July 3, 1908] 



SCIENCE 



31 



These experiments, and others of the same 

 sort, seemed to indicate that sand filtration 

 gives reasonably accurate results if the sand 

 used be as fine as .3 mm. The crucial test of 

 this point, however, must be made by drawing 

 a given sample of air through sand filters and 

 a culture bottle, so arranged in tandem that 

 the bacteria which pass the sand shall be col- 

 lected in the bottle. The table below shows a 

 series of such experiments and makes it clear 

 that the efiiciency of the filtration method 

 depends upon the size of sand grain employed. 



EELATIVE NUMBER OF BACTERIA PASSING THEOUGH 

 SAND FILTERS; AND RETAINED IN THEM 



In seven tests with tandem sand filters, each 

 containing 2.5 cm. of sand, with grains be- 

 tween .1 and .3 in diameter, the bacteria 

 passing the sand were — once 30 per cent, of 

 the number retained by the sand, twice 17 

 per cent., once 2 per cent, and three times 

 1 per cent, or less. On the other hand, in 

 three tests with the Sedgwick-Tucker appa- 

 ratus holding a single layer of sand, 2 cm. 

 deep with grains between .5 mm. and 1.0 mm., 

 nearly half the bacteria present passed the 

 sand in one case and about two thirds escaped 

 in the other two instances. 



It seems clear that sand over .5 mm. in 

 diameter is inadequate for filtering out bac- 

 teria. On the other hand, a sand finer than 

 .3 mm. is generally efficient though not 



wholly reliable, since at times it allows a 

 considerable proportion of bacteria to pass. 

 This is not remarkable when the relative size 

 of sand and bacteria is considered. 



It is, of course, obvious that sand can not 

 operate in the removal of bacteria by any 

 process which can properly be called strain- 

 ing. In an editorial discussion of the re- 

 moval of fine particles from water the Engi- 

 neering News (LIX., 344) has described the 

 phenomenon as " adhesion " ; and the term de- 

 serves general acceptance in this connection. 

 The size of the sand must affect the removal 

 of fine particles in two ways. First, in a 

 given depth, the number of surface contacts, 

 which permit adliesion, must vary inversely 

 with the size of the particles. Second, the 

 velocity of flow, which tends to tear off adher- 

 ing particles, must, under given conditions, 

 increase with the size of the particles. Coarse 

 sand might, therefore, bo used with success by 

 filtering through a deeper layer and by cutting 

 down the rate of flow. It is simpler, how- 

 ever, to use sand sufficiently fine to regulate 

 the rate of filtration automatically. 



On the whole, the culture-bottle method 

 seems to offer a more accurate procedure for 

 bacterial examination of air than any yet 

 available. The sand-filter method is fairly 

 accurate as a rule, but occasionally gives low 

 results. The filter method is more con- 

 venient than the culture bottle method for 

 examinations outside the laboratory, since for 

 the latter it is necessary to carry two 1,500 

 c.c. bottles for each examination. Aside from 

 this difficulty of transportation, however, the 

 technique of the culture-bottle method is to 

 be preferred. Bottles are easier to prepare 

 and to sterilize than sand filters and the 

 actual examination is simplified by the omis- 

 sion of sand washing and subsequent plating. 

 C.-E. A. WiNSLOW 



Massachusetts Institute of Technology 



tiOaiETIES AND AOADEMJEH 



THE NEW YORK ACADEMY OF SCIF;NCIi;S, SECTION 

 OP ASTRONOMY, PHYSICS AND CHEMISTRY 



A meeting of the Section of Astronomy, Physics 

 and Chemistry was held at the Museum of Nat- 

 ural Histoi-y on Monday, January 20, at 8 ; 15 p.m., 



