Report on the Bacteriology of Water. 323 



and to clouds passing, &c., &c., it seemed utterly hopeless to expect 

 the accurately comparative results required. 



Taking all these drawbacks into consideration, the Table B never- 

 theless shows some significant and instructive facts. 



The experiment denoted la, for instance, shows that, even in March, 

 the solar action can be detected clearly after so short an exposure as 

 half an hour to an hour, while exposures of 1^ to 2 hours resulted in 

 sharp, clear figures. 



After a large number of these comparative trials, however, I con- 

 cluded (1) that while it seems impossible to overcome all the diffi- 

 culties, and to express the nature of the exposure in words, the 

 general impression gathered was that on certain bright, sunny days 

 in the spring days when the sky is blue and cloudless, and the air 

 peculiarly clear, the bactericidal power of the direct or reflected solar 

 rays is very great much greater than has been supposed. A very 

 slight amount of haze makes a vast difference in the times of expo- 

 sure (e.g., Cases A (1), and B (1), where a much better result was 

 obtained in two hours in the one case than with three hours in the 

 other) ; (2) that very long exposures are necessary if the sky is over- 

 cast with clouds, even though the light is otherwise bright. In other 

 words, the direct rays of the sun are needed for the purpose of rapid 

 action ; (3) with solar light, direct from the sun, very little if any 

 difference can be detected between exposures where the rays fall 

 directly on the plate, and where they are once reflected from a thin 

 plane glass mirror silvered at the back.* 



Summed up in the shortest terms, the conditions of exposure are 

 practically the same as those required in ordinary photography, the 

 chief difference being that the duration of the exposures amounts 

 roughly to hours or half hours in the cases under consideration, 

 instead of minutes or seconds, as in quick plate photography. All 

 this, of course, points to the blue end of the spectrum as the effective 

 one, a conclusion which is abundantly justified, and, in fact, fully 

 proved in the sequel, and by my experiments with the spectrum since 

 publish ed.f 



111. 



In the following sets of exposures (Table C), I employed quartz, 

 instead of glass, as a covering to the Petri's plates, so that, except in 

 the cases '12A, 12B, and 12C, the light traversed no glass before 

 reaching the spore-laden film. 



The results show little additional information, excepting that in the 

 cases 12D and 12E it was interesting to find that the action was 

 approximately as pronounced after one hour of exposure as after 



* This remark refer? particularly to this kind of exposure, 

 t ' Proc. Eoy. Soc.,' rol. 54, p. 472 (Abstract). 



