LIBRARY OF THE UNIVERSITY OF CALIFORNIA. Class SECOND EDITION, REVISED AND EXTENDED. With 7 Plates and Numerous Other Illustrations. SYO. 28s. net. THE PRINCIPLES OF ELECTRIC WAVE TELEGRAPHY AND TELEPHONY BY J. A. FLEMING, M.A., D.Sc., F.R.S. Fender Professor of Electrical Engineering in the University of London, etc. PRESS NOTICES OF THE SECOND EDITION "Dr. Fleming's 'Principles' is still the classic on the subject." Westminster Gazette. "The book, which is fully illustrated, is the result of an immense amount of research and careful compilation, and should be a valuable work of reference on what is one of the most fascinating of present-day engineering studies." Glasgow Herald. "For those who desire a thorough understanding of the nature and action of electric waves, and of the machinery by which inventors have up till now secured control of this subtle agency, this book may be said to provide the latest word." Scotsman. " There is no doubt that this book should be regarded as the standard work on the subject, for there is no other volume with anything like its completeness, or with as thorough a treatment of the theoretical side of the subject." Electrical Engineering. PRESS NOTICES OF THE FIRST EDITION " The most complete treatise on the subject in the English language." Electrical Review, New York. "The student of wireless work will of course obtain this book of necessity as well as by choice." Gassier } s Magazine. " It is a work of real value,' that will remain a standard text-book for a long time to come, and we can cordially recommend it as such." Electricity. "Of the many books published on the subject of wireless telegraphy, the one before us will deservedly take a prominent position. ... It will prove a valuable and delightful possession." The Engineer-in-Charge. " Cet ouvrage est extremement remarquable. L'auteur a traite d'une fa9on tout a fait scientifique quoique avec grande simplicite et nettete, 1'ensemble des questions relatives a la telegraphic sans conducteurs." L'Eclairage Electrique. "Any book by Dr. Fleming is always both eagerly looked for, and as eagerly read. Moreover, unlike many technical books Dr. Fleming's works go at once to the 'standard' shelf in the bookcase. . . . The volume now before us is the most exhaustive treatise on wireless telegraphy we have seen." The Electrical Magazine, " In Dr. Fleming's book we have undoubtedly the one to be recommended to students especially interested in wireless telegraphy. ... In Dr. Fleming's book is to be found a treatment of the subject which is exhaustive and thorough both on the theoretical and practical sides. It is a book which has been long wanted, and will be warmly welcomed." Nature. LONGMANS, GREEN, AND CO. LONDON, NEW YORK, BOMBAY, AND CALCUTTA By the same Author The Principles of I Electric Wave Telegraphy and Telephony SECOND EDITION (REVISED AND EXTENDED] With 7 Plates and Numerous other Illustrations 8vo. 28/- net. LONGMANS, GREEN, AND CO. I ONDOX. NEW YORK, BOMBAY, AND CALCUTTA AN ELEMENTARY MANUAL OF RADIOTELEGRAPHY AND RADIOTELEPHONY FOR STUDENTS AND OPERATORS BY J. A. FLEMING, M.A., D.Sc., F.R.S. FENDER PROFESSOR OF ELECTRICAL ENGINEERING IN THE UNIVERSITY OF LONDON PROFESSOR OF ELECTRICAL ENGINEERING IN UNIVERSITY COLLEGE, LONDON MEMBER AND PAST VICE-PRESIDENT OF THE INSTITUTION OF ELECTRICAL ENGINEERS OF LONDON FELLOW AND PAST VICE-PRESIDENT OF THE PHYSICAL SOCIETY OF LONDON MEMBER OF THE ROYAL INSTITUTION OF GREAT BRITAIN ETC., ETC. WITH ILLUSTRATIONS NR W EDITION LONGMANS, GREEN, AND CO 39 PATERNOSTER ROW, LONDON NEW YORK, BOMBAY, AND CALCUTTA 1911 T PREFACE THE Author published in 1906, through Messrs. Longmans, Green and Co., a volume on the Principles of Electric Wave Telegraphy, in which an attempt was made to provide a fairly advanced treat- ment of the subject, not limited to mere descriptions of various so-called systems of wireless telegraphy, but explanatory of the scientific principles underlying Eadiotelegraphy in general. It was, however, represented to him that a smaller manual on the subject, suitable for the use of students, practical operators, and the general reader, on the same lines, but somewhat more elementary, might meet with acceptance. The Author has, therefore, endeavoured to put together in the present volume the information most likely to be of use for this purpose. Where it has been deemed advisable to introduce some little mathematical reasoning to supplement the verbal descrip- tions, it is limited to the use of simple operations and expressions. For the proof of many of the formulae given in this Manual, which require rather more extended mathematical discussion than can be given here, the reader must be referred to the Author's larger book above mentioned. It is assumed, however, that any user of this Manual has a general acquaintance with the elementary facts of electrical science. It has not been considered necessary to encumber the pages with many references to original papers or patent specifications, 222532 Vi PREFACE since the student who masters this elementary treatise will be able at once to take advantage of the more complete information and references given in advanced text-books. The subject has now acquired a position of such importance in connection with naval and military signalling and marine inter- communication generally, that means are required for adequately teaching the subject to electro- technical students and to practical operators, and thus equipping them with the initial scientific information necessary to enable them to follow intelligently its practical development, and also fit them for extendiDg their knowledge of it by the study of advanced books and original papers. Although there is no want of books upon the subject, many of them are occupied to a large extent with historical matter, and expositions of electrical phenomena which are either unnecessary for the practical radiotelegraphist, or can be obtained from other text- books. Hence reference to these matters is as far as possible omitted in this Manual, and the information given is confined to that which is necessary to enable a student familiar with the elementary facts of electricity and magnetism to proceed to the study of more advanced treatises on the subject of Eadiotelegraphy. In conclusion the writer desires to express his thanks for permission to make use of diagrams and illustrations of appa- ratus to the following firms and gentlemen : To Chevalier G. Marconi, and Marconi's Wireless Telegraph Company for the loan of blocks illustrating Marconi apparatus and stations ; to the Amalgamated Eadiotelegraphio Company and the Kilowatt Publishing Company for blocks of the Poulsen apparatus ; to the Electrician Publishing Company for the use of many illustrations which have appeared in The Electrician of late years ; to Mr. W. Duddell, F.R.S., for some curves employed in Chapter IX. ; PREFACE Vll and to the Editors of The, Electrical Review, and to the Gesellsckaft fur Drahtlose Telegrapliie, of Germany, for views of the Nauen Station and of radiotelegraphic apparatus. To these and others who have assisted in the preparation of the book, the Author desires to express his full acknowledgments. J. A. F. THE FENDER ELECTRICAL LABORATORY, UNIVERSITY COLLEGE, LONDON, August, 1908. TABLE OF CONTENTS CHAPTER I ELECTRIC OSCILLATIONS AND ELECTRIC RESONANCE High frequency alternating currents and electric oscillations The mag- netic field of an alternating current Definitions of periodic time and frequency Low and high frequency alternating currents and electric oscillations Mode of delineating an alternating current graphically. 2. Undamped and damped electric oscillations Characteristic qualities of undamped oscillations Mechanical illustrations The logarithmic decrement of damped oscillations Strongly and feebly damped oscil- lations. 3. Electric circuits and their qualities High frequency resistance Resistance for steady and for oscillatory currents Distri- bution of electric oscillations on conductors Formulae for high frequency resistance of round wires High frequency resistance of spiral wires Experimental and theoretical results. 4. High frequency inductance Definition of inductance Formulae for the inductance of circuits of various forms. 5. Electrical capacity Condensers for radiotelegraphy Formulae for the capacity of various forms of condensers Table of dielectric constants of various insulators. 6. Time period of oscillatory electric circuits Deduction of formulae of the natural frequency of a circuit Oscillation constant of a circuit. 7. Electric resonance Nature of resonance Loosely and closely coupled inductive circuits Tuning of oscillatory circuits The re- sonance curve of an oscillatory circuit Various forms of resonance curves . CHAPTER II DAMPED ELECTRIC OSCILLATIONS 1. Production of damped oscillations by condenser discharges Dead beat and oscillatory discharge of a condenser Mechanical illustrations Apparatus for the production of oscillations by condenser discharges. 2. Induction coils for radiotelegraphy Mode of constructing induc- tion coils Characteristics of a coil required for radiotelegraphy. 3. TABLE OF CONTENTS PAGK Interrupters for induction coils Hammer, mercury jet, and electro- lytic interrupters Advantages and disadvantages of each type The coal-gas mercury break Modifications of the Wehnelt break. 4. Alternators and transformers for radiotelegraphy Type of alternator most useful Types of transformers used in radiotelegraphy Methods of preventing the arc discharge when using transformers in radio- telegraphy. 5. Spark discharges and spark voltages Forms of silent discharger for spark telegraphy Spark discharges in com- pressed air Table of spark voltages between metal balls Spark resistance and methods of measuring it. 6. Condensers for electric oscillations Mode of constructing condensers used in radiotelegraphy Various forms of condensers with liquid and gaseous dielectrics. 7. Inductances and oscillation transformers for radiotelegraphy Sketch of the theory of the oscillation transformer Mode of making various inductances. 8. Multiple transformation of oscillations for high tension condenser discharges Fleming multiple oscillation transformer Multiple method of Mandelstam and Papalexi 37 CHAPTER III UNDAMPED ELECTRIC OSCILLATIONS 1. High frequency alternators Early attempts in constructing high fre- quency alternators Tesla's machines Machines of Duddell, Lamme, Brown, Fessenden, and others Difficulties of constructing high fre- quency alternators. 2. The production of undamped electric oscilla- tions from a continuous current Elihu Thomson's experiments. 3. DuddelFs musical arc The characteristic curve of an electric arc Static and dynamic characteristics of the arc Simon's researches Explanation of the mode in which a continuous arc can make oscilla- tions in a condenser shunt circuit 4. Poulsen's method of producing undamped high frequency oscillations The characteristic curve of the arc in hydrogen. 5. Other researches on the transformation of con- tinuous currents Simon's investigations Investigations on the electric arc made at University College, London S. G. Brown's method of producing oscillations Improved forms of the Poulsen arc apparatus. 6. Inductive effects of undamped oscillations. 7. Resonance effects in connection with undamped oscillations Formula for calculating the current in a secondary circuit .... CHAPTER IY ELECTROMAGNETIC WAVES 1. The electromagnetic medium Maxwell's researches. 2. Electric and magnetic quantities Definitions and fundamental formulae The lumiferous aether Velocity of propagation of a disturbance through it. 3. The nature of a wave Relation of wave length, frequency and TABLE OF CONTENTS XI PAGE velocity Formula for the velocity of a wave disturbance Properties of the electromagnetic medium. 4. An electromagnetic wave A sketch of Maxwell's theory Description of the mode in which an electromagnetic wave is propagated. 5. Velocity of an electromag- netic wave Experimental determinations of the electromagnetic velocity and comparisons with the velocity of light. 6. Practical production of electromagnetic waves Hertz's researches Creation of electromagnetic waves by electric oscillations Formula for the energy radiated by a Hertzian oscillator 107 CHAPTER Y RADIATING AND RECEIVING CIRCUITS 1. Varieties of radiative and receiving circuits The law of exchanges Open and closed magnetic oscillators. 2. The open circuit oscillator The mode of production of electromagnetic waves by an open circuit oscillator Formulae for the capacity and the decrement of a linear oscillator Energy radiated from a linear oscillator The closed circuit oscillator Energy radiated from a closed circuit oscillator Open and closed circuit oscillations compared. 3. Formula for the energy radiated by a closed circuit oscillator. 4. Receiving antennae or absorbing circuits Explanation of the mode in which electromotive force is set up in the receiving oscillator. 5. The practical construction of antennae Various forms of simple antennas. 6. Multiple wire antennae Mode of construction. 7. Earthed and con-earthed antennae The propagation of electro- magnetic waves over the sea and the earth's surfaces The absorp- tive effect of the earth's surface Discussion of the effect of earthing the transmitting oscillator. 8. The establishment of the fundamental harmonic oscillations in open and closed circuits Oscillations on spiral wires Experiments of Seibt and the Author. 9. Modes of exciting oscillations in open or closed radiating circuits Direct exci- tation of a transmitting antenna Direct coupling of an antenna to a condenser circuit Inductive coupling of an antenna to a condenser circuit Electrostatic coupling Waves emitted by inductively coupled antennae. 10. Appliances for giving direction to electromagnetic radia- tion Directive antennae Early experiments Marconi's bent antenna The Author's theory of the same Marconi's experimental results Braun's method of directive telegraphy Closed circuit directive tele- graphy of BeJlini and Tosi Method of locating the position and direction of a sending station 135 CHAPTER VI OSCILLATION DETECTORS 1. Classification of oscillation detectors. 2. Spark detectors Hertz resonator. 3. Imperfect contact detectors Development of the coherer Branly's researches Improvements in the coherer by Lodge Xli TABLE OF CONTENTS and Popoff Marconi's telegraphic detector Arrangements for tapping Forms of self-restoring coherer due to Lodge and Muirhead, L. H. Walter and Solari-Castelli. 4. Magnetic oscillation detectors Researches of Rutherford, E. Wilson, and others Marconi's magnetic detector Fleming magnetic detector Various researches on the operation of a magnetic detector Walter-Ewing magnetic detector. 5. Thermal and thermoelectric detectors Fessenden's barretter Fessenden's liquid barretter Thermoelectric detectors of Duddell, the Author, Austin, and others. 6. Electrolytic oscillation detectors of Fessendenand Schlb'milch S. G. Brown's peroxide of lead detector. 7. Valve or rectifier oscillation detectors Fleming oscilla- tion valve or glow lamp detector Crystal detectors Dunwoody's carborundum detector Pierce's researches on crystal detectors. 8. Electrodynamic oscillation detectors Fleming copper disc galvano- meter. 9. Mode of employing oscillation detectors in combination with recording instruments to detect electric waves Construction of a relay Telegraphic and telephonic methods of reception . . . CHAPTER VII RADIOTELEGRAPHIC STATIONS General principles of radiotelegraphy Method of signalling by the Morse alphabet General arrangements of transmitting and receiving stations The early development of radiotelegraphy. 2. Short distance radiotelegraphic apparatus Antennse or radiator supports Selection of site Forms of mast employed for sustaining the antenna Earthed and non-earthed systems of radiotelegraphy Construction of the earth- plate. 3. The arrangement of the transmitting apparatus for short distance radiotelegraphy Mode of exciting oscillations by an induction coil Signalling keys Mode of exciting oscillations in the antenna by direct charging, direct coupling and inductive coupling Forms of condenser used Silent dischargers Transmitting appliances for the production of undamped waves. 4. Short distance receiving apparatus Form of apparatus employing imperfect contact detectors and Morse printer Form of receiving apparatus employing self-restoring contact detectors Forms of receiving apparatus employing electrolytic receivers and telephones Arrangements for receiving and detecting undamped electric waves. 5. Systems of intercommunication by short distance radiotelegraphy Establishment of ship to shore communica- tion by the Marconi Wireless Telegraph Company Methods employed for communication with ships at sea Obstacles to radiotelegraphy Atmospheric disturbances Interference of stations. 6. Long distance radiotelegraphy The general design of power stations Arrangements of first long distance radiotelegraphic station at Poldhu Marconi's achievements in Transatlantic radiotelegraphy Voyages of the Carlo Alberto Long distance communication with Atlantic liners established by Marconi Marconi's dischargers for long distance telegraphy Description of Marconi long distance stations at Glace Bay, Nova Scotia, and Clifden, Ireland Description of the German radiotelegraphic station at Nauen Description of the Poulsen station TABLE OF CONTENTS Xlii at Cullercoats Relative advantages of the arc and spark method of signalling^ 7. Effect of atmospheric conditions and terrestrial obstacles on long distance radiotelegraphy Effect of atmospheric ionisation Admiral Sir Henry Jackson's observations on the effect of inter- posed land Effect of earth curvature on long distance communication 216 CHAPTER VIII RADIOTELEGRAPHIC MEASUREMENTS 1. Importance of radiotelegraphic measurements. 2. The measurement of high frequency currents Hot wire ammeters Construction of hot wire ammeters for high frequency currents Various forms of hot wire ammeter. 3. Measurement of high frequency potential difference Voltmeter method Interpretation of voltmeter readings Spark volt- ages. 4. Measurement of capacity Absolute and comparison methods Method of measuring small capacities used in radiotelegraphy Fleming and Clinton commutator Method of measuring the capacity of an antenna. 5. Measurement of inductance Ander- son's method of measuring inductance Modifications introduced by the Author Measurement of mutual inductance. 6. Measurement of frequency The oscillation constant of a circuit The cymometer Method of employing the cymometer for measuring frequency The use of the cymometer in determining the logarithmic decrement of damped oscillations The measurement of capacity and inductance by the cymometer. 7. The measurement of wave lengths Relation of wave length to frequency and velocity of propagation The wave lengths in use in connection with radiotelegraphy and their measurement. 8. The measurement of damping and logarithmic decrements by the cymometer and other means Determination of resonance curves by the methods of Bjerknes and Drude Examples of the determination of the damping of a given oscillatory circuit - The measurement of high frequency resistance Measurement of spark resistance Methods of Slaby and Bjerknes Experiments of the Author and others . . . 272 CHAPTER IX RADIOTELEPHONE 1. The problem of radiotelephony Various methods Radiotelephony by electromagnetic waves Necessary appliances Difference between telephony with wires and without The nature of a sound wave in the case of articulate speech Diagrammatic representations of articulate sounds. 2. Transmitting arrangements in radiotelephony High fre- quency alternator method Type of high frequency alternator required Fessenden's high frequency alternator Combined steam turbine alternators Conditions to be fulfilled by high frequency alternators XIV TABLE OF CONTENTS PACK for use in radiotelephony. 3. Electric arc transmitters for radio- telephony Poulsen's arrangements Recent forms of electric arc oscillation generators Construction of condensers and inductances. 4. Microphonic control of electric oscillations Construction of a microphone Parallel microphones Mode of using a microphone to control electric oscillations and varying intensity of electric waves Various ways of employing a microphone as a speech transmitter in radiotelephony 5. Other arrangements employed as transmitters in radiotelephony Electric arcs in series used as oscillation generators Arrangement of Fessenden employing continuous current dynamos in series Difficulties of employing condenser discharges in connection with radiotelephony. 6. Receiving arrangements in radiotelephony Conditions to be fulfilled by the oscillation detector Types of available oscillation detector Electrolytic, glow lamp, thermoelectric, and crystal rectifiers Construction of a receiving circuit Special advantages of radiotelephony. 7. Present state and achievements of radiotelephony Poulsen's experiments between Berlin and Copen- hagen, and Lyngby and Esbjerg Fessenden's experiments and demonstrations Conclusion . , 309 AN ELEMENTARY MANUAL OF RADIOTELEGRAPHY AND RADIOTELEPHONY CHAPTER I ELECTRIC OSCILLATIONS AND ELECTRIC RESONANCE 1. High Frequency Alternating Currents and Electric Oscillations. Since the art and practice of radiotelegraphy and radiotelephony involve the employment of electric currents which alternate or change direction in their circuits very rapidly, it is necessary to commence the study of the subjects by considering some of the general properties of alternating currents. By means of various appliances, such as a dynamo, voltaic cell, thermopile, or other source of so-called electromotive force, we can produce in certain bodies, known as electrical conductors, a state in which they are said to be traversed by an electric current. We recognise the presence of a current by the production of heat in the con- ductor and a magnetic field around it. Exploring the space near a conductor carrying an electric current, by means of a freely suspended magnetic needle, we find the latter sets itself so as to indicate that round the wire there is a distribution of magnetic flux along closed lines embracing the wire. Experi- mentally this is best illustrated by pass- ing a stout copper wire through a hole in a card just large enough to let it pass, and connecting the ends of the wire to a powerful battery or dynamo (see Fig. 1). If we then place a small pocket compass on the card and move FlG - ! it about, we shall find that the needle places itself at every point transversely to the wire. If iron filings B are sprinkled on the card and the latter gently tapped, the filings will be found to collect themselves more or less along certain circular lines, thus revealing the form and distribution of the invisible closed lines of magnetic flux round the conductor, whilst if the compass is placed on the card over the filings, it will be seen that the needle sets itself at all places so as to be in the direction of a tangent to these circles. If the connections of the wire with the terminals of the battery or dynamo are interchanged, it will be found that the compass needle reverses its direction, but the lines of magnetic flux remain circles as before. We may therefore speak of the currrent as having direction in the wire, since the magnetic field as indicated by the setting of the magnetic needle has direction one way or the other with reference to the wire. We are ac- customed to call the direction of a line of magnetic flux the direction in which the north-seeking end, or, as it is usually called, the north pole N, of the compass needle points when placed on that line. The direction of the magnetic field is conventionally related to that of the current in the same manner as the twist and thrust of a corkscrew. Hence, if we imagine a watch laid face upwards on the above-mentioned card and that the circular lines of magnetic flux have the same direction as the rotating hands of the watch looked at from above, then the current creating them would be said to have a downward direction or to be flowing from the face to the back of the watch. If the direction of the magnetic flux remains constant from instant to instant, it is said to be due to a continuous, unvarying, or direct current (D.C.). If, however, the field changes its direction at regular intervals, being first right handed or clockwise in direction and then left handed or counter-clockwise, we say that the current is due to an alterna- ting current (A.C.). The interval of time between two consecutive reversals of direction is called a semi-period, and the interval between two consecutive reversals in the same direction is called the periodic time, or complete period. The number of complete periods executed in a second is called the frequency of the alterna- tions, and is generally denoted by the sign ~. Thus 100 ~ means that the frequency is 100, or there are 100 complete periods per second or 200 reversals of direction of the field and current per second. If the frequency has any such value as 50 or 100 ~ the current would be referred to as a low frequency alternating current. If, however, the frequency were of the order of 1000, 10,000, or 100,000, it would be described as a high frequency alternating cur- rent. There is, of course, no hard-and-fast line of demarkation. The terms high and low in this connection are relative or conventional. When the frequency rises to a value of a million or so the current is generally called an electric oscillation. ELECTRIC OSCILLATIONS AND ELECTRIC RESONANCE 3 In radiotelegraphy we are chiefly concerned with high frequency currents or electric oscillations of a frequency between 100,000 or so and 1 or 2 million, and we have first to consider the mode of representing them and their peculiar qualities. It is most convenient to delineate an alternating current by means of a wave diagram, as follows : Draw a straight horizontal line and let distances marked off from one end represent time. Set up perpendicular (dotted) lines either above or below the line at equidistant points, the length of these lines representing the current in the circuit at that instant, the lines being drawn upwards when the current is in one direction and downwards when it is in the other direction. The gradual increase and decrease of the alternating current first in one direction and then in the opposite direction is then represented by the ordinates of an undulating curve, as in Fig. 2. A curve of this FIG. 2. kind, called a sine curve, may be constructed in the following manner. Take any line OX on which to mark off time (see Tig. 3), and with as centre describe a circle with diameter AB. Divide its FIG. 3. circumference into 12 parts, and through these points draw horizontal lines. Take any length BD on the horizontal line and divide it also into 12 parts and through these points draw lines perpendicular to OX to intersect the horizontal lines drawn through the 12 points on the circumference of the circle. Then mark dots at the intersections of the corresponding vertical and horizontal line, that is, at the intersection of horizontal line through point 1 on the circle with the vertical line through poiqt 1 on the time line, and so on. Through the twelve 4 RADIOTELEGRAPHY intersection points draw a wavy curve. This curve is called a sine curve, because its ordinate PM is proportional to the sine of its abscissa MB, reckoning the whole length BD as divided into 360 parts or degrees. Thus if BM is ^ of BD, the ordinate PM at that point M is proportional to the sine of the angle of 30, or is 0*5 and on the same scale that the maximum ordinate of the curve is taken as unity. This maximum value of the ordinate is called the amplitude of the wave curve. An alternating current which is represented by such a sine curve is called a simple harmonic or simple periodic current. We may, however, have alternating currents represented by any form of wave curve provided it is single valued, that is, has only one value of the ordinate for one given abscissa; in other words, by any periodic curve which does not cut or double back on itself.* In connection with alternating currents we are some- times concerned with the maximum value or amplitude at given moments, but more frequently with a mean value of a particular kind called the effective or, usually, the root-mean-square (E.M.S.) value, denned as follows : The rate at which a current is producing heat in a conductor is at any instant proportional to the square of the current, and also simply proportional to the true effective resistance which the conductor offers to that current. If, then, the conductor has such a form that its true resistance for continuous currents is the same as for the alternating current in question, the mean value of the heat produced in any time is proportional to the mean or average value of the square of the current during that time. Hence we may ask the following question. If an alternating current of any given wave form exists in a circuit, find the value of the con- tinuous or unvarying current which will produce heat at the same rate in the same conductor. Suppose we have the wave form of the current given. Then if we draw many equidistant current ordinates during one complete period and square them, that is, multiply the number denoting their length by itself, we can set off a new curve whose ordinates drawn at the same instant represent the square of the instantaneous values of the varying current. This curve is represented in Fig. 3 by the dotted curved line. If, then, we take the mean of the value of these squares and the square root of this mean, we have the value of the continuous current which would produce heat in the conductor at the same rate. The square root of the mean of the squares of the various instantaneous values of the current at equidistant intervals of * Various curves representing alternating currents of complex wave form are shown in Fig. 2, chap. ix. ELECTRIC OSCILLATIONS AND ELECTRIC RESONANCE c time is called the root-mean-square or RM.S. value of the alternating current. When, therefore, we speak of an alternating current of 1 ampere we mean a periodic current which would produce heat at the same rate as an unvarying or continuous current of 1 ampere, assuming that the effective resistance of the conductor is the same in the two cases. The importance of tliis proviso will be seen later on. 2. Undamped and Damped Electric Oscillations. When an alternating current of very high frequency exists in a circuit and continues uninterruptedly, it is usually called a persistent or undamped electric oscillation. It may be represented by a regularly repeated curve (see lowest curve, Fig. 4). We are, however, con- cerned in radiotelegraphy with a kind of alternating current of very Strongly Damped Oscillations Feebly Damped Oscillations Undamped Oscillations FlG. 4. high frequency which consists of separate groups of alternating electric currents, each group beginning with the same amplitude, but then tailing away or damping down more or less quickly to zero, and after an interval of rest beginning again. Such a current may be represented graphically by the upper and middle curves in Fig. 4, and is spoken of as a series of trains of damped electric oscillations. In this last case we have therefore to consider : (i.) The initial amplitude of each train. (ii.) The number of oscillations in each train. (iii.) The number of trains per second. (iv.) The rate at which the amplitude dies away in each train or the damping, as it is called. 6 RADIOTELEGRAPH 'Y As these damped electric oscillations are much used in radio- telegraphy, we shall consider more carefully some of their properties and the manner in which the above qualities are related to each other. Suppose a pendulum to have a hollow bob with a hole at the bottom, the said bob being filled with ink or with sand. If the pendulum is set in vibration over a sheet of paper which is moved with uniform speed transversely to the direction of the plane of oscillation of the pendulum, the outflowing ink or sand would describe upon the paper a wavy line, gradually decreasing in amplitude as the vibrations of the pendulum die away (see Fig. 5). The vibrations of the pen- dulum each take place in the \ same time that is, they are isochronous and the interval of time between two move- ments of the bob in the same y direction across its lowest position is called the periodic FIG. 5. time of the pendulum. If the pendulum is slightly displaced from its position of rest, the bob is raised, and the action of gravity creates a torque or restoring couple, tending to bring it back again to its lowest position. If this torque and the corresponding angular displacement are both measured in suitable units, the quotient of the torque by the angle of displacement is called the torque per unit angle. Again, if we suppose the whole mass of the pendulum divided into small portions, and the mass of each portion multiplied by the square of its distance from the axis of rotation, the sum of all such products for the whole pendulum is called its moment of inertia. It is shown in books on mechanics that the complete periodic time of a small vibration of a pendulum is obtained by dividing the square root of its moment of inertia by the square root of the torque per unit angle, and multiplying the quotient by the circular constant 2?r or by 6'283. Accordingly, we may say that the time of vibration is given by the rule, The complete periodic time of the pendulum Moment of inertia of pendulum Eestoring torque per unit angle of displacement If the amplitude of the vibrations does not exceed a few degrees, then the time of vibration is independent of the amplitude. If ELECTRIC OSCILLATIONS AND ELECTRIC RESONANCE 7 then we were to measure the amplitude of successive vibrations as they die away, we should find that each amplitude bears the same relation to the preceding one in magnitude. Thus, suppose the first or initial displacement or amplitude is represented by the number 100, and the second one by 90, then the third one would be fa of 90, or 81, and the fourth f - of 81, or 72'9, and so on. Thus the numbers representing the successive amplitudes or excursions of the bob would be 100, 90, 81, 72'9, 65*6, 59*04, 5314, 47-82, 43'04, 3874, 34'86, etc. These numbers are said to be in geometrical progression because each bears a constant ratio to the one preceding or following it in the series. If the logarithm of these numbers are written down, we obtain another series of numbers in arithmetic progression, successive terms having a constant difference. The reader is probably aware that there are two systems of logarithms in use, one to the base 10, which is employed in the construction of the ordinary slide rule, and the other called the Napierian system, to the base 271828. The Napierian logarithms are obtained from those to the base 10 by multiplying the latter by the modulus or factor 2-30259. Accordingly, if we take the Napierian logarithms of the above series of numbers representing the gradually decreasing amplitudes of the pendulum as the vibrations die away, we have the following table : Amplitude of successive excursions which exhibit a constant ratio of 10 : 9. Napierian logarithms of successive excursions which exhibit a constant difference 0'1053. 100-0 4-6052 900 4-4998 81-0 4-3945 72-9 4-2891 65-6 4-1838 59-04 4-0783 53-14 3-9730 47-82 3-8677 43-04 3-7622 38-74 3-6568 34-86 3-5514 etc. etc. The constant difference between the Napierian logarithms of the amplitudes of two successive swings is called the logarithmic decrement (abbreviated into log. dec.) of the motion. It is denoted by the symbol S. 8 RADIOTELEGRAPHY The reader should notice that we may speak of the log. dec- per complete period or the log. dec. per half period, according as the amplitudes plotted are the successive excursions of the pendulum in the same direction or in opposite directions. Thus in the above example the numbers 100, 90, etc., are the amplitudes of the successive swings in the same direction or to the same side. Hence the number 0'1053 is the log. dec. per complete period. If the oscillations of the pendulum are permitted to die away, strictly speaking, this should take an infinite time, because each succeeding oscillation has a definite ratio to that of the preceding one. As a matter of fact, however, there will come a time when friction only just permits the pendulum to return to its zero position from its last small excursion, and the oscillations will then have ceased. We may, however, fix a definite useful limit to the final amplitude, and say that when the last excursion is so much reduced that it is only 1 per cent, or yj^- of the initial amplitude, the oscillations have for all practical purposes ceased. We can then easily find how many oscillations will take place before this happens. It is obvious that between the 1st and 10th swing in the same direction there are 9 complete oscillations, and between the 1st and 100th swing there are 99 complete oscillations. Again, in any system of logarithms the logarithm of 1 is always zero. Hence if we divide the number 4*6052, which is the Napierian logarithm of 100 by the log. dec., the quotient will give us a number which is one less than the number of complete swings in which the amplitude has been reduced to 1 per cent, of the initial amplitude. Thus in the case of the pendulum above mentioned we have 4*6052 -f- 01053 = 43'7. Accordingly, in 44 to 45 complete swings the pendulum would be practically at rest, since the amplitude of its excursions would then have become reduced to 1 per cent, of the initial amplitude. The measurement of the logarithmic decrement enables us therefore to count the number of oscillations composing a train and therefore to say how fast they die away. A train of very few oscillations, say 5 or 6, is called a highly damped train, and a train of very many oscillations, say 100 or more, is called a feebly damped train (see Fig. 4). These facts with regard to mechanical vibrations have their analogues in electric oscillations. We can by special devices, explained in the next chapter, set the electricity in certain forms of circuit in motion by giving it a sudden impulse or release. It then oscillates to and fro in the circuit, and creates rapid alternating currents, which, however, continually decrease in their maximum value because' their energy is being dissipated by ELECTRIC OSCILLATIONS AND ELECTRIC RESONANCE 9 various causes such as resistance. Hence the oscillations are gradually damped out. We shall also consider in another chapter the manner in which this damping and the logarithmic decrement of the oscillations can be measured. 3. Electric Circuits and their Qualities. High Frequency Resistance. Before discussing the production of electric oscilla- tions we must refer to some of the qualities of electric circuits which are important in connection with high frequency currents. One of these is the effective resistance of the circuit. Electric resistance may be defined as the quality of an electric circuit in virtue of which the energy of an electric current existing in the circuit is dissipated as heat in the conductor. It may therefore be measured by the energy dissipated per second per unit current that is, by the power dissipated per unit current. The practical unit of current is the ampere, which is denned as the unvarying current which, when flowing through a neutral solution of nitrate of silver between silver electrodes, deposits on the negative electrode O'OOlllS gram of silver per second. The practical unit of power or work done per second is the watt. The practical unit of resist- ance (called the ohm) is therefore the resistance of a conductor which dissipates 1 watt as heat when a current of 1 ampere flows through it. The resistance of a conductor depends, however, upon the mode in which the current is distributed over the cross-section of it. Imagine a rod of copper of uniform section of any shape, and suppose it built up by laying together fine copper wires of square section placed parallel and closely packed. When a current flows through the rod we may picture to ourselves the current as uniformly divided between the small constituent wires, or we may suppose that these are insulated from each other and that some of the components carry more and others less current than the average, so that the total current is not distributed uniformly over the cross-section, but is denser in some places than in others. We can then very easily prove the following statement to be true. The resistance of a conductor for uniform distribution of the current over its cross-section is less than that for any non- uniform distribution of the same current. Let the large square in Fig. 6 be the section of the rod, and the small squares into which it is divided be the component elements. Let us suppose the total current to be equally dis- tributed over the cross-section as indicated by the uniform shading. Then the conductor has a certain resistance, and dissi- pates a certain energy per second per unit of current flowing through it. In the next place, let us suppose that current is removed from one of the little elements and added to that in 10 RADIOTELEGRAPHY $SSSSS5$! FIG. 6. another, so that, whilst the current in one component element or wire becomes zero, represented by the small white square, that in the other selected constituent, represented by the doubly shaded square is doubled. This is in effect making the current non-uniform over the total section, without altering the total current flowing. The heat produced per second in any conductor by Joule's law is proportional to the square of the current, so that if the current is doubled, the heat is quadrupled. Hence, if we consider the energy dissipated as heat in each element or filament of the whole number into which we have considered the conductor to be divided, we see that when the current is uniformly distributed it is uniformly or equally dissipated as heat, and if there are, say, 1000 elements or little component conductors, then the total heat produced is 1000 times that in one element. If, however, we assume that the current is taken away from one element and added to that in some other element, then, as far as regards these two elements, the heat produced per second in the first is now zero, since the current is zero, and in the second the heat is quadrupled, since the current is doubled. In all the other elements it remains at the original value. Accordingly, although the total current flowing through the whole conductor is the same as before, the total heat generated has been increased by rendering the distribution of current over the cross-section non-uniform. From which it follows at once that the production of heat per unit current flowing through the conductor is a minimum for uniform distribution of the current over the cross-section. One of the particular characteristics of an alternating electric current, especially of a high frequency current, is that it is not distributed uniformly over the cross-section of the conductor, as is the case with direct currents, but is concentrated in a surface layer of the conductor or principally confined to the skin. We may illustrate the difference between the two cases by a thermal analogy. Imagine an iron ball put into a furnace and left there for some time; it would become equally hot all through, and have the same temperature at the centre as at the surface. If, however, after being in the furnace for a short time, it is taken out and cooled, and then put into the furnace again, and these operations rapidly repeated, it would experience the changes of ELECTRIC OSCILLATIONS AND ELECTRIC RESONANCE u temperature only at the surface layers, and the interior would hardly change in temperature at all. To make clear the reason for this superficial concentration of the current on conductors when it is rapidly alternating, we must consider a little more closely the manner and meaning of estab- lishing a current in a conductor. For this purpose it is best to consider a simple case. Let there be two sheets of metal AB, CD (see Fig. 7), placed parallel to each other, and charged with electricities of opposite sign. These plates are shown in section in Fig. 7 and indicated by the thick black lines. This / ',,-'.";>, \ arrangement constitutes a condenser. ' The insulator or dielectric between the plates is in a peculiar state of strain along certain lines called lines of electric strain. In the diagram the directions of the strain at various parts is denoted by the dotted lines. The bodies we call conductors do not permit the creation in ^ ^+*"1''' - them of electric strain. If the strain in v ^>~\-''' a dielectric exceeds a certain value, the j? IG ~ 7t insulator is ruptured and a spark dis- charge takes place. We may compare a dielectric with an elastic wire, which can endure a certain twist before breaking, whereas a conductor is like a thread of honey or some such plastic body, on which we cannot put any twist at all, because it yields immediately to the strain. The so-called charge of the condenser is the energy of this electric strain in the dielectric, and each cubic centimetre of the insulator stores up a certain amount of the whole energy. For the sake of giving definiteness to our conceptions, we may think of the mass of the dielectric as divided up into closely com- pacted tubes, the sides of these tubes being bounded by lines in the direction of the electric strain. At the point where the tubes terminate on the conductors there is a positive or negative charge, and it is convenient to so select the size of these tubes that the charge on each end is a unit of electric quantity. We know that electricity can spread over the surface of conductors, but not over the surface of insulators. Accordingly, we must suppose that the ends of the tubes of electric strain are quite free to move over the surface of the conductors on which they abut. In books on the theory of electricity, it is shown that the state of electric strain in a dielectric is equivalent to a tension or pull along the direction of the tubes of electric strain and a pressure at 12 RADIOTELEGRAPHY right angles to them. The whole mass of the dielectric may be considered to be in a state of stretch and squeeze, and in fact the attraction between oppositely electrified bodies is only the pull exerted by the tubes of electric strain extending between them. Suppose, then, that the two oppositely electrified plates of the condenser in Fig. 7 are connected by a wire. The opposite ends of the tubes of electric strain move along it and approach each other, the tube shrinking up in the process in virtue of the ten- sion along it. The disappearance of the tubes nearest the wire relieves the lateral pressure on others lying outside, and they all in turn collapse in the same manner. Each tube, however, represents a certain amount of potential energy stored in the dielectric, and it cannot disappear without leaving an equivalent behind it in some other form. The movement of positive and negative charges of electricity along the wire in opposite directions involves, how- ever, the production of another effect in the dielectric, namely, a distribution of magnetic flux along closed curves embracing the wire constituting what we call the magnetic field due to electricity in motion. As, therefore, the energy of electric strain or the electrostatic field in the dielectric vanishes, due to the shrinking up of the tubes of strain, it is replaced by energy of magnetic flux or by a magnetic field distributed along endless lines enclosing the conductor. This magnetic flux begins to be created at the surface of the conductor where the tubes of electric strain are vanishing, and it spreads outwards into the dielectric and also soaks or penetrates into the conductor much more slowly. It cannot yet be said that we understand fully the mechanism by which this energy trans- formation is effected, or how it is that the lateral movement of the tubes of electric strain which stretch from plate to plate, causes them to transfer their energy to another form in which it exists in a state called magnetic flux distributed along closed lines round the discharging wire. In electromagnetic phenomena we recognise, however, that we are concerned with energy which may exist either in the form of electrostatic energy due to an electric strain in a dielectric produced by an electric charge, or with magnetic energy resulting from magnetic flux produced by an electric current. The electrostatic form of energy has a close similarity to the potential energy of ordinary mechanical strain or distortion, and magnetic energy to the kinetic energy of moving masses. Electro- static strain can exist in dielectrics, but not in conductors. Magnetic flux can exist both in dielectrics and in conductors, but the change of position of magnetic flux or movement of lines of ELECTRIC OSCILLATIONS AND ELECTRIC RESONANCE 13 magnetic flux through conductors causes a dissipation of some of the energy as heat. Eeturning, then, to the case of the condenser which is being discharged, it must be understood that the lines of magnetic flux which replace the lines or tubes of electric strain do not spring into existence simultaneously at all parts of the field, but originate at the surface of the discharging wire and spread outwards into space, and also penetrate much more slowly into the wire itself, generating heat in the latter as they move through it. If, then, the magnetic field outside the wire reaches its full and final state very quickly, the field inside the wire will have only penetrated by that time a very little distance into the metal. The speed with which the magnetic field outside the wire reaches its full development depends on the form of the circuit that is, upon the inductance, of the circuit. The magnetic field, however, does not remain permanent, but in turn begins to disappear, its lines contracting in again upon the wire. It can be shown that this process recreates the electrostatic field, but with electric strain directed in the opposite direction to that strain, the collapse of which gave rise to the magnetic field. We shall consider this process more in detail in Chapter IV., when discussing electromagnetic waves. Meanwhile it will be sufficient if the reader is able to grasp clearly the following ideas. Let the small thick line circle in Fig. 8 represent the section of the wire discharging the condenser, and let the black dots round it represent the section of the lines or tubes of electric strain, and the fine line circles the direction of the lines of magnetic flux. Then, before discharge, we have in the dielectric only an electrostatic field, and the dielectric is permeated by lines or tubes of electric strain. When the discharge wire con- nects the plates of the condenser the electro- static field begins to disappear, the lines of electric strain moving laterally in on the wire, but in so doing give up their energy to, and create closed lines of magnetic flux which move outwards from the wire. There comes an instant then when the electrostatic energy has all disappeared, the condenser is dis- charged, but a magnetic field has taken its place, and an electric current is said to flow in the discharge wire. This current then begins to die away that is, the magnetic field collapses in upon the wire but recreates as it disappears a fresh electrostatic field, the lines of magnetic flux fading away and being replaced by lines of electric strain in the opposite direction to the original strain. 14 RADIOTELEGRAPH? If those transformations of energy succeed each other very rapidly in the space outside the discharge wire, then the magnetic field, which is more slowly soaking into the conductor, will never have time to penetrate far into it, and both the interior magnetic field, and therefore the heat evolution, will be confined to a thin outer layer of the material or to a mere skin. The better the conductivity of the material of which the discharge wire is made, the more slowly does the magnetic field travel into it, and there- fore the thinner the skin. These matters cannot be explained in full detail without the use of mathematical reasoning of an advanced character, but the above brief statement is perhaps sufficient to make it clear that when we are dealing with electric currents which change their direction very rapidly in other words, with high frequency alternating electric currents the current in the wire is not distributed uniformly over the cross-section of the wire, but is confined to a thin surface layer or skin. Hence, in accordance with what we have proved above, the resistance of the conductor for such alternating currents is not the same as that for steady or continuous currents, but is considerably greater. The usual tables for the resistance of copper and other wires give us what is called the ohmic or steady resistance, or resistance to continuous currents, of wires of various diameters and lengths. These values are not usually applicable in the case of conductors used in radiotelegraphy. We have, then, to consider the high frequency resistance of the wire. In the case of steady or direct currents, the resistance of a wire of given material, length, and cross-section varies directly as the length, inversely as the section, and directly as the resistivity or specific resistance of the material. These resistivities are given in books of reference either in absolute units or else in microhms or millionths of an ohm per centimetre cube. Thus, in the case of copper, the electric resistivity is 1*6 microhms per centimetre cube at ordinary temperature. If, then, we have a wire of 1 square millimetre in cross-section, or O'Ol square centimetre and 100 centimetres long, its steady resistance is 1*6 X 100 -f- 0*01 = 0-016 ohm. The ratio of the high frequency to the steady resistance has been investigated by many physicists. Lord Kayleigh has given two formulae which enable us to calculate the resistance of a straight conductor for high frequency currents applicable in cases of moderate and very high frequency respectively, to straight or but slightly curved wires of any material. We have transformed his formulas to meet the case of straight, round sectioned wires, as these are nearly always used in practice. Also, we have ELECTRIC OSCILLATIONS AND ELECTRIC RESONANCE 15 considered the wire to be of a non-magnetic material, say copper, as these are mostly employed in practice. Let the letter c denote the circumference of the round wire measured in centimetres, n the frequency of the current, and p the steady current resistivity of the material of which the wire is made, reckoned in absolute units on the centimetre, gramme, second system of measurement. Thus for a copper wire p = 1600 at ordinary temperatures. We then calculate the value of the quantity nc 2 /p = k. Thus if n = 1000, e = 1, and p = 1600, we have k = % . Let R' be the high frequency resistance of the wire and R its ordinary or steady resistance. Then, provided that 7c is TW less than unity, we can calculate the value of the ratio from R the formula R' &2 7,4 R = " 48 ~ 2880 Thus for a copper wire of 1 centimetre in circumference or about J- inch diameter, and for a frequency of 1000 ~, we have |f = 1 + li* - TF874 = 1*008 nearly Accordingly, the increase in resistance for this wire and frequency is rather less than 1 per cent. If the frequency and diameter of the wire are such as to make the value of the quantity k very much greater than unity, say at least equal to 6, then Lord Rayleigh has given another formula which, as modified by Dr. A. Russell, is equivalent to Thus, suppose the frequency n = 1,000,000 and the wire is a copper wire of one centimetre in circumference, then p = 1600, c = 1, and n = 10 6 . Hence k = u$p, and ^//T^ 2 5, and ?V^= 12'5 Therefore the resistance of this wire for currents of 1,000,000 per second would be 12^ times greater than its ordinary or steady resistance. Suppose the wire to be of one-tenth the diameter, or O'l centimetre in circumference, equivalent to about ^ of an inch in diameter. Then for n = 1,000,000, p -- 1600, and c = O'l, we have & = JA = 6, and \/k = 2-5 16 RADIOTELEGRAPHY Therefore ~ - 1'25, or the high frequency resistance of the wire is approximately 25 per cent, greater than its steady resistance. If, however, we consider the case of a copper wire still smaller, say of O01 centimetre in circumference, then for a frequency of 1,000,000 this would make k = -f%, and the last formula is then no longer applicable, since k is now less than unity, and we must revert to the previous formula. Accordingly for this last case we have f = 1 + Ad's) 2 - WTO (A) 4 = 1 + 12*00 Hence for so small a wire the resistance is not perceptibly increased even when using frequencies of a million and upwards. The conclusion from the above statements is that, when using rather thick copper wires to convey high frequency currents, the effective resistance may be very much greater than it is for con- tinuous or low frequency currents. On the other hand, if the wire is of very small diameter, then even for high frequency currents this increase in resistance due to the concentration of the current at the surface of the wire is not very serious. Accord- ingly, conductors for conveying high frequency currents should not be solid metal thick wires but stranded conductors, made by bunching or twisting together very numerous fine silk or copper covered wires not thicker than No. 32 or No. 36 S.W.G. in size. There is, however, another point in connection with this matter of equal importance. When a wire is coiled into a helix of many close turns it is found that its high frequency resistance is considerably greater than that of the same wire stretched out straight, even as calculated by Lord Eayleigh's formula. The reason for this is as follows : We have already shown that anything which causes a non- uniformity of distribution of current over the cross-section of the wire causes an increase in resistance. In the case of a straight wire we have also shown that high frequency currents concentrate themselves near the surface of the wire. If the wire is straight, this surface layer of current is uniformly thick all round the section of the wire. If, however, the wire is coiled into a spiral, the current is not merely concentrated in a surface layer, but is furthermore concentrated on the inner side of each turn, so that the surface layer of current is of different thicknesses at different parts of the circumference (see Fig. 9). From this it follows that we have an increase in the resistance to high frequency currents as com- pared with the resistance of the same wire when stretched out ELECTRIC OSCILLATIONS AND ELECTRIC RESONANCE 17 straight. This effect has been investigated experimentally by Dolezalek, Battelli, and Magri, and by T. B. Black, and theoreti- cally by Wien, Sommerfeld, and by L. Cohen. Battelli and Magri, as well as Black, placed equal lengths of two similar wires in the glass bulbs of two similar air thermometers, one of the wires being straight and the other tightly coiled. A high frequency current or train of oscillations was then sent through the two wires in series, and the amount of heat generated in the two cases measured by the increase of air pressure created in the bulbs. Using two sizes of wires about 0*15 and 0*3 cm. in diameter, and oscillation frequencies FIG. 9. of 1,000,000 and 5,000,000 respectively, Black found that the resistance of the wire in the form of a spiral was to that of the same wire stretched out straight in some ratio between 1*25 and 1*89. Hence we may say that the effective resist- ance of a wire for the above frequencies may be increased from 25 to 90 per cent, by coiling it in a helix. This is over and above the increase in resistance which results from the surface disposition of the current as compared with the resistance to steady currents. Wien and Som-merfeld have investigated the matter theoretically. Sommerfeld considered the case of a tightly wound spiral or solenoid of one layer made up of square- sectioned wire with extremely thin insulation, and he deduced from this an expression for the ratio of the high frequency resistance E" of a spiral of one layer of circular-sectioned wire to the resistance to continuous currents E which is equivalent to where Ic has the same meaning as in the Eayleigh formula given above. The constant 2\/7r = 3'54 and ^\/& is the value of the ratio of the high frequency resistance E' of the wire stretched out straight to the steady current resistance E. Black's experiments show, however, that the actual ratio is not much more than half the above value. L. Cohen has also given a formula for the ratio of the resistance of a helix to that of the same wire when stretched out straight when the frequency is of the order of 10 6 and the diameter of the wire (d) not extremely small, say 1 mm., which is equivalent to i8 RADIOTELEGRAPHY where N is the number of turns per unit length of the helix. This formula, however, appears to give ratios which are greater than those furnished by experiment. ~R" Black found that for very high frequencies the ratio p-, viz. that of the helix to that of the same wire stretched out straight, was independent of the frequency when the frequency was very high, but determined by the number of turns per unit of length and by the diameter of the wire, and also that it was also deter- mined by the ratio of length to mean diameter of the helix, at least up to a ratio of 10, beyond which, however, all spirals behave like infinite spirals. None of the formulae so far given by theory appear, therefore, to be capable of predicting very accurately the high frequency resistance of such solenoids or coils as are used in radiotele- graphy. It is, however, important to bear in mind that for wires of the diameter of 1, 2, or 3 mm., and for frequencies of 10 or so, the resistance of the wire when stretched out straight will be about four, eight, or twelve times that of the steady or ordinary resist- ance, and that when coiled into spirals of eight or ten turns per inch this augmented resistance may be again increased by 50 to 100 per cent., that it may become half as much again, or double that which it is for the same wire subjected to the same oscillations, but stretched out straight. It is, therefore, difficult to predict the power dissipated as heat by electric oscillations in spirals made of wire not very thin. 4. High Frequency Inductance. The inductance of a circuit is that quality of it in virtue of which energy is conserved or associated with the circuit when a current flows through it. There is a very close analogy between the dynamical quantity called the mass of a body and the inductance of a circuit, and it may be that the so-called mass of matter is in fact due to the same ultimate qualities which create inductance. If an ordinary body is in motion without rotation, with a certain linear velocity v y then it possesses a certain store of kinetic energy. This energy is measured by the product of half the square of the velocity and a certain constant, M, called the mass of the body, or by M> 2 . In the same manner, if an electric current i is circulating in a circuit it involves a certain storage of energy which is proportional to half the square of the current and to a certain constant, L, called the inductance of this circuit, or to . Again, if a heavy body is in rotation round an axis it has a ELECTRIC OSCILLATIONS AND ELECTRIC RESONANCE 19 certain angular velocity, w, at any instant, and a certain angular energy. This last is proportional to half the square of the angular velocity, and to a coefficient, K, called the moment of inertia, or to Kw a . Furthermore, if the linear or angular energy and velocity of a body are changing, the rate at which the energy changes with the velocity is called the linear or angular momentum (M0 or Ko>), and if the momentum is changing with time, the time rate of change of the linear or angular momentum is called the force or torque acting on the body. In the case of an electric circuit, if the energy stored up is changing with the current, -then its rate of change (Li) is called the electrokinetic momentum of the circuit. If this latter is changing with the time, the time rate of change of the electrokinetic momentum is the electromotive force acting in the circuit due to inductance. Having regard to the fundamental discovery of Faraday that a change in the number of lines of force linked with or passing through a circuit gives rise to and is a measure of the electromotive induced in it, we easily see that the electrokinetic momentum Li is merely another name for the total magnetic flux due to the current in a circuit which is self-linked with the circuit. The quantity called the inductance may, therefore, be con- sidered to be a coefficient or number which denotes the number of lines of magnetic flux which are self-linked with a circuit when unit current flows through it. If the unit of current is the ampere, then the corresponding unit of inductance is called the henry, and a circuit having an inductance of 1 henry when traversed by a current of 1 ampere produces a total self-linked magnetic flux which is generally called 1 weber. If this flux is removed uniformly from the circuit in 1 second it would generate in it an electromotive force of 1 volt whilst the removal lasts. The inductances of such circuits as we have to deal with in radiotelegraphy are generally small, and are conveniently measured in millihenrys or microTienrys, these being respectively one-thousandth and one-millionth of a henry. The above statements may be perhaps made clearer by con- sidering a concrete example. Imagine a wire bent into a circle to have some source of electromotive force, say a battery, inserted in it of such magnitude as to make a current of 1 ampere circulate round it. The current will create a magnetic flux which is linked with its own circuit, and if the inductance of this circular wire is 1 henry, then a total flux of 1 weber will be self-linked with the circuit. A flux of 1 weber is equal to 100 million " lines of force " in the usual mode 20 RADIOTELEGRAPHY of reckoning. Let the battery be short circuited, then the current in the circuit commences to die away, and in so doing removes magnetic flux from the circuit. This action creates an induced electromotive force in the circuit acting in the same direction as the battery. If the flux is removed uniformly in 1 second, and if the constant induced electromotive force during that time is equal to 1 volt, then the total initial self-linked flux has a value called 1 weber, and the inductance of the circuit is 1 henry. The in- ductance of a circuit therefore depends on its geometrical form, and is determined by the amount of linkage of flux that takes place with itself when 1 ampere flows in the circuit. The in- ductance is therefore increased by closely coiling the circuit. Also it is determined by the mode in which the current is dis- tributed over the cross-section of the conductor and is therefore not the same for steady currents uniformly distributed as for high frequency currents which are concentrated at the surface of the conductor. The actual calculation of the inductance of any con- ductor is impossible, or very difficult, except in a few particular cases in which the circuit is of a simple form, such as a straight wire with return at an infinite distance, a circular, square, rect- angular, or spiral circuit. The principle on which such pre- determination proceeds is that of calculating the work done in dividing up the current into a series of elementary filaments of current, and removing them all to an infinite distance from each other. We may consider the whole current to be divided into a number of parallel streamlets or filaments of current in the con- ductor. Now, parallel currents moving in the same direction attract each other. Hence, if we assume that these filaments of current can be separated from each other to an infinite distance, as if we were separating a rope into its constituent strands, the work so done in taking the current to pieces must be equal to the energy it possesses as a whole before separation. This last is equal to ^Li 2 , where i is the whole current and L the induct- ance of the circuit. The quantity L is therefore sometimes called the potential of the circuit on itself. It was shown by Neumann that the potential of a circuit on itself is obtained by summing up all the products obtained by multiplying together the lengths of all possible pairs of small elements of the circuit multiplied by the cosine of the angle between their direction, and divided by their distance apart, regard being taken in this process of the distribution of the current over the cross-section. The actual calculation is even in the simplest cases somewhat difficult, and for details the reader must be referred to larger treatises, but we shall here give the formulae for some cases often required in ELECTRIC OSCILLATIONS AND ELECTRIC RESONANCE 21 practice. These expressions, with the exception of the formula (6) for the inductance of a helix of one layer, are the high frequency values applicable in cases in which the current is confined simply to a surface layer of the conductor. In these formulae we suppose the circuit to consist of a circular- sectioned wire of diameter d centimetres and of length I centi- metres, made of a non-magnetic material, the diameter being small compared with the length. The inductance L in these f ormulse is expressed in microhenrys, and this can be reduced to centimetres or absolute electromagnetic units by multiplication by 1000. (1) Inductance of a straight wire of length I centimetres. L= 2l I 2-3026 logio-j - 1 j microhenrys The above formula is applicable in calculating the inductance of a single wire antenna of copper or other non-magnetic wire when we are concerned with high frequency oscillations in it. (2) Inductance of a square circuit formed of circular -sectioned wire. Length of side of square = S centimetres. 1000 (2-3026 logio ^ - 2-853 j microhenrys (3) Inductance of a circular circuit formed of circular -sectioned wire. Mean diameter of the circle = D centimetres. 10 ~ -- 2*45 microhenrys The above formulae (2) and (3) are applicable for calculating the high frequency inductance of a square and circular circuit respectively. (4) Inductance of two parallel wires D centimetres apart, length of each wire being I centimetres. 4:1 C 2T)"l L = iooo I 2 ' 3026 loSl TJ micronen3 7 s (5) Inductance of a rectangular circuit of circular-sectioned wire. TJic sides of the rectangle are respectively A and B centimetres, and the diagonal D centimetres. 22 RADIOTELEGRAPHY 9-2104 (, 4AB L = "looo" ( (A + B) loglo ~7T " A loglo(A + D) - B logio(B + D) - - 7~ : microhenry 3 (6) Inductance of a helix of one single layer of circular-sectioned wire. N = number of turns per centimetre of length of the helix. D = diameter of helix in centimetres. / = length of helix in centimetres. L . 0'0156.(y) \ microhenry s In making actual concrete standards of inductance the formulae (2), (5), and (6) will be found very useful. A ver^Jairlj accurate standard of inductance may be made as follows : Chuck in a lathe a rod of ebonite or hard dry wood and turn it truly cylindrical. Then with the screw-cutting gear cut a screw groove in it of S or 10 turns to the inch, and wind up in this groove bare copper wire of No. 14 or No. 16 S.W.G. size. The length of the helix may be from 10 to 50 times its diameter. The ends of the copper wire should be secured to terminals. Then measure the mean diameter of the helix of wire, taking into account the thickness of the wire. This gives the value of D in the above formula. The length of the helix (I) is then measured and the number of turns (N) per unit of length, and these quantities inserted in the formula (6) will give the inductance of the helix in microhenry s, each of which is 1000 absolute electromagnetic units. Since the ratio y appears in the formula (6), we may measure the diameter and length of the spiral in inches if more convenient, and also count the turns per inch. It will then be seen that the quantity (TrDN) which appears in the formula is the length of wire wound on one unit of length of the helix, and is the same number whether D is measured in inches and N in turns per inch, or D is measured in centimetres and N in turns per centimetre. Hence, if we adopt the inch as the unit of length, the tfnly change we have to make is to insert for the value of I in the formula (6), where it appears alone in front of the bracket, the length of the helix measured in centimetres. ELECTRIC OSCILLATIONS AND ELECTRIC RESONANCE 23 The above formula is due to Dr. A. Kussell, and is of considerable practical value when using currents of low or moderate frequency, but when high frequency currents are employed, the actual in- ductance of the spiral will be slightly less than that given by the formula by an amount which can be predetermined theoretically by a formula due to L. Cohen. It is necessary always to bear in mind that the inductance of a coil or wire depends to some extent upon the. manner in which the current is distributed over its cross-section. The resistance of a wire,, we have shown, is a minimum for uniform distribution of current over the cross-section and is increased by any cause tending to make it non-uniform. On the other hand, the inductance is a maximum for uniform distribution over the cross-section, and is diminished by any cause tending to make it non-uniform. Hence it is less for high frequency currents than for steady currents. Also, in the case of coiled wires, the increase in frequency has a greater effect in diminishing the inductance than is the case with straight wires. Dolezalek has suggested constructing standards of inductance of bunched or stranded insulated wires, each wire being not more than 0*1 mm. diameter, or, say, No. 40 S.W.G., to compel uniformity in the current distribution over the section. 5. Electrical Capacity. If we communicate a charge of positive electricity to a conductor, we raise its potential, that is, we increase the quantity of work required to be done to transfer a unit of positive electricity carried on a small conductor from the earth to the surface of the conductor in question. The charge which must be given to a conductor to raise its potential by one unit is a measure of its electrical capacity. If this capacity is denoted by the letter C, and the potential of the conductor by V, then the quantity of electricity on it is Q, where Q = CV. The practical unit of electric quantity is the coulomb, and the practical unit of potential is the volt, and the practical unit of capacity is the farad. Hence, to charge a capacity of 1 farad to a potential of 1 volt, we must place on it a charge of 1 coulomb of electricity. Faraday showed that it is impossible to create a charge of electricity of one kind only. Hence, if any conductor is charged with 1 coulomb of positive electricity, there must be an equal charge of negative electricity on some other conductor or conductors, or on the earth. A Condenser is any arrangement of two conductors on one of which a charge of positive electricity can be placed, and on the other an equal quantity of negative. The magnitude of either quantity is called the charge of the condenser. The potential difference (P.D.) between the conductors is called the terminal 24 RADIOTELEGRAPHY P.D. of the condenser. The quotient of charge by terminal P.D is called the capacity of the condenser. In radiotelegraphy the capacities with which we are concerned are generally very small and best measured in microfarads (mfds), that is, in million ths of a farad, or in micro-microfarads (mmfds), that is, in millionths of millionths, or billionths of a farad. In the case of certain conductors of symmetrical form we can pre- determine the capacity from the geometrical form and dimensions. Thus, for instance, consider the case of a conducting sphere placed in free space at a great distance from all other conductors. If we place on this sphere a charge of electricity, it will distribute itself equally. It is a fundamental principle that all parts of a conductor must be at the same potential, and it can be shown from first principles that the potential in electrostatic units at any point at a distance x from a charge of positive electricity q, also reckoned in electrostatic units, collected on a very small sphere, is - x Hence, in the case of the charged sphere the charge on its surface may be divided into small elements of charge which are all equidistant from its centre. Hence, if the radius of the sphere is E and its whole charge is Q, its potential is g But its capacity C is equal to the quotient of charge by potential. Hence its capacity is numerically equal to its radius when reckoned in electrostatic units. The electrostatic unit of potential is equal to 300 volts. The electrostatic unit of quantity is such that 3000 million electrostatic units are equal to 1 coulomb. Accordingly 900,000 million electrostatic units of capacity must be equal to 1 farad, or 900,000 to 1 microfarad, or 0*9 of a unit to 1 micro-microfarad. Accordingly, we have the following rule : To reduce capacities reckoned in electrostatic units to their equivalent reckoned in microfarads, divide "by 900,000. Hence if a sphere has a radius K centimetres, its electrostatic capacity is E electrostatic units and its capacity in microfarads is E 900000* For convenience of reference we give below the formulae for the capacity of conductor's or condensers of various forms expressed in microfarads (mfds). (1) The capacity of a sphere of radius It centimetres when at a distance from all other conductors and from the earth. ELECTRIC OSCILLATIONS AND ELECTRIC RESONANCE 25 C= 900FOO mfds - (2) The capacity of a thin, flat circular disc of diameter D centimetres when at a distance from all other conductors and from the earth. D D mfds. 3-1415 x 900000 2827431 (3) The capacity of a condenser formed of two flat plates each of area A square centimetres placed parallel to each other at a distance d centimetres, d being small compared with C= " 4-rrd x 900000 " 11309724 x d^ (4) The capacity of a condenser formed of two thin concentric tubes I centimetres long and. of diameters DI and D& centimetres respectively. ^ v , C = -- ~ - mfds. 4-6052 Iog 10 = X 900000 (5) The capacity of a vertical wire of length I centimetres and diameter d centimetres at a considerable distance from the earth and from all other conductors. C = ~~97~~ ~ mfds - 4-6052 logio^-x 900000 it (6) A horizontal wire of length I centimetres and diameter d centimetres stretched parallel to the earth's surface at a distance h centimetres above it. C = - -1- - mfds. 4-6052 log 10 ^X 900000 Concerning the above formulae a few remarks are necessary. In the first place the assumption made in all cases is that the dielectric medium surrounding or between the conductors is air or some other material of unit dielectric constant. If, for instance, in the formulae (3) and (4) the dielectric between the plates or cylinders is glass, ebonite, oil, or other material, the formula must 26 RADIOTELEGRAPHY have the constant K prefixed as a multiplier, K denoting the dielectric constant of the material as taken from the following table : DIELECTRIC CONSTANTS OF VARIOUS INSULATORS FOR AIR K = 1. Dielectric constant Dielectric. Katl5C. Glass, flint 6-57 to 10-1 Glass, crown 6'96 Ebonite 2-05 to 3'15 Indiambber, pure 2-12 Indiarubber, vulcanised .... 