The work of Trescot on potassium nitrate is reported separately, as follows:

Impurities in reagents, reckoned on the nitrogen as the basis of percentage, were reported as follows by two analysts only:

Percentage of impurities in reagents based on nitrogen present.

Carpenter, reporting the work of Cooke, says:

The Kjeldahl method, or in fact any of the methods described in the "Official Methods of Analysis," are very unsatisfactory for the determination of nitrates. In our own work for the analysis of nitrate of soda, nitrate of potash, etc., we use an electrolytic method.

Robb says:

I noticed a very perceptible loss of nitric oxid when I added the salicylic-acid mixture both in the case of the nitric-acid solutions containing a nitrate-free fertilizer and those containing cane sugar.

Rudnick says:

Inasmuch as all of the work could not be carried out in full, such features as seemed to me of less interest from the standpoint of a fertilizer manufacturer's laboratory were omitted. These had to do chiefly with the determination of nitrate nitrogen per se. We are not so much interested in this feature, inasmuch as there are several good methods now in use for this purpose. Of these we prefer the Schloesing-Wagner method, as described in Bulletin 107 of the Bureau of Chemistry, page 111.

It is not the determination of nitrate nitrogen that we are so vitally interested in but the determination of total nitrogen where nitrates are present.

The results with the fifth-normal nitric acid were very unsatisfactory indeed. This was probably due to the great dilution with water, which we never meet with in our work. Every precaution was taken to prevent undue heating on adding the strong salicyl-sulphonic acid.

A great many more determinations than appear in the preceding report were made in this laboratory, but none of them was any more satisfactory than those reported on, and there seemed to be no special significance in the results obtained. The Ulsch method as given in Bulletin 107 with a number of variations has been tried in this laboratory at various times during the past years, but has never given satisfactory

results. The zinc-iron method works very smoothly, and some of the results are quite as satisfactory as those obtained by the Schloesing-Wagner method.

But, as stated above, we are mostly interested in the Gunning and Kjeldahl methods for total nitrogen in the presence of nitrates. These methods have not given satisfactory results in the past. There are some details in the manipulation of these methods not described in Bulletin 107 which seem to tend materially to a closer approximation of the true amount of nitrate nitrogen, such as quick covering of the sample with enough salicyl-sulphonic acid to absorb all of the evolved nitrogen oxids, sufficient time for thorough nitration of the acid, thorough cooling of the mixture before adding the thiosulphate, and heating to a free boil before adding the mercuric oxid or the potassium sulphate, respectively. We find also that the tendency to foam is very largely obviated by heating the flask containing the mixture in a boiling water bath for an hour, or possibly less, prior to adding the thiosulphate.

In view of the poor results obtained with nitric acid it seemed that the use of a pure nitrate weighed directly into the digestion flask might prove much more satisfactory for determining the accuracy of the various methods. Our work following this plan has given much better results than by using nitric acid as the source of the nitrate.

In the large majority of cases we have to do with mixed fertilizers containing a nitrate. Any inaccuracies due to the water present in nitric acid or in the solution of a nitrate could easily be obviated by evaporating to dryness in the digestion flask, preferably in vacuo, after neutralizing when necessary to prevent the loss of nitric acid.

But even in the determination of an added nitrate the results leave much to be desired. This is not only the case in this particular work, but has been a generally recognized fact among fertilizer chemists for a long time. If the secret of success lies in certain details of manipulation not commonly known, which make the method reliable, these details should certainly be mentioned in the description of the method. Trescot says:

I inclose results on neutralizing with soda and evaporating down, also on mixtures of potassium nitrate and blood, and potassium nitrate and bone by the Gunning modified method for nitrates and the Ulsch-Street method. These results only confirm my previous work on such materials with these methods. The zinc-iron method is a failure in my hands. I never could get concordant results. If there is anything more I can do, let me know and I will be glad to do it; only I do not wish to repeat my work on the Ulsch-Street and Gunning modified methods, for after the most careful checking for many years I am convinced that if handled properly, and on dry materials, both methods will give all the nitrate present.

DISCUSSION OF RESULTS.

