In column, the serial order is based upon the complete mineral analysis.

When every mineral is taken into consideration, it is very evident that the soil ranking first in serial order will be the one with the largest amount of mineral matter dissolved, which is to say, that the soil lowest in insoluble residue will rank first in serial order. Accordingly column i is obtained direct from the column headed "Insoluble Residue," in Table B.

If these averages are arranged in two divisions, the serial order up to 17 should be found in the first division, in order that they may agree with the actual productive capacity, whereas, upon inspection, it is found that in every instance but half as many are found in the first division as should be there theoretically. If arranged in three divisions, the first division should contain eleven, ranging from eleven down. However, those found in the first division that come within this range, vary from one-fifth to less than one-half of the theoretical number. Again, if arranged in four divisions, in the first should be found eight, ranging from that figure downward, whereas, there is found but from one fourth to one-half of the theoretical number.

At the bottom of the table are found the number that are found in each column in the first half, first third, and first fourth respectively.

The only deduction that can be obtained from this, is that the figures are remarkably adverse to the conclusion desired, as is well illustrated in pot 1 (which will undoubtedly rank first), according to all of the standards suggested in this table, while it grades but one-third of the maximum crop. The pot ranking lowest, will be unquestionably pot 89, when it is seen that the lowest and highest are consecutive, the lowest preceding.

It is believed that no such collection of complete analyses, on such a variety of soils whose agricultural value has actually been determined, has previously been presented, and that it would be difficult for them to be shown in a more unfavorable light. However, many such analyses are made, and in some cases even the mineral matter insoluble in acid is determined by means of fusion.

Such analyses might be of value in determining the geological origin, but this could be better done by an examination of the

rock, in the case of sedentary soils, whereas in the case of transported soils it could not be done in either case.

If it is expected that the digestion in strong acid will reveal what is to be the future state of fertility, it is suggested that the effect of weathering, upon this insoluble mineral matter, is so infinitesimal that it would be out of consideration in so far as agricultural purposes are concerned.

It will not be overlooked that the discussion throughout this paper, pertains to the oat plant only. As to whether the oat plant's habits and requirements of the oats are comparable with those of another plant, or whether the conditions of every plant must be established independently, is not within the scope of this paper. On lines similar to those followed in this paper, it would be possible to establish solvent conditions as representing the feeding ability of any plant, whereupon the desired crop would be specified when the soil sample is forwarded for analysis. If a diagnosis of a soil is made, it must be with reference to a specified plant, as plants vary so in the nature of their feed, and their ability to obtain it.

16

If plant food can be identified by a laboratory method, there is no doubt as to a method of procedure in the taking of soil samples from the field. A succession of similar depths should be taken in order to ascertain how deep the available food existed, and with this, compare the depth to which the feeding roots of the intended crop are known to penetrate. For actual practice, the writer has constructed a very simple form of sampling cylinder, made out of 7-inch wrought-iron pipe. The pipe is cut 6 inches in length, and turned down to a thickness of 1⁄4 inch, leaving a collar on one end, to strengthen and drive upon, while a cutting edge is turned at the other end. This makes a strong cylinder weighing about 4 pounds. The cylinder is driven down to the top, and the enclosed soil taken out. The soil is dug from around the sides of the cylinder as it is driven down for the second 6 inches, and so for a third 6 inches. The separate portions are weighed and subsampled for analysis, and from such data the pounds per acre of plant food to definite depths are obtained.

It is clearly obvious that the depth to which the plant food is supplied to the plant, is of great importance. To fix a definite depth at which a soil should be sampled, is about the same as at

tempting to fix a definite depth at which a plant shall feed. If the first 6 inches of one soil should contain, say 20 parts per 1,000,000 of available phosphorus pentoxide, and the second 6 inches should contain none, and in a second soil the first 6 inches should contain 10 parts, and the second 6 inches also 10 parts, it is reasonable to suppose that in the growth of plants, which will feed to a depth of 12 inches or more, two such soils will be equally fertile.

The more divisions into which a soil sample can be divided, the more data there will be for study. If the successive depths could be reduced to 3 or 4 inches (it being understood that definite areas are taken), and a series of samples taken until the vanishing point of the available food is reached, from such data a curve could be drawn which, if compared with the root system of a plant, would illustrate the amount of available food which could be assumed as being present.

However, if it is limited to two, or even one sample, let the total depth be that to which the intended crop is known to feed. If it is intended to estimate the amount of plant food which is in an acre to the depth of a feeding crop, say 12 inches, and all of the available food chances to be in the first 4 inches, it is immaterial as to whether the soil be sampled to a depth of 4, 6, 8, 10, or 12 inches, as the final calculation will be the same in each case. Consequently, if it is not known to what depth the available food exists, if a single sample to the depth of 12 inches be taken, it covers all doubt, and with no disadvantage in case the available food does not extend so far.

For an illustration, Table N is an arrangement of the first and second 6 inches of the same soil, with the phosphorus pentoxide and the potash of the upper and lower samples arranged accordingly.

In soil a the first 6 inches were in pot 1, with the second 6 inches in pot 31, the others being similarly arranged. The phosphorus pentoxide in soils a and d is seen to be equally distributed to at least a depth of 12 inches, while in soils c and f, practically all of the phosphorus pentoxide is found in the first 6 inches. In soils b and e, there is practically none in the first 6 inches, and, as would be expected, none in the second 6 inches.

If it is undertaken to place a comparative value on these soils with respect to the phosphorus pentoxide, let it first be assumed

that the intended crop can not feed to a greater depth than 6 inches. Then soils a and c are equally fertile, as will be also d and f. If it is assumed that the intended crop will feed to at least a depth of 12 inches, the soil a will be one-half more fertile than c, while d will be two times as fertile as ƒ, soils b and e being deficient in both cases.

Similar comparisons may be made with the potash.

In the continuation of soil study, assume that the available plant food rarely extends to a depth of more than 12 inches, and that plants penetrating below that depth, do so for the purpose of obtaining moisture. If such assumptions should become facts, it would be a simple matter to obtain an accurate sample by sampling through a definite area to a depth of about 12 inches, weighing the total sample, mixing and taking a definite portion as a subsample for analysis. In such a case, the exact depth to which a sample is taken would be of small importance. It might range from 10 inches to 10 feet without variation in the result, provided the area over which the sample is taken, is accurately defined, and a definite portion of the thoroughly mixed total -sample taken.

METHODS OF CHEMICAL ANALYSIS AND FOUNDRY CHEMISTRY. BY FRANK L. CROBAUGH, M.S., Cleveland, O: Published by the author. The analyst, or any person who intends to become an expert analyst, needs to have his methods at hand concisely stated, with all details verified by himself or by an author who has tested the methods thoroughly. In a work recently published by Mr. Frank L. Crobaugh, of Cleveland, Ohio, may be found a collection of selected methods especially in use in analysis of iron and steel, which have been thoroughly tested by the expert hand of the author. On account of the varied experience of Mr. Crobaugh in iron laboratories, in iron and steel plants and in his own business, his compilation of these methods will be of great service to any chemist engaged in similar work.

In Part II of this work will be found a mor complete statement of the principles and methods of foundry chemistry, than can be found elsewhere. Foundrymen are beginning to appreciate the importance of a more thorough knowledge of the chemical composition of pigs and castings, and of the chemical changes involved in foundry practice. They will find this book very useful in their daily work. C. F. MABERY.