AUSTIN: SIGNALS IN KADIOTELEGRAPHY 327 



telegraphy that the increase in strength of signals at night was 

 caused by the decrease in an absorption in the upper conducting 

 layers of the atmosphere after the ionization due to the sunlight 

 and possible cathode rays from the sun had ceased. 



The data accumulated by the U. S. Navy Department during 

 the last three years appear to make this explanation improbable 

 for the two following reasons: (1) It is known that in certain 

 regions and at certain wave lengths the ground absorption is more 

 than twenty times as great as would be the case if the signals 

 were sent over salt water. The sun's rays can hardly be thought 

 to affect the losses in the earth to any extent, yet on some nights 

 these waves travel across the same region, reaching the receiving 

 station with as great strength as would have been the case if 

 there had been no absorption at all.^ (2) Observations on un- 

 damped oscillations from the arc have shown that at night there 

 is a selective strengthening and weakening of the signals with 

 changing wave length. For example, during the recent tests of 

 the arc at Arlington, it was found that when the night signals at 

 the receiving station were faint at the regular wave length of 

 4100 meters, changing to 3950 meters would almost invariably 

 bring them in with greatly increased strength and vice versa. This 

 suggests the light and dark interference bands of optics and, as 

 Dr. De Forest has suggested, ^ the phenomenon may be explained 

 by the interference of a set of waves traveling along the earth's 

 surface with another set which has been reflected from the con- 

 ducting layers of the upper atmosphere. Calculations show that 

 the height of the reflecting surface would be from 40 to 60 miles, 

 which are very probable values for the point at which conduc- 

 tivity would begin. This phenomenon has been observed so con- 

 stantly in the work with the arc that there can be absolutely no 

 doubt of its existence. The fact that it has not been observed 

 in the case of spark waves is due partly to the fact that spark 

 apparatus is generally not changed in wave length by such small 



1 In many well authenticated cases signals have been received with more than 

 twice the strength which the simple Duddell and Taylor law would have indi- 

 cated. / 



* Proc. Inst. Radio Engineers 1 : 37. 1913. 



