Let us suppose again that the compound was made with prime lard stearine, in which case we have the following computation :-

From the above computations the value of the refractive index in determining approximately the respective quantities of cotton oil and stearine in mixed lards is apparent. It is true that in individual cases the variation might be very much greater than indicated above, but as an expression of the mean result it appears to me that the refractive index is fully as valuable if not more so than the specific gravity in the quantitative determination of mixed lards.

I propose to push this investigation somewhat further by more extensive examinations of the specific gravities and refractive indices of lard, oleo-stearine, prime steam lard, and mixed lards.

The refractive index of pure water at 25° as indicated by the instrument employed (Abbe's large model) was 1.3300. When the index of water at the above temperature is taken at 1.3330, 0030 should be added to the numbers given in the above paper.

ON SOURCES OF ERROR IN DETERMINATION OF NITROGEN BY SOD ALIME, AND MEANS FOR AVOIDING THEM.

By W. O. ATWATER.
(Continued from page 59.)

From these results, as well as from consideration of what is known of dissociation in general, it is evidently impossible to predicate anything definite regarding the amount of decomposition that would take place at a given temperature in a glass tube containing soda-lime and such a complex mixture of gases as occur in an ordinary combustion. I regret the lack of facilities which prevented determinations of the temperature at which the combustions in these experiments were made; but so far as one can judge from observation, I should say that what I have here designated as "medium heat" or "usual heat," namely, that in which the combustion tube is dull red and the soda-lime yellowish, is about that at which combustions are very commonly made in the laboratories where I am acquainted.

Prehn and Hornberger have reported observations which imply large losses of nitrogen by oxidation in soda-lime combustions. When sugar was burned in the tube to expel the air before the combustion and to sweep out the ammonia afterwards, considerably more ammonia was obtained than where this precaution was omitted, and the apparent oxidation by the air present in the tube at the beginning was greater than that from the air used in the aspiration at the end. But the experiments were made with ammonium salts, and where no sugar was used there was apparently no gas to drive out the air and prevent its oxygen from acting on the ammonia. In ordinary combustion of nitrogenous materials large quantities of gases are produced, and in the experiments I have just described oxalic acid was used, which naturally drove out the

air.

On the other hand, numerous analysts find that with proper precautions there is no loss from oxidation. In our work we have taken pains to have the tubes well filled with soda-lime and substance so as to leave only very little open space for air, and to heat the anterior layer of soda-lime before applying the heat to the mixture of soda. lime and substance, thus expelling a large part of the air in advance, and have been careful not to aspirate with air until after putting out the flames, so that the contents of the tube have become somewhat cooled before the air has been admitted. In how far omission of these precautions may explain the apparent oxidation elsewhere observed, I, of course, cannot say. Naturally there would be the less danger of loss the more the air is removed by diluting gases before heating. But considerable experience has left me with the very strong impression that with the precautions I have suggested there need be very little fear of oxidation. This must be especially true where, as in ordinary combustions, large amounts of hydrocarbons are present to combine with the oxygen.

Makris has shown that oxidation may take place, but his experiments were made under conditions especially arranged to insure admixture of air with ammonia at a high temperature.*

Prehn and Hornberger have also experimented upon the effect of different degrees of heat on the ammonia in determinations with ammonium salts and potassium ferrocyanide. They find, at what they call ordinary heat or dull redness, no considerable indications of dissociation, but on heating to bright redness they found with tubes 35-40 cm. and the anterior layer of soda-lime 15 cm. long, considerable, and with tubes 55-65 cm. and the anterior layer 30-40 cm. still more indication of the dissociation of ammonia. Their results thus agree with those above given, except that they get more dissociation with the high heat in the long than in the short tubes. This they very reasonably explain by the fact that the ammonia in the long tubes had more time to dissociate, from which I infer that the gases were caused to pass more slowly than was the case in our work.

Makris also tested the dissociation of ammonia by passing a slow current of gas through a tube 70 cm. long filled with pieces of soda-lime and heated to bright redness. Analysis of the gases coming from the tube showed a very considerable amount of dissociation. But, as Gruber has observed, the case here was different from that in an ordinary combustion, in that a large amount of ammonia was exposed for a long time to

* Ann. Chem. (Liebeg) 184, 376.

a high heat and without any considerable amount of diluting gas, whereas in ordinary combustion there is less free space in the tube, and that largely filled with other gases, so that the ammonia is rapidly swept away, and furthermore the diluting gases are largely hydrocarbons which would naturally furnish nascent hydrogen to regenerate dissociated ammonia.

