Guglielmo Marconi's 1896 announcement that he was able to use radio waves for long-distance communication caught most scientists by surprise. However, within a few years these same scientists began to apply their electrical engineering knowledge in order to challenge some of Marconi's assumptions and designs, and suggested new approaches which would lead to much more efficient transmitters and receivers. On November 22, 1899, two meetings discussing "The Possibilities of Wireless" were held by the American Institute of Electrical Engineers, in New York and Chicago. This extract is the transcript of the New York meeting.
1899 Transactions of the American Institute of Electrical Engineers, pages 607-628:

A Topical Discussion at the 137th Meeting of the American Institute of Electrical Engineers, New York, November 22nd 1899. President Kennelly in the Chair, Chicago, Local Honorary Secretary Pierce in the Chair



    PROF. REGINALD  A. FESSENDEN :--Whilst there are many advantages in living in a city in which there is a widespread and intelligent interest taken in scientific work, there is this disadvantage, that when any discovery of a striking nature is made, the professor's friends and the directors of the institution he is connected with expect him to immediately lay aside his own work; work which he may personally consider of much greater importance than the novelty of the hour, and to assist in the development of the new discovery.
    Having thus been forced, some years ago, into X-ray work, with much loss of time and very little results to show for it, I considered myself proof against the seductions of liquid air and wireless telegraphy. Consequently when, having suggested to one of the editors of the New York Herald that they report the international yacht race by the new method, I was invited by them in December, 1898, to undertake the work myself, I declined and put them in communication with Signor Marconi.
    It was found later, however, that there were some exceedingly interesting questions, which had not been solved, in this connection. In none of the work hitherto done had any exact measurements of the quantities involved been made. Consequently many points were still in doubt. The theory of electromagnetic waves had been well and thoroughly covered by Heaviside. But if the theory, as put forward by him, is correct, it is extremely difficult to account for Marconi's law, that the distance of transmission, other things being equal, varied as the product of the heights of the sending and receiving wires. And yet the great experimental ability shown by Signor Marconi in his admirable work rendered it necessary to suppose that in forming our theory we had omitted some consideration which, known to Marconi and his assistants, were as yet unknown to us.
    Unfortunately the receiver in general use, i. e., the coherer, is very ill adapted to quantitative measurements. A simple and very sensitive form of receiver having suggested itself, a few experiments were made in June, but the matter would have been dropped had it not been that my former assistant and present colleague, Professor Kintner, with the greatest of kindness offered to help me, and with the aid of his invaluable constructive and experimental ability, it has been found that it will be possible to carry the work to a successful conclusion, though as yet but a few of the points which it is proposed to investigate have been covered, and these not as yet fully.
    As we have already had one paper this evening and there are some other speakers to follow, I will only touch on one or two points which may be of interest, and on them as briefly as possible. I shall leave the discussion of the question of long distances to others, and shall give, as my contribution to the discussion, an account of the first quantitative measurements which have been made in this line, and some of the results obtained.
    I will first show a model which is supposed to represent the form of the waves which are concerned in the phenomena, so far as we can tell from the latest theories, (or rather from the theories on the subject, because some of them are not of recent date. Heaviside showed a number of years ago what the general shape of these waves ought to be.) I will then show and describe some forms of receivers which I have designed, and which have shown themselves especially valuable for making quantitative measurements. Of course Marconi's apparatus is peculiarly adapted for practical work; but it is not so well adapted for giving exact measurements of the amount of energy or voltage. Moreover, it is not nearly so sensitive as the receivers I shall show. I will then state briefly how my own experiments have agreed with the theory and take up one or two practical points which have developed in the working.
    The piece of wire and pasteboard here shown [Fig. 1,] is supposed to represent the waves: A is supposed to be the sending apparatus; the pole which is connected to the induction coil. When the break of the induction coil occurs the vertical conductor is charged, and the electrostatic lines, according to Heaviside, come out and strike the ground at B. I don't remember exactly into what detail Heaviside goes. He gives the general shape of it. The electrostatic waves come out all around, of course; this model being merely a slice out of the wave front. The model represents the condition of affairs when, after some waves have been emitted, the tension is a maximum. As soon as the spark passes between the two terminals of the induction coil there is a conducting path formed through the heated air to ground. The ends of the lines can now slide down the conducting wire to ground and form a sort of a half hoops as shown at C,--and travel outward. The whole thing, for a cylinder, is worked out in Prof. J. J. Thomson's supplement to Maxwell. Fig. 1
    As it goes out, as you will see from Fig. 1, the rear ends of the half hoops bend more and more in, as at D, until finally when you get some distance away from the exciting source the fronts and backs of the waves are parallel to one another; and form parts of concentric spheres whose center is the originating point A.
    This kind of wave was first experimented with by Lodge, so far as I know. It differs from a true light wave in this, that the two ends of the wave slide over the conductor. If you can imagine the pasteboard strip, E, a mirror, and suppose that the other half of the wave was underneath it at X, as if it were its reflection in the mirror, you then get the full wave, such as you get from light, or from the regular Hertz oscillator. But, as I say, this differs from it, in that the ends of the electrostatic lines slide along the surface of the conductor as they go out. The wave-length is approximately four times the height of the exciting pole. The model is made to scale. The wavelength from F to G being four times the height of A. Consequently, if you double the height of your pole you double the length of the waves.
    K. L, represent the electrostatic lines. M. N. represent the magnetic lines, sticking out at alternate ends, to indicate their direction. When a wave in going along strikes a conductor, such for instance as the wire P, the magnetic lines cut it and in cutting it they create a voltage in it. This voltage, incapable of being supported in the conductor itself, concentrates itself down at the bottom of the pole across the gap of Marconi's receiver and sparks across, reducing the silver sulphide or oxide, and when the current has once passed, the insulation being destroyed, the relay works. The periodicity of these waves, is pretty high. Traveling at the same velocity as light, we find that if we have a 150-foot pole the periodicity comes out about three millions per second. If we take a pole such as would be used to send signals across the ocean, a thousand feet high, we find that the periodicity is 400,000 per second.
