92 EXPLOITATION OF PLANTS 
able scatter of the observed groups of plants, each group 
numbering about 200 plants ; equally, they can con- 
stitute a very serious waste of time and energy if the 
observed groups are not scattered so as to be compar- 
able, as a recent publication has shown. They can be 
utilised in a number of ways; in the first place, they 
abolish that bug-bear of agricultural experiment, the 
Probable Error of a Single Plot, since by showing exactly 
how the final yield was built up, they show how plot-to- 
plot differences arose, and so convert the probable error 
into.a recognisable error due to definite physiological 
causes. They serve as seasonal or site-records of crop 
behaviour, and, if obtained at a sufficient number of 
observing stations, they could depict the crop of a whole 
country in a form which appeals to the eye, and is further 
of exactly definable precision. Moreover, such records, 
as they grow day by day along the squared paper on 
the laboratory wall, very soon begin to foretell their 
own form; the rates of stem-growth indicate what will 
be the flowering-rates three weeks later, and these again 
determine the rates of fruiting two months afterwards, so 
that a system of cold-blooded Forecasts for any country 
can be developed at small cost of organisation to replace 
the present subjective expressions of more or less dis- 
interested expert opinion. They not only show the 
effect of weather conditions on the crop, but even in 
our present state of knowledge they can occasionally 
be used as records of the environment. Lastly, variety- 
testing can be conducted with absurdly small quantities 
of seed, thus enabling real agricultural trials of new 
strains or varieties to be made at a very early stage of 
the breeding work ; for this purpose the observed groups 
