118 
shown in the trend of the yield — curves of controls. The facts 
have to be investigated. Results of utility could be possibly 
procured after a number of years but on the lengths of the period it 
is difficult to hazard a guess. i. 
Some of the facts are being investigated on Tumbuk Estate 
(Mr. F. H. Dale) and will, it is hoped, be communicated from time to 
time but the difficulties of any manager in executing work of this kind 
are very great. 
The case of old coconuts presents a simpler issue. After a 
period of years of bearing the palms reach a stationary maximum- 
production-period which we may assume is a fairly long span ere 
the loss of fruit due to oncoming senility supervenes. In the 
absence of recoi’ds, it would perhaps be safe to assume, that Eifter 
10 to 15 years, all the palms in any field of an estate would have 
reached this stage or have so nearly approached it as to cause little 
fluctuation, and now the disturbing factor so bafflingly difficult in 
the young palm is excluded. Thus it becomes possible to procure 
more comparable data of the yields of blocks of palms over a period 
of time, the collection of which data constitutes the preliminary work 
mentioned above as a basis on which to calculate the size of 
experiments for the attainment of results of accuracy. 
Without going into the long explanation as to the particular 
method it may be stated that this has actually been done on Grapis 
Estate where by taking preliminary records of yield over a period of 
six months, it was possible to initiate experiments on eight blocks 
of 100 trees each, using two as controls, and one pair each for 
the trial of three manures, in the detection of differences over 
15 per cent. 
In Ceylon experiments have for some years been conducted on a 
basis — not of absolute yield — but of proportionate increase in plot- 
yields from period to period, that is, that each plot used has been 
used largely as its own control. There is little doubt that this 
method provides information of great utility. Manurial treatment 
to permanent crops is a cumulative business and records over a 
period of years on blocks of palms irrespective of control plots gives 
very useful information. In the Chilaw experiments recently 
published the yields from January to February in two consecutive 
years were compared, the result of treatment being displayed in 
proportionate increase in yield from January to February of one year 
to January to February of the next year. The method given as in 
vogue there is not without its utility in the detection of differences 
of considerable magnitude, but there its utility (and in this 
particular instance it is well illustrated) seems to end, for the yields 
of the plots untreated show tremendous . variation and the causes of 
these are quite unexplained. It seems impossible to put any general 
interpretation to the results obtained. 
