236 



percentages of injured plants. In the case of the wheat sawfly {Cephus 

 ? pygmaeus) it was found that samples containing 200-300 stems gave 

 quite unreliable results, and that the probable error is fairly high even 

 in samples containing 1,000 stems ; so that in order to obtain an 

 approximately reliable standard, still larger samples must be taken. 

 The attempt to arrive at an index of infestation in the case of Oscinella 

 frit was still less satisfactory, for the inequality of the samples was 

 even more pronounced. Even when each sample comprised 1,000 

 plants, it was found that the percentage of infestation in one sample 

 might be nearly double that in another. 



A further point which had to be considered was the method of 

 taking the samples. At first, the stems or plants were selected at 

 random throughout the field, but this was abandoned as unsatisfactory^ 

 Then limited areas of 5| square feet were investigated in detail ; 

 but this system also proved unreliable, principally owing to the 

 uneven distribution of the pests. The method finally adopted was to 

 select a stem or plant at every alternate pace up and down the plot ; 

 but the author is by no means satisfied that even this will give sufficiently 

 consistent results. 



Another question, which naturally arose during the work, was that 

 of studying the influence of the injury on the plant. This is dealt 

 with in a paper by Andreieva and the author (see above p. 230). The 

 loss due to insects can be determined in two ways : — Statistically, by 

 examining a sufficient number of plants, both injured and uninjured, 

 taken from the field ; and experimentally, by placing the plants in 

 quite similar conditions and then infesting them artificially. Up to 

 the present the statistical method only has been used. The author's 

 observations have shown that the majority of insects prefer for oviposi- 

 tion strong plants, rather than weak ones ; so that if the effect of insect 

 injury is tested merely by comparing the weights of attacked and 

 immune plants, the apparently absurd conclusion may be arrived at, 

 that the average weight of the infested plants is higher than that of 

 healthy ones, and that the infestation by an insect thus apparently 

 increases the harvest. In so far as such pests as the wheat sawfly 

 are concerned, it is not difficult to correct the error, by selecting for 

 comparison plants which are approximately of the same size ; for the 

 sawfly does not attack the wheat until the plants have already attained 

 their full growth. In such circumstances it is quite possible to deter- 

 mine the loss by the statistical method. 



In the case of Oscinella frit the matter is different. It was not 

 found possible to select plants from the fields between which a reliable 

 comparison could be made. The whole subject will have to be examined 

 under definite artificial conditions. 



It is not only very desirable, but very necessary, to investigate 

 more closely the injury itself from an anatomical and physiological 

 point of view, as at present little or nothing is really known. The 

 chief outcome of such work will be the determination, within the limits 

 of such accurate experiments as may be possible, of the influence of 

 agricultural methods on the infestation of the plants by insects. The 

 questions themselves and the methods of investigation are sufficiently 

 diverse and every worker will perforce become an innovator. 



During the discussion, A. Sopotzko suggested that when studying 

 the influence of pests on plants, it is necessary to take into consideration 



