98 



ment, all the controls were the same for the entire experiment. These 

 control data points were averaged later and used for comparison to 

 each chemical independently (Figures 25-33; Table 24). These graphic 

 comparisons (Figures 25-33) demonstrate the large differences between 

 the control and treatment data. 



ANOVA on the control treatments in the 1975 test (Table 25) gave 

 a high degree of significance (.01 level) for the following sources: 

 blocks (b), treatments (T), weeks (W), BxW, TxB, TxW, TxBxW. Differences 

 between blocks were expected. For this reason the blocks were incor- 

 porated in the test. The difference between the treatments simply 

 meant that all the chemicals were not controlling the same. The SSD 

 with the treatment by weeks interaction meant that a Duncan's new 

 multiple range test on the chemicals had to be run on a week to week 

 basis. The Duncan's test gave an ordered comparison between treatments 

 and denoted the statistical significant differences (SSD), if any. 



ANOVA for block, treatment, and TxT for each week demonstrated 

 that only weeks 2, 3, and 6 were significantly different (Table 26). 

 Week 1 gave total control by all chemicals tested. Weeks 2 and 3 

 proved to have SSD for the treatment. Weeks 4 and 5 should have had 

 varying treatment control, but this was not evident. Clos analysis 

 of the data (Figures 25-33) indicates that a natural population decline, 

 possibly by an epizootic by Hirsutella thompsonii , Fisher, drove the 

 populations back down, not allowing for any differences to develop. 

 During week 6, even with this epizootic-like condition, significant 

 differences did exist within populations of mites on the chemically 

 treated treatments. In many cases the treated populations exceeded 

 the control populations. 



