and the other, up to 1 kilo. in weight, is preserved in tightly-closed glass bottles in a cool place for a quarter of a year, reckoning from the day of despatching the report of the analysis, unless some other arrangement has been agreed on with the party sending the sample in question. 4. In crude phosphates and animal charcoal the moisture is determined at 105-110°, as a proof of identity. In samples which, during desiccation, may lose ammonia in any form whatever, this must also be determined. 5. Only average samples, carefully taken, and of at least 250-500 grms. and packed in well-fitting glass bottles, should be sent for analysis.

6. The weight of the samples sent in should be stated on the certificates of analysis.

7. In substances the moisture of which varies on pulverising the proportion of water must be determined both in the powdered and in the rough state, and the result of the analysis should be calculated according to the moisture of the original rough substance.

B.-Examination of Phosphatic Manures.

1. The extraction of superphosphates is effected by placing 20 grms. of the sample in a litre flask, pouring upon it 800 c.c. of water, and shaking it strongly and continually for half an hour, for which purpose a special machine has been invented, and may be set in action by hand or by any motor. About 150 rotations per minute are recommended. The flask is then filled up to the mark with water, the total liquid well shaken up and filtered. This process was introduced in the laboratories of the Association of Agricultural Experimental Stations on January 1st, 1891.

2. The solutions of double superphosphates must be boiled with nitric acid before the precipitation of the phosphoric acid, in order to convert any pyrophosphoric acid present into the tribasic acid. To 25 c.c. solution of superphosphate, 10 c.c. of concentrated nitric acid (14 specific gravity) are to be used.

3. In arbitration analyses the molybdenum method is to be used. The comparative determinations of superphosphate by the citrate method and the molybdenum methods as instigated by the commission must be made known at the next general assembly.

4. For determining the iron and alumina in crude phosphates the method of Glaser is for the present accepted as decisive :-5 grms. phosphate are dissolved in the known manner in 25 c.c. nitric acid of specific gravity 1.2, and in about 12.5 c.c. of hydrochloric acid of specific gravity 112, and made up to 500 c.c. 100 c.c. of filtrate, representing I grm. of the original substance, are placed in a litre flask, and 25 c.c. sulphuric acid of specific gravity 1.84 are added.

The flask is let stand for about 5 minutes and shaken for a few times, adding then 100 c.c. alcohol at 95 per cent, and cooling the flask, which is then filled up to the mark with alcohol and well shaken, when contraction takes place. The stopper is lifted, the flask again filled up to the mark with alcohol, and shaken again. After standing for half an hour it is filtered. 100 c.c. of the filtrate, representing o'4 grm. of the original substance, are evaporated down in a platinum dish until the alcohol is expelled. The liquid, free from alcohol, is mixed with about 50 c.c. of water in a beaker and heated to boiling. Ammonia is then added until the reaction is alkaline, but not during the boiling, to avoid a violent effervescence.

The excess of ammonia is driven off by boiling. The liquid is let cool, filtered, the precipitate washed with hot water, ignited, and weighed as ferric phosphate plus aluminium phosphate; half the weight found is assumed as consisting of Fe2O3+Al2O3. By this method the determination can be effected in from 1-2 hours.

5. The fine meal in basic slag meals is determined according to the method agreed on at Bonn. The operation is as follows:-50 grms. ground slag are placed in a sieve with a surface of not less than 20 c.m. in diameter, and made of the wire tissue, No. 100, by Amandus Kahl,

of Hamburg (smooth texture), and shaken for fifteen minutes by hand or in a suitable machine. 6. For determining the moisture in superphosphates 10 grms. of the sample are heated to 100° for three hours; the loss of weight is assumed as water.

7. For determining phosphoric acid in bone-meal, fish guano, meat manure, and crude phosphates, as also total phosphoric acid in superphosphates, 5 grms. are dissolved in 50 c.c. aqua regia, consisting of 3 parts hydrochloric acid of specific gravity 1'12, and I part nitric acid of specific gravity 125, or it is boiled with 20 c.c. nitric acid of specific gravity 142 and 50 c.c. sulphuric acid of specific gravity 18 for half an hour.

