LEAF-AERATION IN NATAL PLANTS. 155 



iirst propounded to the writer by Prof. Bews. The principle is 

 that pioneer types are more variable in their physiological structure 

 and functions than subsequent types, and for that reason able to 

 adapt themselves to the more varied and more variable conditions 

 presented by the habitat during early stages in plant succession; 

 whereas subsequent and climax types are less variable in physio- 

 logical structure and functions, and for that reason limited to the 

 stable and often specialized environmental conditions which obtain 

 in the late and final stages of succession. 



The work of which the main results are here briefly presented 

 is the first attempt at obtaining precise quantitative confirmation 

 of Dr. Bews' suggestive hypothesis. The results obtained indicate 

 that further work along similar lines — the comparison of the 

 physiology and physiological structure of pioneer species with that 

 of subsequent species in various plant successions — may be 

 expected to provide results of equal importance in either confirm- 

 ing or modifying the original hypothesis. 



AGRICULTURAL EXPERIMENT: ITS DESIGN AND 

 INTERPRETATION. 



By E. Parish, B.Sc, 

 Department of Agriculture. 



Read July 15, 1921. 

 Summary. 



1. The variation in yield of single plots similarly treated is 

 so great as to render valueless comparisons made from the result 

 of single plot trials; similarly with experiments with small num- 

 bers of animals, owing to the high variation in the individual. 



2. Replication in agricultural experiments is absolutely 

 necessary. This should bs obtained rather by replication of the 

 plots in a field trial, or animals in a feeding trial, in any year, 

 than by continuing the experiments over a number of years. It 

 is desirable, however, even when the experiments are properly 

 designed, that they be repeated in successive years. 



3. In field trials the replication of the plots must be systematic 

 and so designed as to vitiate the effect of progressive differences in 

 the soil. Suggestions for the conduct of manurial and variety 

 trials and co-operative experiments will be made in the full text of 

 this paper. 



4. In animal feeding trials the individuals must be carefully 

 selected and be uniform in age, breed, sex and condition, and 

 should be at least five in number in each lot, and preferably 10. 

 The animals in the lots under comparison should be subjected to 



13 



