July 15, 191S Methods of Bacterial Analyses of Air 353 



In the first series of tests reported in Table 11 the joints between the 

 two filter tubes were not sealed to prevent leakage, but were carefully 

 examined to discover possible leakage. As the results show (see tests 

 2, 9, and 10), some leakage probably did occur through the cork 

 stoppers which were used to connect the tulaes. For this reason, in the 

 second series of tests all joints which could possibly leak were sealed 

 with paraffin after the apparatus had been sterilized. 



The first series of tests (Table II) gave an average of 217 colonies per 

 plate for the first aeroscopes of the tandem pairs in the modified form 

 of aeroscope, with 12 per plate (5.7 per cent) for the second aeroscopes. 

 In comparison with this, the first tubes of the selected (p. 349) stand- 

 ard aeroscopes gave an average of 175 colonies per plate, with 29 colonies 

 per plate (13.2 per cent) for the second aeroscopes of the tandem pairs. 



Twenty-five comparative tests in the second series (Table II) under 

 the rigid conditions just described gave an average of 400 colonies 

 per plate for the first aeroscope of the modified form and 3 per plate 

 (0.75 per cent) for the second. The selected standard aeroscopes gave 

 an average of 384 colonies per plate for the first aeroscopes and 42 per 

 plate (about 10 per cent) for the second ones. Thus, while the standard 

 aeroscopes allowed an appreciable number of bacteria to pass through 

 into the second tube, the modified form probably did not allow any 

 bacteria to pass through into the second tube, the small numbers of 

 colonies occurring on the plates being scarcely more than the numbers 

 that appeared on the check plates. 



The second series (Table II) shows conclusively that the standard 

 aeroscopes allow bacteria to pass through the sand layer, since all other 

 leakage was cut off by the paraffin seal. It is probable that a more 

 careful selection of those standard aeroscopes in which the layer of 

 sand was least affected by the steam sterilization would have resulted 

 in a higher average efficiency than was found in these tests. On the 

 other hand, if no selection had been made, the results would undoubt- 

 edly have shown a much lower average efficiency. The mere fact that 

 such a selection is necessary in order to secure reliable results is of itself 

 a serious weakness in the standard procedure. 



The idea that the low numbers on the plates of the second aeroscope 

 in the train of the modified standard aeroscope really represent contami- 

 nations is further supported by the fact that an examination was made 

 of the cotton plugs used to support the sand filter in the modified aero- 

 scope. In 25 of the tests, after the sand in the first tube had been 

 poured out, the upper end of the tube that had been in contact with 

 this sand was carefully flamed and the cotton plug blown out into a 

 test tube containing 10 c. c. of sterile water. After a thorough shaking, 

 this bacterial suspension was plated in duplicate in i c. c. quantities. 

 In the entire series the average number of colonies that developed per 

 cotton plug was only 50 (5.0 colonies per plate), the highest number 



