190 AUSTIN! WIRELESS TELEGRAPHY NOTES 



150 v. voltmeter with error e z = 10e x . By (2), p = b/dx(xz) = z, 

 while q = x; also R = ze x + xe z ; dR/dx = e z , dR/8z= ex. Substi- 

 tuting and reducing, 



x = \t 



3.2 ohms. 



The condition for minimum probable error is z/x = 4.6 ohms. 



II. Find best order of interference x in determining relative 

 wave-length y = C/x, if an error of observation of 1/100 wave- 



1 / 2x 2 \ 



length at zero path-difference becomes 1 nn ( 1 + /^ on sepa- 



rating plates. Using (5), with /' — — , /" = 2C/x 3 , and <£' 



-C 



x 2 



= 4x/10 9 gives at once the result x = 22,000. 



In practice a distinction must be drawn between precision and 

 accuracy. 2 If Ax denote the precision of observation of x, the 

 criterion leads to the best attainable precision in y; but if, from a 

 knowledge of the reliability of his method, the investigator can 

 go further and employ for Ax an estimate of accuracy, the formulae 

 lead to the conditions for best accuracy in y. 



In either event, it is the aim of the criterion to afford rules of 

 interest to the designer and experimenter rather than to the com- 

 puter. 



For their interest and criticism acknowledgment is gratefully 

 made to President Woodward of the Carnegie Institution and to 

 Drs. Burgess, Dorsey, Buckingham and Mr. L. A. Fischer of the 

 Bureau of Standards. 



RADIO-TELEGRAPHY.— Notes from the U. S. Naval Wire- 

 less Telegraphic Laboratory. L. W. Austin. 



1. Wave Length and Ground Absorption of Electrical Waves 



Since the early days of wireless telegraphy, the country lying 

 to the north and north-east of Newport, Rhode Island, has been 

 known to show a very large absorption for electrical waves. 



2 In short, accuracy = (reliability of method) it (precision of observation). 



