Set I.-Taken by excavation of holes 9 x 4 inches to subsoil (3-8 inches; average, 6 inches). Excavations, 11 to the series, taken in zigzag lines crossing the plat from side to side, beginning at the Northwest end and concluding at the Southwest end. Series A includes the excavations begun at the West corner of Plat I; Series B, those beginning at the North corner of that plat; Series C and D, the corresponding excavations for Plat IV. The subsamples from the excavations were weighed, air-dried, weighed and prepared separately for analysis, but only the corresponding fine soils (passing a sieve of inch mesh) were used in aliquot portions to make up series composites for analysis. The individual subsamples weighed from 2.25–5.75 kg. each, after air-drying.

Set II.-Taken by borings with a inch soil auger, using only the surface soil thus removed from each boring. The series in this set were lettered E to H, and correspond in distribution to the respective series, A to D, above described. Each series represents 40 borings, which were composited as they were taken. Weights of composites, air-dry, 4.5-4.9 kg.

Set III.-Taken and composited in the same manner as Set II, but distributed along the lines of an X over the respective plats, 64 borings to the plat. Series I includes the borings from Plat I; Series J, those from Plat IV.

All the results are based upon the air-dry soils, whose hygroscopic moisture content ranged from 0.72 to 0.95 per cent. The difference would affect only the acre weight determinations and these in amounts within the limits of error of the determinations.

ACRE-7 INCH WEIGHTS.

The acre-7 inch weights, calculated from the results obtained by the excavation method, with the correction described in the following paper1, afford the following data. The computations were based upon apparent specific gravities for soils to the average depth of 61⁄2 inches. A slight, but nearly uniform error is thus introduced.

The probable error of these apparent specific gravity determinations, computed by the use of the Gauss formula, rarely affects the results to more than the third decimal place. Expressed in pounds to the acre-7

1 J. Assoc. Official Agr. Chemists, 1920, 4: 103.

inches, the probable errors of the single series of eleven subsamples each range from 9,300 pounds to 15,500 pounds, and for the twenty-two determinations to the plat, from ± 8,700 pounds to ± 9,800 pounds. The differences between series were therefore distinctly larger than the probable errors of the determinations. The extreme range of apparent specific gravity found for single subsamples (1.209 to 1.425) emphasizes the importance of the duplication of such determinations.

FRACTIONING THE SAMPLES.

The excavated subsamples were so large that, to save labor, their fractioning or parting before sifting out the fine soil was highly desirable. The accuracy of the usual simple parting method was first tested. The air-dry soils from two subsamples were rolled on an oilcloth fifty times, the fine soil was brushed radially to the central heap, the latter quartered, and each quarter pestled with a rubber pestle and sifted separately. The following results were obtained:

This method of mixing evidently failed to secure a uniform distribution of the larger particles through the aliquots. Probably no serious difference in the mechanical groups of the fine soil would have resulted from its use, but the gravel component might be very considerably affected. Since this point was of importance in the present study, the entire amount of each subsample was sifted.

SIFTING RESULTS FOR PLATS I AND IV.

The results from the sifting of the two lots of subsamples follow:

Differences of the magnitude reported above must obviously affect the weight of fine soil found for a particular area and depth. Whether the chemical composition also of the fine soil is similarly affected remains to be considered.

The respective fine soils (series composites, in most cases) were separately analyzed, usually in duplicate, with respect to a number of the constituents. The nitrogen and phosphoric acid for all samples, and

the loss on ignition for Set III samples, were determined by Walter Thomas; all the other determinations were made by E. S. Erb. The results, together with the year of analysis, are presented in the following table:

Phosphoric acid, soluble in N/5 nitric acid, 40°C., 5 hours. 1916 0.0020§

Recalculated to 0.95 per cent moisture basis.

† Recalculated to 0.82 per cent moisture basis.

All of these samples represent unusually large amounts of soil taken, an unusually large number of subsamples, and subsamples distributed with care, probably greater than usual; also, no error in parting or quartering the samples is involved in the case of Sets I and II. In contrast with these, each series of Set III represents, on the one hand, a greater number of subsamples than was used for the series of Sets I and II; but, on the other hand, a somewhat different distribution of subsamples and a mode of quartering that the writer's later work has shown to be inadequate for the purpose of a mechanical analysis.

Comparison of the respective series results for Sets I and II, shows that, despite unusual care, the sampling error remains greater than the analytical error. The series averages from the boring samples, Set II, agree as well with one another as do those obtained by excavation;

hence the one method represents the composition as well as the other. The importance of careful distribution of subsamples and preparation of composites is shown by the differences in results for the samples of Sets II and III.

EXCAVATION METHOD FOR DETERMINING THE APPARENT SPECIFIC GRAVITY OF SOILS.

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

Various methods have been proposed and used for determining the apparent specific gravity of soils. They may be divided into two classes: First, those in which, by means of a tubular or prismatic implement with a cutting edge, a block of soil of definite cross-section and depth is separated and removed from the body of the soil; second, those in which the measured block of soil is separated from the surrounding soil by trenching, is shaped to exact dimensions in place, and, with or with