TlIK WOJU.KN KXrKKI.MKNTS : TIlKlli OBJECT AND .MKTlloD. ^1 
would consequently be 1-^,000 times less certain than in the case of the 
wheat plot. 
Another great dilliculty is the length of the life of trees. With 
annual farm crops the experiment can be completed in a year, but with 
a fruit-tree the life history may extend over fifty years or more, and we 
can hardly expect certain of our conclusions to be valid unless they are 
based on observations for at least half this period. 
There is also another serious difficulty, in the existence of the enormous 
number of varieties of each sort of fruit-tree. For instance, to do any 
one experiment properly on all the varieties of Apples known would 
require a much larger area than could reasonably be devoted to an entire 
experimental station. Yet we can never feel sure, if we restrict our 
experiments, as needs we must, that our conclusions may not be fallacious 
when applied to other existing or forthcoming varieties. 
There are several other, though less important, difficulties in 
experiments on fruit-trees as compared with those on agricultural crops. 
A great deal depends on culture in the form of pruning, and it is quite 
possible that the adoption of different methods in this respect might lead 
to contradictory results in different hands, even where the same cultiva- 
tion of the soil was adopted. 
As to experiments with manures, also, there is this difficulty, that if 
the tree is to be allowed to attain maturity, a great deal more space must 
be allotted to it than it can utilise in its early years. If we manure the 
whole of this space we shall be unduly enriching the soil which is now 
unoccupied, whereas a partial manuring of the ground round the stem 
lands us in the difficulty of determining how much shall be manured, 
and to what extent the manure is washed out into the neighbouring 
unmanured soil. Of course, there are other difficulties which apply to 
experiments in agriculture as well as in horticulture — the great difficulties 
of the differences of soil, climate, and seasons ; the latter only can be 
surmounted by extending the experiments over a sufficient time, the other 
two cannot be combatted in any one expeiimental station. There are 
some points which we can hope to solve only when we shall have many 
such stations distributed over the country. 
Great as these difficulties are, I scarcely think that they should be 
used as arguments for not attempting experimental work in horticulture. 
No doubt they may lead us to many uncertain and erroneous conclusions 
at first, but if only they induce others to start similar work under some- 
what different circumstances we may be sure that correct conclusions of 
general application will be arrived at in the end, and our work will not 
have been valueless. 
One essential feature of experiments such as ours is that they must 
be comparative— that the results of any particular treatment cannot be 
gauged unless we have another plot under ordinary or normal treatment 
with which the special plot can be compared. A second essential is 
that in each experiment one condition of treatment, and one only, must 
differ from that in the normal plots, for if two or more are altered at 
the same time it is obvious that we shall not be able to decide to 
which of these alterations the difference of results is due. This is but an 
instance of the mathematical principle that one equation can evalue but 