The opinions quoted and results reported show wide variations. By methods (c) or (d) with and without organic matter, Robertson and Trescot alone get satisfactory results; those of the other chemists are low, mostly impossible, in fact. This may be due to heat generated by the acid and water, as many chemists think and as Rudnick's and Trescot's work on potassium nitrate seems to show. On the other hand, Robertson's work shows no appreciable loss. An important question is, Must not this possible loss from heat be reckoned with even with a comparatively dry substance? Even with potassium nitrate Rudnick falls far short of acceptable accuracy. In view of these facts may not a too speedy application of heat in methods (c) and (d) cause loss of nitrate vapors? It is a practice of some careful analysts to allow nitrate samples to stand several hours after the thiosulphate or zinc dust is added. Hence, recommendations are offered fixing a minimum time limit in methods (c) and (d). Method (g) as carried out by Bailey, Morrison, and Trescot gives good results; as carried out by the other analysts uncertain and varying results, mostly far too low. It must be noted, however, that to test the completeness of the reduction of nitric acid to ammonia the distillation is made at first off magnesia and completed off soda, except in Trescot's work, the amount of ammonia from each distillation being estimated separately and the sum taken. It appears further from this process that the distillation from magnesia is usually far from complete, giving in several instances less than one-fifth of the total amount of ammonia; that is, less than one-fifth of what

should have been obtained. As method (g) permits the use of magnesia only, no soda, in distilling, it is evident how very inaccurately this method is practiced by some chemists. It would seem that the magnesia distillation process is conducted by many chemists less successfully than any other analytical process.

Furthermore, the work of Bailey and Morrison shows that by method (g) some ammonia is obtained from cotton-seed meal, about 2 or 3 per cent of the total nitrogen in the substance coming off as ammonia, when the distillation is off magnesia, and about three times as much in addition, when soda is used in the distillation. This is fortunately a compensating error and contributes something toward making up the deficit just described due to distilling ammonia off magnesia.

As practiced by some chemists the process of distilling off magnesia is worse than useless. As method (g) does not affect the amount of total nitrogen, errors are not so serious, however, as in methods (c) and (d).

Method (h) shows better agreement among different chemists and also gives figures somewhat approximating theory.

The only report on method (f) is to the effect that excessive frothing prevented distillation.

Of the 54 answers to the 14 questions stated above, 39, or nearly three-fourths, must be considered unfavorable, either as failing to reach the 98 per cent standard of accuracy or as showing reactions contrary to the plan and purpose of the method.

It must be admitted, whatever may be the inherent accuracy of the several methods here discussed, that most of them in the hands of some experienced chemists fail to give reasonably reliable and accurate results. It must also be admitted that these methods with more or less variation in detail are commonly accepted by chemists as reliable, and, as this report shows, may be made to give exceptionally accurate results. Obviously methods so firmly established, and by some analysts so successfully employed, are hardly to be condemned, or even seriously questioned, without a study of them by a large number of chemists. The data obtainable for this report are too meager to justify any criticism or proposed radical change.

The referee is still of the opinion that the methods here discussed are fairly accurate when properly followed, less accurate perhaps in the hands of some analysts than methods applied to simpler determinations, such as of nitrate-free nitrogen, but yet as accurate as the difficulty of the case permits; and furthermore, that unquestionably these methods are not always successfully followed, as evidenced by the criticisms of chemists as well as by the reported results, and that probably some analyses made according to these methods are erroneous, giving too low results; that the complaint of the officers of the National Fertilizer Association may possibly be based on fact in certain cases, probably due in part to erroneous analyses and in part to actual loss of nitric acid, but that it is not within the power of this association at present to remedy the evil complained of, if it exists.

RECOMMENDATIONS.

Two recommendations of 1907, referred to the referee for 1908, are recommended for adoption as official. (Nos. 2 and 4, Circular 38, page 1, or Bulletin 116, page 129.) These changes relate to the use of copper sulphate in the Kjeldahl and Gunning methods. For detailed statement of changes see page 183.

Recommendation 3: Bulletin 107 Rev., page 8, fourth line from top, after the word "time" insert: "Allow the flask to stand without heat for not less than six hours." Recommendation 4: Same reference, page 8, under (d) (3) “determination," fifth line of paragraph, after word "and" insert: "Allow the flask to stand without heat for not less than six hours; then". So changed the sentence beginning with "Add 5 grams" would read: 'Add 5 grams of sodium thiosulphate and allow the flask to stand without heat for not less than six hours; then heat the solution for five minutes; cool; add 10 grams," etc.

66

At the close of the reading of the nitrogen report, the president announced the following committees:

Committee on amendments to the constitution: J. P. Street, J. T. Willard, P. F. Trowbridge.

Committee on nominations: R. J. Davidson, C. H. Jones, B. B.

Ross.

Committee on resolutions: L. L. Van Slyke, A. J. Patten, V. K. Chestnut.

REPORT ON INORGANIC PLANT CONSTITUENTS.

By H. D. HASKINS, Referee.

The work on inorganic plant constituents has been along lines recommended by the referee of the preceding year, particularly with reference to the development of a method for the determination of iron and aluminum in ash. The sample which has served for the work was prepared by thoroughly mixing the ash of a species of Lycopodium, known to contain a large proportion of aluminum, with a finely ground and incinerated sample of wood ashes, the latter being known to contain considerable quantities of iron.

