142 
A Studji of Trypanosome Strains 
It will be seen that the following strains, T. rhodesiense, T. brucei, T. gamhiensei 
the ilf^^imta, and wild game, give either reasonable or excellent results as conabined 
frequencies of T. minus and T. majus. On the other hand the G. morsitans and 
the human strains break up into reasonable pairs of components, but the goodness 
of fit test is not fulfilled. In the case of the human strain, we better matters 
somewhat by taking the strain through the rat only, but the fit is still bad. If we 
confine our attention to a single human being, the case of Chituluka, we still do 
not get a satisfactory fit, although few statisticians could look at the four diagrams 
published by Sir David Bruce and others for Chituluka*, and not recognise the 
character of the material as being at least bimodal. The same applies to the 
Mkanyanga data of an earlier paperf, it is distinctly bimodal. But besides this 
bimodal character there are certain other features in the human data, and to a 
lesser extent in the G. morsitans, which appear to some extent to disguise the 
bimodal features. I am not prepared to assert definitely that this is the appearance 
of a third component. It is of course easy to improve the fit of the distribution 
by the introduction of such a third component, but the remarkable excellence 
of a bimodal resolution for T. rhodesiense, T. gambiense, and the wild-game strain 
makes me hesitate at present to adopt such an expedient. 
Owing to the courtesy of Sir David Bruce (who heard from Sir John Rose 
Bradford tliat I was much puzzled over the differentiation of strains) I have been able 
to examine a series of drawings of the various strains of trypanosomes. There is 
no other morphological differentiation which impresses itself a priori on the layman 
and statistician, and which might serve as a new measure of the possibility of differen- 
tiation into T. minus and T. majus. But it occurs to me that an index of breadth to 
length of the nucleus might just possibly serve as a differential character of even 
more importance than the lengtli. It is only a suggestion and considerable caution 
would have to be used in selecting only nuclei not near the dividing stage. But 
it would be of striking interest to see how far the resulting frequency distributions 
for the nuclear indices were or were not bimodal. I think a classification according 
to nuclear index might possibly— to judge from the drawings — cut across the 
forms " intermediate " in length. But this is only a suggestion which may appear 
idle to the student of the subjectj. Some difficulty might also arise from the 
doubt as to whether the index was really greater than 100, or the nucleus as 
a whole had set itself athwart the "length" of the trypanosome. This difficulty 
would certainly have to be considered in the " stumpy " T. brucei and T. gambiense 
■■■ R. S. Proc. Vol. 86, B, pp. 291 to 293. 
t R. S. Proc. Vol. 85, B, p. 428. 
t Several students of the subject with whom I discussed the matter stated that they considered the 
nucleus so mobile and so impermanent in form, that a "nuclear index" would prove of little value. 
I think much objection could a priori be raised to the use of the trypanosome "length" on the same 
grounds. The problem is rather, whether in dealing with large numbers we do reach an average type. 
It would only be possible a posteriori to justify the use of a nuclear index, i.e. if it were found to differ 
sensibly from one pure strain to a second, and if it confirmed in such cases as T. rhodesiense resolutions 
based on length frequencies. 
