay, C. H., letter to, 5. Characteristic. Of crystal contact, 164, 165, 167, 169, 171, 180, 181; rising, 171; falling, 171; of arc, 255, 258. Claims of Marconi's 1896 patent, 90. Clifden, Ireland, antenna at, 315. Coal gas, arc in, 258. Coefficient of coupling. Denned, 236, 250; advantage of varying, 289; effect on resonance, 289. Coefficient of .induction. Self, 19; Mutual, 18. Coherer, 80, 85, 143; and circuit, 81, 86; applied to study of electric waves, 81; theory of, 144. Cole, A. D., short waves, 59. Compressed-air condenser, 318. Conduction, electric, in gases, 9. Conductivity. Unilateral, 164, 170, 172; high-frequency, 337. Conductor, earth not a perfect, 125. Condenser. Discharge of, 1, 2, 3, 26, 28, 31; definition of, 24; energy and e.m.f. of, 25; work of charging, 27; air, 114; sending, 317; receivmg, 323. Condenser circuits. Resonance be- tween, 42; measurement of wave length of, 246. Conrad, F., wave length of oscil- lator, 116. Construction of apparatus, 312. Continuity of arc-oscillations, 260. Corpuscular theory of electricity, 8, 10. Coulomb, torsion balance, 330. Coulomb, 24. Coupled circuits, 93; two types, 95; introduction of, 97; reasons for using, 228; oscillations in, 228; detuning of, 251. Coupled pendulums, 232. Coupling. Close and loose, 241 ; per- fect, 241; coefficient of, see coef- ficient. Criterion of oscillation, 30, Crystal detectors. See Rectifiers, Crystal. Crystal rectifiers, 156, 157, 175. See Rectifiers, Crystal. Cuba, U. S. Stations in, 326. Current distribution in oscillator, 108, 111, 115. Current- voltage characteristic. Of carborundum, 164, 171; of molyb- denite, 180; of electrolytic detector, 203; of arc, 255, 258. Curvature of earth, effect of, 128. Cut-over switch, 90, 322. Cycle, hysteresis, 149. Cyclic change, 21. Cymometer, 221. Damped discharge, 34. Damping, 30, 146, 291. Dawn, effect of, on transmission, 134, 136. Daylight, effect of, 133. Decohering devices, 85. De Forest. Audion, 214; arc in steam, 259. De la Rive. Spark in oil, 57; Hertz's experiments, 68. Detectors, 140; why needed, 142; classification of, 143; magnetic, 145; thermal, 153; crystal, 156, 157, 175; rectifiers as, 171; why recti- fiers act as, 173; effects of resistance of, 291; electrolytic, 201; vacuum, 212; and potentiometers, 324. Detuning, 251. Dielectric, definition of, 24. Dielectric constant, 25 ; relation to in- dex of refraction, 40; table of, 341. Diffusion, of electric current, 63. Direct-coupled circuits, 95, 96. Directive antenna. Marconi, 298; intensity about, 298; theory con- cerning, 299. Directive wireless telegraphy, 296; limitations to, 303. Discharge. See Condenser. Displacement assumption, 37. Displacement current, 37; about an oscillator, 39. INDEX 345 Dissipation, of charge by ionized air, 138. Distance. Law of, 130; of trans- mission over different soils, 131. Doenitz, Johann, wave meter, 216. Dolbear, Amos, wireless telegraphy of, 77. Double oscillation, spark photograph of, 248. Drude, Paul. Calibration of wave meter, 117; resonance method of measuring wave length, 216; wave meter, 216. Duane, velocity of waves on wires, 68, 69. Duddell, W. Law of distance, 129; thermo-galvanometer, 129, 154; singing arc, 254, 255. Dunwoody, H. H. C., carborundum detector, 160. Duplicity of vibration of coupled cir- cuits, 235. Dynamometer, high-frequency, 113. Earth. Propagation of electric waves over, 122, 125. See Ground. Edward VII, message from Pres. Roosevelt to King, 107. Einthoven galvanometer, 141, 263. Electric force, related to potential gradient, 334. Electric waves. Maxwell's theory of, 5, 36, 38; Hertz's experiments on, 5, 43, 50, 51, 66; properties of, 