The colourless liquid is allowed to cool thoroughly, and the flask is then fitted with a tapped funnel and tube, as shown in Fig. 18. This tube must not be less than 4 mm. in diameter, and with the end in the flask cut diagonally to facilitate drops of liquid falling back again into the flask. It rises obliquely for a height of 12 to 15 inches, is then bent over as shown in the figure and connected by a rubber tube[49] to a 100 c.c. pipette, or similarly shaped tube, the other end of which dips just below the surface of a volume of exactly 50 c.c. of “normal” hydrochloric acid contained in a second flask. About 50 c.c. of distilled water is introduced into the flask containing the treated sample, and after this 100 c.c. of a solution of 50 grams of caustic soda in 100 c.c. of water is carefully and slowly run into the flask by means of the tapped funnel with which it is provided. The contents of the flask are now boiled for about half an hour,[50] the normal acid in the receiving flask being kept cool by immersing the latter in cold water. The liquid in this second flask is then titrated with normal sodium carbonate, using methyl orange as indicator. The difference in c.c. between 50 c.c., the volume of acid used, and the quantity of normal sodium carbonate required to neutralise it, when multiplied by 0·014 represents the amount of nitrogen (in grams) in the weight of the substance used for the determination; or if multiplied by 0·0786 shows the weight of hide-fibre in the same quantity of material. Some chemists add copper sulphate, or a drop of mercury before boiling up the substance with the strong sulphuric acid, but the use of such substances introduces complications in the process without, in the case of gelatinous matter, securing more accurate results. It is absolutely necessary that the acids and alkali used should be free from ammonia, and a blank experiment should be made using pure sugar which contains no nitrogen, and a correction applied if necessary for the ammonia they contain.[51] [49] The ends of the glass tubes should fit closely together, so as to expose the rubber as little as possible to the action of ammoniacal vapour. [50] “Bumping” is often very troublesome at this stage, and may be prevented by passing a current of steam from another flask, or ammonia-free air through a tube with a capillary opening into the boiling liquid; fragments of pure zinc, of platinum, or broken tobacco-pipe are much less efficient. It is an additional safeguard against the escape of ammonia to fix a small absorption-tube containing fragments of glass to the absorption-flask. The normal acid is run through this tube into the flask, so as to wet the broken glass, and is finally rinsed into the absorption flask before titrating its contents.