



























< 
APRIL 7, 1923] 
quantity appear to do actual harm. It is obvious, 
therefore, that a complete manure drawn up on the 
basis of the Rothamsted experiments would fail in 
practice to give the best results on a lighter soil. As 
an instance the following may be quoted, this being 
one of a general scheme of experiments organised 
from Rothamsted : 
- 
Barrey: Licur SAnpy SoIL IN SUFFOLK, 1922. 
Bush. per acre. 
Complete artificial manure Sete igs: 
Incomplete manure: phosphate 
omitted . ‘ . : 27°5 
No manure 16:0 
In this instance the omission of phosphates has raised 
the yield by 6 bushels per acre. As against this, an 
array of instances might be brought from clay farms 
where a phosphate is the one and only thing that 
causes crop increases. Any one who had to deal with 
a es 
tease resces » farmyard manure 
_ Fic. 1.—Yields of wheat from Broadhalk plots manured with complete 
artificial manures, and farmyard manure, respectively, compared 
with the average yield for the whole country. 
farmers’ problems could multiply apparent contra- 
dictions and inconsistencies of this kind. When one 
collects, as we have done at Rothamsted, the results 
field trials with artificial manures made in different 
arts of the country they seem at first to be simply 
tangled mass of unrelated facts. 
Now it is the business of the man of science to sort 
out a tangle of this kind, to reduce it to order, to find 
the general principles running through it, and finally 
to prove the correctness of his conclusions by being 
able to predict with certainty what will happen in 
given conditions. The recognised method of pro- 
cedure is to discover the various factors at work and 
‘investigate them one at a time. This is being done 
at Rothamsted in two ways: by field observations, 
and by quantitative laboratory measurements. Ob- 
servations in the field show that each of the fertilis- 
ing substances— phosphates, potassium compounds, 
‘nitrogen compounds, etc.—in addition to its general 
effect in increasing plant growth, produces certain 
specific effects which may be of advantage, or may 
be a disadvantage to the plant in the particular con- 
ditions in which it happens to be growing. Thus, 
phosphates have a special influence in hastening the 
_ ripening processes, which no doubt accounts for the 
- Suffolk results just quoted. In the dry conditions 
_ of a sandy soil, ripening is already too early, and any 
reduction in an already short growing season cuts 
No. 2788, voL. 111] 
NATURE 
are less needed, and, indeed, beyond a certain | down the yield; in cold, wet districts, however, this 
467 

property is very valuable. 
' In the early stages of the plant’s life phosphates 
stimulate root development to a marked degree ; this 
is well shown in their effect on swedes. Nitrogen 
compounds tend to increase leaf development and 
give greater vigour of growth, but beyond a certain 
point the advantage is counteracted by a loss of 
resisting power, and the plants may fall victims to 
attacks of disease. Crops—especially cereals—may 
be unable to stand up against the weather and may 
become “lodged.” Indeed, the proper adjustment of 
plant nutrients affords plant pathologists one method 
of dealing with plant diseases. 
Qualitative observations of this kind, while of high 
value, are not entirely sufficient: it is necessary to 
have quantitative measurements of as high an order 
of accuracy as possible. At Rothamsted this is done 
by means of water cultures and pot experiments ; 
all the factors are controlled as closely as possible 
and the results are plotted on curves which can be 
studied in detail. This method was developed ex- 
tensively by Hellriegel and is now in common use in 
agricultural laboratories. 
The method naturally invites mathematical treat- 
ment, and attempts have been made, notably by 
Mitscherlich, to express the curve by equations. 
There is a seductive look about a mathematical formula 
which rarely fails to appeal to the biologist, but as 
a rute the number of experimental points obtained is 
much too small to justify mathematical treatment, 
and it is not surprising that investigators fail to agree. 
Ten years ago the fashion was for logarithmic curves ; 
now it is for sigmoid curves, which are probably 
nearer the truth, though not yet a complete expression. 
This method of studying single factors is pushed 
to a high degree of refinement in plant physiology 
laboratories, such as that of the Imperial College 
under Prof. Blackman, or that under his brother at 
Cambridge, and there can be little doubt that the 
effect of individual factors on the plant will ultimately 
be well known. All this work is giving valuable 
information as to causes and principles. 
These curves show the relationship between yield 
and plant food supply at one particular temperature 
which remains constant, and one particular water 
supply which also remains constant. But a completely 
different set of figures would be obtained if the tempera- 
ture were different or if the water supply were altered. 
Supposing one wished to take account of the effect of 
water supply as well as food, one would draw a series 
of curves, which would properly be expressed as a 
surface, and this has been done by one of the Rotham- 
sted workers—Mr. J. A. Prescott—to show the effect 
of nitrate supply and spacing on the yield of maize 
in Egypt. The experiments had the advantage that 
the climatic conditions are less fickle there than here. 
It would be of the greatest interest to obtain such 
surfaces for other pairs of factors. 
If an attempt were made to study factors three at 
a time, it would be necessary to prepare a series of 
surfaces and to embody them in a figure in four dimen- 
sions, which is certainly beyond the capacity of the 
ordinary agricultural investigator. But in agricultural 
field work the factors do not vary one at the time, or 