40, 48; interference of, 45, 46, 47, 48; refraction of, 40, 55; velocity of, in air, 40, 50, 69; of short wave length, 51, 56, 59, 60; polarization of, 54; table of, 60; on wires, 62, 66, 74; velocity of, on wires, 66, 68, 69, 70, 74; from grounded os- cillator, 124. Electricity. Theories as to nature of, 6; and magnetism, 12; elementary facts about, 329. Electrolytic detector, 201; current- voltage characteristic, 203; oscillo- graphic study, 205; conclusions re- garding, 211. Electromagnetic theory of light, 5, 36,41. Electrometer. Capillary, 142; abso- lute, 329. Electromotive force. Of condenser, 25; definition of, 334. Electron, mass and charge of, 9. Electroptatics, 23, 329. Energy. Relation of magnetic field to, 21; and e.m.f. of charged con- denser, 25. Engineering details, 312. Fahie. History, 75, 82; letter to, 158. Falling characteristic, 171. Farad, 24, 336. Faraday, Michael. Electrolysis, 8, 9; current from magnetic field, 16; electrostatics, 23; dielectric, 24; basis of Maxwell's theory, 36. Feddersen, rotating-mirror photo- graphs, 3. Fessenden, R. A. Barretter, 154; electrolytic detector, 201, 203; high-frequency alternator, 306; tower at Brant Rock, 316. Field of electric force about oscilla- tor, 49, 124. Field of magnetic force, 13, 50. Fizeau, velocity of electric propaga- tion, 62. Fleming, J. A. Dynamometer, 113; note on Zenneck's theory, 125, 133; oscillation valve, 212; wave meter, 220; method of measuring capacity, 224; on directive antenna, 299. Formulas. For current during dis- charge, 31; period of discharge, 35; for two wave lengths in coupled circuits, 236; for period of arc, 260; for sending capacity, 319; for high-frequency resistance, 337; for calculating capacity, 339; for cal- culating inductances, 341. Franklin, Benjamin, theory of elec- tricity, 7. Frequency meter. See Wave meter. 346 INDEX Galvanometer. Principle of, 13; sen- sitiveness of, 141; Einthoven's string, 141. Geissler tube detector, 70, 216. Glace Bay, Nova Scotia, 107. Gounelle, Velocity of electric pro- pagation, 62. Graph of current, 32. Grating used in showing polariza- tion, 54. Ground. Quarter-wave, 121; prac- tical, 317. Grounded circuits, 83, 108, 118. Guided electric waves, 124. Hack, F., effect of sub-surface water, 133. Harmonic oscillation, 279. Heat, 21, 60. Heaviside, Oliver, theory of waves on wires, 63, 65. Height of antenna, dependence on, 271, 275. Helix, 321, 344, 345. Helmholtz, on elementary portions of electricity, 9. Henry, Joseph. Magnetism by Ley- den-jar discharge, 2; obtained cur- rent from varying magnetic field, 16. Hertz, Heinrich. Existence of elec- tric waves in air, 5, 41, 42, 43; oscillator, 44; resonator, 44; re- flection of electric waves, 44; attempt to measure velocity of electric waves, 50; short waves, 51 ; waves on wires, 66. Hertz oscillator. Displacement cur- rent about a, 39; distribution of current and potential in a, 108. Hewitt. Mercury interrupter, 270, 278. Hot-wire ammeter, use in tuning, 249. Hughes, D. E. Microphonic detec- tor, 158. Hydrocarbon gas, arc in, 259. Hydrogen, arc in, 258. Hysteresis, 21; dielectric, 25; sup- pression of, 148. Identity of electric waves and light, 56. Images, electrical, 122. Image theory of ground, 118, 122. Index of refraction, 40. Indicating instruments, 140. Inductance. Mutual, defined, 18; self, 19, 20; tuning by, 93, 272; formulas for calculating, 341. Induction. Electromagnetic, 17; mu- tual and self, 17, 18; resem- blance to inertia, 20. Inductively coupled circuits, 95. Inertia, 20, 21. Infra-red and electric waves, 60. Installation, 322. Interference, 45; possibility of pre- venting, 271; curves showing