116 MCEWEN AND MICHAEL. 



Selecting the middle group of each series (groups 2 and 5) as the 

 "standard" (see p. 103) the equations, derived as on page 104 for 

 determining the corrected averages, are 



A = 12.38 + l{2d+ 5/) (50) 



B = 12.47 + ~ (2(i + 3/) (51) 



C = 8.93 + ^ (5cZ + If) (52) 



D = 9.69 + - (2a + 5c) (53) 



E = 11.68 + - (2a + 3c) (54) 



F = 12.41 + ^ (5a + Ic) (55) 



where 



a = B - A (56) 



c = B - C (57) 



f/ = E - D (58) 



/ = E - F (59) 



Following the method of solution given in table 2, the first approxi- 

 mations to d and /^ (d^ = 11.68 - 9.69 = 1.99, f = 11.68 - 12.41 

 = — 0.73) substituted in the second members of equations (50), 

 (51) and (52) give 



AA" = - [2 X 1.99 + 5 (- 0.73)] = 0.037 (60) 



y 



AB" = ^ [2 X 1.99 + 3 (- 0.73)] = 0.199 (61) 



1 In order to save labor and reduce the number of required approximations 

 to a minimum, correction should be made first for the variable having the 

 greatest apparent effect. In this particular case the apparent effects of 

 temperature and precipitation are essentially the same, and first approxima- 

 tions to d and / are ased rather than those to a and c merely because it conforms 

 to the procedure in table 2. 



