There is one more precaution in connection with the purchase of transformers. They should do the same thing for all the important frequencies which they are to transmit. If they do not, the speech or signals will be distorted and may be unintelligible. If you take the precautions which I have mentioned your radio receiving set formed by a detector and one amplifier will look like that of Fig. 94. That is only one possible scheme of connections. You can use any detector circuit which you wish,[10] one with a grid condenser and leak, or one arranged for feed-back In either case your amplifier may well be as shown in the figure. [Illustration: Fig 100] The circuit I have described uses an audion to amplify the audio-frequency currents which come from the detector and are capable of operating the telephones. In some cases it is desirable to amplify the radio signals before applying them to the detector. This is especially true where a "loop antenna" is being used. Loop antennas are smaller and more convenient than aërials and they also have certain abilities to select the signals which they are to receive because they receive best from stations which lie along a line drawn parallel to their turns. Unfortunately, however, they are much less efficient and so require the use of amplifiers. With a small loop made by ten turns of wire separated by about a quarter of an inch and wound on a square mounting, about three feet on a side, you will usually require two amplifiers. One of these might be used to amplify the radio signals before detection and the other to amplify after detection. To tune the loop for broadcasts a condenser of about 0.0005 mf. will be needed. The diagram of Fig. 100 shows the complete circuit of a set with three stages of radio-amplification and none of audio. [Footnote 10: Except for patented circuits. See p. 224.] LETTER 20 TELEPHONE RECEIVERS AND OTHER ELECTROMAGNETIC DEVICES DEAR SON: In an earlier letter when we first introduced a telephone receiver into a circuit I told you something of how it operates. I want now to tell why and also of some other important devices which operate for the same reason. You remember that a stream of electrons which is starting or stopping can induce the electrons of a neighboring parallel circuit to start off in parallel paths. We do not know the explanation of this. Nor do we know the explanation of another fact which seems to be related to this fact of induction and is the basis for our explanations of magnetism. [Illustration: Fig 101] If two parallel wires are carrying steady electron streams in the same general direction the wires attract each other. If the streams are oppositely directed the wires repel each other. Fig. 101 illustrates this fact. If the streams are not at all in the same direction, that is, if they are at right angles, they have no effect on each other. [Illustration: Fig 102]