ex- tent of, 294. Interrupter. Mercury, 270, 278; vi- brator used with persistent oscil- lations, 262. lonization of air, 137, 138. Irish Channel, experiments on, 129. Iron wires, velocity on, 69. Jackson, H. B., transmission over various surfaces, 133. Johnson, K. S., assistance with rec- tifier, 176. Kelvin, Lord. See Thomson, Sir Wm. Kennelly, A. E. , explanation of day- light effect, 139. Key, 321. Kinraidy, spark gap, 270. Kirchhoff, theory of waves on wires, 63. Kites, 89, 91. Klemencic, thermal junction, 59, 154. Korda, air condenser, 114. Lampa, short waves, 60. Lebedew, short waves, 59. Lecher, waves on wires, 70. Lepel, Baron von, arc, 265. Leyden jar discharge. See Condenser. Leyden jars, copper-plated, 317. INDEX 347 Light. Electromagnetic theory of, 3, 41; identity of electric waves and, 56; table, 60; effect on transmis- sion, 133. Lindsay, J. B., signaling through water, 76. Loadstone, 12. Lodge-Muirhead-Robinson, coherer, 144. Lodge, Sir Oliver. Resonance ex- periment, 42, 215; use of coherer, 81; patent of resonant circuits, 93; system of wireless telegraphy, 97. Loops of potential and current, 45, 46, 47, 48, 111. Lyman, Theodore, ultra-violet light, 60. Macdonald, wave length of oscil- lator, 116. Madelung, E., on magnetic detec- tor, 151. Magnet, 12. Magnetic detector, 145, 146, 147, 151, 153. Magnetic Field, 13, 14, 15, 16; about a Hertz oscillator, 50. Magnetism, relation between elec- tricity and, 12. Magnetization by condenser dis- charge, 2. Mandelstam, phase-difference exci- tation, 298. Map of stations, 326. Marconi, Guglielmo, 80; first patent, 83; 1896 apparatus, 83; grounded circuits, 85; coherer, 85; deco- hering devices, 85; " claims," 90; achievements between 1896 and 1898, 91; coupled circuits, 103; duplex apparatus, 105; achieve- ments in 1901-1902, 106; effect of daylight, 133; company, 139; mag- netic detector, 146; reflectors, 296: directive antenna, 298. Maurain, C., suppression of hys- teresis, 149. Maxwell, James Clerk, electro-mag- netic theory, 5, 36, 41. Medium, influence of intervening, 23. Mercury-arc oscillator, 307. Method of wireless telephony, 305. Microfarad, 24. Microphone. As detector, 158; ap- plied to wireless telephony, 309. Mirrors, cylindrical metallic, 51. Molybdenite, 161, 177, 178; oscillo- grams of, 186; thermo-electric prop- erties, 189. Monarch, repair ship, 129, 131. Morse, S. F. B., telegraphy by con- duction through water, 75. Mounting for molybdenite detector, 179. Muirhead, coherer, 144. Nasmyth, G. W., period of arc, 260. National Electric Signaling Co., 139. Navy, U. S., Stations on Atlantic Coast, 326. Nodes, 45, 46, 47, 48, 111. Northrup, E. F., dynamometer, 113. Oersted, H. C., relation of electricity to magnetism, 13, 14. Oil, castor, condensers submerged in, 317. Oil, vaseline, spark in, 57. Optics of electric oscillations, Righi, 56. Oscillation. Spark-photograph of, 3; period of, 35; number, 87; nature of, 108; of coupled systems, 228; photograph of double, 248; har- monic, 279. Oscillator. Hertz, 44; field about, 49; rectilinear, 51; for short waves, 59; Marconi, 83; wave-length of, 116; mercury-arc, 307. Oscillatory discharge. See Conden- ser. Oscillographic study, of crystal recti- fiers, 181; of electrolytic de- tector, 205, 208; of pendulum motion, 233. Paalzow, waves on wires, 72, 73; bolometer, 154. Panama, U. S. Stations in, 326. 