ARNETH'S NUCLEAR CLASSIFICATION. 407 



of sepsis. Solis-Cohen and Strickler were not able to find any shift to the 

 left in tuberculous patients at any stage of the disease although the poly- 

 morphonuclear cells were reduced in numbers. (6) Arneth considers a shift to 

 the left an evidence of lowered resistance to the disease affecting the patient 

 whose blood he is studying, as is indicated by the fact that when a tubercular 

 patient improves the blood picture tends to return to the normal. (13) 



OUR METHOD OF STAINING AND COUNTING. 



Most of our counts were made with very thin smears stained with 

 Wright's stain. A few were stained with hematoxylin and eosin after 

 the method of Bushnell and Treuholtz. This method gave excellent 

 pictures but did not seem to offer sufficient advantages to compensate for 

 the trouble of making a second stain for the differential count. Whatever 

 staining method is used, it is important that smears be thin. We have 

 tried the technique of Wiedenreich, but without success. (16) 



Even in a well-spread and well-stained slide there will always be 

 an uncertainty as to the group in which some nuclei should l)e classified. 

 Pottenger considers that these doubtful cells constitute 12 per cent of 

 the total when Wright's stain is used. To obtain uniform results, we 

 adopted the following rules in counting: 



( 1 ) Nuclear masses connected by a distinct isthmus were counted as 1 

 nucleus, while masses connected by only a thread were counted as 2 nuclei. 



(2) Nuclear masses clearly superimposed were considered as separate nuclei, 

 but where the superposition was not distinct the body was considered a single 

 nucleus. 



(3) In all instances when there Avas doubt as to whether a cell should be 

 grouped in one or the other of two classes, it was always placed in the higher 

 class. 



By a careful adherence to rule 3 we have avoided the possibility of 

 producing an artificial shift to the left. If all doubtful cells had been 

 recorded alternately, the first in a lower and the second in a higher class, 

 the shift to the left which we found for bloods in the Tropics would have 

 been even more pronounced than we now claim. 



In our work we enumerated for differential counts 200 cells for each individual. 

 For the Arneth work we counted 200 cells each from a part of the men and 

 100 each from the remainder. While an enumeration of 100 cells for Arneth 

 work and of 200 cells for differential counts may in a few instances lead to 

 slight errors for the individual, it is believed that it is ample to give accurate 

 average results such as we were seeking in our investigation. 



The following table indicates how close an agreement in the results 

 for both differential and Arneth counts was obtained by our tAvo in- 

 dependent observers working at different times. 



