16 



cases, and produced diarrhoea when fed to monkeys and 

 young rats. It was rarely encountered in children not 

 suffering from acute diarrhoea. 



Morgan's bacillus has been found in flies captured from 

 houses in which cases of diarrhoea occurred, but until its 

 causal relation to the human disease is proven too much 

 significance cannot be attributed to these observations. 



We are, therefore, entirely dependent upon epidemic- 

 logical facts in testing whether the belief in the dominant 

 influence of the house fly in the occurrence of epidemics of 

 infantile diarrhoea, a view which has been warmly espoused 

 by many observers, is valid. 



THE RELATIONS IN TIME BETWEEN CASES OF DIARRHOEA 

 AND PREVALENCE OF FLIES. 



Unlike the case of typhoid fever which has just been dis- 

 cussed, the epidemics of infantile diarrhoea in towns exhibit a 

 close time relation with fly prevalence. Epidemic diarrhoea 

 of children does not occur except during that season of the 

 year when flies are abundant and active, and, as will be seen 

 on studying the charts, the relation between fly population 

 and cases during the summer is so striking as to suggest 

 something more than a mere accidental dependence upon 

 the same phenomena. 



The charts, Figs. 7, 8, 9, 10, and 11, show the relation of 

 diarrhoea and fly prevalence in time. Figs. 7, 8, and 10 are 

 derived from Dr. Hamer's observations in London in 1907, 

 1908, and 1909. Figs. 9 and 11 are constructed from Dr. 

 Niven's (1910) observations in Manchester in 1904 and 1906. 

 I have selected these particular years from the Manchester 

 data because they represent what happened in warm summers 

 in which the number of cases was considerable. 



The charts have undergone some manipulation at my 

 hands. I have antedated the deaths by ten days, which, 

 according to Dr. Niven's observations, represents the average 

 period intervening between onset and death. They ought, 

 no doubt, to be antedated further to include the incubation 

 period, but I have no data to determine what this should be. 

 In the case of Dr. Hamer's observations I have summed his 

 tri-weekly observations and plotted the fly catch for the 

 whole week. In this way I have eliminated most of the 

 irregularities and obtained smoother curves. I have also added 

 the weekly mean temperature for London for the period 

 comprised by Dr. Hamer's observations, and in Fig. 8 for 

 about two months previously. 



The procedure adopted by Dr. Niven and Dr. Hamer to 

 obtain data regarding fly prevalence was common in all 

 cases. The number of flies caught on sticky paper or 

 entrapped in a series of domestic dwellings, generally in 

 neighbourhoods inhabited by poor people, were recorded 



