892 



SCIENCE 



[N. S. Vol. XXXI. No. 806 



ghums, etc. What is needed is not so 

 much descriptions and detailed classifica- 

 tion of these varieties, as a classification 

 and undei-standing of their principal 

 hereditary characteristics. In other 

 words, the knowledge of them needs to be 

 arranged not only with regard to the ex- 

 isting forms, but also as far as possible 

 with regard to their characters and po- 

 tentialities. Such a monograph does not 

 exist for a single one of our principal 

 crops, though there is an increasing num- 

 ber of contributions to the subject. The 

 field is a vast one in which there is not 

 only a great work to be done in compiling 

 what is known of our cultivated plants, 

 but a greater one in clearing up the many 

 problems concerning their origin. 



In a very different way plant breeding 

 is beginning to do much to better agronomic 

 methods. I have before stated that the 

 most accurate plot work being done in this 

 country is by the plots devoted to fertil- 

 ity investigations. How accurate are 

 these? Hall, of Kothamstead, thinks no 

 results with fertilizers are at all trust- 

 worthy unless the yield difference is at 

 least 10 per cent. In much of the Amer- 

 ican breeding work going on 10 per cent, 

 increase by selection would be deemed good 

 progress. The question is, can any feas- 

 ible system of trial plots measure accu- 

 rately such a difference? Very recently 

 several men have looked into this subject, 

 more or less independently. The most 

 comprehensive work has been done by 

 Lehmaun at the Mysore Experiment Sta- 

 tion, India. Similar work has been done 

 by Lyon at Cornell, Montgomery at Ne- 

 braska, Shoesmith in Ohio and Smith at 

 Illinois. AU of these investigators find 

 a surprising dift'erence in plots due to dif- 

 ferences in soil. On what was considered 

 the most uniform soil at the Nebraska 

 Experiment Station the variation between 



plots on one acre was 35 per cent. — a 

 much greater difference than the breeder 

 of wheat expects to get. Lehmann found 

 differences varying from to 300 per cent. 

 — and further that on many plots the dif- 

 ference was increased or diminished ac- 

 cording to the season. He proposes to 

 use in his work with fertilizers only the 

 plots that give uniform results for at least 

 two similar seasons, a method that he calls 

 standardization. In this country agrono- 

 mists have used mainly the system of 

 check plots — a system which it now ap- 

 pears may be absolutely misleading. In- 

 deed, a study of the check plot records in 

 various experiments shows that they vary 

 in just the way that Lehmann founi his 

 plots to vary. 



Some American agronomists are em- 

 ploying the method of duplicate plots— a 

 plan that is rapidly growing in favor. 

 The number of duplications for the most 

 accurate work will necessarily vary ac- 

 cording to the evenness of the soil, four to 

 sis duplications apparently being neces- 

 sary for very accui'ate results even on 

 faii-ly uniform soil. The subject is, how- 

 ever, one that needs much additional in- 

 vestigation, as the disturbing effects of 

 soil inequalities have evidently been 

 greatly underestimated. 



The results of plant breeding seem 

 likely, therefore, to have a profound ef- 

 fect on agronomy as a whole, demanding 

 as it does both the most accurate plot 

 methods to determine relative yields and 

 a much more intensive knowledge of our 

 crop plants— the material with which 

 breeding must work. 



There is still another botanical method 

 that needs to be brought more intensively 

 into agronomy — namely, the method of 

 pure cultures, which has brought so great 

 results in our knowledge of the lower 

 plants. It is this method that enabled 