    Another point in which these waves have a great advantage over the ordinary light waves is that they follow the surface of the conductor. If, for instance, we have a hill, the waves slide up and over that hill. It is for this reason that I think we can never hope to obtain anything from direct Hertz waves, thrown by a reflector, because those will not work that way; they go straight out, and the only chance that they have is through striking the water and being cut off or reflected a little bit and then sliding along in the same way as the Lodge waves.
    Since these waves travel over the surfaces, the surface must be a conductor, and this naturally has considerable to do with the strength of the waves received. On land, the waves are frittered out very much more quickly than they are on water. For sea water and land the proportion seems to be about three to one, so far as one can judge from Mr. Marconi's experiments. I have made none myself as yet. But I have noticed that there is a tremendous difference on land on different days. On a muggy day you get perhaps five to ten times the throw of the receiver that you will on a dry day. Ice does not seem to make much difference, so far as we can tell; possibly for the reason that in that case the wave goes right down to the warm ground underneath the surface and travels along there, and of course the ice being of very high specific inductive capacity it would not hurt matters very much.
    It is interesting to calculate what the effect of distance ought to be. In the present case you see that at twice the distance you have the same number of lines on the wave front; but in order to get the whole wave front, your pole has got to be twice as high. In other words, to get the same voltage piled up on the receiving conductor over at P you must have a pole just twice as high as it has to be at C. Or, for a given height of pole the voltage available decreases directly as the distance, and with a receiver like Marconi's which depends mostly on the voltage--the coherer would have to be twice as sensitive to work at twice the distance, the pole being of the same height, or in other words at double the distance a given coherer should be worked with double the spark-length at the sending end. The energy, however, is only a quarter as much for a given height of pole. At the receiving end, if you double the height of the pole you would get double the voltage on it. At the sending end, if you double the height of the pole and at the same time double the length of the spark, then you will get a wave exactly the same, except that all its linear dimensions are doubled, and you will get double the voltage at the far end. So that theory indicates that Marconi's rule should be modified to this: "If you double the height of the receiving pole and at the same time double the height of the sending pole and also double the length of the spark, the distance at which you can receive should be four times as great." I cannot find that Marconi's rule ought to hold unless you also double the voltage of the spark, because the effect of doubling the height of the sending wire ought not theoretically to have nearly as much effect as doubling the height of the receiving wire.1
    The energy, however, comes out somewhat differently. If you double the height of the pole at the far end, you get double the voltage. If you double the height of the pole here, and at the same time double the voltage, you will get four times the energy near the origin in a given slice. But if your receiving pole is over at P and the magnetic lines come along and strike it, the conductor gathers in a certain fraction of the width of the wave and this fraction is approximately, as near as I can figure it, about one-seventh of the wave-length, the amount of energy that the conductor scoops out of the wave front depending on the length of the wave itself. Consequently, when you double the height of the sending pole and double the wave length you scoop out a bigger area, twice as much, doubling the energy, but not the voltage received. Take your receiving pole, its height here gives you voltage. Then lay your sending pole crosswise, like this, so as to get a rectangle; the product of these two is proportional to the total amount of energy per wave. If at the same time you have doubled your voltage you would get eight times the energy, but you only have half as many waves per second. So that using a receiver which works by energy and not by voltage, time distance to be sent ought to vary as the product of the heights of the poles if the spark-length is proportional to the height of the sending pole. Fig. 2
    I will now describe some of the instruments that I have made. The first instrument is shown in Fig. 2. It is heavier than necessary, but is nevertheless quite sensitive. The ring A is a bit of No. 26 wire hung on a quartz fibre with a mirror M attached. B is the collecting wire. The magnetic lines strike this and make a voltage, and the resultant current goes around this coil C, around through the coil D, and then to ground. The closed ring is at 45º to the coils C and D. The principle on which it works was first discovered by Elihu Thomson. The current coming in through C and D makes an alternating current field. That alternating current field sets up a voltage in the ring A. The voltage in the ring makes a current in the ring, and the current in the ring reacts on the original magnetic field. The consequence is that the ring tends to turn. It is a very nice galvanometer to work with. You have a quartz fibre. Your zero always stays the same. There is nothing magnetic about it. For the best results it should be connected with a condenser right across it. Of course, the reason of that is obvious. You generate a voltage in it, and if you have got a condenser as at F, then for a given voltage you can get a considerable current in the coils, that current simply depending on the resistance of the coils, and you do not fritter any energy away in current up and down in the receiving wire; whereas if you put a condenser in series with the instrument, your whole current flows through the receiving wire, tapering off, of course, as it goes up, and you lose a great deal of energy in resistance. This is the first and simplest form, and is the form that is used mostly. It works very nicely. Fig. 3
    Fig. 3 is a form which I have tried a little but have not had time to work out very thoroughly: A is a little block with silver wire on it, B is a V-shaped piece of carbon. C is another block with wire on it. D is a closed silver ring which is laid right on top of A, B and C, so that it almost balances on A and C. F is the collecting wire and the current comes down from the pole and goes through the coil X as before, and out again to ground, with a condenser across X as before. You put this ring D on so that it lightly rests on the microphonic contact B. T is a source of A. C. voltage, very low, and H is a telephone. When the current passes through the coil X, it presses the ring down and the previous slight noise in the telephone is made louder, and on the duration of loud noise depends whether the signal is a dot or a dash. This seems to be pretty nearly as sensitive as this other instrument here. Fig. 4
    The third instrument which is more sensitive still, but which I have not had time to work very much on, is shown in Fig. 4. This looks, at first, a little like a single-phase motor. A is a wire ring made in a figure 8 form. B is a piece of iron wire bent into the shape shown. As a matter of fact, in the instrument made, this wire went three or four times through A. F is the collecting wire. The magnetic flux is linked with a part of the ring, and the ring itself has one edge in the magnetic field. The result is, a voltage is induced in the ring, and the current flowing around in the ring acts on the air-gap flux and the ring tends to twist. It is quite sensitive.
