THE MICROBIOLOGY OF THE ATMOSPHERE 



simple form is described by Gilbert (1950). Improved forms of bubbler 

 have been described by Wheeler et al. (i 941), by Moulton et a I. (1943) who 

 added an atomizer stage before the bubbler, and by duBuy et al. (1945). 



(iv) The Rosebury-Henderson capillary impinger (Plate 4) differs 

 somewhat in principle from the bubblers. A small flask carries a wide 

 inlet tube, and the inner end of the tube is fused to a short length of 

 capillary tubing which dips at least 5 mm. below the liquid but remains 

 at least 4 mm. above the bottom. The diameter of the capillary tubing 

 acts as a limiting orifice, controlling the air-flow rate under suction (for 

 example at 2-3 litres per min., or even up to 20 litres per min.) (Rosebury, 

 1947). This impinger has been tested and found to have a high retention- 

 efficiency in work on experimental airborne infection of animals. If used 

 to sample clouds containing large particles, however, substantial wall-loss 

 may occur at the bend of the inlet tube. This can be prevented by adding 

 a pre-impinger, originally devised by K. R. May & Druett (1953) to 

 collect particles over 8 /x in diameter. 



(v) Centrifugal samplers. Centrifugal impaction is used in the Wells 

 (1933) air centrifuge, in which 30 to 50 litres per min. of air are drawn 

 through an agar-lined glass cylinder rotating at 3,500 to 4,000 r.p.m. ; 

 this makes the air rotate and throws all suspended particles onto the 

 wall of the cylinder by centrifugal force. After exposure the cylinder is 

 incubated, and colonies developing on the agar-coated walls give a direct 

 volumetric reading of the organisms present — so far as these are cultivable 

 on the medium used. This apparatus has been used very extensively 

 for routine bacterial sampling (e.g. Pincus & Stern, 1937; Wells, 1955). 

 Its disadvantages include the difficulty of examining the catch inside the 

 curved tube, and a deposition efficiency falling from 100 per cent with 

 2-3 fji particles to 50 per cent with 0-77 /x particles (Phelps & Buchbinder, 

 1 941). Wells has also illustrated a model of the air centrifuge for collecting 

 pollen for microscopic identification {see Committee on Apparatus in 

 Aerobiology, 1941). 



A more elaborate development is the 'conifuge', designed for micro- 

 scopic study of size-distribution of particulate clouds (Sawyer & Walton, 

 1950; Green & Lane, 1957). It appears to be ideal for comparing terminal 

 velocities of spherical with irregularly-shaped particles, but has the 

 disadvantage that it samples only 25 cc. of air per minute. 



Different in appearance, but essentially similar in principle, are the 

 cyclone dust-collectors extensively used in industry for removing dust 

 from air, and which have the advantage of allowing a large through-put 

 with small pressure-drop. Small cyclone dust-collectors have been used 

 by Tervet (1950), Tervet & Cherry (1950), and Ogawa & English (1955), 

 to collect large quantities of rust spores for inoculation experiments, and 

 we have found them satisfactory for the much smaller spores of Ustilago. 

 (For the design of cyclone dust-collectors, see C. N. Davies, 1952). 



(vi) hipaction samplers. Tmpactor' samplers use suction from a pump 



98 



