line curve the equation of which is y-4.47x; y being the voltage
across the primary. See Curve Sheet No. 1. Figure 3.

Then the matter of obtaining the ratio between the primary
and secondary voltages was undertaken in the following manner.

Reference was made here to the A. I. E. E. standard spark gap voltages and also to the results of Fisher's investigations as found in references mentioned under head of Bibliography.

Since Mr. Fisher's curve of voltage versus spark gap distances in air, cross the A. I. E. E. curve at the point corresponding to a spark gap distance of 1.3 inches; he recommends that, for relative comparison and the rating of any high tension transformer, the exploring coil voltage be carefully found, which corresponds to the voltage across the secondary terminals necessary to cause a spark between the points of sharp needles No. 12, when placed 1.3 inches apart.

This was done many times under various conditions of atmosphere, using new needles for each test. The primary voltage was measured in place of the exploring coil, since they have a straight line ratio; and on account of obtaining greater accuracy due to the larger readings.

Voltages were also measured corresponding to spark gap distances of one inch and two inches. The following table shows the average results of the readings taken.

According to the advice of Fisher, the secondary voltage required to cause a spark of 1.3 inches was assumed to be 25,000 volts.

The ratio between primary and secondary is therefore

25,000 63.5 = 394.

This ratio was then multiplied by the primary voltages 50, and 90, corresponding to spark gap distances of 1.00 and 2.00 inches respectively, and these values plotted along with the curves of Fisher and the A. I. E. E., for the purpose of comparison. The curve drawn through these points, gives Bristol's Interpolated Curve as labelled on Curve Sheet No. 2. Figure 4.

For voltages above 150 Multiplier No. 11625 was used. Ratio 3.8.

MEASUREMENT OF DIELECTRIC HYSTERESIS OR ENERGY LOSS.

To get the energy loss of any given reading, the watt loss at open circuit for the corresponding voltage was subtracted from the watt loss in the reading under consideration.

The justification for this was on account of the extremely small variation in the regulation of the transformer under such load variations as apply in the tests; and because great care was taken to eliminate all possible leakage. The value, however, herein found and so named dielectric hysteresis, is not intended to mean absolutely that alone, but rather includes what other losses there may have been, such as those due static discharges and leakage.

ARRANGEMENT OF APPARATUS.

The arrangement of apparatus for measurements of dielectric strength, arcing voltages, and dielectric hysteresis which was found most convenient and satisfactory, is shown in Figure 2 and Plate LXVIII.

CONDITIONS OF TESTS, MATERIALS, TESTING DEVICES AND GENERAL METHODS.

CONDITIONS OF TESTS.

In this thesis it has not been my object to investigate the effect of various conditions of tests; as mentioned in the introduction.