172 BELL SYSTEM TECHNICAL JOURNAL 



The mean frequency difference or spacing interval between suc- 

 cessive minima for the records given in Fig. 20 is approximately 

 2,200 cycles. Therefore, the mean wave length difference in length of 

 path from equation (5) is 277 wave lengths, or 136.5 kilometers. 



It is evident that the errant waves following the second path must 

 have been led a devious route. While this is about all the informa- 

 tion which can be deduced directly from these data it is interesting 

 to speculate further with the information along the lines of some of 

 the theories which have been proposed to account for such wave 

 deflections. For instance there is the Heaviside layer theory in which 

 there is supposed to be a more or less well defined reflecting layer in 

 the upper atmosphere. For this we would visualize our high alti- 

 tude waves as proceeding in a straight line up to the layer, being 

 reflected, and striking back to earth at the receiving station. 



Since the distance from transmitter to recei\'er was 110 kilometers 

 the length of the secondary path was 110+136.5 or 246.5 kilometers. 

 By triangulation the height of the assumed reflecting layer may be 

 determined as very nearly 110 kilometers or equal to the distance 

 from transmitter to receiver, and the angle of incidence is 26.5 degrees. 



As yet no positive information has been acquired concerning the 

 variation of difference in length of two major night-time transmission 

 paths with direct distance from the transmitter. If the path differ- 

 ence is due to reflection from an overhead layer, the expected rela- 

 tion by triangulation becomes quite simple. 



When Ad is the difference in length of path, y is the direct distance 

 and h is the vertical height of the layer. 



An investigation of this relation would probably do much to prove 

 or disprove the reflection theory. 



At this point it is well to recall the results of earlier tests in which 

 it was observed that single frequency waves separated by 1 ,000 cycles 

 faded in approximately an inverse relation also indicating a spacing 

 interval of about 2,000 cycles. The agreement of these earlier records 

 is particularly noteworthy since about three weeks elapsed before 

 the more detailed band fading records were made. 



Fig. 20 shows a time variation in the frequency position of the 

 minima which is explained as due to a variation in the difference of 

 path length. If we indulge in further speculation along the line of 

 layer phenomena we conclude that the reflecting layer is rising and 

 falling. It is improbable that the whole layer would rise and fall 