    So far as the results obtained, I would say that they seem fairly well to agree with the theory, but as yet not so well as I could have hoped. For instance, doubling the spark-length, you ought to get four times the deflection; whereas I did not get that, though, perhaps, this was due to a weaker stream of sparks. Here is a record of one set of observations made:--
    When the spark-length was 40, the deflection was 60.
    When the spark-length was 32, the deflection was 62.
    When the spark-length was 24, the deflection was 33.
    When the spark-length was 21, the deflection was 30.
    When the spark-length was 13, the deflection was 10.
    When the spark-length was 9, the deflection was 6.
    When the spark-length was 5, the deflection was 3.
    And so on down. The deflection does not quite vary as the square of the spark length, and it ought to.
    The time during which the spark stream is made, seems to influence the deflection very nearly as it ought to. For instance, one result was: Time during which the spark was occurring, 2 seconds, deflection was 20. Time 10 seconds, deflection 94.
    The results obtained are merely preliminary, and before giving them in detail, I want to repeat them under better conditions. To get accurate results you must have a good ground. You must go for the best results to some place near salt water. The ground seems to depend a good deal on the state of the weather, as I say. Sometimes running a wire direct from one station to the other will not improve the ground more than 20 or 30 per cent. At other times, the ground will go away back on you. And it does not seem to depend on contacts, because all the contacts made were amalgamated with mercury.
    The results are not always consistent. Sometimes you get five results the same. Then suddenly, owing to something in the sparks you will get a kick five or ten times as big. One curious thing I might mention was that we got some very hard kicks while we were working. Finally we traced the cause up to the fact that about three or four hundred yards away there was an electric street railroad, and as a car went around a certain corner there was an arc owing to the trolley sparking over a contact. Every time that arc occurred we got a tremendous kick. The trolley itself seems to act as a sending pole.
    The following are the points which have been or will be investigated:
    1. Effect of change of spark-length.
    2. Effect of different kinds of break.
    3. Effect of condensers in sending wire, in shunt to air-gap.
    4.     "          "     in series.
    5.     "          "          "     with receiving wire.
    6.     "          "     in shunt to receiver.
    7.     "     varying capacity of sending wire.
    8.     "          "          "     receiving wire.
    9.     "          "     heights of sending and receiving wires.
    10. Efficiency of different portions of sending and receiving wires.
    11. Efficiency of different kinds of grounds, especially of what may be called the crow-foot ground.
    12. Efficiency of different forms of receivers.
    Some work has been done in nearly all of these lines, but it has been found necessary to turn all our attention to the question of a reliable ground, and when this is settled we shall return to the other questions.
    I have to express my sincere thanks to Professor Kintner for his invaluable assistance, without which it would have been impossible for me to have taken up the subject.
    MR.  W. J. CLARK :--I don't know that I have very much to say this evening on the subject of wireless telegraphy after listening to the very interesting account of Prof. Fessenden's experiments. Perhaps the best thing I can do is to give you my experience. I suppose you all know that during the yacht races I had the pleasure of being associated with Mr. Marconi in reporting the races for the New York Herald. During that time Mr. Marconi had a receiver on the Highlands at Navesink connected with a wire about 115 feet in height. He also had a receiver on the cable boat Mackay-Bennett connected to a wire I think about 115 feet in height. I had a transmitter and receiver on the Grand Ducchesse, 115 feet of wire, and a Thomson induction coil for my transmitter. I used the ¾ in. balls which are always furnished with the Thomson Roentgen ray apparatus, for my oscillator. The balls were in very bad shape. They had been used for a long time without either being cleaned or polished, and in spite of this fact we had no trouble whatever in sending the signals in to the cable boat and also in to Navesink station. Mr. Marconi was on the Ponce, and had no trouble whatever in reading our signals. I have constructed two different styles of receivers. One is a very expensive instrument, provided with a great many adjustments for convenience. Another is a smaller instrument intended principally for demonstration work. I made a test with my expensive instrument, and then with the other one, and I found that the simpler instrument gave me much better results over the long distance, and some days I was receiving from Mr. Marconi at the rate of between 10 and 15 words per minute from a distance of fifteen miles. I have noticed one thing in experimenting; that it is necessary to have the capacity of the two vertical wires equal as nearly as possible. During the recent electrical exhibition I was asked to send the message from the Pulitzer Building to Madison Square Garden. I was only given about one or two days at the outside to get ready, and of course that meant a great deal of work in a very short time. We placed about 260 feet of wire on the Pulitzer Building and about the same amount on Madison Square Garden, placing the transmitter in the Pulitzer Building, and the receiver at the Garden. We found it impossible to get any results whatever. We could not get the slightest sound from our receiver, although it was a very sensitive instrument, and the transmitter was a very powerful one. Our wire was rubber covered, and thoroughly insulated from each building. The way we insulated these wires was by hanging them from a rod of hard rubber about two feet in length and about an inch in diameter. We used a large rubber rod in order that it might be able to resist the strain upon it. At the lower end of the wire we used a similar rod, and then took a tap from our wire and led it in through a window on whichever floor we happened to have our instrument. After the exhibition was over, I made a test between Madison Square Garden and our laboratory, which is not over a block and a half distant. Using about 260 feet of wire on the Garden and about 60 feet at the laboratory, I could get no results whatever, no matter which end I placed the transmitter on, and I was rather puzzled. I found, however, in conducting some other experiments on the bay, and also in the city, that it was absolutely necessary to have the two wires of about the same capacity. We have been able to transmit altogether, aside from what we did at the yacht races, to a distance of about seven miles, and our signals were very clear, indeed. However, I must accord to Mr. Marconi the credit of having by far the best apparatus, better than any I have seen in America, including my own. His receiver is extremely non-sensitive. The receiver which Mr. Marconi was using at the races was a receiver which, under ordinary conditions, that is, under the conditions under which I have experimented, would not receive a message from my transmitter at a distance of more than a mile. Now, Mr. Marconi tells me, at least he hints to me, that the way he accomplishes this result is by measuring the capacity of each wire, and having them precisely alike.
