SUiMMER DIARRHOEA 157 



consequently his figures do not necessarily represent the fly 

 population in houses. Nevertheless his curves representing 

 diarrhoea fatality and fly prevalence closely resemble those of 

 Niven. Hamer, Peters (1910) and other critics have pointed out 

 that if variations in the number of fly transmitters of an infective 

 agent existing in the stools of infants were the sole factors con- 

 cerned, the curve for diarrhoea cases should shoot up beyond and 

 come down later than that representing the number of flies, 

 because the opportunities afforded to flies to pick up the infective 

 agent increase with the development of the epidemic. This, 

 however, does not happen. On the contrary the epidemic is 

 arrested while flies are numerous, and declines more quickly 

 than the fly population. In this exhaustion of susceptible 

 material probably plays some part. The decrease in the activity 

 of the flies owing to Empusa disease in the autumn, and to colder 

 weather also probably has a very important effect. Possibly the 

 fall in the temperatures has a decided effect in checking the multi- 

 plication of the infecting agent as Martin (1913, p. 5) suggests. 



Peters (1910, p. 717), who most carefully studied epidemic 

 diarrhoea in the town of Mansfield, but who did not make 

 accurate estimations of the numbers of flies in houses, gives 

 the following summary of the evidence specially favouring fly 

 carriage. " ( i ) The low level of the winter cases — in the absence 

 of flies. (2) The facts suggesting that the house, and not the 

 individual, is the centre of infection. (3) The sudden outbreak 

 in a ' clump,' supervening upon solitary preceding cases, suggests 

 that flies have suddenly gained access to infection. (4) The fact 

 that within the ' clump,' infection appears to some extent to rain 

 down equally upon all persons and houses included, suggests 

 systematic dissemination by flies. (5) Variations of prevalence 

 with variations of temperature. (6) The almost identical tem- 

 perature limitations of fly and diarrhoea prevalences, as regards 

 their rise and fall : the significant immobility of the diarrhoea 

 curve till the first fortnight of favourable fly temperature has 

 passed by. (7) The correlation of the fly and diarrhoea curves 

 is such as to be quite compatible with a theory of causal connec- 

 tion between flies and diarrhoea. (8) The large amount of 

 evidence for personal origin of infection is mostly also evidence 