2-69 Mica 6-64 Sulphur 2-9 to 4-0 Shellac 2-7 to 3'0 Paraffin oil 2-00 Turpentine 2-23 Benzol 2-38 The values vary a good deal for different specimens of materials of the same name, especially in the case of solids of complex composition. In the next place it should be noted that the formulae for the plate condenser (3) and cylinder condenser (4) have been obtained on the supposition that the lines of electric strain stretching between the conductors are straight lines. As a matter of fact, at the edges they are curved and there is a correction for the fringe of lines of force at the edges which has been omitted. It is not large when the plates are near together. In the third place the formulae for the capacity of the sphere (1), the disc (2), and the vertical wire (5) have been obtained on the supposition that they are removed a long way from the earth. In the case of actual conductors of this kind, the measured capacity would be found to be greater than that given by the above formulae. The manner in which this capacity can be experimentally obtained will be described in a later chapter. In those cases in which a condenser has to be constructed of definite and measurable capacity for radio telegraphic purposes the best type is a condenser consisting of flat metal plates (sheet tin) immersed in paraffin oil. If two such plates are placed in oil separated by a distance very small compared with their linear dimensions, the capacity can very approximately be calculated by the formula (3) and the dielectric constant may be taken as equal to 2*0, and a condenser of predetermined capacity thus made. For other purposes a collection of vertical or partly vertical and partly horizontal wires are used, and these form with the ELECTRIC OSCILLATIONS AND ELECTRIC RESONANCE 27 earth a condenser of definite and measurable capacity. For many purposes in radiotelegraphy the familiar Ley den jar, consisting of a glass bottle or vessel, coated with tin foil on part of its interior and exterior surfaces, is a convenient form of condenser, but one which is rather bulky in comparison with its electrical capacity. 6. The Time-period of Oscillatory Electric Circuits. If we join in series with each other some form of inductive circuit and some form of condenser, we have a circuit which is called an oscillatory electric circuit, and such circuits are broadly divided into two classes, respectively called open and closed oscillatory circuits. Thus, suppose we bend a thin wire into the form of a square and connect the two ends to two metallic plates placed very near to each other (see Fig. 10), we have a circuit which is called a closed oscillatory circuit. It possesses inductance owing to the size and form of the square circuit and capacity in virtue of the proximity of the two plates which form a condenser. If, however, we affix a metal sphere or disc to the end of a long vertical wire which has its lower end in close proximity to but not touching the earth (see Fig. 11), we have an open oscillatory FIG. 10. circuit which possesses inductance in the wire and capacity since the sphere or disc and the wire itself forms with the earth the two surfaces of a condenser. Between these extreme types of oscillatory circuit we can have many others, all characterised by possessing inductance and capacity in series. If the electric charge in such a circuit is disturbed in any way, as by introducing 28 RADIOTELEGRAPHY into the circuit a sudden electromotive force, this charge oscillates backwards and forwards with a definite and constant time period, just as a pendulum when disturbed and left to itself executes mechanical vibrations. It will be necessary to consider a little more in detail how these oscillations are produced. Let us suppose the condenser has a capacity of C farads and that the circuit connecting its plates has an inductance L henrys and a resistance of E ohms. Then let the condenser plates be charged until they have a potential difference of V volts. Suppose that the circuit is now completed through the inductance, the condenser commences to discharge. Electricity passes in the form of a current round the circuit, and as the charge in the condenser diminishes the current in the circuit increases and the energy is transformed from electrostatic to electrokinetic form. At the outset the energy imparted to the condenser was equal to JCV 2 joules, and when the discharge is complete the result is that we have a current in the circuit say of A amperes, and therefore an electrokinetic energy JLA 2 joules associated with it, and also some portion of the original energy has been transformed into heat in consequence of the resistance of the circuit. This current energy, as already explained, expends itself again partly in reproducing electrostatic charge, and the transformation repeats itself again and again until the whole original energy is dissipated as heat in the circuit or by any other dissipative action. These electrical operations may be compared with the similar energy changes which take place in the case of mechanical vibrations. Imagine a heavy mass M hung up by a spiral spring. The spring can be stretched, and to stretch it work must be done on it and energy of strain stored up in it. If then the heavy bob is pulled down and released, the energy of strain stored up in the stretched spring expends itself in creating motion in the heavy mass. When the spring has returned to its original dimensions it has expended this energy in imparting a kinetic energy -JM^ 2 to the bob, with the exception of that small part which may have been expended in making eddies in the air or in heating the metal of the spring. This kinetic energy, however, expends itself in turn in compressing the spring and then is again transformed into the form of energy of strain, and the process repeats itself again and again until the original store of energy is frittered away and the bob comes to rest. Hence, just as the original energy represented by the stretch of the spring is transformed into energy of motion of the bob, and then back again, so the energy of electric strain originally created in the condenser expends itself in creating an electric current in the ELECTRIC OSCILLATIONS AND ELECTRIC RESONANCE 29 inductive circuit, and then this in turn re-creates the energy of electric strain, and the process repeats itself until the energy is frittered away into heat or removed by radiation. In the one case we have mechanical oscillations, consisting in the up-and- down motion of the heavy mass, and in the other electric oscilla- tions consisting in a high frequency alternating current in the inductive circuit. Suppose, then, that at any instant during the discharge of the condenser the current in the inductive circuit is a amperes, and at that moment the charge of the condenser is q coulombs, the maximum or original charge being Q coulombs. At that instant the potential difference of the condenser plates must then be j volts. Again the fall in potential down the inductive circuit is due partly to its resistance K and partly to its inductance L. The part due to resistance is measured by the product Ea in accordance with Ohm's law, and the part due to inductance is measured by the rate at which La is changing with time, which may be denoted by La. The sum of all these differences of potential round the circuit must be zero, Hence we have La+Ra + = \j The current in the circuit at any moment is measured by the rate at which the condenser is gaining or losing charge. Therefore we have with the above notation q = a, and hence the rate at which a is changing is the rate at which q is changing, which may be written q = a. Substituting these values for a and a in the above equation, we have, L jf + E q + 1 = This equation establishes a relation between the charge q in the condenser at any moment, and its rate of change q, and its rate of rate of change q, and it is called a differential equation. In all oscillatory circuits which present themselves in radiotelegraphy the high frequency resistance E measured in ohms, the inductance L measured in henrys and the capacity C measured in farads have T> such values that the second term .=- q may be neglected in com- L parison with tTl cos Pt But -^ = q or the time rate of change of ON or q. If, then, we draw a line OT at right angles to OP and make OT = p . OP = pQ, it is easy to see that its projection OZ on Y is equal to ^>Q cos_p, and hence the length of OZ is a measure of the velocity of the point N or of q. If we repeat this process, and draw a line OU at right angles to OT and p times as long or equal to jo 2 Q, then its projection OW on OY is equal to ^? 2 Q sin pt, and by similar ELECTRIC OSCILLATIONS AND ELECTRIC RESONANCE 31 reasoning it can be shown that the length OW is a measure of the acceleration of N or of q. If, then, the point N executes a simple periodic motion to and fro along YY, so that its displacement at any instant is ON = q, its acceleration q at that instant is equal to p 2 times ON, or q = p 2 q. Eeturning then to our electrical oscillation, we may make the supposition that the charge q of the condenser varies periodically from Q to zero, and from zero to Q in accordance with a simple periodic law, so that q = Q sin pt where p = 27r-times the periodic time T. We have already shown that for the oscillatory electric circuit Also that for any harmonic motion q + p*q = Q It follows that and since p = 7^-, we have for the periodic time T the expression T = 27rVEC The electric oscillations, therefore, are executed with a frequency n = , and 2WIC These two last formulae accordingly give us expressions for the time-period of the oscillations and their frequency, where T is measured in seconds, L in henrys, and C in farads. As, however, the capacity is most conveniently measured in microfarads and the inductance in microhenrys, the time-period will be expressed in microseconds, or in millionths of a second. This time-period is called the natural time-period of oscillation of the circuit. Just as every pendulum has its own natural time- period of oscillation, in which it executes vibrations when disturbed and left to itself, so every oscillatory electric circuit consisting of a capacity and inductance in series with each other has its own natural time-period in which an electric charge given to it oscillates if disturbed and left to itself. 32 RADIOTELEGRAPH? We may put the formula for the natural frequency of an oscillatory circuit in a more convenient form for calculation. Let L cms. stand for the inductance of the circuit expressed in absolute electromagnetic units, that is in centimetres. This is equal to the inductance in microhenry s multiplied by 1000. Let C mfds. stand for the capacity measured in microfarads. Then it is easy to show that 5-033 x 10 6 n = ,- V/C mfds. . L cms. 1000 The number 5 '033 is equal to 27T Thus, suppose that the circuit consists of a Leyden jar having a capacity of -3 Jo tn f a microfarad, and the inductive circuit has an inductance of 30 microhenrys or 30,000 centimetres. Then the quantity \/0 mfds. . L cms. has a value of 10. This last quantity is called the oscillation constant of the circuit. The natural frequency of such a circuit is then half a million, and the natural time-period of oscillation is two microseconds. When two oscillatory circuits have the same natural time-period they are called syntonic circuits. This is the case when the product of the capacity and the inductance of the two circuits is the same, although the individual values may be very different. 7. Electric Eesonance. If an oscillatory circuit has a periodic or alternating electromotive force set up in it and if the frequency of this E.M.F. agrees with the natural frequency of the circuit, then an immensely greater current will be produced than if the periods do not agree. This increase in the amplitude of the alternating current created in the circuit by exactly syntonising the frequency of the impressed E.M.F. with the natural frequency is said to be due to electric resonance. The term is borrowed from acoustics. It is a familiar fact that we can set up a very considerable amplitude of vibration in a pendulum by administering to it small blows, or even puffs of air, provided that these are timed to agree exactly with the natural period of the pendulum. In the same manner oscillations of great amplitude can be created in a heavy elastic beam supported at both ends by very gentle blows given at the right intervals in the centre. Supposing that an electric condenser has its plates connected by a wire, part of which forms a coil, and that over this last coil we wind another wire in which a high frequency current can be set up. Then alternating currents in this coil, called the primary coil, generate by Faradaic induction a secondary electromotive force in ELECTRIC OSCILLA TIONS AND ELECTRIC RESONANCE 33 the other coil, and theiefore a secondary cuirent. The secondary current will be small unless the two circuits are syntonised, but it increases very rapidly as the circuits are brought into syntony with each other. This is called tuning the circuits. Two such overlaid or neighbouring coils constitute an oscillation transformer, and the two circuits are said to be covpled together. If the primary and secondary coils are very close together, or intertwined, the circuits are said to be closely coupled. If they are far apart, they are said to be loosely coupled. If both circuits possess inductance and capacity, they are called coupled oscillatory circuits, and they can be tuned by varying the capacity or inductance of one or both until the oscillation constants of each circuit are either equal or in exact integer ratio to each other. If oscillations are set up in a circuit which are in agreement with its natural frequency, they are called free oscillations. If, however, oscillations are maintained which have a frequency different from the natural frequency of circuit, they are called forced oscillations. Suppose, then, that we form an oscillatory circuit consisting of a condenser, C, the capacity of which can be varied gradually, and an inductance coil, P, which forms one coil of an oscillation transformer, and insert in the circuit some instrument, such as a hot wire ammeter, A, adapted for measuring high frequency currents (see Fig. 13). Let the other circuits of the oscillation transformer have undamped oscillations set up in it of any frequency. If then we take obser- vations of the reading of the ammeter, beginning with a very small capacity in the circuit, and steadily increase the capacity, we shall find the current as read on the ammeter increases with the capacity up to a certain capacity, and after that decreases again. If we plot these ammeter readings as ordinates of a curve drawn to capacity values as abscissae, \ AAAAA/V we shall obtain a curve rising up to a peak ^\ 5 more or less sharply, corresponding to the maximum observed value of the current, and to a certain value of the capacity (see Fig. 14). This maximum current is called FlG - 13< the resonance current, and the value of the natural frequency of the circuit corresponding to the value of the capacity which produces this maximum current is called the D ./ 34 RADIO TELEGRA PHY CAPACITY. FIG. 14. resonance frequency, whilst the curve itself so drawn is called a resonance curve. Such resonance curves can be used, as we shall show in Chapter VIII., for de- termining the logarithmic de- crement or damping of the oscillatory circuit for which they are described. If the resonance curve is very sharply peaked, it shows that a very little want of tuning of the oscillatory circuit to the exact frequency of the im- pressed electromotive force has a very great effect in reducing the current induced in the oscillation circuit. This implies that exact tuning has a great effect in exalting the current. This again shows that the damping or sources of energy loss in that oscillation circuit are small, and hence the loga- rithmic decrement is small. On the other hand, a resonance curve with a rounded summit, as in Fig. 15, implies that the departure from absolute CAPACITY. syntony has not much FlG - 15 effect in reducing the current, or, vice versa, that exact tuning has not much effect in exalting it, and this means that the circuit in question has a large decrement. Under some circumstances the resonance ourve is a double-humped curve, as in Fig. 16, which indicates that oscillations of two different frequencies exist in the cir- cuit. Let there be two oscillation circuits, each con- sisting of a condenser and an inductive circuit. I^et these * circuits be coupled induc- tively and closely together by overlaying one circuit on the other, so that oscillations set up CAPACITY. FIG. 16. ELECTRIC OSCILLA T1ONS AND ELECTRIC RESONANCE 35 in one circuit induce oscillations in the other, and let the two circuits be syntonised. Then, if free oscillations are excited in one circuit, the result will be to create in both circuits a complex oscillation, which is alternately of greater and of less frequency than the natural frequency of either circuit taken alone. The effect, however, is capable of simple explanation by a mechanical analogy. Let a string be fastened loosely across a room, and from it let two rods be hung like pendulums placed a little way apart. Draw one rod on one side and let it go, setting it in vibration like a pendulum in a plane at right angles to the line of the sustaining string. It will at once begin to set the other pendulum in vibration, because the first pendulum in its vibrations imparts little jerks to the string, and so administers impulses which set the second pendulum in vibration. It will be found that as the second pendulum begins to take up the vibrations the first one comes to rest. The reason is obvious. By the Third Law of Motion, action and reaction are equal and opposite, and the first pendulum cannot accelerate the second without retarding itself. When the first pendulum has come to rest the second one is in full swing. The process then repeats itself, and the motion is gradually handed back from the second to the first. If, now, we consider the rate at which each pendulum is swinging, it is not difficult to see that when one pendulum is acting as the driver it must be swinging at a slightly less rate than its natural or free rate. Also, when it is being driven it must be swinging slightly faster. Accordingly, each pendulum is alternately going faster and slower than it would go if vibrating freely and alone. Piecisely the same thing happens in the case of the two coupled electric circuits. The oscillations in the primary can only create secondary oscillations by generating an electro- motive force in the secondary circuit, and a counter- electromotive force in their own circuit. This results from the insertion and withdrawal of lines of magnetic flux into and from the circuits, and the effective inductance of the primary circuit is changed by the presence of the neighbouring secondary circuit. Accordingly, the two circuits act and react on each other like the two pendulums above mentioned, and create oscillations which are alternately quicker and slower than those of the natural free period of each circuit. If we place near to either the primary or the secondary circuit another or tertiary circuit, which contains a variable capacity and some means for indicating the current strength in it, we can, by altering the capacity in this last circuit, tune it to syntonise with either frequency, and if we plot a resonance curve by continuously varying the capacity in the 30 RADIOTELEGRAPHY tertiary circuit, we find it to be a double-humped curve as in Fig. 16, where the two maximum ordinates correspond to the two separate frequencies of the oscillations set up in the secondary circuit. We shall return to this matter again in a later chapter in considering the inductively coupled transmitter used in radio- telegraphy, and also the cymometer for measuring electric wave length. CHAPTER II DAMPED ELECTRIC OSCILLATIONS 1. The Production of Damped Oscillations by Condenser Dis- charges. If a condenser such as a Leyden jar is electrically charged and then discharged through a wire having a high resist- ance, the flow or movement of electricity set up in the discharge circuit is always in the same direction. The current begins by being zero at the instant of completing the circuit, rises to a maximum very quickly, and then gradually falls in strength to zero again. The variation of the current during discharge may be represented by the ordinafces of a curve as in Fig. 1. This form of discharge is called the deud- beat or non- oscillatory dis- charge. It takes place when the resistance E of the dis- charge circuit is greater than the value of 2\/_, where C is the capacity of the con- denser measured in farads, L the inductance of the circuit in henrys, and R the resistance of the circuit in ohms. Thus suppose that a Leyden jar having a capacity of 01 of a microfarad or O'Ol x 10 6 of a farad is discharged through a circuit having an inductance of 1 microhenry. Then, in these units, the quantity J is equal to 20, and hence if the circuit has a resistance of more than 20 ohms the discharge will be dead beat or unidirectional in type. If, however, the resistance is much less, say 1 ohm, then the discharge will not be unidirectional, but will be an oscillatory discharge, electricity moving backwards and forwards in the discharge circuit with gradually decreasing amplitude, in the manner already explained in the previous chapter, so that 38 RADIOTELEGRAPHY the discharge current is represented by the ordinates of a decrescent wavy curve, as in Fig. 2, constituting a train of damped oscillations. A mechanical illustration of these two modes of discharge is as follows : Imagine two metal vessels, one of which is exhausted of its air and the other contains air under pressure. The difference of the pressure of air in the two vessels corresponds to the difference of potential between the two plates of the electric condenser. Let the above-named vessels be connected by a long, narrow pipe FIG. 2. which offers considerable obstruction to the movement of air through it. Then when the connection is made by opening a tap in the pipe, the air pressure will sink in one vessel and rise in the other, but the motion of the air in the pipe will always be in the same direction. This corresponds to the dead-beat electric dis- charge. If, however, we suppose the vessels to be connected by a short, wide pipe offering but little obstruction to the motion of air through it, then, if the connection is suddenly established by opening a valve, the air pressure in the two vessels will only be equalised after a series of rushes of air to and fro in the pipe. The air first flows through the pipe in one direction, and then in virtue of its inertia overshoots the mark and moves back again, and this action is then repeated, each of the air oscillations being less than the previous one, until they subside and leave the two vessels at last at equal pressure. To create trains of damped electric oscillations we must therefore charge some form of electric condenser and discharge it suddenly through an inductive circuit of low resistance, and repeat the process again and again. The simplest way of doing this is by means of an induction coil having its secondary terminals connected to two metal balls separated by a slight air gap, these balls being also connected to a condenser and low resistance inductance coil joined in series, as in Fig. 3. DAMPED ELECTRIC OSCILLATIONS 39 The induction coil I has its primary circuit supplied with current from a primary or secondary battery or from a continuous current dynamo, and in this circuit is an automatic interrupter which continually opens and closes the battery circuit. At each interruption an electromotive force is set up in the secondary cir- cuit, and the condenser C becomes charged with positive electricity on one side and negative on the other. When the difference of potential of the two plates reaches a limit fixed by the length of the gap between the two spark balls, a discharge of the condenser takes place through the inductive resistance L and across the spark gap S, and if the gap is not too long the conditions are fulfilled for FIG. 3. the production of the oscillatory form of discharge of the condenser which accordingly takes place. Several such oscillatory discharges may take place during the rise and fall of the electromotive force created in the secondary circuit at each interruption of the primary current. Hence, when the coil is set in operation a bright crackling spark appears in the spark gap, and the experienced ear can decide from the sound whether this spark is accompanied by electric oscillations or not. This apparatus is called the spark apparatus for the production of damped oscillations. A perspective view of it is shown in Fig. 4. The oscillations die away in amplitude during the train because the energy of the original condenser charge is being dissipated by resistance or in other ways, part of the dissipative resistance residing in the spark itself. We shall then consider each of the element of this apparatus in succession. 2. Induction Coils for Radiotelegraphy. We shall assume that the reader is acquainted with the general construction and mode of operation of an induction or spark coil. For the purposes of radiotelegraphy a type of induction coil frequently employed is one 40 RADIOTELEGRAPHY called a 10-inch coil, meaning an induction coil which can produce a spark 10 inches long between points or small balls attached to the ends of its secondary circuit. Such a coil consists of a primary circuit of thick wire, generally No. 12 S.W.G. in size, wound in 300 or 400 turns on an iron core composed of a bundle of well annealed soft Swedish iron wires, each not more than No. 22 S.W.G. in thickness, the bundle being 2 inches in diameter and about 18 inches long. In some cases makers provide several FIG. 4. primary cores or wind the primary coil in several distinct layers, the ends being brought out so as to combine them in different ways in parallel or in series This, however, is not so necessary in radiotelegraphy as in Kontgen ray work. The resistance of the primary of a 10 -inch coil made as above would be about 0'3 of an ohm, and its inductance 0-02 of a henry. This primary coil is enclosed in an ebonite tube with very thick walls, at least J of an inch thick, and this tube must be very carefully made and perfectly free from flaws. The tube should be 2 inches longer than the iron wire bundle and closed by ebonite caps at the ends, the ends of the primary wire being brought out through holes in one of the caps. This tube carries two thick ebonite cheeks between which the secondary coil is wound. This last consists of a very considerable length (10 to 17 miles) of No. 34 or No. 36 S.W.G. copper wire, double covered with white silk. The secondary circuit is wound in a very large number of flat coils or sections, several hundred such coils being sometimes employed. These are prepared by winding the silk covered copper DAMPED ELECTRIC OSCILLATIONS 41 wire between paper discs in a flat spiral, as a sailor winds up a spare rope. These coils are then slipped on to the ebonite tube enclosing the primary coil, and the ends of the coils are then jointed together. To enable this to be done, the coil sections are wound in double flat layers with a disc of paraffined paper between, the beginning and end of the wire thus being at the outside, and the two layers so wound that the windings follow on in the same direction. There is then no difficulty in making the joints between the various flat coils composing the secondary circuit. Mr. Leslie Miller has, however, invented an ingenious machine for winding the flat disc coils consecutively with no joints between them at all, as shown in Fig. 5, in which, however, the discs are FIG. 5. shown widely separated for the sake of clearness. The object of this mode of winding is to secure that no two points on the secondary wire which are at great differences of potential come near to each other. The whole of the very numerous flat coils forming the secondary circuit are compressed together between the two thick ebonite cheeks, and it is usual to immerse the whole finished secondary coil in very hot melted paraffin wax to exclude air and insulate it thoroughly. The coil so made is finished by enclosing in an outer sheath of thin ebonite and mounting it on a box baseboard on supports of ebonite. The ends of the secondary wire are brought to two terminals carried on ebonite pillars and provided with adjustable spark points (see "Fig. 6). Two other adjuncts are then necessary. In the first place some means has to be provided for rapidly interrupting or reversing the primary 42 RADIOTELEGRAPHY current, since it is only by so doing that we can create electro- motive force in the secondary circuit. This appliance is called an interrupter. Also it is necessary, as first shown by Fizeau, to place a condenser of a certain capacity across the points between which interruption of the primary circuit occurs. This is called the primary condenser. In reference to the construction of induction coils for radio- telegraphy, it should be noticed that the value of a coil is not to be judged simply by the length of spark it can give between 'the ends of the secondary wire when the primary circuit is interrupted, PIG. 6. [Reproduced by permission of Messrs. Newton LiIi and pLv\2 respectively. Again, each circuit not only is linked with its own lines of flux, but with some of those of the other. Let M be a quantity, called the coefficient of mutual inductance, such that MI 2 is the flux due to the secondary circuit which is linked with the primary, and MIi is the flux due to the primary which is linked with the secondary circuit. Then pMI 2 and pMIi are the induced electromotive forces in the primary and secondary circuits respectively. Now, in each circuit the potential difference of the condenser terminals is the sum of the reactance voltage and induced voltage. Hence we have for the two circuits the electromotive force equations Vi V 2 = Eliminating Ii and I 2 by the help of the equations Ii = I 2 = >C 2 V 2 , we have finally the equations 2 = jp a MCiVi 4- (1 - ^ 2 L 2 C 2 )Y 2 = Furthermore, eliminating Vi and V 2 by cross multiplication, we arrive at the biquadratic equation - M 2 )^ 4 - (dl* + C 2 L 2 )/ +1 = M 2 If we write & 2 for v-^-, and if we assume the two circuits are LiL 2 tuned so that CiLi = C 2 L 2 = CL, then, making these substitutions in the above equation, it is easy to see that 2 _ 1 1 _ "CL Hence there are two values of p, according as we take the positive or negative sign. Since the natural frequency n of either circuit alone is given by the equation n = ^ /TTT, we may denote the two roots of the above equation by the symbols p 2rrni and pz = 27r7i 2 , and then we have 1 n\ = 7i 2 = n DAMPED ELECTRIC OSCILLATIONS 69 The quantity Jc is called the coefficient of coupling, and it denotes the ratio of the coefficient of mutual inductance M to the square root of the product of the separate inductances LI and L 2 of the two circuits. M is always numerically less than x/CE because the number of lines of flux a circuit can send through a neighbouring circuit is less than those linked with itself. Hence k is a proper fraction, and \/l - k is less than unity, and \/l 4- k is greater. Hence we see that n\ must be greater than n, and n% less than n. Accordingly we derive the following important conclusion. When two oscillatory circuits are inductively coupled together so that oscillations in one excite oscillations in the other, and if these circuits are tuned to the same frequency when separate, then when coupled together, oscillations of two frequencies are set up in them both, that of one being greater, and that of the other less, than the natural frequency of either when alone. The difference of these two frequencies depends upon the co- efficient of coupling of the circuits. If k is a small fraction, the circuits are said to be loosely coupled, and then % and % are not very different. If k is a large fraction near to unity, then % and w 2 are very different, and the circuits are said to be closely coupled. For many purposes a very convenient form of variable , in- ductance may be made as follows : two cylinders of hard dry wood or of ebonite have a coarse screw groove cut on their surface with a pitch of 8 or 10 to the inch. In this groove, bare copper wire, No. 14 or No. 16 S.W.G-., is wound, and the ends of the wire attached to screw terminals in the ends of the cylinders. The groove should be so deep that the wire lies halfway deep in it. These cylinders are mounted up on a board side by side. A thick and heavy strip of copper is bent as in Fig. 20, so as to fit the XMk FIG. 20. curved surface of both parallel cylinders and lie on them. An ebonite handle serves to slide it along. The current enters at one end of one spiral, passes down it -a certain distance, then cuts across the copper strap to the other, and returns by the other spiral. Hence, by moving the strap more or less along the cylinders, a variable amount of inductance can be inserted in any oscillatory circuit. The inductance of any length of the spiral can 70 RADIOTELEGRAPHY be predetermined by the formula given in Chapter I., in the section on Inductances. Another mode of making an inductance which can be varied, is to construct two circular coils of equal number of turns, one of which is rather smaller than the other and can revolve on an axis arranged as a diameter to the larger coil. The two coils are joined in series with each other, and when they are arranged in the same plane with the windings following on in the same direction, they have their maximum inductance. If, however, the inner coil is turned round so as to be at right angles to the outer, or turned right about face, so that the windings are in the opposite direction, then the two coils in series have a greatly reduced joint inductance. By the use of appropriate induction coils, we can transform electrical oscillations, increasing the current and lowering the potential, or vice versa. Oscillation transformers for creating extra high potential discharges by means of lower potential oscillations are often called Tesla coils, though as a matter of fact employed by many physicists prior to the date of Tesla's researches. A coil of this description consists of a primary circuit, which should have few turns, and a secondary circuit of many turns. These circuits must be constructed of highly insulated wire, and the secondary circuit should be preferably formed of one single layer of wire wound on an ebonite or glass tube. The primary circuit is in series with a condenser and spark gap. The whole coil must be immersed under some highly insulating oil to prevent brush discharges. When oscillatory discharges are sent through the primary circuit, oscilla- tions are created in the secondary circuit of higher potential, and long sparks and powerful electric brush discharges can be taken from the ends of the secondary circuit. The effect may be increased by " tuning " the two circuits by adding capacity to the secondary circuit. 8. Multiple Transformation of Oscillations for High-Tension Con- denser Charging. We can employ the high frequency oscillations produced by a Tesla coil or oscillations transformer to charge an oscillation circuit, and so produce trains of oscillations which are not only of great amplitude, but succeed each other with great rapidity. The following arrangement, devised by the author in 1900, is such a system of multiple transformation for high potential con- denser charging. A high-tension alternator, D (see Fig. 21), provides an alternating current having a frequency, say, of 50 at a pressure of 2000 volts. This current passes through the thick wire of an ordinary high-tension transformer, T 1 , and is transformed DAMPED ELECTRIC OSCILLATIONS 71 up to 20,000 or 30,000 volts. Across the secondary terminals of this transformer are connected a pair of spark balls, S 1 , a con- denser, C 1 , and the primary coil of an oscillation transformer, T 2 . The secondary circuit of this last is connected in turn to a pair of spark balls, S 2 , and to a condenser, C 2 , and the primary circuit FIG. 21. of a second oscillation transformer, T 3 . The secondary circuit of this last transformer then provides oscillatory discharges of extra high tension and high frequency, and a large number of trains of oscillations per second. The operation of the apparatus is as follows : At each alterna- tion of the current in the alternator, a current traverses the first transformer T 1 , and creates alternations of potential which charge the condenser C 1 . If the circuit composed of the secondary circuit of the transformer T 1 , the primary circuit of the transformer T 2 and the condenser C 1 has its capacity and inductance adjusted to be in resonance with the low frequency (say 50) of the alter- nator, then powerful oscillations will accumulate in it, which at intervals will discharge across the spark gap S 1 . Thus there will not be 100 sparks per second at S 1 , corresponding to the 50 fre- quency, but perhaps 10 or 12. At each of these sparks the con- denser C 1 discharges with oscillations and gives rise to a long train of damped oscillations. These are transformed up in potential by the transformer T 2 , and in like manner charge the condenser C 2 , and, if the circuit of this condenser is properly tuned to the circuit of the condenser C 1 , then, in like manner, powerful oscillations are set up in the circuit composed of C 2 and T 3 , and when sparks occur at the second spark gap S 2 we have high potential high frequency oscillations in the circuit of C 2 which consist of multiple trains of oscillations, a group of trains in the circuit of C 2 corre- sponding to each one in the circuit of C 1 . Special means have to be provided for preventing the arcing at the primary spark balls, which will be described in a later chapter. 7 2 RADIOTELEGRAPHY We have already shown that when two oscillatory circuits are in tune and coupled together inductively, oscillations of two frequencies are created in them by their mutual reaction. Hence in the above-described arrangement the effect produced in the last oscillation circuit is a very complex one, and cannot be described as a simple series of trains of oscillations of one period. Other arrangements with multiple spark gaps have been devised, by which oscillations can be created in definite relative phase differ- ence to each other, such, for instance, as that due to Mandelstam and Papalexi, which is as follows : A circuit is constructed as in Fig. 22, which contains two con- FIG. 22. densers, d and 02, two spark gaps, Si and S 2 , and so arranged as to form two oscillation circuits, Oi and 2 . The spark gap Sa is short-circuited by a large inductance, L, and the other portion of the oscillatory circuits comprise inductances LI and L 2 . If, then, the spark gap Si is connected to an induction coil or transformer, the inductance L offers no obstacle to the slow charging of the condenser Ci, which accordingly becomes charged, but the con- denser C 2 is not charged. When the potential of Ci reaches a certain value, it discharges across the spark gap Si with oscilla- tions which take place in the circuit comprising the two condensers and the two inductances with a frequency determined by the in- ductances LI and 1^ and the capacities Ci and C 2 . At a certain phase of the discharge determined by the constants of the circuit a discharge takes place across the spark gap S 2 , and sets up inde- pendent oscillations in the circuit 2 , which have a frequency determined only by the capacity and inductance C 2 , L* Hence we have oscillations started in the circuit 2 which bear a definite relation to those in Oi as regards phase of maximum value, deter- mined by the inductances and capacities and the spark-gap length S 2 . The full theory of the action can only be explained by the aid of mathematical analysis, but a comprehension of the principles involved may be obtained by considering a mechanical analogy. DAMPED ELECTRIC OSCILLATIONS 73 * Imagine a heavy weight suspended by means of a spiral spring from a fixed support. If the weight is pulled down and released, it vibrates up and down at a rate determined by the mass of the bob and the stiffness or extensibility of the spring. The mass of the bob corresponds to the inductance of an electrical circuit and the extensibility of the spring per unit force to the capacity of the condenser and the mechanical rate of vibration to the electrical frequency. If we suppose the weight pulled down by a thread which breaks when the tension in it exceeds a certain value fixed by the extension of the spring, we may regard this thread as acting like the air in the spark gap in suddenly releasing the strain and starting the oscillations. Next suppose that a second weight of different mass is hung alongside of the first, suspended by a spring of different ex^ tensibility, and let the two weights be connected through a pivotted lever, as in Fig. 23. The left-hand weight and spring must be supposed to be connected to the lever by a thread which will be ruptured by a certain tension. Hence on pulling down the right-hand mass by the thread Si the corresponding spring is extended and the first system alone stores up strain. When the thread Si, in Fig. 23, corre- sponding to the spark gap Si, in Fig. 22, snaps, the right-hand mass LI oscillates, and it communicates its oscillations to the left-hand mass L 2 , and as these systems are not in tune there is a strain brought to bear on the thread 82, which finally snaps and releases the mass L 2 , which thereafter executes free oscillations in a period determined by the mass L 2 and the resilience of the spring FIG. 23. C 2 . In this manner the free oscillations of the system L 2 C 2 are started by the oscillations of the system We shall refer in Chapter V. to a system of directive telegraphy by Prof. F. Braun in which this method has been applied for the production of two trains of oscillations having fixed phase relations. The special claim made by A. Jollas for this method of charging devised by Mandelstam and Papalexi is that by this means it is possible to convey to a condenser a greater energy than corre- sponds with the length of the spark gap used. In the ordinary or single oscillation circuit we cannot give to the condenser more energy than CV joules, where C is its capacity in farads and V 74 RADIOTELEGRAPHY * is the spark potential corresponding to the length of spark gap I employed. Hence the attempt to increase V involves an increase in I, and this involves an increase in the resistance of the spark, and therefore a larger damping in the circuit, and therefore fewer oscillations per train. Hence the integral effect of the oscillations as estimated by the heating effect of the whole of the train when passed through a fine wire is not necessarily increased, but may be decreased by increasing the spark length. We diminish the number of oscillations in a train if we try to increase the ampli- tude of the initial oscillation. It appears therefore that there is a certain length of spark gap which gives the least damping and therefore greatest integral effect, and this appears to correspond to a very short spark length in air of about 1 mm. or less. CHAPTER III UNDAMPED ELECTRIC OSCILLATIONS 1. High Frequency Alternators. It has already been explained that undamped or persistent electric oscillations are extra high frequency alternating currents, the frequency of which may be from 1000 to 10,000 times greater than those of the so-called low frequency alternating currents used for electric lighting and the transmission of power. The production of these undamped oscillations has attracted great attention of late years, and several methods have been found by which they can be produced. One of these is by the use of a high frequency alternator. The invention of these machines dates back to about 1889 or 1890, when arc lighting by alternating currents began to attract attention, and it was hoped that by the employment of alter- nating currents of a frequency of 10,000 or more, the sound of the alternating arc which was very noticeable at 50 or 100 ~ would be annulled. Elihu Thomson and Nikola Tesla were successful in constructing such machines. Tesla constructed one form of high frequency alternator as follows (see Fig. 1) : It consists of a fixed ring-shaped field magnet with magnetic poles projecting inwards and a rotating armature in the form of a fly-wheel. This wheel, J (see Fig. 2), was turned down on the edge, forming a kind of flanged pulley, and this groove is wound full of annealed iron wire insulated with shellac. Pins, L, were set in the sides of the ring J, and flat coils, M, of insulated wire wound over the periphery of the armature wheel and around the pins. These coils were connected together in series, and the ends of the series carried through a hollow shaft, H, to slip rings, P, P, from which the currents were taken off by brushes, 0, 0. The field magnet consisted of a kind of toothed wheel, with the teeth turned inwards (see Fig. 2), and an insulated wire or strip was wound zigzag fashion between these teeth, so that when a con- tinuous current was passed along this conductor, the teeth were RADIOTELEGRAPHY made alternately North and South magnetic poles. It is quite possible thus to produce a magnet having 400 radial poles in the FIG. 1. circumference and also easy to put 400 coils on the armature. Hence if such a machine is driven at a speed of 3000 revolutions D E FIG. 2. per minute, or 50 per second, it produces an alternating current having a periodicity of 10,000 ~. A machine of this kind can be UNDAMPED ELECTRIC OSCILLATIONS 77 constructed to give a current of, say, 10 amperes. In the machine above described, which was capable of giving an alternating electromotive force of about 100 volts, the field magnet consisted of a ring of wrought iron 32 inches outside diameter, about 1 inch thick, the inside diameter was about 30 inches. The distance between the teeth was about $*- inch, and each field magnet tooth was about -^ inch thick. On the armature 384 coils were connected in two series. The width of the armature was 1 J inch. With magnetic teeth placed so close it was necessary to have an extremely small clearance between the armature coils and the magnet, to avoid excessive leakage or loss of useful magnetic flux, hence, it was impossible to use wire for the armature thicker than No. 26, Brown and Sharp gauge. This size is equivalent to No. 28 J British S.W.G. The armature wires must be wound with great care, otherwise they are apt to fly off in consequence of the great peripheral speed. It is practicable to run such an armature at a speed of 3000 revolutions per minute, equivalent to a peripheral speed of 375 feet per second. In another type of machine constructed by Tesla, magnetic leakage was avoided by making adjacent poles on the same side of the armature of the same polarity. In this second form the armature consisted of a copper plate in the form of a disc with a large hole in it (see Figs. 3 and 4). The plate was cut through by radial slits alternately at the inside and outside edge, so as to divide the plate up into a zigzag strip. This plate was clamped on a central boss fixed on a shaft (see Fig. 4) and caused to re- volve between the two parts of a field magnet having a large number of inwardly projecting poles, all those on one side being of the same polarity and facing an equal number of like poles on the opposite side, of the opposite polarity. In this manner, the disc was perforated by the magnetic flux passing across from one set of poles to another, and the passage of the strips into which the disc is cut up, into and out of these streams of magnetic flux FIG. 3. 78 RADIOTELEGRAPHY gives rise to the electromotive force in the armature. The arma- ture winding therefore consisted of a single disc-shaped conductor equivalent to a zigzag winding, and this was driven at a high speed so that the radial elements of the armature cut across streams of magnetic flux. A very strong excitation could there- fore be employed without producing any wasteful leakage flux. The chief defect of this design of armature is that unless the slits in the disc are very close together, so that the width of the radial bar or slice is not more than -^ inch, there is considerable heating of the armature, due to eddy currents set up in it. In one FIG. 4. machine of this type, constructed by Tesla, the field had 480 polar projections on each side, and from this machine it was possible to obtain a current having a frequency of 15,000 complete periods per second. When a machine of this description having a disc of considerable diameter is driven at a speed of 3000 E.P.M., very accurate balancing is necessary, or otherwise dangerous vibrations will be set up in the machine. Great rigidity and accuracy of work is therefore necessary in all parts of the machine, because the clearance between armature and field magnets must necessarily be very small. A plan for obtaining the necessary high relative speed between UNDAMPED ELECTRIC OSCILLATIONS 79 armature and field without exceeding moderate limits of actual rotation was adopted by Sir David Salomons and Mr. Pyke, who constructed in 1891 a high frequency alternator on the following lines. It consists of two iron discs, both having teeth like a crown wheel and each revolving independently on a shaft turning in its own long bearing. The wheels are placed on the ends of the shafts in line with each other so that the projecting teeth are in apposition and can be brought almost into touch with each other by shifting the bearings upon the bed plate, in grooves made for the purpose of facilitating this adjustment. The discs are each 12 inches in diameter, and one of these discs is so wound as to constitute both the armature of an alternator and the armature of a continuous current motor. With this object, the greater part of the centre of the disc is filled up with a Gramme-wound flat ring armature and the usual commutator, whilst the edge of the disc consists of a large number (about 360) of small iron teeth, round which a fine insulated wire is coiled. These teeth project outwards perpendicularly to the surface of the disc, and by means of insulated slip rings the alternating current can be drawn off from this alternator armature. The other disc or wheel constitutes the field magnet both of the alternator and the motor. It has a transverse bar, round which insulated wire is wound forming an electro-magnet, which provides the field for the Gramme armature, and the current also passes in shunt through a wire wound zigzag fashion between projecting teeth on this magnet disc, similar to the winding on the armature disc. A continuous current is supplied to this field magnet by means of a pair of slip rings and brushes, and there is also a brush holder carrying a pair of brushes fixed to the disc which press against the commutator of the Gramme armature fixed in the other disc. When a continuous current is supplied to the machine at a pressure of 100 volts, it commences to rotate, the two discs running in opposite directions, the continuous current field magnets being pushed backwards as it drives the Gramme armature forwards. In this manner, a differential velocity can be given to the discs equivalent to a speed of 3000 E.P.M. in its effect on the alternating armature. .Since there are ten teeth to the inch in the peripheries of the discs, and 360 poles in the whole of the circumference, it follows that with an absolute speed of each disc of 1500 E.P.M. an alternating current will be produced in the wire wound in the teeth of the armature disc which will have a frequency equal to 180 times 50, viz. 9000 periods per second. A description has been given by B. G. Lamme of a small 8o RADIOTELEGRAPHY alternator of 2 K.W. capacity, having a frequency of 10,000. This alternator was built by the Westinghouse Company for Leblanc, who required it for experiments in connection with telephonic research. The alternator is of the inductor type, with 200 polar projections. The armature was of sheet steel only 3 mils (= 0*003 inch) in thickness. The rotor consisted of a forged steel disc 25 cms. in diameter. Driven at a speed of 3000 E.P.M., the frequency was 10,000 complete periods per second. Since steam turbines, such as the De Laval turbine, are now built which run at a speed of 30,000 RJP.M. or more, this motor, when it can be employed, offers a means of obtaining very high frequency currents from any suitable form of alternator direct coupled to the turbine. It is most convenient to make the revolving part the field magnet and have the armature stationary. Generally speaking, it is not easy to obtain by any of the devices above described a frequency higher than 30,000 periods per second. Very excellent mechanical workmanship and perfect balance is necessary to be able to run any form of disc armature, having a diameter of 30 cms. or so, at a speed of 100 revolutions per second. Such an armature must carry 300 coils to be enabled to give even this frequency. In consequence of the difficulty of balancing a wound armature, the inductor form of alternator has been adopted in some cases for high frequency machines. The revolving part is then merely an iron disc having teeth or notches cut on its edge. If two chisel- shaped magnetic poles are placed on either side of such a disc, and if these poles carry armature coils wound on them, then as the notched iron disc rotates it varies the magnetic reluctance of the magnetic circuit, and hence the flux passing through the armature coils. In this manner an electromotive force is created in them wluch has a frequency determined by the speed of the iron disc and the number of its teeth. W. Duddell has described the construction of a high frequency alternator of the inductor type. It consists of a laminated soft iron ring having two inwardly projecting poles. This ring is wound with an exciting circuit, so that a direct current flowing in this circuit tends to make one of these poles North and the other South. In addition, another or armature circuit is laid upon the ring. Between the pole pieces a laminated soft iron disc revolves which has V-shaped notches cut on its periphery. The exciting circuit on the ring had inductance coils inserted in it, so as to prevent high frequency currents being generated in it. The iron inductor disc was revolved by a cotton belt passing UNDAMPED ELECTRIC OSCILLATIONS round a pulley on the inductor shaft and round two large metal disc pulleys which in turn were driven by an electric motor. In this manner the inductor disc was driven at 30,000 or 40,000 E.P.M. Alternating currents could be obtained from the armature circuit having a frequency up to 18,000 per second. The machine gave a current (R.M.S.) of 1 ampere and an electromotive force of 40 volts. Subsequently inductors with 50 or 60 teeth were used and driven at speeds up to 600 revolutions per second. This furnished an alternating current having a frequency of 50,000. Finally an inductor disc was made with 204 teeth, merely a sort of laminated iron disc with a milled edge. Coils of wire were wound on the iron pole tips as armature coils, and with this * FIG. 5. High Frequency Alternator of S. OK Brown. arrangement it was finally found possible to create an alternating current having a frequency of 120,000 when the disc was driven at a speed of 600 revolutions per second. On the other hand, the output of the machine was then very small, being only O'l ampere at 2 volts. The alternator gave 3'6 volts on open circuit. This machine was constructed for experiments on the electric arc, and not primarily for the purpose of electric oscillation work. An inductor type of alternator has been constructed by S. G. Brown, having a normal speed of 6000 K.P.M. and a frequency of 12,000 ~ but capable of running up to a frequency of 20,000 ~. The output of the machine is, however, not more than a fraction of a horse power (see Fig. 5). a 82 RADIOTELEGRAPHY The only inventor who so far has claimed to have constructed an alternator of larger power and much higher frequency is E. A. Fessenden, in the United States, who has made an alternator said to give a frequency of 80,000 ~. In practice it seems to have been limited to a frequency of 60,000 ~ with an output of 250 watts when running at a speed of 10,000 E.P.M., and having an electromotive force of 60 volts. As far as any details have yet been published it appears that the machine is of the Mordey type, consisting of a fixed armature in the form of a thin disc and a revolving field magnet with 360 poles. At a speed of 139 revolutions per second, or about 8400 E.P.M., an alternating current of 50,000 ~, and with a terminal E.M.F. of 65, is generated. The maximum output, however, is hardly one-third of a kilowatt. Hence, although sufficient for the experimental purposes for which it was required, such a machine falls far short of the power required for long distance radiotelegraphy. Eecently, however, the same inventor has constructed some high frequency alternators of the same type but larger output, to which reference is made in Chapter IX, of this manual. It is found by experience that the attempt to run alternator armatures or inductor discs at very high speeds, say, above 5000 E.P.M., involves considerable power expenditure, due to resulting air friction and air churning. Hence, if we attempt to gain the high frequency by extremely high speeds, the alternator cannot have a very high efficiency. On the other hand, there are practical limitations also to the size of the armature and its peripheral velocity. The simplest form of high frequency al- ternator is the inductor type of machine in which the only moving part is a steel disc having its edges cut into teeth, so that by its revolution the magetic flux through the armature coils is varied and hence an electromotive force created. The inductor type labours, however, under the disadvantage that the attempt to take a current out of the machine generally results in a large drop in the terminal potential difference. It is therefore an exceedingly difficult matter to combine in one alternator the properties of high frequency, high potential, and large power output. It cannot be said that such machines are yet commercial articles, or that they can be easily made for frequencies above 20,000 ~. The progress yet made in constructing such alternators has not yet enabled them to command much use in radiotelegraphy, and we are not yet able to obtain commercial high frequency alternators having even a power output of 10 kilowatts at a frequency of 50,000 or more. Hence the alternator method of producing undamped oscillations has up to the present only come UNDAMPED ELECTRIC OSCILLATIONS 83 into limited use, although there is still the possibility that it may be improved, and a considerable field for its employment exists in connection with radiotelephony, as described in a later chapter. 2. The Production of Undamped Electric Oscillations from a Continuous Current. The discovery that it is possible to produce undamped electric oscillations from a continuous electric current by means of the electric arc opened up a wide field of research. In 1892 Elihu Thomson patented in the United States (U. S. Patent, No. 500,630. Applied for July 18, 1892) the following method intended to effect the above-mentioned transformation. From the terminals of a direct current dynamo or a storage battery B, having an electromotive force of 500 volts, a circuit is taken which passes through a coil of very high inductance K and is interrupted by a spark gap S between two metal balls. These balls are adjustable as to dis- tance, and are also connected by another 0&Q&Q&&. circuit consisting of a condenser C and | K an inductance L in series (see Fig. 6). -=- B The operation was stated to be as follows : When the spark balls are put in contact, a current is drawn from 0000 the supply and passes through the large FIG. 6. inductance coil. If the balls are sepa- rated, an electric arc is formed and the condenser becomes charged by the difference of potential between the balls. The formation of the arc between the balls involves, however, the passage of a current through the large inductance, which causes a drop in voltage, so that the potential difference of the balls is decreased. The inventor stated that the ball distance, inductance, and capacity can be so adjusted that the condenser is regularly charged and discharged across the spark gap. The electromotive force in the direct current circuit charges the condenser and then forms an arc across the spark gap, but the rush of current which then ensues through the large inductance causes an arc between the balls and a drop in their potential difference, and the condenser then discharges back across the gap. In his specification Elihu Thomson says that he was able easily to obtain persistent oscillations in the condenser circuit of 30,000 or 40,000 per second, but no proof was given in this publication that the oscillations were really unintermittent. Nevertheless, it is clear that he realised the utility of undamped oscillations, and was endeavouring to find means for producing them, as shown by remarks made subsequently in 1899 on the matter in an address to the American Association for the Advancement of Science (see the Electrician, 84 RADIOTELEGRAPHY September 22, 1899, p. 778). He does not mention the use of any other material than metal for the balls, but it was affirmed that the effect is improved by the use of a strong magnetic field across the arc, or an air blast applied to the space between the balls. These observations of Elihu Thomson did not at the date of first publication attract much attention, probably because no apparent immediate application presented itself, and it was not until a fresh discovery was made by Duddell that interest in the matter revived. It is clear, however, that Elihu Thomson had proved in 1892 that the shunting of a direct current arc by an oscillatory circuit containing capacity and inductance provided a means for converting some of the energy of a direct current into energy of electric oscillations, whether the transformation was into groups of intermittent damped oscillations or into true persistent oscillations. 3. Duddell's Singing Arc. In 1900 W. Duddell described some very interesting observations on the behaviour of an electric arc between solid carbon terminals when shunted with a condenser and induc- tance in series, in a paper to the Insti- tution of Electrical Engineers of London, entitled, " On Kapid Variations in the Current Through the Direct Current Arc." In these experiments he formed FIG. 7. . an electric arc A between rods of solid carbon of the kind generally used as the negative rod in an ordinary plain continuous current arc, but he connected the two arc carbons by an oscillatory circuit consisting of a condenser C and inductance L in series (see Fig. 7). Using carbons 9 mm. in diameter, with an arc current of 3*5 amperes and a potential difference of 42 volts, and a condenser of 1 to 5 microfarads capacity in series, with an inductance of 5 millihenrys, he found that the electric arc gave forth a musical note, the pitch of which depended upon the capacity and inductance in the oscillatory circuit. Also that in the condenser circuit undamped electric oscillations were set up. He showed that the effect could only be well produced with solid carbons, and this only when the capacity in the shunt circuit was of the order of a microfarad, and the resistance of that circuit rather small. He also noticed that to obtain the effect, 'the arc must be supplied with continuous current from some steady source, such as a dynamo D or storage battery, and a resistance K of several ohms must be put in this circuit in series with the arc. The resistance of the inductance in series with the UNDAMPED ELECTRIC OSCILLATIONS condenser must, on the other hand, be low, not more than about 1 ohm. To explain the manner in which these oscillations are set up in the condenser circuit by the continuous current passing through the arc, we must consider some of the properties of the carbon arc itself, that is, of the electric arc taken between hard carbon rods. In the case of every conductor of electricity there is some relation between the current flowing through it and the potential difference between the ends of the conductor. Thus if we take a metallic conductor and keep it at a constant temperature, and create various potential differences, reckoned in volts, between its ends, and measure the resulting current flowing through the con- ductor in amperes, we find that the current is strictly proportional to the terminal difference of potential, provided there is no internal source of electromotive force. Hence the relation of current to voltage for various current values can be represented by a straight line, as drawn in Fig. 8, in which the abscissae represent current in amperes and the ordinates the potential difference of the ends in volts, and the tangent of the angle of slope of the line, or the ratio of voltage to current is, by Ohm's law, constant and equal to the re- sistance of the conductor. Any line representing the relation of the potential difference of the terminals of a conductor or a generator to the current flowing through it or out of it is now called a characteristic curve. Hence, for ordinary metallic conductors, the characteristic curve is a straight line rising upwards with increasing current, and is called a straight rising characteristic curve. If, however, a series of observations are made by means of a voltmeter and ammeter on an electric arc between carbon rods, we find a totally different form of characteristic curve. If we measure the current through the arc and the potential difference of the carbons for various constant lengths of arc, and plot a curve showing the relation between the two for various currents through the arc, we obtain a curve which is concave upwards and slopes downward as the current increases ; in other words, we have a falling characteristic (see Fig. 9). An increase in the arc current is accompanied by a decrease in the potential difference of the carbons. Hence the arc considered as a conductor differs essentially from a metallic conductor, and does not obey Ohm's law. CURRENT FIG. 8. 86 RADIOTELEGRAPHY CURRENT. FIG. 9. The characteristic curve of the arc is therefore a curve which slopes in the opposite direction to that of conductors which do obey Ohm's law. Moreover, H. T. Simon has furthermore distin- guished between the so-called static characteristic and the dy- namic characteristic of the arc. The former is a curve which delineates the relation between the arc current and the arc elec- trode potential difference (P.D.) when these quantities are slowly varied and in one direction, and the latter is a curve which de- lineates the relation between current and P.D. when these quantities vary periodically and in a cyclical manner, as in the alternating current arc. The static characteristics of carbon arcs have been determined and described for many different arc lengths and carbon sizes and qualities, particularly by Mrs. Ayrton. The static characteristics for arcs between carbon and metal electrodes and in various gases have been studied by Upson and others. In all cases they are curves sloping downwards with varying curvature indicating that as the arc current increases the arc P.D. diminishes (see Fig. 10). On the other hand, the dynamic characteristic of the arc is a closed loop like the hysteresis or cyclical magnetisation curve of iron. When the current and voltage vary periodically, the arc P.D. corresponding to a decreasing current of any value is lower than for the same current when it is increasing. Hence the P.D. is not merely a function of the current, but of the direction in which the current is changing. If we then consider the operations which take place in the carbon arc when so shunted by a condenser and inductance in the light of the above facts, it will be seen that they are as follows : Suppose the arc to be burning steadily, and that the condenser shunt circuit is suddenly applied to the carbons. Electricity rushes into the condenser, and the current through the arc is momentarily diminished. The potential difference (P.D.) of the carbons is thereby increased, and thus tends still further to charge the condenser. When the condenser is fully charged, the arc current again slightly increases, and this is accompanied by a small fall in the P.D. of the carbons. The condenser then begins UNDAMPED ELECTRIC OSCILLATIONS 87 to discharge current through the arc, and still more increase the arc current, and therefore lowers the carbon P.D. Owing to the inductance in series with the condenser, it not only completely discharges itself, but more, it becomes charged in the opposite direction. It is then in a condition to repeat the process with even better conditions, and in this manner persistent electric oscillations are set up in the shunt circuit, the condenser alter- nately drawing current from the arc and then giving it back again, and owing to the falling characteristic of the arc the accompanying changes of carbon P.D. are such as to sustain the operation. The process may be described as the electrical equivalent of the action of a closed organ pipe. In this last case we have a steady blast of air from the bellows blown in at the foot of the pipe. This corresponds to the direct current through the arc. This air is made to impinge against the lip of the pipe and create a sudden compression in the pipe near the lip. This compressed aerial region is propagated up the pipe, reflected at the top and returns to the mouth. The air pressure then being rather greater inside the pipe than externally, the impinging air jet is forced outside the lip and tends to produce a reduction of pressure or rarefaction inside the pipe near the mouth. This rarefied region is in turn propagated up the pipe and returns again, and on reaching the mouth of the pipe causes the jet of air to play inside the lip, and thus reproduce the wave of compression. In this manner the fluctuating change of air pressure in the pipe controls the behaviour of the issuing air jet, so that it tends to maintain that varying condition and establishes regular periodic changes of air pressure in the pipe and outside it, which are appreciated by the ear as a musical sound. Some part of the energy of the steady air jet is thus converted into aerial oscillations. In the electrical experiment the condenser and inductance constitute an oscillatory circuit which has a natural time period of electric oscillation of its own. It answers to the column of air in the organ pipe, which also has its own natural time of aerial vibration depending on the length of the pipe. The changes of air pressure just outside the mouth of the organ pipe correspond to the changes in potential difference of the carbons between which the arc is formed. The air jet supplied by the organ bellows corresponds to the continuous current supplied to the arc. The organ pipe therefore converts some of the kinetic energy of this issuing jet into energy of aerial oscillations in the pipe. The condenser and inductance shunted across a direct current arc likewise convert some of the energy of the direct current supplied to the arc into energy of electrical oscillations in the condenser circuit. We have furthermore to 88 RADIOTELEGRAPH? notice that if the condenser circuit is radiating, that is, giving up energy to the surrounding dielectric, it must be drawing that energy from the arc. Hence, the power taken up by the condenser from the arc when charging must be greater than the power given back by it to the arc when discharging. It follows that the path of the characteristic curve of the arc when its current is increasing must be different from the path when the arc current is decreasing ; and, in fact, the complete cycle must form a closed loop. For the characteristic curve is a curve connecting the two variables, current and P.I)., which are the factors of power, and the diagram drawn in terms of these is like the indicator diagram of an engine, which is a curve connecting pressure and volume of the working substance. The area of the closed indicator diagram is a measure of the work done by the steam and the area of the closed characteristic curve is a measure of the work done on the con- denser circuit. To have any oscillations produced at all in a condenser shunt circuit we must, therefore, work between two points on the arc characteristic curve, at which it is not flat but sloped, and the steeper the curve in this region the greater will be the power taken up by a condenser of given capacity for a fixed current variation through the arc. If, therefore, the characteristic curve is not very steep at the point at which we are working, it is necessary to employ a con- denser of somewhat large capacity if we are to draw from the arc any appreciable power at each charge, because the voltage variation which creates the charge is small. Under these conditions the time period of oscillation cannot be very small. When using two solid carbons as arc electrodes the characteristic curves for currents up to about 8 or 10 amperes are not very steep, as shown in Fig. 10, for various arc lengths. Duddell found that for such arcs it was necessary to employ a condenser having a capacity of the order of 1 microfarad or more to obtain oscillations of any sensible energy, and that it was not easy to obtain oscillations of a frequency higher than 10,000 ~. Other physicists, who, no doubt, used smaller arc currents and worked, therefore, on the steeper part of the characteristic, were able to use smaller capacities and therefore obtained higher frequencies. Banti, Wertheim-Salomonson, and Maisel found it possible to obtain oscillations having a frequency of up to 400,000 by employing a carbon arc operated with a small current and large voltage. In all probability this was due to the use of smaller arc currents, enabling a smaller capacity to be employed, and yet charged with appreciable energy by working at a steeper part of the characteristic curve. Meanwhile it may be UNDAMPED ELECTRIC OSCILLATIONS 89 said that the method generally was not regarded, prior to 1903, as affording a very promising means of obtaining powerful undamped oscillations of the high frequency and energy required in radio- telegraphy. Nevertheless, Duddell had foreseen its application for this purpose, and specially mentioned it in a British Patent Specification, No. 21629, of 1900. This suggestion, however, lay dormant, but was revived again by E. A. Fessenden, who patented in 1902 and 1903 in the United States plans for effecting radiotelegraphy by setting up oscillations in a circuit consisting of a condenser and inductance connected in w ao 00 O 70 CO at 60 I" o 0. 6mm 5 mm i 5mm 45 6 78 9 10 II 12 13 K CURRENT THROUGH ARC IN AMPERES. FIG. 10. The numbers on the curves denote the arc length in millimetres. series with a spark gap ; the said spark gap being also connected through a high resistance with the terminals of a continuous current dynamo. The specification taken alone furnishes no proof that persistent oscillations were produced and not a series of inter- mittent condenser discharges caused by the drop in voltage resulting from the rush of current through a high resistance. Hence, in the absence of definite proof that undamped and un- interrupted oscillations were obtained by this method, we have a right to suspend judgment upon the result. 4. Poulsen's Method of Producing Undamped High Frequency 90 RADIOTELEGRAPHY Oscillations, In 1903 Valdemar Poulsen, of Copenhagen, described important improvements in the arc method of creating electric oscillations which give fresh importance to the matter. He produced an electric arc between a carbon rod as the negative and a copper rod as the positive terminal, the latter being kept cool by a water circulation within it. The arc was at the same time surrounded by an atmosphere of hydrogen or a hydrocarbon gas or vapour, and crossed transversely by a strong magnetic field. On shunting this arc with an oscillation circuit consisting of a small capacity and a large inductance, he found that he obtained in this circuit very powerful undamped or persistent oscillations, the FIG. lla. frequency of which, by a proper selection of capacity and inductance, could be made to be as high as a million or more, and quite within the range of those required for radiotelegraphic work. Before proceeding to explain the reasons for the increase in frequency obtainable by the above means, it will be best to describe more in detail the construction of Poulsen's apparatus, and some of the modifications of it which have been suggested. The electric arc is formed with a direct current voltage of 400 to 500 volts between the end of a solid carbon rod, about 1 inch in diameter, and the end of a water-cooled copper pole (see Fig. lla). The latter consists of a copper tube, which is closed at both ends UNDAMPED ELECTRIC OSCILLATIONS and has an inlet and exit pipe for circulating cold water through it. The end of this tube terminates in a sharp copper nose piece, which is removable and can be renewed when burnt away. These two electrodes pass through holes in brass sockets let into two marble slabs, which form the ends of a brass cylinder, which passes through a brass box shown in section in Fig. lla. By this means the cylinder can be cooled outside by water to remove the heat created by the electric arc formed between the copper and carbon poles in its interior. Ar- rangements are made whereby the carbon rod can be slowly rotated on its axis by a motor, and the arc is "struck" and length regulated by a screw attached to the copper pole. In addition to this, the polar ends of a powerful electromagnet M (see Fig. 11&) project into the cy- linder gas tight so as to form a powerful magnetic field transversely to the arc. By means of a pipe placed under the arc, coal gas can be admitted to the cylinder, and the gas passes out through an exit pipe. When the cylinder is full of gas, the arc is formed by means of the current taken from a 5 00- volt continuous current dynamo, and the magnetic field excited. The copper pole must be the positive pole and the car- bon the negative. If, then, an oscillatory circuit, consisting of a condenser of small capacity in series with an inductance of such magnitude as to give to the oscillatory circuit a natural frequency of 500,000 to 1,000,000, has its ends connected to the copper and carbon rods, powerful electrical oscillations are set up in the condenser circuit. The conditions for obtaining this effect are, however, as follows : Choking coils or inductances must be placed in the wires bringing FIG. lib. 9 2 RADIOTELEGRAPHY the continuous current to the arc, so as to prevent the oscillations passing back through the generator. The gas or other hydrocarbon must be supplied freely but not too fast. The magnetic field must be strong, and the arc length must be adjusted to it. There is a particular arc length (called the active arc) which gives the best results. The copper terminal must be kept as cold as possible, and the carbon rod must have its edge square and sharp and be kept in slow but very regular rotation. When these conditions are all ful- filled, the oscillations in the condenser circuit are powerful and prac- tically undamped or persistent. In place of coal gas, pure hydrogen gas may be used, but it is rather more difficult to main- tain a steady electric arc in an atmosphere of pure hydrogen than in coal gas or in air. We may also use the vapour of a volatile hydrocarbon liquid such as pentane, or petrol, or even alcohol may be introduced drop by drop into the arc chamber and allowed to evapo- (Seproduced from the "The Electrician" by permission of the F^ 6 ' ThlS liquid Can proprietors. be supplied from a sight-feed lubricator as shown in Fig. 12, which gives a general view of the Poulsen Arc apparatus. It is found that almost any gas which does not contain oxygen will exalt the frequency of the oscillations obtainable from a carbon-copper arc, even an inert gas like nitrogen, but a hydrocarbon gas or vapour is found to give the best effect. Poulsen proceeded immediately on the discovery of FIG. 12. UNDAMPED ELECTRIC OSCILLATIONS 93 these facts to apply them in the practice of radiotelegraphy, but we shall return to the consideration of this application in a later chapter. 