    I have recognized for a long time that the coherer was a very troublesome piece of apparatus, and that it was something which we should try to get away from as much as possible. Consequently, I have been experimenting for some little time on a new kind of receiver. I should have said that when Mr. Marconi was here, although his instrument was receiving from his own transmitter at the rate of but from 18 to 22 words per minute, be said to me that he thought his receiver had about reached the limit of the transmitting operator's speed. I would have liked to have the opportunity of trying to transmit to his receiver using our American Morse code, but that unfortunately was impossible, because Mr. Marconi's assistant could not decipher that code. Now, I have found that by using a new arrangement in the receiver and doing away with the coherer entirely, that we are able to transmit at a very much greater speed; in fact, that we are able to transmit as fast as we can operate the key. I have only experimented with this very recently, only during the last few days; that is, I have only brought it to a state anywhere near perfection during the last few days.
    Another peculiarity I have observed is this: that in the transmitter we get the best results by far, when the secondary of our coil is wound with the finest possible wire. You know when the X-rays first came out, everybody said that we must wind our secondaries with a much coarser wire than they were wound with before. The consequence was that nearly everybody was provided with induction coils with coarse secondaries, giving apparently a heavy spark. Of course we all experimented with those coils because they were convenient; but I happened to have access to one of the old style coils with very fine wire on the secondary, No. 40, I think, and I found the results were very much superior to what I was able to get with the coils wound with a heavy secondary, although the coil with the fine secondary was a very much smaller one, very much shorter spark-length than the coil with the coarse secondary. I think from this, that there are indications of the fact that the balls of the oscillator are not necessary at all. You will remember that when Mr. Marconi first began to experiment, he used balls of four inches in diameter, and stated in some of his publications at that time that the larger the balls to a certain extent, the better the result. Now for 110 miles transmission he is using balls one inch in diameter, and if I had my apparatus here to-night, I could show you that it is quite possible to transmit signals across this room with the balls so far separated that you do not get any discharge at all, and I could also prove to you conclusively that waves did not emanate from the spark at the interrupter. I think from all this, as I said before, that the indications are that we will get along without the balls at all or without any spark. When Mr. Marconi was here he had the kindness to visit our laboratory and I was showing him one of our small outfits and showing him how it would operate by opening a motor switch on the wall with 220 volts and a very small current, and Mr. Marconi expressed his very great surprise that a spark of that kind would work any receiver, because he claimed it was not a static spark.
    Another thing I have noticed is this, that for instance if we have a receiver on this table, and a transmitter on the other table, a very small one, say giving a half-inch spark, if we separate the balls to the full distance of one-half inch, we will not get anything like as good results as if we bring them to about one-sixteenth of an inch, and in a five-mile transmission I have found much better results from a spark half an inch in length than from one inch in length; but in each case I was using a coil of 10-inch spark length, so that the distance of the balls from each other would not interfere with the continuity of the spark.
    I do not know that I have anything further to say; but before sitting down I would like to ask Prof. Fessenden what kind of spark he has been using in his experiments and what difference he has found, if any, between the heavy spark and the light spark, so to speak.
    PROF. FESSENDEN:--I used an ordinary induction coil for the work. Part of the time I used it with the ordinary break, and part of the time with a Wehnelt break. I did not find very much difference between the two. I found a one one-hundredth of an inch spark to work very well at about 200 yards with 15 ft. poles, and it did not seem to make very much difference what the kind of spark was. I did notice however that when we got a flaming spark, the results were not quite so good. There is always a little risk in making experiments with the sender too near the receiving instrument, because electro-magnetic effects are apt to come in. I noticed this when working at first from one end of the laboratory to the other. The core of the induction coil, I found, would make sufficient stray magnetism to work the thing.
    I will say, as regards the sensitiveness of this particular instrument, that it would seem as if a mechanical device ought to be better than the coherer, for this reason--we can make a galvanometer which will work with a current of about 10-11 amperes and a resistance of about 100 ohms or less. That makes about 10-20 watts or 10-13 ergs per second. I do not think that you can make a coherer work much under a fraction of a volt. What is the smallest that you have succeeded with, Mr. Clark?
    MR. CLARK:--I have not experimented in that line at all.
    PROF. FESSENDEN:--I do not think you can make it to work very much under a volt; that is of excessive voltage above the regular voltage of the battery, and in that case, roughly taking the quantity of electricity as equal to the smallest amount possible, that is to say, enough to charge it electrostatically, which would make its capacity about unity, we find you would get about five millionths of an erg. Now reducing ten seconds period down to a tenth of a second, you find the galvanometer should be about ten thousand times as sensitive. Of course you won't get this; but it seems to show there is some hope that way2
    DR. M. I. PUPIN:--I would like to ask Mr. Clark why he did not clean the balls. When he said he worked with dirty balls, it seemed to me it was a very small matter to clean the balls. Why didn't he clean them? Was it on purpose or accidentally?
    MR. CLARK:--I did not clean them because Mr. Marconi said to me that he found better results at the present time with rough balls than with smooth ones, and that rather tended to confirm the results of some of my own experiments.
