Agricultural investigations have, in the search for methods useful in guiding fertilizer practice, turned to the chemical analysis of the crop produced upon a given soil, and to direct trial of yields by experimental use of different fertilizers on a small scale.

In examining the composition of the plant as an indicator of the soils' deficiencies, the results of analysis of different organs from plants grown on soils deficient in particular ingredients, and from plants grown on plats of the same land in which the deficiency has been made good by fertilizers, are used.

The results of analysis are expressed in the form of ratios existing between the three important fertilizer constituents present in the dry matter of the particular plant part used. The quantity of nitrogen is usually taken as 100. From such examinations, Atterburg, examining the seed, concluded that if the ratio of phosphoric acid to nitrogen were less than 50: 100, the maximum crop could not be obtained, and if the ratio were less than 20: 100, it might be concluded that a serious deficiency of phosphoric acid existed. Stahl-Schroder from similar studies fixed upon the ratio 30-40 : 100 and 20-30: 100 as more nearly correct than those fixed by Atterburg. Mr. H. H. Pingree,* latè assistant chemist in my laboratory, made a careful study of the oats crop grown on a number of our fertilizer plats at the station, and obtained results similar to those just mentioned. It is probable that this method of examination of a soil's condition would require as a basis of judgment a careful analysis of the crops contained from a series of fertilizer plats on the soil in question. The trials of this method have not been sufficiently numerous to permit a safe judgment as to its limitations.

Another class of methods for testing present needs of the soil involves a comparison of the crop yields obtained from the unfertilized land, and from that treated with various fertilizer constituents and combinations thereof. I will refer first to those methods of this class, which are designed to be conducted on a small scale, and, it may be, at some point remote from the farm under study. In such case representative portions of the soil are packed into pots, the latter being first provided with a layer of coarse material to insure the proper drainage. The upper layers of the soil in some of the pots receive the additions of fertilizer, and the weight of crops produced in the several pots is accurately determined. In such experiments care is of course taken to preserve, so far as practicable, similarity of conditions with respect to the manner in which the soil is packed into the pots, the temperature and the moisture during growth. Heretofore it has been customary to use rather large pots. and to grow the crop to maturity in order to gain the comparative data. It is a common fault of such experiments, no matter how carefully conducted, that they do not exactly reproduce the temperature, moisture and aeration conditions that influence the crop grown on the undisturbed soil, and it is well known that each of these conditions is an important factor in determining crop growth. No matter how carefully conducted such experiments may be, it is clear that they are of value as preliminary trials suggesting probable needs of the lands rather than as conclusively showing both the kind and the proportion in which the several fertilizers may be required.

*See Report of the Pennsylvania Experiment Station of 1905-6, pp. 43-53.

Pot experiments of the kind just mentioned are somewhat cumbrous and extensive, and require an entire growing season for their completion.

The Bureau of Soils of the United States Department of Agricul ture has recently proposed another method in which baskets, three inches in depth and diameter, made of paraffined wire gauze, are used instead of large pots. A single species of plants has been selected for use, namely, a hardy variety of wheat, a half-dozen seedlings being employed for each pot. The plan involves the observation of the manner of growth during a period of five weeks, and a determination of the green and dry weights of crop developed by the end of that period. It is clear that where so small quantities of soil are used that the difficulty of securing representative portions is increased. This tendency to error is somewhat offset by the possi bility of conducting the same treatment upon a number of pots, say five or six, and using the average result rather than that obtained from the trial of a single pot culture. It is very certain also that the results obtained during so small a portion of the cycle of growth of a plant cannot accurately represent the results that would be obtained in the whole cycle. To illustrate, much of the soil nitrogen is not directly available to plants, but becomes available by the action of nitrifying bacteria during the season of growth. The rate of nitrification under field conditions is not uniform through the growing season, nor is all the nitrogeneous material of the soil equally susceptible to nitrification. These facts make it very improbable that even under field conditions the first six weeks' growth of the wheat crop will accurately represent the soil changes and plant gains through the whole season. The results thus far obtained by the use of the wire-basket method indicate that it has considerable value as a means of determining the kind of fertilizer deficiency in the soil, but that it does not afford conclusive evidence respecting the proportions in which the deficient constituents must be supplied in order that a maximum wheat crop may be secured, assuming that the conditions of culture and season are most favorable. Since it is clearly established that one crop may do fairly well where another species fails, it is clear that the wire-basket test with wheat, or, indeed, a test of the soil by any method with wheat, cannot afford very certain information as to the deficiencies of the soil for the crops of other species, such, for example, as clover, potatoes or turnips.

For these reasons, while the pot methods are to be regarded as highly available means of soil examination to be conducted at central stations, where continuous and skilled supervision can be given, the need remains for tests also upon the soil in place. To reduce the costs of such tests, and at the same time to make possible their more accurate performance, experiments upon small plats rather than upon large fields have been preferred. Another reason for this preference is that by duplicating the series of plats receiving the several treatments under comparison, it is possible to generally reduce the errors due to local variations in the character of the soil.

The Experiment Station has not been able to enter largely into the test of Pennsylvania soils by pot-culture methods. It has, however, recently inaugurated a series of soil tests with fertilizers fitted for the growing of corn, wheat and potatoes. In 1906 I secured the

co-operation of over 100 farmers, and this year there are over seventyfive tests for corn and potatoes in progress. Part of the tests of this year, I am glad to say, are repetitions of those made last year. It is important that the results through several seasons be obtained, since fertilizer effects are very considerably influenced by the character of the season. It has not been possible as yet to report the details of these co-operative tests. I have, however, called out some of the more interesting results.