entirely satisfactory for the determination of the phosphorus variations of soils under investigation in this laboratory. The articles by Robinson1, Rost, and Peters have led to the belief that other workers might be interested in these modifications.

METHOD.

Place 10 grams (except for muck and peats) of the prepared air-dry sample in a 250 cc. graduated Kjeldahl flask, add 0.7 gram of mercuric oxid and digest by the regular Kjeldahl method, as for total nitrogen. Add a crystal (about 0.5 gram) of pure sodium nitrate or potassium nitrate to complete the oxidation. When it is partially cooled, add about 200 cc. of water, and when cooled to room temperature, make to volume.

The solution is filtered, through a good grade of folded filter, as follows: Shake the Kjeldahl flask and pour its entire contents on the filter. Pour back the filtrate until it comes through clear. The solid material settles down and soon prevents everything but the clear solution from passing through. Pipette out 25 cc. of the solution into a 250 cc. beaker, and add 15 grams of dry ammonium nitrate. Heat the solution to boiling, stirring to insure solution of the nitrate. Add, with constant stirring, approximately 30 cc. of ammonium molybdate solution. Place the beaker in a water bath at 60-65°C. for 1 hour. Proceed from this point according to the official methods.

The objects sought and attained by this technique are:

(1) The use of a representative amount of the soil sample.

(2) Uniformity of treatment of all samples not attained in fusion methods.

(3) A single speedy operation to remove the organic matter and prepare the solution.

(4) Conditions for the precipitation of the ammonium phosphomolybdate which may be duplicated.

The concentrations of the solutions used and the addition of the dry ammonium nitrate without preliminary neutralization always yield a clean yellow precipitate of ammonium phosphomolybdate.

Acknowledgment is made to S. D. Conner (Agricultural Experiment Station, La Fayette, Ind.) for cooperation in the testing out of this technique.

1 J. Ind. Eng. Chem., 1916, 8: 148.

2 Soil Science, 1917, 4: 295.

3 J. Ind. Eng. Chem., 1915, 7: 39.

J. Assoc. Official Agr. Chemists, 1917, 3: 149.

Assoc. Official Agr. Chemists, Methods, 1916, 2.

DETERMINATION OF MOISTURE IN FIELD
SAMPLES OF SOIL1.

By H. A. NOYES2 and J. F. TROST (Agricultural Experiment
Station, La Fayette, Ind.).

The amount of field soil to be used for making moisture determinations has been left largely to the individual analyst. It is obvious that differences in the structure of soils must bear a relation to the amount of sorting which will occur in the taking and transferring of specific amounts of soil for the determination of moisture.

Four different soils were studied—a fine gravel, a fine sand (Wabash sandy loam), a loam (Sioux silt loam), and a black sand, high in organic matter. The structural character of these soils when air-dry is brought out in Table 1.

The table shows that the relative proportions of the different sized particles varied considerably in the different samples.

Moisture determinations were made on each of these soils when they contained little more than hygroscopic moisture and also when they contained the moisture content found under average field conditions in early spring. A sample of the loam containing an intermediate amount of moisture was also analyzed. The samples were surface soil and were brought to the laboratory in sealed Mason jars. The jars were kept sealed except during the time that portions of the sample were being taken from them.

The average moisture content of the four soils under study was de

1 Presented by C. B. Lipman.

? Present address, Mellon Institute of Industrial Research, Pittsburgh, Pa.

termined as follows: The cover was removed from the jar, the spatula inserted and the soil mixed. By means of the spatula, the approximate weight of soil desired was transferred to a weighed dish and the jar was resealed. Triplicate 2, 5, 7, 10, 15 and 20 gram portions of each sample were weighed quickly but accurately to milligrams. The samples were dried to constant weight in a drying oven regulated to run at 102°C.±2°. The average moisture content shown by the triplicate determinations is given in Table 2.

(1) That the moisture content determined on different weights is not the same.

(2) That the average of all determinations made bears different relations to determinations made on the same weights of the different soils.

(3) That the amount of moisture in the different samples affected the results obtained.

Table 3 gives the variations between the triplicate determinations that were averaged to get the results reported in Table 2.

The following factors require that the weight of field soil taken for moisture determinations be large: (a) high moisture content; (b) variable proportions of coarse to fine material; (c) tendency of the soil particles to sort out; and (d) change in moisture during weighing (personal factor).

It was found advisable to ascertain, by testing, the weights of differ

TABLE 3.

Effect of size of samples on variations between triplicate moisture determinations.

ent types of soil that must be taken to have the moisture results agree to 0.1 per cent. This gives the relation between factor (d) and the others. It was noted that, although different individuals may not decide on the same weight of soil, due to a difference in individual working errors, they obtain the same moisture results when each one uses an amount large enough to make his duplicates check well.

SUMMARY.

(1) A quantity of field soil weighing less than 10 grams was found to be unsatisfactory for the accurate determination of the moisture present in a soil.

(2) The weight of a particular soil necessary for an accurate moisture determination depends on the soil, the amount of moisture present in it, and the technique of the person making the analysis.

(3) With all kinds of soil, the optimum amount required varies with the moisture content and physical condition of the soil, and therefore it is necessary to determine the weight of soil which the average analyst should use.

J. B. Rather1 (Agricultural Experiment Station, Fayetteville, Ark.), submitted a paper on "An Accurate Loss-on-Ignition Method for the Determination of Organic Matter in Soils"?.

1 Present address, Standard Oil Company, Chemical Laboratory, Brooklyn, N. Y.

* Ark. Agr. Expt. Sta. Bull. 140: (1917); J. Ind. Eng. Chem., 1918, 10: 439.

C. R. Wagner and W. H. Ross (Bureau of Soils, Washington, D. C.), presented by title a paper on "A Modified Method for the Determination of Fluorin, with Special Application to the Analysis of Phosphates"1.

A STUDY IN SOIL SAMPLING.

By WILLIAM FREAR and E. S. ERB (Agricultural Experiment Station,
State College, Pa.).

The official directions for soil analysis omit all mention of the method of soil sampling, and manuals for the guidance of soil analysts are reticent upon the procedure necessary to secure a sample. Reports of soil examinations discuss as significant small differences in composition, but furnish no data to establish within what limits analytical results vary for different portions of the same soil solution or for different samples of the same soil. In fact, the literature furnishes few data for guidance, and an inquiry, very limited in scope, showed that even those investigators who are directing extensive soil studies, involving large expenditure of time and money, are employing sampling methods quite different in detail.

The writers made a study which, it was hoped, would give information as to the precautions necessary to obtain representative samples of the surface soil upon which the general fertilizer series of plats of the Pennsylvania Agricultural Experiment Station are located. The soil is not of a single type, but is chiefly Hagerstown silty clay loam. The problem was, however, to represent the plats studied, not the individual soil types or subtypes. Earlier studies have shown a marked lack of uniformity as to chemical composition in the soil adjacent to these plats.

Concretely, the question was how many subsamples must be taken from well and symmetrically distributed points over the respective plats in order that duplicate composites from the same one-eighth acre plat may agree satisfactorily with respect to the point of composition in question.

The plats sampled were Nos. 1 and 4, Tier II, of the series abovenamed.

The samples were taken in July and August, 1916, in three sets: I and II, by Erb and Kern, for one study; III, by G. J. Kuhlman, for another study. The sets differed in the following respects:

J. Ind. Eng. Chem., 1917, 9: 1116.

* Annual Report of the Pennsylvania State College, 1908-1909, 215; 1909-1910, 163; 1910-1911, 313.