41 



discuss these questions here. All the examiners whose work is 

 under consideration were specially trained for this work, and not 

 employed in it until I was personally satisfied of their compe- 

 tence. They all used similar methods, and may, I think, be 

 trusted to have exercised due care and diligence.) The number 

 of examinations made per case in the new series is roughly as 

 follows : 



In the Birmingham series (101 cases), all but three cases were 

 examined thrice each. At Brighton (35 cases), most cases had 

 only one examination, but a few two or three. At Bristol, about 

 a half of the 227 cases studied had three examinations apiece, and 

 the rest one or two. The Leeds and Sheffield cases (338 and 168 

 cases respectively) mostly received one examination each, but a' 

 number received two or more. The 120 cases at Reading were 

 examined only once each. It is thus evident that the majority 

 of the cases in the whole series received but one examination 

 apiece ; though a significant proportion received two or three, and 

 a negligible number more than three. 



Now from figures which I have published elsewhere (1917), 

 from those recorded by the Liverpool workers and others, and 

 from those which I have accumulated since, it is clear that if a.ny 

 series of cases is examined only once apiece, the number of 

 infections found is very considerably less than that which actually 

 existed. The errors due to deficient examination are not easily 

 determinable with precision ; but it is probable that, if the series 

 examined is sufficiently large, they are approximately of the 

 following order of magnitude : A series of cases examined once 

 apiece may he expected to disclose about one-third of the infections 

 actually present. The number of infections found will, at all 

 events, almost certainly be less than one-half of the real number. 

 The error for a series examined thrice per case is more easily 

 determined : and it is probable that, in such a series, between 

 one-half and two-thirds of the actual number of infections will 

 have been detected. 



From these considerations it is clear that the percentages 

 given in the last table — since they are based, for the most part, 

 on series examined only once per case — must be multiplied by at 

 least two, or even by as much as three, in order to obtain the true 

 incidence of infection. To multiply by two would almost certainly 

 give too low a value, since a relatively small number of cases 

 received more than one examination ; but if we multiply by three, 

 we shall run the risk of obtaining too high a figure, since this 

 would be the factor appropriate to a series examined throughout 

 only once apiece. It is not at present possible to deal with findings 

 of this sort with great mathematical accuracy, but it is probably 

 safe to assume that the true figures lie between the limits which 

 can be approximately determined in this manner. I believe, 

 therefore, that the following figures give a fair estimate of tha 

 true incidence of infection with the various protozoa, as indi- 

 cated by the minimal actual findings exhibited in the preceding 

 table : 



