18 



After the eggs were laid in the autumn of 1910 all localities 

 were carefully inspected, and the number of the fresh clusters 

 was determined as nearly as possible. For examination of the 

 single eggs, five clusters, when possible, were collected from each 

 locality. Five clusters from a place in the western part of 

 Concord, Mass., served for control. These eggs must be con- 

 sidered as normal because they come from a place in which 

 positively no disease whatever has occurred. In collecting 

 clusters for examination the first five found were taken; and 

 were not taken at random. 



Every single egg of all these clusters was examined micro- 

 scopically. This, it is true, required very much time, but it 

 was nevertheless indispensable. The eggs were separated into 

 three kinds : first, those that showed no signs of development of 

 the embryo (empty or unfertilized eggs) ; second, those in 

 which the more or less developed embryo had been killed before 

 its complete development; and finally, those which showed a 

 perfectly developed embryo (normal and supposedly living 

 eggs). In every case in which the egg could not be absolutely 

 assigned to the second series it was counted with the third 

 series. 1 



In the details of all the work everything which seemed to 

 suggest the economic value of the wilt has been purposely re- 

 garded from the most unfavorable standpoint, for the purpose 

 of meeting any objections. When estimating the egg clusters 

 from the selected localities there was always present either the 

 division agent or the local superintendent, and at that time 

 written notes were made of these estimates before the intro- 

 duction of the disease. In the spring we made the lowest 

 possible estimate on the number of clusters present, while 

 with the second estimate in the autumn, after the disease had 

 produced, its effect, the fresh clusters were estimated as liberally 

 as possible. 



The average which resulted from the examination of the 

 single clusters must not be taken as absolutely fixed, since the 

 small number of the clusters examined admits of a variation 



1 The eggs of the control clusters could all be distributed with exactitude into the respective 