My results above detailed accord very exactly with those of Gruber and of Johnson and Jenkins above cited, who found no indications of loss by dissociation. But it is evident that both the complete transformation of nitrogen into ammonia, and the loss of ammonia, depend largely upon the manipulation. Thus Kissling gets very good results with an anterior layer 35 cm, long, and evident loss of nitrogen when the anterior layer is only from 7 to 10 cm. long, the combustion being carried on slowly.*

The experiments described by Mr. Ball and myself in a previous article bear upon the questions of dissociation of ammonia and of incomplete ammonification of nitrogen of distillation products. They were made with casein, the conditions being varied so as to allow in some cases very little, and in others considerable opportunity for the ammonia formed to be dissociated. The results, given in detail in the article referred to, are more concisely set forth in Table IV.

(To be continued.)

REPORT OF RECENT RESEARCHES AND IMPROVEMENTS IN
ANALYTICAL PROCESSES.

ESTIMATION OF GLYCERIN IN THE COMMERCIAL ARTICLE. F. FILSINGER. Zeitschr f. Angew Chemie, No. 1, 1889.-The acetin process recommended by Benedict and Cantor (see ANALYST, 1888) gives good results with the purer kinds of glycerin, but is unreliable for the titration of samples got from soap leys, as these retain impurities which affect the process. Better results would no doubt be obtained if the impurities were first got rid of, but this would involve dilution, and the process only works with fairly concentrated samples.

L. DE K.

TEST FOR ANTIFEBRIN IN PHENACETIN. M. J. SCHRÖDER. Nederl. Tydschr. v Pharmacie, etc. January, 1889. Phenacetin, when Phenacetin, when taken internally, yields phenetidin and para amido phenol, both harmless bodies, whilst antifebrin yields aniline, which is decidedly poisonous. It is therefore of importance to test for the presence of antifebrin in phenacetin. The author found the best test to be Plugge's reagent, which consists of a solution of mercurous nitrate with a little nitrous acid. 5 grm. of the sample is boiled in a test-tube with 8 c.c. of water, allowed to cool, and filtered off from the re-crystallized phenacetin. The filtrate is boiled with a little potassium nitrite and dilute nitric acid, then mixed with some of Plugge's reagent, and again boiled. If no red colour is got, the sample may be considered as practically pure; at all events, there cannot be more than 2 per cent. of antifebrin.

*Ztschr. anal. Chem, 24, 1885, 448.

L. DE K.

ZINC SALICYLATE. L. VAN ITALLIE. Nederl. Tydschr. v. Pharmacie, etc February, 1889.-The author tests the salt by incinerating the compound, moistening the ash with nitric acid, and finally igniting the residue, which should be not less than 21 per cent. A commercial sample only yielded 18 per cent. The author estimated the degree of solubility of this salt in various fluids. One part dissolves in 25-2 parts of water at 16° C. One part of the anhydrous salt dissolves in 36 parts of ether of 725 sp. gr. at 16° C., and in 450 parts of chloroform of 1.495 sp. gr. at 15° C. One part of the salt dissolves in 3.5 parts of spirits of wine, sp. gr. 819-4 at 15° C. In petroleum spirit it is quite insoluble. L. DE K.

MODIFICATION OF KJELDAHL'S NITROGEN PROCESS. J. W. GUNNING. Nederl. Tydschr. v. Pharmacie, etc., February, 1889.-The author operates as follows: One part of potassium sulphate is fused with two parts of sulphuric acid. This mixture gets semisolid in the cold, but it readily melts, and may then be treated like a fluid. About one gramme of the substance to be analysed is put into a 300 c.c. flask with round bottom and short neck. If liquids such as milk or beer have to be tested, a suitable quantity must first be evaporated to dryness in the flask itself. About 30 c.c. of the acid mixture are now added, and the whole heated with a Bunsen burner. At first strong frothing occurs and white fumes escape, consisting chiefly of water vapour. To prevent loss of strong acid, the neck of the flask is now fitted with a funnel, which is then covered with a watch-glass. This simple arrangement will now cause the acid to condense and run back into the flask. The operation is finished when the acid looks colourless, which will be generally the case after about an hour. After cooling, the ammonia is now estimated as usual. The author prefers the standardising of the volumetric acid by the iodine and sodium hyposulphite method. The test analyses are very satisfactory. Uric acid yielded 33.3 per cent. of nitrogen, theory requiring 33.33 per cent. yielded 10 per cent., theory requiring 10.1 per cent.

Aniline oxalate

L. DE K.

ASSAY OF CARBOLIC ACID. L. DE KONINGH. Zeitschr f. angew Chemie, No. 5.— The author, in reply to Mr. Williams, still believes the salt test to be a good one, both for the liquefied acid of the B.P. and hydrated cresylic acid. If the acid contains much tar oil, it cannot be in the hydrated state, but a little uncombined water may be shown by the benzol test, which, however, is not meant to be used for the purer forms of the acid. The percentage of tar oils is best estimated by agitating the sample with four times its volume of 10 per cent. soda ley if necessary, with addition of a fixed quantity of benzol. Williams' process for the estimation of free acid in carbolic powders is no doubt a most excellent one, but the author fails to see in what essential particular it differs from the one published in the ANALYST, vol. xii., by Dr. Muter and himself. To all intents and purposes it is just the same, though perhaps not quite so accurate, because not performed in specially constructed apparatus. W. H. D.