348 INDEX Parallel, capacities in, 340. Pendulum, coupled or sympathetic, 232. Persistent oscillations, 262; train pro- duced by, 306. Period. Of oscillation, 35; relation of wave length to, 50; of singing arc, 260. Peukert, rotating quenched spark,268. Phase-difference oscillator, 297. Pickard, G. W. Daylight effect, 134; crystal detectors, 161; copper- plated jars, 317. Pierce, G. W. Spark potential along glass, 56; number of oscillations, 87; current distribution, 111; dyna- mometer, 113; wave length of oscillator, 116; test of image theory, 119; crystal-contact detectors, 162; anatase, brookite, molybdenite, 177; sound measurements, 178; oscillographic study, 181, 205; thermoelectric properties, 189 ; con- clusions regarding crystal recti- fiers, 199; electrolytic detector, 205, 211; wave meter, 221; photo- graph of double oscillation, 248; mercury interrupter, 270, '278; resonance, 271, 281. Platinized contacts, 169. Poincare", H., concerning error, 50. Polarization. Of dielectric, 37; of electric wave, 53. Poldu, England, 106; directive an- tenna at, 298. Popalexi, phase-difference excita- tion, 298. Popoff, receiving apparatus, 81. Porto Rico, U. S. Stations in, 326. Potential, 332; distribution in oscil- lator, 108; gradient, 334. Potentiometer, 324. Poulsen, V. Singing arc, 256; in- terrupter at receiving station, 263. Power, 228, 318. Practical units, 336. Preece, Sir Wm. Telegraph by elec- tromagnetic induction, 78; co-oper- ation with Marconi, 91. Principles, reliance on, 325. Pupin, M. J. Loaded-line telephony, 65; electrolytic rectifier, 202, 204. Quenched spark, 253, 266, 267, 268, 269. Radiant heat and electric waves, 60. Radium, renders gases conductive, 9. Rathenau, telegraphy by water con- duction, 77. Ratio of units, 40, 332. Rays and shadows, 52. Receiving circuits. Resonance of, 271; forms of, 285. Receiving condensers, 323. Receiving transformer, 323. Recording device, 263. Rectification. Of A. C. by crystal contact, 170; facts adverse to ther- . mo-electric explanation of, 196; in- sufficient heating of contact to account for, 198. Rectifiers. Crystal, 145, 157, 162; with or without battery, 172; why, act as detectors, 173; permanence of, 176; molybdenite, 178; oscillo- graphic study of crystal, 181 ; sum- mary of conclusions regarding crystal, 199; electrolytic detector a, 211. Reflectors, metallic, 51, 296. Refraction, 40, 55. Reichmann, Fritz, dynamometer, 113. Relay, sensitiveness of, 140. Repulsion, electrostatic, 329. Resistance. Contrast of inductance with, 21; damping by, 34; pro- tective, 88; temperature coefficient, 194; effect of, on sharpness of resonance, 225; of detectors and effect on resonance, 291; high- frequency, 337. Resonance. Between condenser cir- cuits, 42; of circuits, 92; electrical, 215; effect of resistance on, 225; of sending station, 243; of receiving circuits, 271; curves, 281; relation in coupled system, 286; sharpness of, with coupled circuits, 291. INDEX 349 Resonator. Hertz's circular, 44; rec- tilinear, 51; Righi's, 56; with thermal junction, 59. Righi, Augusto, apparatus, 56. Rising characteristic, 171. Robinson, coherer, 144. Robison, S. S. Manual of Wireless Telegraphy, 219. Roentgen rays make gases conduc- tive, 9. Rogers, telegraphy through water, 76. Roosevelt, President, message, 107. Rubens. Waves on wires, 72, 73; telegraphy by water conduction, 77; bolometer, 154. Rutherford, E., magnetic detector, 145. Sarasin. Repetition of Hertz's ex- periments, 68; spark in oil, 57. Saunders, velocity of waves on wires, 68. Schloemilch, electrolytic detector, 201, 203. Schmidt-Wilkes telephone receiver, sensitiveness of, 140. Schumann, V., ultra-violet light, 60. Seawater, propagation of electric waves over, 125, 131. Sending station. Tuning of, 243; construction of, 312. Shadows, cast by metallic screens, 52. Shoemaker, electrolytic detector, 203. Shunt capacity, tuning by, 273. Shunted telephone, used with