Jan. 3, 1921 



Correlation and Causation 



583 



to a considerable extent (d E . = o.i6i) by outstanding factors. The addi- 

 tional term in this case would have this value and when added to 0.839 

 gives 1, as it should. With one exception, the calculated correlation 

 between transpiration and evaporation is a little smaller than the actual 

 correlation. This means either that there is some additional factor 

 which should be allowed for or else that the path coefficients with W, T, 

 R, and H are not given quite their due weight, owing perhaps to lack of 

 complete linearity in the correlations. 



Table II. — Table of calculated path coefficients 



Wet-bulb depression 



Evaporation (shallow tank) 

 Transpiration: 



Small grains 



Rye 



Sorghum and millet .... 



Sudan grass (inelosure) . 



Sudan grass (open) 



Dent corn 



Algerian corn 



Cowpea and lupine 



Alfalfa 



Amaranthus 



Average transpiration 



/>x. 



o. 298 

 •395 



238 



2 00 

 234 



539 



339 



297 

 349 

 35 J 



3°3 



052 

 -79 



Tempera- 

 ture. 



pn.T 



0.896 

 •544 



Radia- 

 tion. 



/>X.R 



.249 



. 207 



■ 203 



• 130 

 •°59 



■ 109 



• 194 

 .214 



■ 117 



• 105 

 .181 



Absolute 

 humidity. 



-o. 811 



" -437 



Correlation with 

 evaporation. 



Calcu- 

 lated. 



o. 830 

 (■ 839) 



826 

 852 



741 



Actual. 



0.83 

 1. 00 



87 



91 



713 



93 



82 



79 



85 



77S 



705 



560 



781 



Table III. — Coefficients of determination 



Wind 



Tem- 

 pera- 

 ture. 



rfx-i 



Radi 

 ation 



Abso- 

 lute 

 hu- 

 mid- 

 ity. 



Joint determination. 



</x.i 



Residual. 



</x-c 



Wet-bulb depression 



Evaporation 



Transpiration: 



Small grain , 



Rye 



Sorghum and millet 



Sudan (inelosure) . . , 



Sudan (open) 



Dent corn 



Algerian corn 



Cowpea and lupine. . 



Alfalfa 



Amaranthus , 



Average transpiration . . 



o. 0S9 

 156 



OS 7 



044 

 055 

 290 

 "5 

 088 

 122 

 123 

 092 

 oo_< 



07S 



0.803 



. 296 



. 607 



. 728 



.516 



■757 

 .861 

 .664 

 .724 

 • S°4 

 ■364 

 •3M 

 •537 



o. 156 



. 062 

 •043 



■ 041 



■ 017 



• 003 

 . 012 



• 038 

 . 046 



• 014 

 . on 

 •033 



0.657 

 . 191 



. 240 

 •34° 

 ■ 177 

 •047 

 . 141 

 . 164 

 • 153 

 . 120 

 . 180 

 .183 

 . 176 



OK) 



01 8 



013 



004 

 003 



007 



Ol 2 



on 



OOS 

 007 



(- 



•125 

 .038 

 •293 



. 062) 

 .O96 



■237 

 .057 

 ■ 247 



•474 

 . 607 

 .277 



The coefficients of determination are given in Table III. The differ- 

 ence between their sum and unity is given in the last column as d x . 0> 

 the determination by outstanding factors. As suggested above, the 

 assumption that all the fundamental correlations are linear may involve 