C.-Examination of Nitrogenous Manurial Substances. 1. Nitrogen in the form of blood, flesh-meal, and similar organic matters, can be determined either by the Kjeldahl's process or with soda-lime.

2. Nitric nitrogen in mixtures may be determined according to Schlösing, Grandeau, or Lunge. Total nitrogen is determined according to the KjeldahlJodlbauer or some similar process.

In Peruvian guano, whether crude or rendered soluble, the nitrogen, on account of the nitrates present must be determined by the Kjeldahl-Jodlbauer or by some analogous method.

3. For the determination of nitrogen in saltpetre a direct method should be sought for.

4. The total nitrogen in commercial ammonium salts should be determined by distillation with soda-lye.

Examination of Cattle-Foods.

The extraction of fats, which ought to be complete, is effected exclusively by means of ether free from alcohol and water. The weighed ethereal extract should dissolve in anhydrous ether without residue.

The commission is requested to explain the determination of fat in linseed cake by further experiments.

It was resolved to institute experiments concerning the degree of acidity of the extracted fat. It was recommended to use phenol-phthalein as indicator, and decinormal soda-lye as standard solution, as proposed by H. Fresenius, and to calculate the results as oleic acidC18H3402

282

1 c.c. decinormal lye=0'0282 grm. oleic acid. It was recommended to throw further light upon the value of the "iodine number" for the analysis of foods by its more frequent determination.

It was resolved to express the proportionate moneyvalue of proteine, fat, carbohydrates, by 3:2:1.

(To be concluded).

ON THE DETERMINATION OF NITRATES
IN WATER.*

By ALLEN HAZEN and HARRY W. CLARK.
(Concluded from p. 122).

II. Aluminum Process.

As usually described (Sutton's "Volumetric Analysis," Fifth Edition, p. 364), the aluminum process for the determination of nitrates consists of boiling a portion of the water with caustic soda, to expel ammonia, and after cooling, adding a piece of aluminum foil. The hydrogen evolved reduces the nitrates and nitrites present to ammonia. The hydrogen is taken through dilute acid to hold any ammonia that may be given off. After reduction is complete, the acid is washed into the water, and the ammonia distilled and nesslerised.

* Journal of Analytical and Applied Chemistry, June, 1891.

The process may be divided into two parts: the reduction to ammonia, and the estimation of the ammonia formed. For convenience we shall first consider the determination of the ammonia, and afterwards the conditions of complete reduction.

The amount of ammonia removed by the hydrogen is ordinarily so small that it seemed better to estimate its quantity and apply a correction, rather than use the troublesome absorption tubes of glass moistened with acid. The proportion of ammonia removed depends upon temperature and the amount of aluminum dissolved, Fifty c.c. of water saturated with ammonia at 20 degrees holds 26 grms. If the water contains o oo1 grm. of ammonia it is 1-26,000 saturated. One grm. of aluminum in dissolving gives off 1200 c.c. of hydrogen, and, supposing the ammonia to obey exactly the laws of gases, this should carry 1-26,000 of its volume of ammonia, or 0046 c.c., weighing 0.000035 grm. or 3'5 per cent of the whole amount of ammonia present. With different amounts of aluminum, and the same volume of water, the loss is directly proportional to it, being 1'75 per cent when grm. is dissolved. With different amounts of ammonia, the same percentage is removed. At higher temperatures the loss is greater, becoming 4.6 per cent at 30 degrees for 1 grm. of aluminum in 50 c.c. With reduced pressure the loss is greater, and is inversely proportional to the pressure. The loss from a nitrate solution is slightly less than from an ammonia solution, because there is no ammonia to be lost when the first hydrogen is given off.

