LONG-WAVE RADIO TELEPHONE TRANSMISSION 681 



In any particular study, the field strengths of the radio transmitters 

 were corrected to a constant antenna current and expressed in decibels 

 above one microvolt per meter. Where a study involved a comparison 

 between data observed at different locations, or on different fre- 

 quencies, corrections were applied to reduce the data to a common 

 basis by the application of the Bell System Long- Wave Radio Trans- 

 mission Formula.^ 



The transmitters and transmission paths involved are indicated in 

 Table I. 



TABLE I 



Years Transmitter Rocky Point, N. Y., to 

 West-East 



1923-1924 57 Kc. 2XS New Southgate, England 



1925 57 Kc. 2XS Chedzov, England 



1926 (to Sept.) 57 Kc. 2XS Wroughton, England 



1926 (Oct.) to 1928 60 Kc. WNL Wroughton, England 



1927 (Mar.) to 1933 60 Kc. WNL Cupar, Scotland 



East-West 



Rugby, England, to 



1927-1930, part of 1931 60 Kc. GBT Houlton, Maine 



Remainder of 1931 68 Kc. GKA Houlton, Maine 



1932-1933 68 Kc. GBY Houlton, Maine 



Statistical Frequency Distributions 

 In making a correlation study the best results are obtained when 

 observations of the variables involved have the same type of statistical 

 "frequency distribution." As illustrated in Fig. 1, the cumulative 

 frequency distribution curves of long-wave field strength observations, 

 when plotted (in decibels) on "arithmetic probability" paper, tend to 

 be straight lines and thus of the Normal Law type. The best corre- 

 lation will therefore be obtained when the other variables involved have 

 approximately straight-line distributions on the same type (arithmetic 

 probability) of coordinates. 



Accordingly, cumulative distribution curves were so made for all the 

 variables used in this study. It was found that good approximations 

 to the Normal Law distribution were obtained with logarithmic scales 

 for radio field strength (db above 1 fJLv/m), monthly averages of solar 

 limb-prominences, sunspot-numbers, and Ui measure * of magnetic 

 activity. However, for the "C" measure of magnetic character of 

 days and monthly averages of ultra-violet radiation, linear scales were 

 found to be better than logarithmic scales. 



Correlation by Years 

 In spite of the irregularities in the long-wave field strength curves in 

 Fig. 2, it is evident that during the sunspot maximum period the radio 

 * See Appendix L 



