1905.] with reference to the Purification of Sewage. 



245 



starch-zinc-iodide solution, and with metaphenylene-diamine, and estimated by the use of 

 the latter. When both nitrites and nitrates were present, they were estimated together by 

 the indigo method {Tii'mann-Giirtner's " Wasseranalyse ") and the figure for nitrates 

 obtained by subtraction. This method was afterwards given up in favour of the Copper- 

 mine couple method,* where, to allow for. traces of ammonia originally present in the 

 solution, or introduced during analysis, a control estimation was always made ; this control 

 was carried out in every way like the real analysis, except that no couple was introduced, 

 and the ammonia obtained was subtracted from that found in the actual estimation. 



Total nitrogen was estimated in the sewage by Kjeldahl's method, a small amount of 

 Na. : S0 4 only being added during the preliminary heating. 



The oj idisability (" Oxidirbarkeit," or measure of oxidisable substances present) was 

 estimated by reduction of permanganate in alkaline solution on boiling for ten minutes, 

 care being taken to keep the external conditions (such as concentration, size of flask, total 

 amount of liquid present, temperature, and time of reaction) constant in all determina- 

 tions. Periodic examination of sewage and filtrate in this way gave useful comparative 

 results and a means of following the course of the general oxidation. 



Course of General Oxidation During Maturation: First Appearance oj 

 Oxidised Nitrogen, — -The filters were started on February 20, 1901, and their 

 action carefully controlled by means of analysis from that time onwards. 

 Especial attention was paid to the period of maturation from the time of 

 first using to that of full efficiency, as it was thought this should throw 

 light generally upon the manner of their working. The sewage employed 

 contained on an average : — total nitrogen 10 parts, organic matter (by 

 evaporation and ignition) 21 parts; oxidisability (expressed in terms of 0) 

 about 11 parts per 100,000 by weight. The course of the oxidation will 

 be seen by reference to Tables Ia, Ib and II. On March 7 (see Table II, 

 analyses 1 to 4) there was a marked amount of general oxidation taking 

 place in all three filters, or at least a reduction of oxidisable substances, but 

 there was no trace of oxidised nitrogen in the filtrates nor was there any 

 diminution in ammonia. Oxidised nitrogen first appeared in the tall filter 

 on March 18 (the filtrate being then clear, bright and without smell), three 

 days later it was detected in the medium filter and seven days later in the 

 short filter. During this first period (after starting and before the occurrence 

 of nitrification) the amount of ammonia in the sewage was frequently 

 compared with that in the filtrates and found to be always the same.f On 

 March 7 this was proved to be so for all three filtrates, and even on 

 March 28, when the first trace of oxidised nitrogen appeared in the filtrate 

 from the short filter there was no diminution of the free and saline ammonia 

 coming through {cf. analyses 10 and 13). So, generally, Tables II and III 



* Sutton, 'Vol. Analysis,' 8th ed., p. -15i. 



t As the sewage varied considerably in composition from time to time, care was always 

 taken that the sewage and filtrate analysed for comparison should correspond to one another 

 as nearly as possible. 



