ccii Tables for Statisticians and Biometrwians [XXXVIII XLI 



-i +nj+ ... + w 5 ') 9 (n/ + w a ' + . . . + <) 



(4-5)3 (3-5)? 



6Z) = 



15 (%'+< + %') _ Il(n 1 / + n 2 / ) 3%' 

 (2-5)? (1'5) ('5)5 



, - + <+...+ T^sQ . 4 ( < + wg' + . . . + Q 

 ~"~ 



(3-5) 





(2-5)2 (1-5)2 (-5)2 ' 



and to determine q from ?? 6 ' 



_-3622(n 1 ' + - 2 ' + ...+rc 5 / ) 13899 



(xv) continued 



7? 



21885 + W+W) 15601 +'^) , 



"~5^~ ~3^~ 



...... (xvi). 



In this particular case we have, from (xvi), on substituting for %', n^ ', n s ', n 4 ' 

 and n 5 ' at the first terminal 



- 1034 19505-979 70870422 104181-719 66954*292 J 



"ll^ + "~9^ "7^ ~V~ "3^~ I' 



We find the following values: 



q= 0, 08, -10 , -12, 1-00, 



rc 6 = - 436-33, -165-10, - 67'75~, +43-52, +57221-11. 

 Therefore q = "12 is the best value for q. 

 Substituting this value for q we have 



^=-702-128,820, NB = - 70*292,106, NC= 10746,671, 



ND = - -315,428, NE = - -082,135-. 

 Hence our auxiliary curve at the first terminal is 



Z= 1715 + a' 12 (- 702-128,820 - 70-292,106^ 



+ 10-746,671^ - -315,428* 3 - -082,135^). 



We now deduce from this curve by putting x in succession 0, 0*5, 1, 1*5, ... 

 the frequencies on the half-unit subranges up to and including the frequency 45 

 at cloudiness 5. 



In precisely the same way we deduce from equations like (xv) and (xvi) an 

 auxiliary curve at the other terminal of the distribution. There we find q = "44, 

 and the following table for cloudiness in twenty half-units arises : 