    If I may be permitted I would like to ask one more question. I do not profess to be a theoretical man. I am working entirely on practical lines, and perhaps some of you this evening, perhaps Prof. Fessenden, can set me straight on one or two points. Prof. Fessenden mentioned that he was able to get transmission across the room by means of the core of the coil. My question to him may not seem a very bright one, and if so I ask it in ignorance. Why is it then that if we take a very powerful magnet and bring it in the neighborhood of the coherer and move it around all that we please, we cannot affect the coherer in any way.
    PROF. FESSENDEN:--I think that the reason of that would be that when the interrupter breaks the circuit the iron demagnetizes in about the ten-thousandth part of a second, or something of that sort, and you would have to agitate yourself very vigorously in order to get equal results.
    DR. PUPIN:--What was the spark-length that Mr. Marconi had?
    MR. CLARK:--One inch--about two centimetres.
    MR. F. V. HENSHAW:--I would like to be enlightened a little bit on the spark-gap question. I had an impression that we had some good rules to go by in regard to proper height of pole and length of spark-gaps, etc.; that if you wish to transmit a long way, you should have a long spark, and if you want to transmit a short way, you should have a short spark. Now it appears that the length of the spark doesn't make any difference, and later on we learn that we can do it all right with no spark at all. If there are any quantitative results with regard to the ratio of the spark-length to the other factors, I think it would be very interesting to have them.
    PROF. FESSENDEN:--I have just given my measurements on the relation between the length of spark and the effect at the receiving end. I pointed out that the effect at the receiving end did not quite vary as the square of the spark-length. So that in doubling the spark-length you do not get quite four times the effect, as you should.
    MR. HENSHAW:--Then there must have been something in Mr. Clark's results that masked that. Didn't I understand you to say that you got better results with a one-sixteenth inch spark than with a long spark?
    MR. CLARK:--Yes, for a short distance, and Prof. Fessenden explained how that occurred. He has experimented to a greater extent in that line than I have. But Mr. Marconi has transmitted 110 miles with a one-inch spark, and he uses a one-inch spark from five to ten miles, just the same. In fact, he has adopted a spark of that length as a standard.
    MR. C. E. DUNN:--Mr. Clark, I believe, said a few moments ago that he found that he got results without any spark at all. I would like to ask him if he had any idea from what the effect arose.
    MR. CLARK:--I do not claim to give any reason for it at all. I simply state the fact. Prof. Fessenden, as you have heard, states that he thinks it was due to the magnetic influence from the core of the coil. But I have tried the action of a very powerful magnet in the neighborhood of a coherer, and making and breaking the circuit of that magnet with a vibrator, placed at a safe distance, I have found no results. But in case any of you may be experimenting with the interrupter of your coil in a vacuum, I want to tell you something that happened at the exhibition over in Boston. We were blowing up a boat every day, four times a day, by wireless telegraphy from across the hall, and we found, to our astonishment, that at about two o'clock each day something interfered with us, and we consequently had to blow up the boat a few minutes before two. Now, Mr. D. McFarlan Moore had his artificial daylight down in the basement, several hundred feet away from us, and we found that he started up about two o'clock, and after a little experimenting we found that we got these waves from Mr. Moore's apparatus. You all know that his interrupter is in a vacuum, and we could not find any sparking or anything that we thought would influence our receiver, except the sparking in the vacuum; so after that we claimed to blow the boat up by artificial daylight.
    DR. PUPIN:--In reference to getting the effect without the spark, I think I was the first to show Mr. Clark that under certain conditions it could be obtained. Mr. Clark, two or three years ago, showed his apparatus to the New York Electrical Society. I was its president then, and heard Mr. Clark's lecture. I tried the apparatus that Mr. Clark showed, and I saw that it would work whether the interruption of the primary current produced a spark or not. I think Mr. Clark was very much surprised that such was the case. Of course, the distance was the length of a table, and the receiving line was connected to a water pipe, and the transmitting line was connected to the same water pipe, so it was equivalent to having the receiving and transmitting line connected, being one line, of course, and as soon as the primary was broken, of course the transmitting line as well as the receiving line was charged, and that charge produced a spark in the coherer and made it operate. It is, of course, not necessary for a short distance like that to have a spark at all. But, if you take 100 yards distance, I am perfectly sure you could not get any effects at all unless you have a spark.
    MR. CLARK:--On the occasion which Dr. Pupin mentions, one wire was connected to the water pipe, and the other to the gas pipe--at least, I supposed that was the case. But I found, to my delight the next morning, that our man had not carried out his instructions, and the transmitter was not connected to a ground at all. At the present time, with the instruments that we are now using, I find no difficulty at all in transmitting 30, 40 or 50 feet with only a perpendicular wire of about eight or ten feet in height, with no ground connection whatever, which of course is very much more interesting. I have also succeeded in directing the movements of a model automobile, quite a large-sized piece of apparatus, weighing about 50 pounds, with no trouble at all.
    THE PRESIDENT:--Was that with a spark?
    MR CLARK:--Yes, with a spark.
    THE PRESIDENT:--If there is no further discussion, I think that we ought to extend a vote of thanks to Prof. Fessenden and to Mr. Clark for their entertaining descriptions. The subject is an extremely fascinating one, fraught, perhaps, with many possibilities in the future, and the discrepancies which are bound to occur between the evidences of practice and the prophecies of theory are only the more interesting, since they show the great complexity of this subject and the many facts which enter into each case. But, no doubt, as we get more experience on the one hand, as we acquire better apparatus on the other, and as we also get more and more insight into the nature of the phenomena which are involved, we ought to be able to bring theory and practice into a more complete degree of uniformity.
