June, 1912.] 



each direction throughout all the plots. Bach plot should therefore have contained 

 two sets of 200 seedlings each, but as there were not enough seedlings available to 

 fill plots C, D and E, the vacant space was filled up with other seedlings planted at 

 the same distance, the position of the plants under experiment being care- 

 fully noted. 



The plants were harvested at the beginning of May, 1912, and the result is 

 given in the following table* 



o 



A 1 



2 



B3 

 4 



C5 

 6 



D7 



8 



E9 

 10 



5 a a 



£ 5 Oh 



bo 



24 

 23 

 42 

 30 

 28 

 24 

 18 

 14 

 13 

 12 



155 

 161 

 172 

 130 

 75 

 141 

 66 

 23 

 22 

 17 



-a 



- as 

 2 3 



CD 60 £ 



& 60 ^ 



4,458 

 3,565 

 6,131 

 4,811 

 . 2,232 

 4,004 

 1,251 



848 



508 



194 



=4—1 . 



-CD 



'3 <s s. cs 



60 60 



CD • .— v 



-a ?j3 



3 c( O 



29 

 22 

 36 

 37 

 30 

 28 

 19 

 36 

 23 

 11 



45 

 37 

 61 

 51 

 45 

 43 

 31 

 63 

 34 

 22 



u 



cd a 



cn " a w 



•— !+H CD 1^ 



CD 2 O 



2 5 ft ft 



m « 



18 

 18 

 52 

 52 

 60 

 60 

 60 

 60 

 37 

 37 



.2 a cd » 



a * -5 



o8.2 a& 



SQ.2 



12 

 12 

 10 

 10 



If we disregard plots 8, 9 and 10*, where the irregular result may be attri- 

 buted, in part at least, to the very small number of plants grown, several interest- 

 ing conclusions present themselves. 



The first section of plots A, B, C, etc., contained in each case the offspring of 

 the better of the two plants selected in the preceding generation. In the present 

 generation the yield per acre is better in each case for the offspring of the heavier 

 yielding parent. 



The average weight of grain per plant from the best plot (D) of the preced- 

 ing crop was 17 grammes. The average weight from all the present plots was 29 

 grammes. That is to say the average for the whole of the second crop was nearly 

 double that for the best plot of the first crop. The 1911-12 crop from unselected 

 seed was probably rather better as a whole than the 1910-11 crop, but a consider- 

 able part of the improvement on the experimental plots can only be attributed to 

 selection. 



The calculated crop per acre from the offspring of the best selected plant— 

 the plant which produced 42 grammes of seed in the preceding generation— (Plot B 

 3) exceeds the average crop from all the selected plants by 16 bushels or 35 per cent. 

 There can be no doubt that the crop from this plot (B 3) would exceed the crop 

 from unselected seed by a very much greater amount. 



From the evidence so far available we arrive at the conclusion that the crop 

 from an acre of paddy can be increased 100 per cent, by transplanting alone as com- 

 pared with the usual method of broadcasting. The further improvement which 

 can be obtained in a single generation by rigorous seed selection may safely be put 

 down at least at 50 per cent. 



Selection of plants which tiller well further enables the distance of trans* 

 planting to be considerably increased without loss of crop, thus materially re- 



* Plot E gave a very bad result in both seasons. There was no reason for supposing 

 that there was any material difference in the conditions affecting plant growth between the 

 Remaining plots. 



