1905-6.] Studies in Immunity : Theory of an Epidemic. 497 
20), is also given. The fit here is not good, hut the statistics of 
London are not nearly so definitely homogeneous. As, however, 
the divergence of the actual from the theoretical is, though much 
greater, of the same nature as that observed in the divergence of 
Diagram Xili. 
the actual and theoretical curve for Glasgow, it is possible that 
some other factor affects both. 
So far, no exceptions have been noted to the form of epidemic 
distribution given by type IY. These have occurred in my work 
only in the case of some small epidemics of which it was difficult 
to give the beginning or the end, and in the case of outbreaks of 
endemic diseases where the tail of one epidemic runs into the 
beginning of another, — cases difficult to treat, as will presently be 
PROC. ROY. SOC. EDIN. — VOL. XXVI. 32 