PROPOSED METHODS.

The method proposed for study contains some of the features incorporated in the official method of determining ferric and aluminic oxids and phosphates in soils (Bul. 107, p. 15). See also Bul. 56, Proceedings of the Fifteenth Annual Convention of the association in which recommendations are made by Hartwell in regard to a method described in Crooke's Select Methods of Chemical Analysis. The method as outlined for the work this year was in detail as follows, a hydrochloric acid solution of the ash being used:

SEPARATION OF FERRIC AND ALUMINIC OXIDS IN ASH ANALYSIS.

Use a solution corresponding to 0.2 gram of ash. After removing the phosphoric acid the filtrate from the precipitate of ammonium phosphomolybdate, consisting of a nitric acid solution of molybdic acid, ferric oxid, alumina, lime, and magnesia, is placed in a beaker and cautiously neutralized with ammonia, care being taken that the temperature does not rise above 40° C. and that the alkali is added only in slight excess; allow to stand in a warm place until the precipitate completely settles, filter the clear supernatant fluid, wash the precipitate with hot water by decantation, then transfer it to the filter, and finish the washing. Next, redissolve the precipitate through the filter in weak, hot nitric acid (1 to 5), reprecipitate with ammonia, filter, and wash in the same careful manner. The precipitate is dried, ignited, and weighed as ferric oxid and alumina.

METHOD (b).-The weighed precipitate of ferric oxid and alumina is dissolved, on the hot water bath, in a covered flask by the addition of about 20 cc of dilute sulphuric acid (1 part sulphuric acid to 4 parts water). The iron is reduced to the ferrous state by adding iron-free metallic zinc (about 5 decigrams at each addition) until the solution is completely decolorized and the iron is all reduced; cool by immersing in cold water, dilute with cold distilled water which has been recently boiled, pour off and wash into beaker, leaving behind any residue of zinc. Titrate with standard permanganate solution.

METHOD (c).-An aliquot part of the original solution A, corresponding to 0.2 gram of ash, is evaporated on hot water bath with the addition of 10 cc of sulphuric acid until all hydrochloric acid is expelled; dilute with water, reduce with zinc, and estimate iron by standard solution of potassium permanganate. The per cent of ferric oxid obtained is deducted from the per cent of ferric oxid and alumina, corrections being made for filter ash, to obtain the per cent of alumina.

MAKING AND STANDARDIZING PERMANGANATE SOLUTION.

Dissolve 2.82 grams of pure crystallized potassium permanganate in distilled water by the aid of heat; cool and dilute to 1 liter and preserve in stoppered flask. Standardize this solution by titration with metallic iron solution as directed in the second American edition of Fresenius's Quantitative Chemical Analysis, pages 268–269.

A copy of the method as outlined above, together with the ash sample, was sent to eight chemists who had signified their intention of cooperating in this work. The results received from five analysts are given in the tabulated statement which follows: Cooperative work on ash sample.

J. A. Le Clerc found that it was necessary to redigest the ash residue with hydrochloric acid in order to remove all of the iron present. In Method C some organic matter seemed to interfere with the end point of permanganate titration.

Andrew J. Patten used an approximately hundredth-normal solution of permanganate solution, this being preferred on account of the small amount of iron present. O. M. Shedd found objection to the determination of iron by Method B in that it takes a long time to dissolve the oxids in the sulphuric acid solution, and at this point recommended the fusion of the oxids with potassium hydrogen sulphate.

The referee is of the opinion that a weaker solution of permanganate is preferable to the one recommended.

The average results obtained by the two methods agree very closely and indicate but little choice in method of procedure.

The variations obtained between the various chemists in the results reported may be due to the method of standardizing the permanganate solution. The referee has used for this purpose a solution made from iron of a known composition furnished by the Bureau of Standards, Washington, D. C. The personal equation must also enter into cooperative work of this nature to a greater or less extent.

RECOMMENDATIONS.

In view of the fact that the two methods gave results agreeing within 0.01 of a per cent, the referee feels justified in recommending the one which involves the least manipulation. The following is therefore recommended as an official method for the separation of iron and aluminum in inorganic plant constituents:

Use an aliquot part of solution A corresponding to 0.2 to 0.5 gram of ash for the determination. After removing the phosphoric acid, place the filtrate from the precipitate of phosphomolybdate, consisting of the nitric acid solution of molybdic acid, ferric oxid, alumina, lime, and magnesia, in a beaker and cautiously neutralize with ammonia, care being taken that the temperature does not rise above 40° C., and that the alkali is added only in slight excess; allow to stand in a warm place until the precipitate completely settles. Filter the clear supernatant fluid, wash the precipitate a couple