5. Other Researches on the Transformation of Continuous Currents into Electric Oscillations. In addition to the researches already mentioned made on the singing arc, the effect has been the subject of experiments by Simon and Keich, and researches by H. Th. Simon, and the last-named investigator has done much to elucidate the matter as well as to explain the reasons for the discrepancies between the results of other observers. Simon and Eeich in 1903 found that when a high potential arc was formed between metal balls in vacuo, strong oscillations were set up in a circuit including a capacity and inductance in series placed as a shunt across the balls. H. Th. Simon in 1906 subjected the whole phenomena to very careful scrutiny. He pointed out, as already explained, that the production of oscilla- tions by a continuous current arc essentially depends upon the arc having, as already explained, a falling characteristic curve, and showed that the effect was best produced by using high potential arcs and small currents and electrodes which are good conductors of heat like metals. The reason for this is clear when we examine the form of a characteristic curve of a carbon arc for constant arc length, but varying current as in Fig. 10. We see that the slope of the curve continually increases as the current decreases, so that for small currents the characteristic is much steeper than for large ones. This implies that for small currents a certain absolute decrease in the current is accompanied by a much greater increase in the P.D. of the electrodes than is the case for larger currents. Hence we can communicate the same energy to a condenser of small capacity by an arc using small arc currents as to one of larger capacity by using larger arc currents. In the former case, however, we have a higher frequency of oscillation made possible. As long, therefore, as we are using a large current to form a carbon arc in air, say, 10 amperes, we are working on the nearly flat part of the characteristic curve, and we can only trans- form a sensible part of the continuous current energy into oscilla- tions by making use of a condenser of relatively large capacity, and therefore permitting only relatively low frequency oscillations to be created. If, however, we employ a high potential arc and a small arc current, then we are working on the steep part of the characteristic curve, and can transform a sensible portion of the direct current energy into electric 'oscillations by using a small capacity in the shunt circuit, and hence obtain oscillations of high 94 RADIOTELEGRAPHY frequency. This, then, probably explains the discrepancy between the results of various observers. They have used different arc currents and voltages. The problem of creating undamped oscillations of a frequency high enough for radiotelegraphy resolves itself, therefore, into the discovery of methods for making the characteristic curve of the arc steep for that part corresponding to the current used. It appears that this can be achieved to some extent by the use of artificially cooled metallic electrodes, or at least a metallic anode or positive pole, and partly by causing the arc to be formed in a strong transverse magnetic field. Poulsen's important dis- covery, however, was the effect of a hydrogen or hydrocarbon atmosphere in increasing the steepness of the characteristic curve of the copper- carbon arc. This was well shown by some experiments carried out in the Fender Electrical Laboratory by Mr. W. L. Upson, in 1906 and 1907, at the author's suggestion. In these experiments, amongst other results, the characteristic curves were obtained for electric arcs formed between electrodes of different materials in air and in hydrogen gas. Very con- siderable differences were found in the form of curves. Thus, for instance, the diagram in Fig. 13 shows the forms of the characteristic curves of a carbon- carbon arc in air, and of a copper-carbon arc in hydrogen, the arc in the last case being formed between a cooled copper rod as the positive and a carbon rod as the negative electrode. It will be seen that for the same arc current the copper-carbon arc in hydrogen has a much steeper characteristic curve than the carbon arc in air. Also the above curves show the importance of using small arc currents and high potentials. Upson found that a carbon-aluminium arc in air, the carbon being the positive electrode, had almost as steep a characteristic for small currents as a copper-carbon arc in hydrogen, copper being the positive electrode. No explanation has yet been given why a. 80 4 8 12 ARC CURRENT IN AMPERES, FIG. 13. UNDAMPED ELECTRIC OSCILLATIONS 9$ the hydrocarbon atmosphere so greatly increases the steepness of the characteristic curve of the arc burning in it. Another matter of considerable interest is whether the oscillations set up by the Poulsen arc are truly persistent or are intermittent. The author examined this question as follows : A Poulsen arc apparatus was constructed as described in the previous section, and an oscillatory circuit was connected to the arc electrodes. This consisted of a condenser, C (see Fig. 14), made of metal plates separated by ebonite sheets, the whole being immersed in highly insulating oil, joined in series with a copper wire spiral inductance L. The condenser had a capacity of 0*0029 microfarad, and the inductance was 215 microhenrys. Hence the oscillation period of the circuit was nearly 5 microseconds, fr or ^-g of a second. To this oscillation circuit was connected a long helix H of silk-covered wire, wound on an ebonite rod 4*78 V H FIG. 14. cms. in diameter, the helix having 5470 turns of No. 30 S.W.G. copper wire, and a length of 210 cms. Also a. strip of zinc as long as the helix was laid on the table, the helix and the strip being connected to the two ends of the inductance of the oscillatory circuit, as shown in Fig. 14. When the Poulsen arc P was set in operation, the condenser circuit had oscillations set up in it, and these communicated to one end of the helix a series of electromotive impulses. Such a wire helix possesses inductance and capacity, and periodic electro- motive forces acting on one end of it set up therefore a condition of electrical vibration in it. The helix has a certain natural period of electrical oscillation of its own, and if the frequency 96 RADIOTELEGRAPHY of the electromotive impulses agrees with this period, then the variations of potential at the far or open end of the helix will become very much greater than the variations at the end attached to the condenser circuit. This is a case of electric resonance, and the effect is exactly analogous to that by which the sound of a tuning fork is greatly exalted if it is held over a tall glass jar of suitable depth. If the jar has a depth about equal to one-quarter of the length of the air wave corresponding to the note emitted by the fork, then the conditions are fulfilled for producing this exaltation of the sound. The jar, or column of air in it, is then said to be in resonance with the fork. In the same manner a helix of wire of suitable length may be brought into electric resonance with an oscillatory circuit. In the experiments described, the far end of the helix of wire was furnished with a number of needle points. The arc was formed by the currents taken from 440 volt supply mains, but some part of this voltage was dropped in the regulating resistance of the arc. The actual potential difference of the arc electrodes (copper and carbon) was from 300 to 350 volts continuous, and the arc current from 5 to 10 amperes. The potential difference of the condenser of capacity 0*0029 mfd. in the shunt circuit was, however, as much as 1200 or 1500 volts RM,S. value, and the current in this condenser circuit was about 5 amperes, as measured by a hot wire ammeter. This showed that the amplitude of the potential variations of the condenser plates was very much greater than the steady potential difference of the arc electrodes. The helix, however, effected a further rise in the potential, for from the needles at the far end powerful electric brush discharges took place, which showed that the voltage variations at the open end of the helix vastly exceeded even those of the condenser plates. The helix produces a powerful electric field all around it, and in this field vacuum tubes of all kinds, or glass bulbs filled with rarefied gases, glowed brilliantly. A vacuum tube V of the spectrum type (see Fig. 14) filled with rarefied Neon gas glows with an intense orange light when held near the helix when the arc is in operation. If the tube is waved rapidly to and fro, or attached to a turn table and rotated (see Fig. 14), the persistence of vision causes its image to be expanded into a band or disc of light. This image will, however, be found to be crossed by black lines transversely, for the to and fro movement, and radially for the rotation, and this indicates that the light of the tube is intermittently extinguished. The tube is not con- tinuously luminous because the electric field and the electric oscillations are not absolutely uninterrupted. This experiment shows that although undamped oscillations are set up in the UNDAMPED ELECTRIC OSCILLATIONS 97 condenser circuit, they are not quite without interruption or discontinuity. The cause of these interruptions in the oscillations seems to be the sudden shifting of the point on the carbon from which the arc takes its departure as the carbon rod rotates. On the other hand, if the carbon does not rotate it is rapidly worn away at one point and changes in arc length occur. Nevertheless, whilst the oscillations are taking place they are undamped in the sense that their amplitude is maintained. In the chapter on radiotelegraphic measurements (Chapter VIII.) we shall describe the methods in use for measuring the damping and logarithmic decrements of electric oscillations. Measurements have been made of the logarithmic decrement of the oscillations set up by the Poulsen arc by Eausch von Traubenberg, and he has found that the log. dec. is practically zero. This implies that the amplitude of each oscillation is the same as that of the preceding or following one, in other words, that the oscillations are persistent. Nevertheless, we must conclude from the author's experiments that under some circumstances short interruptions in the uniform flow of the oscillations may take place. The c Poulsen method may, however, be properly described as a method for the production of undamped oscillations. The claims that have been made for other arc or spark methods, such as that of Elihu Thom- son, or the modification of it suggested by Fes- senden or S. G. Brown, have not yet been justi- fied to the same extent, by measurement of the logarithmic decrement, or established themselves by definite evidence as proved methods for the production of persistent oscillations. S. G. Brown, in 1906, devised a modification of Elihu Thomson's method, as follows : A disc of metal W, preferably of aluminium, is fixed to a shaft and kept in slow rotation (see Fig. 15). Against the edge of this disc a copper block C rests, pressing lightly, and a direct current under a pressure of about 200 volts is passed through a resistance EI and large inductance LI and across the loose contact between the block and the disc. A condenser K and small inductance L 2 in series are also joined as a shunt between the block H [Reproduced by permission from " The Electrician.*' FIG. 15. 98 RADIOTELEGRAPHY and the disc. When the direct current passes, oscillations of high frequency are set up in this condenser circuit, and these can be transformed up or down by an oscillation transformer. We cannot, however, conclude without proof that this method produces persistent oscillations and not a very rapid series of intermittent oscillations. The only convincing evidence that in any particular method provides a means for the production of truly persistent undamped oscillations, is afforded when an actual measurement of the logarithmic decrement shows it to have a zero value. In addition, some evidence should be forthcoming to show that there are no interruptions in the series of oscillations. Various modifications have been suggested in the actual apparatus for the production of undamped oscillations by the electric arc in hydrocarbon vapour. The author has made use of the following arrangement : To get rid of the necessity for rotating the carbon, the magnetic field itself may be made a radial or conical field, and the arc caused to take place transversely to it. The arc itself will then rotate round the edge of the carbon. To obviate the necessity for a closed box to contain the gas enveloping the arc, the anode or cooled copper terminal may be made cuplike in form at the upper end, and the carbon or cathode may be made hollow, and various liquid hydrocarbons allowed to drop down it so as to generate a hydrocarbon vapour just where it is required. Ac- cordingly the arrangement of the apparatus is as follows : On a cast-iron base-plate is placed a cylinder of iron 15 cms. in diameter, and 15 cms. high, and about 2 cms. in thickness. Inside this cylinder is a vertical iron pin, which is enclosed in a brass tube, and cold water can pass up a hole bored in the pin and down again in the space between the pin and the brass cylinder. The top of the brass tube is closed by a recessed copper cap, which is kept cool by the circulating water. This tube is surrounded by a magnetising coil, and when a current passes through this coil it magnetises the pin and creates a flux, which passes up the pin, then spreads out radially and completes its magnetic circuit through an iron disc with a hole in it placed on the top of the cylinder. Through this hole passes a hollow carbon rod, and the current to form the arc passes up the brass tube through the cap, leaping across at some point to the carbon, and thus forms an arc between the carbon and the inner edge of the copper cap, which is constructed with a replaceable ring of copper for renewals. This arc is formed in a strong radial magnetic field, and therefore tends to rotate round the carbon rapidly. Some suitable liquid, such as turpentine, petrol, pentane, or amyl alcohol, is allowed to trickle down the hollow carbon and drop into a copper cap and be UNDAMPED ELECTRIC OSCILLATIONS 99 volatilised, and thus surrounds the arc with the necessary non-oxy- genic atmosphere. To protect the upper iron disc of the cylindrical tubular electromagnet from being overheated, a sleeve of insulating fire-proof material, such as porcelain or silica, is placed round the carbon. The coil exciting the magnetic field may be arranged as a shunt coil on the main arc circuit or be placed in series with the arc. If used in the latter way, it obviates the necessity for any other choking coil in series with the arc, but in any case there must be such a coil, to prevent the oscillations excited in a con- denser and inductance circuit joined as a shunt across the arc from passing back into the generator circuit. An arrangement of this kind will produce persistent oscillations when the arc is worked with an electromotive force of 220 volts taken from the electric supply mains of an ordinary house service, provided that it is a direct current supply. As already mentioned, to secure the best results there is a particular adjustment of magnetic field strength, supply of hydro- carbon or non-oxidising gas, and length of arc is necessary, and these can generally only be found by trial and failure. The oscillatory circuit which is shunted across the arc must be one in which the capacity is small and the inductance large. If we reckon the capacity C in microfarads and the inductance L in microhenrys, then the ratio must be a large number, something of the order of 10,000. In other words, we must keep the capacity small relatively to the inductance. If made large, that is, any- thing like a considerable fraction of a microfarad, it will be found impossible to keep the arc alight. The condenser robs the arc of so much current at each oscillation that it is extinguished. There can be no doubt that the arc apparatus for producing undamped oscillations is a somewhat troublesome appliance to manipulate when it is desired to obtain a prolonged production of oscillations, and one difficulty is to get rid of the large amount of energy which is dissipated in the form of heat in the arc itself. The single arc works well up to about 10 amperes and 440 volts, but when it is desired to obtain more energy in an oscil- latory circuit, then it is better to employ a number of arcs joined in series rather than attempt to put more current through a single arc. 6. Inductive Effects of Undamped Oscillations. The inductive effects of undamped electric oscillations are very striking. Some of them may be exhibited by the following experiments. A Poulsen arc apparatus is set up as described in the previous loo RADIOTELEGRAPHY sections, and part of the inductance in the oscillatory circuit is made to consist of a coil of insulated wire of 8 to 12 turns, say, of 7/20 S.W.G. indiarubber-covered wire, wound on a square wooden frame, say 60 cms. or 2 feet in the side. If the capacity in the oscillatory circuit is a condenser of 0'003 of a microfarad, and the total inductance is about 300 microhenry s, then the natural time period of oscillation of this circuit would be 5*96 or nearly 6 microseconds, and if the arc were operated with 400 to 500 volts, and taking 7 to 8 amperes, a hot wire ammeter placed in the oscillatory or shunt circuit would indicate a current of 5 or 6 amperes. Hence, although the condenser has a small capacity, 1 s it is charged and discharged so many times per second, viz. _ 6 or nearly 166,000, that the actual quantity of electricity passing and repassing each section of the conductor is very large. It must be remembered that an electric current is measured by the quantity of electricity passing any section of its conductor per second, and, as regards heating effects, it does not matter whether this passage is uniformly in one direction or an ebb and flow backwards and forwards. Hence, a small quantity of electricity oscillating very rapidly may produce the same thermal effects as a larger quantity oscillating more slowly. If the quantity of electricity in, or the charge of, a condenser varies from instant to instant, in accordance with a sine function of the time, so that the actual charge q in the condenser is to the maximum charge Q in the relation q = Q sin pt where p stands for Zirn and n is frequency of the oscillations, then the current a flowing into or out of the condenser at any instant is the time rate of change of the charge, and is given by the expression a = pQ, cos pt Hence, in accordance with the principles already explained in Chapter I., the maximum value of the current, which we will denote by A, is connected with the maximum value of the charge Q by the relation A = pQ, If the condenser has a capacity C microfarads, and is charged to a potential V volts, then the maximum value of the current into or out of it is A amperes, such that CV UNDAMPED ELECTRIC OSCILLATIONS TOT Accordingly, if the capacity C is, say, 0'003 mfd. and the charging 10 6 voltage 1500, and the frequency n is -rp then A should be 4'7 amperes nearly. This calculation proceeds on the assumption that the charge of the condenser varies in accordance with a sine law. This, however, is not strictly true in the case of the oscillatory circuit of a Poulsen arc. Therefore the relation between the current A in the oscillatory circuit and the oscilla- tory potential difference of the condenser terminals is not strictly, but only approximately, in accordance with the equation A = 10 6 We must note in passing how the true potential difference of the condenser terminals denoted by V in the above equation is to be measured. If we apply a direct current voltmeter to the electrodes of the Poulsen arc, we should obtain a reading Y , which might be anything between 200 to 500 volts, according to the electromotive force of the direct current dynamo supplying the current and the controlling resistance in series with the arc. If we apply an electrostatic voltmeter to the terminals of the con- denser in the shunt circuit, we should find a much larger potential difference, say Vi, which is alternating. But this observed potential difference Vi is due partly to the direct P.D. between the arc terminals and partly to the oscillations in the shunt circuit, and the true oscillatory potential difference of the con- denser terminals, denoted by V in the equation above, is connected with V and Vi by the relation V 2 = V 2 V 2 Hence V = \/V - V 2 For the actual variation of the P.D. of the condenser terminals may be represented by the ordinates of a line which is made up of a sine curve, the ordinates of which are all increased by a constant amount YO, and it is not difficult to show that the mean of the squares of the ordinates of such a curve is equal to the sum of the square of the constant ordinate and the mean square value of the ordinate of the sine curve taken alone. To obtain therefore the value of V suitable for insertion in the equation A = !^ 6 V for 102 RADIOTELEGRAPH? giving the current, we have to take the square root of the difference of the squares of the readings of an electrostatic voltmeter applied to the condenser terminals and a direct current voltmeter applied to the arc electrodes. The power taken up in the arc in watts cannot be precisely measured by the product of the direct current through the arc measured in amperes, and the reading VQ of the direct current voltmeter. It can only be properly measured by a wattmeter. Nevertheless, experiment shows that the power factor of the Poulsen arc, viz. the ratio of the true power taken up by it as measured by a wattmeter to the ampere-volts or product of the current and arc P.D., is not far from unity, generally about 0'97. Ee turning then to our experiments; if we hold near to the above- mentioned square circuit another circuit consisting of a few dozen turns of highly insulated wire, say, 50 turns of No. 16 S.W.G. india- rubber covered wire wound into a circular or square coil 30 cms. in diameter, and attach to the ends of this circuit a 50-volt carbon filament glow lamp, we shall find that when this secondary circuit is held near to the square primary circuit the incandescent lamp glows up brilliantly. This shows that in the secondary circuit a very considerable voltage is induced. This electromotive force is created in the secondary or lamp circuit by the rapid change in direction in the lines of magnetic force due to the primary oscillation circuit which are linked with the secondary circuit. The oscillatory current in the primary circuit creates around it a rapidly alternating magnetic field. Some of the lines of this field are thrust through or linked with the secondary circuit, and the insertion or withdrawal of these gives rise to the secondary electromotive force. This last is proportional in magnitude to the rate at which the flux linked with the secondary is changing. If Ii is the maximum value of the high frequency current in the primary circuit, then we may denote the maximum value of the total number of lines of force due to it which are linked with the secondary by MIi, where M is a quantity called the mutual inductance of the two circuits. The secondary E.M.F. is then measured by the rate at which the flux linkage with the secondary circuit varies. Hence, the maximum value of the induced secondary electromotive force E 2 is equal to pM.Ii or to 27raMIi, where n is the frequency. Accordingly, although M and I may be small, yet if n is very large the secondary E.M.F. may become very great. If this secondary circuit has an effective resistance B 2 and an inductance L 2 , then the secondary current created in it has a value I 2 such that UNDAMPED ELECTRIC OSCILLATIONS 103 " as shown in treatises on alternating currents. 1 In the case of high frequency circuits the resistance E is nearly always negligible in magnitude compared with the re- actance pL, and hence we may say that for such a simple inductive circuit the secondary current I 2 is given by the equation M 12 = L, 11 It is therefore increased by reducing the inductance of the secondary circuit as much as possible, whilst increasing the mutual inductance. This may be illustrated by the following experiment : Construct a circuit consisting of, say, 50 or more turns of highly insulated wire No. 16 S.W.G., the wire being wound in a flat, circular coil about 45 cms. or 18 inches in diameter. Insert this coil in series with a condenser of small capacity, say, O003 mfd., as a shunt circuit on a Poulsen arc. Then bend a very thick copper wire, say, of 5 mms. diameter or 0*2 inch in thickness, into a nearly complete circle of 45 cms. or 18 inches diameter and connect the ends by a few inches of thinner copper wire, about No. 22 or 26 S.W.G. Fix the copper ring to a wooden handle. Then excite undamped oscillations in the first-mentioned coil and bring down over it slowly the thick copper secondary circuit. The fine piece of copper wire will be rendered red hot and melted. We have here constructed a pair of circuits, the primary of many turns and the secondary of only one turn, so that the mutual inductance M is large, but the secondary inductance L2 is small. Hence the secondary current induced is many times greater than the primary current and easily melts part of the secondary circuit. In this case we have constructed a " step-down " transformer for undamped oscillations, decreasing the voltage but increasing the current. On the other hand, we may reverse the process and construct a step-up transformer. If we insert the primary coil of an ordinary induction coil as part of the inductance in the shunt circuit of a Poulsen arc, we can obtain from the ends of the secondary circuit a flaming discharge which resembles an alternating current arc in being a lambent mobile flame rather than a spark, not unlike the discharge produced by a Wehnelt break. 1 See " The Alternate Current Transformer," by J. A. Fleming. Vol. I., p. 178. 104 RADIOTELEGRAPHY 7. Resonance Effects in connection with Undamped Oscillations. All the inductive effects of persistent oscillations are vastly increased if we avail ourselves of the exalting influence of resonance. The oscillatory circuit shunted across the Poulsen arc has a certain time period of oscillation of its own, determined by its capacity and inductance. If we cause this circuit to act inductively upon a secondary circuit which contains no condenser, the effect on this last circuit is to produce a forced oscillation. If, however, the secondary circuit has a condenser inserted in it and possesses inductance, then it also has a natural time period of its own, and by adjustment of the capacity and inductance can be tuned to the period of the primary. When this is the case, the electromotive force produced by the reversal of the direction of the magnetic flux due to the primary circuit which passes through the secondary has a cumulative action, each impulse adding its effect to the previous ones, just as when we apply small blows or puffs of air to a pendulum, striking or blowing exactly in time with the natural time period of the pendulum. In this last case small repeated blows soon create a very large vibration in the pendulum, and in the electrical case the feeble but repeated inductive impulses finally create a very much larger current than would be the case if the circuits were not in tune. These facts may be illustrated in the following manner. Let the shunt circuit of the arc consist of a small condenser, having a capacity say of 0*003 mfd. and an inductance of about 300 microhenrys, part of this inductance consisting of a circuit of 12 turns of insulated wire wound on a square wooden frame, about 60 cms. or 2 feet in the side. The inductance in series with this square circuit should consist of two parallel spirals of copper wire, about 10 turns to the inch, the spiral being 1 inch in diameter and say 36 inches long. These spirals are provided with a short-circuiting bar, as already described in 7, Chap. II., by means of which the total inductance can be gradually varied. The secondary circuit should consist of a similar square coil, sliding spiral inductance and condenser, which may conveniently be a large Leyden jar. In some part of the secondary circuit is inserted a small 4- volt carbon filament glow lamp, taking a current of about 0'5 ampere to render it incan- descent (see Fig. 16). The secondary circuit is placed about 6 feet, or say 2 metres, away from the primary circuit, the square coils having their planes parallel to each other. If, then, oscillations are set up in the primary circuit by the arc, and the distance between the coils is sufficient, we shall find that little or no effect is produced in the secondary circuit, as judged by the emission of UNDAMPED ELECTRIC OSCILLATIONS 105 light by the glow lamp, whilst the two circuits are out of tune. It is, however, possible to so adjust the inductance and tune the circuit that the little lamp glows brilliantly, thus indicating that the secondary current is immensely increased by tuning, but is extinguished by a very small alteration in the inductance either of the primary or the secondary circuit. This experiment illustrates very well the effect of syntony in exalting the secondary current. It can be conducted also with undamped oscillations set up in the primary circuit by the use of a spark gap, but it is found that when using undamped oscillations the tuning, as it is called, is much sharper than with damped oscillations. The reason for this is generally as follows : The current set up in the secondary circuit is determined as to strength by several factors. Let J stand for the root-mean-square value of the secondary FIG. 16. current, that is the value which would be indicated by a correct hot-wire ammeter inserted in the secondary circuit. Then J 2 is called the mean square value of the oscillations, or, by some German writers, the integral effect of the oscillations. Let GI be the capacity in the primary circuit and C 2 that in the secondary circuit, and Vi the potential to which the primary condenser is charged, and let k be the coefficient of coupling of the circuits, and furthermore let S t and S 2 be the logarithmic decrements of the two circuits. Then when the circuits are adjusted to resonance so that they both have the same time period of oscillation n the secondary current comes to a maximum value. We shall denote this maximum value by J max , so that J 2 max is the mean-square 106 RADIOTELEGRAPHY or effective value of the current in the secondary circuit when it is tuned to the primary. It was shown lay P. Drude by an elaborate course of reasoning that the value of this maximum current is given by the formula, J =~ C V 2 The quantity * l is the energy put into the primary condenser a initially or at each charge. Hence it is seen that the secondary current (mean-square value) is proportional to the initial energy given to the primary condenser and to the capacity of the secondary condenser, and to a function which depends upon the coupling and the damping of both circuits. For the proof of this formula, the original paper or advanced treatises on the subject must be consulted. The important point to notice is that the value of J 2 max i g increased by decreasing the log. dec. of either the primary or secondary circuit. This points to the importance of securing small resistance in the circuits. It will be shown in the next chapter that part of this damping depends upon the power of the circuits to radiate their energy, and hence in practical work it is never possible to make the decrements absolutely zero, or else the secondary current would become infinite in value. The above formula is of use in calculating the current in the antenna of a radiotelegraphic transmitter apparatus when the parfc which radiates is inductively coupled to the part which stores the energy used. CHAPTER IV ELECTROMAGNETIC WAVES 1. The Electromagnetic Medium. Although the study of distance actions in electricity and magnetism, such as the attraction and repulsion of magnetic poles, and of conductors conveying currents, as well as the effects of magnetic and electrical induction, long ago suggested the idea of an electromagnetic medium to the minds of Ampere, Faraday, and Henry, as the means by which these effects at a distance are produced, the notion was hardly more than a surmise until James Cierk Maxwell, in 1864, communi- cated to the Eoyal Society a classical memoir on " A Dynamical Theory of the Electromagnetic Field." In this paper he pre- sented the chief facts of electromagnetism in such a manner as to show that electric and magnetic effects cannot be produced instan- taneously at a distance, but must be propagated through space with a finite velocity. The phenomena of optics had already been found to be best explicable 011 the hypothesis that all space is occupied and all matter interpenetrated by an imponderable medium capable of undulation, and that waves in it of a certain kind and range of wave length constitute light considered as a physical agent. Maxwell showed that the velocity with which an electro- magnetic effect is propagated through a dielectric is dependent upon the known electric and magnetic qualities of it, and from such measurements as were available at that date, he proved that it must be identical with that of light. Hence, the conception that the assumed luminiferous sether must be identical with the hypothecated electromagnetic medium was advanced to the position of a scientific theory supported by some important evidence. 2. Electric and Magnetic Quantities, To explain the steps by which this conclusion was reached, we must prepare the way by some definitions of terms and special conceptions. In addition to its ordinary state, every material substance can be put into a con- dition in which it is said to be electrified or charged with electricity. io8 RADIOTELEGRAPHY We are able, for instance, to bring about this state by the friction of two bodies one against the other, but the simplest investigation shows that the resulting electrical condition of the two bodies is not the same, and that whilst both are electrified, the electrifica- tion on them are different in kind. Thus the friction of glass against silk produces so-called positive or vitreous electricity on the glass and negative or resinous electricity on the silk, whilst the friction of ebonite or shellac against flannel produces negative electrification on the ebonite or shellac and positive on the flannel. If the two rubbed bodies are held at a little distance, an attractive force is found to exist between them, and if another small electrified body, say, a pith ball, is placed in the space between the silk and the glass which have been rubbed together, it will tend to move one way or the other, according to the nature of the electrification on the pith ball. The space between two such oppositely electrified bodies is called an electric field, and is said to be the seat of electric strain, or displacement. The force between the electrified bodies is not merely an action at a distance, but is determined by the nature of the material substance, whether solid, liquid, or gas, which occupies the interspace. If a couple of bodies are electrified, one positively and the other negatively, and are found to attract each other with a certain force in air when placed at a certain distance, then if immersed under turpentine or paraffin oil, they would at the same distance only attract each other with about half that force. If, on the other hand, they were placed in a highly perfect vacuum, they would attract each other with a slightly greater force. Hence, we see that the attraction essentially depends upon some quality of the surrounding material medium, but is not entirely dependent on it, for it exists even when all surrounding matter is removed. This quality is capable of being numerically defined, and is called the dielectric constant of the medium. We shall denote its magnitude by the letter K. The dielectric constant of an absolute vacuum is taken as unity. We may, therefore, say that the electrified bodies produce a state called an electric strain in the material, then called the dielectric, around them, and the degree or intensity of this state is determined by the dielectric constant of that material. It is convenient to regard the state of electric strain in a dielectric as produced by an agency called electric force, but determined as to degree or intensity by its dielectric constant. We can then say that electric force produces electric strain in a dielectric to an extent determined by its dielectric constant, just as we say that the bending, flexure, mechanical strain of a beam is produced by mechanical force to an extent determined by the elasticity of the beam. ELECTROMAGNETIC WAVES 109 The dielectric constants of some well-known materials have been already given in Chapter I., but owing to the variation in the constant of different specimens of the same substance the numbers given can only be taken as approximations for that of any particular sample. We can then define a unit of electric quantity or electrifica- tion and a unit of electric force as follows : Let two very small spheres be supposed to be equally charged with opposite elec- tricities and placed with their centres 1 cm. apart in a vacuum. If the electric quantity is such that the attractive force is -g-| T of the weight of a gramme in London, viz. 1 dyne, then each body is said to be charged with 1 electrostatic unit of elec- tricity. At a distance of 1 cm. from its centre each such small electrified sphere would exert a unit electric force when placed in vacuo. We can then, by appropriate methods, measure the charge or quantity of electricity on a body in electrostatic units, and also the electric force at any point in an electric field in the above-mentioned units of electric force. Turning next to the elementary facts of magnetism, we find we can make similar statements. In addition to their ordinary condition, certain bodies, such as steel, can be put into a state in which they are said to be magnetised. If a pair of steel wires are magnetised in the direction of their length, we find that between the ends of these wires, when near together, there are attractive or repulsive forces. The space between these ends and all around the magnetised steel is called a magnetic field, and is said to be traversed by magnetic flux. A simple experiment with a pair of such magnetised wires suffices to show that the ends are not identical in properties, but differ like positive and negative electricity. Moreover, a very careful experiment would show that as in the case of the electrified bodies so for the magnetised ones, the force between them depends to some extent on the interposed medium. Two long uniformly magnetised steel wires, whose polar ends attract or repel each other at a certain distance with a certain force in air would attract or repel each other with a slightly less force if placed in liquid oxygen or in a solution of feme chloride, but the force would not vanish even if the magnets were in a perfect vacuum. Hence we are led to conclude that the force depends to some extent upon a quality of the surrounding medium which is called its magnetic permeability. The permeability of empty space is taken as unity, and that of any other substance is denoted by the symbol ILL, and is measured by a number greater or less than unity. It is then convenient to consider that this state, called magnetic flux, produced in the space I io RADIOTELEGRAPHY around a magnetised body is due to an agency called magnetic force, the degree or intensity of the flux depending upon the permeability of the medium. Hence, we can say that electrified bodies exert electric force (E) in their neighbourhood, and produce electric strain (D) in the material or medium by which they are surrounded to a degree depending upon its dielectric constant (K). Similarly, magnetised bodies and also electric currents exert magnetic force (H) in their neighbourhood, and produce magnetic flux (B) in the material or medium by which they are surrounded to a degree depending upon its magnetic permeability \i. The fact that electrified bodies or magnets attract or repel each other at a distance, and that electric currents can create other currents in wires at a distance, and that these actions are not entirely dependent upon the presence of any material sub- stance in the interspace, but can take place also through a perfect vacuum, has always impressed competent thinkers with the idea that there must be an electromagnetic medium by means of which these actions are transmitted across intervening space. The observation that light takes time to pass from one place to another, and that it comes to us from far distant stars across interstellar space, which, as far as we know, is not full of ponder- able matter, is a proof that it must either be a substance bodily transmitted like a letter sent by post or a physical state or change of state which is propagated through a stationary medium. In- numerable facts of optics prove that it is, in fact, an undulation, and that, therefore, there must be something which undulates. The velocity with which this undulation travels has been measured with considerable exactness, and has been found to be very close to 300,000 kilometres per second, or, roughly, about 1000 million feet per second. The speed with which any disturbance travels through an elastic medium is, as -shown further on, determined by the square root of the ratio of its elasticity to its density. No known form of tangible or gravitative material has such a large ratio of elasticity to density as to permit an undulation or impulse of any kind to travel through it at a speed of 300,000 kilometres per second. The atmosphere, for instance, is a material possessing elasticity or resistance to compression, and likewise density. If, however, a sudden compression is created in it at any place, this state of compression is propagated through it at the rate of only about 330 metres per second at Cent., viz. with the velocity of sound. Even in hard steel the velocity of propagation of a com- pressional or extensional strain travels only at the rate of 5612 metres, or about 18,600 feet per second. Accordingly, we are ELECTROMAGNETIC WAVES in compelled to admit that if light is due to vibrations propagated through a medium at the rate of 300,000 kilometres or 186,000 miles per second, the medium capable of this must possess qualities very different from those of any form of tangible or ponderable matter with which we are acquainted. The medium called the sether must necessarily be universally diffused, and must inter- penetrate all ordinary matter. It cannot be exhausted or removed from any space, because no material is impervious to it. As far as we know, it is non-gravitative, but ordinary matter stands in some very close relation to it. It must also possess some form of elasticity, that is resistance to a change of state of some kind produced in it, and it must also possess inertia or a quality in virtue of which a change so made in it tends to persist. We are not justified in making the assumption that its elasticity like that of ordinary matter is a resistance to change of bulk or form, or its inertia necessarily an inertia with regard to motion. It is, how- ever, clear that the medium has the power of storing up energy in large quantities and transmitting it from one place to another, as shown by the fact that enormous amounts of energy are trans- mitted from the sun to the earth. It is only in virtue of some form of elasticity and some type of inertia that a medium can thus transmit energy in the form of undulations through it. When we consider the relations of electric strain, electric force, and dielectric constant, we see that they are quite analogous to the relations which ordinary mechanical stress or force and material strain or displace- ment and electric resilience bear to each other. The term strain in physics means any deformation of a body or change in relative position of its parts, and the word stress is applied to denote that which produces strain. If the strain increases with the stress and disappears spontaneously when the stress is removed, the body is called elastic. The ratio of stress to strain at any stage expressed in appropriate units is then called the elasticity corre- sponding to that strain. Thus, gases do not resist change of form but resist change of bulk, and their elasticity is thus a resistance to change in volume, the elasticity being defined as the ratio of the increment in pressure to the decrement in volume produced by it. Solid bodies resist change of form, say extension. Thus a longi- tudinal stress produces an extension of a bar of metal. The ratio of the pull or tension to the extension, expressed as a fraction of the original length is called the longitudinal elasticity, or Young's modulus of elasticity. The characteristic of an elastic substance is therefore that it experiences some kind of strain under the action of a corresponding stress, and that within certain limits the strain is directly proportional to the stress and inversely as the elasticity. 112 RADIOTELEGRAPHY Moreover, the strain disappears more or less completely as soon as the stress is removed. If, then, we consider a dielectric of any kind we find that a certain state can be produced in it, called electric strain, by an agency called electric force, and that the strain is within limits proportional to the force and to a constant called the dielectric constant. Also, just as the removal of the mechanical stress causes the resulting strain to disappear, so the removal of the electric force causes the dielectric strain to dis- appear. Accordingly, we notice that the reciprocal of the dielectric constant is analagous to the elasticity of a material substance. There is more than a mere analogy between mechanical strain of a solid body and dielectric strain. It is well known that the optical properties of transparent bodies are affected by mechanical strain. Thus the mechanical strain produced in the particles of a bar of glass when bent can be rendered evident by examining the bar by polarised light when bands of colour indicate the lines of strain. In the same manner the electric strain set up by electric force in dielectrics affects their optical qualities, and can be rendered evident by the employment of polarised light. Further- more, in the case of mechanical strain, the production of a strain involves the expenditure of energy, and the strain when produced is a store of potential energy. The energy stored up per unit of volume is measured by the product of the strain and the average stress. Hence it is equal to half the product of the strain and the maximum stress, or to half the quotient of the square of the stress by the elasticity, or to half the product of the elasticity and the square of the strain. In precisely the same manner it can be shown that when a dielectric is in a state of electric strain under electric force, the energy stored up in it per unit of volume is numerically equal to half the product of the square of the electric force and the dielectric constant, or to half the quotient of the square of the electric strain by the dielectric constant. 3. The Nature of a "Wave. We have then to consider the production of a wave in an elastic and dense medium. Physically speaking, a wave is defined as a cyclical change taking place in a medium which is periodic in space as well as in time. Put into less formal language, it means that each particle of the medium executes some movement or experiences some change which is repeated over and over again, all particles performing the same motion or experiencing the same change in succession but not all simultaneously. Thus, for instance, a surface wave on water is caused by the particles of water rising and falling periodically, so that along a certain line the particles execute this motion succes- sively. At regular intervals, however, along the line will be found ELECTROMAGNETIC WAVES 113 particles which are in the same phase of their motion at the same instant. These are said to be separated by one wave length. The wave length, therefore, is a distance which comprises one set of particles in all possible stages of their periodic motion. In regarding a wave we may suppose ourselves to remain fixed at a certain point, and to watch the cyclical changes which take place at that point in a time called the periodic time (T). Otherwise we may imagine ourselves to travel with a uniform velocity along the line of propagation so as to remain always in contact with the same phase of the motion. This velocity is called the wave velocity. Thus a bather standing in the sea fixed at one point finds the water rise and fall over him as the sea waves travel past him, and he can note the interval of time between two successive greatest elevations of the water. But a seagull, flying along over the surface of the sea, by adjusting his speed, can keep himself constantly poised over the summit of a hump of water or place of greatest elevation. His velocity is then that of the wave motion in the direction of his flight. If we call the wave length X, and the wave velocity V, and the wave period T, then the relation V = I = n\ holds good in all cases of wave motion. The wave velocity is equal to the quotient of wave length by periodic time or to the product of wave length and frequency. On certain assumptions we can obtain an expression for the wave velocity in terms of the elasticity and density of the medium as follows : Let us suppose that there are a row of particles each of unit volume and of mass m which lie in a straight line, and let each particle in succession execute a simple har- P. monic motion up and * down along a line at ? * a right angles to the first. A simple harmonic mo- tion is the motion of the projection on any dia- FlG - * meter of a point which moves with uniform velocity round a circle. This motion is equivalent to assuming that the small masses lying in one line are attracted back to it with a force varying as the distance when displaced perpendicularly from it. Thus let P (Fig. 1) be one of I U4 RADIOTELEGRAPHY the particles, and at any moment let its displacement from the zero line be y, and let ey be force drawing it back, then ey is the stress, and the displacement y is the strain, and the elasticity or ratio of stress to strain is e. Then the force acting on the d?"u particle of mass m is wvf and this must be equal to ey, because the force tends to reduce the displacement. Hence, the equation of motion is The above is called a differential equation, and the reader who has some slight knowledge of the differential calculus will be able to see that a particular solution of this equation is = Y sin \/-. t v m where Y is the maximum displacement during the phase. This can be at once proved by differentiating the last equation twice with respect to t, and substituting the differential coefficient so obtained in the equation of motion, when it will be found to satisfy it. Suppose, then, that the particles all execute this motion successively, so that at any one instant their positions delineate a sine curve. If we reckon the abscissae x from the point on the line at which one particle is on the axis at the zero of time, then the equation to the space distribution of all the particles at any fixed time is y = Y sin x. Again, suppose that whilst the particles are in oscillation we move uniformly forward so as to keep in contact always with a displacement of constant value y whilst varying our abscissa x. Our velocity will be -, and will be that t of the wave motion V. For that ordinate y of constant value we have the relations y = Y sin V . t = Y sin x y v m Hence m In other words, the wave velocity is the square root of the quotient of the elasticity by the mass of the particle. If, then, we are ELECTROMAGNETIC WAVES 115 considering the oscillations, not of a single row of particles but of elements of volume of a continuous medium, we can write the symbol p instead of m, where p is the mass of the unit of volume or density of the medium, and we arrive at the formula given above for the wave velocity, viz. -A This expression is the well-known formula for the velocity of a sound wave or wave of compression and rarefaction in any medium, as, for instance, through air, and from it we can deter- mine the wave velocity if we know the elasticity and the density. Thus, the density of tempered steel is 0*285 pound per cubic inch. The elasticity (Young's modulus) for the same steel is approximately 16,100 tons per square inch. Accordingly, if we reduce this last figure to absolute units in terms of the inch, pound, and second as units, we have e = 16,100 x 2240 x 32*2 x 12 and p = 0.285 Hence the velocity of propagation of a longitudinal wave of compression and extension in tempered steel is 16100 x 2240 x 32-2 x 12 ~~ 0-285 = 221>0t i This velocity, however, is in inches per second, but reduced to feet per second it is 18,400, which agrees with the observed velocity of sound through steel. The same expression would enable us to predict the velocity of a wave of any kind, say, a wave of transverse displacement, or shear wave, provided we can obtain the numerical value in absolute measure of the elasticity of the material for that particular kind of strain. Let us then consider what are the qualities of an electro- magnetic medium which would correspond to the elasticity and density of a material substance. We know nothing about the mechanical structure of the aether, nor what forms of strain can be imposed upon it, but we do know that electric force produces in a dielectric and in the aether a state called an electric strain, and that the reciprocal of the dielectric constant is a measure of the ratio of the electric stress to the electric strain, or, in other words, of the electric elasticity. Elasticity is that quality of matter in virtue of which energy can be stored up by strain in it in a potential form. A bent rod, a stretched spring and compressed 1 16 RADIOTELEGRAPH? air, are all cases of strained materials which possess potential energy in virtue of the elastic resilience concerned. If we electrically strain a dielectric we store up in it per unit volume ID 2 energy equal to ^^, where D is the electric strain and K the A j\. dielectric constant, and this energy is potential and it corresponds to the potential energy stored up in a mechanical form when a material substance of elasticity e experiences a configurational strain S, for the energy so stored up per unit of volume is then ieS 2 . Again, when an electric current A flows through a circuit embedded in a dielectric, the energy associated with that circuit is measured by -J^iLA 2 , where A is the value of the current in amperes, and L is a constant depending on the form of the circuit and fj. is the magnetic permeability of the dielectric. When a material body is in motion its kinetic energy is measured by IpBV 2 , where p is the mean density of the body, B is its bulk or volume, and V its velocity. We know that an electric current is a form of energy, and there are good reasons for considering that the energy involved is kinetic. Hence, we see that in the ex- pression for the electrokinetic energy the symbol p, or the magnetic permeability of the medium, takes the place of the symbol p or the density of the body in the expression for the motional or kinetic energy of ordinary matter. Accordingly, we have reasons for comparing the quality we call the magnetic permeability of a dielectric with the density of material substances and the reciprocal of the dielectric constant with the elasticity of matter, when we are comparing the electrical with the mechanical qualities. If this is the case, then the expression _ which is analogous to the expression \/-, should be the expression for the v P velocity of an electromagnetic wave; but before we can decide this matter we must consider more carefully what is meant by an electromagnetic wave. 4. An Electromagnetic Wave. In order that a wave may be produced in a medium, this last must possess two properties, it must resist and persist. Thus, in the case of a surface wave on water, the water surface resists being made unlevel. If at any place it is suddenly de- pressed, as by dropping a stone into it, a force is brought into play to restore the displaced water to its original level. In so doing the water is set in motion, but in virtue of its inertia when ELECTROMAGNETIC WAVES 117 so set in motion it persists in motion, and not only moves until it is back in the original position, but moves beyond it and creates an elevation in place of a depression. Then, again, the force of restitution comes into action to depress the surface again, and so an undulatory motion is set up at that point. Moreover, the water at one point is in close connection with the water around it by cohesion, so that elevation and depression at one place cause the water in the immediate proximity of the initial disturbance to share in the same motion, but to lag behind a little in imitating the motion of the first displaced portion. Hence, to create a wave in a medium we must produce some form of strain which in disappearing as potential energy transforms itself into an equivalent in motional or kinetic energy. One of the important contributions Clerk Maxwell made to this subject was to explain clearly the manner in which a con- nection between the potential or electrostatic and kinetic or magnetic forms of energy is established in the case of dielectric media. He pointed out that an electric strain, or displacement, uhilst it is changing, that is, whilst it is increasing or diminishing, is equivalent to an electric current, and must therefore create magnetic flux along a closed line embracing the varying electric strain. A varying electric strain is therefore called a displacement current. Again, Faraday had shown that the variation of magnetic flux through a closed metallic circuit creates electromotive force in that current. Maxwell extended this idea also to a circuit in any material, not only conductive but dielectric, and stated that the variation of magnetic flux through any area enclosed by a line drawn in a dielectric must create electric strain along that line. Suppose we have a currrent flowing in a straight infinitely long conductor, magnetic flux is distributed in circular lines round that conductor in the space outside and inside the wire. If a magnetic pole, that is the end of a very long thin magnet, is held in the field of the current, it will tend to rotate round the wire, being urged by a force proportional to the strength of the pole and to the magnetic force at the point where it is held. Experiments proving this are described in every book on physics. If, however, a short magnet is laid on a disc of card, the card being suspended so as to be free to rotate in a plane perpendicular to the current, but the magnet not free to move on the card, there will be no rotation of the latter (see Fig. 2). When this experiment is carefully considered, it will be found to prove that the magnetic force (H) at any point near the long straight current must be inversely as the perpendicular distance of the point from the current, and further an analysis shows that u8 RADIOTELEGRAPHY it is proportional to twice the current and inversely as the distance. If, then, we consider a single circular line of magnetic flux of radius r, its length is 2irr, and the magnetic force all along 20 it has a value where C r is the current in the wire. 20 The product 2irr X - - = 4?rC is called the line integral of the magnetic force along that line, and we see that it is independent of the radius, and therefore of the form of the path. Accordingly, the line integral of the magnetic force along any closed line embracing a current is equal to 4?r times the total current flowing through that closed line. Applying then Max- well's principle, we see that if in any dielectric the electric strain (D) is changing with time, its rate of change -77 O r I) mul- U>t ) tiplied by 4rr, gives us the line integral of the magnetic force due to it along a boundary-line perpendicular to the direction of the electric strain. When this boundary-line encloses a very small area, the quotient of the line integral by the area is called the curl, and we may, therefore, express the above statement in symbols as follows : 47rl) = curl of H In the next place, the electromotive force in any circuit is defined as the line integral of the electric force (E) along that circuit, and Faraday's law of induction, as extended by Maxwell to dielectrics, tells us that the time variation of the magnetic flux through any area is a measure of the electromotive force in, or line integral of electric force round, that bounding line. Hence, when dealing with a unit area we can express the above fact symbolically, thus - B = curl of E FIG. 2. ELECTROMAGNETIC WAVES ng The minus sign is prefixed to B because a diminution of the flux is required to produce a right-handed or positively directed electric force. The two equations 47rl) = curl of H (magnetic force) B = curl of E (electric force) establish a cross connection between the quantities D and B and E and H. There are also two direct relations, viz. B = M H 47rD = KE which express the fact that the magnetic flux B is proportional to the magnetic force H and to the magnetic permeability /u, and also that the electric strain D is proportional to the electric force E and to the dielectric constant K. This last equation is obtained in the following manner. If we suppose a sphere of radius r described in a dielectric, and that a small conductor charged with Q electrostatic units of electricity is placed at the centre, then through the surface of the sphere there will be an electric strain D produced, which is everywhere directed outwards along the radius, and the product 4?rr 2 D, or the surface integral of the strain through the whole surface, will be equal to the quantity Q put at the centre. Therefore 47ir 2 D = Q 4TTD = K(jg- 2 ) But ~-^ is the radial electric force E at the surface of the sphere. Hence 4;rD = KE Accordingly, from the above four expressions we can deduce the two important equations of electromagnet ism, viz. 47rt) = KE = curl H - B = fjLK = curl E The dot over a letter signifies the rate of change with time of the quantity denoted by that letter, and the expression " curl " signifies the line integral round a unit of area, or the quotient of the line integral round any small area by the magnitude of that area. If we then translate these conceptions into common language we can 120 RADIOTELEGRAPHY describe the production of an electromagnetic or electric wave in a dielectric as follows : To produce an electric wave we must first create at some place in the dielectric or in the aether a very sudd en change in an electric strain. This may, for instance, be the sudden release or destruc- tion of a very intense localised electric strain. The results of this, in accordance with Maxwell's first principle, is to generate magnetic flux along a circular line embracing the decreasing strain, the said flux line having its plane perpendicular to the direction of the strain. This initial strain is represented as to direction by the arrow in Fig. 3, and the embracing circular line of flux by the FIG. 3. FIG. 4. double elliptical line, being a circle seen in perspective. We may also represent the end- on view of the strain by a small circle, con- taining a cross as in Fig. 4 , and the embracing flux line is then represented by the larger embracing circle. When the end- on view of a line of electric strain or magnetic flux is represented by a small circle, we may represent the direction of the strain or flux, whether to or from the reader, by placing a dot or a cross in that circle, the dot representing that the strain or flux is towards the reader, and the cross that it is away from him. The reader should also bear in mind that a diminishing electric strain is by Maxwell's first principle equivalent to a current in an opposite direction to the strain, and an increasing strain to a current in the same direction as the strain. Also that the relation between the direction of a current and of the direction of its circular embracing magnetic flux is that of the thrust and twist of a corkscrew. Again, by Maxwell's second principle, the creation of a line of magnetic flux, or its strengthening, involves the production of electric strain along lines embracing the flux line, so related as to direction that on any section plane transverse to the flux line the ELECTROMAGNETIC WAVES 121 electric strain is counter-clockwise in direction round that part of the section in which the flux is away from the reader, and clock- wise around that section of the flux which is towards the reader. Bearing this in mind, it will be seen that the diminution of the original central electric strain is accompanied by tbe production of a series of concentric circular lines of magnetic flux all embracing the original line of strain, and with directions as denoted in Fig. 5. ELECTRIC STRALN FIG. 5. These flux lines are not, however, created simultaneously. The flux lines nearest the original strain are generated and increase as the original strain dies away, and this creation of magnetic flux involves the production of other closed loops of electric strain linked with these flux lines, which new strain lines tend, as they are produced, to create in turn other circular lines of magnetic flux in the same direction as the first created, but lying further away from the line of original strain. In this manner a state of alternate electric strain and magnetic flux is created at continually increasing distances, and is propagated out into the medium. At the instant when the original central electric strain has disappeared the line of flux nearest to it, due to its variation, has disappeared also, but the disappearance of this flux involves the creation of lines of electric strain linked with it, which in the interior of the circle of flux have a direction opposite to the original strain and outside of it the same direction. This is equivalent to a reversal of direction of the inner original or central strain, and it therefore involves a reversal in direction of the whole system of embracing and co-linked lines of flux and strain. This reversal in direction does not, however, take place simultaneously at all points of space, but successively from point to point outwards into space. A careful consideration of the diagrams will therefore show that as the original energy of electric strain imparted to the medium disappears, it expends itself in creating an equivalent in the form 122 RADIOTELEGRAPHS of an embracing magnetic flux, and that this flux in turn expends its energy in creating electric strain in the medium outside its line. Hence the energy imparted at one point is transferred from point to point in the medium by an action between contiguous parts of it, the principle involved being that variation or change of electric strain produces embracing magnetic flux, and variation or change of magnetic flux produces electric strain. The directions of the strain and flux are at right angles to each other, whilst both take place along lines which are self-closed or circuital. The portion of the medium which is the seat of these actions is con- tinually changed ; in other words, the operation is propagated through the medium with a velocity which is definitely measurable and related to the specific qualities of it. Hence the sudden release of an electric strain at one place is felt after a time at regions far removed from it. If this phenomenon is considered it will be recognised to be exactly analogous to the effect produced upon the surface of still water when we make a sudden depression at one point, as by a throwing a stone into it. We have then a depression of water level at the point of impact succeeded by eleva- tion, and we have water motion set up as the result of these changes of level, when the water moves up or down to remove them. The changes of level correspond to the electric strain, and the water motion to the magnetic flux. Change of level of water results in the production of motion in the water, and motion in virtue of inertia results in the production of change of level, just as in the dielectric, change of electric strain results in the production of magnetic flux, and change in flux produces electric strain. If we take a section of the electric field transversely to the original flux, we find that the field is occupied with concentric lines of magnetic flux, and orthogonally, or at right angles to these, and intermixed are packed the cross-sections of lines of electric strain, the strain being orthogonal or at right angles to the flux. Likewise, if we take a section of the field in the plane of the original electric strain, we find it to be occupied by closed loops of electric strain, intermixed with which are found the cross-section of lines of magnetic flux. Moreover, the direction in which the transmission of the energy is taking place is at right angles to the plane which contains the directions of the lines of magnetic flux and electric strain. Thus, an electromagnetic wave is, so to speak, woven out of electric strain and magnetic flux which constitute respectively the warp and the weft of the fabric. The magnetic flux and electric strain are called the component magnetic and electric vectors which make up the wave. ELECTROMAGNETIC WAVES 123 The magnetic component at any one spot changes cyclically in strength and reverses or alternates in direction, although it may maintain a constant direction. So also does the electric component, but at the same point in space the electric component is a maxi- mum at the instant when the magnetic component is zero, and vice versa; in other words, the two vectors differ 90 in phase. We can imagine ourselves endowed with special senses of such a kind as to enable us to detect the presence of lines of magnetic flux and lines of electric strain in space, and appreciate their direction and movement. Suppose, then, that we took up our position at a fixed spot in space through which electromagnetic waves were passing. We should detect these regions of magnetic flux and electric strain, alternately succeeding one another at that place. On the other hand, we can imagine it possible for us to fasten our attention upon a particular phase of the strain or flux, and to move along so as to keep always in contact with that same phase of the force or flux. We should then find ourselves travelling in a certain direction normal to the direction of the flux and strain, with a velocity called the wave velocity. 5. The Velocity of an Electromagnetic Wave. Let us next consider the propagation of a plane electromagnetic wave in which the electric and magnetic components are at right angles to each other and to the direction of pro- pagation. Suppose the electric force (E) is everywhere parallel to the axis of x, the magnetic force (H) to the axis of y, and the direction of propagation of the wave to the axis of z (see Fig. 6). The student who has even a small knowledge of the principles of the differential calculus knows that if the ordinate of a curve corresponding to any abscissa x is denoted by y y then the ordinate corresponding to an abscissa x + &c, where $x is a small increase in x, is y -f --x. Hence, if at the dx origin the electric force has a value E along the axis x t then the electric force in the plane of xz in a direction parallel to the axis of x, but at a distance $z from it, is E + -r-Ss, being greater the further we remove from the origin. 2 ' &/"" s f H 6x E dz^ y^ ^"' * ^'c- . dE s-. E+ dz 6 * X Fia - 6 - 124 RADIOTELEGRAPHY If, then, we take the line-integral of the electric force round a small rectangle, whose sides have lengths &r, $z respectively, lying in the plane of xz, with one corner on the origin, we have to sum up all round this rectangle the product of the length of each side by the electric force along that side, travelling always in the same sense round the rectangle, and reckoning the product as positive when the force is in the same direction as the motion, and negative when it is against it. We note, then, that parallel to the two sides of length &c, the force has a value E and E 4- -= $z respectively, and that parallel to the two sides the electric force is zero, because there is no component in the direction of propagation. Hence the line integral is equal to If we divide this line integral by the area SzSx of the rectangle, we have the curl of the electric force in the plane of xz t viz. denotes the electric moment of the oscillator. This last term is defined as follows: Imagine an ideal oscillator consisting of two small spheres connected by a thin rod. We may consider that the rod possesses inductance but negligible capacity, and the whole of the capacity is in the sphere. During the oscillations a certain quantity of electricity may be considered to oscillate backwards and forwards, and the product of this quantity by the distance between the spheres is the electric moment. In an actual linear oscillator the effective length is something less than the real length, just as in the case of the moment of a magnet, the actual distance between the poles, which is one of the factors of the magnet moment, is indeterminate, but something less than the real length. The electric moment of an oscillator can, however, be measured as a single quantity, like the moment of a magnet, and is proportional to the original charge given to it, and therefore to the capacity of one-half of the oscillator with reference to the other and to the initial potential difference to which they are charged. It is also proportional to the length of the oscillator. Since, then, the wave length of the emitted radiation is proportional to the length of the oscillator, we see that the above formula for the radiation per period of the linear oscillator shows us that this radiation will be proportional to the square of the capacity, to the square of the charging voltage, and inversely as the wave length or directly as the frequency of the oscillations. In his historical researches Hertz constructed his oscillator by attaching to one end of two short, rather stout rods of brass, two square zinc plates, and furnishing the other ends of the rods with spark balls. These rods were placed in line with spark balls in apposition, and when so arranged the plates with the surrounding air as dielectric formed a condenser of a certain capacity. When plates are charged by connecting them to the secondary terminals of an induction coil, and the spark balls on the rods approached to within 4 or 5 millimetres of each other, a bright crackling spark passes and oscillatory discharges take place across the gap. The result of these oscillations is to create, as we have seen, electric radiation, and the oscillations expend their energy in creating the state of alternate electric strain and magnetic flux 134 RADIOTELEGRAPH? which constitutes an electric wave. The characteristic of a wave is that energy is conveyed away from the wave-making body and exists in the surrounding medium in a double form partly potential and partly kinetic, the energy of the complete wave being half potential and half kinetic, and transferred from point to point in the medium with the velocity of light. We must consider a little more closely the operations which are taking place. When the two rods or parts of the oscillator are charged to different potentials, energy is stored up in them. If C is the capacity of one-half of the oscillator with respect to the other in microfarads, and V is the potential difference in volts, then n -r^V 2 is the energy storage in joules and -^-CV 2 is the 2i ID 2t storage reckoned in ergs. When the discharge takes place a certain proportion of this energy is dissipated as heat and light in the spark, and also a small proportion as heat produced in the the rods by the oscillatory current taking place in them. The larger portion, however, is communicated to the dielectric in the form of magnetic flux and electric strain, and this part does not return in again upon the oscillator, but is permanently imparted to the dielectric as an electromagnetic wave which travels away from the oscillator. The damping of the oscillations is thus partly due to resistance and partly to radiation, and the total logarithmic decrement is made up of two parts, viz. the resistance decrement and the radiation decrement, and accordingly as this last coefficient is large or small, so is the oscillator called a good or a poor radiator. The laws which govern electric radiation are closely analogous to those of radiant heat and light, and this is what we might expect, seeing that in both cases we are con- cerned with the vibrations of the same medium the aether although with different wave lengths. CHAPTER' V 00 RADIATING AND RECEIVING CIRCUITS 1. Varieties of Radiative and Receiving Circuits. Broadly speaking, circuits in which we can establish oscillations may be divided into open and closed circuits which correspond respectively to good and bad radiators of electromagnetic waves. If we place two rods or wires, each having at one end a spark ball, and at the other end a metal plate, in one line with the two spark balls in close proximity, and the plates as far apart as possible (see Fig. 1 (a)), we then construct a cir- cuit called an open radi- ative circuit which is a verj good radiator or producer of electromag- netic waves. If, how- ever, we bend round the wires, so that the plates come into approximation (see Fig. 1 (&)), we in- crease their electric capa- city with respect to each other and form a closed radiative circuit which is relatively to the open one a poor radiator. Be- tween these two extremes there is, however, no hard and sharp separation, and we may have every possible intermediate variety of radiative circuit. These various electric circuits correspond as regards their electric radiative power to good and bad thermal radiators. A rough black surface is a good radiator of heat, and a bright polished silver surface a poor one. Moreover, it is a fundamental principle in thermal radiation that good radiators are good absorbers. Thus, a lamp-blacked surface is not only a good radiator of heat; but 136 RADIOTELEGRAPHY readily absorbs heat falling upon it. In the same way good electromagnetic radiators are good absorbers of electromagnetic waves. If such waves travelling through space fall in the right direction upon an open or closed oscillatory circuit, that is, one having capacity and inductance in series, the energy of the waves is absorbed to a greater or less extent, and expends itself in setting up electric oscillations in the receiving circuit. A precise statement was long ago formulated as regards radiant Heat and Light called the Law of Exchanges, which is the foundation of spectrum analysis, viz. that a body absorbs best those particular luminous or thermal radiations which it emits if heated. An identical law holds good in the case of electric radiation, which may be enunciated as follows : An oscillatory circuit, or one having capacity and inductance, in which therefore electric oscillations can be excited, absorbs some of the energy of electromagnetic radiation falling upon it, and it absorbs best radiation of that kind and wave length which it would itself emit if set in oscillation, and absorbs it most readily when arriving in the direction in which it would itself radiate most strongly. This law of electromagnetic radiation is of the greatest importance, and may be said to be the foundation on which the art of radiotelegraphy is erected. In the case of an open circuit electric radiator the free ends become at intervals the seat of electric charges which create an electric potential in space, and as the electric strain at any point in the surrounding space is partly due to these free charges, an open radiative circuit is also called an electric oscillator. In the case of a closed radiative circuit the plates being near together and carrying electric charges of opposite sign, these tend to neutralise each other's effect in external space, and the pre- dominant agency in creating the radiation is therefore the current in the circuit. Hence, a closed radiative circuit is sometimes called a magnetic oscillator. A large variety of open and closed or intermediate forms of electric and magnetic oscillator are employed in radiotelegraphy. 