146 



THE AGRICULTURAL NEWS. 



May 6, 1916. 



be intermixture of distinct varieties showing marked 

 differences in habit and behaviour. (4) Trees often 

 require special branch or soil treatment which intro- 

 duces an unknown fector. (.5) In manurial experi- 

 ments with trees the effect of the manure is cumulative; 

 the applications go on year after year to the same 

 plants in the same soil. Each manurial plot, therefore, 

 is to some extent its own control; its progress in itself 

 •can be measured quite irrespective of any other plot. 

 This is an important feature, and should be appreciated 

 by experimenters since it gives reliable information. 



We may now consider methods. It doing 

 so, it will be taken for granted that actively 

 bearing trees are used in all the experiments, 

 and that they are moderately uniform in every 

 way as far as is practically possible. The question 

 •of the size and number of the plots will not 

 be discussed since this leads to considerations some- 

 what outside the scope of this article. There is, 

 however, one point which must be referred to, and that 

 ■concerns the question of the value of control or 

 untreated plots as a standard of comparison. There is 

 reliable evidence to show that one control plot is 

 totally inadequate in a manurial experiment with 

 trees. As a matter of fact opinion seems to be in 

 some quarters, that control plots are altogether value- 

 less, as will be seen later; at any rate there can be no 

 question that if a control is used, it is necessary to 

 have. more than one in any series of manurial trials. 



Taking coco-nuts first, the method which has 

 been in progress for three years in Trinidad may be 

 -expressed as follows: — 



• Gain ur lo.ss=- (Vi - v.j) - (Cj -c^) • - in 

 Where 



v,= value of yiulit fniiii maiiureil plot 



v.^= ,, ,, ,, ,, same i)l<>t in 1011-12 



(yield befure manure took eftect the following 

 year, the so-cjilled 'natural yield'). 

 Cj= value of tlie average of the two control plot yields 

 c„ = value of tile average of the two ojntrol plot yields 



in lin J -l-_> (natural yield) 

 ■m = cost- of iiianuring. 



This method appears to be a sound one if the 

 1911-12 crop can be accepted as giving the 'natural 

 yield'. Taking the ditfcrence between the control 

 averages is an important feature of the method, and is 

 to be commended. These experiments show that with 

 only one or two exceptions, manuring is accompanied 

 by financial loss, nor does there appeal- to have been 

 any physiological benefit gained by the trees. 



The coco-mit experiments in Nevis have been 

 conducted according to the plan expressed as follows: — 



Increased, or decreased yield = < n+ (n._, — nj) !- — c 



Where 



n = average number of mature nuts per tree picked 

 between the two annual applications of manure. 



n, = average number of manure nuts per tree at begin- 

 ning of experimental year. 



n._.=average numlier (jf mature nuts per tree at end 

 of experimental ye;u'. 



c = the n + (no -- n j) value of control. 



The financial gain or loss would be given by sub- 

 stituting values: for yields, and by subtracting the cost 

 of manuring. 



There is only one control in these experiments, 

 and its employment in interpreting the results is 

 of doubtful value. The n.>— Uj operation is the 

 notable feature of this method, n, is a measure of the 

 natural productivity or yield of the tree in the first 

 year, and of the improved productivity or otherwise in 

 subsequent years, n is a measure of the rate of 

 production. Each manured plot is therefore self- 

 sufficient in itself as far as it goes. The trees in 

 these experiments are young. Had the experiments 

 been delayed until the trees had been in bearing 

 for some time, it would have been possible to have 

 determined the natural yield. Then with the controls 

 in triplicate it would have been possible to compare 

 the yields with the yields from the manured plots. 

 Under those conditions the system would have been 

 superior to that adopted in Trinidad. This, however, 

 would have involved a delay of several years. It may 

 be possible in the future to ascertain the natural 

 yield of certain plots and then, later, to conduct; 

 experiments upon them. 



The Nevis experinHiits indicate so far that con- 

 siderable benefit is to be derived from the use of 

 manures. Poorly bearing trees especially respond to 

 manurial treatment. 



Coming now to the experiments with cacao, the 

 Dominica plan is simple and can be expressed thus:— 



Gain or loss = (v - m) - e 

 Wliere 



v = value of average of results in any manured plot. 



m = cost of manuring. 



c = value of average of results in tlie controls. 



There is no duplication in the main or 'original' 

 series of experiments, but they have now been con., 

 ducted for fourteen years. The 'additional' series, 

 however, started in 1907, provide extra controls and 



