68 



random sampling of organisms in some medium, or insect and mite counts 

 in field plots (Steel and Torrie, 1960). For this and all Poisson val- 

 ues in these studies the transformation of Y = (X' + .5)^ was used 

 (Table 7). 



The derived quadratic equation (Table 8) was used to give a repre- 

 sentation of the effects the snail populations had on the citrus rust 

 mites (Table 9; Figure 24). Control areas (X=0), after day one, dis- 

 played a constant value through the experiment. The reason for the 

 apparent initial decline in mite population was due to the variability 

 in average surface area throughout the experiment. The surface areas 

 of the nine cubic feet of test area (X-i) were averaged as were the 

 averages of all the untransformed initial mite counts (X ). This al- 

 lowed for comparison of the various cubic units. Quadratic equations, 

 having basis in all of these factors, are affected by any averaging 

 (Figure 24). Thus by averaging X, and X values, a decline in population 

 values is depicted. The average of these two values, X-, and Xp,,were 

 combined with the intercept value to yield the control values. 



Y (x = 0) = Intercept - .003 X, + .0260 X, 

 (n = 1-6) ' 2 



Y (x - 0) = 2.8613 - .5122 + .2503 

 (n = 1-6) 



Y (x = 0) = 2.599 (transformed data) 

 (n = 1-6) 



To untransform the data to original state, the following 

 formula was used. 



X' = Y^ - .5 



X' (x = 0) = 6.2548 (Table 7) 

 (n = 1-6) 



Test areas using one snail per cubic foot (X = 1) displayed a 



sharp decline in citrus rust mite populations for the first four days 