    DR. PUPIN:--I was invited to take part in this discussion, and I came to discuss the subject, but I would not carry out my resolution, if I did not think that we have not yet touched upon the possibilities of wireless telegraphy. So far we have had an exchange of personal experiences. These experiences are all very interesting, and I am sure that we are very much obliged to Prof. Fessenden and to Mr. Clark for their very valuable information. It is extremely important that we should know from actual personal experiences what are the difficulties involved in wireless telegraphy. But the subject of this evening is "the possibilities of wireless telegraphy," and I do not think that the AMERICAN  INSTITUTE should adjourn without saying something relative to these possibilities. Of course when a man speaks about the possibilities of anything, it is all in the future, and he has poetical license--he can indulge to his heart's content. Now, if you will allow me to indulge a little bit, I will tell you something about what I think might be the possibilities of wireless telegraphy. My ideas on the subject are not an offhand notion. I have thought a great deal about the subject. I teach in a college; am called upon quite frequently by young students to tell them something about wireless telegraphy, and if I go out into society or among business men, and they know that I am a professor at a college, they think that I know all about wireless telegraphy; so they come to me and ask me all about it, and of course I try to give them an intelligent answer. So, whether I want to or not, I have to think a great deal about it. It is in the line of my business. If I make a mistake, in my figures, or if I should be a little too sanguine in my expectations, you will be very indulgent with me, because I am speaking about mere possibilities.
    One of the possibilities is the tuning of receiving apparatus. The chief difficulty to-day is that if two transmitting stations are working at the same time, the message received at the receiving apparatus is unintelligible, because they interfere with each other. It is, therefore, very desirable that the receiving apparatus should be tuned so that it will respond to one transmitting apparatus only, and to no other. The second important point is that we should extend the distance as much as possible. These are the two important points, and I think the possibilities in those two directions are very great.
    Now, if you will allow me to go to the blackboard, shall try by diagrams to make myself clearer than I might otherwise do. The evening is still young, and we have lots of time; so don't get impatient. In case you want to go, why do so, I won't feel offended. Fig. 5
    The transmitting wire as it is constructed to-day consists of a long vertical wire, a b, Fig. 5, and a spark gap a c, and an earth connection G, and the receiving wire is e d, the coherer is at f, and the local circuit g f h, the induction coil A, is also indicated connecting to the two sides a c of the spark-gap. I would like here to ask Mr. Clark a question. How many breaks per second did Mr. Marconi employ:
    MR. CLARK:--I cannot answer that question definitely. He had a very rapid rate, the most rapid one that I have heard. I can only guide by the ear.
    DR. PUPIN:--The fastest mercury break used with induction coils that I have seen, did not go very much over thirty a second.
    MR. CLARK:--Mr. Marconi said that all he required for dots was one spark between the balls.
    DR. PUPIN:--That is not the point, though, that I wanted to know.
    PROF. FESSENDEN:--You can call it 500 per minute possibly. When it gives a high pitched note, it would be somewhere around there.
    DR. PUPIN:--That is about 8 per second. When you excite two wires like a b and c G until the potential is high enough to make the spark in the spark-gap, then each spark is followed by a series of oscillations. They are the so-called free oscillations of this wire--the wire b a c G. The wave-length will depend on the nature of the excitation. I do not care to go very deeply into that for two reasons--because I don't know much about it, and secondly, it is not necessary--two very good reasons. I will tell you why it is not necessary, and why I do not know much about it. Take a wire like b G, the capacity of which is a variable quantity, because the capacity of any element depends on its length and its distance from the earth. As you go up along the wire the capacity of the wire is continually diminishing. The free vibrations of a wire like that, are like the free vibration of a string, the elasticity of which varies from point to point. Now, no man on earth would be able to tell you with any degree of exactness what sort of oscillations you would get there. I don't think you would get a pure harmonic oscillation. There will be an oscillation of all sorts of unrelated frequencies. If we had an ordinary wire where the capacity per unit length is a constant quantity from beginning to end, the problem is very easy; it as been solved; the wave-length that the wire emits is a multiple of the length of the wire. Prof. Fessenden told us that the wave-length is about four times the length of the wire. Now, these oscillations are very much dampened, for this reason--that you have a great deal of dissipation of energy right in this spark-gap. If you could excite oscillations in a wire without a spark-gap, these oscillations would have some damping, but not very much. They would be, so to speak, sonorous oscillations, as when you strike a bell made of fine bell-metal, it continues to ring for a long time after the stroke is delivered. But if you put a finger on the bell and strike it then, or put in some resistance, the sound dies out very rapidly. So that is the state of affairs in this transmitting wire with a spark-gap. You have the high resistance, dissipation of energy, and the oscillation is a very damped one; you have only a very few waves sent out after each spark. Now, when you have a train consisting of very few rapidly diminishing waves they cannot produce much resonance. Fig. 6
    That is the reason why they have not been able to tune their receiving apparatus in England, that is, I think so. I don't know anything in this line for certain. I know only one fact, and that is that nobody has been able to tune the receiving circuit. What I mean by tuning is this if the receiving wire has the same period as the transmitting wire, then the wave coming from the transmitting wire will excite strong oscillations in the receiving wire under proper conditions; but if the receiving wire has not the same period as the transmitting wire, then the oscillations induced in it will be small. Now, even if the two are in unison, and the transmitting wire sends only a few rapidly decaying oscillations at stated long intervals, then you will get no appreciable resonance. To produce strong resonance you must send forth oscillations which have little damping. That has been shown experimentally in the case of the Hertzian oscillations, and these are, in a certain sense, like Hertzian oscillations, because they are of high frequency. Hertz showed that if you have a series of slowly decaying waves, and they strike a conductor having the same periodicity as the conductor from which these waves proceed, you then get very strong resonance effects. Here is another method of producing strong resonance effects. If you produce a series of