July 25, 1872] 



NATURE 



245 



may exist in combination as calcic carbonate, and during 

 the evaporation the remaining sulphurous acid reduces 

 and expels the nitrous and nitric acids present, in which 

 the ferric chloride greatly helps it ; the sulphuric acid 

 which is thus formed is neutralised by the calcic sulphite 

 (formed during the preliminary expulsion of carbonic 

 arid) with liberation of fresh sulphurous acid, or should 

 no calcic salt be present, the sodic sulphite is added to 

 effect this. 



As soon as the whole of the water is evaporated, the 

 residue is carefully detached from the dish by means of a 

 flexible steel spatula, and thoroughly mixed with some 

 fine cupric oxide ; it is then introduced into a stout piece 

 of combustion tube about 430 millimetres long, one end 

 of which has been closed, after having been very care- 

 fully cleansed, and which has about 30 millimetres of its 

 length filled up with coarsely granulated cupric oxide.* 

 After the mixed residue and cupric oxide have been 

 introduced, the tube is filled up to within 100 millimetres 

 of the open end with granular cupric oxide ; a tightly- 

 rolled cylinder of copper gauze covered with sheet copper, 

 and which has been recently ignited and cooled in a cur- 

 rent of hydrogen, is then put in, lastly a few millimetres 

 moi'e of granular cupric oxide, and the tube is then drawn 

 out in the blowpipe flame, put into a combustion furnace, 

 connected with a sprengel pump, and, while the anterior 

 part of the tube is being heated to redness, thoroughly 

 exh.austed of air. The pump is then stopped, and the 

 heat gradually carried backwards until the whole of the 

 combustion tube has been heated to redness. Any gas 

 which may come off is collected from the bottom of the 

 Sprengel pump over mercury, and as soon as the gas 

 ceases to be evolved, the furnace is allowed to cool slightly, 

 and the tube again exhausted. The gas pumped off is 

 transferred to an apparatus for the analysis of gases, and 

 measured, after the absence of sulphurous anhydride has 

 been insured by the introduction into the gas of a drop of 

 a solution of dipotassic dichromate. The carbonic anhy- 

 dride is then absorbed by potassic hydrate, and the gas 

 again measured, the difference being the carbonic anhy- 

 dride. To the residual gas a minute bubble of oxygen is 

 added to decompose the nitric oxide, should any be present, 

 and after the excess of oxygen has been absorbed by pyro- 

 gallic acid, the gas again measured consists of nitrogen ; to 

 this half the difference between the two last readings is 

 added, as this represents the nitrogen which had existed 

 as nitric oxide, and the result is the total nitrogen. 



All the above data are reduced to measures of weight 

 •0012562 



by the use of the formula. Log 



\— — - which 



(i + 0036 7/ ,7 60 

 gives the weight of cubic centimetres of nitrogen in 

 grams. This table of logarithms is carried out for each 

 tenth of a degree from 0° to 3o°C. 



To those uninitiated in gas analysis, the above may 

 sound very complicated ; but in practice it is found ex- 

 tremely simple. The whole analysis, including the calcu- 

 lations, can easily be carried out in half an hour, and the 

 combustion itself need not occupy more, as a rule, than 

 forty- five minutes. 



The value of the above process is at once perceived 

 when it is remembered that by it 00000005 grrmi. carbon 

 and o'o.~^ooi gram, nitrogen are distinctly measurable 

 quantities-. The methods of analysis involving the use of 

 gas-.metric measurements are by very far the most accu- 

 rate in the whole range of a science v.'hose very founda- 

 tions rest on the possibility of accurately weighing and 

 measuring varying quantities. The great capabilities of 

 this method of inquiry have received a splendid demon- 

 stration in the researches of Sir Benjamin Erodie on 

 Ozone ; and Profs. Williamson and Russell, recognising 

 the peculiar excellence of these methods, have endea- 

 voured with considerable success to make them applicable 



to a more extended range of work, in connection with 

 which it will only be necessary to remind the reader of the 

 careful determinations of the atomic weights of cobalt and 

 nickel made by the latter chemist. 



Notwithstanding these advantages, the application of 

 this method to the determination of the organic consti- 

 tuents of water has encountered the most strenuous oppo- 

 sition in many quarters. 



But until a more absolutely certain method of determining 

 the quantities of carbon and nitrogen shall have proved 

 Frankland and Armstrong's process to be in error, it would, 

 indeed, be the height of folly to adopt in its place a method 

 which, like the Albumenoid process, only professes to give a 

 fraction of the nitrogen present (and has absolutely failed to 

 prove that it even gives any known fraction whatever) ; 

 while it does not even attempt to estimate the carbon at 

 all. Again, over 60 per cent, of the errors given by 

 Frankland and Armstrong in their paper are minus errors, 

 and thus directly tend to favour the Albumenoid process 

 when analysis of waters by the two methods are compared ; 

 for the Albumenoid process almost invariably gives less 

 quantities of ammonia than Frankland and Armstrong's 

 does of nitrogen. It must be borne in mind that these 

 absolute errors are almost certainly due to errors in 

 weighing the small quantities of substances used to test 

 the process, quantities so small that they are admitted by 

 the objectors to be only about one-tenth of those usually 

 used in organic analysis. 



But in the case of water no weighing at all occurs. A 

 measured quantity of water is taken, and the quantity of 

 carbon and nitrogen which happen to exist in that 

 measure is ascertained, whether it be small or great. 



Again, it has been urged that ammonia is lost during 

 the evaporation. This was found to be the case ; but it 

 was a difficulty only requiring to be known in order to be 

 remedied. Accordingly, solution of salts of ammonia 

 made acid with sulphurous acid in one case, and with . 

 metaphosphoric acid in the other (the last acid replaces 

 the first when sewage is operated on), were evaporated and 

 the nitrogen detetmined, and from the loss found two 

 tables® constructed, in which each alternale term was an 

 absolute determination, and the intermediate ones cal- 

 culations. The wonderful constancy* with which the 

 numbers alter for each strength of solution shows how 

 accurate the determinations must have been in order to 

 obtain them. 



Another valuable proof of the trustworthiness of a 

 process is to be found in the degree of agreement exist- 

 ing between duplicate determinations made with it ; and 

 when the process is examined ia this way it certainly 

 passes the test in the most satisfactory manner, and the 

 following examples will show how well duplicate determi- 

 nations agree with each other : — 



Grand Junction Water Company 

 East LoHclou ,, ,, • 



New River ,, ,> ■ 



The Don .it Alford .... 



Lady Well Spring (Dundee) 



Indeed, the numbers given in the Registrar-General's 

 monthly reports exhibit so remarkable an agreement 

 among themselves that any unprejudiced judge must 

 admit the accuracy of the method by which they were 

 obtained. We shall consider the remainder of the subject 

 in a concluding article. 



'■ See Stilton's " Volumetric Analysis," p. »76, and Editioo. 



