Variance analysis of variance: The assumption of normality must also carry the assump- 

 tion that the variances of the means of the variables are homogeneous before parametric meth- 

 ods can be used. Like means with like variances are considered to be of one population; 

 like means with unlike variances are from different populations. 



The data on attack density and brood were converted to a common basis- -1 sq. ft.; 

 gallery lengths were converted to a 1/10 sq. ft. basis. 



Complete cross -comparisons of variances were made for attack density and gallery 

 length. The 10 trees in each d.b.h. category were considered as replicates (table 1). The 

 cross -comparison of variance for brood density was incomplete because the between-plot dif- 

 ference was not tested. The difference is obvious because the plots were selected on this 

 basis. The basis for replication for brood density was the 13 trees per plot. 



70 r 



_l L. 



6 8 10 12 14 16 18 20 



2 



Mean Inches Egg Gallery Per 1/10 Ft. 



Figure 4. --Teton plot 

 --gallery density: 

 X for 1/10 sq. ft. 

 rectangular sam- 

 ples, o for 1/2 sq. 

 ft. samples. 



+ 6 8 10 12 14 16^ 18 20 



Mean Inches Egg Gallery Per 1/10 Ft. 



Figure 5. --Wasatch 

 plot --gallery den- 

 sity: X for 1/10 sq. 

 ft. rectangular 

 samples, o for 1/2 

 sq. ft. samples. 



4 



