THE NITROGEN CYCLE 227 



of sewage at first, increasing the quantity as nitrification 

 becomes established. 



Dr. Chick found that in sewage filters, as in the experiments 

 Avith solutions, the nitrification took place in two well-defined 

 stages, first nitrites and then nitrates being formed. The 

 length of time required for complete nitrification to become 

 established depended on the amount of ammonia present, 

 either actually as ammonium carbonate, or potentially as un- 

 fermented organic matter in the sewage applied. Tempera- 

 ture also has a marked efiect, as might be expected, in deter- 

 mining the time necessary for nitrification to be established ; 

 for this reason it is advisable always if possible to bring new 

 sewage filters into work during the warmer months of the 

 year. 



Finally, mention may be made of the importance of the 

 character of the material used in the construction of the filter 

 beds. Practical experience has shown that better results are 

 obtained with a medium which offers a maximum of surface ; 

 thus irregular material, such as clinkers, gives better results than 

 when a smoother material, such as gravel, is used. Experi- 

 ments by the author and Percy G-aunt have shown that, in 

 addition to the efEect of surface in giving an extended habitat 

 for bacteria, the majority of vesicular or porous materials have 

 the power, to a greater or less extent, of retaining ammonium 

 salts, either in their smaller pores or in their larger interstices ; 

 such materials, therefore, afford a somewhat longer time for 

 the nitrifying action to take place when a solution containing 

 ammonium salts is brought in contact with them. 



The purely physical side of this question has also been 



carefully investigated by W. Chfford. He allowed known 



amounts of water to trickle through media of different kinds 



and dimensions at defined rates. When equilibrium was 



established between the rate of inflow and outflow, a known 



amount of sodium chloride solution of known strength was run 



on to the filter. The amoimt of chlorine emerging from the 



q2 



