4 



JOURNAL OF THE ROYAL HORTICULTURAL SOCIETY. 



to produce four distinct races or varieties. One of these was to be 

 particularly rich in proteids (which is important for its use as feeding 

 stuffs) ; another variety was to have as low a percentage of proteid as 

 possible (for use in the manufacture of glucose). Another race was to 

 have a high oil content (such a race would be useful in many arts), 

 and the fourth race was to have the smallest possible percentage of oil. 



The experiments began in 1896, and until 1906 the seeds were 

 specially tested and selected for all these four different qualities. The 

 seed selected in 1897 as being highest in protein contained 12'54 per 

 cent, of protein. In 1906 the average of the high protein plot had 

 risen to 14*26 per cent. Individual seeds in 1906 had as much as 

 16'3 per cent. 



In the reverse experiment to reduce the percentage of protein, the 

 results were not quite so remarkable. The average proteid content in the 

 original crop was 10' 92 per cent. The seeds selected as lowest in 1896 

 had 8'96 per cent. In 1906 the average in the low proteid plot was 

 down to 8'64, and seeds had been sown with only 7*21 per cent. 



So that in ten years from an average of 10*92, seed as rich in pro- 

 tein as 16*3 and as low in proteid as 7*21 had been obtained. 



Similar results were obtained in the development of high and of 

 low oil percentages. In 1896 the average oil content was 4' 7 per cent. 

 The first seeds selected for high oil had 5*39 per cent. ; but the average 

 in 1906 had risen to 7*37 and seeds with 7*86 had been obtained. 



The seeds sown for low oil percentage in 1897 had only 4*03 ; the 

 average in 1906 was 2'60 per cent, and seeds with only 2*20 per cent, 

 had been obtained. 



These results are of the greatest interest not only in practice but 

 in theory. They show what extraordinary results can be obtained by 

 steady and careful selection. 



But in any ordinary crop what is wanted is the largest possible 

 return, not so much any particular difference in chemical constitution. 



To improve any ordinary crop, therefore, the obvious course is 

 simply to select the largest and heaviest seed. 



Of recent years many experimenters have tried this method with 

 success. Arthur, for instance, sowed large, medium, and small peas on 

 plots of equal size. He found that the largest seed resulted in plants 

 with more stem, more foliage, but especially a greater proportionate 

 increase in the weight of seeds harvested. 



I could mention many other experiments which confirm these 

 results, but shall only give details of those which I think are of the 

 very first importance. 



I allude to the great series of elaborate cultures made by Arnold 

 Engler and Cieslar with the seeds of Scotch Pine, Larch, and Syca- 

 more in Switzerland. 



One of the most interesting of these experiments, or rather series 

 of experiments, was an attempt to test carefully the difference in 

 general vigour of the descendants of large, medium, and small seeds. 



There were four sets of each size of seed, and as in every case the 



