In order to eliminate group frequency effects, the sending circuit was next driven with an arc whose cathode was a massive copper rod, and a pilot lamp connected to a few turns of wire was placed 10 cm. above the sending helix. All the previous experiments were repeated with similar results, except that in this case. the galvanometer deflections were much stronger, while the telephone recorded only variations in the intensity of the radiation produced by irregular burning of the arc.

In the course of these experiments a condenser of 1 mf. capacity was put in parallel with the detector, which was exposed to a very strong radiation, and the condenser was discharged thru a ballistic galvanometer, giving a deflection which indicated that it had been charged to about 1.5 volts potential. The consistency with which this experiment could be repeated indicated that the oscillations were fairly continuous and regular. When this experiment was repeated with the condenser in series with the aerial and detector, no charge was accumulated on it.

It seemed advisable, for purposes of comparison, to carry out a parallel series of observations with an aluminum cell operated at

60 cycles. Here we have to do with a true valve action, which is due primarily to a large asymmetry in the cell resistance. Fig. 7 shows the scheme of connections employed for the first series of tests. A comparison with Fig. 4 shows that the capacity A of Fig. 7 corresponds to the aerial capacity, the aluminum valve V to the detector, and the capacity C to an additional capacity to be imagined as placed between the aerial of Fig. 4 and the ground. C thus corresponds to the 1 mf. capacity mentioned in the preceding paragraph.

The aluminum valve was a Pb-Al cell with sodium phosphate as an electrolyte. It showed a residual E.M.F. as indicated by

the plus and minus signs, and a conductivity in the two directions, as indicated by the arrows above and below the cell. A ballistic galvanometer G was used to determine the nature of the charge on C, while A was removed from the circuit and tested on the same galvanometer. The condenser A showed scarcely any leak, but the charge on C would disappear in slightly less than a minute.

With no alternating current applied, the residual E.M.F. of the valve would, on closing the switch S, give the condenser C a charge opposite to that indicated in Fig. 7. When the alternating current is applied the terminal potential of C at the instant of test depends on three things: the residual E.M.F. of the valve, the extent of the valve action itself, and the particular phase of the alternating current at the instant of breaking the circuit of C. The first cause is of little relative importance except for very small currents; the last cause produces some irregularity, but the second predominates eight times out of ten, a charge on C resulting as shown in Fig. 7.

Now when S is opened, the case is very different. If A is considerably smaller than C, the residual E.M.F. effect is practically eliminated. The valve action will obviously go on only until A and C are charged up to a sufficient counter E.M.F. to offset the asymmetric impedance of the circuit. Since the valve action of the aluminum cell is small for small currents such as would be transferred by a few microfarads capacity at 60 cycles, it requires several seconds for C to become charged. Indeed, if the circuit is closed only for a few tenths of a second, a charge on the condenser always appears in favor of the residual E.M.F., but if it is kept closed for several seconds, it accumulates a charge due to the valve action. If the circuit is kept closed for a longer time, the galvanometer is just as liable to deflect one way as the other. This is because the charge on C due to valve action has leaked off, and only the third effect remains. This is symmetrical as obviously after equilibrium has been established, the same quantity of electricity must pass in each direction thru the condensers. That A was also charged was determined by testing it with the same galvanometer. In order to repeat the tests A must be discharged previously each time.

If now the capacity A be reduced, the valve action need go on for a much shorter interval to produce the same voltage asymmetry over A, and thus finally the only appreciable charge accumulated on C is that due to the residual E.M.F. This, then, is the clue to the action of the detector in the aerial circuit. The capac

ity of the aerial is so small that only a very small charge is necessary to produce an asymmetry in voltage sufficient to offset the asymmetry in impedance. The previous experiment illustrated in Fig. 3 confirms this explanation. There the aerial capacity was replaced by a much larger capacity K, and a feeble signal was received in the telephone, since the charge on the capacity K leaked back between the oscillation groups. Obviously a deflection of the galvanometer could not be expected.

To carry out the analogy still further, the aluminum valve was connected up as in Fig. 8, which corresponds to the normal detector connection as illus

the galvanometer, the capacity C could have been dispensed with, and the current sent directly thru the galvanometer. The capacity A very quickly accumulates a sufficient charge to offset the asymmetry in the impedance, and thereafter the current is symmetrical, at least as far as average values in the two directions are concerned. The valve action is then made manifest by an asymmetric E.M.F. at the terminals of V, which accumulates a charge on the condenser C. The action is thus strictly analogous to that of the detector.

SUMMARY

1. The series and parallel connections for the carborundum detector have been investigated, and it has been shown, that in general the series connection is of no value, as the aerial current is not rectified to an appreciable extent.

2. The difference between the action with continuous oscillations and with group frequency (spark gap) has been ascribed to a return discharge of the aerial, or capacity K, as the case may

be, thru the detector during the idle periods between group frequencies.

3. For purposes of comparison, the behavior of the aluminum rectifier in similar circuits has been investigated, analyzed, and found to be consistent with the behavior of the detector, when the small capacity of the usual aerial is taken into consideration.

4. The current in the aerial which contains a detector of this type has been found to be symmetrical. The valve action, therefore, makes itself felt as an asymmetric E.M.F. over the detector terminals. The parallel connection is by far the most efficient way of utilizing this action to produce an indication.

The Discrimination of Pitch and its Relation to Training

WILLIAM WELLINGTON NORTON,

Assistant Professor (in charge) of Music, University of North Dakota

THE question of the source of musical ability has brought forth

such a variety of opinions, that it seems worth while to consider some of the phases of a problem which can be scientifically investigated, and may prove of practical value in musical education. Investigators have heretofore tested the upper and lower pitch limits, the threshold of sound perception, discriminative sensibility for differences of intensity, and for differences of pitch. As bearing upon the perception of music, there have been tests for discrimination of groups of simultaneous tones (chords), of tones heard in succession, for perception of rhythm, and for emotional or effective response. These investigations have all dealt with the sensory process of hearing. The study of motor processes of singing will be mentioned later. Our problem would lead to such questions as: Does pitch discrimination exist in "islands"? e.g., can those able to discriminate pitch within the immediate vicinity of International A (435 vibrations) do as well in the next octave higher or lower? What part does attention play in the keenness of discrimination of sounds? If the memory image is involved, what is its nature and importance? Does memory of absolute pitch assist in fine discrimination? Will the ability to discriminate pitch be improved with training? All these questions have their practical bearing upon musical education. Since pitch discrimination is one of the elements in determining musical ability a continued incapacity in this particular might determine whether training would be worth while for those considering the field of music for a career.

In the present research the work was confined to discrimination for pitch and intensity, to discover what correlation between them there may be, to determine the extent of individual differences, and to ascertain the effect of training upon pitch discrimination. In this article the writer will attempt to present the