1922.] Trials of New Varieties of Cereals. 



341 



may reasonably be expected to hold good under a wide range of 

 external conditions. Illustrations may be given both with regard 

 to yield and quality. 



1. It is fairly clear that a race with a long growing period is likely to give 

 a hetter yield than one with a short growing period, and therefore it is probably 

 desirable to select for any locality a race with as long a growing period as the 

 ordinary climate and farming conditions of that locality permit.* 



2. The writer has shown that with some races of cereals (given complete 

 ripening) a greater proportion of the total dry matter of the plant is 

 accumulated in the grain than with others, and there is consequently a better 

 proportion of grain to straw — an obvious advantage, because grain is more 

 valuable than straw, f 



3. Comparative "strength" in wheat appears to be quite definitely a racial 

 character to a great extent independent of external conditions. J 



4. Immunity from certain diseases is also a racial character more or less 

 independent of external conditions. § 



The above are examples of characters which can be tested in 

 the plant breeding stage. Further investigations of the factors 

 of productivity and of quality together with elaborated nursery 

 testing methods will probably tend to lessen the number of new 

 races which are worth carrying forward to the stage of variety 

 testing in the field. 



Whatever may be the improved characters which the plant 

 breeder has aimed at and hopes that he has obtained, the first 

 essential in systematic comparisons on the field scale is that each 

 comparison shall be, as far as is reasonably and practically 

 possible, free from errors of experiment, and the next essential 

 is that some reliable estimate should be made of the probable 

 extent of the errors that are unavoidable. Unless these con- 

 ditions are fulfilled it is not possible to proceed with any 

 advantage to that interpretation of the results which will still be 

 necessary and which should accompany the publication of them, 

 if they are to be of service to future growers. 



Errors of Experiment. — It is admitted that every separate 

 field experiment is subject to unavoidable errors. If in a com- 

 parison of yields on different plots at any station in the same 

 year the errors of experiment exceed the observed differences 

 between the yields of the races under comparison, the results 

 of that particular trial are valueless as yield trials. In variety 



* Pedigree Seed Corn. Beaven. Jnl. Royal Aar. Soc. of England. 

 Vol. 70, 1909. 



f Breeding Cereals for Increased Production. Beaven. Jnl. Farmers 

 Club, Nov., 1920. 



{ Inheritance of Strength in Wheat. Biffen. Jnl. Ayr. Science, iii. 86, 1908. 

 § Inheritance of Disease Resistance. Biffen. Jul. A (jr. Science, ii, 109, 1907 : 

 iv, 421, 1912. 



