EXPERIMENT STATION RECORD.

VOL. VII.

No. 2.

The present number of the Record contains the concluding chapter of an article by Dr. J. B. Lindsey, of the Massachusetts Station, on the organization, equipment, and work of the experiment station at Darmstadt, Germany, which gives the details of Wagner's method of pot experimentation.' The Darmstadt Station is justly considered a model as regards equipment for this kind of investigation. The methods there pursued are the result of years of continuous effort to perfect a system of exact research into the laws of plant nutrition. The development of this system was prompted by a conviction on the part of Professor Wagner that it is impossible to secure results that can be relied on from field experiments on either large or small plats of soil under natural conditions, even when the greatest precautions are taken.

It may not be generally admitted "that the results of such [field] experiments are in no way to be depended on when looked at in the light of scientific research," but there is a growing skepticism among thoughtful investigators in this country as to the reliability of the plat method. Although at first thought comparatively simple, the method is beset with difficulties which experience has shown to be extremely hard to overcome. Frequently the results of a year's work are vitiated by conditions entirely beyond the control of the experimenter. It is only by a thorough appreciation of the difficulties to be guarded against, a strict attention to all details, and a continuation of the experiments through a series of years that the experimenter can hope to secure results of scientific value.

By means of pot culture many of the sources of error in plat experiments are eliminated. The conditions are made more uniform and are to a larger extent brought under the control of the experimenter. For instance, the soil is made uniform in both chemical and physical properties, and may be exhausted of any single fertilizing ingredient if desired; the supply of plant nutrients is measured more definitely; the supply of water is made the same for all the pots; the ravages of insects are very largely prevented; and the factor of meteorological conditions is eliminated. In this way definite laws are worked out with greater certainty

'The methods of conducting pot experiments at the Bernburg Experiment Station, where the assimilation of atmospheric nitrogen has been extensively studied, have already been described (E. S. R., 5, pp. 749, 835).

and in less time than would be possible by plat experiments. As a matter of fact practically all that is at present known concerning the fertilizing ingredients required by plants has been demonstrated by pot and water cultures. To this may be added our knowledge of the assimilation of atmospheric nitrogen by plants, the nitrification of soils, and the availability of many different forms of plant food. The best combinations of fertilizing ingredients for different crops have in many cases been suggested by the results of pot experiments, and this method is being very largely employed in Europe in studying the whole question of plant nutrition.

There are many questions which field experiments in their most improved form are adapted to studying, such as certain general questions relating to the mutual adaptability of soil, climate, and plant; methods of culture; relative value of different crops and of varieties; rotations, etc. They are also useful in demonstrating the practical value of the principles deduced by more exact methods and thus popularizing the results of agricultural investigation. Many of the American stations enjoy exceptionally favorable conditions for field experiments in that it is possible for them in many cases to carry on parallel experiments on virgin soil and on the same soil modified by different systems of culture and manuring. We can not afford to discard the system of field experiments as a means of investigation, but we should not lose sight of its inherent weaknesses, and should endeavor to so systematize and improve it that these sources of error will be reduced to a minimum.

Pot experiments afford a valuable means not only for supplementing but for controlling field experiments. As already suggested, field experiments may serve the purpose of demonstrating the practical value of the results of more exact methods, but pot experiments may be equally useful as a check on field experiments. Experiments in pots are also almost indispensable in those cases where a great variety of typical soils are to be studied at the same time with reference to physical and chemical properties.

The value of the method of pot experimentation as an adjunct to other means of investigation is being more widely recognized in this country every year. Some of our stations are now well equipped for work in this line, and it is believed that the results obtained by them will fully justify further extension of the system.

WAGNER METHOD OF POT CULTURE.

J. B. LINDSEY, PH. D.

In the previous paper1 the objections to field experiments were discussed and brief mention was made of the method worked out by Prof. Paul Wagner at the Darmstadt Experiment Station for studying problems in plant nutrition in cylinders sunk into the ground and in movable pots. The following are the principles on which this method rests:

PRINCIPLES UNDERLYING THE METHOD.

(1) All conditions having any influence upon the development of the plant must be similar for each similar experiment.

As conditions might be mentioned physical and chemical character of soil, the amount of plant food and water present, compactness of the soil, distribution of fertilizer, quantity of seed, depth of planting, number of individual plants, thickness of planting, light, heat, circulation of air, etc.

(2) The limit of error must be ascertained by carrying out parallel experiments, and each experiment must be continued until accurate average results are obtained.

This is extremely important, and in by far the larger number of experiments heretofore made it has been entirely lost sight of. In order to illustrate this Wagner gives the following example:

Suppose a field 1 hectare in area be used for the purpose of studying the comparative values of phosphoric acid in the form of dissolved boneblack and Thomas slag. The common method of experiment would be to divide the field as nearly as possible into three equal parts, apply. ing 30 kg. of boneblack phosphoric acid to one plat and an equal quantity of phosphoric acid from Thomas slag to another, leaving the third plat free from phosphatic manure. Each of the three divisions would be manured with nitrogen and potash in excess. If the field were planted with wheat, the excess of the yield of the phosphoric acid plats over the one not thus treated would be taken to show the relative effects of the two forms of phosphoric acid.

