t48 TIMEHRIL 
course, be eliminated by carrying the experiments over 
some years, but where the trials are only carried out on 
single plots for each manure or for each variety of cane, 
soil differences are accentuated and not eliminated if the 
same plots are used, as they must be when canes are 
ratooned, year after year. 
We cannot eliminate the soil differences, but we can 
arrange our experiments in such a manner as to minimise 
the effeéts of the differences upon the conclusions which 
we draw from the results. 
I have spent some time in working out a fairly simple 
arrangement of experiments by the use of which the 
planters of the colony may, perhaps, be assisted in their 
choice of manures and seleétion of varieties of canes, 
and which appears to me to offer as complete control 
over soil differences as can be obtained in a necessarily 
restri€ted series of field experiments. 
In order to have any control on the accuracy of the 
results, the expetiments must be conduéted in duplicate 
at least, and each series of plots should have two left un- 
manured, or where varieties are being experimented with, 
two plots planted with Bourbon canes ; and these control 
plots must be so placed that the returns upon them may 
fairly represent the mean fertility of the land under ex- 
periment. 
For experiments on a sugar estate where, as a general 
rule, it will not be feasible to weigh the produce from the 
separate plots but only to measure the juice obtained, 
half an acre will, probably, be found the most convenient 
size for each plot. 
Dam beds should not be included in any system of 
experiments, and whete possible the portion of a field 