2. The Open Circuit Oscillator. This radiator consists of a vertical or nearly vertical rod or wire A, the upper end of which is insulated, and may or may not terminate in a metal plate, whilst the lower end of the rod or wire is connected to a good conducting plate (E) buried in the earth, or placed near the surface of the earth (see Fig. 2). It is usually called an aerial, air-wire, or antenna. The simplest method of establishing oscillations in this antenna consists in interrupting the wire just above the earth and interposing a pair of spark balls. These balls are then connected RADIATING AND RECEIVING CIRCUITS 137 to the secondary terminals of an induction coil, and when the coil is in action the upper part of the aerial is charged at intervals, say, with negative electricity. When this charge reaches a certain potential the insulation of the air in the gap breaks down and electric oscillations are thus set up in the antenna. Prior to the discharge the wire itself forms one plate of a condenser, of which the surrounding earth is the other coating and the air and sether the dielectric, In this condition lines of electric strain stretch from the wire to the earth on all sides (see Fig. 2). When the discharge takes place this condenser is discharged and the electric strain in the spark gap disappears, owing to the air in the gap becoming conductive. At this moment the charge in the antenna rushes down into the earth, creating a conduction current in the wire, which varies from point to point, but is a maximum at the earthed end. This conduction current is completed and made circuital by the simultaneous production of a dielectric current by the release of the dielectric strain in the space round the antenna. The antenna is thus surrounded by a magnetic field, the lines of flux being circles with their centres on the antenna (see Fig. 3), As the current in the wire continues, it builds up a new electric strain around the antenna, the direction of which is opposite to that strain, the relaxation of which produced the conduction current, and finally the kinetic energy of the conduction current is transformed entirely again into potential energy of electric strain outside the antenna. This process is continually repeated, and we have then an oscillatory electric current in the wire and periodic changes in the direction of the magnetic flux and electric strain at all points outside the wire, which are propagated outwards with the velocity of light. If we follow out in detail, by mathematical analysis, the movements of the lines of electric strain which 138 RADIOTELEGRAPHY originally stretched from the antenna to the earth, we find they are continually displaced, and the result of the first oscillation is that the ends of these lines which originally terminated on the antenna run down it, and finally so place themselves that they form semi-loops of electric strain with their ends resting on the earth (see Fig. 4). This detachment of the line of electric strain / // j I /;;."::;.,.,.,;:*->. \\ \\ \\ ( < u *//"; jf ///,'---- 1 ::: '->V\i ii f \\\ ^tt Q FIG. 4. corresponds to and expresses the detachment of the energy which takes place from the antenna in the radiation, and at each oscillation a fresh production of such semi-loops of strain then takes place, those first produced moving away from the antenna radially in all directions. For the purposes of radiotelegraphy we may fix our attention entirely on the regions near the earth and at some distance from the antenna. In this district the magnetic flux lines are parallel to the earth, provided the antenna is vertical or rtearly so, and the electric strain lines are perpendicular to the earth. At any one instant the electric strain at the earth's surface is directed alternately upwards and downwards over annular regions or districts which succeed each other radially. At each oscillation these regions are displaced outwards from the wire in every direction. The shortest distance between two adjacent places at which the electric strain has its maximum value in the same direction at the same time, is called a wave length for that antenna. For a simple or plain single wire antenna earthed at the bottom with a spark gap placed near the earthed end, the wave length is not far from 4*8, or, say, 5 times the length of the wire. A plain aerial wire of this kind has a relatively small capacity. If the wire is a circular-sectioned metal wire O'l inch or 2*5 mm. in diameter (d\ and 100 feet or 300 cms. in length (I), its capacity in space far removed from the earth can be approximately calculated by the formula in Chapter I., viz. Capacity in microfarads = y 41454 x 9 x 10 5 x logio - RADIATING AND RECEIVING CIRCUITS 139 07 In this case j is 24000, and logic 24000 = 4-3802, so that the capacity is b -fe- s of a microfarad. The actual capacity of such a vertical insulated wire, with its lower end near the earth, would, however, be about 10 per cent, greater than the value calculated by the above formula, owing to the proximity of the earth. Hence, even if charged to 30,000 volts, which is the equivalent of a 1 centimetre spark at the spark balls, the actual quantity of electricity put into such a single wire antenna would be only 6 microcoulombs, and the energy stored before discharge only about T V of a joule, or 83,000 ergs, for the energy put in is measured by the value of the expression - - reckoned in consistent z units. The natural time period of oscillation, T, of this plain serial is equal to ^ r^g, where X is the wave length of the radiation in o X -LU centimetres, or to ^ -^7^0 = TTTil secon X where I is the length of o X -LU lu the wire in centimetres. Thus, for a single wire serial 100 feet long it is 0'48 microsecond, which means that each complete oscillation takes place in rather less than one-half of a millionth of a second. Such a simple straight wire has, however, enormous radiative power. It can be shown from theoretical considerations that the radiation decrement () per half period of a plain aerial wire of length I centimetre and capacity C microfarads is very nearly given by the formula 8 = 2-5 x 9 x 10 5 y = 2-25 x lO 6 ^ L v It is therefore equal to 2J times the capacity per centimetre of length reckoned in micro-microfarads. Thus, if I = 100 feet = 3000 cms., and C = ^oW microfarad, we have 8 = 0'15, which implies that the Napierian logarithm of the ratio of one oscillation to the next in the opposite direction is 015. The ratio of one oscillation to the next in the opposite direction is c~ s , where e = 2' 7 18, viz. the base of the Napierian logarithms, and 8 is the decrement per semi-period. Hence, in 10 complete oscillations, or in 20 semi-oscillations, the amplitude is reduced to a fraction of the initial amplitude, represented by ~ 2 5 , which for the case in question is equal to (2'718)~ 3 = 0'05 nearly, or 5 per cent, of the initial value. 140 RADIOTELEGRAPHY So that in less than a dozen periods the oscillations are practically damped out. The above formula, however, gives us only the decrement for radiation. In addition, there is some damping due to the resistance of the aerial wire itself and to the spark, so that in practice the actual number of oscillations pro- duced when a plain aerial is charged and discharged is something less than half a dozen. This antenna is therefore called a highly damped radiator. The great radiative power of open oscillators makes it necessary to supply them with very large amounts of power to maintain in them persistent oscillations of high frequency. Thus, if we consider the case of a simple linear oscillator consisting of two rods in one line of total length I and capacity C microfarads with respect to each other, and if undamped oscillations of simple harmonic type and frequency, N, are maintained in this oscillator, the E.M.S. value (a) of the current at the centre, reckoned in amperes, would be given by the expression where V is the maximum potential difference of the rods in volts. The electric moment of the oscillator in electrostatic units would then be given by 9 x 10 5 and the energy radiation (E) Substituting in this last expression the values of < and CV from the two previous equations, we have for the value of the energy in ergs radiated per period the formula and therefore the energy radiated per second, or the power (W) in watts supplied to maintain the oscillations continuously, is given by the equation W = RADIATING AND RECEIVING CIRCUITS 141 Now, for a simple double rod antenna, the ratio -r- is nearly 04, A and, since ?r 2 = 9'87, we have finally W = 126a 2 Hence, to produce persistent oscillations with a current having an R.M.S. value of 10 amperes at the centre of the oscillator, would involve the expenditure of 12 kilowatts. Bearing in mind that NX = 3 x 10 10 , we can also write the value of the power absorbed in the form W = 87 X 10- 20 /VN 2 which shows us that for the linear open oscillator the power absorbed to produce persistent oscillations varies as the square of the frequency, and as the square of the current at its centre. 3. The Closed Circuit Oscillator. A closed circuit oscillator is made by constructing some form of loop of wire which may have its plane vertical or horizontal, and inserting in the circuit a suitable form of condenser. It is quite easy to construct a horizontal closed circuit radiator by driving into the ground four or more stakes or telegraph poles which carry insulators, and then straining round these a wire or wires in parallel, so as to make a large loop of one turn, the ends of which are brought into a house and connected in series with a condenser and to a pair of spark balls, if damped oscillations are to be set up, or to an electric arc or alternator if undamped oscillations are required. More often a closed radiator has its plane set vertically. It is then necessary to erect some form of mast or tower to carry it. The simplest form of closed circuit vertical radiator is made by attaching to the top of a mast or tower two wires of about equal length which are upheld by one or two insulators at the summit, or under some circumstances this point of the loop may be uninsulated. The wires must be of considerably greater length than the height of the mast. The lower ends are then brought into a signalling house and attached to fixed points. To some point or points at or below the middle of the wires guy lines are attached by insulators and strained tightly so as to stretch out the wires and form a lozenge or triangular-shaped closed antenna (see Fig. 5), which may have its plane set in any required direction. Otherwise, two masts or towers are erected in required positions, and between these from insulators one or more horizontal wires are suspended, which are continued downwards in a vertical direction, and the ends brought into a signalling house, the whole H2 RADIOTELEGRAPHY arrangement forming a square or rectangular vertical closed circuit, which may have its plane set in any required direction. Or a single mast or tower may be employed, having two sprits attached to it by means of which an antenna wire is upheld in the form of a vertical rectangle, the two ends being brought into a signalling house. This last arrangement possesses the advantage that by swinging round the sprits the plane of the loop can be altered so as to set in any required direction, as required in the case of directive radiotele- graphy. In any case, the closed loop may consist of either a single wire PIG. 5. or a number of wires arranged in parallel. In the case of the closed circuit antenna the arrangement of a number of wires in parallel not too near together, has the advantage of reducing the inductance of the loop, and thus enabling a large area to be employed without correspondingly increasing the inductance. This same advantage is obtained by arranging wires in parallel in constructing an open circuit antenna, but in the case of the open antenna the objection exists that multiplying the wires increases the total capacity of the antenna although it decreases the inductance, so that one effect to a certain extent nullifies the other as regards the reduction of the time period of the antennae. The closed oscillating circuit as compared with the open has much less radiative power, and, therefore, a much smaller radiative decrement. If a closed oscillatory circuit of area S is traversed by a current of a maximum value I, then the product IS = M is called the maximum magnetic moment of the circuit, just as the product of the length of the open oscillator and the maximum electric charge at the extremity is called the maximum electric moment. It can be shown that the current radiated through a sphere of large radius described round a closed oscillator is given by the formula ,, 167T*M 2 ~3F~ where X is the wave length of the emitted radiation. RADIATING AND RECEIVING CIRCUITS 143 If the maximum current measured in amperes is denoted by A, and the area in square centimetres by S, then the magnetic moment AS of the closed oscillator M is given by -^n. Hence, if a is the R.M.S. value of the current, we have Accordingly, the power absorbed by the oscillator to radiate persistent undamped radiation in watts is given by W = 4 x 10- 88 x S a a a N* where N is the frequency, and NX = 3 x 10 10 . If we compare the above expressions for the power radiated by the closed and open oscillators, assuming them to be in both cases the seat of persistent oscillations, we see that we have the two expressions W = 4 x KT 38 SVN 4 (closed oscillator) W = 87 x lO' 20 JVN 3 (open oscillator) for the power absorbed in persistent radiation in the two oscillators respectively. These last two formulae show us that the power radiated varies as the square of the frequency in the case of the open oscillator, but as the fourth power for the closed oscillator. Hence the power radiated by a closed oscillator increases very much faster with the frequency than in the case of the open one, and conversely for the closed oscillator it is very much smaller for the same frequency, and decreases very much more rapidly as the frequency is lowered. Accordingly, a closed oscillatory circuit has sometimes been called non-radiative, but in truth there is no such thing as an absolutely non-radiative circuit; it is a question of degree, and as in the case of thermal radiation some surfaces are better radiators of heat than others, but no surface is absolutely non-radiative, so in the case of electrical radiation, some circuits are vastly better radiators than others. The closed circuit oscillator has, however, certain valuable qualities as a directive radiator ; that is to say, its radiation is not equal in all directions round its vertical axis. In the case of a linear open oscillator it is obvious that since everything is symmetrical round the axis of the oscillator, the radiation must be equal in all directions in planes passing through this axis of symmetry, and the magnetic field of the open oscillator is, as already shown, distributed in circular lines with 144 RADIO TELEGRAPH Y their centres on the axis. If, however, we consider the magnetic field of a closed circuit, say, in the form of a circle, and consider the distribution of the magnetic field in a plane perpendicular to the plane of the circuit drawn through its centre, it would be seen to be as shown by the dotted lines in Fig. 6, where the large black dots represent the cross-section of the circular circuit. This matter will be dealt with more fully in a subse- FIG. 6. quent section; meanwhile it may be pointed out that the radiation of a closed circuit antenna is at a maximum in the plane of the circuit, and although symmetrical with respect to one plane is not symmetrical with respect to any one line drawn in it. 4. Receiving Antennae or absorbing Circuits. We have hitherto chiefly paid attention to the processes by which oscillations set up in an antenna radiate electromagnetic waves, but we must at this stage consider more fully the action of an antenna or closed circuit as an absorber of electromagnetic radiation. If we move a conducting wire through a field of magnetic force so as to cut across the lines we generate in it an electromotive force which is proportional to the magnetic force, to the length of the conductor, and to its velocity at right angles to the direction of the field. If the conductor is stationary, but if the magnetic force lines move across it, the same effect is produced. This action is the basis of operation of all dynamo electric machines. If, then, an electromagnetic wave falls on a conductor in such a manner that the magnetic component of the wave is transverse to the conductor, we may regard the movement of the wave as being equivalent to a cutting of this conductor transversely by lines of magnetic force. Hence, an alternating electromotive force is created in the antenna which is proportional to its length, and to the intensity of the magnetic force. The wave, however, possesses an electric force component, and this is at right angles to the magnetic component. If the magnetic component is at right angles to the absorbing antenna, then the electric component is in the same direction as the antenna, and as the wave passes over it both these components expend their energy in creating electromotive force in the antenna in the same direction. The wire in fact absorbs the electric component in the direction of its length, and is cut by the magnetic component transversely to its length. Both RADIATING AND RECEIVING CIRCUITS 145 these operations contribute to create in the antenna an electro- motive force. In a complete wave, the energy of the magnetic component is equal to the energy of the electric component, and hence both contribute equally to the production of the electromotive force. It follows from this, and from the Law of Exchanges, that if a linear open circuit receiving antenna is placed at right angles to a similar radiating antenna, it will absorb nothing, and have no electromotive force created in it. If the antennae remain in parallel planes, but if their directions are inclined at an angle 9, then the effect produced in the receiving antenna is equal to cos times that which would be produced if the antennae were parallel. In optical language, a linear antenna emits a plane polarised wave, and the receiving antenna must be parallel to the electric component of that wave. If the receiving antenna is a closed or partly closed circuit, then a third source of electromotive force exists. Consider the case of a closed receiving circuit, placed with its plane vertically to the earth and in the direction in which electromagnetic waves are passing over it, with their electric component perpendicular to the earth, and therefore their magnetic component perpendicular to the plane of the receiving circuit. As the waves advance, their magnetic component cuts through the vertical sides of the closed receiving circuit, and their electric component is more or less absorbed by the same sides. These actions contribute to the production of an electromotive force in the circuit, which is in one direction as the components pass over the near side, and in the reverse direction as they pass over the far side. But, in addition, if we consider the group of lines of magnetic flux in the wave which at any moment fill the closed receiving circuit and perforate through it, we shall see that the effect of the advancing waves is to cause a periodic change or alternation in the amount of magnetic flux thus perforating. If the closed circuit is rectangular and exactly half a wave length long, this action will produce the maximum effect, and we may adopt a phrase used in connection with the theory of induction motors, and call it the transformer effect of the wave, whilst the cutting of the sides of the receiving circuit by the moving lines of magnetic force is called the dynamo effect. It will be seen on consideration that the relative magnitude of the electromotive forces set up by these three actions is capable of variation by many factors. Moreover, it gives the closed receiving circuit a directive power, or power of determining the direction in which the waves are travelling, not possessed by the simple open L 146 RADIOTELEGRAPHY vertical receiving antenna. Thus, for instance, if the closed receiving antenna has its plane perpendicular to the direction of the incident waves, it will not be affected at all. This leads at once to a means of determining this direction. We shall return to this matter in the last section of this chapter. 5. The Practical Construction of Antennae. The simplest form of radiotelegraphic antenna is a long vertical wire, A, which is suspended by an insulator at its upper end from a mast M, tower, or building (see Fig. 7). The lower end is in connection with one of a pair of spark balls S, the second of which is connected to a plate of metal E sunk in the earth. Owing to the fact that the electrical oscillations are confined to the surface of the wire, it is not desirable to employ a wire of large diameter. If a solid wire is used, it should not exceed 0*1 or at most 0*125 inch in diameter, that is, 2'5 to 3 rnm., but it is generally better to employ a stranded wire, say, one made up of 7 No. 20 or 7 No. 22''wires (^ or -fa S.W.G.). As regards material, tinned copper wire is most usually employed when stranded wires are used, but for solid wires aluminium may be used with advantage. The density of aluminium is 2'6 against 8 '9 for copper. Hence, for equal bulks aluminium has only one-third the weight of copper. The price of aluminium is now about 80 per ton, and that of copper (in 1908) 60. Hence, for wires of equal length and diameter the cost will be proportional to the product of the density of the material and the price per ton. Accordingly, the aluminium wire will only cost about one-third of that of a copper one of the same bulk. The tensile strength, however, must be taken into account. That of commercial aluminium is from 26,000 to 40,000 pounds per square inch, and that of soft drawn copper is about 30,000 pounds per square inch. Alloys of aluminium are, however, now made with a density not exceeding 2*7, which have a tensile strength as great as that of copper. In the case of high frequency currents, the electric conductivity does not much matter, as current is carried chiefly on the surface. The wire, however, must not be of iron, as the magnetic hysteresis of this iron would increase considerably the damping. The experience of the author has shown that aluminium withstands the weathering action of the atmosphere very well. The chief precaution which must be taken in its use is to avoid FIG. 7. RADIATING AND RECEIVING CIRCUITS 147 the galvanic action which is set up when aluminium (which is highly electropositive) is brought in contact with other metals. The end of an aluminium wire should not be twisted up with a copper, brass, or iron wire if the junction is exposed to moist air, but insulated from it. Also aluminium wires are more difficult to solder effectually than copper wires, but these disadvantages are more than outweighed by the advantages of its use. We have only one-third of the weight to support with the same windage surface, and as aluminium wire can be obtained in coils of any reasonable length, there is no need to make joints in it. If the upper end is attached to a brass or copper ring, the latter should be wound over with indiarubber tape, to avoid a metal-to-metal contact between the aluminium and copper, which would cause the former to wear away by local galvanic action. There is no necessity to put an insulating covering on the aerial wire, although indiarubber-covered stranded copper wire as used for electric light wiring is sometimes employed for antennae. A wire 100 feet in length and O'l inch in diameter has an electrical capacity of about 0'0002 microfarad or 180 electrostatic units, from which it will be seen that its energy storage in no case can be very large. Accordingly, when more capacity is required, we must either use several wires or else add metal surface at the top. In some of his early work Marconi employed as an antenna a long strip of galvanised iron wire netting, or else a single wire, with a cylinder of such wire netting placed at the top, or else a kite or balloon having its surface covered with tinfoil. The objection to the use of such capacity areas at the top of the wire is that they offer a greatly increased surface for the wind to act upon. Hence multiple antenna? are generally preferred. 6. Multiple Wire Antennae. Several wires may be arranged in many different forms to obtain an antenna of large capacity. The simplest method is the double cone antenna. In this case a light wooden star is formed by crossing a number of laths of wood like a star. Long wires of equal length are passed through holes in the ends of these laths, and tied together at their ends (see Fig. 8). Two such wooden stars or crosses may be employed, and a cylindrical or fourfold antenna thus made (see Fig. 9). Otherwise, wires may be arranged in fan fashion (see Fig. 10), being joined at the upper end by a wire or rope, and bunched together at the lower ends ; or they may be arranged conically. In grouping together wires in this manner, whilst we increase the capacity relatively to a single wire, the increase in capacity is by no means proportional to the number of the wires. If two wires 100 feet long are hung up vertically parallel to each other, 148 RADIOTELEGRAPHY and about 4 or 5 feet apart, the total capacity is not double that of one wire, because the lines of electric strain proceeding from each wire to the earth are not then distributed symmetrically. Those of each wire disturb the distribution of the other. The nearer the wires are brought, the more we make their joint M FIG. 10. capacity less than the sum of their individual capacities when far apart. Thus, if four wires 100 feet long and O'l inch in diameter are hung up vertically very far apart, their joint capacity would be about O'OOOS mfd., but if brought within 4 or 5 feet of each other, it would not be more than 0'0004 mfd. As a rule, a few wires spaced fairly far apart are better than very many wires near together, as far as total capacity is concerned. The multiplication of the wires in an antenna has, however, another effect. It decreases the inductance of the antenna and also its high frequency resistance. It decreases, therefore, the time period of oscillation in so far as inductance is concerned, although on the whole there is generally an increase in the time period. Its chief advantage is that it enables us to accumulate more energy in the antenna in virtue of the greater capacity. It is an advantage to increase this capacity without adding to the height of the wires, because an increase in height involves more cost in supporting them. One way of doing this is to carry the wire up vertically for a certain height, and then extend it horizontally. This may be done in one, two, or more directions, and we have a gallows-shaped, or T-shaped, or umbrella-shaped antenna. RADIATING AND RECEIVING CIRCUITS 149 This last form, with the radiating wires inclining downwards, is a favourite form, since it is easy to erect, and the wires them- selves can act as stays for the central support. It is also a convenient form for the portable antennae used for military radio telegraphy (see Fig. 11). FIG. 11. There are certain forms of antennae which have the property of sending out electromagnetic waves more in one direction than others, and these are called directive, antennce. They will be considered in a later section. For ship antenna, a special gaff or sprit is attached to a mast to give greater height, and a multiple antenna may be suspended from it, or, for some purposes, horizontal wires may be carried between masts, and vertical wires brought down from them from the middle or from both ends. 7. Earthed and Non-earthed Antennae, From what has been already stated it will be evident that in addition to classification into open and closed antennae, we must distinguish between earthed and non-earthed antenna?. If, for instance, we stretch an insulated wire horizontally, cut it in the centre and introduce a spark gap, we construct a non-earthed horizontal antenna. When traversed by oscillations the lines of magnetic force are arranged round the antennae with their planes vertical, and the electric force is in lines which lie in radial planes passing through the wire. If such a horizontal antenna is placed a little above the earth's surface, then at some distance from the antennae in a direction perpendicular to the oscillator, the magnetic force is 150 RADIOTELEGRAPHY vertical to the earth's surface and the electric force is horizontal. If we turn the above oscillator into a vertical position and place it at some distance above the earth's surface, the magnetic force will then be parallel to the earth's surface. In both these cases we construct what are called non-earthed oscillators. Supposing, however, in the last case the vertical Hertzian oscillator is partly buried in the earth so that one of the rods is completely under- ground, the spark balls being just above the surface, we then construct a vertical earthed antenna, and on considering the distribution of electric and magnetic force round it, it will be seen that the magnetic force is at all points parallel to the earth's surface, whilst near the earth's surface, at some little distance from the oscillator, the electric force is always vertical. If one wire of the antenna is placed vertically and the other horizontally, the spark gap being at the angle where they meet, both wires being insulated, we have a form of non-earthed antenna, in which the horizontal wire is sometimes called the balancing capacity. This capacity may take the form of a sphere or metal cylinder also insulated from the earth (see Fig. 12). On the S////////////S/S//SS FIG. 12. other hand, if a vertical earthed antenna is constructed, and if a condenser of large capacity is inserted between the spark balls and the earth (see Fig. 13), we do not in fact make a non- earthed antenna. A condenser can be traversed by a high fre- quency alternating current, and hence if the capacity is large enough to pass the same current as a dielectric current that would pass as a conduction current if the condenser were removed and the RADIATING AND RECEIVING CIRCUITS conductive earth connection restored, the presence of the condenser does not render the antenna in effect non-earthed. It is well to bear in mind that generally speaking when we are concerned with high frequency currents a condenser acts as a conductor, whilst a large inductance acts as a non-conductor to these currents. A form of radiator which is sometimes employed consists of a sheet of insulated metal placed close to the earth and connected by a vertical wire or wires with another plate elevated above the earth, the wire being interrupted by a spark gap placed near the lower plate (see Fig. 14). The real distinction to be made between these various forms of open circuit radiator is the mode of distribution of the lines of electric force before discharge. In the case of perfectly non- earthed or Hertzian radiators, the lines of electric force which start from one-half of it, ex- tend through space and terminate in the other half, the two parts being separated by the air gap, and constituting the two plates, as it were, of a condenser. In the case of the conductively earthed or Marconi radiator, the lines of electric force before discharge stretch from the antenna and terminate on the earth's surface in its neighbourhood. In the case of antennae comprising a balancing capacity of any form placed near the earth, there are a triple set of lines of electric force. One set extend from the vertical wire to the balancing capacity on which there is an opposite charge of electricity, others from the antenna to the earth, and a third set extend from the balancing capacity to the surface of the earth. If the balancing capacity is not removed a considerable distance from the earth, then during the oscillations we have rapid changes of potential of the earth in the neighbour- hood of the antenna just as in the case of the conductively earthed or Marconi radiator. The difference between the radiating effect of these forms of radiator has been much discussed. All practical experience shows that to produce telegraphic effects at a great distance, the lower ends of the radiating and receiving antennae must be conductively connected to the earth, or what is equivalent to it must be connected to the earth through a condenser of large capacity ; that is to say, currents of electricity must flow into and out of the /T/77/////////////////////////////// FIG. 14. i$2 RADIOTELEGRAPHY earth in the neighbourhood of the antennae. Also the radiating antenna must be so arranged that the lines of its magnetic force are parallel to the earth's surface, and the lines of its electric force at a distance from the antenna and vertical to it. The reason for this is that the propagation of an electro- magnetic wave over the surface of the conducting sea or land requires that the lines of electric force should terminate on the earth's surface perpendicularly to it, or else that they should be detached from the oscillator in the form of complete loops with their planes perpendicular to the earth's surface. The magnetic force is then parallel to the earth's surface. If the oscillator is placed in any other position, say, horizontal, then the energy of its oscillations expends itself more or less in making induced or secondary currents in the earth's surface beneath it, and there is a corresponding diminution in the energy radiated. To produce the most effective radiation, the conduction current in the oscillator should be perpendicular to the earth's surface; it then creates rapid alterations of potential in the earth's surface beneath it, and detaches from itself lines of electric force and therefore radiates, but it is not then so situated as to enable it to expend its energy in creating induced or secondary conduction currents in the earth. It is found, therefore, that for effective long distance radiation the antenna must either be vertical or have a considerable part of its length vertical, and that its lower end must be in such con- nection with the earth, that electricity can flow into and out of the earth out of and into the antenna very freely. Over short distances it is quite possible to produce sufficient radiation for telegraphic purposes with a non-earthed antenna having a balancing capacity, but when long distances have to be covered the direct connection to earth must be adopted or else a condenser of large capacity interposed between the base of the antenna and the earth which is equivalent to an earth connection. The practical construction of the "good earth" required will be considered in the chapter on Eadio telegraphic Stations. As the law of exchanges necessitates an identity in the nature of the radiating and absorbing circuits, it follows that when employing earthed radiating circuits we must employ a similar circuit for receiving and absorbing the energy of the waves radiated. The undoubted advantage gained by the employment of earthed radiating and receiving antennae, and the known fact that the distance at which radiotelegraphy can be conducted with a given energy expenditure depends upon the nature of the earth connection, and of the intermediate soil or surface whether land RADIATING AND RECEIVING CIRCUITS 153 or sea, over which the waves pass, raises the important question of the true function of the earth in this matter. The materials of which the surface crust of the earth is composed are very poor conductors of electricity when perfectly dry. They owe their conductivity chiefly, if not altogether, to the water contained in them. Sea water, owing to the presence of salts in it, is a better conductor than rain or river water. Hence, the conductivity of the soil or sea is essentially electrolytic conductivity. For this reason also it has a high dielectric constant as that of pure water is 80 compared with air as unity. The resistance per centimetre cube of water and soil, and their dielectric constants, are approximately as follows : Specific resistance in ohms per centimetre cube. Dielectric constant k air = i. Sea water .... 100 80 Fresh water 100,000 80 Damp soil Dry soil 10,000 to 100,000 1 000 000 and upwards 5 to 15 2 to 6 Dry rocks practically insulators Electric theory shows that the electric force outside and very near the surface of a good conductor must always be perpendicular to it, and also that the electric force must be zero inside the con- ductor. Hence, if the surface soil were as good a conductor as a metal, and we set up at any point on it a radiating earthed antenna, the electric component of the waves at or very near the earth's surface would be normal to the surface, and there would be no sensible penetration of the wave into it. There would be no absorption of the energy of the waves, but they would glide over the surface without other weakening than that due to the diffusion of the energy over a greater space. If, on the other hand, the soil was a perfect insulator there would be a penetration of the wave into the soil, but no loss of energy by absorption. The sea or soil is, however, a poor conductor or bad insulator, and this implies that there is a weakening of the electromagnetic wave moving over its surface by which some of the wave energy is frittered away as heat in the surface soil. This absorption reaches its maximum for a certain degree of specific resistance, and beyond a certain value of this specific resistance it also decreases with the dielectric constant. This matter has been treated theoretically by J. Zenneck, who has given his results in the form of curves calculated for an assumed wave length of 1000 feet or a frequency of one million. 154 RA DIO TELEGRA PHY Supposing such long electric waves, which are of the wave length mostly used in radiotelegraphy, to travel over surfaces of various conductivities and dielectric constants, Zenneck has cal- culated the distance at which the wave amplitude would be reduced to Ot ^ or to - (where t is the base of the Napierian ,4/JLo logarithms) of the amplitude at the origin, apart altogether from reduction in amplitude due to the spreading of the wave energy over a larger area with increasing distance. His results are exhibited in the curves in Fig. 15, in which the ordinates repre- KILOMETRES 10000 -9 -10 -II -12 -13 -14 -15 -16 -17 SEA I PRPOLJ WATFR ! DAMP WATER I Fl i WATER | SQ|L FIG. 15. DRY SOIL sent this critical distance in kilometres and abscissse the logarithms of the specific resistance of the surface of the material over which the waves travel in ohms per centimetre cube. We have marked the range of resistivity corresponding to sea water, fresh water, and dry soil respectively, and it will be seen that there is an extraordinary reduction in the distance corresponding to a given reduction in the wave amplitude, with increasing resistivity of the surface over which the waves travel. It is this absorption which accounts for the well-known fact that radiotelegraphy is con- ducted with much more difficulty over dry land than over sea water. In radiotelegraphy the radiating and receiving antennae are RADIATING AND RECEIVING CIRCUITS 155 placed at the separating surface of two media, one the air, having an almost perfect non -conductivity and unit dielectric constant, and the other the earth or sea water, having a more or less imperfect conductivity and a dielectric constant greatly exceeding unity. The wave glides over the bounding surface, but partly penetrates into the soil or water, and on one side suffers absorption or loss of energy in so doing, and is thereby weakened. Although, however, the above facts give some reason for the greater difficulty of conducting radiotelegraphy over dry land than over sea, they do not account for the great improvement effected by making the earth connection at both ends. Eadiating and receiving antennae both connected to the earth respond at greater distances for the same energy expenditure than antennas connected to perfectly insulated balancing capacities placed at some distance above the earth. In the former case the two antennas and the earth virtually form one oscillator connected through a conductor of a certain capacity. If we may assume that the upper layers of the atmo- sphere are non-conducting then the capacity of the earth considered as a sphere in space is only about 800 rnfs., or about equal to that of an Atlantic cable, and the addition to or subtraction from it of small quantities of electricity is therefore able to alter quite appreciably its potential relatively to some fixed zero. On the other hand, if the upper layers of the atmosphere have any appreciable conductivity the terrestrial capacity may be much greater. Hence, if we have a conductor which is charged and discharged alternately from or to the earth, it is in fact giving or taking electricity to or from the earth, and therefore raising or lowering the earth's potential at that point relatively to some absolute zero. These sudden changes of potential at one point must be propagated over it, and an oscillation detector placed at a distant spot in the circuit of another syntonic antenna will detect these changes. From this point of view the earthed radiating antenna and the earthed receiving antenna and the earth itself constitute one single oscillator, and rapid variations in the electric distribution at the radiator are felt and detected at the receiver in virtue of operations taking place in the crust of the earth and not entirely confined to the superincumbent dielectric. It has been asserted by Sir Oliver Lodge that the radiation from a non-earthed antenna is less damped than is the case with earthed antenna, but the difficulty is to institute a comparison in which the initial energy storage is the same and all other circum- stances identical except in regard to the connection or not to the earth. 156 RADIOTELEGRAPHY 8. The Establishment of Fundamental and Harmonic Oscillations in Open and Closed Circuits. It is well known that a stretched string, such as a violin string, can not only vibrate as a whole but can divide itself into oscillating sections, having lengths respec- tively equal to one-half, one-third, and one-quarter, etc., of the whole length, in which case it emits notes of higher frequency than when vibrating in a single undivided length. The stationary points on the string are called the nodes, and the places of greatest amplitude of motion the ventral points, antinodes or loops. The skilful violinist can by touching the string lightly at certain places and bowing at other places thus cause a single string to vibrate in harmonics and emit a series of notes having frequencies 2, 3, 4, etc., times that of the fundamental note due to the vibration of the string as a whole. The same is true of the oscillations of the air in an organ pipe, as described in any book on acoustics. In the case of electric circuits possessing capacity and induc- tance, we have a similar phenomenon. Thus we can set up electric oscillations in a linear Hertzian oscillator or Marconi antenna, called the fundamental oscillation. In this case there is no variation in potential at the centre of the Hertzian oscillator or at the earthed end of the Marconi antenna, and this point is therefore called a node of potential. At the free ends or upper end the variations of potential are a maximum, and these points are therefore called loops or antinodes of potential. The variations of potential increase from the centre or lower end to the free or upper end, and in the case of the Hertzian oscillator the charges at the two free ends are always opposite signs. Hence, we may represent these varia- tions of potential by the distance of a dotted curve from a thicker line repre- senting the oscillator or antenna (see Fig. 16 (a)). At the same time there are variations in the conduction current in different parts of the oscillator or antenna which may be represented in the same way by a fine, firm line. At the centre of the Hertzian oscillator or the earthed end of the Marconi antenna the amplitude of the current is a maximum, and that point is called an antinode or FIG. 16. RADIATING AND RECEIVING CIRCUITS 157 loop of current. On the other hand, at the free end, the conduction current is zero, and therefore these points are nodes of current. The current therefore increases gradually from the node to the antinode, and is not the same at all points on the antenna. Its amplitude at any point may thus be represented by the distance of a fine, firm line from a thick line representing the oscillator, as in the diagrams in Fig. 16 (a). Again, we may set up oscillations in an open or closed circuit, which are called harmonics of the fundamental. For example, in the case of the earthed Marconi antenna, we may set up a first harmonic oscillation in which in addition to the node of potential at the earthed end, there is another i node of potential at about one-third of the length of the rod from the open end, and a node of current at about one-third of the length of the rod from the earthed end, the variation of potential and current amplitude along the rod being represented by the fine dotted and firm lines, as in Fig. 16 (6). A second harmonic may also be set up in which there are two nodes of potential in the rod in addition to one at the earthed end, and a corresponding distribution of current, the rule being that the earthed end must always be a node of potential, and the free or insulated end an antinode or loop of potential, whilst the earthed end is an antinode of conduction current and the free end a node of conduction current, nodes of current coinciding with the antinodes of potential, and vice versa (see Fig. 16 (c), (d)). In the case of a linear Hertzian or non-earthed symmetrical oscillator having the spark gap at the centre, there is at that point a node of potential and an antinode of current. If we consider the nature of the distribution and the lines of electric force round a vertical earthed antenna, we sfee that in the case of the fundamental oscillation the lines of electric force stretch from the antenna to the earth in all directions round it, and when oscillations are excited by the discharge, these lines are detached as semi-loops of electric force with their feet on the earth. When, however, harmonic oscillations are produced in the antenna, we have as it were a superposition of a complete non-earthed antenna, on the top of an earthed one, and we must therefore have a detachment not only of semi-loops but of complete loops of electric strain which move outwards into surrounding space, and together with the corresponding expanding circular lines of magnetic force which together constitute the radiation. When a plain earthed antenna has electrical oscillations excited in it which are the first harmonic of its fundamental, the frequency of these oscillations is three times that of the i$8 RADIO TELEGRA PH Y fundamental oscillation, whilst the wave length is one-third. In the case of the second harmonic the frequency is five times and the wave length one -fifth of that of the fundamental. In the case of a closed or nearly closed oscillatory circuit, we can also set up oscillations which are either a fundamental or a higher harmonic. Thus, in the case of a closed circuit consisting of a condenser and a circle of wire, with a spark gap opposite to the condenser (see Fig. 17), the fundamental oscillation involves a conduction current to and fro in the circuit with a node of potential at the centre of the spark gap and loops or antinodes of potential at the condenser plates. The condenser plates always FIG. 17. FIG. 18. carry electric charges of opposite sign, so that the distribution of potential for the fundamental oscillation is represented in Fig. 17 by the radial distance of the dotted line from the thick black one representing the oscillatory circuit. The amplitude of the conduction current in the wire may be represented by the radial ordinate of another line, and this is a maximum at the spark gap, and has a minimum but not zero value at the condenser surfaces. In the first harmonic oscillation of such a circuit there are three nodes of potential and three antinodes, as shown by the radial ordinates of the dotted line in Fig. 18, and similarly, three places of minimum amplitude of the conduction current. A second harmonic can also be excited with five nodes, and a third with seven nodes of potential and so on, the frequency being pro- portional to the number of nodes and the wave length inversely as the number. This production of fundamental and harmonic oscillations in an antenna can be very beautifully exhibited by an experiment RADIATING AND RECEIVING CIRCUITS 159 devised by G. Seibt, which has been improved by the author. A very long helix of fine silk- covered wire is made by closely winding in one layer the covered wire on a long round rod of ebonite. The helix constructed by the writer was nearly 2 metres or 80 inches in length, and 5 cms. or 2 inches in diameter. This helix is supported on insulators in a horizontal position about 2 feet above a table. A closed oscillatory circuit is then formed of a condenser coDsisting either of one or more Ley den jars Ci C 2 or of sheets of ebonite coated with metal and placed in oil and a variable inductance made as described in Chapter II. (see Fig. 19). Also a FIG. 19. spark gap, S, must be provided in this oscillation circuit consisting of a pair of zinc or brass balls adjustable as to distance, and enclosed in a wooden box. These balls are connected to the secondary terminals of an induction coil, and one of them is connected to the earth E. The helix is connected to this oscillatory circuit as shown in Fig. 19. The first step is to so adjust the capacity and inductance in the oscillatory circuit as to tune it to the natural time period of the helix. In the case of the helix made by the author, the helix consisted of 5465 turns of wire on a rod 215 cms. in length, and its natural time period of oscillation was nearly ^ooVoo" ^ a secon d or about 5 microseconds. The condenser used in the oscillation circuit had three sections of capacity 1461, 2887, and 5835 micro- microfarads respectively. The inductance used could be varied from 5000 to 120,000 absolute units or from 5 to 120 microhenrys. The oscillation circuit was in tune with the helix when the former was composed of an inductance of 110 microhenrys and a capacity of 5885 micro-microfarads, since these last constants correspond to a frequency n, where, ' 5-033 X 1Q = >/0.-005885 X 110000 which is equal to a time period of nearly 5 microseconds. 160 RADIOTELEGRAPHY When oscillations are set up in the condenser circuit by connecting the spark balls to an induction coil they will excite other oscillations in the helix, and these create around it an electric field which may be detected by holding near the helix a vacuum tube preferably filled with Neon. If this tube is held in various positions, it will be found to glow with increasing brilliancy as moved from the condenser end towards the free end of the helix, owing to the gradual increase in the potential amplitude as the free end is approached. If, however, the inductance and capacity in the condenser circuit are altered so as to make the frequency 3, 5, or 7 times that of the fundamental oscillation of the helix, it is found that a state of vibration is set up on the helix in which there are nodes of potential near which the vacuum tube does not glow. Thus we can set up a state in which there is one such node of potential about one-third of the length of the helix from the free end, and likewise states in which there are 2, 3, etc., such nodes. 9. Modes of Exciting Oscillations in an Open or Closed Radiating Circuit. We have next to consider the various modes of exciting oscillations in a radiating circuit. The simplest method is the production of damped oscillations by inserting a spark gap in the circuit. Thus, to excite such oscillations in an open earthed plain Marconi aerial, a spark gap is inserted near the earth. The spark balls are connected to an induction coil or transformer, and the antenna is there- fore charged either intermittently or alter- nately, and discharged across the spark gap. In this case the charge is limited by the capacity of the antenna and the voltage which the coil or transformer can give, and this process results, as we have seen, in the production of highly damped oscillations. The energy stored is necessarily small, and hence is soon frittered away (see Fig. 20). In the case of a closed circuit, including a condenser of large capacity, a much larger energy storage is possible, and since the cir- FIG. 20. cuit is a worse radiator than an open circuit, the trains of oscillations, and therefore of the waves emitted, are less damped, and contain more oscillations per train. This method of excitation by a spark gap is called the direct excitation. The second method is a method of direct coupling. In this case oscillations are excited in a closed circuit containing a RADIATING AND RECEIVING CIRCUITS 161 condenser inductance and a spark gap, and since the circuit is a bad radiator and the condenser can have a large capacity, a considerable storage of energy is possible. To this circuit at one point is con- nected an antenna or good radiating circuit (see Fig. 21), and another point on the closed circuit is con- nected to the earth. The antenna or open circuit has its own natural period of vibration like that of the closed circuit, and the two must therefore be syntonised together. This is most easily achieved by inserting an inductance coil between the antenna or open circuit and the earth, and making a variable part of this inductance coil by means of a movable contact the inductance in the closed circuit (see Fig. 21). In this manner the oscillations of the two circuits can be varied until they are equal, and this equality can most easily be discovered by connecting a hot wire FlG - 21 - voltmeter across one or two turns of the inductance coil near the foot of the open circuit and then varying the inductance or capacity in the open and closed circuits until this hot wire voltmeter gives its maximum reading. The two circuits are then said to be tuned together. With such an arrangement we provide a much larger storage of energy than is the case in the direct method of excitation, since we associate with a feebly radiative closed circuit of large energy storage power a good radiative or open circuit. The method of direct excitation may be compared with a thermal radiator, in which a small quantity of a hot fluid, say water, is enclosed in a metal vessel covered with lamp black. The water then cools quickly for two reasons : first, because the surface of the vessel is a good thermal radiator ; and secondly, because there is only a small store of liquid to cool. If, however, we were to connect this vessel with a reservoir made of polished metal containing a larger store of the hot fluid, then by the continuous circulation through the small but good radiative vessel of the large mass of hot fluid from the other the radiation could be maintained for a much longer period. A third method of exciting the oscillations in an open radiative circuit is by the method of inductive coupling, using an oscillation transformer. In this case, we insert in the open radiative circuit one coil of a transformer comprising two interwound coils of wire, the second coil forming the inductance of the closed oscillatory M 1 62 RADIO TELEGRA PHY FIG. 22. circuit. Hence, when oscillations are excited in the latter by means of a spark gap, this will induce other oscillations in the open or radiative circuit, provided that the two circuits are brought into tune or syntony with each other (see Fig. 22). For this purpose both circuits must be provided with variable inductances, as was first done by Marconi, so that by variation of the inductance in the open circuit and variation of the capacity or the inductance in the closed circuit, the two circuits may be brought into tune with each other. We may discover when this is the case by con- necting a hot wire voltmeter over one or two turns of the inductance placed in the open radiative circuit and then altering the inductance in one or both circuits, or the capacity in the closed circuit until the reading of this voltmeter becomes a maxi- mum. The two circuits are then tuned. Unless this is done, the oscillations in the closed circuit will produce very little effective radiation in the open circuit. The two circuits, however, may be so adjusted that the time period of one circuit is a harmonic of that of the other, in which case they will operate as if coupled for fundamental oscillations. There is a fourth method of connection called the electrostatic coupling, which however is not frequently employed. In this case, the open or radiative circuit terminates in a plate which is placed in contiguity to one of the plates of the condenser in the closed circuit. An arrangement employed by Hertz and also by Lecher makes use of this method of coupling. Oscillations are excited in an open or Hertzian oscillator, consisting of two plates at the ends of two rods, these rods being in one line and separated by a spark gap. In apposition to the two plates of the Hertzian oscillator are placed two other plates which are respectively connected to two long wires. These long wires must have their capacity adjusted by varying their length until they are in syntony with the primary circuit in which the oscillations are set up, and these last will then create secondary oscillations in the long wires, which will attain their maximum amplitude on exact syntonisation. The most usual modes of connecting the good radiating circuit to the energy-storing circuit are by the direct or inductive coupling. The latter method has the great advantage that the closeness of the coupling can be varied, and hence also the nature of the oscillations RADIATING AND RECEIVING CIRCUITS 163 sec up in the secondary circuit. We have already shown that when two circuits having the same time period of oscillation are inductively connected, then, on establishing free oscillations in one circuit, oscillations of two frequencies are set up in both circuits, and that if the coupling is close these two frequencies are well separated, but if the coupling is loose they are merged into one. Hence, if we desire to set up in an antenna induced oscillations of one definite period the antenna must be connected in series with the secondary coil of an oscillation transformer, the primary coil of which is in the closed or condenser circuit, and the two coils must be well separated. If, however, the coupling is close we then have oscillations of two frequencies set up in the antenna and waves of two wave lengths radiated. It can be proved by experiment and by theory that in the last case the wave train of largest wave length is the least damped, and the train of shortest wave length is most damped or has fewest oscillations. If the closed and open circuits respectively when separated have decrements Si and 2 then when coupled together with a coupling coefficient k Wien has shown that the decrements of the two waves radiated by the open antenna will be DI and D 2 , such that D 1= Thus, for- instance, in an experiment made with a coupled antenna at University College, London, the open antenna had an oscillation constant of its own of 7'0, and hence a frequency 5 x 10 6 n = = = 072 x 10 6 , corresponding to a wave length Ao = 1400 feet. This is the wave length of the wave which would have been emitted if this antenna had been used as a plain self-excited antenna. Its decrement S 2 when so used was found to be 0'175. It was then coupled to a closed circuit containing a condenser of capacity 0'025 microfarad and an inductance of 2 microhenry s, and had therefore a natural frequency of 071 X 10 6 . It had a decrement Si = 0'09 when the spark gap in the circuit had a length of 2 mm. Accordingly, when the two circuits were coupled with a coupling coefficient k = 0*5 waves of two wave lengths were emitted from the antenna, viz. 164 RADIOTELEGRAPHY Ai = AoVT+1 = 1400 x 0-7 = 980 feet A 2 = AoVlT^ = 1400 x 1-224 = 1714 feet and these had decrements DI and D 2 , such that 0-09 + 0175 2 X 1-224 .The longer wave of 1714 feet has the least damping, and therefore the longest train of waves. In the case of the direct coupled antenna there is also an emission of waves of two wave lengths when the open and closed circuits are syntonised. In this case, however, the difference between their wave lengths depends upon a coefficient p, which is the square root of the ratio of the capacity c of the antenna with respect to the earth to the capacity C of the condenser in the closed circuit, and the wave lengths of the two waves emitted are given by Ai = A \/l + p A 2 = Ao\/l p where p \/ Hence, if c is small compared with C, waves of only one wave length are radiated. This is usually the case, and the mode of direct coupling is therefore one which is very generally employed. 10. Appliances for giving Direction to Electromagnetic Eadia- tion. Directive Antennae. We have seen that in the case of a vertical open circuit antenna the radiation is necessarily sym- metrical in all directions, and is, therefore, equally detected by receiving antennae placed at the same distance round it in all azimuths. A problem which presented itself very soon in connection with radiotelegraphy was the limitation of this uniform all round radiation to a certain direction. It was obvious that some means was required to effect that which a lens or mirror effects in the case of light. Hertz had shown that electromagnetic radiation of short wave length could be reflected by metallic mirrors, and followed the laws of reflection of light. By means of parabolic RADIATING AND RECEIVING CIRCUITS 165 mirrors he thus concentrated electromagnetic radiation into a beam. This, however, can only be done if the wave length of the radiation is not large compared with the dimensions of the mirror. Thus, for instance, Hertz placed two metal rods, each about a foot in length, in line with each other, and placed them both in the focal line of a parabolic cylindrical mirror. At a distance he placed another similar mirror with a receiving antenna in its axis, and, on setting up oscillations in the transmitting oscillator, he was able to direct a beam of electromagnetic radiation on to the other mirror and concentrate it in the focal line and hence detect it. Marconi was successful in projecting electromagnetic radiation by this means for a distance of about 2 miles when using waves of short wave length. But this can only be done by the aid of mirrors when the dimensions are comparable with the length of the. waves employed. In radiotelegraphy, however, the wave length of the waves now employed is from 500 or 1000 up to 10,000 feet or more in length, and hence there is no possibility of constructing mirrors of sufficient dimensions to concentrate such radiation in any required direction. A new line of investigation was, however, opened up by the observation that the radiation of a sloping antenna, and particularly that of a vertical loop or closed circuit radiator, was not symmetrical. In the case of the closed circuit it is greater in the plane of the loop than at right angles to it. Some preliminary investigations concerning this phenomenon were made by Sigsfield, Braun, Zenneck, Strecker, Slaby, Garcia, de Forest, and others, on the non-symmetry of radiation of inclined open antennae, and Stone and de Forest, noting also cases of asymmetry in receiving antennae, suggested means for locating the direction of an electro- magnetic wave. But although claims were made for arrangements said to be effective, these various researches were not pressed to such logical issue as to disclose any definite scientific principle, whilst in some cases results said to have been obtained are clearly in contradiction to well ascertained facts. The problem of loca- ting the direction in which an incident wave was arriving seems to have first attracted attention. If two vertical antennae are erected on a plane at a distance apart equal to half the wave length of electromagnetic waves travelling over that plane, which have their magnetic force horizontal and electric force vertical, then their action upon these two antennae will depend upon the direction or propagation of the waves. If, for instance, the waves are travelling in a direction parallel to the plane in which the two antennae are placed, oscillations will be created in these two receiving antennae which are opposed to one another in phase. If, i66 RADIOTELEGRAPHY however, the wave is travelling in a direction perpendicular to the plane containing two vertical antennae, then the oscillations set up in them will be coincident in phase. If, then, these two vertical antennae are insulated from the earth, and horizontal wires are brought near the earth from their base to a middle point and then earthed at the middle point, it is possible to make the oscillations propagated .along these horizontal wires combine together at their junction in their action upon some form of oscillation detector made as described in the next chapter, so that when the oscilla- tions are of the same phase, the oscillation detector is affected, but when the oscillations in the antennas are opposed in phase, the oscillation detector is not affected. If, then, we could move round the two antennas into various positions, keeping them half a wave- length apart, we could ascertain the direction in which the waves are travelling by ascertaining how the antenna must be placed to produce the greatest effect. It will be obvious, however, that when dealing with waves of a thousand feet or more in wave length, this movement, although possible in idea, is not practicable in fact, and accordingly the method, although theoretically correct within certain limits, fails to give a solution of the problem for practical purposes. The first real solution of the problem came from an observation made by Marconi that if an antenna is bent so as to have a short part of its length vertical and the greater part of its length horizontal, the lower end of the vertical part being connected to the earth and the outer end of the horizontal part being insulated, then such an antenna provides the means for producing a non- symmetrical radiation, and also for detecting the direction in which electromagnetic waves are passing over it (see Fig. 23). He found O To Induction. Coil FIG. 23. that such a bent antenna emits a less intense radiation at any given distance in the direction in which the free end points than in the opposite direction. Also, since the law of exchanges holds good for electric radiators, this form of bent antenna receives or absorbs best electric waves which reach it from a direction opposite to that in which the free end points. Hence, two similar bent antennae when set up back to back, that is, with their free ends RADIATING AND RECEIVING CIRCUITS 167 pointing away from each other, form a system of radiator and receiver which has a greater range in that position than in any other at the same distance, and hence has directive qualities not possessed by the ordinary vertical antennae. Although the full explanation of this phenomenon requires the application of some- what advanced mathematical analysis, it is not difficult to give a general explanation of it in non-symholic language. Consider, for example, a square or rectangular circuit A, B, C, D (see Fig. 24), B E 4-H+h' +H-h H+b' H-h FIG. 24. in which electric oscillations are taking place. The magnetic field of such an oscillator consists of closed lines which embrace the circuit of the oscillator. These lines are all perpendicular to the plane A, B, C, D in that plane, and if the current is in any instant going round in the same direction to the hands of a watch, that is, in the direction A, B, D, C, the lines of magnetic force in the included space are proceeding away from the reader, and returning on all sides outside the area towards the reader. If we represent the section of these lines of force by little circles, and indicate that the line is coming towards the reader by a dot put in the centre, and that it is moving away from the reader by a cross, then we can represent the section of a pair of such lines of magnetic force outside the oscillator returning back on both sides in the equatorial line by the two little circles marked 4- H, in which the magnetic flux is towards the reader. On the other hand, if we consider a simple open antenna EF of the same height as the side of the rectangle BD, and consider the nature of the magnetic force round it when a current is flowing upwards in it, it will be seen that these lines are circles lying in planes at right 168 RADIOTELEGRAPHY angles to the antenna, and that the sections of these lines in that plane may be represented by the little circles -f ti and h marked respectively with a dot and a cross. If, then, we suppose the open and closed circuits to be placed so that the open one is in close contiguity to one side of the closed one (see Fig. 24), and that the oscillations in these parts of the two circuits in contiguity are always in opposite directions, then it is quite easy to see that the field due to the open circuit antenna will assist the field due to the closed circuit antenna on the left-hand side, but tend to weaken it on the right-hand side. So that if we call the field due to the open antenna on the one side A, and on the other side ti, the result in the field due to the combined open and closed antennae will be H 4- ti on the left-hand side, and H h on the right-hand side. We can now imagine the two oscillations in the continuous wires BD, EF which are opposed in direction to annihilate each ether, and the result is that we are left with a bent antenna as in the lower Figure 24, in which if oscillations are set up we are able to produce a field which is non -symmetrical, being greater on the side away from, which the open ends point. Such an antenna is called a bent antenna, and if we imagine it half buried in the earth, the surface of the earth being a plane of zero potential, it produces the same effect above the earth's surface as one-half of a complete double bent antenna. It follows, then, that an earthed antenna partly vertical and partly horizontal must produce a non-symmetrical radiation. 1 Marconi made this discovery experimentally as follows : Setting up at some place a bent antenna as above described, he took observations of the strength of the field, that is, of the intensity of the radiation by means of a vertical receiving antenna placed at equal distance but in various directions around the bent antenna. Marking off then on a polar diagram of radial lines (see Fig. 25) the intensity of the radiation in different azimuthal directions, he obtained a closed curve something like a figure 8 with two unequal loops, the radii of this curve representing the intensity of the radiation for various angular directions round the bent transmitter. It will be seen that the radiation is greatest in one direction, and that is the direction away from which the free 7 The above almost self-evident explanation of the action of a bent antenna has, however, not been accepted by German writers, although it is essentially confirmed by the experiments of Bellini and Tosi described below. J. Zenneck has advanced another theory in which he states that as far as regards the bent receiver antenna, the asymmetry depends on the alternating field of the trans- mitter being inclined to the vertical, and having therefore a horizontal component. See Science Abstracts, Vol. II. B., abs. 705, June, 1908. RADIATING AND RECEIVING CIRCUITS 169 310 350 10" rro 100 20.0" 190 i8ff> 170"^ PIG. 25. end of the bent radiator points. It is also a minimum in another direction approximately 110 from the maximum direction, and it has a secondary or intermediate minimum 180 in the opposite direction, that is, in the direction in which the free end of the bent antenna points. The shape of this curve can be fully accounted for theoretically by assuming as above that the bent antenna is a combination of a closed or magnetic oscillator and an open or electric oscillator. A large number of ob- servations were obtained by Marconi with bent transmit- ting antennae and vertical or open receiving antennae, and also with vertical or sym- metrical radiating antennae and bent receiving antennae placed in various relative positions, and these observa- tions all confirm the statement made above that an antenna which radiates best in any one direction absorbs best as a receiving antenna, waves which are coming from that direction, and also that when an antenna is constructed which is partly vertical and partly horizontal, the radiation is non- symmetrical, being greater in some directions than in others. Marconi's observations were made with radiating and receiving antennae from 30 to 45 metres in length, separated by distances varying from about 250 metres to 600 or 700 metres, and he then found that for the same distance between the antennae the intensity of the radiation, as measured by a thermal or magnetic oscillation detector, was sometimes as much as four times greater in the direction away from the free end of the bent radiator pointed than in the same direction. The wave length of the waves used in his experiments was about 150 metres, and hence the maximum distance at which experiments were carried out was only about four or five wave lengths. Practical experience, however, shows that the same directive qualities exist at very much greater distances, but theory points to the fact that at extremely large distances the asymmetry tends to vanish, and that any bent oscillator, however arranged, has no asymmetry of radiation for very large distances. In one experiment he employed a horizontal wire 100 metres in length, placed at a slight distance above the earth's surface, and 1 70 RADIOTELEGRAPHY connected at one end through a spark gap with the earth. Such a transmitter sent out waves approximately 500 metres in length. The receiving antenna was a vertical wire 8 metres in length, tuned to the period of the transmitter by means of a syntonising coil and connected to the earth through a magnetic oscillation detector (see next chapter). The signals were quite distinct at 16 kilometres when the horizontal part of the radiator pointed away from the receiver, but only very weak at 10 kilometres when the free end of the transmitter pointed towards the receiving wire, and quite undetectable at 6 kilometres when the free end of the transmitter pointed at right angles to the line joining the transmitter and receiver. Again, at Clifden, Connernara, Ireland, by means of a hori- zontal conductor 230 metres in length as a receiving antenna, and connected into the earth through a magnetic oscillation detector, Marconi found it possible to receive with clearness all the signals transmitted from the Poldhu station at a distance of 500 kilometres, provided that the free end of the horizontal receiving antenna pointed directly away from the direction of Poldhu, whilst no signals at all could be received if the horizontal wire at Clifden made an angle of more than 35 with the line of direction of Poldhu. Furthermore, he found that he could receive signals from the Admiralty Station on the Scilly Isles at Mullion in Cornwall, a distance of 85 kilometres, by means of a horizontal receiving antenna 50 metres in length placed 2 metres above the ground, one end of the wire being connected to the earth through a magnetic oscillation detector provided that the free end of the wire at Mullion pointed away from the Scilly Isles, but that no signals could be received if the horizontal portion was swivelled round so as to make an angle of more than twenty degrees with the line joining Mullion with Scilly. Also by means of a horizontal wire 60 metres in length, supported 2 metres above the ground and being connected at one end to the earth through a magnetic oscillation detector, Marconi was able to locate the direction of an invisible ship sixteen miles away, sending out electromagnetic waves, by noticing the direction in which the free end of the horizontal receiving antenna had to be placed in order to make the signals most strong. This direction was a direction opposite to that from which the waves were arriving (see Fig. 26). Some experiments of the same kind were made by the author in the same year. A vertical radiating antenna was employed consisting of a single wire which could be bent over at various heights from the ground, so as to make a bent antenna partly RADIATING AND RECEIVING CIRCUITS 171 vertical and partly horizontal, the ratio of the horizontal to the vertical lengths being varied at pleasure. A vertical receiving antenna was employed at distances varying between 80 to 150 feet, and in the receiving an- tenna a hot wire oscillation detector of the thermoelectric type (see Chapter VI.), de- vised by the author, was em- ployed to measure the E.M.S. value of the current created in the receiving antenna. The transmitting antenna had its horizontal part swivelled round in various directions at intervals of 15, and in the several positions the current created in the receiving an- tenna was measured, the oscil- lations being excited in the transmitting antenna by means of a spark gap of con- stant spark length. The total length of the transmitting antenna was 20 feet, and the height of the receiving antenna was the same length. The following table shows the current in the receiving antenna in arbitrary units for each position of the horizontal part of the transmitting antenna Hadiation from a Bent Earthed Transmitting Antenna 20 feet in total length. Receiving Antenna vertical and 20 feet high. Distance between receiver and transmitter 138 feet. Length in feet of vertical part of Transmitter of Transmitter FIG. 26. of horizontal part x if f* j. r 15 16 17 18 19 ength in feet .... 