Numerous experiments, keeping the acid separate from the bulk of the water, show that the above estimates are slightly above the truth, but they form a satisfactory basis for corrections, which practically never exceed 2 per cent. We have found that, with ground waters at least, distillation of the ammonia after reduction is unnecessary, quite as good results being obtained by diluting an aliquot portion and nesslerising direct. The conditions of success of this process are that the water after reduction shall be practically colourless, and free from the black residue of the aluminum, and that no alumina shall precipitate before nesslerisation. The reducing action of the nascent hydrogen, with the caustic alkali and the carbonate always present, clarifies a majority of waters sufficiently, and it is only with the yellow swamp waters with low nitrates, where large volumes must be taken for the determination, that the colour interferes with the result. Waters usually settle so clear after the action that suspended matters are not troublesome. The separation of alumina causes the greatest difficulty, clouding the tube, and often seriously lowering the readings. This can be prevented by using for dilution distilled water entirely free from carbonic acid. Such water can be easily prepared by blowing steam through ordinary good distilled water. This treatment will also remove every trace of ammonia. Care

must also be taken that the carbonic acid of the breath does not come in contact with the water while measuring with a pipette. When the aluminum-mercury couple suggested by Ormandy and Cowhen is used, direct nesslerisation cannot be applied, for the separated alumina invariably holds a portion of the ammonia, and the results are too low. With caustic soda, however, at the end of the reaction, everything is in solution, excepting a slight residue of iron, &c., from the aluminum, and the portion nesslerised contains its full share of ammonia.

To determine how much aluminum must be dissolved to completely reduce the nitrates, the following experiments were made. The reductions were made in 50 c.c. Nessler tubes, on account of the convenience with which large numbers of tubes can be handled in racks, and the ease with which they are cleaned. (See Table below). The amount of nitrite present at any point is comparatively small, and disappears promptly when the nitrate is all reduced, thus marking the end of the reaction. If we assume that the first centigrm. of aluminum in dissolving reduces 20 per cent of the nitrate, and each additional centigrm. 20 per cent of that remaining unreduced, we obtain the figures given in the last column. These agree fairly well with the observed reduction. The proportion of nitrate reduced by insufficient amounts of aluminum is almost exactly the same, whether the standard contains I or 10 parts nitrogen as nitrate in 100,000. With more dilute solutions, the proportion was nearly the same, but with stronger solutions the ratio is lower, although the absolute amount of nitrate reduced increases.

With higher temperatures, the solution of the aluminum is much more rapid and a smaller portion of the hydrogen is effective. At 30°, each centigrm. of aluminum in dissolving reduces only about 10 per cent of the unreduced nitrate, and so twice as much must be used to obtain a given reduction as is required at 20°. With temperatures lower than 20° the action is very slow, and a long time is required for enough aluminum to be dissolved.

If a large excess of caustic soda is used, the action is also very rapid, and the hydrogen is less effective, but the difference is not so marked as when the action is hurried by increased temperature.

The time required for the action depends upon the surface of the aluminum; the form of the metal has no other influence on the reduction. The same ultimate result is obtained with equal weights of aluminum, other conditions being the same, whatever its shape. With tin foil, the reaction is complete in a few hours, while with thick wire days may be required. When the tubes are allowed to stand over night at room temperature, foil o'005 inch thick is thin enough, and gives less trouble by floating than thinner foil. Jour. Chem. Soc., 1890, 811.

At one time we used a cheap caustic soda, containing zinc, because it was entirely free from nitrogen. It dissolved the aluminum freely, but for some unknown reason, the hydrogen evolved was less effective than that from other samples of caustic soda and potash. We thought this might be due to the zinc, but when zinc was added to pure caustic soda no corresponding effect was produced. The use of the soda was discontinued; but no explanation of its action has been found. The presence of sodium carbonate, even in large quantity, does not affect the reduction in any way.

NEWS

the amount of ammonia obtained is too small for accurate estimation. In such cases, 100 c.c. or more of the water is boiled in a flask with the caustic soda to 50 c.c. The reduction is then made as usual, after which the ammonia is distilled in a current of steam and nesslerised. If the water is concentrated in an evaporating dish, the results are high, owing to absorption of nitrous acid from the gas, but in a flask, the blank is not more than o'003 part, and is quite constant.

The following determinations of nitrates were made by one of us in solutions prepared by the other :—

Solution taken.

Direct reading.