STANDARDISING Permanganate. R. JAHODE. Zeitschr f. angew Chemie, No. 4, 89. ---Many analysts still prefer to standardise their permanganate with pianoforte wire, but the trouble always is to exclude the air after the iron has dissolved. The author operates as follows: The iron is dissolved in boiling acid in a flask, which is closed by a cork,

through which goes a doubly-bent tube, the end of which is made to dip into a beaker containing a solution of sodium bicarbonate. When solution is complete and the liquid allowed to cool, the soda solution finds its way into the flask, but no sooner have a few drops got in, than an evolution of carbonic acid gas setting in drives the fluid back.

L. DE K.

CORRESPONDENCE.

[The Editor is not in any way responsible for opinions expressed by his correspondents.]
To the Editor of the ANALYST.

SIR,-In the number of the ANALYST for February last, there is a report of a paper by Mr. Rowland Williams on the "Determination of Citric Acid in Lemon Juice," brought before the Society of Public Analysts, "to promote," as Mr. Williams said, "a discussion among the members." I should have been glad to have taken part in such a discussion, had there been one, if Mr. Williams had, in pursuance of his object, kindly informed me that his paper was to be read, and that the greater part of it had reference to what he interpreted as being my method of "Estimating Citric Acid in Concentrated Lemon Juice," and in condemning the use of a solution of an alkaline carbonate for the titration. If Mr. Williams had informed me some months ago, when he asked for some information, that my very general reply was to form the foundation of a paper to be read before the Society, I might have described to him the precautions necessary to be observed in sufficient detail to have saved him from being misled into the belief that there is any objection to the use of alkaline carbonate for the purpose, or of running the risk of misleading others, beyond a little more time being required for the experiment than is necessary when caustic alkali is employed. There is absolutely no inaccuracy involved in use of alkaline carbonate, with litmus as an indicator, when the proper conditions are appreciated and observed,

I have no kind of intention of taking up the space of the journal with a criticism of Mr. Williams' paper as a whole, but I cannot help expressing my surprise at the figures given in the two tables which accompany it.

In the first of these are given comparative results of the analyses of six samples by the caustic method and by alkaline carbonate, showing, according to Mr. Williams, that the carbonate gives uniformly and exactly one ounce of citric acid more in a gallon than the caustic solution.

The only possible explanation of there being difference at all, is that there was interference by carbonic acid retained in the solution; but if Mr. Williams allowed carbonic acid to remain during the final testing, how is it there is anything like agreement in the results and that the differences are almost constant?

In the second table the results of the analyses of seven samples of pure citric acid by the two solutions are given, and the same astonishingly uniform differences are shown.

I cannot think Mr. Williams means to suggest that the alkali in an alkaline carbonate has less saturating power for citric acid than it has as it exists in the caustic condition, and yet this seems to be the only possible inference to be drawn from the tables themselves. My explanation, of course, is that these differences are really errors of manipulation.

I do not think I should have taken up the space of the journal with this letter if I not lately found that Mr. Williams' paper has been distributed widely in pamphlet form amongst manufacturers and dealers commercially interested in the sale of lemon juice, of whom some, at all events, are not in a position to estimate correctly the value of Mr. Williams' criticism of what he believes to be my method of analysis, but which, in fact, is only his mode of employing it. At the meeting at which the paper was read, there seems to have been no discussion, Mr. Allen only expressing what appears to me, to say the least of it, a very basty "regret" that "chemists of repute" should adopt an alkaline carbonate with litmus as an indicator for the purpose under discussion. Now, in reply to Mr. Williams and to Mr. Allen, I should have thought it might have been assumed by the Society of Public Analysts that chemists of repute, who had been engaged in a special analysis for five-and-twenty years, had carefully examined and tested every process that had been suggested, and under their sense of responsibility, and as the result of their long experience, had adopted the process, which in their hands gave the most reliable returns under all circumstances. Many hundreds of analyses have been made in my laboratory with caustic and with alkaline carbonate solution side by side, and observing the precautions necessary in each case, I, unlike Mr. Williams, obtain identical results, and I protest as strongly as possible against the inferences to be drawn from Mr. Williams' tables, as being opposed to all my experience. There are reasons why sodium or potassium carbonates are frequently to be preferred. After removal of the carbonic acid with properly prepared litmus paper of the proper tint, there is absolutely no objection to their use. As a matter of fact, I do not think Mr. Williams finds it easy to obtain caustic alkali free from carbonate.—I am, sir, yours truly,

39, Lime Street, London, E.C.,

26th March, 1889.

G. H. UPTON.