100 100 105 106 110 mzontal part of angular degrees. Current in the receiving Antenna in arbitrary units. 100 100 100 100 100 15 98 97 94 92 93 30 92 85 96 83 75 45 82 79 79 77 67 60 78 74 70 71 58 75 77 67 59 56 45 90 72 66 57 52 48 05 71 65 57 46 41 20 70 66 62 53 49 35 72 64 60 54 48 50 73 80 58 67 59 65 70 74 56 69 60 80 82 69 64 63 68 172 RADIOTELEGRAPHY These observations clearly confirm Marconi's observations that the radiation from a bent antenna is imsymmetrical, being greatest in a direction opposite to that towards which the free end of the antenna points. It was also found that by bending down the free end towards the earth, as in Fig. 27, the radiation became still more unsymmetrical, as shown by the polar curve in Fig. 27, liter iver 21 ^ t. vertical istance 138 135 FIG. 27. in which the radii represent the strength of the currents in the receiving antenna corresponding to various relative positions of the horizontal or inclined part of the transmitting antenna. It will be seen from Fig. 27 that the tipping down of the horizontal part causes nearly the whole of the radiation to be sent out towards that side opposite to which the free end points. Another entirely different method of giving direction to electric waves has been devised by F. Braun, which depends upon the interference of electric waves travelling in the same direction but different in phase. In Braun's method, three simple vertical wire antennae are set up in positions corresponding to the angular points of an equilateral triangle, and oscillations are created in these antennas which differ from one another in phase. These oscillations with phase differences were produced by a method devised by Papalexi and Mandelstam. By these arrangements it is possible to cause the waves emitted by the free antennae to combine together and assist one another in certain directions, but to neutralise .one another in certain other directions. It is well RADIATING AND RECEIVING CIRCUITS 173 known, for instance, as described in books on Optics and Acoustics, that waves of light or waves of sound can in this way interfere, so that two light waves may actually destroy one another and produce darkness, and two sound waves neutralise each other and produce silence. This effect is called the interference of waves. Braun found that by a proper arrangement of the antennae and adjust- ment of the phase difference, the radiation of the three antennae could be combined together in a certain region out of the whole azimuth of 360. The experiments were carried out on a large open space near Strasburg. Wooden poles 20 metres high were planted at the corners of an equilateral triangle whose sides were 30 metres long. Antennae wires each approximately 33 metres long terminated in wire netting stretched parallel to the ground and at a small distance above it. These constituted the balancing capacities. In the centre of the triangle an observation hut was constructed from which the wires ran out horizontally to the masts at a height of 2| metres above the ground. At a distance of 1300 metres a receiving station was constructed and a receiving wire erected attached to a pole 20 metres high. In the circuit of this receiving wire was placed a hot wire oscillation detector (see Chapter VI.), by means of which the current in the receiving wire could be measured. In a number of the experiments the oscillations in two of the transmitting antennae were of the same phase, but differed from these in the third antenna by a definite amount, say, by 100, this definite difference of phase being secured by the method of producing multiple spark discharges due to Mandelstam and Papalexi. The amplitude of the oscillations in the two antennae in the same phase was half that in the third antenna. Under these conditions, if observations are taken of the current in the receiving antenna at equal distances, but in different azimuths round the triple transmitter, it is found that in one direction the radiation is a maximum, and in the opposite direction it is nearly zero, -varying in accordance with the radii of a polar curve, as shown in Fig. 28. The method, although ingenious, has not the simplicity and practicality of the bent receiving and transmitting antennae em- ployed by Marconi. Another very ingenious system of directive radiotelegraphy has been devised by E. Bellini and A. Tosi. They employ a nearly closed circuit transmitting antenna, consisting of two aerial wires suspended from one mast, the upper ends being insulated and the lower ends brought into a signalling house, the wires being 1/4 RADIOTELEGRAPHY stretched out, as shown in Fig. 29, so as to give them the form of a triangle. If oscillations are set up either by the direct coupled 120 S3* 150 330 300 III II FIG. 28. FIG. 29. or inductive method, radiation takes place from this nearly closed antenna which is not symmetrical, and is greatest and equal in the two directions in the plane of the antenna and zero at right angles to that plane ; in other directions varying in accordance with the radii of a figure of 8 polar curve, as shown in Fig. 30 (a), (&), (c). Fig. 30 (a) shows a theoretical curve of which the radii in various RADIATING AND RECEIVING CIRCUITS 175 directions are proportional to the energy sent out in these directions by a closed circuit oscillator with plane perpendicular to the paper, and in the direction of the maximum radius. Fig. 30 (&) shows a curve obtained by actual observations, employing the Duddell thermoammeter (see chapter VI.) to measure the energy radiated in various directions. Fig. 30 (c) is a similar polar curve, the radii of which denote magnetic field strength in various azimuths at equal distances from the transmitter. These inventors employed a transmitting antenna and a receiving antenna of the same form. When used as a receiving antenna the oscillation detector, of whatever type it may be, is placed at the centre of the lower or horizontal side of the triangle. When such a circuit is used for reception the intensity of the oscillations created in it by the incident waves is a maximum when the plane of the circuit coincides with the direction of the waves, and zero when it is at right angles to it. Such a circuit may be employed, therefore, to discover the direction in which its waves are travelling which fall upon it by swivelling round the circuit into various positions around its vertical axis, but Bellini and Tosi prefer to construct and erect two such circuits at right angles to one another at each station. Each of these circuits contains in its lower part a coil which can be acted upon inductively or can act inductively upon another circuit RADIOTELEGRAPHY placed in aii intermediate position, which last circuit either contains the oscillation producing arrangement, if it is a trans- mitter, or the oscillation detecting arrangement, if it is a receiver. The arrangement is shown in Figs. 31 and 32, the pair of closed Detector Figs. 29, 30, 31, 32 are reproduced from " Electrical Engineering " of November 14, 1907, by permission of the proprietors. FIG. 32. antennae forming the transmitting and receiving arrangement respectively are placed with their planes at right angles, and the coils to be acted upon inductively, which are inserted in their lower portions respectively, being shown in plan and therefore at right angles. A third coil, which forms as it were the primary or secondary circuit of an oscillation transformer, is placed close to and within the other two coils just mentioned, and in the transmitter this last-named coil is connected respectively with the condenser and a spark gap, and in the receiver with an oscillation detector. The coil in which the oscillations are either set up or are detected is capable of being swivelled round so as to be parallel with either of the coils contained in the circuits of the pair or closed antennae. Supposing now the waves are incident on the receiving arrangement coming from a certain direction. In order to determine that direction, all that it is necessary to do is to swivel round the secondary coil in direct connection with the oscillation detector so as to place it parallel to one or other of the RADIATING AND RECEIVING CIRCUITS 177 primary coils inserted in the closed circuit antennae or in some intermediate position. Some position will then be found in which the indications of the oscillation detector are a maximum, and when that is the case, the waves must be falling on the compound antennae in the direction of the plane of the secondary coil attached to the oscillation detector. In the same way, to send out radiation, which is a maximum in any given direction, the coil in which the oscillations are being produced is swivelled round so as Reproduced from "Electrical Engineering" by permission of the proprietors. FIG. 33. to be parallel to one or other of the secondary coils inserted in the circuits of the two closed circuit antennae, and the radiation will then be a maximum in the direction in which that primary coil points. Experiments with this system showed that good results could be obtained with an expenditure of less than 500 watts between Dieppe and Havre (55 miles overland) and Dieppe and Barfleur (110 miles over sea). The angles between the stations, Dieppe- Havre-Barfleur is 23, but the Dieppe-Barfleur transmission did N 178 RADIO TELEGRAPHY not affect the Havre, nor did the Dieppe-Havre transmission affect Barfleur. The height of the antennae was 48 metres, the wires being 60 metres long at the base and 60 long in the inclined side, forming an equilateral triangle, each side of which was 60 metres in length. It was also found that this closed circuit system was more proof against disturbances from atmospheric electricity (to which further allusions will be made in Chapter VII.) than the system employing open circuit antennae. The great advantage of the methods of Bellini and Tosi is Reproduce! from "Electrical Engineering " by permission of the proprietors FIG. 34. that no movement of the antennae themselves is required to locate the direction of the radiant centre or to give direction to the radiation. The only movable part is a small coil which acts upon, or is acted upon, by the fixed antennae. This arrangement for locating the direction of a station is called a radiogoniometer by its inventors, and promises to be of considerable use in connection with radiotelegraphy. By employing a single vertical antenna in combination with two nearly closed antennae at right angles to each other (see Fig. 33), Bellini and Tosi have been able to confine the radiation of this compound antenna entirely to one side ; so that the polar curve of radiation is represented by a cardiod (see Fig. 34), the antenna being situated at the cusp. It may be noted also that A. Artom had previously employed bent antennae for radiotelegraphy, and also independent crossed straight antennae with the object of creating circular and elliptically polarised electric waves. CHAPTER VI OSCILLATION DETECTORS 1, Classification of Oscillation Detectors. In the previous chapters it has been shown that electric oscillations set up in an open or closed radiative circuit create an effect in the space around which is propagated through the dielectric as a wave, and that when these waves impinge in the right direction upon another open or closed syntonic receiving circuit placed at a distance, they set up in the latter similar oscillations, which, however, are far more feeble than the oscillations in the radiating circuit. These induced oscillations are not directly appreciable by our senses, and their existence can only be ascertained by the employment of special appliances, called oscillation detectors, in- serted in or associated with the receiving circuit. The oscillation detector in its turn can be made to actuate some form of tele- graphic or telephonic instrument, and thus to make evident to the human eye or ear the commen cement, end, or continuance of the oscillations set up in the transmitting circuit. In this chapter we shall consider the various forms of oscilla- tion detector which have been invented. Of whatever type they may be, an oscillation detector is an appliance which enables us to detect the existence of a very small, high frequency alternating current in a circuit, or alternating difference of potential between two points on it. Hence, a first classification of oscillation detectors may be made by dividing them into potential indicators and current indicators. From this point of view they may be regarded as very sensitive forms of alternating current voltmeter or ammeter adapted for detecting or measuring exceedingly feeble but high frequency alternating electromotive forces or currents which exist in a circuit traversed by electrical oscillations. A large number of forms of oscillation detectors have now been devised, depending upon the power of electric oscillations to effect various changes, as follows : 1st. The simplest and the original method of detecting the i8o RADIOTELEGRAPHY presence of oscillations in a circuit consists in the observation of spark discharges passing between points on the circuit, between which there is a difference of potential. Any arrangement for making such observations may be described as a spark detector. 2nd. The next in historical order is a meteod depending upon the power of electric oscillations when passed through a loose or imperfect contact between certain substances to make the electric conductivity of that contact better or worse. In the majority of cases the contact is improved and the conductivity increased, and since the surfaces are then more or less perfectly made to cohere together, such devices have been usually called coherers, but the term is not sufficiently general, and it is therefore better to speak of them as imperfect contact oscillation detectors. 3rd. Those depending on the power of electric oscillations to affect the magnetic properties or state of magnetic metals. These are called magnetic detectors. 4th. A large class which depend upon the ability of electric oscillations to heat a fine wire or substance of high resistance. These are called thermal detectors, or if a thermoelectric junction is employed, they are called thermoelectric detectors. 5th. A fifth type of oscillation detector operates in virtue of the power of electric oscillations to affect chemical action. These are called electrolytic detectors. 6th. A sixth type of detector depends for its action upon the unilateral conductivity of rarified gases, which are permeated by negative ions or corpuscles. These are called valve detectors, or ionised gas detectors. . 7th. A detector has been invented based upon the possession by certain crystals of the curious property of unilateral con- ductivity, or power of conducting a current much better one way through them than in the opposite direction. These are called crystal rectifiers. 8th. Detectors based upon electrodynamic actions between fixed and movable conductors, one of which is traversed by oscillations. These are called electrodynamic detectors. We shall briefly consider each of these classes of detector in turn. 2. Spark Detectors. In the original researches by which Hertz experimentally established the production and properties of electromagnetic waves, he made use of a simple, closed, resonant receiving circuit having a small spark gap in it as a means of detecting electric oscillations set up in that circuit when held in various positions, in a space in which electromagnetic waves were falling upon it. This circuit was generally circular, and the OSCILLATION DETECTORS 181 spark gap consisted of two small balls, adjustable by a micro- meter screw (see Fig. la). If such a circuit is held with its plane perpendicular to the direction of the incident electromagnetic wave, and the line joining the spark balls of the resonator parallel to the direc- tion of the electric component of the wave, small sparks are seen, due to the electric force in the wave setting up a potential difference between the spark balls. No such spark is seen if the resonator is held with the line joining its spark balls at right angles to the direction of the FIG. la. electric component of the wave. The above arrangement constitutes a closed receiving circuit. An open one may be made by using two rods of adjustable length, which can be placed in one line, having adjustable spark balls between them (see Fig. 16). If this open resonator is held FIG. 1&. parallel to the electric component of an incident wave, and if the lengths of the rods are adjusted to be about two-fifths of the wave length, then small sparks will be seen at the gap, due to the electromotive force set up in the rods by the incident wave. This spark method of detection is, however, chiefly useful in laboratory experiments, and has a limited range of utility in radiotelegraphy on account of want of delicacy. We need not, therefore, discuss the theory of the Hertzian resonator in this manual at greater length, but refer the reader to larger treatises for a fuller description of its properties and uses. 3. Imperfect Contact Detectors. The first type of oscillation detector which was devised, having sensibility sufficient to be useful in radiotelegraphy, was developed out of the researches of numerous physicists on the peculiar conductive properties of metallic filings and loose contacts. As far back as 1835, Munk had observed that a mixture of tin filings, carbon, and other materials in a loose condition was non-conductive to electricity, but became conductive on passing the discharge from a Ley den jar through it. The same fact was observed again by Calzecchi- Onesti, in 1884. S. A. Varley, in 1852, noticed a remarkable fall in the resistance of masses of metallic filings under the action 182 RADIOTELEGRAPHY of atmospheric electric discharges. In 1866, C. and S. A. Varley applied this discovery in the construction of a lightning protector for telegraphic instruments. In 1878, D. E. Hughes appears to have discovered that a tube full of zinc and silver filings placed in series with a voltaic cell and a telephone, became conductive under the action of an electric spark at a distance. In 1890, E. Branly, of Paris, rediscovered this important fact, and confirming the observations of previous researches, added much new know- ledge. He noticed that an electric spark had the power of suddenly changing the electric conductivity of a loose mass of metallic filings placed a long way from the spark. In the majority of cases the change is from a poor conductivity to a much better one, but in a few cases, such as a loose contact between lead and peroxide of lead, the change is from a fairly good conductivity to a worse one. Branly constructed his metallic filings spark detector by placing in a tube of non-conducting material some metallic filings loosely packed between two metal plugs (see Fig. 2). He connected this appliance in series with a galvanometer and a single cell, and by adjusting the pres- sure of the plugs was able to prevent the current from the voltaic cell affecting the galva- nometer, because the loosely aggregated filings are non-con- ductive for the feeble electro- motive force of a single cell. Under the influence of an electric spark at a distance, the filings, however, change quite suddenly into a condition of better conductivity, and the current from the cell passes through the mass and deflects the galvanometer. Branly found the same effect takes place when a loose or imperfect contact is formed between two pieces of slightly oxidised metal, such as copper or steel wires, and he found that this contact drops in resistance from many thousands of ohms to a few ohms under the influence of an electric spark made some yards away. He also noted that a slight tap or blow destroyed the improved contact. These observations did not attract attention in England until described by Dawson Turner, in 1892, and they were then re- peated by Croft before the Physical Society, in 1893, and carefully examined by Minchin and Lodge in the same year. In 1894, Lodge made use of a tube of glass full of loosely aggregated iron or brass filings contained between two plugs to detect the existence of electric waves created by a spark discharge. He christened this device a coherer -, because he considered that the action of the FIG. 2. OSCILLATION DETECTORS 183 oscillations produced by the incident electric wave was to make the particles cohere together. A similar type of detector was employed by Popoff in Kussia in researches on atmospheric electricity, and described by him at the beginning of 1896. In the same year G. Marconi described in a British Patent Specification a greatly improved form of metallic filings oscillation detector constructed in the following manner : In a small glass tube about 3 or 4 cms. long and 5 mms. internal diameter, he placed two silver plugs fitting the tube tightly. To these plugs were attached platinum wires sealed through the closed ends of a tube. The inner ends of the plugs were polished and slightly amalgamated with mercury and brought within a couple of millimetres of each other. The interspace was filled with a very small quantity of nickel and silver filings, 95 per cent, nickel and 5 per cent, silver, carefully sifted. The glass tube was then, exhausted and sealed. Subsequently the ends of the silver plugs were bevelled off so as to make the interspace wedge- shaped (see Fig. 3). Marconi thus constructed an extremely sensitive form of imperfect contact oscillation detector, which under the influence of very feeble oscillations set up by electric waves passed from a condition of high resistance to a condition of low resistance. Branly had previously shown that a contact detector of this kind could be brought back to its original high resistance and sensitive condition by giving it a slight blow or tap, so as to wrench asunder the surfaces connected together by the action of the oscillations which had passed through it. Hence a metallic filings tube of this kind when employed for the detection of electric waves must be associated with some arrangement for continually tapping or decohering the filings. Lodge did this originally by means of a clockwork arrangement for continually shaking the metallic filings tube, or used an electric bell hammer to administer small blows 1 84 RADIOTELEGRAPHY continuously to the tube itself. The same was done by Popoff, in 1906, but he arranged the metallic filings tube in series with an electric relay, which, in turn, set in operation an electric bell, the hammer of which administered one or two blows to the coherer tube, shaking up the filings and bringing them back again to a non-conductive condition. Marconi improved upon this by making all the adjustments capable of very nice regulation, and arranging the electromagnetic tapper so as to administer a delicate blow to the metallic filings tube from the under side of exactly the right strength to destroy the conductivity of the metallic filings immediately a current passes through them in virtue of the passage through the tube of electric oscillations (see Fig. 4). The final outcome of all this work was the invention of a complete apparatus, com- prising a tube with metallic filings contained between two metallic plugs. This was placed in series with one or two voltaic cells and a tele- graphic relay, so that when a small current passed through the metallic filings tube the moment it became conductive, this current was made to close another circuit and set in operation other telegraphic instruments. The relay was caused, in addition, to close a circuit through another electromagnet operating the tapper, so that the moment the metallic filings tube had passed into the conductive condition a blow was administered to it which brought it back again into the non-conductive state, whilst, at the same time, the relay was made to actuate some form of recording instrument, as more particularly described in Chapter VII. The complication introduced by the necessity for tapping these metallic filings conductors back into a non-conductive condition led to inventions to construct self-restoring imperfect contact detectors. Branly had discovered that a loose contact formed with powdered peroxide of lead increased in resistance by the action of an electric spark. S. G. Brown has more recently constructed and described an auto-restoring oscillation detector, in which a small plug of dry compressed peroxide of lead is gently compressed between a platinum plate and a lead plate, but it is FIG. 4. OSCILLATION DETECTORS 185 FIG. 5. claimed for this device that it acts as an electrolytic valve, and will be, therefore, referred to again in a following section. Lodge, Muirhead, and Eobinson devised a form of contact detector consisting of a steel disc rotated by clockwork, the edge of which just touched the surface of some mercury covered with oil (see Fig. 5). The steel disc has a sharp knife-like edge, and under ordinary circumstances the film of oil carried round by the steel pre- vents good electric contact between the steel and the mercury ; but if electric oscillations are passed through the contact they break down the insulation of the film of oil and make a good electric contact between the steel and the mercury which, however, disappears as soon as the oscillations cease. A form of self-restoring contact detector was devised by Italian naval officers, attributed by Captain Bonomo to Ca,stelli, a signalman in the Italian Navy, but also claimed by the Marquis Solari. In this ap- pliance a small globule of mercury is contained between a steel and carbon plug fitting tightly in a glass tube (see Fig. 6). A tele- phone and a single voltaic cell are in- cluded in series. By adjusting the pressure, the current from the cell can be prevented from flowing through the contact points be- tween the mercury and the steel and carbon plugs, but if electric oscillations are passed through these contacts the conductivity is improved and a current passes, but the conductivity automatically disappears on the cessation of the oscillations. FIG. 6. 1 86 RA DIO TELEGRA PH Y One of the best and simplest of these self-restoring imperfect contact oscillation detectors is the Tantalum-Mercury contact invented by L. H. Walter. A fragment of tantalum wire from a tantalum lamp is attached to a platinum wire, and the tip of the tantalum immersed in mercury. The wire and mercury are connected respectively to a shunted voltaic cell in series with the telephone. This fraction of a volt of electromotive force is unable to send a current across the surface of contact of the tantalum and mercury, possibly because the tantalum is not wetted by mercury. When, however, oscillations are passed across the junction, the contact is improved so far that a current can pass through the telephone, creating a sound ; but this contact between the tan- talum and mercury ceases the moment the oscillations stop hence the contact is self-restoring. The mercury with the tantalum point dipping into it can be sealed up air-tight in a glass vessel, so that the mercury is preserved from oxidation. It will be seen, therefore, that all these forms of detector consist of some form of contact, either single or multiple, between various substances, the conductivity of this contact being either greatly improved or else diminished by passing electric oscillations through the contact point. In the construction of imperfect contact detectors for radio- telegraphy the most important requirement is certainty of action. To secure this only a small testing current must be passed through the contact to work the relay or other indicating instrument which reveals the resistance change. The oxidisable metals are not so suitable for making these oscillation detectors as mixtures of rather oxidisable metals with others which are non-oxidisable. Hence, nickel and silver or aluminium and steel contacts are better than copper with copper. The magnetic metals nickel and iron, for some reason not fully understood, have a marked superiority over other metals. Hence, good contact detectors have been made by a light contact between the steel balls contained in a tube or, as in the tripod coherer of Branly, constructed by making a small copper three-legged stool, having the feet slightly oxidised, placed on a polished steel plate. Many devices have been employed for restoring the contact to its sensitive condition, not only by tapping, but by rotating the tube or surfaces, or acting upon the metallic filings, if of iron or nickel, with a magnet. It is advisable, however, to exclude the air, so as to prevent the surfaces from becoming too much oxidised. The metallic filings detector, in whatever form it may be made, must be regarded as a potential indicator, depending, as it does, OSCILLATION DETECTORS 187 upon the production of a certain small alternating difference of potential between two surfaces in loose or imperfect contact, the result of which is to improve or diminish the conductivity between these surfaces. Much discussion has taken place upon the exact nature of the processes at work when electric oscillations pass through an imperfect contact. The original idea was that the effects were due to thermal action, heat being developed at the imperfect contact which welded together the junction. This hypothesis, however, fails to explain the contact action when such unweldable substances as carbon or other non-metallic conductors are employed. Also, it does not explain the decreased conductivity in the case of a lead and peroxide of lead junction. When the oscillations begin to take place across the junction there is a certain difference of potential between the points or surfaces in imperfect contact. We may think of the surfaces as initially separated by an extremely thin film of air and forming therefore a condenser of small capacity. Another view, therefore, taken of the effect is that the electrostatic attraction between these surfaces squeezes out the air and brings the surfaces into molecular contact, thus effecting an improved conductivity. At present, however, our knowledge of the true nature of electric conduction and of the reason some substances are better conductors than others is too imperfect to enable us to account for the fact that oscillations passing across a loose contact between two surfaces do not always cause an increase of conductivity. Neither the welding theory nor the electrostatic attraction theory explain why the magnetic metals, particularly nickel, are so much better than most others in making imperfect contact oscillation detectors. In spite of much research, therefore, the scientific problems in connection with this coherer action are by no means solved. 4. Magnetic Oscillation Detectors. It was well known in the early part of last century that the discharge of a Leyden jar can magnetise steel sewing needles. Between 1842 and 1850, Joseph Henry, in the United States, examined this effect with great care, and only obtained a clue to numerous puzzling phenomena when he realised that the discharge of a condenser through a circuit of low resistance is oscillatory in character, and sometimes has the effect of magnetising and at other times of demagnetising the steel. In 1870, Lord Rayleigh in discussing some electromagnetic phenomena pointed out that the effect of an oscillatory discharge of electricity upon iron or steel depends upon the direction of the maximum value of the current during the oscillations, and also that there may be superimposed magnetic effects in the needle. The i88 RADIOTELEGRAPHY subject was taken up again by E. Kutherford, in 1895, and in a Paper published in 1896 he described a number of important experiments in which the demagnetising power of electric oscilla- tions was employed as an indicator for electric waves, making therefore the first magnetic detector, as follows : About twenty pieces of fine steel wire, each about 0'07 mm. in diameter and about 1 cm. in length, were insulated by shellac varnish and bound together into a little bundle. A fine copper wire insulated with silk was wound over the bundle in two layers of 80 turns. This small electromagnet was fixed at the end of a glass tube, and was placed at the back of a small suspended magnetic needle. If a current was passed through the coil it magnetised the steel wires to saturation, and they then caused a certain steady deflection of the suspended magnet. If, then, electric oscillations were sent through the coil wound round the steel wire, these demagnetised the steel, and caused the magnetometer deflection to diminish. Kutherford connected the two ends of the magnetising coil with two long horizontal rods acting as antennae, and half a mile away he set up a Hertzian oscillator with its rods also in a horizontal position. After magnetising the steel wires the Hertzian oscillator at a distance was set in action. The electromagnetic wave passing out through space fell upon the receiving rods, and set up in them electric oscillations which, passing through the coil, demagnetised the steel wires, and therefore caused a diminution in the deflection of the associated magnetometer needle. In 1897, E. Wilson took up the subject, and endeavoured to make this magnetic detector self-acting, so that the deflection of the magnetometer needle should close the circuit, and again magnetise the steel bundle. In 1902, G. Marconi described two other forms of magnetic detector. In one a thin bundle of iron wires was surrounded by a magnetising coil and its central part also embraced by a second coil in series with a telephone. Over this small bundle of iron wires a horseshoe magnet was slowly rotated so as to carry the iron through a cycle of magnetic changes, magnetising it first one way and then the other. If electric oscillations are sent through the coil wound round the iron they exercise an oscillatory magnetic effect on the iron, and these cause sudden magnetic changes in it which set up electro- motive forces in the secondary coil embracing the iron, which make themselves evident by short sounds or ticks heard in the telephone. Whether the change in the iron be from a less to a greater or greater to a less state of magnetisation, an inductive effect on the coil in series with the telephone is equally exerted, and the great OSCILLATION DETECTORS 189 sensitiveness of the telephone to sudden but small changes in the current through it makes it an extremely delicate means of detecting these small changes in the magnetic state of the iron. For telegraphic purposes Marconi invented a far more perfect instrument, automatic in action, and capable of giving telegraphic signals, made as follows : Two wooden discs e, e, (see Fig. 7), grooved on the edges, are driven round slowly by clockwork. An endless band, a, a, made of a bundle of fine silk-covered iron wires, is arranged like a belt over these pulleys, and moves forward at the rate of 7 or 8 cms. per second. At one place the iron band passes through a glass tube, g, b, on which is wound a coil of insulated wire through which oscillations can be passed, and this coil is embraced in the centre by another coil, c, connected with the telephone, T. A pair of horseshoe magnets are placed with their similar poles together opposite to the last- mentioned coil, as shown in the diagram. If electric oscillations are passed through the coil wound round the band, they change the magnetic state of the iron, and generate an induced current in the secondary coil, and hence a current and a sound in the tele- phone. If oscillations con- tinue to pass through the demagnetising coil. at short intervals, the telephone will FIG. 7. emit a sound which is practically continuous, and if the oscillations are in longer or shorter groups, the corresponding sounds are produced in the telephone, and may be interpreted in accordance with an agreed code of signals. The extreme sensitiveness of the telephone to induced currents bestows upon this apparatus very great power in detecting feeble oscillations. The operation of this instrument was considered by Marconi to be due to the power of electric oscillations passing through the coil surrounding magnetised iron to annul the hysteresis of the iron. A reduction in magnetic hysteresis does not, however, invariably accompany the action of electric oscillations on iron or steel. Walter and Ewing discovered that in hard steel an increase in hysteresis may result when oscillations are sent through iron whilst being submitted to cyclical magnetisation. They devised a form of magnetic oscillation 190 RADIOTELEGRAPHY detector based on this fact as follows : An electromagnet rotates round a vertical axis so as to produce a revolving magnetic field (see Fig. 8). In the centre of that field is suspended a closed coil of hard drawn insulated steel wire suspended by an elastic wire. When the magnet rotates it tends to carry the suspended coil round with it in virtue of a dragging action due to magnetic hysteresis, and the torque so produced is resisted by the control of the elastic suspension. When electric oscillations are passed through the closed coil of steel it is found that the hysteresis of the metal is increased, and the coil twists more in the direction of the rotation of the magnet. This instru- ment, therefore, can be employed as a means of measuring the intensity of electric oscillations as well as merely to indicate their presence. Another magnetic oscillation de- tector suitable for quantitative work has been devised by the author as follows : In a pasteboard tube are included 7 or 8 small bundles of fine iron wire, each wire well painted with shellac varnish, and each little bundle wound over uniformly with a magne- tising coil of fine silk-covered wire, and over this coil and separated from it by guttapercha tissue another de- magnetising coil of somewhat thicker insulated wire. The magnetising coils are all connected in series in such a manner that when a current passes through them, it magnetises the whole of the wires so that contiguous ends have the same magnetic polarity. The outer or demagnetising coils are joined in parallel. Over this bobbin is wound a long secondary circuit consisting of 6000 turns of very fine silk- covered copper wire. Associated with this induc- tion coil is a rotating commutator (see Fig. 9) carrying four insulating discs secured to a steel shaft. These discs have brass sectors let into their circumference, and against them four springs of brass wire press. As the commutator rotates, these springs and discs are made to close or open various electric circuits. The function of the first disc is to make a break in the circuit of the magnetising coils placed round the iron bundles, to magnetise [Reproduced from "The Electrician permission of the Proprietors. FlG. 8. OSCILLATION DETECTORS 191 them during one portion of its rotation, and leave them magnetised during the other portion. The function of the discs 2 and 3 is to short circuit the terminals of the secondary coil during the time the magnetising current is being applied by disc 1. A sensitive galvanometer is connected to the ends of this secondary coil, one being permanently connected and the other through the intermittent contact made by disc 4. The operations which go on during one complete revolution of the discs are as follows : First the current of a battery of secondary cells is employed to magnetise the iron bundles, and during this time the terminals of the fine wire secondary coil are short circuited, and the galvanometer disconnected. Shortly ^ after the magnetising current is interrupted, the secondary bobbin is unshort circuited, and immediately afterwards the galvanometer cir- cuit is completed. Hence, during a large part of one revolution the iron wire bundles are left magnetised and surrounded by a secon- dary coil connected to a galvanometer. If during this period electric oscillations pass through the de- magnetising coils, an electromotive force is induced in the secon- dary bobbin by the demagnetisation of the iron, and this causes a deflection of the galvanometer coil. Since the interru-pter discs are rotated very rapidly, if the oscillations continue, this intermittent electromotive force produces a practically steady current through the galvano- meter which is proportional to the demagnetising force being applied to the iron. Hence the arrangement becomes not merely a means of detecting oscillations, but of measuring their intensity. This instrument has been employed by Buscemi for quantitative measurement of the opacity of various dielectrics to electric waves. The endeavour to account for these interesting magnetic effects FIG. 9. 192 RADIOTELEGRAPHY of electric oscillation has given rise to much research and dis- cussion. The chief contributions to it have come from Maurain, L. H. Walter, Ascoli, Arno, Piola, Foley, C. Tissot, P. Duhem, W. H. Eccles, and J. Eussell. Eussell has carefully distinguished between the two conditions under which we can work. (i.) Iron or steel may be placed in a constant magnetic field, and then subjected to the action of electric oscillations. (ii.) Iron or steel may be subjected to continuing electric oscillations, and then the magnetic field around it changed. In the case of Eutherford's experiment, hard iron or steel having considerable retentivity is subjected to a magnetic force, which is then removed, leaving remanent magnetisation in the iron. The action of oscillations taking place round the iron is then always to remove or diminish this magnetisation, and this can be detected by a change in the position of a suspended magnetic needle in the neighbourhood. In Marconi's first form of magnetic detector, a horseshoe magnet is rotated slowly (about one turn in two seconds) over a thin bundle of hard iron wires, which are surrounded by two separate coils of wire. The iron is thus carried slowly through a cycle of magnetising force of equal positive and negative values. The magnetisation induced in the iron lags behind the magnetising force in virtue of so-called hysteresis, and, therefore, if ordinates representing the magnetisation are plotted out in terms of the magnetising force as abscissae, we obtain a magnetisation curve of the well-known looped form. The area of this loop is proportional to the work expended in carrying the iron through one complete magnetic cycle. Eussell states, as the result of his experiments, that if oscillations act continuously on the iron whilst it is being carried round the magnetic cycle, the area of this loop is greatly increased, thus showing an increase in hysteresis loss. If the oscillations come intermittently, as they would do in radiotelegraphic signalling, then the effect depends upon the particular point in the cycle at which the oscillations arrive. The result, in any case, is to produce a sudden change in the magnetisation of the iron. Hence, if the oscillations are sent through one coil wound round the iron, the sudden change in magnetisation produced by them creates induced or secondary electric currents in another coil wound over the oscillation coil, and therefore causes a sound in a telephone in series with the latter coil. In Marconi's iron band form of magnetic detector the action is somewhat different. The band is passing through a magnetic OSCILLATION DETECTORS 193 field, so as to be always subject to a longitudinal magnetising force, which is first in one direction and then is quickly reversed, because the two horseshoe magnets are placed with their poles against the wire and have similar poles in contact. Under these conditions, Eussell found that the effect of a longitudinal oscil- latory magnetic force is to increase the magnetisation due to the steady force by an amount which is greater for an increasing than for a decreasing field. If the double north poles of the magnets are in the centre, and the iron moves from left to right, then the moving iron band distorts the field, and the effect of the oscillations passing round the iron is to increase the magnetisation of the iron more on the left hand than on the right of the north poles. Both these effects alter the number of lines of magnetic flux through the secondary coil in series with the telephone, and therefore cause an induced current to flow through it, and the telephone in series emits a sound. Hence, Eussell considers that Marconi's second form of magnetic detector acts in virtue of the increase of magnetisation in iron which occurs when an oscillatory field is superimposed upon a slowly changing or stationary field near a cyclic extreme, whereas Kutherford's form of detector operates in virtue of a decrease of magnetisation produced when the magnetising force has been applied and has been removed. The function of the moving band is twofold : it supplies the hard iron or steel in a condition of low permeability to be raised by the oscillations to a condition of higher permeability, and it distorts the field in the direction of motion. This view is rather different from that taken by Marconi himself and others, who have expressed the opinion that the action of the oscillations is to annul the hysteresis of the iron. According to W. H. Eccles, the effect produced when a bundle of iron wires is taken slowly round a magnetisation cycle and an oscillatory magnetising force applied at any point is to bring the iron back to the condition of magnetisation it would have under the final steady impressed magnetic force acting on it if the hysteresis was suddenly annulled. The action of the oscillations is therefore to cause a return to the normal curve of magnetisation. In the Walter-Ewing form of detector we have different magnetic conditions. The field is then revolving somewhat rapidly, so that a drag is produced oil the suspended iron, due to the so-called rotational hysteresis. Oscillations increase this hysteresis, and therefore the deflection of the suspended iron at least in fairly strong fields. According to L. H. Walter, all magnetic detectors may be divided into two classes. First, those in which the oscillations 194 RADIOTELEGRAPHY act on the iron after the magnetising force has been applied and withdrawn. Examples of this type are the original Kutherford detector and the Fleming quantitative detector above described. In this case the available energy is limited to the remanent magnetism in the core, and the action of the oscillations is to reduce or destroy this remanent magnetism. The second class, represented by Marconi's moving band detector, derive their energy from an external magnetic field and from the motive power driving the band, and the action of the oscillations is merely to release some of this energy. If the iron is moving through a field of increasing magnetic force, it is on the lower side of the hysteresis loop, and the action of the superimposed oscillatory field is to increase the magnetisation when not actually at the peak of the curve, the increasing effect being, as E. Wilson first showed, greatest at or near the point of inflection of the lower branch of the hysteresis curve. This increase in magnetisation of the portion of the iron band partly enclosed by the coil in series with the telephone creates the induced current in the latter. We may, therefore, in a sense, speak of this increase in local magneti- sation of the iron as due to an annulment of hysteresis. The action of the moving band detector is, however, essentially dependent on the supply of energy from an external source to magnetise the iron and move it against the magnetic force. The action of the oscillations is only a trigger action, which creates a sudden increase in the magnetic flux in a part of the iron embraced by the secondary or telephone coil. This causes in turn induced currents to flow through it, first one way and then the other. Accordingly, with the band detector, the only possible signal receiving instrument is a telephone, unless we provide some means of sifting out the direct from the inverse induced current in the telephone coil. This has been achieved, however, by means of one of the author's oscillation valves or glow-lamp detectors, described in a subsequent section, and by its use it is possible to obtain from a Marconi moving band magnetic detector, asso- ciated with a Fleming oscillation valve, intermittent but uni- directional currents, which can operate a relay, and therefore work any ordinary telegraphic printing instrument. Another method has been devised by L. H. Walter, by which a detector of the Walter-Ewing type, depending upon rotational hysteresis, can be made to furnish continuous currents. In this case oscillations are made to act on a magnetic mass undergoing reversals of magnetism in a rotating field in such a manner that the changes of magnetism produced by the oscillations create alternating induced currents in embracing coils of wire, which are OSCILLATION DETECTORS '95 rectified by a commutator in the usual dynamo machine manner. The inventor has described his apparatus as follows : Two ebonite bobbins, B, B (see Fig. 10), mounted on the same spindle, are rotated in the field of two horseshoe permanent magnets, NS, NS, these bobbins being wound, in a similar manner to those illustrated in connection with the pivoted bobbin detector previously referred to, with some feet of steel wire of suitable resistance. A wind- ing of two coils, W, W, at right angles to one another, of a hundred turns, is placed on each bobbin at right angles to the plane of the steel wire winding, as in a drum armature, corresponding coils, i.e. W and W, W and W, being connected in such a way that the E.M.F.'s generated are equal and opposite. The ends of the windings are connected to the segments of a four-part commutator, C. (For the sake of clearness, only one pair of corre- sponding windings, of one turn each, is shown connected in Fig. 10.) The steel wire windings of the two bobbins are exactly alike, the ends of one winding being insulated, while those of the other are connected to a pair of slip-rings, r, r, and brushes, by means of which the oscilla- tions can be passed through the winding. On testing this apparatus, with no oscillations acting, there was no potential difference at the brushes. On waves arriving, a steady deflection of the gal- vanometer was obtained in a direction corresponding to an increase of E.M.F. generated by the armature acted upon by the oscillations. By suitably proportion- ing the turns in the winding the sensibility was considerably increased. The usual speed employed is about five to eight revolutions per second. Higher speeds have been tried, and give a larger effect, but the zero is not so steady. Telephonic signals can, of course, be received simultaneously by connecting to the winding at some point before the E.M.F. is commutated. When a relay alone has to be actuated, however, it may be advantageous to so arrange matters that the generated E.M.F/s do not exactly [Reproduced from " The Elec- trician " by permission of the Proprietors. FlG. 10. 196 RADIOTELEGRAPHY balance, and a small initial current, insufficient to actuate the relay, passes all the time through it. The change can be rapidly effected by a very slight shift of the brushes. 5. Thermal and Thermoelectric Detectors. An electric oscil- lation, being a form of electric current, produces heat in a circuit through which it flows. The heat produced per second is pro- portional to the resistance of the conductor and to the square of the effective or K.M.S. value of the current. This mean square value is called the integral 'value of the current. Accordingly, if the oscillations consist of separate trains, these not being very close to one another, the integral value of the oscillations may be small, although the maximum value in each train may be very considerable. If we call J the integral value of the current J being the R.M.S. value and if we call I the maximum or initial value of the current in each train, S the decrement of a semi-period, n the frequency of the oscillations, N the number of trains per second, then I and J are related as follows : J = provided that the oscillations are not very strongly damped, as is the case with a closed oscillation circuit. If the oscillations are persistent or undamped, then if I is the maximum value, we have . provided the oscillations are of simple sine form. To detect feeble oscillations by thermal effects, we must pass them through a wire of high resistance but small heat capacity, so that there may be a relatively large rise of temperature, and we must provide some means of detecting this rise of temperature in the wire. Since the introduction of a resistance in an oscilla- tion circuit damps the oscillations, it is necessary to make the resistance introduced as small as possible and to localise the heat produced in it. Hence, what is used is a short length of not more than a few millimetres, or even less, of extremely fine wire made of a material of high resistivity, such as constantan or platinoid, and we must provide some sensitive means for deter- mining a small rise of temperature in the wire. Since the rise of temperature measured depends upon the relative rate at which the wire gains and loses heat, we can increase the rise of tempera- ture for a given heat production in the wire by reducing the convection of heat by the air. In other words, if we put the wire OSCILLATION DETECTORS 197 in a high vacuum we can obtain the greatest rise of temperature for a given integral value of the oscillations. Such an arrange- ment of a short length of high resistance wire enclosed in a vessel which may or may not be exhausted of its air, is called a bolometer, or, by Fessenden, a "barretter. The small rise of temperature may be detected in several ways. (1) We may enclose the wire in the glass bulb of an air thermometer, and make a form of electric thermometer, which was devised in the last century by Snow Harris, but by conti- nental writers generally attributed to Eeiss. To render it sensitive to changes of external temperature, this thermometer should be of a differential form, consisting of two glass bulbs connected to a glass U-tube, containing some non-volatile liquid, such as sulphuric acid. One of the bulbs has a fine wire sealed through it. The level of the liquid is not then affected by external changes of temperature, but if oscillations are passed through the fine wire, heat is produced, the air in that bulb is expanded, and the level of the liquid falls in one tube and rises in the other until a stationary condition is reached. Such an instrument may be calibrated for integral values of the oscillations by passing various continuous currents through the wire and noting the position of the liquid in the U-tube. This instrument is generally called a hot wire thermometer. (2) We may detect a rise of temperature by the change in resistance produced in the bolometer wire itself. This was first done by Eubens and Eitter in 1890, who constructed a form of Wheatstone's bridge, of which two of the arms consisted of rectangles of fine wire, the other two arms of variable resistance as usual, the galvanometer G and battery B being connected, as shown in Fig. 11. If, then, electric oscillations are passed through one of the rectangles, E, from corner to corner, this arm of the bridge is heated, its resistance is increased, and the balance of the bridge upset, the galvanometer G showing a deflection, which indicates a change in resistance, and therefore serves to measure the rise of temperature produced in the wire, and therefore the integral value of the oscillations. C. Tissot has employed this method in a more refined form as a detector of oscillations produced by the electric waves utilized in radiotelegraphy. He makes use of a fine platinum wire not more than O'Ol mm. in diameter as the bolometer wire to be heated by the oscillations, and detects and measures this heat by the variation in resistance of the wire. With such an arrangement he states that he has detected electric waves at a distance of 40 kilometres from the radiator. IQ8 RA DIO TELEGRA PH Y E. A. Fessenden has devoted much attention to the improvement of thermal oscillation detectors, made as follows : An extremely fine platinum wire, about 0*08 mm. in diameter, is embedded in the axis of a silver wire, about 2 mm. in diameter, like the wick of a candle. The compound wire is then drawn down until the diameter of the platinum wire is reduced to about 0'0015 mm., in the manner first suggested by Wollaston. A short piece of this Wollaston wire is bent into a loop and the ends attached to stouter wires and enclosed in a glass bulb. If the tip of the loop is dipped into nitric acid, the silver coating is dissolved off, leaving a short length of exquisitely fine platinum wire sealed vwwwvww ^vwwvwww FIG. 11. into the bulb like the filament of an incandescent lamp (see Fig. 12). The bulb is then closed and the air exhausted. Furthermore, the glass bulb may be enclosed in a silver bulb to shield it from external radiation. To detect the small rise in temperature produced in this fine wire by very feeble oscilla- tions passed through it, the ends of the loop are connected to a telephone, a single shunted cell being interposed. The cell then transmits a feeble continuous current through the telephone and the bolometer wire. If oscillations are then sent through the latter, they still further heat the loop and suddenly decrease the current through the telephone. The ear then detects this decrease of telephonic current by a sound made in the telephone. OSCILLATION DETECTORS 199 Fessenden found that in place of the platinum wire a fine tube filled with a high resistance liquid could be employed, or even the two ends of a very fine platinum wire dipped into a liquid, thus forming what he calls a liquid barretter. A number of such bolometer wires can be arranged in parallel to be heated by the same oscillations. (3) We may use a thermoelectric couple to detect a small rise of temperature in the wire. Such an arrangement was first employed by Klemencic. The oscillations are sent through a fine constantan wire, and against this rests the junction of a thermoelectric couple made of iron and constantan, or some other pair of suitable metals. The ends of the couple are connected FIG. 12. FIG. 13. to a suitable low resistance galvanometer, and on passing oscil- lations through the high resistance wire the galvanometer gives an indication. An improved arrangement of this kind was devised by the author in 1906, taking advantage of the position of tellurium and bismuth in the thermoelectric series. A sort of double test-tube of glass was constructed (see Fig. 13), the interspace between the two tubes being subsequently exhausted. Through the bottom of the inner test tube were sealed four wires, two of these, a, &, were connected to a fine constantan wire, and the other two, c, d, were connected to a tellurium-bismuth therm o- junction T formed of very fine wires of bismuth and tellurium, the 200 RADIO TELEGRA PHY junction being soldered by a special solder to the centre of the constantan wire. A high vacuum was then made in the interior space. When oscillations are passed through the constantan wire, a suitable low resistance galvanometer being connected to the leads from the thermoj unction, the galvanometer deflects, the deflection being proportional to the square of the integral value of the oscillations. This detector has proved to be of great use in quantitative researches. Another form of thermoelectric detector was devised by Mr. Duddell. He employs a form of Boys' micro-radiometer in which a thermocouple of bismuth and antimony wire forming a loop is suspended by a quartz fibre, in a strong magnetic field. A mirror attached to the thermo- couple enables small deflec- tions to be measured. Under- neath this thermocouple he placed a very fine narrow strip of metal, say gold leaf, through which the electric oscillations are passed. These heat the strips feebly. One junction of the small suspended thermo- couple rests just above the strip, but not quite touching it, and is therefore heated by radiation and convection. The couple is therefore traversed by a current and deflected in the magnetic field. This thermal detector is also quantitative. The complete instrument is shown in Fig. 14. (4) We may detect the heating of the bolometer wire by its own expansion, as first done in 1889 by W. G-. Gregory. This arrangement is more suitable for detecting large and powerful oscillations than feeble ones. The author has devised and used a hot wire ammeter made as shown in Fig. 1, Chapter VIII., where the oscillations are passed through a fine wire, the ends of which are fixed. The expansion of the wire therefore produces a sag of the middle part, which is magnified by an index needle or by the movement of a mirror and a ray of light. The advantage of enclosing the bolometer in a vacuum is very considerable, because in the case of fine wires heated by a current the removal [Reproduced by permission of the Cambridge Scientific Instrument Company. FlG. 14. OSCILLATION DETECTORS 201 of the heat is chiefly effected by air convection, and if therefore we stop this convection by removing the air, the temperature of the wire rises higher for a given integral effect of the oscillations. These thermal detectors are very useful for quantitative work in the laboratory, because they can be so easily calibrated by passing through them a continuous current, and they are therefore not merely oscillation detectors but are measurers of the integral value of the oscillations or of their mean-square value. Moreover, the thermal detectors are especially valuable in connection with the measurement of undamped oscillations, as there we have to do with integral values which are large compared with the integral values of damped oscillations. Another form of thermal detector can be made by dispensing with the bolometer wire and heating the thermoelectric junction itself directly. Thus, for instance, a form of detector has been made by L. W. Austin which consists of a fragment of tellurium pressed against the edge of an aluminium disc, the pressure being adjustable by a screw. A telephone is connected across this junction. If oscillations are sent through the junction they produce heat at the point of imperfect contact and therefore a thermoelectric current, which passes through the telephone and will therefore give an indication, if the oscillations are started or stopped, by a sound in the telephone. Such an arrangement, therefore, is not suitable for metrical work, but is suitable as a radiotelegraphic detector where trains of oscillations are intermittently created and stopped. 6. Electrolytic Oscillation Detectors. Numerous isolated observations were made between 1898 and 1902 which indicated that electric oscillations possess the power to affect the polarisation of small metallic surfaces immersed in an electrolyte. In 1898, A. Neugschwender showed that if a deposit of silver was made on a sheet of glass and divided into two parts by a sharp cut with a razor, and a film of moisture deposited on the glass, then electric oscillations taking place across the gap had the power to alter the resistance of the film of moisture. Similar observations were subsequently made by E. Aschkinass and L. de Forest, arid the latter made an oscillation detector con- sisting of a tube closed at the ends with metallic plugs, the inter- space being filled with a mass of peroxide of lead and glycerine having in it metallic filings. The modern form of electrolytic detector originated in 1903 with E. A. Fessenden, and was shortly afterwards independently invented by Schlomilch. It consists essentially of a vessel having as one electrode a very fine short wire of platinum, offering there- fore an extremely small surface. This is generally made the anode 202 RADIOTELEGRAPHY or positive pole. The other electrode is a platinum or lead or silver plate of much larger surface, and the two are immersed in an electrolyte, which may be nitric acid, dilute sulphuric acid, or any other aqueous electrolyte yielding oxygen or hydrogen on electro- lysis (see Fig. 15). The electrode of small surface is generally prepared from a Wollaston wire by drawing down a platinum wire coated with silver, or else a platinum wire coated with iron, until the platinum wire itself is less than O'OOl mm. in diameter. A short length of the compound wire is then fixed to the end of a screw, so that it can be lowered by a very small amount into the electrolyte. If, for instance, the electrolyte is nitric acid, then a silver-coated wire is employed, and on lowering the tip of this into Dilute - --Sulphuric Acid -- fl u FIG. 15. FIG. 16. the acid the silver is dissolved away, leaving a platinum elec- trode of microscopic dimensions immersed in the liquid. If dilute sulphuric acid is employed, then an iron-coated platinum wire is used, If a small electromotive force is applied to such a cell by means of a shunted voltaic cell, B, and a telephone, T, or sensitive galvanometer included in the circuit, a small current will flow through the electrolytic cell, E, and will polarise the electrodes, which will thereupon reduce the current practically to zero (see Fig. 16). Under these conditions, if oscillations are sent through the electrolytic cell they destroy the polarisation of the small electrode and the current suddenly increases, but it returns to its former small value as soon as the oscillations cease. According to Schlomilch and Lee de Forest, this electrolytic OSCILLATION DETECTORS 203 detector is sensitive to oscillations only when the small electrode is connected to the positive pole of the voltaic cell ; that is, when it is polarised with oxygen gas, but Fessenden, Kothmund, and Lessing state that the cell is equally sensitive when the point is negative. According to L. W. Austin, for feeble oscillations the cell appears to be about equally sensitive both ways, but for stronger depolarising currents, acts best with the small electrode positive. There is likewise a difference of opinion as to the reason for the operation of this detector. Fessenden advocates a thermal theory, according to which the chief effect is due to the change of resistance produced at the surface of the small electrode, but J. E. Ives has found that if platinum black is deposited on the small platinum electrode this deposit, as is well known, reduces the polarisation effect and stops the detector action. If the cell is used with a telephone in series with a shunted voltaic cell, and if one of the electrodes in the electrolytic cell is connected with an antenna, and the other with the earth or a balancing capacity, then the impact of electric waves upon the antenna will cause oscilla- tions to pass through the electrolyte and the sudden increase in the current through the telephone is heard as a short sound or tick. If the trains of oscillations are intermittent but rapidly succeed each other, then these short sounds in the telephone run together to a continuous noise, and in this manner by acting upon the cell by a series of trains of oscillations more or less prolonged, signals according to the Morse alphabet can be conveyed. The valuable feature of this electrolytic detector is that it is not merely qualitative but quantitative. The degree to which the polarisation is destroyed is in some sense proportional to the amplitude of the oscillations. Hence, when employing the instrument with a galvanometer, the deflection of the galvanometer will vary accord- ing to the amplitude of the oscillations passing through the cell. It is also sensitive to a wide range of frequency and can be employed with alternating currents of low frequency, and under these circumstances can detect a voltage of a few ten-thousandths of a volt applied to the terminals of the electrolytic cell. The resistance of the electrolytic cell with slowly alternating currents may vary from a few hundred to many thousand ohms. Hence from one point of view we may regard the electrolytic oscillation detector as a detector of the same type as the imperfect contact detector, in that the action of oscillations is to produce an effect on it equiva- lent to a sudden decrease in resistance. If acted upon by undamped oscillations, the amplitude of which vary continuously, then the equivalent resistance of the cell varies continuously, and in some degree proportionately to the intensity of the electric oscillations. 204 RADIOTELEGRAPHY This important fact has been applied, as shown in Chapter IX., in connection with radiotelephony, and has been made the means by which undamped electromagnetic waves are made to convey articulate speech between two points without the use of con- tinuous connecting wires. A form of detector which is by some classified as an imperfect contact and by others as an electrolytic detector is that invented by S. G. Brown. It consists of a pellet of peroxide of lead held between a plate of lead and one of platinum. If an external E.M.F. from a single secondary cell is impressed upon it so that the current flows through the peroxide from platinum to lead, this current will experience a counter electromotive force due to the electro-chemical action of the lead-peroxide of lead-platinum [Reproduced by permission of the Cambridge Scientific Instrument Company. FIG. 17. couple. According to Mr. Brown, when oscillations pass through this couple they increase its counter-electromotive force by stimu- lating chemical action and so reduce the current sent through it by the external cell. The couple acts, therefore, as a conductor of which the resistance is increased by electric oscillations. The pellet of peroxide is mounted up in a holder so as to apply to it an adjustable pressure (see Fig. 17), and is placed in series with a galvanometer and single cell. When oscillations are created through the peroxide the deflection of the galvanometer decreases but increases again when they cease. 7. Valve or Rectifier Oscillation Detectors. Since electric oscillations are alternating currents of high frequency, the means of detecting them simply as electric currents which OSCILLATION DETECTORS 205 have been already described are based upon actions which are independent of the direction of the current. Thus, for instance, the heating effect of an oscillation being determined by the square of the strength of the current at any instant, is indepen- dent of its sign or direction, and the same is true of the coherer or anticoherer action, as it is sometimes called, viz. the increase or decrease of the electric conductivity of an imperfect contact. Speaking generally, the methods available for measuring an alternating current are vastly inferior in sensibility to our means of measuring a direct or continuous current. Thus no form of alternating current ammeter has yet been devised which is at all comparable in sensibility with the ordinary direct current mirror galvanometer. An instrument of the latter class can quite easily be made to produce a very large deflection of a spot of light upon a screen due to passage through the coils of the galvanometer of a current of one hundred millionth of an ampere. But it is extremely difficult to make an alternating current ammeter which will give any large deflection of a spot of light across a scale for a current of much less than one-thousandth of an ampere. Accord- ingly, we are much limited in our ability to detect electric oscilla- tions by the fact that they are alternating currents. If, however, means are available for rectifying these high frequency alternating currents, that is to say, eliminating the movement of electricity in one direction, and converting them into continuous currents, then all ordinary sensitive mirror galvanometers become available as instrumental means for rendering visible or evident the presence of oscillations in the circuit. Hence any appliance for rectifying electric oscillations is an important addition to our means of detecting them. To achieve this it must possess unilateral con- ductivity, in other words, it must permit the passage of electricity through it in one direction but not in the other, or permit it to flow in one direction under much less electromotive force than in the other direction. A very simple but effective form of oscillation valve was invented by the author in 1904, based upon researches made many years previously, viz. in 1890, upon the Edison effect in incandescent electric lamps. An ordinary incandescent lamp with carbon filament has a metal plate included in the glass bulb, or a metal cylinder, C, placed round the filament, the said plate or cylinder being attached to an independent insulated platinum wire, T, sealed through the glass (see Fig. 18). When the carbon is rendered incandescent by electric current, the space between the filament and the plate, occupied by highly rarefied gas, possesses a unilateral conductivity, and negative electricity will 206 RADIO TELEGRAPHY FIG. 18. Fleming Oscillation Valve. pass from the incandescent filament to the plate, but not in the opposite direction. This effect depends upon the now well-known fact that carbon in a state of high incandescence liberates electrons or negative ions ; that is to say, point charges of negative electricity. These electrons, or cor- puscles,!are constituents of the chemical atom. Hence, a carbon filament in an incandescent lamp is discharging from its surface negative electricity, which may even amount to as much as an ampere or even several amperes per square centimetre. If, then, an incan- descent lamp made as described has its filament rendered incandescent by a continuous current, and if another circuit is formed outside the lamp connecting the negative terminal of the filament with the insulated metal plate or cylinder in the bulb, and if oscillations are set up in this circuit, negative electricity will be able to move through this circuit from the filament to the plate inside the bulb, but not in the opposite direction. Hence, if an ordinary continuous current gal- vanometer is included in the external part of this last-named circuit, it will give a deflection when oscillations are set up in that circuit. For this purpose the author employs a small incandescent lamp with a rather thick carbon filament, or better still, a double or treble fila- ment, taking altogether about 2 amperes at a terminal voltage of 12 volts. The filament must be so constructed that it is brightly incandescent when OSCILLATION DETECTORS 207 12 volts are applied to the terminals of the lamp, or, to use a lamp manufacturer's phrase, the filament must be working at an efficiency of at least 3 watts per candle, Another coil in which oscillations are being set up is then connected with the negative terminal of the filament and the external terminal of the metal cylinder. This circuit also includes a sensitive galvano- meter, relay, or telephone. On setting up oscillations in that circuit they will be rectified, and the galvanometer will give a steady deflection as long as the oscillations last. In using this oscillation valve or glow lamp detector as a receiving arrangement in radiotelegraphy, the author places the valve, 0, in series with the relay or galvanometer, G, in the secondary circuits of an oscillation transformer, the primary circuit, p, of which is inserted in between the re- ceiving antenna and the earth (see Fig. 19). The valve then rectifies the oscillations pro- duced in the circuit, s, and the relay or galvanometer is affected. Marconi has employed this glow lamp detector with a tele- phone in a slightly different manner, as shown in Fig. 20. The antenna, A, is coupled through an oscillation trans- - former with a circuit which includes the valve, 0, and the fine wire coil of an ordinary FlG< 20 - 10-inch spark induction coil, I, the low resistance coil of which is in circuit with the telephone, T^. Condensers are placed across the secondary circuit of the oscillation transformer and also in series with the fine wire coil of the large induction coil. By suitable adjustments of the capacity of this condenser the circuits are brought into resonance. Oscillations taking place in the antenna, due to the impact of electric waves upon them, are then transformed by the oscillation transformer, rectified by the oscillation valve, and sent through the fine wire coil of the large induction coil in the form of unidirectional but intermittent currents, and these oscillations are again trans- formed up in current value by the large induction coil, and create 208 RADIOTELEGRAPHY (MM/WWWWWV a sound in the telephone. So used, the oscillation valve becomes one of the best long-distance receivers for electric waves yet devised. It is a characteristic property of ionised gases that the current through them does not increase proportionately to the electro- motive force, but soon reaches a value called the saturation current, beyond which no further increase takes place with increasing electromotive force. In the case of the author's glow lamp detector the current read on a galvanometer in series with the valve does not increase in value proportionately to the electro- motive force of the oscillations rectified beyond a certain limit. At or about 20 volts potential difference be- tween the cylinder and negative terminal of the carbon filament valve there is a very marked rise in the current for a small increase of voltage. It is possible, however, to calibrate the arrangement of valve and galvanometer, so as to enable the integral value of the oscillations to be obtained from the galvanometer deflections. This glow lamp detector has been much used by Lee de Forest, disguised under the name of an FIG. 2], audion, and claimed as his own invention. It was, however, described in scientific papers and in patent specifica- tions by the author long previously. The author has since found that greatly improved results can be obtained by employing a particular type of glow lamp with a Tungsten filament and a copper insulated cylinder surrounding it. The electronic emission from the Tungsten is greater than that from carbon, probably because it is a better conductor and can be raised without volatilisation to a much higher temperature than carbon. The author now uses this Tungsten glow lamp detector as follows : The battery which supplies current to incandesce the filament has a variable resistance, r, placed in series with it, and a high resistance, E, as a shunt across its terminals. A movable & OSCILLATION DETECTORS 209 contact on this shunt, K, is connected through a telephone, T, with one plate of the condenser in the oscillation circuit of the receiv- ing antenna (see Fig. 21), and the metal cylinder is connected to the other plate of the condenser. The two resistances can be so adjusted that the action of the oscillations taking place across the vacuous space in the vaWe is to increase the electronic emission, and therefore the currents through the telephone. If, therefore, trains of waves are infringing on the antenna, they are heard on the telephone as long or short sounds, depending on the number of trains incident. Another class of valve or rectifier detector is based upon an interesting property possessed by certain crystals of rectifying electric oscillations. It was discovered by General H. H. C. Dun- woody, in the United States, that a crystalline mass of carborun- dum, which is an artificial silicate of carbon, when supplied with electrodes, acts as an oscillation detector, and converts these oscillations into a continuous current. The crystal is inserted in the circuit of the antenna, and shunted by another circuit containing a telephone and a battery. When oscillations are set up in the antenna sounds are heard in the telephone, and it was found that the battery may be dispensed with and yet sounds continue to be heard. This phenomenon has been carefully investigated by G. W. Pierce, and he found that the current through the crystal in one direction under a given electromotive force was very much greater than with a current of the opposite direction under the same electromotive force. In other words, the carborundum possesses a marked unilateral conductivity. This property had, moreover, been previously found in some crystal metallic oxides and sulphides by F. Braun, but none of these showed such striking asymmetry as that shown by car- borundum. Thus, for instance, if a continually increasing voltage is applied to a mass of carborundum crystals, the following table, taken from a paper by G. W. Pierce, shows the currents pro- duced in one direction or the other in microamperes by that voltage. It will be seen that under an impressed electromotive force of 10 volts, the current in one direction is a hundred times greater than in the opposite direction, but this ratio decreases with the rise of voltage. 