The proportion reduced by insufficient aluminum is considerably greater than with potassium nitrate solutions in which the aluminum dissolves more rapidly. Care must be taken in such cases that enough aluminum dissolves. Conditions which are best adapted to the reduction of potassium nitrate often fail completely with waters, because in the time which suffices for the complete reduction of the standard, the reduction of waters with slow action is far from complete. As with the phenolsulphonic acid process, the fact of good results with standards is insufficient evidence of good results with waters.

The absence of nitrites, as shown by the sulphanilic acid and naphthylamine test, is the best evidence that reduction is complete. As long as a decided red is given by these reagents, we may be sure that reduction is incomplete, but when only a slight reaction or no reaction is given, reduction is complete, and no higher results can be obtained by dissolving more aluminum. To obtain this result with waters, in a moderate length of time, it is necessary to use twice as much caustic soda as is required by potassium nitrate standards.

To make the determinations of nitrates in waters, we proceed as follows:-A 50 c.c. Nessler tube is filled to the mark with the water, and about 0'4 grm. aluminum foil 0'005 inch thick added with 2 c.c. of a 40 per cent solution of caustic soda. After standing eighteen to twenty-four hours at room temperature, a portion, I to 25 c.c., depending on the amount of nitrate present, is taken out with a pipette, and put in a tube of distilled water free from carbonic acid, which has previously been brought to the same temperature as the ammonia standards. All are nesslerised and compared in the usual way. Blanks and standards are frequently done as a control, and a portion of the waters, after reduction, is tested for nitrites. When nitrites are found in considerable quantity, the determinations are repeated, usually giving a higher result. In calcolating the results, the correction for the volume of the caustic soda solution added, and the loss of ammonia with the hydrogen, and the reduction from ammonia to nitrogen, can all be combined in a single factor. When the determination is carried out as above, this factor is o'88. Thus if 5 c.c. of the reduced water contains o'05 m.grm. ammonia, 5 c.c. of the original water contained 0.88 of this, or o'044 m.grm. nitrogen as nitrate. The blank should not exceed o'005 part nitrogen in 100,000, and can often be neglected. Deduction is made for the animonia and nitrites, but when the ammonia is a considerable fraction of the total nitrogen, it must be removed by boiling, after adding the caustic soda. In these cases, and also with waters having very low nitrates, especially yellow surface waters, the results obtained in this way are often unsatisfactory, both because the ammonia reading is influenced by the colour of the water, and because

་

10 parts peptone

10 parts peptone with potassium nitrate

20 parts peptone

20 parts peptone with potas. sium nitrate

Io parts egg albumen
Sewage, albumenoid
monia, o'50
Sewage, albumenoid
monia, o'72

am

am

Ο ΟΙ

0'02

Distilled.

With large amounts of organic matter, the action of the caustic soda and aluminum decomposes a portion of the organic matter with formation of ammonia, as is shown by the experiments with peptone and albumen. As far as our experiments have been carried, this error does not exceed from 2 to 4 per cent of the albumenoid ammonia. With almost all ground waters, and with most surface waters, this error is quite insignificant. It is only in presence of large quantities of decomposing organic matter that the result becomes uncertain. In such cases nitrates are invariably accompanied by nitrites, for if nitrate is added to such a water, a portion, if not the whole, is quickly reduced to nitrite, and in turn, the nitrite is often reduced, after all the nitrate is gone. We are thus able, in many cases, as for example with sewage, to infer the entire absence of nitrates from the low results, and the absence of nitrites. With excessive organic matter and the presence of nitrites, the results obtained can be taken as a maximum limit, and the same less 4 per cent of the albumenoid ammonia, as a probable minimum limit. It should be observed that the nitrate determination in such unusual cases is of minor importance, and that the phenolsulphonic acid process is also quite unreliable under these conditions, the organic matters giving a yellow colour which often corresponds to more nitrate than is shown by the aluminum process.

With potassium nitrate solutions in distilled water, the results are quite satisfactory, being as accurate as the strength of ammonia solutions can be determined at a single trial by direct nesslerisation, and we may believe that our results upon waters, with the above mentioned exceptions, are equally accurate.

Support in Weighing Substances.-H. Schweitzer (Chemiker Zeitung) recommends zylonite, a paper made from nitro-cellulose, camphor, and alcohol.