rapidly decaying waves, you will get small resonance, but this small resonance effect would be magnified if you could repeat these dampened impulses at very quick intervals. Instead of having eight sparks per second, suppose you had one hundred thousand sparks per second, each succeeding train of rapidly decaying waves will strengthen the effect of the preceding train. Then the current here in the tuned receiving circuit would swell up, and ultimately you will get a very large resonant oscillation by a sort of accumulative effect. So that in the present method of wireless telegraphy, we are laboring under two difficulties: one difficulty is that the waves proceeding from the transmitting wire are very much damped. The second difficulty is that the sparks do not proceed in a sufficiently rapid succession. Eight sparks per second--that is nothing at all, it is the coarsest kind of dilettante work. It may be good enough for the present, but it is not enough for the possibilities of the future. If we are to speak about possibilities of the future, then it seems to me we ought to consider these two points: One is to produce electrical oscillation in this transmitting end, which are less damped; secondly, to produce a much more rapid succession of sparks, a much more rapid succession of impulses. Now, I am going to tell you one way that I think would operate and be successful in transmitting less dampened oscillations. I shall describe a method which I have tried, in connection with another scheme, for which there are several claimants besides myself. Suppose we have a wire a b c G (Fig. 6) and a condenser a c. Suppose that the condenser plates have a shunt with a gap a1 c1. Connect the induction coil with the two sides of this spark-gap. This is a condenser having a considerably larger capacity than the capacity of the vertical wire, and it is preferable that the shunt wire have considerable self-inductance. This arrangement represents exactly the following mechanical analogy. Fig. 7
    In Fig. 7, A B C is a tuning fork, with its neck, C, rigidly fixed, and a string, A D, attached to it. One end, D, of the string is fastened. Suppose you strike this tuning fork, it will vibrate and make the string vibrate, and the vibration of the string will continue as long as the tuning fork vibrates. If the tuning fork is heavy, its vibrations will continue for a long time after each stroke. The string performs forced vibrations, and not natural vibrations, which it would perform if it were fastened at its terminals and then struck. The forced oscillations of the string are not rapidly decaying, because the body vibrating it--the tuning fork--is a sonorous body. The condenser, a c (Fig. 6), with a shunt, is an electrical oscillator, an electrical tuning fork. The Hertzian oscillator is an electrical tuning fork. The analogy is perfect. It is not a mere superficial analogy; it is a perfect analogy. Every time you pass a spark, that is equivalent to giving the tuning fork a stroke. You start oscillations in this circuit just as you start the vibrations of the tuning fork by a stroke. These oscillations then keep up the oscillations in the vertical wire. Oscillations like that have been produced by Hertz himself. The wire was not connected to the earth, to be sure, but that does not make a very essential difference. I will give you an illustration of one Hertzian arrangement which is very interesting. It is represented in Fig. 8. A, is the induction coil, B C D E are four plates 16 in. square, and at a distance of 4 in., and a b is the spark-gap and E H and D K are two long wires with bridge, F G. The circuit, C B E F G D C, is the oscillator, the electrical tuning fork; the wires, E H and G K, are the electrical strings vibrated by the electrical tuning fork. This arrangement, the so-called Lecher arrangement adopted by Hertz, gives electrical oscillations, when the spark-gap is small, which are very sonorous. After each spark a long series of waves will be emitted from a system like this. I don't care where these waves go, a part of them will certainly strike the receiving wire and will continue working there for some time. Now all you have to do at the receiving wire is to have another electrical tuning fork synchronized with that at the receiving end. Such an arrangement is given in Fig. 6. The condenser l with its shunt o m n is the resonator. By varying the dimensions of the shunt m n o or the capacity of condenser l we can tune this circuit. I do not see why such a circuit should not resonate. I would like some one to tell me why it should not. It does it in the ordinary conditions in the Hertzian experiments. I do not see why it should not do it here. There is nothing mysterious or even strange about these waves employed in wireless telegraphy, they being perfectly simple waves like any other electrical waves and can be made to obey the same rules. Now the receiving resonator will be even more sonorous, a great deal more sonorous than the transmitting oscillator. Two circuits may have time same period but different decrements. Two bells may sound the same note, but one will continue for a long while to ring and the other may die out very shortly, depending on the internal friction and depending on the material from which the bell is made. The receiving resonator will have the same period as the transmitting oscillator when they are in unison, but the first one will have a larger decrement. That is because its frictional losses are larger. The coherer, f is, of course, supposed to be in a secondary winding as shown in Fig. 5. The oscillations once started in the receiving resonator will continue a much longer time than those in the transmitting oscillator. Now that is a very important point. Prof. Fessenden has told us that these oscillations are about three millions per second. I think he is very near the truth. They are considerably slower than the Hertzian oscillations on account of the length of the transmitting wire. Now these oscillations, excited by a resonator like the one in Fig. 6 are very persistent. Five thousand wave lengths, that is five thousand complete oscillations, will be emitted. Oscillations will take place before the resonator becomes exhausted. The oscillations in the oscillator are not so persistent on account of the air-gap. Say 500 oscillations take place after each spark before the oscillator is exhausted. That is, for five hundred three-millionths of a second, or for five thirty-thousandths or for one six-thousandth of a second, these oscillations will persist after each spark, and then the electrical tuning fork is ready again to be struck again by an electrical blow, by a spark. You see that this oscillator is a tuning fork, that you cannot strike before the oscillations have subsided, because the air-gap conducts too well while the oscillations are going on. On account of the air-gap you have to wait until the oscillations die out before the induction coil can produce another spark. So that only after one six-thousandth of a second the oscillator will be ready to receive another spark. Now, the receiving resonator will continue considerably longer, there being much less damping. That is, after one six-thousandth part of a second, after the last wave-length is reached here the induced oscillations will continue for a considerably longer time, say ten times as long; that is, one six-thousandth of a second, plus ten six-thousandths of a second. This is of course very