210 RADIOTELEGRAPHY EELATION OF CURRENT TO VOLTAGE, SHOWING UNILATERAL CONDUCTIVITY OF CARBORUNDUM. Current in microamperes. Volts. C C' C Commutator left. Commutator right. C' 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 1000 30 33 23-2 1200 50 24 25-0 1500 . 27-5 2000 120 17 Pierce has discovered that various other crystals possess the same unilateral conductivity, and may therefore be used in the same manner as oscillation detectors, when placed in the oscilla- tion circuit and shunted by a telephone and battery. One of these is the material called hessite, which occurs in nature as a telluride of silver or gold; and it has also been found that a crystal of oxide of titanium acts in the same manner. This crystal occurs naturally as a mineral known as octahedrite or anatase. The cause of this unilateral conductivity has been much discussed. It is possible that the cause may be thermoelectric, but sub- merging the crystal in oil does not appreciably change its behaviour, nor heating one junction more than the other. Pierce measured the current voltage or characteristic curve of carborundum, considered as a conductor, and found that it was not linear, but that the apparent resistance of the sub- stance falls as the current is increased, which implies a decrease of terminal potential difference with increase of current. In this respect, therefore, the carborundum and the other crystals resemble the electric arc and ionised air rather than metallic conductors. This, however, is not a complete explanation, because, as Dun- woody points out, carborundum may be used as a detector of electric waves without any battery in the circuit. Nevertheless, OSCILLATION DETECTORS 211 the property of the material above mentioned, viz. the non-linear character of its characteristic is sufficient to account for its unilateral conductivity, for, as H. Brandes points out, all con- ductors or combinations of conductors which do not follow Ohm's law, are capable of acting as detectors of electric oscillations owing to their rectifying effect. Hence, in any case in which the characteristic curve of a substance or its volt-ampere curve is not a straight line rising up from the origin, the material will be capable of acting as a rectifier for electric oscillations. Braun found that a number of substances, such as copper pyrites, iron pyrites, galena, and copper or antimony sulphide, possess unilateral conductivity, the current in some cases being twice as great in one direction as the other. He also found that no thermoelectric action could explain the phenomenon. Nothing, however, seems to approach carborundum in this peculiar property. Pierce found that in one specimen platinised on one side so as to make an improved contact, the current under an impressed electromotive force of 34' 5 volts was 527 times as great as the current in the opposite direction with the same voltage. In another case with 30 volts pressure the current in one case was 3000 or 4000 times greater in one direction than the other. As the current increases the efficiency of rectification decreases, but up to the present no theory has been proposed which satisfactorily explains this remarkable effect. 8. Electrodynamic Oscillation Detectors. Since high fre- quency alternating currents or electric oscillations create magnetic fields varying in a similar manner around the con- ductors through which they pass, and can induce oscillations of a similar frequency in neighbouring conductors, there are therefore attractions and repulsions produced between circuits through which oscillations are passing and others in which oscillations are induced by them, in virtue of the electrodynamic force between conductors conveying currents. We may, there- fore, construct oscillation detectors which depend for their operation upon these forces of attraction and repulsion. It was discovered independently by the author and by Elihu Thomson that if a metallic disc or ring is suspended by a fine wire within a circular coil through which an alter- nating current is passing, the ring or disc being held at an angle of 45 degrees to the plane of the winding of the coil, then the alternating current in the coil induces secondary currents in the disc or ring, and these create a mechanical force tending to make the disc turn, so that its plane is at right angles to the plane of the winding of the coil or parallel to 212 RADIO TELEGRAPHY the axis of the coil. It can be shown from first principles that when a closed inductive circuit is placed in an alterna- ting magnetic field it will be acted upon by a torque, compelling it to move into a position in which it includes the least number of flux lines. The author devised, as far back as 1887, an alternating current galvanometer depending on this principle, in which a copper or silver disc was suspended by a long fine wire in the above manner in the interior of a coil which could be traversed by alternating currents (see Fig. 22), The same principle has been applied as a detector of electric oscillations, in 1899, by Fessenden, who used a sus- pended silver ring and two fixed coils on either side of it through which the oscillations pass. More recently, G. W. Pierce has in- creased the delicacy of the instrument by employing a disc of silver paper suspended by a long quartz fibre, the plane of the disc being hung at an angle of 45 to the axis of an ebonite tube, on the outside of which was wound a coil of insulated wire convey- ing the electric oscillations. A small fragment of silvered glass attached to the disc serves to reflect a ray of light upon the scale and to indicate a move- ment of the disc. The average torque on the ring and therefore its deflecting moment is proportional to the square of the current in the coil and to the square of the frequency for the same instrument. Hence, if the frequency is constant and the ring is suspended by a quartz fibre of constant size and length, the restoring torque varies as to the deflection, and the deflection would measure the mean square or integral value of the oscillations passing through the coils. This conclusion has been confirmed by Pierce experi- mentally. This form of detector, therefore, like the thermal detectors, measures the integral value of the oscillations ; but since the mechanical forces are small, such an electrodynamic detector is not nearly as sensitive as the best forms of thermal detector. Nevertheless, in some quantitative researches it has proved itself to be very useful. FIQ. 22. OSCILLATION DETECTORS 213 9. Mode of employing Oscillation Detectors in combination with Recording Instruments to detect Electric Waves. In considering the above-described oscillation detectors it will be seen that, with the exception of the electrodynamic detectors, they may all be divided into three classes (1) Those which under the action of electric oscillations undergo a change which in effect is equivalent to an alteration of resistance. (2) Those which under the action of electric oscillations undergo a change which induces an electromotive force in another associated circuit. (3) Those which possess a unilateral conductivity, offering therefore a greater resistance to the passage of a current in one direction than in the opposite direction. As regards the first class, viz. those which under the action of electric oscillations undergo a change equivalent to a change in resistance, if we make the oscillation detector form a part not only of the oscillation circuit but of another circuit in which there is a continuous impressed electromotive force, and also some instrument capable of being affected by a change in this direct current, we can make the oscillation detector act the part of a relay in the following manner : The influence of the oscillations is to produce a change in the oscillation detector which virtually alters its resistance, making it greater or less. This action then in turn creates a change in the continuous current passing through it, generated by the con- tinuous current appliance such as the voltaic cell, and this change in the continuous current is then able to affect some instrument capable of recording it by a visible or audible indication. Thus, for instance, an imperfect contact detector, such as the metallic filings coherer, undergoes a change under the influence of electric oscillations which cause it to become a better conductor. If, there- fore, the metallic filings tube forms part not only of the oscillation circuit, but of another circuit containing a small unidirectional electromotive force, such as that provided by a voltaic cell, and some instrument capable of detecting a change in this current, such as a relay or galvanometer, then the influence of the oscilla- tions on the metallic filings tube will cause it to undergo a sudden decrease in resistance, and therefore there will be an increase in the continuous current passing through it. This increase in the continuous current may be made to record itself permanently by employing some form of telegraphic relay. A relay is a device by means of which a small increase in a very feeble current, or the passage through it of a very feeble 214 RADIOTELEGRAPHY current, is made to close another circuit containing a larger electromotive force and capable of carrying a larger current. A common form of relay consists of an electromagnet with coils of many turns. When a small current, say, of one milliampere, is passed through these coils the electromagnet attracts an arma- ture, and this is made to close another circuit which then permits the passage of a much larger current, capable of working any form of telegraphic recording instrument. Thus one may associate together the following devices : A metallic filings coherer and a single voltaic cell may be joined in series with the electro- magnet of a relay made as above described, and the second circuit of the relay may contain a battery of half a dozen cells and a Morse telegraphic printing instrument. If then the ends of the metallic filings coherer are connected with an antenna and balancing capacity, or with a pair of antennae, and if electric waves fall in the right direction on this antenna they create electric oscillations which pass through the metallic filings and suddenly lower the resistance of the mass. At this moment the single cell passes an increased direct current through the coherer and through the magnet of the relay, causing it to attract its armature and in turn to close the circuit of the larger battery and printing instrument, and in this manner to record a mark on a paper tape. In place of the coherer, relay and a printing instrument we may employ some form of self-restoring detector in series with a telephone, and then the increase in the current through the oscilla- tion detector due to its change in resistance under the action of oscillations suddenly alters the current flowing through the tele- phone, and it emits a sound which is heard as a short tick. These two methods are called respectively telegraphic and tele- phonic methods of receiving. Thus if in place of a metallic filings coherer we employ a thermal detector, then the action of the oscillations is to heat the fine wire of the detector and to increase its resistance, and if the wire is joined up in series with a single cell or telephone as above described, so that it is traversed not only by the oscillations but also by a continuous current due to the single cell, then the increase in temperature due to the oscilla- tions decreases this continuous current suddenly, and also causes a sound in the telephone. On the other hand, the magnetic detectors act in virtue of the change in magnetisation produced by the action of oscillatory magnetising forces brought to bear upon iron or steel. This change in magnetisation is made to create an induced electromotive force in another circuit embracing the iron or steel, and if this last circuit contains a telephone, the change in current through the OSCILLATION DETECTORS 215 telephone will give rise to a sound. Broadly speaking, we may say that the imperfect contact detectors and the thermal detectors and the electrolytic detectors experience under the action of oscil- lations a change which is equivalent to a change in resistance, whilst the magnetic detectors and the thermoelectric detectors act in virtue of a change which is equivalent to the production of an electromotive force in another circuit connected with the detector. In all cases the influence of the oscillations on the detector is made to bring about the increase or decrease of another current in another circuit, either by a variation of resistance or the introduc- tion of an electromotive force, and this last current is made to indicate itself either on a telegraphic instrument, by the inter- position usually of a relay, or else directly and audibly by means of a telephone. In the case of rectifying detectors the process is somewhat different. Here the oscillation detector is of a material, either rarified gas or crystal, which has a unilateral conductivity, and converts the oscillations directly into a continuous current, capable of being appreciated and indicated either by a galvanometer or bv a telephone. Taking the whole arrangement together, the oscillation detector and the associated circuit and the antenna connected to the oscil- lation detector, we have a sensitive appliance for detecting the passage of electric waves through space, which may therefore be called a cymoscope or wave detector when so used. In reviewing the action of oscillation detectors generally, we remark that in some of them the energy of the oscillations created in the receiving circuit is allowed to expend itself directly in affecting some indicating instrument, such as a telephone or galvanometer. In other cases the energy of the oscillations merely releases the energy of some external source, and it is this which affects the indicating instrument. In these last cases the action is called a trigger action, because it is similar to the operation by which the pressure of the finger on the trigger releases the energy of the powder which in turn propels the bullet. The coherer, electrolytic detector, and magnetic detector are instances of this last class, whereas the thermal detector, gaseous and crystal rectifiers are instances of the first class. In the next chapter we shall consider how these appliances are utilised for conducting radiotelegraphy. CHAPTEE YII RADIOTELEGRAPHIC STATIONS 1. The General Principles of Radiotelegraphy. Having described in the previous chapters the methods employed for generating electromagnetic waves and radiating as well as detecting them at distant places, we have next to consider the combination of these processes into a practically operative system of radiotelegraphy. To convey information to a distance we must be able to produce at the receiving station at pleasure certain visible or audible signals signifying letters, words, or ideas. For this purpose the most commonly used code is the International system of Morse signals, according to which each letter of the alphabet is denoted by a collocation of elementary signals of two kinds, one of short length or duration, called a dot, and the other of three times the length or duration, called a dash. Groups of these dots and dashes are made to succeed each other, with an interval equal to the length or duration of a dot between them, to form the various letters or numerals. Thus, the International Morse Code usually employed is as follows : THE ALPHABET. RADIOTELEGRAPHIC STATIONS 217 THE NUMERALS. I - - 6 - 2 - _ _ 7 3 - . - 8 - 4 - _ _ _ __ 9 _ 5 full a4-t\t\ Full stop Repeat - Call Signal If, then, we have the means of marking upon paper a collection of these signs or making them audible as short and long sounds in a telephone, we can signal out letters, and, therefore, words. A space equal to a dash is left between letters and a longer space between words. In the case of non-alphabetic languages, like Chinese and Japanese, the ideographs are numbered, and the numbers trans- mitted and translated. If the Morse characters can be printed on paper strip as received, we have a permanent record. If they are received by telephonic sounds, the observer translates them men- tally, and writes down the letter on paper as received. Accordingly, the broad principles of all radiotelegraphy are as follows : At one place, called the transmitting station, there must be an antenna or radiative circuit, called the sending antenna or radiator, and means must be provided for creating electric oscil- lations in this circuit, which may be damped or undamped. A circuit closer, called a sending key, must be included in the transmitting circuit, by which the oscillations or successive trains of oscillations can be started or stopped at pleasure, and hence electromagnetic waves radiated in trains or groups of trains, of shorter or longer duration, to correspond to the signals of the Morse alphabet or any other similar signal code. At some other place, called the receiving station, there must be a similar antenna or absorbing circuit to absorb these waves. In the course of this circuit, or connected with it, there must be some form of oscillation detector to be influenced by the oscilla- tions set up in the absorbing circuit by the impact on it of the waves sent out from the corresponding transmitter. This oscilla- tion detector must have some appliance, such as a telephone or telegraphic recording instrument, connected with it, which it influences, and then there must be an observer to hear or note the 218 RADIOTELEGRAPHY signals so given, which correspond with those made by the sending key at the distant place. By this means we transmit signals without continuous wires, by means of electromagnetic waves, from one place to another, which are interpretable as alphabetic or intelligible signals conveying information, and this constitutes the art and practice of radiotelegraphy. The precise appliances employed differ according to the distance of the stations and their locality. The particular characteristic of radiotelegraphy, as compared with conductive or fixed wire tele- graphy, is that one or both of the stations may be in motion. Hence, it forms an ideal means of communication between two ships at sea or between ships and the shore. Also, since sea water is a good conductor relatively to dry land, communication by radiotelegraphy over sea is facilitated, as already explained, and it is therefore especially marked out as a means of super- marine communication. Nothing is more remarkable than the rapid rate at which the maximum distance of communication over-sea by radiotelegraphy has been extended. Prior to 1896 no one had been able to demonstrate the detection of an electro- magnetic wave at a greater distance from its generator than about half a mile. In 1897, Marconi gave the first demonstration in England of actual radiotelegraphy over a distance of several miles, and in 1898 had an operative system at work between Bourne- mouth and the Isle of Wight, a distance of about twelve miles. In 1899 he accomplished the feat of radiotelegraphy across the English Channel, and for the first time drew public attention strongly to the possibilities of the new telegraphy. By the end of that year, or the beginning of 1900, a distance of 100 miles was covered by him, and inventors all over the world were endeavouring to follow him in these achievements. At the beginning of 1901 he had signalled in this manner 200 miles, from the Isle of Wight to Cornwall, and at the end of 1901 had succeeded in sending alphabetic signals 3000 miles over the Atlantic Ocean. Since that date its achievements have steadily been extended by him and others, and in 1908, or seven years from the transmission of the first transatlantic signals, regular radiotelegraphic communication was established between Ireland and Nova Scotia by the ingenuity and perseverance of Mr. Marconi and those associated with him. By that date also every navy in the world had adopted it as an indispensable means of signalling. We shall, then, in the following sections consider in detail the apparatus now employed for short and long distance radiotelegraphy, RADIOTELEGRAPHIC STATIONS 219 taking the various elements of the apparatus in succession, and dealing first with that employed for short distances, viz. up to 100 or 200 miles, both on shore and on ship, for the purposes of supermarine intercommunication. 2. Short Distance Radiotelegraphie Apparatus. Antennae or Radiator Supports. In establishing a short distance radio- telegraphic station, the first question to be considered is the site and the erection of a support for the antenna. Another important matter is the possibility of securing a "good earth," FIG. 1. Marconi's Wireless Telegraph Station at Poole, Dorset. in a telegraphic sense. As the greater part of short distance radiotelegraphy is conducted over sea and is concerned with communication with ships, such a station is nearly always established on the coast. It is desirable that the soil at the point selected should not be too dry or rocky, as hard dry rocks are poor conductors, and render it difficult to obtain the necessary earth connection. It is also desirable that the site should not be overtopped by hills, and that it should have an open outlook to the sea in the direction in which the transmission is chiefly 22O RADIOTELEGRAPHY desired. The soil, however, should be sufficiently firm to enable good foundations to be obtained for the mast or masts used as the antenna support. Assuming the station to be for short distance work, a single mast generally suffices for this purpose. This is usually erected in three sections, 50- or 60-feet poles being employed, each of which must be well stayed, and if galvanised iron wire is employed for the stays, these should be interrupted by insulators at intervals, so as to avoid having long wires which might have a natural time-period of oscillation equal to that of the antenna employed. At the top of the mast a gaff is erected with pulley and tackle for hauling up the antenna wire (see Fig. 1). In the case of ship installations it is usual to provide either the main or fore mast with an additional gaff to carry the antenna. The antenna itself is preferably constructed of hard drawn tinned copper, phosphor bronze, or aluminium wire, and it is nearly always in the form of a multiple wire antenna. If a fan- shaped multiple wire antenna is employed, then it is generally necessary to erect two FIG. 2. masts with a horizontal or triatic stay between them, from which the antenna is suspended (see Fig. 2), but if a single mast is employed, then the antenna may be of a double cone, or preferably of the umbrella form (see Fig. 11, Chap. V.). When the spark system is employed, then, owing to the high potential of the extremities of an open antenna, the upper end must be extremely well insulated. A form of insulator devised by the author, suitable to this purpose, is made as follows : A thick- walled ebonite tube, about 2 feet or 60 cms. in length, has a brass wire down the centre ending in a loop at the bottom and a button at the top. The top button is covered over with an ebonite cap fitting perfectly watertight. The ebonite tube is gripped by a cross oak bar just below the top, and below this a conical or tubular rain shield is attached watertight to the top, so as to RADIOTELEGRAPHIC STATIONS 221 keep the ebonite as far as possible dry. One or two such insulators may be used in series. Similar insulators are used to draw out the wires into desired positions. A closed or loop antenna may be constructed of two or more antenna wires connected together at the top and sustained by an FIG. 3. insulator, if required, the middle or some lower points of the two wires being pulled out by rope stays attached to insulators, so as to form a lozenge-shaped or triangular closed or loop antenna. In cases in which bent or directive antennae of Marconi's form are employed, then two or more masts have to be erected to sustain the hori- zontal portion of the antenna in the required direction (see Fig. 3). In some cases a group of four masts are arranged at intervals about equal to their height, the tops being joined together by triatic stays, and from these a fan or conical multiple wire an- tenna brought down to a central point (see Fig. 4). It is hardly necessary to enter Fia - 4 - into the details of the erection and staying of such masts, as these are well under- stood by the riggers or mast builders, who would in any case be employed. Unless the mast or masts are erected in contiguity to some already existing building or lighthouse, it is then necessary to 222 RADIOTELEGRAPHY erect near the base of the mast or masts in some convenient position a building which serves as a radiotelegraphic station, which may be of wood, brick, or galvanised iron, as most con- venient. Into this building the antenna is then led, and an important detail of construction is the means necessary to secure the necessary insulation and yet weather tightness. One method is to cut a hole in a pane of glass in a window, or to make a window glazed with a single sheet of thick plate glass. In this a hole is bored, through which passes a thick-walled ebonite tube made tight with indiarubber washers. It is advantageous to protect this glass from the weather as far as possible by an overhanging hood facing in the direction opposite to that of the prevailing wind. Through the ebonite tube passes a thick stranded copper wire cable well insulated with indiarubber. The outer end is connected to the bottom of the antenna, and the inner end is brought to a highly insulated switch, by means of which it can be connected as required to the trans- mitting or receiving apparatus. In nearly all cases the same antenna serves alternately as a sending or radiating and as a receiving or absorbing antenna. Hence, a long break switch is required for throwing it over quickly from one apparatus to the other. As already observed, an important element is the earthplate or connection to earth, which is unquestionably necessary for effective radiotelegraphic communication for any considerable distance. This is made by burying in the earth strips of sheet copper or sheet zinc, or thick copper wires radiating from a convenient point like the roots of a tree. Galvanised iron should not be used, as it soon corrodes in damp soil. The object should be to obtain the greatest possible surface exposed to moist earth, as in the case of a lightning conductor, and in order to be able to test the resistance of this earth, it is desirable to construct this earth-plate in two portions separated from one another. The resistance between these two portions can then be measured by a Wheatstone's bridge or any other method, but if a single earth is employed the earth resistance cannot easily be measured. In any case a thick stranded cable or wide copper strip must be connected to the earth-plate or earth-wires and brought into the sending and receiving room of the radiotelegraphic station ; and during all times when the apparatus is not being used, the antenna should be connected to the earth-plate directly, so as to form a well-earthed lightning conductor. A thoroughly good antenna should resemble a tree, in having almost as much beneath the ground as above it, the trunk and RADIOTELEGRAPHIC STATIONS 223 branches corresponding to the exposed portion of the antenna and the ramified rootlets to the earth- wires. In the case of a ship no difficulty exists in obtaining a good earth, because a connection can be made to the copper sheathing or to the outer plates of the ship. In dry places it is desirable to make arrangements for putting water down on the earth-plate, so as to keep the soil round it moist. One of the great difficulties of effecting radiotelegraphy in inland places in many tropical countries is the dryness of the soil and the difficulty of getting a thoroughly good earth connection. It is possible to operate over short distances without an earth connection. It is then usual to make a balancing capacity by spreading out over the earth a sheet of metal or galvanised iron netting, which in some cases is insulated from the earth. This forms with the earth a condenser, or the metal plate takes the place of the earth in the ordinary mode of working. 3. The Arrangement of the Transmitting Apparatus for Short Distance Radiotelegraphy. In the case of radiotelegraphy con- ducted with damped electrical oscillations, or so-called spark telegraphy, the transmitting apparatus comprises three elements. (1) Some means for charging the antenna or other condenser to a high potential. (2) A discharger permitting this charge to flow out of the con- denser with oscillations which are either communicated directly or inductively to the antenna. (3) Means for controlling these oscillations or repeating them in long or short groups in accordance with the Morse signals. Tor short distance spark telegraphy the necessary high potential is always obtained by the use of an induction coil or transformer ; either a single instrument or a number of induction coils or transformers may be employed, having their secondary circuits joined in series and their primary circuits in parallel. The most usual appliance is an induction coil of the ordinary type giving a spark 30 to 60 cms. and taking current either from primary or secondary batteries or from a small alternator. The induction coil is placed on a table in the transmitting station, or it may be fastened to the wall. It may be operated either by alternating currents or by an interrupted continuous current. If operated by a continuous current, this may be taken from a battery or a dynamo. In isolated places and on lightships and in lighthouses it is usual to employ secondary batteries which are charged by primary batteries. The ordinary 10-inch induction coil usually requires a primary current of 10 amperes at 16 volts. Hence, eight or ten 224 RADIOTELEGRAPHY secondary cells are sufficient to work it. These, however, must be charged by a battery of primary cells, say, 20 primary cells in series, each giving 1/5 volts E.M.F. One arrangement which may be employed is to connect up 100 large dry cells, five in parallel and twenty in series, and join them in parallel with ten secondary cells, and connect the ends of the secondary cells to the terminals of the primary circuit of the induction coil. The primary cells then charge the secondary cells, and the secondary cells give up current to the coil as required. In land stations a convenient arrangement is a small oil engine and continuous current dynamo, which is employed to charge the secondary cells, which in turn are used with the coil. On board ship the current can be taken from the lightning circuits, which generally furnishes a continuous current at 50 or 100 volts. In cases where the continuous current is employed, some form of interrupter must be used with the coil. The ordinary hammer break with platinum contacts is still used on board ship on account of its simplicity, but mercury turbine breaks are also much employed. In this latter case it is usual to suspend the mercury break in gimbals. The Wehnelt or electrolytic break is not often used, as it involves the use of acid and is troublesome to keep in order and somewhat irregular in action. The bulk of the work now done may be said to be divided between the hammer break with platinum contacts and the mercury break. The next element in the transmitting apparatus is the signalling key, for interrupting the primary circuit in accordance with the signals of the Morse alphabet. This must be a quick break key with a long ebonite handle easily operated, and has generally a magnetic blow-out in connection with the platinum terminals between which the interruption takes place. When alternating currents are employed, a key has been devised by means of whicji the circuit is only actually interrupted at a time when the alter- nating current passes through its zero value. This is achieved by means of an armature, which when once pressed down by the movement of the hand, keeps the circuit closed until the alter- nating current in the course of its cycle passes through its zero value, when the key automatically opens its circuit again. In cases where alternating current is supplied, or can be provided by means of a small alternator, it is usual to substitute a closed iron circuit transformer, insulated in oil, for the induction coil, and this is also done on board ship in the case of high power trans- mitters for long distance work. We have next to consider the methods of creating the oscilla- tions in the antenna. The simplest way of doing this is by RADIOTELEGRAPHIC STATIONS 225 employing the induction coil to charge the antenna directly, as in Marconi's original invention. (See Fig. 5). In this case, the antenna, A, is connected to one of the secondary terminals of the induction coil, I, the other terminal being connected to the earth, E, and the two terminals also joined to a pair of spark balls, S, which can be more or less approximated. At each inter- ruption of the primary circuit an electromotive force is created in the secondary circuit, which charges the antenna, and when this potential difference reaches the spark potential corresponding to the distance at which the spark balls are placed, the antenna dis- charges itself across the gap with oscillations, which, however, as already explained, are highly damped, probably not more than half a dozen oscillations at most taking place.- The most usual way to charge the antenna, however, is by connecting it directly or inductively to a closed condenser circuit interrupted only by the spark gap. In this case the secondary terminals of the induction coil, I, or transformer are connected to the adjustable spark balls, S, and these balls are * also connected by a condenser, C, con- sisting usually of a battery of Leyden jars, and by FlG> 5< an adjustable inductance coil, L. The antenna, A, is connected directly to one point on this circuit (see Fig. 6), some other point being connected to earth. The antenna and condenser circuits must be syntonised. This may be done by connecting in series with the antenna another adjustable or tuning inductance by means of which the natural time period of the antenna is made to agree with that of the condenser circuit. This tuning may be achieved by connecting a hot wire voltmeter over one or two turns of the inductance in series with the antenna, and then altering this inductance, or else altering the capacity and inductance in the condenser circuit until the indications of this voltmeter are a maximum. Marconi prefers to connect the antenna, A, inductively with the energy-storing circuit by means of an oscillation trans- former called a transmitting jigger. This jigger consists of a Q 226 RA DIO TELEGRAPHY wooden frame on which are wound two circuits, p, s, a single ^^%^^<^^ FIG. 6. turn or two of a small inductance which forms part of the con- denser circuit, and another overwound circuit 5 to 10 turns in FIG. 7. series with the antenna, and with a tuning inductance, L (see Fig. 7). RADIOTELEGRAPHIC STATIONS 227 This last method possesses the great advantage that the closeness of coupling of the antenna and condenser circuit can be varied. It has already been shown in Chapter I., that when two oscilla- tion circuits are connected together inductively, and in tune with one another, oscillations set up in one circuit result in the produc- tion of oscillations in both circuits, having two frequencies, one greater and the other less than the natural frequency of each circuit when separate. Hence, when we are employing an induc- tively coupled antenna which has been syntonised with the condenser circuit, it is necessary to bear in mind that oscillations of two frequencies are set up in the antenna, and waves of two wave lengths radiated from it, one greater and the other less than the wave length corresponding to the natural frequency of the antenna taken alone. One of these waves has greater amplitude than the other, the longest wave length being the least damped, and, therefore, generally speaking, having the greatest integral value. The wave lengths approximate to one another in propor- tion as the coupling is made weaker, but then they also diminish in amplitude, so that by the employment of a weak coupling, which can be done by separating the primary and secondary cir- cuits of the oscillation transformer, we gain a radiation which is relatively feeble, but of one single wave length, whereas by coupling closely together the primary and secondary circuit of the oscillation transformer we have more powerful waves but waves of two wave lengths radiated, and the receiving antenna must accordingly be syntonised to one or other of these wave lengths. As regards condensers for the oscillation circuit, although a Ley den jar is a bulky form of condenser in comparison with its energy-storing power, nevertheless its simplicity still recommends it. The main condenser generally consists, then, of a battery of Leyden jars, a certain number being joined in parallel or partly in parallel and partly in series. It is very important that these jars should have their capacity marked upon them, and that they should be selected so as to be exactly equal. When oscillations are taking place in the condenser circuit, it will be seen that an electric brush discharge takes place from the free edge of the outer tinfoil. It has been shown by Eickhoff that this brush discharge involves a considerable expenditure of energy, and, therefore, tends to damp out the oscillations. The author has also shown that it increases the capacity of the jar by an irregular amount. Both these effects can be stopped by putting the jars in insulating oil, and in place of Leyden jars it is in every way preferable to employ a condenser consisting of sheets of glass or ebonite coated with 228 RADIOTELEGRAPHY metal and immersed in oil, or the glass and ebonite may be omitted, and the condenser constructed simply of metal plates immersed in oil. Brush discharges are thereby prevented, and accuracy of tuning is secured by preserving a constant known capacity in the condenser circuit. In the case of either the inductive coupling or the direct coupling of the antenna to a condenser circuit, it is of the greatest importance to secure an exact syntonisation between the antenna and condenser circuit, as a very little want of syntony immensely decreases the strength of the current flowing into and out of the antenna. Other things being equal, the radiation from the antenna will be proportional to the mean square value of the current flowing into the base of the antenna. This current may be measured by inserting in that point a hot wire ammeter. Another important element in the transmitting arrangement is the spark discharger. Where large capacities are being employed, the noise of this spark is very distressing, and moreover it enables the messages to be read at a great distance by any one familiar with the Morse alphabet. Hence, for many reasons, the spark balls should be enclosed, as first suggested by the author, in a cast-iron chamber with a glass peephole in it. The chamber is preferably kept full of nitrogen or carbonic dioxide gas, and should also have some lime or alkaline material in it for absorbing acid vapours. In this manner the discharger can be rendered perfectly noiseless. It is a great advantage to blow a jet of air upon this spark gap to quench the arc. This can easily be done by means of a small Lennox blower operated by an electric motor. In the case of short distance transmitting plant, the discharger generally consists of a pair of balls of iron or brass, with arrange- ments for turning them round in such a way that every now and again fresh surfaces can be brought into apposition, as the continual action of the spark erodes or cuts away material from the spark balls, leaving them rough. The general appearance of this Marconi transmitting plant is as shown in Fig. 8, which gives a view of the Marconi transmitting apparatus used on Atlantic liners, the set comprising two induction coils and two lots of Ley den jars for producing signals of two wave lengths. At one time it was considered an advantage to divide the spark up between several spark surfaces by putting balls in series, but this does not appear to be the case, and it is usual now to employ a single spark gap. In those cases in which an induction coil is used with a thick secondary wire or a trans- former is employed for charging the condenser circuit, inductances RADIOTELEGRAPHIC STATIONS 229 must be inserted between the spark balls and the coil or trans- former, to stop the arcing which would otherwise happen. So far we have only considered the use of damped oscillations for transmitting, with which at present (1908) the bulk of the short distance radiotelegraphic work of the world is conducted. If [By permission of Marconi's Wireless Telegraph Co., Ltd. FIG. 8. Marconi Wireless Telegraphic Transmitting Apparatus. undamped .waves are employed, then a Poulsen arc takes the place of the spark balls. The copper-carbon arc apparatus, already described in Chapter III., would have its electrodes connected by a small capacity con- sisting of metal plates in oil and a large inductance coil, this oscillation circuit being so arranged as to have a frequency of something of the order of a million. The antenna is generally 230 RA DIOTELEGRAPHY directly connected to this oscillation circuit and tuned to it. The signals are then made by short circuiting, by means of a key, a few turns of this inductance so as to throw the oscillation circuit out of tune with the antenna. The arc has to be fed from a con- tinuous current dynamo giving an E.M.F. of 400 to 500 volts, suitable regulating resistances being interposed. The arc is enclosed in a metal box with cooling flanges attached, clockwork r r _ i or an electric motor being employed to rotate the carbon. In place of hydrogen or coal gas the vapour of alcohol is now used to exclude the air. Alcohol is admitted drop by drop to the arc chamber from a sight feed lubricator attached to it (see Kg- 9). For short distance work an arc formed with 220 volts and a small current of 2 or 3 amperes is sufficient, but for long distance work an arc taking 8 to 10 amperes at 440 volts is employed. In this last apparatus the magnetic field in which the arc burns is pro- vided by a large electro- magnet on which the arc box is fixed (see Fig. 9). 4. Short Distance Receiving Apparatus. Turning next to the receiving arrangements, we find that they differ chiefly in the nature of the actual oscillation detector employed and the recording or signal making appliance. It is a very great advantage to possess a permanent automatic record of the signals made by some form of telegraphic instrument. On the other hand, the mechanical and electrical inertia involved in some appliances limits the speed. "VVe may avoid this, and [By permission of the Amalgamated Radiotelegraphic Co., Ltd. FIG. 9. Poulsen Arc Apparatus. RADIOTELEGRAPHIC STATIONS a self-recovering increase the speed by using a telephone with oscillation detector or a photographic recorder. Accordingly, the receiving apparatus may produce self-recording visible signals on telegraphic paper tape, or audible signals by means of a telephone. In the first case, some kind of imperfect contact oscillation detector may be employed in conjunction with a telegraphic relay, and print- ing telegraph or recorder. In the other case, a magnetic, electrolytic, FIG. 10. Arrangement of Marconi Wireless Telegraphic Receiving Apparatus, employing a Coherer, Relay, and Printer. or rectifying detector with a telephone is used, or a thermoelectric detector in combination with a photographic recorder. The arrangements for telegraphic reception employed by Marconi for short distance working are as follows : The antenna, A, is connected to earth through the primary coil, /!, of an oscillation transformer, and a tuning coil is inserted in between the earth plate and this primary coil. The secondary 232 RADIOTELEGRAPHY coil, j 2 , of the oscillation transformer is cut in the middle and a condenser inserted. The outer ends of the secondary coil are connected to a Marconi metallic filings coherer tube, T. The terminals of the tube are also connected by a second or tuning condenser (see Fig. 10). The ends of the first-named condenser [By permission of Marconi's Wireless Telegraph Co., Ltd. FIG. 10A. Marconi Eeceiving Apparatus. are connected through two small choking coils, Ci, C 2 , with the electromagnet circuit of a telegraphic relay, E, and with a couple of dry cells, B, so that when the coherer tube is made conductive by the oscillations a current from these cells flows through the tube and actuates the relay. The relay is then made to work a Morse inker by means of 6 or 8 more cells. The whole RADIOTELEGRAPHIC STATIONS 233 apparatus, coherer tube, tapper, relay, oscillation transformer, condensers and cells, is mounted up on a stout baseboard, and enclosed in a metal box with sliding door so that the operator can get his hand in to the numerous set screws which adjust the relay and tapper to their best positions (Fig. 10 A). The relay used is the so-called polarised relay. In this a horseshoe electro- magnet is fixed on one end of a permanently magnetised steel bar, the other end of which is bent round and carries a delicately pivoted steel tongue, which is held with its free end between two soft iron pole pieces fixed on the ends of the electromagnet. The steel tongue then tends to stick to one or other of these poles. It is pulled away from one pole by an adjustable spring, and is held in balance against a stop, so that if a very small current is sent through the coils of the electromagnet, it magnetises one of the soft iron poles North and the other South, and causes the movable tongue to fly over against the other platinum point stop and make a contact with it, whilst at the same time it closes a second electric circuit. In the case of relays used on board ship, the tongue must be so balanced that there is no tendency for it to move merely by the pitching or rolling of the ship, and the relay must furthermore be enclosed in an airtight box to prevent the damp sea air depositing moisture on the platinum contacts. Such a relay will generally have a resistance of 1000 ohms or more in its electromagnet coils, and will work with a current of O'l of a milliampere and close a circuit, enabling a current of O'l ampere or more to be passed through the other circuit which is closed by the relay. The adjustment of the relay is effected by turning a screw which moves over the pair of soft iron poles together one way or the other, and so alters the pressure of the tongue against one of its stops. The recording instrument generally used in telegraphic receiving is called a Morse inker, and consists of a clockwork mechanism which drives a strip of paper tape under a roller at a uniform speed (see Fig. 11). At one end of the instru- ment an electromagnet, M, acts upon an armature carried on the end of a lever, to the other end of which is attached an inking wheel, W, which dips into a well of ink, I, so that when the armature is attracted by the magnet the wheel is pressed up against the underneath side of the paper tape, p, and marks upon it a short or long line, according to the time during which the electromagnet is excited. Hence, when the relay closes the circuit of a local battery, B, which last is in series with the magnets of the Morse inker, a mark will be made upon the paper. 234 RADIOTELEGRAPHY At the same time the relay closes the circuit of another electro- magnet, to the armature of which is fixed a hammer like that of a trembling electric bell. This tapping magnet is so arranged that the hammer strikes the underneath side of the coherer tube, and there are adjusting screws which regulate the range and frequency of the blows. The whole operation taking place in the receiver is then as follows : When an electric wave falls upon the antenna it excites oscillations in the antenna. These are transformed by the oscillation transformer or receiving jigger, as it is called, and these oscillations pass through the coherer tube, causing it to become conductive. This then passes a current from the single or double cell through the relay magnet, and actuates the relay. This last in turn closes the circuit of the larger battery and sets in operation the Morse inker and the trembling tapper, which at FIG. 11. Scheme of Connections of Relay and Morse Inker. R, Relay magnet. , TZ, Terminals of relay. a, b, Relay contacts. t, Relay tongue. My Morse inker magnet. B, Local battery. W, Ink wheel. P, Paper strip wheel. I, Ink well. p, Paper tape. once taps the coherer tube back to a condition of non-conductivity, and at the same time a mark is made upon the paper as long as the waves or trains of waves continue to fall upon the antenna. A number of adjustments are necessary to get a good result. The relay must be adjusted to proper sensibility, and also the magni- tude and force of the blow given by the trembling hammer on the underneath side of the coherer tube, by the adjusting screws. Also it is necessary to regulate the sensitiveness of the Morse inker by altering the tension of a spring which pulls the armature away from the electromagnet. The apparatus requires a certain skill in management, and the training required in an operator is not merely that of learning to send accurately and quickly the Morse signals with proper spacing by the transmitting key, but much more in adjusting the numerous regulating screws of the RADIOTELEGRAPHIC STATIONS 235 receiving apparatus, so as to make it record accurately and unintermittently the signals received. In the Lodge-Muirhead receiver using the self-restoring mercury and steel disc detector described in Chapter VI. no tapper is necessary. By means of a shunted voltaic cell an electromotive force of a fraction of a volt is applied between the mercury and the wheel through the circuit of a Kelvin syphon recorder. When oscillations are passed across from the mercury to the disc they temporarily break down the insulation of the oil film and the external E.M.F. then causes the glass pen of the syphon recorder to make a deflection, which, however, subsides as soon as the oscillations cease. Hence, on the paper tape the uniform straight line drawn by the pen is broken lay a sudden sharp notch or hump representing a dot and a larger square-shouldered hump representing a dash, according as the pen is kept deflected for a shorter or longer time. Although there is a great advantage in possessing a permanent record on the tape, the numerous adjust- ments then necessary and the reduction in speed of reception due to mechanical and electrical inertia in the receiving arrangement cause preference to be given in many cases to the telephonic method of reception, which possesses much greater sim- plicity. In this case, the receiving apparatus consists of the antenna and the oscillation transformer, the primary circuit of which is inserted in the antenna circuit, and the secondary circuit connected to a condenser. To the terminals of this condenser are connected the electrolytic receiver or other detector employed, and to the terminals of the electrolytic cell are also connected a circuit including a telephone and a single voltaic cell, shunted by a variable shunt (see Fig. 12). In this case, the oscillations set up in the antenna, A, give rise to secondary oscilla- tions in the circuit comprising the tuning condenser, C, and the secondary circuit of the oscillation transformer, and these oscilla- tions, when they have reached a certain amplitude, act upon the electrolytic receiver, V, and cause its resistance to fall. The local voltaic cell, B, then sends a current through the electrolytic cell and through the telephone, T, causing a sound in the latter '; but as soon FIG. 12. 236 RADIOTELEGRAPHY as the oscillations cease in the antenna, the electrolytic cell rises again instantly to its former resistance. Accordingly, if trains of oscillations fall upon the antenna, a rapid succession of short sounds is made in the telephone, which run together into a sound of continuous duration, prolonged as long as the trains of waves fall on the antenna. Hence, audible signals, corresponding with the dot and dash of the Morse alphabet, can be made by a suitable emission of long or short trains of waves from the transmitting antenna. The local receiving circuit, comprising the condenser and an inductance, may be either inductively coupled to the receiving antenna or directly coupled. In any case, variable inductances are inserted in the antenna circuit and in the connected oscillation circuit, to bring the two into syntony with each other. If the coupling in the transmitting circuit is inductive, and also that in the receiving circuit, then there are four circuits which have to be brought into syntony with each other, viz. the closed sending circuit, the associated sending antenna, the receiving antenna, and the closed local receiving circuit, and unless this tuning is accurately done, the apparatus will be wanting in sensibility. So far, we have been considering only short distance receiving apparatus for spark telegraphy. If, however, an electric arc is used in the transmitting circuit, or other means for emitting from the sending antenna continuous or undamped waves, then some modifications are necessary in the receiving arrangements. Thus, if an electrolytic receiver is employed, it must be intermittently disconnected from the local receiving oscillation circuit, or other- wise no sound will continue to be produced in the telephone during the emission of the waves from the sending antenna, and it would not be possible, therefore, to signal a dash as well as a dot. If, however, an electrolytic detector is rapidly connected and discon- nected from the oscillation circuit, two effects are produced. In the first place, during the time the electrolytic detector is not connected to the local receiving oscillation circuit, the feeble oscillations produced in the receiving antenna have time to work up more powerful oscillations in the inductively connected closed receiving circuit in virtue of the principle of resonance, by the accumulated effects of properly timed impulses. The amplitude of the potential variations of the terminals of the condenser is therefore increased as much as possible. If, then, the electrolytic cell with which is associated the voltaic cell or telephone is connected to the condenser, a sound will be heard in the telephone, and if this connection and disconnection is made rapidly, it will produce in the telephone the effect of a continuous sound as long RADIOTELEGRAPHIC STATIONS 237 as the trains of undamped waves continue to fall on the receiving antenna. Hence, V. Poulsen introduced a device he calls a ticker, which is nothing more than a vibrating contact worked by an electromagnet, or other device for rapidly connecting or discon- necting the detector from the local receiving circuit. The same effect might be produced by cutting up the persistent waves at the sending station into small disconnected portions. Practically, it comes to this, therefore, that to conduct Morse signalling by means of undamped electromagnetic waves, it is necessary to rapidly intermit the trains of oscillations either at the sending station or at the receiving station to be able to create a dash as well as a dot signal. In connection with his methods of reception, V. Poulsen has devised an ingenious photographic recorder. The oscillations in the receiver circuits are made to affect a thermoelectric detector, and the current so produced is passed through a string galva- nometer, consisting of a strong electromagnet having a single wire traversed by the current from the thermocouple placed in a very narrow air gap of the magnet. Hence, even a very feeble current passed through this wire causes it to be displaced across the magnetic field, and in so doing it is caused to uncover a small slit in a plate and permit a ray of light to fall upon a moving strip of sensitive photographic film. It thus records on the film a dot or a dash, according to the time during which the aperture is uncovered by the deflected wire. The film, after being exposed, is immediately developed and fixed, and can be inspected after a few minutes. Views of the recorder and photographic slip with signals on it are shown in Fig. 13. 5. Systems of Intercommunication by Short Distance Eadio- telegraphy. Owing to the fact that telegraphy is in most countries a Government monopoly (the principal exception being the United States), and having regard to the special advantages of radiotelegraphy for supermarine communication, the great field of operations for it is found in communication between ships and between ships and the shore. The Marconi Wireless Telegraph Company, Limited, formed in 1897 to work the Marconi system, began in 1899 to create such a system of intercommunication, and has since established all over the world a very large number of stations on the coast for communication with ships. The stations are established at very many places on the coast of Great Britain, Canada, the United States, Italy, and in other places, and vessels of numerous lines working on the Atlantic Ocean are equipped with corresponding sending and receiving apparatus. A complete " wireless exchange" 2 3 S RADIO TELEGRA PH Y has thus been established, by means of which these vessels can communicate with each other when at sea, and with various ports. The Marconi Company was not only the first in the field to establish, but even to-day (1908) is the only company operating a fully organized system of intercommunication with ships equipped with apparatus suitable for communicating with numerous shore 110 words per minute 5 x 10 6 amp. 60 words per minute 1 x 10 6 amp. [Reproduced from " The Electrician " by permission of the Proprietors. FIG. 13. Poulsen Photographic Eeceiver and Jlecords. stations. For this purpose certain wave lengths had to be selected, such as 300, 500, 1000, or 1500 feet, two very commonly used wave lengths for supermarine communication being 1000 feet and 2000 feet, or 300 metres and 600 metres. Each station and ship on which a wireless installation is made has its sending and receiving apparatus tuned for the same or for certain wave-lengths. It is also designated by a letter or a pair of letters, called a " call-signal." RADIOTELEGRAPHIC STATIONS 239 Thus, the Marconi wireless station established at the Lizard, in Cornwall, is denoted by L.D., and in the same way the Crook- haven station at the South of Ireland is denoted by O.K., and of the vessels on which installations are made, such as the ss. Campania, of the'Cunard Company, the call signal is C.A., and of the ss. Deutschland it is denoted by D.L. Some of these vessels are equipped with short distance apparatus : that is, for communicating with one another and with the shore up to 200 miles. Others are equipped with long distance apparatus for communicating with long distance stations, as described in a subsequent section. Each station established on the coast and each ship has, therefore, a certain range of opera- tions, which, to some extent, depends upon the atmospheric con- ditions. On board the vessels a special tele- graph cabin (see Fig. 14) is set apart for the radiotelegraphic work, and, equipped with the apparatus for sending and receiving, as already described, worked by a skilled operator. A view of the interior of one of the shore stations is shown in Fig. 15. The transmitting apparatus is seen on the right hand and the receiving apparatus on the left, and the Morse printer and sending key in the centre. On the ship the antenna is suspended from a gaff attached to the main or fore mast, and brought in through an insulator to the cabin. The Marconi shore stations are, by special agreement with the British postal telegraphic service, connected with the land lines of the country, so that they are in close correspondence with every place in which there is a postal telegraph office. Supposing, then, that the ss. Campania, approaching the south of Ireland, desires to communicate with Crookhaven station in the south of Ireland. Her operator sends the call signal C.K. several times at intervals, waiting in between to see if there is any response, and on getting the return call signal from Crookhaven, [By permission of Marconi's Wireless Telegraph Co., Ltd, FIG. 14. Marconi Wireless Telegraph Cabin on ss. MinnetonJca. 240 RADIO TELEGRAPHY RADIOTELEGRAPHIC STATIONS 241 he establishes communication hy stating the name of the vessel, its approximate position, and course. Then follows the message which the ship desires to send, and the shore station would acknowledge receipt of same, and perhaps repeat it for safety. If the message is for a private person, it would then be despatched from the coast station over the postal telegraph lines to its destination. In the same way, a message sent to a coast station for a particular ship is communicated to that ship. The operators in the coast stations are provided with charts showing the times of sailing from the various ports of all the vessels equipped with the wireless tele- graph apparatus belonging, say, to the Marconi Wireless Telegraph Company. By consulting this chart, the operator can tell what vessels are at any time within range of his station. He can then call up some particular vessel by signal and establish communica- tion with it if it happens to be within range. If, however, the desired vessel is not within range, but another vessel also equipped with wireless apparatus of the same system is within reach, the message can be despatched to the vessel within reach, which is then requested to forward it to the vessel lying beyond, within reach of itself but out of reach of the shore station. Thus, for instance, the operator at Crookhaven might have a message for the ss. Campania when four hundred miles out at sea, and not being able to reach the vessel at this distance, might despatch it to another vessel, say ss. Lucania, which is at a distance of a couple of hundred miles, and request the ss. Lucania to fling the message forward on to the Campania. In this manner a message may be made to jump over three or four ships, arriving at its destination after two or three retransmissions. This, of course, is on the assumption that the vessel is not provided with long-distance receiving apparatus, for if it is so provided it can be reached directly by long-distance stations, to be described presently. The difficulties with which radiotelegraphy of this kind has to contend arise chiefly from atmospheric electric discharges, from possible interferences or cross conversation, and very occasionally from deliberate interference. It has already been explained that a transmitting and receiving apparatus can be made syntonic by proper tuning of the circuits, and will therefore become receptive only of signals approximately agreeing in wave length with the frequency for which the stations are tuned. Thus, for instance, long-distance stations, to be presently described, operate generally with a long wave which does not in the least degree affect apparatus used on board ships and shore stations, operating with a wave length, say, of 1000 feet. On the other hand, receiving stations tuned for 1000 feet will be receptive for waves of that length, and also for R 242 RADIOTELEGRAPHY others not differing very greatly therefrom. The difference in wave length which can exist between a station's own proper wave length and that of the incident waves without stopping reception is a somewhat variable quantity, and depends essentially upon the damping or decrement of the sending station, and also the damp- ing or decrement of the receiving station, that is, upon the form of the resonance curve of the two stations taken together. This resonance curve, as already shown, is more peaked or sharper in proportion as the decrements of the sending or receiving station are smaller. Hence, if the sending station is emitting undamped waves, a more exact tuning or syntony will be required on the part of the receiving station in correspondence to obtain the best effect, than if the transmitter is sending out damped wave trains. That which is really inimical to the privacy of communication is the emission by various transmitting stations of powerful highly damped waves with large initial amplitude, whilst greater privacy is secured by the emission of feeble undamped or very slightly damped trains of waves. Hence the employment of strong damped waves should be as much as possible repressed in the general interest. In the same manner the receiving station will secure its own privacy far better when its receiving circuits are circuits only slightly damped and at the same time largely inductive, because then a slight difference between the frequency for which the re- ceiving circuit is tuned and the frequency of the incident waves will reduce the amplitude of the oscillations set up in the receiv- ing circuit by a very large amount, and hence enable the receiving operator to cut out those signals he does not desire to receive by exact tuning with the wave length of those he does desire to receive. Hence what may be called the unintentional interference and the picking up of messages it is not desired to receive is to some extent a matter of organisation, to a large extent a matter of apparatus, and also of personal skill on the part of the receiving operator, and if he possesses the requisite means and apparatus, he can render himself, so to speak, deaf to everything except the aethereal vibrations to which he desires to be sensitive. On the other hand, there are certain disturbances which are produced by atmospheric electric discharges which create so-called vagrant waves, and these, if sufficiently powerful, affect even syntonic apparatus. They are technically termed atmospheric X's. These atmospheric disturbances are particularly marked in tropical regions at certain times of the day and year. They exhibit them- selves in the case of the telegraphic recording receivers by making dots and dashes irregularly upon the tape, which are mixed up RA DIO TELEGRA PHIC S TA TIONS 243 with the dots and dashes belonging to the Morse signals being received, and render them more or less unintelligible. In the same way, when using the telephonic receivers, the atmospheric X's cause sounds of irregular duration and magnitude sufficient to blur the signals. These atmospheric disturbances have been particularly studied by naval officers in various navies. Many valuable observations have been put on record by Admiral Sir Henry Jackson, 1 and further reference to them is made in a later section of this chapter. Even when these disturbances are not sufficient to prevent communication altogether, which is seldom, they yet reduce FIG. 16. considerably the distance over which intelligible communication can be set up. Means have, however, been devised for stopping out the oscillations due to these irregular atmospheric disturbances to some extent. One of these, devised by Marconi, has received the name of an X-stopper. According to this invention, the antenna is not directly connected to the earth through the primary coil of an oscillation transformer, but a condenser and inductance coil, or a series of condensers and inductance coils, are included, as in Fig. 16. The operation then is as follows : If the antenna, a, 1 For details of these observations the reader may either consult the original Paper of Admiral Jackson (see Proc. Roy. Soc. Lond., vol. 70, p. 254), 1902, or a summary in the author's book, " The Principles of Electric Wave Telegraphy," p. 606 (Longmans & Co.). 244 RA DIO TELEGRA PHY is influenced by an irregular disturbance from a solitary wave or a short train of waves non-syntonic with the period for which the antenna is tuned, then these oscillations pass to earth, bub they do not set up oscillations in the chain of connected condensers. If, however, a syntonic train having passed a considerable number of waves falls upon the antenna, then the repeated oscillations excite sympathetic oscillations in the chain of condensers and inductances, and finally influence the receiving instrument connected with the circuit, r. In regard to radiotelegraphy, these atmospheric disturbances occupy the same position that earth currents and magnetic disturbances do towards telegraphy with wires. It is well known that at certain times the earth's magnetism is in a state of great disturbance called a magnetic storm. Periods of most frequent magnetic storms approximately coincide with the periods of most frequent sun spots and most frequent aurorse, and at the time of these magnetic storms electric currents circulate in the crust of the earth which sometimes interrupt ordinary telegraphic communication by wires altogether. 6. Long Distance Radiotelegraphy. The chief difference between radiotelegraphic stations established for short distance work, say, up to 200 miles or so, and those required for com- munication over 1000 miles or more, is in the power which must be expended for the longer distance. This requires certain modifications, chiefly in the transmitting apparatus, with the object of producing long electromagnetic waves of great amplitude. In power stations emitting damped waves the methods employed for production on a large scale are in principle the same as in small stations, but the apparatus has to be suitably modified. In place, therefore, of induction coils operated by batteries, we have to employ alternating current transformers to charge large condensers and alternating current dynamos to supply these transformers. Hence, a radiotelegraphic power station comprises, in the first place, a source of motive power, which may be a steam or oil engine. Steam is in every way preferable where water can be obtained, but in isolated places an oil engine is a necessity. The engine may be coupled directly to an alternator or may drive it by a belt, the alternator generally being of a type known as a revolving field fixed armature alternator, and having an electro- motive force at the terminals, say, of 2000 volts, and a frequency of 50. The current from this alternator is supplied to a battery of high tension transformers which may have their primary coils joined in parallel and their secondary coils joined in series. These trans- formers should be oil insulated. The transformers are connected RADIOTELEGRAPHIC STATIONS 245 through certain choking coils or inductances with a battery of condensers, which may be glass plate or tube condensers, which in turn are connected in series with one coil of an oscillation trans- former, the secondary circuit of which is inserted between an antenna and the earth, and with a spark gap which is across the terminals of the transformers. In some cases, the antenna is directly connected to the condenser circuit. Owing to the large quantity of electricity which passes at each discharge, it is desirable that this spark discharger should be of the revolving ball or disc type, such as that devised by the author, in which the spark balls are slowly revolved by means of electric motors, so as to continually expose fresh surfaces, or else the greatly improved high-speed-disc dischargers invented by Mr. Marconi. The currents are controlled and the signals made by short circuiting an inductance coil, which may be included in the primary circuit of the transformers. The oscillations transformer or inductance in circuit with the condenser is generally oil insulated. One of the most important elements of a power station of this description is the antenna. The creation of a long wave necessitates a corre- spondingly elaborate structure as an antenna. This antenna, as in the Marconi power stations at Poldhu in Cornwall, and Cape Breton, Nova Scotia, may be supported by wooden towers, 215 feet or more in height, and 25 or 30 feet square at the base (see Fig. 17). From this may be supported a multiple wire cone or fan- shaped antenna, or else the rising portion of a Marconi bent antenna, partly horizontal and partly vertical, the horizontal portion being sustained by other towers or masts (see Figs. 2, 3, and 4, above). The first long distance radiotelegraphic power station in the world was that undertaken in 1900, on the decision of Mr. Marconi and the Marconi Wireless Telegraph Company to make a serious attempt to achieve by his methods radio telegraphy across the Atlantic. A site was accordingly obtained in August, 1900, at Poldhu, near Mullion, in Cornwall, at a place far removed from large towns, and where work could be conducted with privacy. Mr. Marconi designed a multiple wire antenna of a cone shape to be supported by a ring of masts 200 feet in height, the station being placed in the centre. Work was commenced in October of the same year on this great enterprise. The author was entrusted with the duty of designing and arranging the machinery in this first power station for creating the powerful electrical oscillations necessary to excite oscillations in a large antenna. As the site selected for the station was on a cliff some way from water, an oil engine (25 h.-p.) was employed as a prime mover, and this drove by means of a belt, an alternator 246 RADIOTELEGRAPHY giving an alternating current of 50 periods and an E.M.F. of 2000 volts. This voltage was raised by transformers to 20,000, and employed to charge glass plate condensers. In the early experiments the author arranged a double trans- formation system, in which the secondary terminals of the transformer were connected to a pair of spark balls, and these were also connected by a condenser in series with the primary circuit of an oscillation transformer (see Fig. 18). The secondary circuit of this oscillation transformer was again connected with a pair of [By permission of Marconi's Wireless Telegraph Co., Ltd. FIG. 17. Cape Breton Long Distance Wireless Telegraph Station. spark balls, and these again with a second condenser and primary of an oscillation transformer, the secondary circuit of which was inserted between the antenna and the earth. Between the alter- nator, D, and the transformer, T 1 , was inserted a pair of choking coils, H 1 , H 2 , by the short circuiting of which the signals were created. This plant was completed in August or September, 1901. In course of time, however, these early arrangements were con- siderably modified, but they sufficed to create sufficiently powerful oscillations to produce electromagnetic waves which were detect- able across the Atlantic. After much experimenting over shorter RADIOTELEGRAPHIC STATIONS 247 distances, Mr. Marconi went, in December, 1901, to Newfoundland, taking with him balloons and kites as means for raising a wire to form a temporary receiving antenna and various forms of oscillation detector; and on December 14, 1901, he was able to announce that he had received signals which undoubtedly were the pre- concerted signals at that time being sent out from the Poldhu station. It was thus demonstrated that the electromagnetic waves made by no extravagant power expenditure in Cornwall could be detected at a distance of nearly 2000 miles, in spite of the considerable curvature of the sea surface in that distance. This achievement sufficed to give encouragement to Mr. Marconi and his supporters to proceed with the enterprise with the object of establishing regular commercial radiotelegraphic communication across the Atlantic. FIG. 18. Apparatus for Multiple Transformation of Oscillations (Fleming). Marconi returned to England in February, 1902, and at once made arrangements for the erection at Poldhu of a permanent structure for carrying a large antenna. This consisted of four wooden lattice towers, 215 feet in height, placed at the corners of a square 200 feet in side. These towers are strongly stayed by wire ropes (see Fig. 17). At first a conical antenna was employed, but later on, after Marconi had invented the bent antenna, other masts were erected to carry an antenna partly horizontal and partly vertical. New buildings for the generating plant were erected in the centre of the square, and more powerful machinery employed, and improvements introduced which experience had indicated. At the same time similar stations were erected at Cape Cod, in Massachusetts, U.S.A., and Cape Breton, in Nova Scotia. Whilst these improvements were in progress, Marconi returned 248 RADIOTELEGRAPHY to Canada, and on the way across conducted interesting experi- ments on board the Atlantic liner ss. Philadelphia. An insulated antenna wire, 60 metres high, was fixed to the ship's masts. Messages sent from Poldhu were received on board as the vessel went west, and printed down on the Morse tape. Keadable messages were obtained in this way up to 1551 miles from Cornwall, and communications or signals up to 2099 miles, by means of Marconi's printing telegraphic apparatus. In July, 1902, he conducted similar experiments on board the Italian warship Carlo Alberto, placed at his disposal for this purpose by the Italian Government, and on this occasion he employed his magnetic detector as a receiving instrument (see Chapter VI.), the invention of which he had patented some time previously. The first voyage of the Carlo Alberto was to the Baltic, and messages were received on board from Poldhu as far as Cape Skagen in Denmark and Cronstadt in Eussia. In August, 1902, the Carlo Alberto proceeded to the Mediter- ranean, and continued to receive wireless messages all the way ; and later on in the same year went across the Atlantic to Nova Scotia, receiving messages from Poldhu during the voyage, and whilst the ship was lying in Sydney Harbour. Towards the end of 1902 the stations being erected in Nova Scotia and at Cape' Cod were sufficiently advanced to enable a preliminary test to be undertaken, and on January 19, 1903, a wireless message was transmitted across the Atlantic from Welfleet, Cape Cod, Massachusetts, U.S.A., to Poldhu in Cornwall, from President Eoosevelt to King Edward VII., whilst at the same time other messages were sent from Cape Breton in Nova Scotia to Poldhu, and a large number of messages were trans- mitted both ways across the Atlantic in that and the following year. In 1904, Marconi established a system of long distance wireless telegraphy, communicating news to the principal Atlantic liners in the course of their voyages from Liverpool to New York, small newspapers being published on board containing news paragraphs transmitted by radiotelegraphy from the mainland on both sides, and by 1905 this system had become established as an indis- pensable means of communication with vessels en voyage across the Atlantic. In order to conduct this system of communication between ships without interfering with the transatlantic work, new stations were erected on both sides of the Atlantic by the Marconi Company one at Clifden in Ireland, and the other at Cape Breton in Nova Scotia in which many improvements were RADIOTELEGRAPHIC STATIONS 249 RA DIO TELEGRA PH Y RADIOTELEGRAPHIC STATIONS 251 r* e r * -a fl ! 