Edible Products. 



222 



[September, 1911. 



These authors rind that the size of the 

 plot has little or no effect upon the 

 result so long as the plot is more than 

 one-hundredth of an acre in extent. 

 They find that in the case of similar- 

 plots the odds against the yield of one 

 plot being better than the average by 

 more than 5 per cent, are 3 to 1. 



The odds are 10 to 1 against a difference of 



more than 10 per cent. 

 Do 44 do do 15 „ 



Do 290 do do 20 



Do 2,700 do do 25 



Now, odds of 10 to 1 do not by any 

 means represent a certainty. If they 

 were commonly so regarded, book- 

 makers would find it difficult to make a 

 living ; consequently the 10 per cent, 

 increase obtained in our supposed ex- 

 periment by no means proves that the 

 manure is doing any good at all. The 

 chances are only 10 to 1 that the whole 

 difference observed is not entirely 

 accidental, having nothing to do with 

 the manure. Still less is it possible to 

 deduce from such an experiment the 

 amount of benefit which the manure 

 is likely to produce. This information 

 can only be obtained by repeating the 

 experiment. 



Wood and Stratton have therefore cal- 

 culated the number of times an experi- 

 ment must be repeated in order to give 

 any desired degree of precision. They 

 assume that odds of 30 to 1 represent a 

 practical certainty, and their conclu- 

 sions are given in the following table :— 



Precision desired 





in Percentage 



Required 



Difference 



Number 



between Yields. 



of Plots. 



20 per cent. 



1 



15 



2 



10 



4 



8 



6 



6 



10 



4 



23 



2 



91 



That is to say, the yield of two experi- 

 mental plots must differ by upwards of 

 20 per cent, before we can safely con- 

 clude from a single experiment that 

 there is any real difference between 

 them, whilst, in order to detect a real 

 difference of 2 per cent., the experiment 

 must be repeated almost a hundred 

 times. With these facts in view, I think 

 it may fairly be conceded that agricul- 

 tural science does not fall much behind 

 the other sciences in point of laborious- 

 ness. The man who deduces the value 

 of a manure, or of transplanting, or 

 some other point from a single experi- 

 ment resulting in a difference of 10 per 

 cent, may fairly be called unscientific, 

 simply because his deduction is not 

 necessarily true. 



2. The Transplanting of Paddy. 

 I now propose to illustrate what has 

 gone before by some actual experiments 

 undertaken by the Botanic Gardens 

 Department- Mr. D. Clouston, in a 

 paper published in the "Agricultural 

 Journal of India," has recorded the 

 following results of transplanting in the 

 Central Provinces on irrigated land. The 

 figures given are pounds weight per 

 acre, and a bushel of paddy weighs 

 about 44 lb. :— 



Transplanted. Broadcast. 



1904- 05 ... 2,000 ... 960 



1905- 06 ... 1,940 ... 1,190 



1906- 07 ... 1,940 ... 1,220 



1907- 08 ... 1,550 ... 1,180 

 This works out at an increase of 63 per 



cent, on the average of four years. Our 

 degree of precision for four experiments 

 being 10 per cent., we may say that 

 under the conditions of the experiment 

 transplanting may be expected to give 

 an increase of not less than 53 per cent, 

 and not more than 73 per cent, over the 

 broadcasting method. Similar results 

 have been obtained elsewhere, and we 

 may take it as established that the trans- 

 planting of paddy leads to a greatly 

 increased crop, although we may remark 

 in passing that the results of Mr. Clous- 

 ton's experiments differ from one another 

 by an amount which cannot possibly be 

 accounted for by the laws of chance. 



Mr. Clouston's paper says nothing 

 about the distance between the trans- 

 planted seedlings, but we understand 

 that the recognized distance in India is 

 about 9 inches. The question of distance 

 is, however, a most important one. It is 

 clear that there must be some particular 

 distance which, other things being equal, 

 will give the largest crop per aeie ; and 

 that if we transplant more closely than 

 this, we not only waste both seed and 

 labour ; but lose on the total crop as well. 

 Moreover, it is better to plant too widely 

 than too closely, since by the former 

 method we save labour and seed, even at 

 the expense of some reduction of crop. 



The discovery of what is actually the 

 best distance entails a very long series 

 of carefully conducted experiments, and 

 my own preliminary experiment in this 

 direction by no means settles the point. 

 I describe it here, partly in order to 

 point out the precautions which must 

 be taken and the sources of error which 

 arise in experiments of this kind, and 

 partly because it is doubtful whether 

 the work will ever be carried to a definite 

 conclusion. 



On the paddy field ab the Peradeniya 

 Experiment Station an area was selected 

 which was bounded by a single bund, 