rough. I do not pretend to great accuracy. I only wish to make my meaning clear, and to illustrate it by a numerical example. One six-thousandth plus ten six-thousandths--well, make it twelve six-thousandths per second, which is one five-hundredth of a second. You see, the oscillations in the receiving resonator will continue for one five-hundredth of a second after the first spark in the transmitting oscillator. Now, suppose that you pass a spark every thousandth of a second. You see, then, that when the second spark takes place, and the wave-train sent out by this spark begins to arrive at the receiving end, the oscillations there have not died out yet by one five-hundredth of a second; there is still wave-energy left at the receiving end. Now, the second spark will add more wave-energy to the receiving resonator, the third still more, and so on. There will be an accumulative effect in the receiving resonator. It is just like having two tuning forks that are tuned to each other. Strike one tuning fork the second will resonate to some extent. Then, before the energy which the second tuning fork has received, has decayed, if you strike the first tuning fork again and again, the resonance of the second tuning fork will continually increase until it readies a maximum effect, and you cannot go beyond that. That may be called the accumulated effect of resonance. You see, then, the utility of numerous sparks rapidly succeeding each other; I think a thousand per second is none too much. Well, the scheme would operate all right if you could do all these things. But here are the difficulties, and this brings us to the second point of my discussion. I have made a short calculation, using for the condenser capacity at the transmitting end about two-hundredths of a microfarad. That, is, the condenser plates are about a meter in length, and they are at a distance of one centimetre, and using a spark-gap of a quarter of an inch, requiring ten thousand volts, roughly. The coil will have to supply a kilowatt to the electrical tuning fork at the transmitting end, and a very large fraction of that kilowatt is used up in this spark-gap, because an electric radiator is not very much more efficient than an ordinary radiator; a great deal of it goes into heat. I think the efficiency is just about as bad, if not worse; so that perhaps 95% of that kilowatt would be used up in the spark-gap, and the rest is radiated into space, a very small fraction of which operates the receiving apparatus. What do you think becomes of the spark-gap spheres if the heat generated there is not carried off? They will be dirtier than the balls of Mr. Clark's instrument. I think that in less than no time they will be fused, and you won't be able to get any spark through them. There is the difficulty. The rest of that I leave for you to solve. We must put a great deal of energy into our transmitter in order to increase our distance of transmission. The more energy you put into any radiator, I don't care what it is, the better illumination you get. The more energy you can get into an electrical radiator, the more energy you can get out of it, the longer the distance over which you can transmit a certain amount of energy. There is not the slightest doubt about that. What we want, then, is, in the first place, to be able to put a great deal of energy into our radiators; secondly, very rapid succession of sparks; thirdly, radiators and receivers of small damping, and as a result of all these things, an efficient tuning of the receiving to the transmitting apparatus. The solution of these problems will increase the sphere of future possibilities of wireless telegraphy more than anything else that I know of. Fig. 8
    THE PRESIDENT:--Is there any further discussion?
    MR. CHARLES P. STEINMETZ:--I only want to add that I should not be afraid to dissipate even more kilowatts between a pair of balls. I think the best way would be turn a stream of high pressure air blast on it. You have no idea of the enormous amount of energy you can carry away in an ordinary air blast, twenty to thirty pounds pressure to the square inch. I had experience with that. I had occasion once to dissipate a considerable amount of energy between a pair of balls which I used in a high-frequency condenser charge.
    MR. C. O. MAILLOUX:--I would like to ask Dr. Pupin if he would not be likely to have an additional difficulty in the synchronizing of his set of waves, which it seems to me would be essential for obtaining cumulative effects by resonance. When he starts a set of waves and gets resonance at the other end, the resonance can be kept up and increased only if the succeeding wave is in phase with the preceding one. It seems to me that we have to synchronize with the second set, and with all the sets succeeding each other, because otherwise it would be like vibrating a string, and then trying to vibrate it again in some different phase. Taking Dr. Pupin's example of the two tuning forks, if he struck the first tuning fork anew too soon or too late, or if he did not strike it at a certain critical time, or in the same, phase of vibration, it seems to me the effect would be that the second fork must synchronize anew, and hence I should think there would not be any cumulative effect in the resonance. Each wave would have only its own resonance.
    DR. PUPIN:--No, there is no interference. Wave-energy is superpositive. The transmitting apparatus sends waves of the same period all the time. You can put them anywhere you please. There is no interference.
    MR. MAILLOUX:--Do they accommodate themselves by mutual induction--the inductance effect or electrostatic effect?
    DR. PUPIN:--Do you mean to say that when I sing ah, ah, ah, I synchronize the emission of my sounds, and that unless I do that you won't hear me? It is the same thing.
    1. At the close of the discussion, one of the officers who was engaged on the recent tests of the Marconi system for the Navy here, informed me that this was the case, i. e., that it was found in practice that if two vessels had wires of different lengths, communication should be carried over a longer distance when the vessel with the higher pole was the receiver than when it was the sender. This is a very satisfactory corroboration of the theory.
    2. NOTE B. The original instrument described can be used for receiving signals up to 5 miles with a 20-foot pole and 5-inch spark, even without the use of condensers. A slight change recently made, has increased the sensitiveness about 40 times, still without the use of condensers or tuning. For those who may wish to experiment in this line, I would say, that the following dimensions will give an instrument much more sensitive than the coherer, and capable of use over long distances. Coils, two in number, 400 turns of No. 32 wire, boiled in paraffin. Coils ½ inch inside diameter and 1/3 inch long. Moveable ring, No. 26 wire, ½ inch diameter, mirror ¼ inch diameter.
    Where great sensitiveness is required, use smaller coils and ring of silver with a scrap of silvered glass, very thin and 1/16 inch diameter fastened to ring. Observe motion by reflected spot of light from glass. A silvered film of mica or celluloid might be used where great dampening is desired.
    It may be of interest to mention that in sending signals across 1½ miles of city buildings, diagonally across blocks, the signals were only about 10 per cent. of what they would have been over free space.