3 S> J ^3 OQ t o * I a 252 RA DIO TELEGRA PHY introduced by Mr. Marconi, having for their object the increase in speed and certainty of sending and receiving. This station was completed in May, 1907, and by October, 1907, he was able to commence a regular system of press message radiotelegraphy across the Atlantic, communicating news for the American daily journals, and also exchanging public and private messages (see Figs. 19, 20, and 21). The Marconi Company have at the present -time (1908) four large long distance power stations for radiotelegraphy, which are equipped on the spark system, namely, those at Clifden in Ireland, Poldhu in Cornwall, Cape Cod in Massachusetts, U.S.A., and Glace Bay in Nova Scotia. The station at Poldhu has been chiefly used for long distance communication with Atlantic liners, a large number of which are equipped with long distance receiving apparatus, whilst the other stations are reserved for the transatlantic communication proper. As at one time statements were made that the working of these large stations would interfere with the ordinary ship to shore telegraphy, special demonstrations were arranged for the purpose of disproving these statements by Mr. Marconi; and these took place under the inspection of the author on March 18, 1903, in which a large number of messages were simul- taneously sent out from the Poldhu station, and also at the same time messages were sent out by short distance apparatus from a small station close to the power station. These were simultaneously received by Mr. Marconi at the Lizard station six miles away, and were also received by another observer at a station near Poole, two hundred miles away; and it was demonstrated that the powerful waves sent out by the long distance apparatus in no way whatever interfered with the messages being sent by the short distance apparatus in close contiguity to the power station. These tests were confirmed some months later by Admiralty officials, when similar demonstrations were made between Cornwall and Gibraltar. Eeturning, then, to the details of the above power stations, we may notice the highly effective form of discharger invented by Marconi for operating with large discharges. It has already been pointed out that in power stations on the spark system it is essential that the discharge across the spark balls should consist wholly of electricity which has been drawn from the condenser, and not be mixed up with a true alternating current arc discharge which is due to current coming RADIOTELEGRAPH 1C STATIONS 253 directly out of the transformer. Marconi, however, made use of the important fact that between metal surfaces in exceedingly rapid relative motion it is very difficult to produce a true electric arc, but, nevertheless, an oscillatory discharge from a condenser can pass between these surfaces. Hence he has devised forms of discharger made as follows : In one form there are a pair of metal discs, C, C (see Fig. 22), which are caused to rotate rapidly by electric motors or other means. Between these, and insulated from them, another disc, A, rotates at a high speed, with its plane at right angles to the other two. The terminals of a dynamo machine are connected with the terminals of a pair of condensers, K, in series, and FIG. 22. Marconi High-speed Disc Discharger. through inductances, L, L, with the discs C, C. The middle point of the condensers is connected to another condenser, E, and this through one coil of the jigger F with disc A. If a key in the dynamo circuit is closed, it charges the condensers, K, and then at a certain potential the condenser E discharges with oscilla- tions across one or other of the air gaps between the rapidly revolving wheels. The arc discharge which attempts to follow in the track of the oscillations is, however, prevented by the rotation of the discs from taking place. In another arrangement (see Fig. 23) the disc A has studs on its circumference at intervals placed transversely to its plane. The action of the discharger as shown in Fig. 22 is as follows : 254 RADIOTELEGRAPHY The dynamo charges the two large condensers, K, and then at some potential that plate of the small condenser E which is in con- nection with the disc A is charged by a spark passing between the central disc and one of the side discs. Suppose this action charges the said plate of the small condenser positively. This then strengthens the electric field between the middle disc and the other side disc, and a discharge happens on that side which reverses the sign of the charge on the small condenser plate attached to the middle disc. These reversals of charge rapidly succeed each other, each taking place with oscillations, and the FIG. 23. Marconi High-speed Disc Discharger. effect is to produce almost unintermittent oscillations in the coil F, in series with the small condenser E. This discharger, there- fore, affords a means of obtaining practically unintermittent oscillations from a continuous current dynamo machine, whilst the rapid rotation of the discs prevents the formation of a direct current arc which would otherwise stop the process. The principle of the discharger is therefore entirely different from that of the Duddell-Poulsen method of obtaining undamped oscillations from an electric arc. In the case of the discharger, as shown in Fig. 23, the discharges are intermittent, but succeed each other very rapidly. RADIOTELEGRAPHIC STATIONS 255 These Marconi dischargers work with great efficiency, and permit very rapid signalling to be conducted with them. The new Cape Breton or Glace Bay station in Nova Scotia has an antenna of 200 wires rising 220 feet vertically, and then extended 1000 feet horizontally, at a height of 180 feet above the ground. The antenna was designed for a wave length of 12,000 feet. The condenser used has a capacity of 1*8 microfarad, and the spark length used is generally 18 to 20 mm., equivalent to a charging voltage of nearly 50,000 volts. The bent antennae at Glace Bay, Nova Scotia, and at Clifden in Ireland, were placed with their free ends pointing directly away from each other. At the end of May, 1907, the Clifden station was completed. It was designed also for a wave length of 12,000 feet, and had a condenser of T16 microfarad, charged to 80,000 volts. This con- denser is an air condenser formed of sheets of metal hung up on insulators, thereby avoiding the dissipation of energy in condensers made with glass dielectric. The dischargers used as spark gaps in these stations are of the revolving disc type above described, and permit of signalling at a rate as high as hand sending can accomplish. Owing to the regularity of the discharges, the Morse dash is heard in the tele- phone at the other side as a clear musical note, and the operator can distinguish between it and the irregular sounds due to atmospheric discharges. These large stations at Cape Breton and Clifden began exchange of radiotelegraphic messages across the Alantic on October 17, 1907, and hundreds of thousands of words in Press and private messages have since been transmitted, whilst communication has never been interrupted by causes within the radiotelegraphic stations themselves, and only for seventeen hours by breakdowns on land lines. The long history of this great achievement was related by Mr. Marconi in a lecture at the Eoyal Institution of Great Britain, delivered on March 13, 1908, in which he recounted the various stages of the work and the steps by which success had finally been attained. The practicability of long distance radiotelegraphy having been thus demonstrated by Mr. Marconi, numerous rivals entered the field with projects for transatlantic telegraphy, and in the case of the stations erected on the spark system, the general arrangements first introduced in the Marconi stations have been closely followed, with the exception that the coupling of the nntenna in these other stations is generally a direct coupling instead of inductive. 2 5 6 RA DIO TELEGRAPH Y As an instance of a Continental long distance radiotelegraphic station, we may briefly describe that at Nauen in Germany. Within 40 kilometres of the north-west of Berlin, in the vicinity of the small town Nauen, the Telefunken Wireless Tele- graph Company of Germany has erected a powerful spark-telegraph station. The antenna is supported by an iron lattice tower 100 metres, or 325 feet in height (see Fig. 24). The station is situated in a plain extending for several miles round. The nature of the ground is very suitable for a radiotelegraphic station as it is moist, thus considerably facilitating a good earth connection. [Reproduced by permission from "The Electrical Review." FIG. 24. Wireless Telegraph Mast at Nauen, near Berlin. This, however, has added to the difficulty of making foundations for the tower. The iron lattice tower has a triangular section, with sides about 4 metres, or 13 feet in length. The three vertical sides of this tower are constructed in lengths of 8 metres screwed together, and are joined by means of diagonal stays to each other. At the bottom these beams converge on to a ball of cast steel placed in a socket in a foundation plate, and forming a hinge joint. This bed plate is insulated from the concrete foundation on which the tower rests. At a height of 96 metres there is a small platform for working and controlling the RADIOTELEGRAPHIC STATIONS 257 three pairs of pulleys disposed at the top and serving for the lifting up of the aerial. At a height of 75 metres three stays are attached to the tower, sustaining it in a vertical position. These stays consist of iron rods joined together by means of strong hinge joints and attached at the lower ends to anchors fastened at a distance of about 200 metres from the base of the mast. Both at the top and at the bottom, these rods are insulated from the tower and from the anchors. In view of the great electric tension the three upper insulators are immersed in oil. The lower fastenings of the three stays have for their anchorage blocks of brickwork. From the top of this tower extends a system of wires which forms an umbrella- shaped antenna com- posed of six parts, and arranged in such a way that the weights of two opposite sections are balanced against each other. This arrangement equilibrates the stresses due to the sections of the network upon the tower as they are lowered or raised in pairs. {Reproduced by permission from "The Electrical Review." FIG. 25. Umbrella Antenna at the Nauen Radiotelegraphic Station. On the other hand, arrangements have been made for lowering each of the six segments independently if necessary. The antenna is constructed of bronze wires, branching out and increasing in number from top to bottom (see Fig. 25). The whole surface of these wires is 60,000 square metres. Each separate segment of this umbrella- shaped network is fastened to the ground by means of hemp cables passed through a series of porcelain insulators and attached to anchors. All the conductors of the antenna are carried from the top of the tower to the station building, the conductors passing along the tower and entering the station building. These conductors are not separated frpm the tower by any insulators, the tower being thus a part of the antenna. The earth connection consists of a large number of branching iron wires, arranged star fashion, underground, and dividing up like the antenna itself. These earth wires extend over an area of about 126,000 square metres. They converge to a central point, and pass inside the station building. 8 258 RADIOTELEGRAPHY The station is a one-storied building occupying an area of 100 square metres. It is divided into an engine and a telegraphic room, and also contains dwelling apartments in the basement, and a condenser room on the first floor. The heating of the premises is effected with exhaust steam from the engine. The prime motor is a 25-h.p. steam engine working at 120 E.P.M. Coal and coke are used for fuel. Water is obtained directly by pumping from the soil. To economise fuel the steam pressure is reduced during non- working intervals, the [Reproduced by permission from " The Electrical Review." FIG. 26. Condenser and Transformer Boom at Nauen Station. reduced pressure being quite sufficient for the transmission of short telegrams. When a period of continued transmitting approaches, a few minutes are sufficient to get up a full pressure, so as to work at full load. From the flywheel of the engine a single-phase alternator is driven by means of a belt, and excited by a direct coupled exciter. This generator produces alternating current of 50 periods at 750 E.P.M. Conductors are carried from the generator to the telegraph- room to a switchboard arranged therein, on which are the controlling switches, fuses, and measur- ing instruments, as follows : A two-pole switch with fuse, an RADIOTELEGRAPHIC STATIONS 259 amperemeter, a voltmeter, a frequency meter, and a key relay. The alternating current produced by the generator supplies [Reproduced by permission from " The Electrical Review." FIG. 27. Eeceiving Apparatus and Desk in Nauen Station. four high-tension transformers connected to a condenser. The secondary terminals of the transformers are connected to a spark 260 RADIOTELEGRAPHY The condenser consists of 360 Leyden jars arranged in three series groups, which give a resultant capacity of 0'44 microfarad (see Fig. 26). The oscillation circuit comprises an inductance placed in series with the spark gap and with the condensers. This inductance is made of a silver-plated copper tube, and has terminals for connection with the condenser and spark gap, as well as for the direct connection with the aerial and the earth [Reproduced from " The Electrician " by permission of the Proprietors. FIG. 28. wire. The signalling is effected by short circuiting inductances in series with the primary coils of the transformers. The changing from transmitting to receiving is effected by a switch disconnecting the antenna and the earth wire from the transmitter inductance, and connecting them to the receiver, and cutting out simultaneously the connection to the alternator. Such an arrangement is absolutely necessary in order to protect the sensitive detectors and coherers from the powerful effect of the exciting circuit. All the receivers as well as the RADIOTELEGRAPHIC STATIONS 261 condensers and coils necessary for tuning are fitted to a desk in the telegraph-room (see Fig. 27). As an example of a station combining in one the spark and the arc methods, we may take that at Cullercoats, on the Nor- thumberland coast, about 8 miles from Newcastle, England, erected by the Amalgamated Kadiotelegraphic Company. The station itself is situated on a promontory running out to sea. [Reproduced from " Tlie Electrician" by permission of the Proprietors. FIG. 29. It comprises a small four-roomed one-storied building and a large umbrella antenna supported by a single wooden lattice tower (see Fig. 28). The mast is built up of baulks of timber 6 inches square, jointed in lengths. It is 220 feet high and 2 feet square at the base, and supported in a foundation of concrete and stayed by wire ropes cut up into lengths by insulators of creosoted wood. The antenna is constructed of bronze wires (see Fig. 29), which extend from the top of the mast and spread 262 RADIO TELEGRAPHY over a circle of 220 feet in diameter. It is made in two parts, each consisting of 12 wires, and stretched out into a wide semi- circle by guy ropes attached to anchors fastened to various rocks. The 24 wires of the complete antenna are connected at their lower ends to one wire which encircles the mast at a height of about 100 feet. The upper ends of the two halves are connected to two cables which come down into the station building. It will thus be seen that if the two cables are not connected together, the wires form a loop antenna, but if they are connected together they form a single antenna. The earth plate at the station consists of a large number of wires buried about 2 feet in the ground, radiating in all directions from a point near the foot of the mast. The station contains both spark and arc apparatus. In the case of the spark apparatus the power is supplied by an 8-h.p. motor driven directly from the town electric supply, and is coupled to a 5-kw. alternator, supplying 14 amperes at 400 volts, and at a frequency of 120. This alternating current is raised by a dry transformer to 50,000 volts. The sending operator can start and stop the alternator by a switch from the operating table. The primary current of the trans- former passes through a sending key, to interrupt it in accordance with the signals of the Morse alphabet. The high-tension alter- nating current is led to a third room, in which there is a large battery of Leyden jars and an associated inductance and spark gap, forming the oscillatory circuit. This inductance is directly connected to the antenna through a large switch which changes over the antenna from the transmitting to the receiving system when receiving. There are the usual tuning coils inserted in the circuit of the antenna. The reception is conducted by means of an electrolytic oscilla- tion detector and a telephone, the antenna being arranged as a loop antenna (see Fig. 30). With this apparatus communica- tion is carried on for about 400 miles between Cullercoats and Christiana. The station also contains a Poulsen arc apparatus. The arc generator consists of a metal box with marble ends, shown at the The Metrician 30 - RADIOTELEGRAPHIC STATIONS 263 left-hand bottom corner of Fig. 31. This box contains the copper and carbon electrodes, the cooling of the copper anode and of the arc box being effected by radiating flanges exposed to the air, and not by water circulation (see Fig. 9, 3). The striking of the arc is accomplished by lifting the copper electrode momentarily by a lever, and then allowing it to fall to an adjusted distance. The box is kept full of hydrogen supplied from a gas cylinder, or from a calcium hydride generator, by which hydrogen is generated by dropping calcium hydride into water. About two pounds of [Reproduced from " The Electrician" by permission of the Proprietors. FIG. 31. Poulsen Arc Apparatus in Cullercoats Station. hydride provide enough hydrogen for 60 hours' continuous work. In some cases coal gas is used instead of hydrogen. The carbon cathode is rotated by clockwork. The usual tele- graphic work is carried on with a single copper-carbon arc having a fine adjustment for arc length, the arc being formed in a strong magnetic field perpendicular to it. The windings of the electro- magnet are in series with the arc as well as with a variable resist- ance, and the arc is formed by a continuous current of 480 volts taking 10 or 12 amperes. The oscillation circuit is arranged as a shunt to the arc with a direct connection to the antenna, as shown 264 RADIO TELEGRA PHY in Fig. 32. It comprises an inductance coil of many turns and a condenser formed of zinc plates immersed in oil. The plates are separated by a distance of 3 mms., and the capacity is arranged in two sections, so that although a point on the induct- ance coil is put to earth, the terminals of the arc remain insulated. A variable condenser is connected in parallel with the fixed condenser to enable changes to be made in the emitted wave length, which is usually between 1200 and 1500 metres. A hot wire ammeter is inserted in the earth connection to show the current passing into the antenna. The signalling is effected by short circuiting a few turns of the inductance coil, and therefore altering the wave length of the emitted waves. The frequency employed is about 200,000, and the current into the antenna about 10 amperes. The receiving [From " The Electrician." FIG. 32. apparatus used with these undamped waves consists of an oscilla- tion transformer of which the two circuits are very loosely coupled, the primary being joined to the terminals of a condenser inserted in the antenna circuit, and the secondary connected to another large condenser, and also intermittently to a telephone shunted by a third condenser (see Fig. 33). The connection between the telephone and the condenser circuit is made by means of a ticker, or vibrating electromagnet, in which a very light rapidly moving hammer closes and opens the circuit. When the circuit is open, energy accumulates in the large condenser, and on closing it some of it passes into the condenser of the circuit, and on the opening of the contact again this condenser discharges through the telephone. The contact points of the ticker are made of crossed gold wires, and the vibrating mechanism is enclosed in a small sound-proof box. The observer, therefore, hears as sounds of longer or shorter duration in the telephone the RADIOTELEGRAPHIC STATIONS 265 more or less prolonged short-circuiting by the sending key of part of the inductance in the transmitting circuit. Owing to the loose coupling, the tuning is very sharp, and it is easy to perceive in this receiver the effect of altering about one-half per cent, in the capacity of the sending circuit The advantages claimed for the arc method of signalling are first its silence, and, secondly, the entire absence of sparking at the sending key, also the greater compactness of the apparatus and the lower voltages dealt with. For example, the maximum potentials which occur at the top of the antenna when the undamped waves are being used are probably not greater than two or three thousand volts, and the insulation required in the apparatus itself is only for voltages of the order of 1000 volts. It [From " The Electrician." FIG. 33. is also affirmed that atmospheric disturbances are much less felt when using the undamped wave apparatus than when using the damped waves. Furthermore, it is claimed that comparative tests of the arc and spark methods, carried out over ranges of about 900 miles, have shown that the undamped waves are less obstructed by mountainous country than are the damped waves of a spark transmitter of the same wave length when using about the same sending power. It remains to be seen, however, whether in actual working these differences will give rise to a marked advantage. The Amalgamated Kadiotelegraphic Company are also erecting another large power station in Ireland, near Knockroe farm, thirteen miles from Tralee, co. Kerry, with a view to correspon- dence with another station to be erected on the coast of Nova Scotia. The station consists of a power house 40 feet square, with 266 RADIOTELEGRAPH Y an accumulator house and two operating rooms at a distance from the power house. Three high masts, each about 3 feet square at the base and 360 feet high, are arranged round the operating house, and nine short masts, each 70 feet high, form a circle 2000 feet in diameter round the high masts. The high masts are built of square timber baulks framed together in a nearly horizontal position on a staging, and then raised into a vertical position by a [Reproduced from " The Electrician " by permission of the Proprietors. FIG. 34. The Poulsen Receiving Apparatus in Cullercoats Station. crab and jury mast. These high masts carry the upper insulated ends of an antenna of about 300 wires which descend in a cone to the lower masts, whence they are gathered into one conductor and led to the operating house. The power house contains two large and two small dynamos driven by a portable steam engine. The larger machines are intended to supply the continuous current at 500 volts for a Poulsen arc. RADIOTELEGRAPHIC STATIONS 267 The condenser included in the oscillation circuit consists of metal plates hanging in air of a total capacity of ^ - of a micro- farad, and capable of being arranged in various ways. With this apparatus it is estimated that a radiation of 10 to 15 kilowatts will be reached, employing a wave 3000 to 5000 metres in length. 7. The Effect of Atmospheric Conditions and Terrestrial Obstacles on Long Distance Radiotelegraphy. The first attempt to conduct radiotelegraphy extending over many hundreds of miles revealed the important influence which atmospheric conditions have upon such telegraphy, and especially the effect of sunlight upon it. In one of his voyages across the Atlantic, when receiving signals on board ss. Philadelphia, Marconi noticed that the signals were received by night when they could not be detected by day. He arranged a programme of experiments for sending signals of given strength on certain days from the power station at Poldhu, from 12 to 1 a.m., 6 to 7 a.m., 12 to 1 p.m., and 6 to 7 p.m. Greenwich mean time, every day for a week. He found that on board the Philadelphia he did not notice any apparent difference between the signals received by day and by night until the vessel had reached a distance of 500 miles from Poldhu. At about 700 miles, signals transmitted during the day began to weaken, while those received at night remained quite strong up to 1551 miles, and were even quite decipherable up to 2100 miles from Poldhu, being recorded with his coherer and printing receiver. He noticed, also, that the weakening occurred at the time when daylight first fell upon the transmitting antenna, and he inferred that the cause of it was the dissipating action of light upon negative charges of electricity. Another explanation is however possible. It is well known that the atmosphere, especially under the influence of sun- light, is in a state of ionisation, and it has been shown by Professor J. J. Thomson that these gaseous ions or point charges of negative and positive electricity are set in motion by a long electric wave travelling through space, and they therefore partially absorb the wave energy. By means of an apparatus devised by Ebert and Gerdien, it is possible to measure the conductivity of the atmo- sphere in any state and so to determine the number of ions or electrons in a cubic centimetre. Experiment shows that the numbers of positive and negative ions are considerable, but generally unequal. Using an Ebert apparatus, Boltzmann had found during an Atlantic voyage from Dover to New York that the number of electric ions present in the Atlantic atmosphere was 1150 positive and 800 negative per cubic centimetre. During 268 RADIOTELEGRAPHY a voyage from Montreal to Liverpool, A. S. Eve found from 600 to 1400 positive and from 500 to 1000 negative per cubic centimetre, the ratio of positive to negative varying from T04 to T83. These numbers do not greatly differ from those found over land areas of large dimensions, such as Germany or Canada. From data given by Strutt as to the amount of radium in sea- water and in various sedimentary and aqueous rocks, Eve draws the conclusion that this ionisation cannot be wholly accounted for by radioactivity of the sea or soil. Knowing, however, that ultra-violet light is a cause of ionisation, and, perhaps, the pene- tration of the upper layers of the atmosphere by cosmical matter carrying electric charges is another possible cause, we may no doubt assign to these agencies some share at least in the produc- tion of atmospheric ionisation. In any case, it is clear that the terrestrial atmosphere, when we are concerned with large volumes of it, and especially on that side of the earth which faces the sun, cannot be considered as equivalent to space occupied merely by free ether, or even air and ether. The presence of these ions or electrons in the atmosphere alters the dielectric constant of the space, and also exercises an absorp- tive action upon the energy of long electric waves. In other words, air exposed to sunshine, although it may be extremely transparent to light waves, acts as if it were a slightly turbid medium for long electric waves ; but the effect is not sensible up to distances of two or three hundred miles. The atmosphere, moreover, when in a state of ionisation exercises a certain selective absorption upon long electric waves, just as various transparent media exercise selective absorption upon light waves of certain wave lengths. A wave of long wave length and small amplitude is less obstructed than one of lesser wave length and larger ampli- tude. Accordingly, by the choice of a suitable wave length and amplitude, waves can be generated which are not much subject to absorption by the day lit atmosphere, and considerable progress has been made of late in a knowledge of the particular wave lengths to employ for long distance radiotelegraphy. Later observations by Marconi have revealed the curious fact that the increase in sending power required for transatlantic radiotelegraphy by this so-called " daylight effect " is greatest when the daylight or darkness extends only part of the way across the ocean, one station being in day and the other still in night. Another matter of great importance is the effect of obstacles, and especially mountains and earth curvature, upon long distance radiotelegraphic transmission. Even in the case of RADIOTELEGRAPHIC STATIONS 269 ordinary short distance work the effects produced by interposed hills and cliffs are quite marked. A large number of observations have been recorded on this matter by Admiral Sir Henry Jackson, and his results were communicated to the Eoyal Society of London, in 1902. The experiments were conducted between ships of the British Navy provided with apparatus on the Marconi system, and the observa- tions proved that the interposition of land, especially rocks of certain kind, greatly reduces the maximum signalling distance as compared with the distance for the same power for open sea. Summarising the results for soft rocks, hard limestone, and limestone containing a large proportion of iron ores, respectively, the percentage of maximum signalling distance through them, compared with the open sea distance, is as follows : Sand, sandstone, shale, etc. Hard limestone. Iron ores. Maximum distance 81 Minimum ,, 56 Mean 72 68 25 58 Less than 40 23 32 The results obtained show conclusively that hard rocks con- taining iron ores, interposed between the transmitting and receiving stations, especially when in the form of high cliffs, undoubtedly exercise a very marked effect in reducing the possible signalling distance for given types of ordinary short distance apparatus. Even when the cliffs do not extend to any height, there is evidence that the passage of the wave over land weakens its energy. This being, of course, merely a limited case of the more general fact that, for a given expenditure of energy, radiotelegraphy can be conducted over a greater distance over sea water than over dry land, the reasons for which have already been discussed. Another familiar cause of disturbance and limitation of range is found in atmospheric electrical states or conditions, which often constitute a most serious obstacle to effective transmission, even over moderate distances. They are much less frequently noticed in temperate than in sub-tropical regions. In the Mediterranean Sea they seem to be particularly prevalent, and are most persistent in summer and autumn. Owing to their sudden advent and sudden cessation, it is difficult to carry out systematic or prolonged experiments. As already mentioned, these atmospheric disturbances exhibit themselves by making 2?o RADIOTELEGRAPHY irregular automatic records upon the tape in the printing appa- ratus, or sounds in the telephone with a telephonic receiver. Admiral Jackson mentions that one of the most frequently recorded of these atmospheric markings is three dots, with a space between the first two, like the letters E I on the Morse Code, and this is very often due to distant lightning. Such disturbances are more frequent in summer and autumn than in winter and spring, and in the neighbourhood of high mountains than over the open sea, and with a falling barometer than with a rising one. In certain fine weather they reach their maximum between 8 and 10 p.m., and frequently last the whole night, with a minimum of disturbance between 1 a.m. and 1 p.m. When these atmospheric disturbances are present, the actual working distance of radiotelegraphy for any given apparatus may be reduced from 50 to 80 per cent, compared with that in perfectly clear weather. Thus, two ships, whose sea signalling distance may be 65 to 70 miles on a calm, bright day, may hardly be able to communicate 20 miles at a time when atmospheric disturbances are frequent. Admiral Jackson also notes the disturbing effect produced by a dry wind, such as the sirocco. On the other hand, it is a matter of common experience that in certain conditions of the atmosphere the expenditure of an extremely small amount of power in the production of electric waves of the right length is effective in creating signals on a syntonic receiver at immense distances, and these occasional feats are not due to special skill on the part of the operator, but to favourable atmospheric conditions. We have yet to mention another remarkable fact in connection with long distance radiotelegraphy, and that is the small degree to which the curvature of the earth seems to affect intercommuni- cation between stations which employ earth-connected antennae. It is well known that rays of light and sound are diffracted to some extent round obstacles, but the long Hertzian waves from an earthed antenna appear to pass round a one- eighth part of the circumference of the earth without extravagant diminution of amplitude other than that due to distance and atmospheric absorp- tion. We cannot compare experimentally the power required to send such waves 3000 miles over a flat surface with that required to send them 3000 miles round the earth ; but the increase, be it what it may, is not such as to make terrestrial radiotelegraphy on a large scale impossible. It has been suggested that the conduc- tivity of the upper layers of our atmosphere is sufficiently great to confine the waves to a spherical shell of the lower atmosphere, but no data at present available give support to the conclusion. RADIOTELEGRAPHIC STATIONS 271 We have already indicated that the possible cause of this advantageous transmission round the terrestrial sphere is due to the earth connection both of the transmitting and receiving antennae, whereby both these antennae and the earth are practically converted into a single oscillator. The full reasons, however, are yet to seek, and such explana- tions as mathematical analysis has afforded have not been generally accepted. Eadiotelegraphy, like every other branch of electro technics, has its unsolved problems, and it cannot yet be said that a complete elucidation of the nature of the propagation of electro- magnetic waves over and round the surface of our globe has been reached which is beyond dispute. CHAPTER VIII RADIOTELEGRAPHIC MEASUREMENTS 1. Radiotelegraphic Measurements. As soon as racliotelegraphy emerged from its earlier stages, the importance of quantitative measurements in connection with it became evident. Any branch of knowledge only becomes science in proportion as it becomes the subject of exact measurements. Hence when operating with radiotelegraph ic apparatus involving condensers, inductance coils, and other appliances, it is necessary to have the means of measuring and expressing precisely the magnitude of these quantities in terms of certain units, and to state our capacities, inductances, frequencies, wave lengths and decrements, in numerical values. The radiotelegraphist is therefore not equipped for his work unless he has a knowledge of the manner in which these measure- ments are made, and of the appliances necessary in making them. We shall consider in turn the principal important measure- ments which have thus to be made. 2. The Measurement of High Frequency Currents. In dealing with electrical oscillations we have to consider not merely the value of the current at any instant, but what is generally more important, its mean-square value, which is the only value capable of being directly measured. Supposing that a single train or oscillations is sent through a fine wire, it would expend part of the whole of its energy in heating the wire, and if a rapid succession of trains of oscillations were sent through the wire they would create heat in it at a certain rate, which would be balanced by the radiation of heat from the wire, or by the removal of heat in other ways, such as by the convection of the air. Since the emissivity of the wire increases with the temperature, a wire in any given surroundings subject to such oscillations will at last attain a steady temperature. Supposing, then, that we pass through the same wire a continuous current having a certain value, J, adjusted to produce the same temperature in the same wire under the same circumstances that is, to produce RADIOTELEGRAPHIC MEASUREMENTS 273 the same quantity of heat as the successive trains of oscillations per second. The current J is called the root-mean-square value of the oscillations, and the square of J is called the mean-square or integral value of the oscillations. We can therefore find the root-mean-square value by passing the oscillations, whether damped or undamped, through a hot wire ammeter, provided that this has a suitable form. As already explained, electrical oscillations concentrate at the surface of the wire, and hence the true resistance of the wire to these oscillations is higher than its resistance to steady currents, by an amount depending on its section. If, however, we employ a wire not larger than No. 36 S.W.G., or a number of such wires in parallel, the high frequency resistance will practically be identical with the steady or ordinary resistance. Accordingly, an ammeter for high frequency currents or oscillations must be constructed in the following manner. It must consist of one or more bare fine wires of any material, which may be copper, platinoid, or constantan, and these wires should not be too tightly twisted together, but somewhat spaced apart. They must be contained in an enclosure such that when heat is produced in the wire at a certain rate it will, after a short lime, come to a constant temperature. The root-mean-square value of any high frequency current is then measured by the value of the steady current which will bring the wire or wires to the same temperature. We may ascertain this temperature in one of three ways. (1) We may make use of the expansion of the wire itself, so that when expanded to the same extent it is taken to be at the same temperature. As this expansion is always small, it is best measured by the sag produced when the wire is held between two fixed terminals. For this purpose, the wire or wires are attached to two insulated terminals, A and B, which should be mounted on marble or slate. To the centre of the bunch of wires is attached one end of another long wire which has its second terminal also fastened to a fixed point, C. From the centre of this last wire another fine wire or thread is attached to the end of an index needle (see Fig. 1). If then the first wire AB expands and the ends being fixed it sags up or down, this sag causes the second wire DC to sag in the middle still more, and that motion is multiplied by the index needle. In this way a very small increase in length of the wire AB is readily made evident by a movement of the needle. The instrument is then calibrated by passing through it certain measured steady or continuous currents, and noting the position on the scale at which the needle finally stands. The scale may thus be graduated in amperes or miliiamperes. If then T 274 RADIOTELEGRAPHY we pass through the wire AB electric oscillations, either damped or UD damped, the wire will be heated and the needle will take a certain position on the scale, and the scale reading gives the root-mean-square value of these oscillations. (2) We may determine when the wire has the same tempera- ture for oscillations or for steady currents by enclosing it in the bulb of an air thermometer (see Fig. 2). German investigators have made a good deal of use of this form of hot wire ammeter, which they generally describe under the name of a Eeiss electrical thermometer. The instrument was in fact, however, invented by our countryman, Sir William Snow Harris, and described by him in 1827 in the Philosophical Transactions of the Royal Society. FIG. 1. In this instrument, a fine wire or number of fine wires are included in the bulb of an air thermometer, consisting of a U-tube attached to a bulb, the bend of the tube being filled with some liquid. When a current is sent through the wire, it heats the wire and the air, and forces the liquid up one leg of the U-tube until a stationary position is reached. The instrument may therefore be graduated by passing various known continuous currents through the wire. To eliminate errors due to changes of external temperature, it is necessary to make the instrument in the form of a differential thermometer with two bulbs, one of which is occupied by the wire and the other of which is closed. RADIOTELEGRAPHIC MEASUREMENTS 275 (3) Another method of determining when the wires are of the same temperature is by attaching to them a thermoelectric junction, and connecting the latter to any form of sensitive galvanometer or voltmeter. If we employ a voltmeter or galvano- meter, the scale divisions of which are of equal length and indicate equal increments of current through the galvanometer, then, using a theraio-j unction, say, of bismuth and iron pressed against the centre of a fine wire, it will be found on passing continuous currents through this wire that the scale readings of the galvanometer attached to the thermo-j unction are very nearly proportional to the square of the current passing through the fine wire. They are almost exactly proportional to the square of the current, if the wire and thermo-junction are enclosed in a vacuum. In any case, such an arrangement affords the means at once of determining the root-mean-square value of trains of oscillations sent through the fine wire, after the instrument has been calibrated by means of continuous currents of known strength. When we are operating with trains of damped oscillations which succeed each other at uniform intervals N per second, and if the oscillations them- selves have a frequency n and a decre- ment S, then it can be shown that there is a relation between the root-mean- square value J of the oscillations and the first or maximum value I, in accord- ance with the equation 7 _ /Mr FIG. 2. Hence from the root-mean-square value and the known number of oscillations per train and their frequency we can determine the maximum value. It is very astonishing to find the large values of the maximum currents that are reached during the trains of oscillations of an ordinary Leyden jar. Taking for instance a condenser having a capacity of J - microfarad, charged to a voltage of 12,000 volts, equivalent to a 3-millimetre spark, and discharged 50 times a second through a circuit having an inductance of 2000 cms., we find that the frequency of the oscillations is 2 % 25 x 10 G , and the maximum current reached during the first half oscillation 276 RADIOTELEGRAPHY is as much as 420 amperes, whilst the root-mean-square value of the discharge current would only be 1*5 amperes. 3. Measurement of Potential Difference. In measuring the potential difference of points on a circuit traversed by oscilla- tions, we may in the same manner desire to know either the maximum value reached by this potential difference at any moment or its root-mean-square value. The only method for obtaining the maximum value at any moment is by measuring the spark length of the spark which can be taken between balls of a known size connected to these two points respectively. We have in Chapter II. given a table showing the spark voltages of sparks of various lengths between balls 2 cms. in diameter. Hence, if two such balls are connected to the points in question, and adjusted to such length that sparks will just not pass, we can obtain from the table an approximate estimate of the maximum voltage between these balls. Nevertheless, the measurements can only be at the best approximate, because if many sparks are allowed to pass the ball surfaces become heated, which tends to promote discharge of a lower voltage, also the air becomes altered in constitution between the balls, acting in the same manner. Again, if daylight or the light from other sparks is allowed to fall upon the balls, the effect of the ultra-violet light contained in such luminous radiation also promotes discharge at a lower voltage. Neverthe- less, the spark method is almost the only method we possess for determining the maximum of potential difference reached during a train, or at any time between two points on a circuit. The root-mean-square value of the voltage is best measured by means of an electrostatic voltmeter, the capacity of which is as small as possible. A voltmeter of the quadrant type, having a needle or movable plate suspended by a quartz fibre, is sometimes used for this purpose, such as that devised by Dolezalek. The voltmeter method, however, must be used with caution, because the capacity of the voltmeter itself is sufficient to seriously disturb the condition of the circuit owing to the potential difference' which exists when the voltmeter is removed. Under some circumstances we can make use of such a volt- meter to measure the maximum voltage between two points, since there is a relation between the maximum value and root- mean-square value of the voltage, similar to that which exists in the case of the oscillatory current. In other words, if V is the maximum potential existing between two points of a circuit in which oscillations are taking place, having a frequency n, the trains succeeding each other at the rate of N per second, then if RADIOTELEGRAPHIC MEASUREMENTS 277 U is the root-mean-square value of the voltage, U and V are related together by the formula T..V 4. The Measurement of Capacity. One of the most frequently needed measurements in connection with this subject is the exact measurement of capacity. The unit of capacity in the electro- static system is the capacity of a conducting sphere 1 cm. in radius when placed in space at a considerable distance from all other conductors. Capacity is defined, as already mentioned, by the quantity of electricity necessary to charge the body to unit potential. Hence a sphere of 1 cm. radius is charged to a potential of one electrostatic unit (equal to 300 volts) by placing upon its surface one electrostatic unit of quantity, since all portions of the charge are then at a distance of 1 cm. from the centre, and create there a potential of one unit, and therefore raise the whole sphere to the same potential. The capacity of any other condenser can therefore be expressed in electrostatic units by stating its values in centimetres, meaning by that the radius in centimetres of a sphere, the capacity of which is equal to that of the condenser in question. The practical unit of capacity for most purposes is, however, the microfarad, or the millionth part of one farad. The microfarad is a capacity nearly equal to 900,000 electrostatic units. In other words, the microfarad is the capacity of a conducting sphere in free space, the radius of which is 9000 metres, or rather more than five and a half miles. The capacity of the whole earth in space, considered as a sphere, is only equal to 800 microfarads, and hence the unit of capacity called a farad, equal to a million microfarads, is far too large a unit for any capacities which have to be measured by terrestrial electricians. The farad is, however, the practical unit of capacity in consistent relation with the ampere, the ohm, the watt, and the joule, and therefore capacities in any other unit have to be reduced very often to capacities in farads, in substituting their numerical values in equations. A convenient practical unit for radiotelegraphic purposes is the micromicrofarad, that is, a millionth part of a microfarad, equal to nine-tenths of an electrostatic unit. Capacities will therefore be sometimes measured in electrostatic units or micro- microfarads, and at other times in microfarads, depending on their magnitude. The reader will find, in text-books on Physics, a large number of methods given for the comparison of capacities. We shall 278 RADIOTELEGRAPH Y consider only those that are well adapted for radiotelegraphic purposes. Two methods may be employed, one of which is a comparison method, which assumes the possession of another condenser of known- capacity, and the other is an absolute method, in which the capacity is determined with reference to absolute units of resist- ance and time, without reference to any other condenser. The comparison methods are generally employed where rather large capacities, something of the order of a microfarad, have to be measured. In every well equipped laboratory will be found a standard microfarad condenser, or half microfarad, generally constructed with mica as a dielectric interleaved between tinfoil sheets. Supposing, then, that we have another condenser, differing not much in capacity from one microfarad, or, at most, a moderate fraction or multiple of it, we may proceed in the following manner. If the standard condenser is charged by connecting its terminals with those of a voltaic cell or battery of cells, it takes up a certain quantity of electricity, Q, measured by the product of its capacity, C, and the voltage, V, to which it is charged. If, then, we connect this condenser to a galvanometer, a charge rushes out of the condenser through the galvanometer and causes the coil or needle to move through an angular deflection called the " throw." It can be shown that the sine of half the angle of the throw is proportional to the quantity of electricity that passes through the galvanometer, and for small deflections this may be taken as proportional to the displacement of the spot of light on the scale, if the galvanometer is a mirror galvanometer. If, then, two condensers of different capacity are charged to the same potential, and successively connected with the same galvanometer, the throws obtained will be nearly proportional to the capacities of these condensers. In order to eliminate error due to the want of precise proportionality of the throw to the quantity of electricity, we may proceed as follows : Connect the terminals of a voltaic cell or battery to a very high resistance, divided into sections, and connect the standard condenser across any fraction, E, of this resistance, and then connect the condenser immediately afterwards with the galvano- meter, and observe the throw. Perform the same experiment with the other condenser, and vary the resistance between the terminals of the condenser, so as to find by trial the value it must have in order that the throws may be the same in the two cases. Then calling the two resistance values R and R', and the two capacities C and C', the capacities are inversely proportional to the value of the resistances. RADIOTELEGRAPH1C MEASUREMENTS 279 The method is more conveniently carried out by arranging two capacities Ci, C 2 and two resistances, KI, K 2 , as in Fig. 3, with the battery and the galvanometer in opposite diagonals. A key is placed in the battery circuit, and the resistances, BI, R2> are altered until, on raising and lowering the key, the galvanometer gives no deflection. Under these circumstances, the capacities are inversely as the resistances, or GI JA.2 FIG. 3. This method is known as De Sauty's method for the comparison of condensers. One source of difficulty in connection with it is found in the unequal absorptions of the dielectrics of the two condensers, if these are made of different materials. When a condenser is charged for a certain time and then discharged, a certain proportion of the charge comes out instantly ; the re- mainder comes out more slowly, and is called the residual charge, or the absorbed charge. Different dielectrics exhibit this effect in different degrees ; hence, if one of the condensers has a dielectric, say, of paraffin paper, and the other of glass or mica, it is sometimes difficult to find any ratio of the resistances which entirely abolishes all movement of the galvano- meter needle when the battery key is raised or lowered. For a further discussion of these difficulties the reader may be referred to the author's " Handbook of the Electrical Laboratory and Testing Boom," vol. 2, chap. ii. The majority of capacities which we have to measure in radiotelegraphy are fractions of a microfarad, and for this purpose the most convenient method is an absolute method, in which the charge put into the condenser by a given voltage is repeatedly discharged through a galvanometer a known number of times per second, and measured as an electric current. If, for instance, one terminal of a battery, condenser and galvanometer are con- nected together and if the other terminal of the condenser is alternately connected to the terminal of the battery, and to that of the galvanometer, and if this process is repeated rapidly n times per RADIOTELEGRAPHIC MEASUREMENTS 281 second, then the galvanometer is traversed by a series of discharges, each of which conveys a quantity of electricity equal to CV micro- coulombs, where C is the capacity of the condenser in microfarads, and V the potential of the battery in volts. If these n discharges succeed each other rapidly, the effect on the galvanometer is equal to the passage through it of a current equal to CV?i microamperes, or y^g- amperes. This process of rapidly charging and discharging the condenser can be most conveniently effected by means of a rotating commutator, the structure of which is shown in Fig. 4. It consists of an electric motor of -H.P., with associated starting and controlling resistances which can be run off in constant voltage circuit of 100 or 200 volts, with a constant speed which may conveniently be about 1500 E.P.M. To the shaft of this motor is connected by a flexible coupling the commutator, which i B FIG. 5. consists of two' gun- metal discs, A, B, each having four pro- jecting lugs like a crown wheel, and between these is placed another wheel, I. These three wheels are keyed together upon a shaft, and insulated from one another, and they form a drum, the surface of which appears as in Fig. 5 on looking down upon it. Against this drum three brass wire brushes press, which are carried on insulating pillars, E. Two of these make a contact with the outer flanges of the crown wheels, and the middle one makes contact with the central portion. It will be seen, there- fore, that as the drum revolves, the centre brush alternately makes contact four times in each revolution, or a hundred times per second, first with the brush on the right hand, and then with the brush on the left hand. The time of contact is also accurately known, and there is no bouncing or uncertain contact, but a smooth, steady contact. The speed of the commutator is deter- mined by attaching to the shaft an endless screw intergeared with 282 RADIOTELEGRAPHY a toothed wheel, so that the wheel makes one revolution for every hundred revolutions of the shaft. At each revolution of the wheel a pin lifts a lever, which strikes a blow on a gong. By means of a stop watch the time of ten revolutions of the wheel, and therefore a thousand revolutions of the commutator, can be ascertained with an accuracy of less than one per cent. If, then, we connect one terminal of a condenser to the middle brush, whilst the two outside brushes are connected respectively to the terminals of a galvanometer and of a commutator, the other terminals of the galvanometer, commutator, and condenser being connected together, when the commutator is set rotating, it will cause a series of discharges to take place in the galvanometer, which will have all the effect of a steady current and create a deflection which remains constant. To determine the value of the steady current, which will give the same deflection, we may proceed as follows : Place a large resistance, E, in series with a galvanometer and a small shunt, S, across the terminals, and let G be the resistance of the galvanometer itself. Apply the same battery used in charging the condenser to the terminals of the galvanometer, and alter the value of the large resistance E to the shunt S until the galvano- meter gives the same deflection with a steady current passing through it as it did with the intermittent series of condenser discharges. Under these circumstances we have the following equation between the capacities, resistances, and number of discharges, viz : riVC = V S 10 6 : ~ p GS ' G + S from which we deduce c= s ' 106 - S) - The value of n, or the number of discharges per second, is accurately determined by counting the number of revolutions per second of the commutator, and multiplying them by four. The voltage of the battery used in this experiment must be determined by the magnitude of the capacity to be measured. If that capacity is a very small one, then it may be necessary to use a well insulated battery of a hundred small secondary cells, whereas, if the capacity is very large, one cell may suffice. The above method is very well adapted for determining the capacity of an antenna. In this case the antenna is connected to RADIOTELEGRAPHIC MEASUREMENTS 283 the middle brush, the two outer brushes pressing against the commutator shown in Fig. 4 being connected respectively to the galvanometer and to one terminal of a well insulated battery of a hundred small secondary cells. The other terminals of the galvanometer and battery must be connected to the earth. On rotating the commutator, the antenna is alternately charged by the battery and discharged through the galvanometer, and this capacity may be determined in microfarads as above described. In using this commutator for the measurement of very small capacities, such as an antenna, it is necessary to take into account the capacity of the commutator itself, which is not altogether negligible. This is done by taking two readings, one with the antenna connected to -the middle brush, and the other with the antenna removed, and the difference of the capacities determined in the two cases may be taken to be that of the antenna. By this means we can easily measure the capacity of an antenna con- sisting of a wire O'l inch in diameter and 100 feet long elevated into the air, a capacity which will generally be found to be approximately ^ oVfr ^ a microfarad. The above method is also a convenient one for measuring the capacity of Ley den jars, or similar condensers, which have a capacity of the order of O'Ol to 0*002 or thereabouts of a micro- farad. In the case of Leyden jars it should, however, be noted that the actual capacity when used with high frequency potentials is increased by the effect of the glow discharge from the edges of the tinfoil. 5. The Measurement of Inductance. Another important measure- ment is that of inductance, which can be measured either absolutely or by comparison with certain standards of inductance. Innumer- able methods have been described for the measurement of induc- tance ; but amongst those with which the author is acquainted, none is simpler or more easily carried out than that due to Professor A. Anderson. Anderson's method is applicable in the determination of inductances from a few millihenrys up to any multiple of a henry, and by the adoption of certain modifications suggested by the author is capable of measuring inductances as small as a few microhenrys. In its simplest form it is carried out as follows : The coil of which the inductance is desired, marked EL in the diagram in Fig 6, is connected with three other resistances, so as to form a Wheatstone's bridge arrangement. In the circuit of the battery B is placed a key k, and in the circuit of the galvano- meter is placed another key, k', and the galvanometer is joined in 2 8 4 RADIOTELEGRAPHY series with a resistance, r, and also a condenser, C, is connected between one terminal of the galvanometer and one angle of the bridge. Let E be the resistance of the coil of which the in- ductance is required reckoned in ohms, and L its inductance in henrys, and let PQS be the resistance of the other arms of the bridge. Then the first step in the process is to vary the values of P, Q and S, so that when the battery key is first put down, and afterwards the galvanometer key, there is no movement of the galvanometer coil or needle. The bridge is then said to be balanced by steady currents. The resistance S should be a plug- box resistance running from to 10,000 or 20,000 ohms, and the condenser C should preferably be a variable capacity. In obtain- ing a steady balance of the bridge, the resistance r should be cut out, and the condenser C removed. When the balance is obtained these instruments are re-inserted, and we then find that if the galvanometer key is first closed, and afterwards the battery key, the galvanometer coil or needle gives a throw or movement, which, however, can be entirely annulled by suitably varying the capacity C or the resistance r. When this is the case, the bridge is said to be balanced for throws. The advantage of the Anderson method is that in obtaining the balance for throws we do not have to upset the steady balance previously obtained. When the RADIOTELEGRAPHIC MEASUREMENTS 285 steady balance is obtained, there is a proportionality between the four resistances P, Q, R, S, expressed by the formula : P = R Q S and when the balance for throws is obtained, there is a relation between these resistances and the inductance L and capacity G and resistance r, as follows : L= 0{r(R + S) 4- QR} For the proof of this formula the student is referred to the author's " Electrical Laboratory Notes and Forms," Form No. 47, or to the " Handbook for the Electrical Laboratory Testing Room," vol. 2, chap, ii., p. 193. The student should note that in the above formula, if P, Q, R, S and r are measured in ohms, and C in microfarads, then the inductance L will be given in the above formula in microhenrys, and must be divided by 1000 to reduce it to millihenry s, or by 1,000,000 to reduce it to henry s. It will be found that when the inductance is very small, the throw of the galvanometer is small also, and there will be considerable difficulty in determining when it vanishes. We may, however, increase the sensibility of the method considerably by the following plan, due to the author : In the circuit of the battery is placed a vibrating electro- magnet or buzzer, Z (see Fig. 6), which continually interrupts the battery current, say two or three hundred times a second. We then insert in parallel with the galvanometer a telephone and a throw-over switch. During the operation of obtaining the steady balance, the buzzer is cut out and the galvanometer introduced into the bridge circuit. When this is done the telephone is substituted for the galvanometer, and the buzzer is re-inserted, and the observer then alters the resistance of r, or the capacity of the condenser, until no sound is heard in the telephone, or, at least, a minimum sound. To do this exactly, the buzzer must be enclosed in a sound-proof box placed in a cupboard, or at a distance from the observer, or otherwise he will mistake the sound due to the buzzer for a sound in the telephone. Those who have acute hearing can carry out the measurement with great accuracy, and in this way measure the inductance down to a few microhenrys. To check the accuracy of this measurement, it is desirable to possess certain standards of inductance of known value. These are best constructed by forming a square circuit of one single turn of round copper wire, say No. 16 S.W.G. Two 286 RA DIO TELEGRA PHY long wooden laths are arranged in the form of a cross, and wires strained round them so as to make a square circuit interrupted at the ends of one diagonal. From the formula L = ^ (2-3026 log, ^ - 2-853) already given in Chapter I., we can predetermine the inductance of this circuit for high frequency currents, knowing the length S of the side of the square, and the diameter d of the wire, both measured in centimetres. The above formula gives us the inductance in rnicrohenrys, each of which is equal to one thousand absolute C.G.S. units of inductance, which are reckoned in centimetres. Where great accuracy is not required, a larger standard of inductance can be constructed by employing the formula due to Russell for the inductance of a spiral circuit. If a bare No. 16 S.W.G. copper wire is wound upon a round rod of ebonite or hard wood in a helical form, either by cutting a screw groove in the wood or else winding a silk string in between the turns of the wire, and if the mean diameter of one turn of the helix is D centimetres, and I is the length of the spiral in centimetres, and N the number of turns per centimetre, then the inductance of the coil in micro- henry s is given by the formula The formula, however, neglects to take account of the fact that in the case of spiral wires traversed by high frequency electric oscillations the tendency of the current to concentrate on the inner sides of the coils tends to increase the inductance by a small amount. Anderson's method is also applicable for the measurement of mutual inductances. If two coils are placed with their axes in one line, they exert on each other a mutual inductance, and a current in one produces an induced current in the other. The mutual inductance or coefficient of a mutual inductance, M, is defined to be the numerical value of the total magnetic flux which is linked with both coils when a unit steady electric current flows in them. Hence, if we have two coils with their planes parallel to one another, and we pass through them a steady unidirectional current, each coil is self-linked with a certain magnetic flux due to the current in it, and it is also linked with the magnetic flux due to the current in the other coil. The inductance of the coil or its RADIOTELEGRAPHIC MEASUREMENTS 287 coefficient of self-inductance may be defined to be the flux which is self-linked with its own circuit when unit current flows through that coil. Hence, if we have two coils whose inductances are respectively L and N, and their mutual inductance M, and if they are joined in series, it is obvious that they may be so joined up that the current flows the same way round in both coils, or in the opposite direction of the two coils. In the first case the circuit consisting of two coils is self- linked with a total flux proportional to L + 2M -f N, and in the second case a flux proportional to L 2M + N. Accordingly, if we join up the two coils in the first manner, and measure by the Anderson method the inductance of the entire circuit and call it LI, and then join it up in the second manner, and measure its inductance again and call it L 2 , the difference of these inductances must be equal to four times the mutual inductance, or M - - 1 - ~- .?* 4 Hence, given two coils, we can measure separately their in- ductances, and also their mutual inductance in any position. The quotient of the mutual inductance by the square root of the product of the separate inductances of the two coils, that is, M the quantity /f ~ is an important quantity called the coefficient V-LiJN of coupling, and is denoted by the symbol k. The above methods, therefore, enable us to determine the coefficient of the coupling to the two coils which form the primary and secondary coils of a transformer of any kind. 6. Measurement of Frequency. A third essential measurement is the measurement of the frequency of the oscillations, whether damped or undamped, in an oscillating circuit comprising a condenser or capacity and an inductance. It has already been shown that the frequency of the oscillations in a circuit having a capacity of C microfarads and the inductance L microhenrys, is given by the formula : 10 6 m fds mbya or if the inductance is measured in absolute electromagnetic units or centimetres, then the frequency is given to the formula : 5-033 x 10 G n = -T== -= ~= V ^ mfds L< cms 288 RADIOTELEGRAPHY If, then, we have a circuit possessing capacity and inductance in which oscillations are taking place, we can determine the frequency of these oscillations by the principle of resonance. If we place near to but not very close to the circuit in question another circuit containing a known capacity and a known inductance which can be varied, and also have some means for determining when the oscillations induced in this second circuit are at their maximum value, we may cause the oscillations in the first circuit to induce others in the secondary or detecting circuit, and we can vary either the capacity or inductance, or both together, of this last circuit until the current in it has its maximum value. In this case the two circuits are said to be in resonance. It has already been shown that if the two circuits are closely coupled, oscillations of two frequencies are set up in these circuits, but if they are loosely coupled, the resonance curve is a curve with a single peak, and the current in the secondary or detecting circuit will have its maximum value when the capacity and inductance are so adjusted that the product of these two quantities for the secondary circuit is the same as the product of the two quantities for the first circuit, and then, knowing the capacity and inductance in the secondary or detecting circuit, we can determine the frequency n of the oscillations in the first circuit from the formula : _ 5-033 x 10 6 __ 5-033 x 10 6 X/C mfds L cm3 The most convenient method of making these measurements is by means of an instrument devised by the author, called a Cymometer. This consists of a condenser of variable capacity constructed of a tube of brass covered with ebonite, on the outside of which another concentric tube fits closely, but not too tightly as to prevent easy movement. If the tubes lie over one another, such a double brass tube with interposed tube of ebonite constitutes a tubular condenser, but if the outer tube is slid off the inner brass tube the capacity is reduced almost proportionately to the dis- placement of the outer tube. Again, if we have a wire wound in the form of a helix round an ebonite tube, the turns being close together but not touching, and if we have some form of clip which can be slid along the helix so as to make use of more or less of the spiral, we have a variable inductance. These two appliances are combined together in the cymometer in such a way as to form a complete oscillatory circuit ; the inner end of the tubular condenser (see Fig. 7) is connected to one end of the helix of wire by a copper bar, and the outer condenser tube RADIOTELEGRAPHIC MEASUREMENTS 289 is connected to the helix by an embracing clip, so that as the outer condenser tube is displaced from the inner tube to reduce the capacity, the effective inductance in the circuit due to the spiral is reduced in the same proportion. The helix and the tubular condenser, which may be formed of two or more tubes, are mounted on a board, and by means of a handle the condenser tube can be moved and the inductance and capacity simultaneously altered, and in the same proportion. If, then, we place the long copper bar connecting the helix and condenser near but not very close to any other circuit in which oscillations are taking place, we can tune the cymometer circuit to the other circuit by moving the handle so as to vary the inductance and capacity of the cymometer. We must then have some means of determining when the current in the FIG. 7. cymometer, or the potential difference of the tubes forming the condenser is a maximum. The author discovered that the most convenient way of doing this was by the use of a vacuum 'tube of the spectrum type, filled with Neon. Neon is a rare gas contained in the atmosphere, about 80,000th part by volume, and it is remarkable for its small dielectric strength and for the great brilliancy of the glow produced in it when placed in an alternating current field. If such a Neon tube is connected to the terminals of the tubular condenser, then when the capacity and inductance are altered and the oscillation in the cymometer circuit thereby increased up to a maximum, it is easy to determine the moment when this maximum takes place by the Neon tube beginning to glow, or glowing most brilliantly (see Fig. 8). u 290 RADIOTELEGRAPHY Another method of discovering when the current is a maximum in the cymometer circuit is by inserting in the circuit of the copper connecting bar a fine wire of high resistance about a centi- metre in length, having in contact with it a very sensi- tive therm o-j unction of bis- muth and iron. This thermo- j unction is connected to a sensitive galvanometer, pre- ferably a Paul single pivot low resistance galvanometer (see Fig. 9). If then by the movement of the handle of the cymometer it is gradually tuned with any adjacent circuit in which oscillations are taking place, the increase in the current up to a maxi- mum will be indicated by a gradually increasing deflec- tion of the galvanometer, and it is quite easy to determine that adjustment of the cymometer in which the current is a maximum. The cymometer has a graduated scale with a pointer moving over it, and the in- strument is calibrated by the manufacturer so as to show at a glance the fre- quency corresponding to any particular adjustment of the tubular condenser. The author has designed such instruments for reading fre- quencies from 50,000 up to 5,000,000, and the appearance of the complete instrument is as shown in Fig. 10. The cymometer may be employed for the measurement of small capacities ,and inductances in the following manner : Each instrument is, or can be, supplied with a standard RADIOTELEGRAPHIC MEASUREMENTS 291 inductance consisting of one or more turns of insulated wire arranged round a rectangular frame. These inductances vary from about 4000 centimetres, or four microhenrys, up to 75,000 centi- metres or 75 microhenrys, depending on the pattern of cymo- meter, in use. If then a certain small capacity, say, that of a Leyden jar, has to be determined, it is done in the following manner. The jar is placed upon a sheet of ebonite, and one coating is connected to one secondary spark ball of an induction coil, the other coating or terminal of the condenser being connected to one end of the above- mentioned standard inductance, whilst a second end of the standard in- ductance is connected to the other secondary spark ball (see Fig. 11). The spark gap, condenser, and in- ductance are all connected in series. The cymometer is then placed with its copper bar parallel, not very near to one side of the standard inductance. On working the coil, oscillations are set up in the circuit of the jar and inductance, and the handle of the cymometer is moved until the Neon tube glows most brightly. The scale reading of the cymometer then shows the oscillation constant of the cymo- meter in that position, that is to say, the value of the square root of the product of its capacity in microfarads, and its inductance in centimetres in its then position. The value of this quantity is called the oscillation con- stant, and is marked on the scale. It then follows that the oscillation con- stant for the circuit containing the unknown capacity must be the same. Hence, if we square the value of the oscillation constant and divide by the value of the standard inductance in centimetres, we have the value of the unknown capacity in microfarads. Thus, for example, suppose that the standard inductance is 5000 centi- metres, and that the maximum glow in the Neon tube occurs when the cymometer pointer indicates that the oscillation constant 292 RA DIOTELEGRA PHY RADIOTELEGRAPHIC MEASUREMENTS 293 is 10, then the square of 10 being 100, and the quotient of 100-7- 5000 being 5^, we know that the capacity of the con- denser in question must be ^ (7 of a microfarad. The rule therefore is as follows : Square the oscillation constant and divide by the value of the standard inductance in centimetres, and the resulting quotient is the capacity of the jar or condenser in fractions of a microfarad. In the same way the cymometer can be used with a standard condenser to determine the value of an unknown inductance, for if we determine as above described the capacity of a condenser by the aid of the cymometer, then join up this capacity with the unknown inductance and the spark gap, to form an oscillation circuit, putting in, if necessary, a yard of straight wire to lie parallel with the bar of the cymometer, and if we then determine B f r~i 4- 1 D c ^ | _ ,.,,r,.,.,.,.>Tj 1 -' *=== , L_ FIG. 11. the oscillation constant of this circuit, and find it to be 0, then the Q2 inductance in the circuit must be equal to -^, where C is the \j capacity of the condenser in microfarads, and this quotient gives the inductance in centimetres. In those cases where a small inductance is measured, it can be determined as the difference between two inductances, viz. by joining up with the condenser of known capacity a standard inductance of known value, and dividing the oscilla- tion constant as above, and then increasing the inductance of that oscillation circuit by adding in the small unknown induct- ance, and making a redetermination of the oscillation con- stant. Supposing, for instance, that the oscillation constant in the first instance is Oi, and in the second 2 , and that the standard inductance was, say, 5000 cms., and the value of 294 RADIOTELEGRAPHY the unknown and small inductance L, then we have the following equations : - = 5000 from which we can at once determine the value of L. A large variety of such tests can be made with a cymometer, provided it is remembered that the oscillation constant marked on the scale of the cymometer is the square root of the product of its capacity reckoned in microfarads and its inductance in centi- metres, corresponding to the position in which the handle of the cymometer is then placed. 