52 A MANUAL OF BACTEEIOLOGY 



from the water and replace the stopper, wipe the exterior of the 

 bottle dry, and replace the paper cap. Instead of this bottle an 

 Esmarch or other type of sampler may be employed. 



3. Samples for bacteriological examination should be used as 

 soon as possible. If they are to be kept more than an hour, they 

 should be packed in ice. 



4. Samples from pumps should be collected after fifteen min- 

 utes of continuous pumping. Samples from a faucet should be 

 taken only after the water has run freely for five to ten minutes. 



Exercise 74. Quantitative Study of the Bacteria in Surface Waters 



1. Obtain samples of water from three different sources, fol- 

 lowing directions given in Exercise 73. 



2. After shaking the sample at least 25 times, remove 1 cc. of 

 water with a sterile pipette and place it in the bottom of a 

 sterilized Petri dish. If the sample is suspected of being highly 

 infected, the water should be diluted 1 : 10 or 1 : 100, using ster- 

 ile water blanks for the purpose. Plates ought not to contain 

 over 200 colonies. 



3. Pour into each dish a tube of melted gelatin (not warmer 

 than 43° C). Tilt the plate gently to mix the water and gelatin. 

 Place on a level until the gelatin has solidified, and incubate at 

 a temperature of 22° C. or lower. 



4. In the same way make plates, using beef-peptone agar and 

 Heyden-Nahrstoff agar. These plates may be incubated at tem- 

 peratures up to 37° C. 



5. Count the colonies on plates incubated at 22° after seventy- 

 two hours, and on plates incubated at 37° after forty-eight hours, 

 using a counting plate. If possible to do so, all the colonies on 

 the plate should be counted. If there are more than 400, it is 

 easier and fully as accurate to count a fractional part of the 

 plate and estimate from it the total number. It is customary 

 in practice to make plates in duplicate or triplicate, thus afford- 

 ing a check for one's own work. In the best of hands the limit 

 of experimental error is large. With very careful work a varia- 

 tion of 10 per cent may be obtained, but a smaller variation is 

 not to be expected. 



