The estimation of free and albuminoid ammonia is made as follows:--100 c.c. of the water to be examined are introduced into a glass retort, which connects with a Liebig’s condenser, and has previously been thoroughly cleansed by boiling with distilled water; one gramme of pure sodium carbonate is added, and the water distilled until 40 c.c. have passed over, the distillate being separately collected in four 10 c.c. cylinders or tubes. About 10 c.c. of the alkaline solution of potassium permanganate is then added to the remaining contents of the retort, and the distillation continued almost to dryness. The second distillate is likewise collected in fractions of 10 c.c. each. It is advisable to so regulate the process of distillation, that only about 10 c.c. pass over in the space of eight minutes. The two sets of distillates are then separately tested by adding 0·5 c.c. of the Nessler solution to each cylinder, well stirring the mixture, and setting it aside for at least five minutes. A series of comparison tubes (10 c.c. in capacity) are prepared by adding ·001, ·003, ·005 up to ·01 gramme of ammonium chloride, and filling to the 10 c.c. mark with pure distilled water; 0·5 c.c. of the Nessler reagent being added to each. The degree of coloration exhibited in the cylinders containing the two sets of distillates is then matched by the comparison cylinders. It is evident, that from the data thus obtained, the amount of ammonia obtained by the first distillation with sodium carbonate (free ammonia), and by the second distillation with alkaline potassium permanganate (albuminoid ammonia), can be determined. It has been previously mentioned that urea evolves ammonia when boiled with sodium carbonate; the amount of ammonia obtained by the first process of distillation will therefore include that actually contained as such in the water, and that generated by the decomposition of any urea possibly present. As the presence of this body is incompatible with a good drinking water, this fact is of little real importance. In case, however, it be desired to make an estimation of the free ammonia really present, 500 c.c. of the water to be tested are treated with 1 or 2 c.c. of calcium chloride solution, then with a slight excess of potassium hydroxide, and the liquid filtered. It is next distilled as directed above, and the remaining contents of the retort made up to 500 c.c. 200 c.c. of the original sample are then subjected to the same treatment with calcium chloride and potassium hydroxide, and filtered. The second solution, which contains all the ammonia originally present in the water, is now tested with Nessler’s reagent, the solution first obtained by diluting the contents of the retort being employed, instead of pure distilled water, for comparison.