detec- tor, 135. Silicon, 161; steel, 198. Silver, removal of, 324. Simon, H. Th., talking arc, 254. Singing arc, 253, 260, 264. Singing spark, 253. Skin effect, 70, 337. Soil, propagation over, 126, 131. Spark. In oil, 57, 268; potential, 29, 56; photographs, 3, 248; quenched, 253, 266, 269; singing, 253. Spectrum of electric waves, 60. Station, diagram of circuits of com- plete, 322. Stationary waves, 48; on wires, 74. Steam, arc in, 259 Steel-carbon detector, 158, 198. St. John, waves on wires, 73, 74. St. Johns, Newfoundland, 106. Strecker, telegraphy by water con- duction, 77. Submarine telephony, limit to, 65. Sunset, effect of, 134, 136. Sun's rays, ionization by, 138. Surface travel, 69. Sympathetic pendulums, 232. Syntonic circuits. See Resonance. Table. Of wave lengths, 60; of di- electric constants, 341 ; of units, 336. Talking arc, 254. Taylor, J. E., law of distance, 129. Telefunken Co. Arcs in series, 259; quenched spark, 267. Telegraphy, by wires, 63. Telephone receiver, sensitiveness of, 140. Telephony. Line, 65; wireless, 265, 305. Tellurium detector, 160, 198. Tesla coil, 93. Thermal detectors, 59, 153, 154. Thermal junction, use in receiver, 59, 154. Thermo electric, 177, 189, 196. Thomson, Elihu. Transformer, 93; dynamometer, 113; continuous spark, 253; singing spark, 265. Thomson, J. J., electricity and mat- ter, 7, 9, 10. Thomson, Sir Wm. Proof of oscilla- tory discharge, 3; criterion, 30; period of oscillation, 35; waves on wires, 63; absolute electrometer, 329. Torsion balance, 329. Tosi, directive wireless telegraphy, 302. Transformer. High-frequency, 93, 95; charging, 320; receiving, 323. Trowbridge, John, 68, 69, 76. Tube. Geissler, 70, 216; cathode, 151. Tuning. See Resonance. 350 INDEX Uller, Carl, on directive antenna, 300. Ultra-violet light and electric waves, 9, 60, 137, 139. Unilateral conductivity, 164, 170, 172. Units, 24, 329, 336. Vacuum detectors, 70, 212, 216. Vail, telegraphy by conduction through water, 76. Varley, S. A., coherer, 80. Velocity of electric waves, 40; attempt to determine, in air, 50; relation of, to wave length and period, 50; on wires, 62; on wires same as in sur- rounding medium, 68; same as of light, 69. Visible spectrum, 61. Volt, unit of e.m.f ., 24. Vreeland, F. K., oscillator, 307. Wasmut, A., on Peukert's spark, 268. Water. Propagation over, 125; sub- surface, 133. Waves, electric. See Electric Waves. Wave length. Determined by sta- tionary system, 45; relation of, to period, 50; of electric waves and light, 60; of Hertz oscillator, 116: absorption conditioned oii, 13&; in coupled systems, 235; three pos- sible, 251. Wave meter. Calibration of, 117, 222; description, 216; observa- tions with, 219; method of using, 223; used in measuring capacity, 224; applied to tuning sending station, 244. Wave-system, stationary, 48. Webb, wave length of oscillator, 116. Webster, A. G., electricity and mag- netism, 63. Wellfleet, Mass., 107, 298. Wheatstone, velocity of electricity, 62. Wien, Max. Detuning, 251 ; quenched spark, 266, 270. Wireless telegraphy. Before Hertz, 75; by Hertzian waves, 80; by resonant circuits, 92. Woodman, wave length of oscilla- tor, 116. Work. Done in charging a con- denser, 27; definition, 332; method of computing electrical, 334. Zenneck, J. Propagation over earth, 125; absorption a function of wave length, 133; on directive antenna, 300. Zincite, 134, 161. OF THE OF f) 14 DAY USE RETURN TO DESK FROM WHICH BORROWED LOAN DEPT. This book is due on the last date stamped below, or on the date to which renewed. Renewed books are subject to immediate recall. AUG 91963 LD 21A-50m-12,'60 (B6.221slO)476B YC 19344 TK 574-1