7. Measurement of Wave Lengths. Another important measurement is the measurement of the wave length of the waves emitted by an antenna, or of the wave lengths being received by an antenna. In all cases of wave motion there is a relation between the velocity of the wave V, its frequency n, and wave length X, expressed by the equation The velocity of the electromagnetic waves being 300 million metres per second, or very nearly one thousand million feet per second, it follows that the wave length is at once obtained by dividing this last number by the frequency. Hence, if the frequency of the oscillations in an antenna is determined, we have the wave length of the emitted waves. If, then, we can determine the oscillation constant of the antenna, or of the circuit which is radiating, we have at once the following rules : Wave length in feet = 195 '56 X oscillation constant. Wave length in metres = 59'6 X oscillation constant. Frequency in million ths of a second is 5 -033 -f- oscillation constant. In order to determine the wave length, therefore, all that is necessary is to place the bar of the cymometer parallel, but not very near to a portion of the lower part of the antenna. For this purpose, a yard or two of the antenna may be laid in a horizontal position, if necessary. On exciting the oscillations in the antenna and moving the handle of the cymometer, we shall find a position RADIOTELEGRAPHIC MEASUREMENTS 295 in which the Neon tube glows most brightly, provided the cymometer used has a range including the wave length in question. In the case of inductively coupled antennae, it will be found, of course, that there are two wave lengths being emitted, and there- fore two positions in which the Neon tube has a maximum glow. In so using the cymometer, it is desirable to put the bar as far as possible from the antenna after having roughly dis- covered the approximate wave length, and then to take a fresh reading, so adjusting the distance of the cymometer bar from the antenna, that the Neon tube only just glows on passing through to a posi- tion of resonance. With a little care it is possible to determine the wave lengths of the order of 1000 or 1500 feet within 10 feet. Four types of cymo- meters are now made, one suitable for measuring wave lengths from about 30 metres to 1000 metres, another up to 1500 metres, a third up to 2000 metres, and a fourth up to 3000 metres, the lowest possible reading being generally about one twelfth part of the highest possible reading for any one instrument, but with special cases, greater ranges can be obtained. Hence a suitable cymometer must be employed for the particular measurements being made, the oscillation constants of the above four types ranging from about 1 to 12, 2 to 25, 3 to 37, and 4 to 50. For certain measurements in which greater accuracy of reading is required, it is better to employ, instead of the Neon tube, a thermoelectric detector, which is placed in the circuit of the cymometer. The circuit of the cymometer is cut in two places, FIG. 12. 296 RADIOTELEGRAPHY or the simple double copper bend with which it is usually pro- vided for completing the circuit can be replaced by a special double bend (see Fig. 12) containing two cuts in it, in one of which is inserted a fine resistance wire, and in the other a fine resistance wire having a thermoelectric junction in contact with it. These resistances and thermoelectric junction are contained in two ebonite boxes attached to the special bend, and a length of flexible connecting wire is provided, by which the thermoelectric junction is connected to a special low resistance single pivot sensitive galvanometer, that usually employed being made by Paul. There are short circuiting straps for cutting out the thermo- electric junction resistance, or the plain resistance. If we insert in the circuit only the resistance with the thermo-junction, and then employ the cymometer as above described, in proximity to any circuit in which oscillations are taking place, we shall find that as the handle is moved, tuning the cymometer more and more in circuit with the circuit under test, the ammeter exhibits a gradually increasing deflection, and at a certain position of the cymometer a maximum deflection is reached. In this position, therefore, the cymometer circuit is traversed by the maximum current, and, therefore, is in resonance with the circuit under test. In another form of wave-meter or cymometer devised by Donitz, the condenser consists of a number of fixed plates interspaced between a number of movable plates attached to a shaft, by the rotation of which the plates can be more or less sandwiched in between each other, and the capacity of the condenser formed by these plates therefore varied within certain limits (see Fig. 13). This condenser is connected in series with certain coils of wire having a known inductance, and in addition a hot wire ammeter or electric thermometer, consisting of a wire enclosed in one bulb of an air thermometer, is employed as an indicating instrument to show when the current in the cymometer circuit is a maximum. The instrument has a scale to show the wave lengths or frequencies corresponding to any possible position of the rotating axis of the condenser ; that is to say, of the capacity included in the. wave- meter circuit. There is, however, a great advantage in employing an instrument like the author's cymometer, in which the capacity and the inductance of the instrument are varied simultaneously and in the same proportion, as then the divisions on the scale indicating the oscillation constant in various positions are equally spaced. It is also necessary to be able to measure the length of the waves which are incident on a receiving antenna. This may be done as follows : A single turn of wire is included in the receiving antenna RADIOTELEGRAPH 1C MEASUREMENTS 297 which is in inductive coupling with a standard inductance of known value. This inductance is in circuit with an oscillation detector, say, of the electrolytic type, and is shunted by a con- denser of variable capacity. The capacity of the condenser is then varied until the coupled circuit is in resonance, and the standard inductance gradually moved away from the antenna coil, the condenser capacity being also varied to keep the tuning right. It will be found possible to put the standard inductance so far from the antenna coil that the slightest variation of the condenser FIG. 13. The Donitz Wave Meter. capacity either way causes the sound in the telephone .to dis- appear. The capacity in microfarads is then noted, and also the value of the standard inductance in centimetres, and the wave length in metres obtained from the formula X = 59 ifds *-* cms 8. Measurement of Damping and Logarithmic Decrements. In connection with the production of electric oscillations by the 298 RADIOTELEGRAPHY spark method, a frequently needed measurement is that of the decrement of the oscillations. When damped oscillations exist in a circuit, they decay in amplitude according to the law that the ratio of any oscillation to the next preceding it is constant, and this constant ratio is called the damping of the oscillations, and the Napierian logarithm of the ratio of one oscillation to the pre- ceding one, is called the logarithmic decrement, or shortly, the decrement. If we assume, as we may do, that the oscillations in a train are practically exhausted when the last oscillation is not more than one per cent, of the initial one, then, as already shown in Chapter I., the number of complete oscillations, M, in a train is given by the rule M _ 4-605 + g S The quantity 8 is the logarithm of the ratio of two successive oscillations in the same directions to one another, or 2'303 times the ordinary logarithm to the base 10 of the same ratio. As far as regards a mere qualitative determination of the damping, that is a proof that the oscillations with which we are dealing are damped or undamped, probably the best method of doing it is by the vacuum tube oscillograph. This consists of a glass tube having two straight aluminium wire electrodes, in line with each other and nearly touching. The tube is exhausted, but only to a low vacuum, about equal to 1 mm. of mercury. When the elec- trodes of the tube are connected to the terminals of a condenser which is in electrical oscillation a glow appears on the electrodes, the length of that glow being proportional to the potential differ- ence. If the tube is viewed in a rapidly rotating mirror, making, say, 500 revolutions a second, or at least a very high number, the alternately glowing electrodes produce separated images, and we can see at once whether the discharge is damped or undamped. The photographs in Figs. 14 and 15 are oscillographs, taken in this manner by Dr. Dieselhorst, of damped and undamped oscillations. In each of these the upper part is the positive and lower negative, and it will be seen that in the undamped oscillations, which are produced by a Poulsen arc, the oscillations have a greater ampli- tude on one side than on the other. This is because there is on the condenser a steady potential difference, which is that creating the current through the arc, and the oscillatory potential difference is superimposed on this, hence creating a non-symmetrical oscillo- gram. As the quality of a train of electric waves and its effect upon a receiver greatly depends upon the damping, the determination of RADIOTELEGRAPHIC MEASUREMENTS 299 the quantity 8 is an important measurement. It is easily effected by means of the cymometer, as follows : The cymometer circuit is, as explained, cut in two places, or else with an extra double bend of copper having two gaps in it, which takes the place of the ordinary simple double bend em- ployed when using the instrument merely with the Neon tube. In these gaps can be inserted the two ebonite boxes which contain fine resistance wires of constantan, against one of which is pressed a fine bismuth or iron thermo-junction. To complete the output, we require a single pivot Paul galvanometer, having a resistance of about 4 or 5 ohms, and reading from zero up to 400 micro- amperes. Also it is requisite to have means of calibrating this instrument. Assuming the possession of a resistance box, a secondary cell, and a small direct-reading milliamperemeter, the FIG. 14. Photograph of Damped FIG. 15. Photograph of Undamped Oscillations. Oscillations. first step is to calibrate the thermoelectric junction, so as to ascer- tain from the readings of the Paul galvanometer connected to the terminals of the thermo-junction the mean-square value of the current passing through the fine wire. For this purpose we con- nect the fine constantan wire in series with the cell, the milli- amperemeter, and the variable resistance, and pass various currents through it, say from 1 to 100 milliamperes. The deflection of the Paul galvanometer connected to the thermo-junction is noted at the same time. Squaring the values of the continuous currents sent through the fine wire, we then plot a curve, the abscissae in which represent the scale readings of the Paul galvano- meter, and the ordinates the square of the value of the current through the fine wire producing this deflection. Hence, if we subsequently pass oscillations through the fine wire, the reading of the Paul galvanometer enables us to determine at once the 300 RADIOTELEGRAPHY mean-square value of these oscillations. It will generally be found that the curve connecting the squares of the currents passing through the hot wire with the deflection of the Paul galvanometer is practically a straight line. When this calibra- tion is completed the fine wire with a thermo-j unction in contact with it is placed in the circuit of the cymometer, and the bar of the cymometer is placed near to the circuit in which oscillations exist, the damping of which is required. In so doing, it is necessary to be careful not to bring the cymometer too near to the oscillation circuit under test at first, or else the oscillations set up in it may be so strong as to burn out the fine resistance wire in the ebonite box. We proceed then to take a series of observations, as follows : Set the cymometer handle at one end of the scale, so as to include all the capacity, and move it forward step by step, noting the read- ing of the oscillation constant for each stage, and at the same time the reading of the Paul galvanometer in connection with the thermo-j unction. It will be found that on approaching the position of resonance the galvanometer reading will increase very rapidly to a maximum, and it may be necessary to make two or three rough trials, first adjusting the distance of the cymometer from the circuit under test until this maximum current in the cymo- meter makes a deflection of the galvanometer just within the range of the scale of the latter. We can then make a more careful experiment, plotting out a curve, the abscissae of which are the oscillation constants, as read on the cymometer for each position of the handle, or the frequency n corresponding thereto, and the ordinates are the mean-square values of the currents in the cymometer circuit, as obtained from the readings of the Paul galvanometer and its curve of calibration. The curve so obtained is called a resonance curve, and it will be found to run up into a single peak very rapidly, unless oscillations of two frequencies occur in the circuit under test, in which case there will be two peaks (see Fig. 16). Let a 2 be the 'mean-square value of the current in the cymometer, in any position of the handle corresponding to which the natural period of frequency of the cymometer is %, and let A 2 denote the mean-square value of the maximum current in the cymometer when it is tuned to resonance with the circuit under test, and let % denote the corresponding frequency as read on the cymometer. Then the oscillation circuit under test has a certain decrement Si, and the cymometer itself has a certain decrement 2 . It has been shown by V. Bjerknes and P. Drude that the RADIOTELEGRAPHIC MEASUREMENTS 301 following relation holds good between the decrements of the two circuits and the frequencies n\ and n 2 , viz. : provided that HI and n 2 do not differ from one another by more than, say, five per cent. Since the frequency is connected with the oscillation constant FIG. 16. A Kesonance Curve. by the formula n = we can also write the above formula of Drude and Bjerknes in the following form: Si + 8 2 = 31416 2 - Oi / a * Plotting out the resonance curve as above described, it is best to take the mean-square value of the maximum curve as unity, and to correct the other currents in the corresponding ratio ; and the same way for the frequencies, viz. the resonance frequency 302 RADIOTELEGRAPHY and any other frequency. If, then, we put x for (l - j and y for , we can write the above formula for the sum of the .A. decrements finally in the form & + & = 3-1416 x Since the resonance curve is not quite symmetrical with respect to its maximum ordinate, it is best to determine from the resonance curve the values of the frequency n^ lying on either side of the maximum current, which correspond to any given value of the cymometer current, and to take the mean of these values as the value to be put into the above formula. It will be seen, then, that from such a resonance curve we can determine the sum of the decrements of the circuit under test, and that of the cymometer. This last has, however, been increased by the resistance of the fine wire inserted in its circuit, by means of which we determine the sum of the decrements. We have therefore to eliminate the latter quantity as follows : It has been shown by Bjerknes and Drude that, if a secondary circuit has extra resistance of known value inserted in it so as to increase its decrement by a known amount 8 2 ', that the maximum current A (R.M.S. value) in the secondary circuit is altered to A', then the following equation holds good or if we put X for Si -f S 2 an( l X' for Si + S 2 + S 2 ', we mav it in the form X'S 2 ' But therefore where To determine Si we have therefore to take two resonance curves, one as above described, and another in which the circuit RADIOTELEGRAPHIC MEASUREMENTS 303 of the cymometer has its decrement increased by a known amount, by the insertion of a second fine wire resistance in the gap provided for it. The details of the measurements will perhaps best be under- stood by going through the calculations in a particular case. A certain oscillation circuit was set up, and by means of the cymometer a pair of resonance curves drawn, one without 'and one with an added small resistance in the cymometer circuit. These curves were as shown in Fig. 16. From these curves measurements were made giving us the E.M.S. values of the currents a, A, a', A', and at the same time of the quantities x = I and y = -r A number of values of y were taken off fii\_ .A. the curve corresponding to various values of x not exceeding 0'05, and tabulated as under, and the value of Si + 2 calculated by the formula above given. a I = (l-^) = X \ n\l X = Si + 8 2 0-95 0-0120 0-115 0-90 0-0165 0-112 0-85 0-0205 0-104 0-80 0-0255 0-107 0-75 0-0293 0-105 0-70 0-0335 0-103 The mean value of X is then 0108. In the same manner, after increasing the resistance of the cymometer circuit, a second set of values was obtained as follows : a' A= y (i-5t). \ ni ' X' = Si + 8-2 + 5 2 ' 0-95 0-0125 0-120 0-90 0-0210 0-138 0-85 0-0255 0-130 0-80 0-0300 0-125 0-75 0-0345 0-124 0-70 0-0385 0-119 Hence the mean value of X' is 0*126. 304 RADIOTELEGRAPHY Accordingly we have from the curves and formulae above ?7 = 2 ' 34 A / S x + S 2 = 0-108 & 4- 82 + &' - 0-126 Hence S 2 ' - 0*018 S 2 = 0-017 1 = 0-091 The greater part of the decrement S 2 is due to the resistance of the fine wire thermo-j unction, and apart from this the decrement of the cymometer in itself is only 0'005. In this case the oscil- lation circuit being tested comprised a condenser or Leyden jar and an inductance of 5000 cms. and a spark gap of 2 or 3 mm. in length. The high frequency resistance of the inductance was calculated from the dimensions of the wire and found to be 0'23 ohm. As this circuit was a nearly closed circuit, the decrement was all due to resistance, partly of the metallic wire E and partly of the spark r, and this can be shown to be equal to 4wiLSi, where L is the inductance of the circuit and n\ the frequency corresponding to resonance. Hence, if E and r are measured in ohms and L in centimetres, we have But E = 0-23, L = 5000, ^ = 0'091, and m = 0'95 x 10 6 . Hence r = 1'23 ohms. Also from the formula M = ^ we can show that each 01 train of oscillations comprised about 50 semi-oscillations, or 25 periods. Accordingly, the measurement of the decrement gives us all information about the nature of the oscillations taking place and the resistance of the spark. If we had been testing the decrement of a radiotelegraphic antenna, we should have found a much larger decrement than 0-091, because then there would have been radiation to increase the damping, and therefore the decrement. It will be seen, therefore, that by the use of the cymometer and the necessary adjuncts to it, we are enabled to obtain all the required information concerning the oscillations in the antenna of a radiotelegraphic transmitter employing the spark method of RADIOTELEGRAPHIC MEASUREMENTS 305 producing damped oscillations. When operating as above upon an antenna which is inductively coupled to the condenser circuit, the resonance curves will be found to be curves with double humps, as in Fig. 16, Chapter I. ; and if these humps are not too close to one another, we may apply the above process to each hump separately, and obtain the decrement of each of the two co-existing oscillations in the antenna. In making these measurements, the cymometer must of course stand on a table, and a certain length of the antenna must be bent round so as to be parallel with, but not too near, the bar of the cymometer. It will also be found necessary that the outer tube of the condenser should be connected to the earth by means of a terminal provided for that purpose. 9. Measurement of High Frequency Resistance. It has been already explained in Chapter I. that the resistance of a wire of high frequency currents to electric oscillations may be greater than its resistance to ordinary or steady currents by an amount depending on the size and material of the wire. Since the wires generally used for conveying oscillations are round copper wires, we can, in general, by the help of the formulae given in Chapter I., predetermine the resistance of such circuits for oscillations of known frequency, provided that the circuit consists only of a single wire having a slight curvature. If, however, the wire is in the form of a helix, or otherwise closely coiled, there is no way of determining the high frequency resistance except by enclosing the circuit in the bulb of an air thermometer, and determining the heat produced in it by oscillations of known mean-square value. If, however, we construct the circuit of fine silk- or cotton-covered wires twisted together, each one not having a larger diameter than No. 36 S.W.G-., then we prevent the change in distribution over the cross-section of the conductor, and so prevent this conductor from having a different resistance to electric oscillations from that which it has for continuous currents. It is therefore very impor- tant that the coils of all oscillation transformers and circuits used in radiotelegraphy should not be formed of solid metal wires, or stranded cables with individual wires of large diameter, but should be formed of stranded cables constructed of very fine insulated copper wires. In this manner we may generally arrange to avoid having to consider the increase in resistance of our metallic conductors to electric oscillations, and from the known or steady current resistance we can calculate that part of the decrement which is due to the resistance of the wire, since it can be shown that in all cases the decrement per half -period due to resistance is equal to the quotient of the resistance of the circuit x 306 RADIO TELEGRAPHY divided by four times the produce of the frequency of the oscilla- tions and the inductance, provided that the inductance and the resistance are both measured in consistent units, that is to say, if the resistance is measured in ohms the inductance must be measured in henrys. On the other hand, when we are concerned with oscillatory circuits in which we have a spark gap, part of this resistance is due to the resistance of the electric spark itself, and this is very variable, depending upon the quantity of electricity conveyed by the spark, the spark length, and also the number of sparks per second. Many measurements have been made of the resistance of electric sparks, but some of these are useless to the radiotele- graphist, because they are concerned only with the resistance of single sparks. Two methods have been adopted for measuring the spark resistance, which lead to different results. In one of these we measure the resistance of sparks of various lengths conveying always the same quantity of electricity, and by the other method we measure the resistance of sparks of various lengths conveying different quantities of electricity. The first method has been employed by Slaby, the author, and others, and it consists in forming an oscillatory circuit with two spark gaps in it, one of which is variable in length. The circuit includes a hot wire ammeter for measuring the mean-square value of the oscillations, and the experiment consists in substituting for one of the spark gaps a conductive resistance of variable amount, adjusting this 1 latter until the mean-square value of the current in the circuit is the same both for the conductive resistance and when the place is taken by a spark gap of known length. Gene- rally speaking, this method, however, has only an academic interest, because in most circuits as used in radiotelegraphy, the quantity of electricity passing will vary with the length of the spark. Thus, for instance, if we form an oscillatory circuit comprising a con- denser, inductance, and spark gap, charging the condenser by means of an induction coil or transformer, then the quantity of electricity put into the condenser depends upon the spark length, because this determines the maximum voltage ; hence, when the spark happens, and the oscillations take place, the quantity of electricity that passes through the spark gap at each oscillation is a function of the spark length. The only way in which the spark resistance in these cases can be measured is by means of the cymometer or equivalent process for determining the total decrement of the circuit. If the radiation is absent or very small, then the total decrement is made up of two parts, a part depending upon the high frequency resistance of the circuit and a part RADIOTELEGRAPHIC MEASUREMENTS 307 depending upon the resistance of the spark. If we call di the part of the decrement due to the high frequency resistance, and d% that part of the decrement due to the spark resistance, and if we call K the high frequency resistance of the metallic part of the circuit and r the resistance of the spark, both measured in ohms, then, provided there is no source of loss of energy in the condenser itself, and no loss of energy hy radiation, the sum of these two decrements is connected with the sum of these two resistances by the formula , (R + r)10 9 where L is the high frequency inductance of the circuit in centi- metres and n is the frequency of the oscillations. 2? 0-5 I 2 3 ^ S SPARK LENGTH IN MILLIMETRES. FIG. 17. Hence, if we can determine by the cymometer the frequency of the circuit, and determine by calculation the high frequency resistance and inductance of the metallic part of the circuit, and determine experimentally, as already described, the total decre- ment, we are able to calculate the spark resistance. In this manner it can be shown that the spark resistance gradually decreases with increasing length of spark, reaching after a time a nearly constant minimum value, and that it varies to some extent with the materials of which the spark balls are made and with the gap in which the spark takes place. For very large spark discharges this resistance will be only a small fraction of an ohm, but for short sparks two or three milli- metres in length, such as take place when Leyden jars are charged 308 RADIOTELEGRAPHY by an induction coil, as in short distance radiotelegraphic apparatus, the spark resistance may amount to several ohms. The curve shown in Fig. 17 embodies the results of observations by the author on the spark resistance of sparks of various lengths taken in the above manner ; the condenser in the circuit consisting of metallic plates immersed in oil, and the inductance a single rectangle of fine wire, the high frequency resistance and inductance of which were calculated from its dimensions. The capacity used was 0-00261 mfd., and the inductance 0'00636 millihenry. CHAPTER IX RADIOTELEPHONY 1. The Problem of Radiotelephony. Before the invention of the methods of radiotelephony described in this chapter, attempts had been made with some degree of success to transmit the sound of articulate speech over moderate distances without the aid of a connecting wire. In addition to a method depending upon the induction of currents and their conduction through the earth, another has been worked out based upon a peculiar property of selenium of varying its resistance under the action of light and of the continuous current electric arc of varying the intensity of its light when a periodic current is superposed upon the continuous one. We shall, however, here confine our attention to the details of the method employing electromagnetic waves which gives the greatest promise of ultimate utility, now generally called Eadio- telephony. Eadiotelephony consists, therefore, in the transmission to a distance of articulate speech through space without wires by means of electromagnetic waves, as distinguished from radio- telegraphy, which is the transmission of intelligence by means of arbitrary signs, whether audible or visible. As soon as radio- telegraphy, as conducted by the methods already described, had made a certain progress, inventors had their minds naturally turned to the problem of the transmission of articulate speech by the same means. It very soon, however, became clear that the attainment of any practical success was bound up with the invention of a transmitter for producing undamped electric radiation, and of a receiver which should be quantitative in action, that is to say, not merely set in operation by oscillations, but produce an effect proportional to the amplitude of the waves incident on the receiving antenna. The oscillation detector to be used in connection with radio- telephony must therefore be of such a character that it is capable of varying the current through a telephonic receiver in exact 3io RADIOTELEGRAPHY correspondence with the variations of air pressure due to the speaking voice taking place in proximity to the particular tele- phonic transmitter employed at the sending station. In electric telephony conducted with wires, the apparatus usually employed consists of a transmitter of the microphone type and a receiver of the magnetic or Bell type. For instance, in the simplest form of short distance transmitter and receiver the microphone transmitter consists of a metal diaphragm which is set in vibration by the variations of pressure taking place in proximity to the mouth of a speaker uttering near it articulate words. Behind the diaphragm is some arrangement by which a variable or imperfect contact between carbon surfaces is altered by pressure. In the ordinary type of granular carbon microphone, the movements of the diaphragm are made to press together more or less small fragments of graphitic carbon contained in a shallow chamber, and so alter the electric conductivity of the mass. This FIG. 1. variable carbon resistance, M, is placed in series with a few voltaic cells, B, and the primary circuit of a small induction coil, T. One end of the secondary circuit of the induction coil is connected to the earth or to one of the line wires, L, and the other to the line wire or duplicate line wire, if a complete metallic circuit is employed (see Fig. 1). At the receiving end the currents in the line pass through the magnetising coils, E, of a magneto -telephone consisting of a permanent magnet having its polar extremities surrounded with these magnetising coils, the pole or poles being placed in proximity to a thin sheet iron diaphragm, d. The motions of the transmitting diaphragm, D, are therefore repeated by the receiver diaphragm, and every sound made near the microphone trans- mitter is reproduced by the diaphragm of the receiver. If, for instance, a musical sound is created near the diaphragm of the transmitter, there will be variations of air pressure which may be represented by the ordinates of a periodic curve. In the case of a perfectly pure musical sound, this curve approximates in form to RADIOTELEPHONY 311 a sine curve, but for any such sound as a vowel sound, the form of the curve will be complicated, although periodic if the vowel sound is continued. By various devices it is possible to delineate graphically the periodic curves corresponding to various musical or prolonged vowel sounds as in the diagrams in Fi^. 2, which are the results of experiments by Mr. W. Duddell, who has kindly given permission to reproduce them here. In the case of articulate sounds, the variations in air pressure are non-repetitive, but they can nevertheless be represented by the ordina.tes of a curve. Thus, for instance, in speaking to a gramophone or phono- graph, the voice creates variations of air pressure in front of the speaking diaphragm, and at the back of this diaphragm, or con- nected with it by a system of levers, is a delicate cutting tool which carves out upon the surface of the moving plastic cylinder or disc which forms the receiving surface a little channel or groove, the bottom of which is irregular, the depth of this groove correspond- ing from instant to instant to the variations of pressure produced against the diaphragm by the speech being made. If, therefore, a section could be made of this groove, and the outline of the bottom enlarged, it would present the appearance of a very irregular non-repetitive curve, each change in the ordinate of which, however, corresponds to a change in air pressure of the air in front of the diaphragm against which the speech is being uttered, and has therefore a signification as far as the ear is concerned. The problem of telephony is therefore to cause some other diaphragm at a distance to be moved from instant to instant in a similar manner to that of the diaphragm against which speech is being made. This receiving diaphragm will then reproduce at the distant end the same variations of air pressure as those which actuated the transmitting diaphragm, and a human ear placed in proximity to it will therefore hear the speech being made at the distant place. In telephony with wires, the movements of the transmitting diaphragm are made to translate themselves into corresponding variations in the strength of an electric current in the connecting wire by means of the variation in resistance which takes place when carbon surfaces are more or less pressed together, and the re -translation of this variable electric current into the movement of a receiving diaphragm is made to take place by means of the variations in the polar strength of a permanent magnet which takes place when an electric current of varying strength circulates round that pole. To achieve radiotelephony we remove the interconnecting wire Simple Form of 06 Sound as in Coo. f| * A A n * ** s * * f , s ', >i s * ', , *, Complex Form of oo Sound as in Coo. Vowel o as in Ho. Vowel e as in Me. K and First Part of e as in Key [By permission from " Tlie Proceedings of the Royal Institution of Great Britain, 1906. FIG. 2. Oscillograms or Wave Forms of various Sounds taken with the Oscillograph by Mr. W. Duddell, F.R.S. RADIOTELEPHONY 313 and substitute for it a train of electromagnetic waves passing through space. Hence the particular inventions required in order to accomplish the desired result, as regards the transmitter, are to devise a mechanism capable of emitting undamped electromagnetic waves and to vary the amplitude of these waves in accordance with the variations in the air pressure taking place against the trans- mitting diaphragm, and at the receiving end to cause these undamped waves of variable amplitude to actuate a mechanism which shall cause them to set in vibration the receiving diaphragm, so that its displacements correspond with the variations in the amplitude of the electromagnetic waves. We have therefore to consider (1st) the transmitting arrange- ments which have been invented, and (2nd) the receiving arrange- ments in connection with radiotelephony. 2. Transmitting Arrangements in Radiotelephony. High Frequency Alternator Method. It is generally agreed that for the perfect transmission of articulate speech by electromagnetic waves an essential condition is the possession of means for producing at the transmitting station, in the sending antenna, undamped or practically undamped electric oscillations. We have at the present time available two fairly effective means of doing this, viz. by the use of a high frequency alternator, or by the use of a carbon-metal continuous current arc in a hydrocarbon atmosphere, as already described in Chapter III. The alternator method has been particularly advocated and employed by K. A. Fessenden, and the arc method by V. Poulsen and E. Kuhmer. An essential condition of success in the transmission of articulate speech by electromagnetic waves is that there shall be no interruptions in the uniform flow of the undamped oscillations, at least not below such a frequency as is equal to the upper limit of the frequency of audible sounds. If regular vibrations are set up in the air, these are appreciated as sound by the normal ear if they lie in frequency between 40 and 20,000 per second. Human ears vary, however, a great deal in the value of the highest frequency which can be heard as a sound. As regards musical sounds, the highest frequency employed does not exceed 4000 or 5000. If intermittent trains of damped waves were employed, even if the frequency of the trains was as much as 4000 or 5000, they would affect the oscillation detector at the receiving station, and hence the telephone in connection with it, and produce in the latter a musical sound of high pitch which would drown out the variations of lesser frequency which constitute the articulate speech. Hence, if an alternator producing an alternating current having 3H RADIOTELEGRAPHY a frequency of even 10,000 were connected to a radiating antenna, it is probable that most persons would hear a sound in a telephone connected to an electrolytic oscillation detector in a receiving antenna. If, however, it had a frequency of 20,000, they would probably not hear any sound. We may say, therefore, that to be of practical use in radiotelephony a high frequency alternator should give a current having a frequency of not less than 20,000, and preferably higher. Fessenden has constructed such an alternator of the Mordey type, as already mentioned, with fixed armature and revolving field, having 360 poles or teeth. At a speed of 139 revolutions per second it gave a terminal E.M.F, of 65 volts, and an alternating current with a frequency of 50,000, the maximum output at this frequency being 300 watts. The attainment of a speed of 8000 E.P.M. in any revolving shaft and disc necessitates very perfect balancing, and if used on board ship special devices are necessary to obviate gyrostatic action, which would cause serious wrenching at the bearings as the ship pitches or rolls. As regards driving power, the best form of motor has been found to be the De Laval steam turbine, which, in small sizes, can be made to run up to 30,000 E.P.M., or 500 revolutions per second. Electric motors have been constructed, however, for the author, which for 1-H.P. size have run up to 6000 E.P.M., and for 5-H.P. size up to 4000 E.P.M., and these speeds can be multiplied by the employment of a thin and very flexible belt. The attainment of high speeds is facilitated by the use of ball or cylinder bearings and adequate lubrication, and, above all things, by perfect balancing. In the general design of the alternator, the choice lies between the inductor type of machine in which the only revolving part is an iron disc with teeth cut on the edge and the type of alternator with wound polar teeth. The inductor type of alternator has the advantage that it is easy to balance it, but, generally speaking, there is a considerable decrease in terminal voltage with increase in current taken out of the machine. Fessenden has pointed out that it is important that a high frequency alternator should give an electromotive force curve having a true sine form, as it is only then that resonance can be employed to multiply the electromotive force. For this reason it is necessary that there should be no iron in the armature, and as the polar teeth of the field magnet must be very narrow and close together, not more than two or three millimetres in width, to secure the necessary frequency with available speeds of rotation, it is essential to place all the teeth of one polarity on the same side to avoid magnetic leakage. This then results in the selection of the RADIO TELEPHON Y 315 Mordey type of alternator with fixed non-iron armature and revolving fields. It is not necessary to generate very high E.M.F., as we can always transform it up by an oscillation transformer or by resonance, but it is necessary to have high frequency. Unfortunately, we have not yet discovered really practical methods of transforming up frequency, as we can transform up voltage, so as to start with an ordinary low frequency current. There is no doubt that if a high frequency alternator of satisfactory mechanical design giving a frequency of 50,000 and upwards can be constructed as a commercial article, it will find a useful field in connection with radiotelephony. {Reproduced by permission from " The Electrician." FIG. 3. High Frequency Turbo-Alternator (Fessenden). R. A. Fessenden (see The Electrician, Vol. 61, p. 441, 1908) has constructed a high frequency alternator direct coupled to a De Leval turbine, giving a current at 225 volts, with a frequency of 75,000 and about 2'5 kw. output. The machine is of the double armature type, with 300 coils on each, and a field with 150 teeth. The two air-gaps are only ^ inch in length (see Fig. 3). The required steam pressure is 100 Ibs. on sq. inch. 3. Electric Arc Transmitters for Radiotelephony. Turning then to the other method already described for the production of undamped oscillations by means of the electric arc in a hydro- carbon atmosphere, we find that V. Poulsen has brought such arc oscillation generators to a considerable degree of perfection, and 316 RADIOTELEGRAPHY applied them also very successfully in the transmission of speech for very considerable distances over land and sea. As the method of producing undamped oscillations by the aid of a continuous current arc formed between a rotating carbon cathode and a cooled copper anode in hydrocarbon gas has already been fully described in Chapters III. and VII. , we need only here consider the most recent modification of the apparatus. To get rid of the necessity for cooling the box containing the electric arc with water, it is constructed with radiator flanges, so as to be air cooled, and it has been found that alcohol vapour is a convenient substitute for [Reproduced from "Electrical Engineering " of April 23, 1908, by permission of the Proprietors. FlG. 4. coal gas or hydrogen, so that methylated spirit is admitted drop by drop into the chamber in which the arc burns by a form of sight- feed lubricator controlled by a hand or electro-magnetic valve (see Fig. 4). By these improvements the arc method has been made independent of the supply of water and coal gas which was necessary in the early forms of apparatus. A small arc can also now be employed using 200 or 220 volts and a current of 1'5 or 2 amperes. It therefore becomes quite easy to operate it off any commercial lighting circuit furnishing continuous current at 220 volts, and its power consumption is not more than 300 or 400 RADIOTELEPHONY 317 watts. With this small power consumption troubles do not arise from the deposit of soot in the arc chamber. The current through the arc is regulated by adjusting the distance of the carbon and copper electrodes by a screw which is hand regulated, that is to say, an assistant keeps the current through the arc constant by watching an ammeter in series with it and adjusting the screw. The arc is formed in a powerful transverse magnetic field, and the carbon is made to rotate slowly by means of clockwork or a geared- down electric motor. Connected to the copper and carbon electrodes of the arc is an oscillation circuit, consisting of a variable inductance formed of a helix of wire, and a condenser of [Reproduced by permission of the Amalgamated RadiolelejrapUic Co. FIG. 5. variable capacity consisting of metal plates in oil. The form of condenser preferred is one in which there are a number of semi- circular plates fixed one above the other in a tall cylinder of highly, insulating oil (see Fig. 5). A number of other semi-circular plates sandwiched in between the first-earned set are affixed to a long metal rod, so that by turning this rod round the movable plates are brought more or less in between the fixed set of plates. The two sets of plates constitute the surfaces of the condenser, and the oil with which the jar is filled the dielectric. Hence, by simply turning round a milled head on the top of the cylinder, large variations of capacity can be created. The variable inductance RADIOTELEGRAPHY joined in series with this condenser is generally a single helix of bare wire with a sliding contact, by means of which more or less in- ductance can be introduced. In constructing the oscillation circuit to be used as a shunt across a continuous current arc, it seems important that the capacity should be kept small and the inductance large. That is to say, that a given oscillation constant for that circuit should be obtained not by the use of a small inductance and a large capacity, but by the use of a large inductance and a small capacity. If the capacity is reckoned in electrostatic units, and the inductance is reckoned in centimetres, then the capacity so reckoned may be to the inductance in the ratio, say, of 1 to 20. This oscillation circuit is coupled inductively to an antenna circuit, the coupling being close, and the antenna circuit syntonised with the condenser circuit by the introduc- tion of suitable inductance, so that oscilla- tions set up in the condenser circuit induce others of equal frequency and a maximum strength in the antenna circuit (see Fig. 6). Generally speaking, the current in the oscilla- tion circuit will be a current of 4 or 5 AAA II amperes, E.M.S. value, and the frequency -, M may be anything between 100,000 and a A_ _9Sefe million. "When the arc is set in operation we have undamped oscillations created in the condenser circuit and undamped waves emitted by the antenna. 4. Microphonic Control of Electric Oscilla- tions. In order to conduct radiotelephony, we have then to control the amplitude of the electric waves emitted by the antenna, so that this amplitude may vary in exactly the same manner and proportionately to the change of air pressure at any point near the mouth of the person uttering articulate speech. This is best done by the insertion of a speaking microphone in the condenser shunt circuit. Such a speaking microphone consists of a shallow metal chamber closed by a flexible metal diaphragm, which is insulated from the metal chamber, the space between the diaphragm and the solid back containing carbon granules which are more or less compressed by the vibrations of the diaphragm. Hence, when speech is made against a mouthpiece terminating on the diaphragm, the aerial vibration sets up similar vibrations in the diaphragm, and this movement, by compressing more or less the carbon granules, varies the resistance of the carbon included between the diaphragm U FIG. 6. RA DIO TELEPHON Y 319 and the solid back. As a single microphone transmitter cannot be operated satisfactorily with a large current, when it is desired to introduce it into a circuit in which a current of 4 or 5 amperes is flowing, it is necessary to employ a number of these microphone transmitters arranged in parallel, so that each microphone may not carry more than 1 ampere. The microphones can be arranged in a box or at the ends of branching pipes, so that they are all simultaneously affected by variations of air pressure due to speech made to a single mouthpiece (see Fig. 7). In the arrangements adopted by Poulsen this microphone transmitter or variable resistance is inserted either in the condenser [Reproduced from " Electrical Engineering " of April 23, 1908, by permission of the Proprietors. FIG. 7. shunt circuit of the arc or else in a tertiary circuit closely adjacent thereto (see Fig. 8). When speech is uttered against the microphones it varies the resistance of this microphone circuit, and therefore alters the resistance of the condenser circuit slightly, and therefore also affects the current in the sending antenna. Words spoken to the mouthpiece, therefore, produce an effect upon the amplitude of the emitted electric waves, and these are, so to speak, moulded into speech form, that is to say, made to vary as the ordinates of a wave curve representing the changes of air pressure taking place in the mouthpiece of the transmitter. In some cases the microphone resistance may be inserted in 3 20 RA DIOTELEGRAPHY the circuit of the electric arc and operate directly upon the con- tinuous current affecting the arc. In this case, a variation of the condenser current and also of the amplitude of the wave radiated from the antenna takes place in the same manner as the variations in the arc current produced by the changes in resistance of the microphone under the action of the articulate sounds. Or again, the microphone may be inserted as a shunt to the secondary circuit of the oscillation transformer connecting the antenna to the condenser circuit, so that the current into the antenna is more or less shunted to earth (see Fig. 6). Finally, the microphone may be inserted in the earth connection of the antenna so as to vary the current flowing into the antenna itself, and therefore the intensity of the radiated waves. In any case, it should be inserted at a node of potential in the oscillatory circuit. [Reproduced from Electrical Engineering " of April 23, 1908, by permission of tlie Proprietors. FIG. 8. Scheme of Circuits of Transmitter and Receiver for Radiotelephony. Another plan that has been suggested is to employ a condenser telephone consisting of two plates of metal near together, one of which constitutes the diaphragm against which speech is made, so that by the vibrations of this diaphragm under the operation of the voice the plates are more or less approximated, and their electro- static capacity varied. If this condenser transmitter is joined in parallel with the main condenser in the oscillation circuit, speech made against it, by altering its capacity, more or less throws the condenser circuit out of tune with the antenna circuit, and there- fore varies the intensity of the emitted waves. By any of these methods the train of undamped waves emitted by the antenna may be moulded into the form of speech, and these waves are then retranslated back into sound by the arrangements employed in the receiving circuit. RA DIO TELEPHON Y 321 The picture in Fig. 9 shows the complete arrangement for em- ploying a Poulsen arc in a radiotelephouic transmitter apparatus. Fig. 7 shows the multiple microphone arrangement, the mouthpiece against which speech is made branching out into a number of parallel pipes, at the end of each of which is a carbon microphone made as described, such a transmitter being suitable for radio- telephony over 250 miles. By these methods Poulsen has succeeded in transmitting articulate speech from Berlin to Copenhagen, a distance of 460 kilometres, or 290 miles. When a high frequency alternator is used as the source of the [Reproduced from " The Electrician" by permission of the Proprietors. FIG. 9. General View of the Poulsen Transmitting Apparatus for Badio telephony . undamped oscillations in the sending antenna, the oscillations created by the alternator are transferred to the antenna through an oscillation transformer. The microphone may then be placed in the antenna circuit, and as near the earth as possible, or it may be placed in a tertiary circuit wound on the oscillation transformer itself. 5. Other Arrangements employed as Transmitters in Radio- telephony. To avoid placing the arc in a strong magnetic field and enclosing it in an atmosphere of hydrogen or hydrocarbon, experi- ments have been made with a number of electric arcs in series. It has already been explained, in Chapter III., that the character- istic curve of a continuous current arc is steeper for small currents Y 322 RADIOTELEGRAPHY than for larger ones. If, then, a number of electric arcs taking a small current are joined in series, and a condenser circuit shunted over the whole series, it is possible to arrange these arcs to take a small current and obtain from them high frequency oscillations. This has been carried out as follows : A copper tube has a curved concave bottom fixed into it, and this is filled with water to keep it cool. This forms the positive terminal of the arc. The negative terminal is formed by a solid carbon rod, and the arc is struck in the cavity formed by the recessed end of the copper tube. Six such arcs are joined in series and a condenser and inductance shunted over the whole series. It is easy to contrive a mechanism by which all the arcs shall be struck at once and controlled together. The arcs are arranged to take a very small current. The oscillatory current set up in the condenser circuit is made to act inductively upon a syntonic antenna, and a microphone or number of microphones are used in parallel to regulate the emitted waves into speech form. This microphone may be placed in the antenna circuit or as a shunt to it. Fig. 10 shows the complete installation of a multiple arc apparatus for radiotelephony as employed by the German Wireless Telegraph Company (Gesellschaft fur drahtlose Telegraphic), in which electric arcs in series are used as a generator for undamped oscillations when shunted by a condenser and inductance. Another form of transmitting arrangement has been described in a patent specification by Fessenden. In it a high tension continuous current dynamo or else a number of such machines joined in series generate a continuous voltage. This current passes through a regulating resistance, and also across a spark gap formed between a metal point carried on the back of a telephone diaphragm and a rapidly revolving disc carried on the shaft of a motor. The spark gap is also shunted by an oscillatory circuit consisting of a capacity and an inductance, and an antenna syntonised to it is connected to the circuit. The precise operation of the apparatus is not described, but it may be assumed that the arrangement operates somewhat in the manner of Elihu Thomson's continuous current arc apparatus (see Chapter III.), and that a series of discharges passes across the spark gap, and that articulate sounds spoken to the diaphragm, by varying the spark length, vary also the frequency of these discharges, and therefore the amplitude of the radiated waves. No arrangement, however, can possibly give good results as a transmitter in radiotelephony in which the waves radiated are not RADIOTELEPHONY 323 [By permission of the " Gesellschaft fur Draktlose Telegraphic," Berlin. FIG. 10. Multiple Arc Transmitting Apparatus for Radiotelephony. 324 RA DIO TELEGRA PHY either perfectly continuous or undamped, or else consist of wave trains succeeding each other at a rate greater than 20,000 or 30,000 per second. Many attempts have been made to utilise for the purposes of radiotelephony the intermittent discharges of a condenser by greatly increasing the rapidity of the successive discharges. Thus Q. Majorana had produced condenser discharges succeeding each other at the rate of 10,000 a second, each discharge consisting of a train of oscillations. This, however, can only be achieved by the use of a very short spark gap and a high impressed voltage, and employing a large inductance in series with the source of electromotive force. Unless, however, the trains of oscillations succeed each other at a rate greater than 20,000 per second, the intermittances are heard in the receiving telephone as a continuous sound. 6. Receiving Arrangements in Radiotelephony. Assuming that a transmitting station is sending out undamped waves which are being moulded into speech form by means of a microphone con- troller, as already described, these waves may be absorbed by a properly syntonised receiving antenna tuned to the wave length employed, and by suitable arrangements can be made to affect a telephone, so as to translate back the ' oscillations, of constant frequency and amplitude induced in the receiving antenna into audible and articulate sounds. For this purpose it is necessary to employ in the receiving circuit an oscillation detector which is quantitative that is, not merely affected by oscillations, but affected to some extent proportionately to their amplitude. Thus, for example, a coherer or an oscillation detector of the imperfect contact type will be no use, because it is only affected by a certain alternating voltage, and is at once affected almost to the full degree when that voltage reaches a certain limit. Several forms of oscillation detector already described are very suitable for radio- telephonic reception, such as the Fessenden electrolytic detector, the author's glow lamp or ionised gas detector, the thermoelectric detectors, and the crystal rectifiers. Thus, for instance, if a receiving circuit is constructed by inductively coupling the receiving antenna to another oscillation circuit comprising a condenser, an inductance properly syntonised to the antenna circuit, and also includes an electrolytic detector coupled as already described, to a telephone and a local cell, the oscillations passing through the electrolytic detector will not merely alter its apparent electrical resistance, but alter it in some sense proportionately to their intensity, and hence, if undamped waves are falling upon the antenna of constant wave length but varying amplitude, a variation in the RADIOTELEPHONY . 325 apparent resistance of the electrolytic detector will take place, which follows and imitates the variation of wave amplitude. Accordingly, the cunent sent through the telephone by the local cell varies in the same manner, and the telephone diaphragm therefore emits a sound which corresponds with the amplitude of incident electromagnetic waves, and therefore reproduces speech being made against the diaphiagm of the transmitting microphone. Fessenden makes use of a form of telephone receiver he calls a " heterodyne " receiver. It consists of a pair of coils of wire, one of which is wound round an iron wire core, and the other is attached to a mica diaphragm held near the core. The last-named coil is traversed by the current in the receiving antenna, and the first by a local current of the same frequency as that of the trans- mitter. There is therefore a mechanical force between the two coils which varies with every variation of the current in the receiving antenna. The diaphragm therefore reproduces the sounds which are made against the diaphragm of the microphone in the transmitting circuit. We have, therefore, in the combined radiotelephonic transmitter and receiver, a wonderful transformation of energy. The varia- tions of air pressure made against the speaking diaphragm produce similar variations in the resistance of the microphone ; this again varies in the same manner the intensity of the electric oscillations set up in the oscillation circuit connected with the arc, and also the oscillations in the antenna. Thus electromagnetic waves are emitted, the amplitude of which is changing in the same manner. A portion of the energy of these waves is then translated back by the receiving antenna into oscillations, the amplitude of which varies also in the same manner as that of the incident waves, and these, acting on the particular detector coupled to the telephone, reproduce movements of the receiving telephone diaphragm which imitate those made by the diaphragm of the transmitting microphone. Although this operation is complicated, yet, nevertheless, it has been so far perfected that articulate speech can now be trans- mitted over a hundred miles by these means. In fact, radio- telephony, or telephoning without wires, seems to have certain undoubted advantages over telephony conducted with wires. It is well known that the reason why telephonic speech cannot be transmitted more than a certain moderate distance through sub- marine cables, is because of the distortion in the wave form which takes place owing to the combined action of the capacity, in- ductance, resistance, and leakage of the cable. The reason for this is that electrical vibrations of different frequency travel 326 RADIOTELEGRAPHY through such a cable with different velocities. Hence, when a complex vibration is impressed upon the cable by means of an ordinary telephone transmitter, the complicated wave form which represents, as already explained, any spoken word, can be resolved into the sum of a number of simple or sinoidal or harmonic vibrations of different frequency. These vibrations travel through the cable at unequal rates, and hence beyond a certain distance the integral wave form is distorted beyond recognition by the ear. This distortion may be compared with that of bad handwriting. In the case of ordinary written words, a single letter is hardly ever perfectly formed ; but if the departure from perfect writing does not exceed a certain limit, our experience enables us to guess pretty quickly what the word really signifies. In the same way, when listening through a telephone, if the distortion of sound does not exceed a certain limit, the ear is able to guess the meaning of the word, but beyond a certain point it is unrecognis- able. Kadiotelephony, therefore, seems marked out specially for the transmission of articulate speech over sea, owing to the greater difficulty of telephoning through submarine cables than through land wires. Moreover, there is, of course, no necessity that the transmitting and receiving station should remain fixed in position. Another form of oscillation detector suitable for telephonic reception is the oscillation valve or glow lamp detector, invented by the author, which has been made use of by Mr. Lee de Forest under the name of an audion. This glow lamp detector has already been fully described in Chapter VI. It consists of a carbon filament glow lamp having an insulated metal plate or cylinder in the bulb carried on an insulated terminal. When the carbon filament is incandescent by an insulated battery it emits negative ions, and a current of negative electricity can pass across from the filament to the insulated plate sealed into the lamp bulb, which varies with the voltage between the negative terminal of the lamp and the insulated plate. If, therefore, a telephone is connected between one end of the filament and the insulated plate, as described and shown in Figs. 20 and 21 of Chapter VI, the oscillations produced in the receiving antenna, varying from moment to moment in strength with the amplitude of the incident waves, will send through the telephone a continuous current which also varies in the same manner, and the telephone therefore repro- duces the articulate sounds made against the microphone of the transmitting station. Another form of detector much employed in radiotelephonic work is the crystal rectifier without local battery. In the RA DIO TEL E PHONY 327 arrangements employed by Poulsen, the antenna and the earth wires are connected to the two terminals of a condenser shown fixed up against the wall in Fig. 11, which forms with a variable inductance an oscillation circuit tuned to the period of the antenna. This inductance is loosely coupled to another oscillation circuit con- sisting of a condenser, inductance, telephone, and crystal rectifier. The coupling or mutual inductance of the two circuits is very weak, the primary and secondary helixes frequently being set a considerable distance apart, as shown in Fig. 11. The damped oscillations of constant frequency but varying amplitude taking place in the antenna induce other oscillations of the same period and similarly varying in the telephone and rectifier circuit, and [Reproduced from " The Electrician" by permission. FIG. 11. Poulsen Receiving Apparatus for Radiotelephony. the rectifier permits the current to pass through the telephone only in one direction. 7. Present State and Achievements of Badiotelephony. The transmission of articulate speech to a distance by means of electromagnetic waves without the aid of an interconnecting wire has made remarkable progress in the last few years, and has considerable possibilities of improvement. Poulsen has succeeded in transmitting phonograph music by this means from Berlin to Copenhagen, a distance of 460 km., or 290 miles. In Fig. 12 is shown the apparatus used by him for this purpose. On the right-hand side of the picture will be seen the transmitting arrangements, consisting of the copper-carbon 328 RA DIO TELEGRA PH Y arc in its box and the inductance and variable condenser of the oscillatory circuit, and on the left-hand side the receiving arrange- ments, including the receiving telephone. Distinct articulate speech is also said to have been transmitted by the same means 1 5 t_2 "Si I from Lyngby to Esbjerg, a distance of 270 km., or 170 miles* The receiver contained a thermoelectric oscillation detector. Eessenden has also described the arrangements and apparatus of the National Signalling Company of the United States, devised by him for radiotelephonic communication between Brant Eock and New York 350 km., or 200 miles. The generator is a RADIOTELEPHONY 329 1 kvv. steam turbine-driven alternator, giving alternating currents of a frequency of 81,700 to 100,000 at 150 volts (see Fig. 3). The resistance of the disc armature is 6 ohms, and the field exciting current 5 amperes. Using a transmitting antenna 200 feet high at New York, and the Atlantic Tower, 400 feet high, at Brant Rock, an energy expenditure of 200 watts in the antenna is required to cover the 200 miles. Successful demonstrations were also made in 1906 by the same inventor between Brant Eock, U.S.A., and Plymouth, Mass., a distance of 11 miles, in which speech was transmitted, said by telephone experts present to be fairly satisfactory. In 1908 similar experiments were made by Prof. Majorana, in Italy, between Monte Mario and Porto Danzig, a distance of 60 kilometres, in which good speech transmission was obtained. In France, Lieuts. Colin and Jeance, and Chief Engineer Mercier have achieved the distinction of transmitting speech radiotelephonically from Paris (Eiffel Tower station) to Dieppe, and musical sounds from Paris to the coast of Finisterre, a distance of 310 miles. We may say, therefore, that the transmission of articulate speech by electric radiation has attained at present (1908) to something like the range and efficiency reached by radiotelegraphy ten years ago, and, doubtless, in the next few years will steadily progress; but much has yet to be done before it can compete with modern methods of radiotelegraphy in providing regular communication between ships and the shore. Nevertheless, it is a most interesting and wonderful application of electrical know- ledge, placing at our disposal new means of communication between distant and even moving stations. T 2 INDEX ABSORPTION of the energy of electromagnetic radiation by the earth and sea, 153, 154 Alternating current, definition of an, 2 the graphical delineation of an, 3 Alternator, high frequency : Brown, 81 Duddell, 80 Fessenden, 315 Tesla, 75, 76, 77 Alternators for radiotelegraphy, 52 Amalgamated Radiotelegraph Co. at Cullercoats, station of the, 261 Anderson's method of measuring inductance, 284 Antenna, bent, theory of the, by Fleming, 167 cone form of, 221 direct coupling of, to oscillatory circuit, 161, 162 directive, 164, 221 earthed and non-earthed, 149 fan form of, 220 for reception, 144 multiple wire, 147 practical construction of an, 146 umbrella form of, 221 Apparatus for producing damped electrical oscillations, 39 for radiotelegraphy, 55 Apps interrupter, 46 Arc, Poulsen, 230 Arrangement of transmitting apparatus for short distance radiotelegraphy, 223 Arrangement of receiving apparatus in Marconi's system, 231 Atmospheric ionisation, effect of, on radiotelegraphy, 207 Austin, L. W., thermoelectric detector of, 201 B. BANTI, A., 88 Battelli, A., 17 Be'clfcre, M., 49 Bellini, E., 173 Black, T. B., 17 Branly, E., researches on coherers by, 182 332 INDEX Braun, F., 165 experiments with directive antennae by, 173 Brown, S. G., 184 oscillation detector of, 204 apparatus for production of electric oscillations by, 95 high frequency alternator by, 81 C. CAMPBELL Swinton A., 51 Capacities, mode of adding, 65 Capacity, electrical, definition of, 23 measurement of, 277 of conductors, formulae for, 24 of open circuit oscillator, 138 Carlo Alberto, Marconi's experiments on the, 248 Castelli-Solari oscillation detector, 185 Characteristic curve, 85 curves of an electric arc, 86, 89 Clifden, Marconi radiotelegraphic station at, 170, 249, 250, 251 Closed circuit oscillator, 141 oscillatory circuit, 27 Coal gas mercury break, 49 Code, Morse signal, 216 Coefficient of coupling, 69 of mutual inductance, definition of, 68 Cohen, L., 17 Coherer, Lodge, 182 Marconi, 183 theories of the action of the, 187 Commutator for capacity measurement, Fleming and Clinton, 280 Condenser, charge of, by an induction coil, 42 discharge, production of damped oscillations by, 37 Condensers for electric oscillations, 63 for radiotelegraphy, mode of constructing, 64 Construction of antennae, 146 Coupled circuits, mutual reaction of two, 35 Coupling of oscillatory circuits, 33 Crystal detectors, 209, .210 rectifiers, investigations on, by G. W. Pierce, 209 Curl, definition of a, 118 Cymometer, Fleming, 289 use of, to determine logarithmic decrement, 300 t delineate a resonance curve, 300 Cymoscopes, or wave detectors, 215 D. DAMPED electric oscillations, 5 Damping of electric oscillations, definition of, 5 measurement of, 299 De Forest, L., 165 INDEX 333 Decrement due to resistance, 133 radiation, 134, 139 of oscillations set up in antenna, 1G3 Detector, glow lamp, 205 Detectors, crystal, 209 electrodynamic, 212 electrolytic, 201 imperfect contact, 180 magnetic, 187 spark, 180 thermal, 196 Dielectric constants, table of, 26 Direct charging of antennse, 160 coupling of antenna to oscillatory circuit, 1G1 current, definition of a, 2 Directive antennse, 164 researches on, 165 theory of, 167 Discharge, dead beat, 37 of a condenser, 11 oscillatory, mechanical illustration of an, 38 Dolezalek, F., 17 Drude, P., formula for secondary current in oscillation transformer, 106 Duddell, W., high frequency alternator of, 80 oscillograph, records of sound forms by, 312 researches on the electric arc by, 84 singing arc, discovery of the, 84 therm oam meter, 200 Dunwoody, H. H. C., carborundum detector of, 209 Dynamic characteristic of an arc, 86 E. EARTHED antenna, 149 and non-earthed antennse, relative advantages of, 155 Eccles, W. H., 192, 193 Effective resistance for alternating currents, 9 Eickhoff, W., 61 Electrical capacity, definition of, 23 Electric and magnetic quantities, 107 arc, production of oscillations by an, 83 theory of, 86 ,, current, magnetic force of an, 1 mode of establishing an, 11 force, 110 moment of oscillator, definition of the, 133 oscillation, nature of an, 2 oscillations, undamped, 5 damped, 5 set up in an inductive circuit, 29 oscillator, nature of an, 136 resonance, nature of, 32 strain, 11, 108, 110 334 INDEX Electrodynamic oscillation detector, 212 Electrolytic interrupter, 51 oscillation detector, 201, 202 of S. G. Brown, 204 of Fessenden, 201 of Schlomilch, 201 detector, mode of using in radiotelegraph y, 235 Electromagnetic medium, the, 107 theory, 117 velocity, experimental determination of the, 128 wave, nature of an, 116 waves, 107 practical production of, 128, 129 Electrostatic energy stored up in condenser, 63 Elihu Thomson's experiments on the production of electric oscillations by an electric arc, 83 Energy transformations of electrical oscillations, 28 stored up in antenna, 139 Experiments on directive radiotelegraphy by Bellini and Tosi, 173 F. FEEBLY damped train of oscillations, 8 Fessenden, K. A., 89 electrolytic detector of, 201 high frequency alternator, 82, 315 researches on radiotelephony by, 328 thermal detector of, 198 Fitzgerald, G. F., 128 Fleming cymometer, 289 J. A., experiments on Poulsen electric arc by, 95 with directive antennae by, 171 magnetic detector, 191 multiple oscillation transformer, 71 oscillation valve, 205 theory of bent antenna, 167 thermoelectric detector, 199 Formula for secondary current in oscillation transformer, 106 Frequency of an alternating current, 2 of an oscillatory circuit, formula for the, 32 measurement of, 287 Fundamental and harmonic oscillations in oscillatory circuits, 156 in Marconi antenna, establishment of, 156 in closed oscillatory circuits, 158 G. GARCIA, 165 Glow lamp detector, Fleming, 205 mode of using, 206, 207 metallic filament, 208 Grisson discharger, 62 INDEX 335 H. HAMMER interrupter,, 46 Henry Joseph, researches of, on magnetisation by electric discharges, 187 Hertz, Heinrich R., 128, 129, 164 Hertz's investigations, 121) Hertzian oscillator, 132 Hertz resonance circuit, 180 High frequency alternators, 75 alternator for radiotelephony, 315 alternating current, definition of a, 2 resistance of conductors, 9, 305 ,, of wires, formulae for, 14 of spiral conductors, experimental investigations on, 17 High speed dischargers of Marconi, 253, 254 Highly damped radiator, 140 train of oscillations, 8 I. IMPERFECT contact detectors, 180 Inductance, hjgh frequency, 18 measurement of, 284 measurement of, 283 nature of, 19 of various circuits, formulae for the, 21 Inductances for radiotelegraphy, 66 Induction coil, action of an, 42 qualities required in an, for wireless telegraphy, 44 coils, mode of winding secondary circuits of, 41 of constructing, for wireless telegraphy, 39 Inductive coupling of antenna to oscillatory circuit^ 162 effects of undamped oscillations, 99 Interrupters for induction coils, 45 K. KLEMENCIC, J., 199 LAMME high frequency alternator, 79 Law of exchanges, 136 Lee, Miss Alice, 131 Leyden jar, time period of discharge of a, 32 Linear radiator, 132 Locating direction of radial point, experiments on, by Bellini and Tosi, 177 Lodge, Muirhead & Robinson, oscillation detector of, 185 Lodge, Sir Oliver, 128 Lodge's coherer, 182 536 INDEX Logarithmic decrement, definition of the, 7 decrement, measurement of, 300 Long distance radiotelegraphy, 244 M. MAISBL, S., 88 Magnetic detectors, 187 detector, Fleming, 191 Marconi, 189 Rutherford, 188 Walter, 190, 195 Wilson, 188 theories of the action of the, 191 detectors, investigations on the action of, 192 field, 108 of an electric current, 12 flux, 108 round an electric current, 1 oscillator, definition of a, 136 permeability, 109 Magri, L., 17 Mandelstam, L., 72 and Papalexi mode of charging condenser, 72 Marconi, G., 165 apparatus for spark radiotelegraphy, 225 arrangement of receiving apparatus, 231 coherer, 183 dischargers, 253, 254 early demonstrations of radiotelegraphy by, 218 experiments with directive antennae by, 166 long-distance station at Clifden, 170 Poldhu,245 communication with Atlantic liners, 241 magnetic detector, 189 radiotelegraphic apparatus on Transatlantic liners, 228 ship installations, 239 shore stations, 240 Wireless Telegraph Co., 237 wireless telegraph station at Poole, 219 X-stopper, 243 Maxwell, J. C., 107 theory of electromagnetism, 117 Mechanical illustration of an oscillatory discharge, 38 Mercury jet interrupter, 48 Metallic filament glow lamp oscillation detector, 208 Microphonic control of electric oscillations, 319 Mode of exciting oscillations in open or closed circuits, 160 Morse alphabet, international code, 216 inker, mode of employing in radiotelegraphy, 234 Moscicki condenser, 65 Multiple wire antenna, 148 transformation of oscillations, 70 INDEX 337 N. NAUEN, radiotelegraphic station at, 256 Neon vacuum tube, 289 Non-earthed antennse, 149 0. OPEN circuit oscillator, 136 oscillatory circuit, 27 Oscillation constant of a circuit, 32 measurement of, 291 detectors, as used in radiotelephony, 325 classification of, 179 electrodynamic, 212 electrolytic, 180, 201 imperfect contact, 180, 181 magnetic, 187 mode of using to detect electric waves, 213 thermal, 196 thermoelectric, 196 electric, definition of an, 2 transformer, 33 transformers fur radiotelegraphy, 68 valve, Fleming, 205 Oscillations damped, photographic representation of, 299 undamped, photographic representation of, 299 produced by Poulsen arc, 90, 92 Oscillator, closed circuit, 141 electric, definition of an, 136 magnetic, definition of a, 136 Oscillatory circuits, open and closed, 27, 135 electric circuits, time period of, 27 P. PAPALEXI, K, 72 Pearson, Karl, 131 Periodic time of an alternating current, 2 Photographic record of radiotelegraphic signals. Poulsen's apparatus for the, 238 Pierce, G. W., 209 crystal rectifiers, 210 Poldhu, Marconi long distance station at, 170, 245 PopofFs inventions, 184 Poulsen, V., 89, 90 apparatus for radiotelegraphy, 264, 265 radiotelephony, 318, 321 producing undamped oscillations, 89, 90, 92, 263 arc apparatus, 90, 230, 316 photographic receiver and recorder, 238 Propagation of an electromagnetic wave, theory of the, 123 338 INDEX Q. QUENCHING noise of an electric spark, 57 R. RADIATION decrement of an oscillator, 134, 139 from various oscillators open and closed, expressions for the, 143 Radiogoniometer of Bellini and Tosi, 178 Radiotelegraphic apparatus for short distances, 218, 223 long-distance station at Clifden, 249, 250, 251 Nauen, 256 Poldhu, 245 measurements, 272 station at Cullercoats, 201 stations, 216 Radiotelegraphy on the spark system, apparatus for, 223 difficulties of, due to atmospheric discharges, 241 effects of atmospheric conditions on long distance, 207 long distance, 244 Radiotelephonic apparatus, Poulsen, 327 of the Gesellschaft fttr drahtlose Telegraphic, 322 Radiotelephony, apparatus for, 313 between Berlin and Copenhagen, 327, 328 principles of, 309 special advantages of, 326 transmitters for, 313 transmitting arrangements in, 313 Rayleigh, Lord, 187 formula for high frequency resistance of wires of, 15 Receiving antenna, 144 apparatus of Marconi for radiotelegraphy, 231, 232 arrangements in radiotelephony, 324 Rectifying oscillation detector, 204 Relay, telegraphic, 233 Rempp, G., 61 Researches on imperfect contact oscillation detectors, 182 on production of oscillations by an electric arc, 89 Resistance decrement of an oscillator, 134 high frequency, 305 Resonance curves, 34 curve, delineation of, by the cymometer, 301 effects of undamped oscillations, 104 electric, explanation of, 32 Resonant circuits, 34 R.M.S. value of an alternating current, definition of the, 4 of oscillations, measurement of the, 273 Rubens and Ritter, thermal detector of, 197 Rutherford, E., 188 INDEX 339 S. SALOMON'S and Pyke high frequency alternator, 79 Schlomilch, electrolytic detector of, 201 Seibt, G., 159 Short distance radiotelegraphy, apparatus for, 218, 223 receiving apparatus. 231 Sigsfield, 165 Silent discharger for electric spark, 57 Singing electrical arc, investigations on the, 88 Slaby, A., 61, 165 Somerfeld, A., 17 Sound forms, oscillograph records of, by Duddell, 312 Spark apparatus for production of electrical oscillations, 39 detectors, 180 discharges, 66 resistance, experiments on, 61 measurement of, 306 voltages for various spark lengths, 59 Static characteristic of an arc, 86 Stationary oscillations set up on helix, 159 Stone, J. S., 165 Strecker, K., 165 Syntonic circuits, definition of, 32 TANTALUM oscillation detector, 186 Telefunken Wireless Telegraph Co., station at Nauen of the, 256 Telegraphic method of detecting electric waves, 214 Telephonic method of detecting electric waves, 214 Telephony, fundamental principles of, 310 Tesla high frequency alternator, 75 Theory of two-coupled circuits, 67 Thermal detectors 196 detector, Fessenden, 198 Tissot, 197 Thermoelectric detectors, 196 detector of Austin, 201 Duddell, 200 Fleming, 199 Klemencic, 199 Time period of oscillatory electric circuits, 27, 31 Tissot, C., 197 Thomson, Elihu, 83 Tosi, A., 173 Transatlantic radiotelegraphy, first attempts at, 245 Marconi's establishment of, 247 Transformation of continuous currents into electric oscillations, 83 Transformers for radiotelegraphy, 53 Turbine mercury interrupter, 48 340 INDEX U. UNDAMPED oscillations, production of, from continuous currents, 83 electric oscillations, 5 Upson, W. L., researches on the characteristic curves of electric arcs, 94 V. VALVE oscillation detectors, 204 Varieties of radiating and receiving circuits, 135 Velocity of an electromagnetic wave, 123 of light, experimental determination of the, 128 of undulation through the sether, 110 WALTER, L. H., 186 magnetic detector of, 190, 195 Wave length, measurement of, 294 of waves radiated from Antenna, 164 motion, theory of, 114 nature of a, 112 production of a, 113 velocity, formulae for, 113 Wehnelt, A., 50 Wertheim-Salomonson, 88 Wien, M., 17, 163 Wilson, E., 188 Wireless exchange, Marconi system, 237 X, X-STOPPER, Marconi's, 243 Z. ZENNBCK, J., 153 I)* WILLIAM CLOWES AHD SOXS, LIMITED, LONDON AND BECCLKS. I 222532