22 



the results being given in parts per million, this means that we aim at 

 estimating one part of nitrogen in one billion parts of water, or less than 

 one-ten-thousandth ot a grain per gallon. In order to give you some 

 idea of how this is done I place in one of these tubes a column of 24 

 inches of water quite free from ammonia, and in another I place an 

 equal quantity of water to which I have added ammonia in the propor- 

 tion of one part nitrogen to one million parts of water. The two sam- 

 ples as reflected to you from the mirrors are of course quite mdistin- 

 guishable from each other. To each I now add a small quantity of 

 a prepared test liquid called Nessler's solution, and you will observe in 

 the course of a minute or two that while the contents of the first tube 

 are unchanged in colour, a faint brownish yellow colour gradually 

 developes itself in the second tube. Of course it is i)0ssible in the 

 laboratory to apply this test in such a way as to obtain still greater 

 sensitiveness, but the illustration will serve to give you confidence in 

 numerical statements of the results of analysis even when fractional 

 parts of a million are expressed. 



8. While the simpler organic bodies containing nitrogen yield this 

 nitrogen as ammonia during decomposition, many of the more complex 

 substances which enter into the composition of animal structure, such 

 as albumen, fibrin, etc., form other proximate products of decay, these 

 possessing the common property of being converted into ammonia when 

 boiled with a strongly alkaline solution of permanganate of potash. The 

 ammonia obtained by treating a sample in this way, after the ammonia 

 already })resent in it has been taken off, is called " Albuminoid " am 

 monia, as suggested by Wanklyn, the author of the process, and is pro- 

 perly Considered as a most important factor in the analysis. Indeed, if 

 it were ever allowable to adjudge a sample of water for drinking pur- 

 poses upon the indications of a single factor in the analysis I would 

 select this estimation as the critical one. The author of the process, 

 who in conjunction with other analysts, worked upon a very large 

 number of samples of all degrees of badness, concludes from his experi- 

 ence that "o. 10 per million begins to be a very suspicious sign, and 

 0.15 per million ought to condemn a water absolutely." This standard 

 would go hard with Ottawa river water, which in 1888 gave from 0.12 

 to 0.27 in different samples ; in March and April of last year gave 0.15 