It would, however, be by no means right to draw such a conclusion, for in so doing it is assumed that each of the three plats had exactly the same conditions of soil and moisture, were injured to the same

1E. S. R., 7, p. 3.

extent by birds or insects, and were manured, planted, and harvested in exactly the same manner. Now, it is hardly possible to have all of these conditions precisely the same. In such an experiment it is quite probable that the error caused by one or the other varying condition would have been so large as to more than overbalance the increase in yield caused by the phosphoric acid applied, and hence the results would be false.

The correct way would be to divide the field not into three, but into twelve equal plats, leaving four without any phosphoric acid and apply. ing the two different forms to the other eight, one form for each four. With the results of four plats in case of each separate test it is possible to note the uncontrollable limit of error, and if not too great variations exist the average of the four plats in each case should give safe figures from which to draw conclusions.

An average of parallel experiments should be taken as a basis from which to draw conclusions only when these experiments are made upon the same soil and under precisely similar conditions.

(3) In conducting experiments the pots or plats containing the fertilizer to be tested should be made to produce the highest possible yields over those not fertilized. In this way the accuracy of the method is increased.

For example, if the error in case of parallel experiments were but 2 per cent it would as a rule be considered very small. If, now, the average of the unmanured plats should be 100 lbs. of grain, and the average of the manured plats 105 lbs., the error of 2 per cent or 2 lbs. would be altogether too large to admit of conclusions as to the effect of the fertilizer being shown. On the other hand, if the plats were so treated that the unmanured produced 100 lbs. and the manured 150 lbs. the error of 2 per cent would be of no practical account.

(4) It is important to select a soil which is as poor as possible in the ingredient to be tested.

If different forms of phosphoric acid are to be studied and a soil is selected so rich in that ingredient that without an application a certain area will produce 400 lbs. of grain, and a like area after a liberal application will show an increase of but 50 lbs. over the unfertilized, such increase is certainly far too small to allow any trustworthy conclusions to be drawn.

Sand would be very poor in all three ingredients, but the physical condition of sand is not suitable for such an experiment. The most suitable soil by far is the surface soil of cultivated fields poor in the ingredient to be studied. If soils can not be procured that are onesidedly exhausted they must be prepared. The method of preparation will be described further on.

(5) In order to prove the value of any particular element of plant food those plants should be selected that have the greatest need of this ingredient.

(6) There should be no excess of the fertilizing ingredient tested left in the soil.

If 30 lbs. of soluble phosphoric acid from boneblack produces 200 lbs. of wheat, and 30 lbs. of phosphoric acid from Thomas slag produces a like quantity, one is by no means justified in saying that the Thomas slag gives equally as good results as the dissolved boneblack. It is quite possible that all of the 30 lbs. of soluble phosphoric acid was not necessary to produce that amount of wheat; perhaps 15 lbs. would have produced as large a yield.

In order to prevent an excess in the soil of the ingredient to be tested, it is of the first importance that the other necessary elements of fertility be present in liberal quantities. Only when this condition is fulfilled is it allowable to draw conclusions. It is by no means an easy problem to secure these conditions, for it is equally important that these ingredients be not present to such an extent as to prove injurious to plant growth. Wagner states the case as follows: These other ingredients are to be applied (a) in proportion to the quantity of the fertilizing ingredient being tested; (b) according to the amount which the soil has at its disposal; (c) in proportion to the absorptive power of the soil; (d) according to the particular need of the plant for these several ingredients; (e) depending upon the sensitiveness of the plants toward an excess of the ingredients, and (ƒ) according to the quantity of plant growth produced on a given area.

A knowledge of the soil and of the plants to be used as well as some experience is necessary in order to supply the excess ingredients in the correct quantities to secure such a favorable development of the plant that the extra amount produced over the unmanured pots is to be attributed to no other cause except the influence of the ingredients being tested.

(7) Different quantities of the ingredient to be tested should be used, and conclusions should be drawn only when the increase in the yield. keeps pace with the increase in the quantity of the ingredient applied. If, for example, 1, 2, 3, 4, and 5 gm. of nitrogen produces 100, 200, 300, 400, and 420 gm. increase in yield, respectively, it is clear that the first 4 yields kept pace with the increased supply of nitrogen, while the fifth application of nitrogen did not for some reason have opportu nity to do its full work. The first 4 applications make it evident by reason of the regular increase that no other ingredient was lacking and that the condition as called for above was fulfilled.

One can therefore depend on the results from the 1, 2, 3, and 4 gm. applications but not on those from the 5 gm. application.

(8) The experiment must be conducted in such a way as to clearly show that the increase in yield was brought about by the single ingre dient under investigation, and not by any other influence.

If it is desired to prove the value of the phosphoric acid in ground bone the fact must not be lost sight of that the bone contains nitrogen