286 



The Joi'RNAL OF Heredity 



character of our orj^anization, we had 

 been so much occupied with the ]^racti- 

 cal appHcation of our discox'cry that 

 we had never had time to elucidate the 

 theoretical question or to lay the result 

 of our observations before the puljlic. 



METHODS IiVH'ROVED. 



It was not without deliberation that 

 the engineer von Nccrgaard, the first 

 scientific director of Svalof, based the 

 work on methodical selection. He sim- 

 plified methods as much as possible, 

 made the experiments more practical, 

 replaced "approximate" measurements 

 of certain characters of the plants by 

 numerical valuation of weight, number, 

 measure, size, etc. In ])ursuit of this 

 aim he invented special instruments 

 and introduced new methods of research. 

 He was thereby able to control ob- 

 jectively the resulting variations and 

 their relative constancy. Thus, me- 

 thodical selection was for the first 

 time made the object of critical scien- 

 tific study, the results of which quickly 

 appeared. 



In 1890, when I became director of 

 the work, I had to meet the desires of 

 the authorities who were subsidizing us 

 by making a hasty general survey of 

 the notes which we had compiled. In 

 spite of the extent and intensity of the 

 selection experiments which had been 

 made, that inspection did not show me 

 the results which we had a right to 

 expect. Rigorous selection pursued 

 during five years had produced only a 

 relative uniformity; we could not show 

 a single new and constant variety- 

 character. And most of all, it wa«: 

 evident that our selected varieties, left 

 to themselves for a year or two, unques- 

 tionably fell back to the condition of a 

 mere mixture of the original varieties. 



Evidently, we were unable to i)roduce 

 what the Swedish farmers wanted — 

 better varieties, which would be con- 

 stant. It was obvious that we must 

 find a new method of oy)eralion which 

 would give better results. 



Such, then, was the desi)c'raLe situa- 

 tion at the Ijcginning of 1891. 



But V)y the summer of 189.^ the difli- 

 culties had been (jvercome ; a new method 

 of work was in use at Svalof, which 



promised perfect results in carrying 

 out the program which the society had 

 laid down. 



During the second year of my resi- 

 dence at Svalof (1889), my botanical 

 instinct had led me to study attentively 

 the numerous strange forms which were 

 found among the old cultivated varie- 

 ties; by jjlanting these strange forms, 

 would it not be possible to produce new 

 varieties as good as those which we had 

 tried to create by the slow and system- 

 atic improvement of the original variety ? 



START ON NEW PLAN. 



The work was started without delay. 

 Preparatory experiments made in 1890 

 with 30, and continued in 1891 with 

 291 different types of spring wheat and 

 vetch had not given satisfactory results, 

 so I increased the number to a thousand 

 in 1892, including all the cereals, peas 

 and vetches which we were studying. 

 At first the result seemed none too 

 encouraging; yet I had occasion to 

 observe, here and there, some novel 

 autumn wheats, which at once opened 

 interesting perspectives to me; they 

 represented only five per cent, of the 

 whole collection. They showed a uni- 

 form and characteristic type, such as 

 we had never previously found in our 

 plantings; it was evident that they 

 differed wholly from the rest in value. 

 The office records showing the origin 

 and treatment of these plots were 

 accordingly examined, and we found 

 that each one of these lots came from 

 a single head of grain, or else, what 

 amounts to the same thing, from a 

 single plant. Our attention was by 

 this means drawn to the im]oortance 

 which the progeny of a single individul 

 has in the j^lant breeding. It was a 

 novel idea for us; apjoreciated from the 

 scientific point of view, it did not fail 

 to inspire us with great hoj^e. 



In order to obtain a definite solution 

 of this problem, we made a general 

 experiment mth sei^arate cultures, in 

 our 1893 planting, using 2000 different 

 numbers and letting each number rei)re- 

 sent the i)lanting of the seeds of a single 

 well-marked individual ])lant. The 

 result was convincing. It left no doubt 

 as to the fact we liad disco\-ered in our 



