2 BULLETIN 137, U. S. DEPARTMENT OF AGRICULTURE. 



In America there are no native stocks. The grain-producing areas 

 are relatively new. The varieties peculiar to a section are usually 

 the result of chance introductions. Breeding material from foreign 

 sources is as likely to contain desirable types as is that already at 

 hand. In this investigation, in order to obtain the proper basis upon 

 which to conduct breeding work, stocks were assembled not only 

 from local sources but from all over the world. Many distinct 

 strains were isolated from each stock, for both the local varieties and 

 the foreign introductions were usually either races that had not been 

 purified or that had become mixed after purification. The isolation 

 was accomplished by head and plant selections, which when grown 

 in pedigree rows formed a surprisingly large collection. When to 

 these were added a still greater number from the progeny of hybrids, 

 the problem became one of elimination. The plant selections from 

 their very nature were made more or less arbitrarily, and hundreds 

 of these forms were necessarily duplicates. These duplicates, espe- 

 cially as long as they were not so recognized, were a drain upon the 

 breeder, and it was soon realized that the efficiency of a nursery was 

 measured, not by the number of stocks it carried but by the number 

 it eliminated. 



It was to accomplish this reduction better that the character 

 studies were made. The distinctions found were of two classes, 

 morphological and physiological. The morphological variations 

 were, in the broader divisions, of taxonomic value, and many of 

 them were practically invariable. The physiological characters 

 were, from their nature, more difficult to appraise. They were found 

 to possess not only more widely fluctuating limits, but the limits 

 often overlapped and at times the characters became inseparable. 

 In physiological characters a further distinction was made between 

 permanent and place variations. Some separations were so wide 

 that they never became confusing, while others became evident only 

 when grown under certain conditions of soil and climate. Such dis- 

 tinctions are worthless as taxonomic features, but have proved very 

 valuable as indications of individual qualities in breeding. Even the 

 lack of stability in a character does not destroy its usefulness, as 

 the tendency of a strain to behave in a certain manner under certain 

 conditions may mark an inherent difference. 



It is realized that distinctions of this kind are only a part of plant 

 breeding, and it is not thought that that part is clarified in any 

 great measure. In this paper are given a few of the observations 

 that have been found useful in barley breeding, and with them many 

 that have been found useless. The data upon which the conclusions 

 are based consist of some 200,000 recorded observations, extending 

 over a period of five seasons and embracing experiments at St. Paul, 

 Minn. ; Williston and Dickinson, N. Dak. ; Highmore, S. Dak. ; Moc- 



