FUNGICIDES 379 



Snapdragon rust. Recently the snapdragon rust caused by Puccinia 

 ardirrhini has been developed as a method representing the rust fungi.^" 

 The method is fundamentally similar to that for the tomato diseases, even 

 to the LD95 point for comparison. However, in this case the disease is 

 suitable for evaluating organic and sulfur fungicides but is resistant to 

 copper compounds, wliich are effective against the tomato diseases. 



Bunt of wheat. The standard method of the American Phytopatho- 

 logical Society for evaluating seed treatments to control bunt of wheat ^ 

 has been adopted for preliminary greenhouse tests. ^^ This disease is of 

 coiu-se a representative of the important seedling infection smut diseases, 

 though somewhat easier to control than others in this group. 



Pea seed decay. The somewhat commonly used method of evaluat- 

 ing seed treatments to control damping off fungi by means of pea seeds is 

 erratic and has not been adequately studied. Research is now in progress 

 in an attempt to understand this disease and its responses more fully. 



Ph3rtotoxicity. Efficient fungicides must not only control diseases but 

 they must be non-injurious to the plant. It has been found that under 

 greenhouse conditions snap beans, tomato, buckwheat, and tobacco are 

 well suited for such tests of phytotoxicity.^^ Responses carmot be recorded 

 in precise numerical terms but they can be grouped into five or seven classes. 

 Comparisons are thus made of the injury follo\^'ing 1 per cent spray or the 

 per cent of spray to give the threshold of injury. 



Cumulative Error Terms 

 The fungicide X replicate test interaction is suggested as the error term 

 for both laboratory and greenhouse methods of testing fungicides. With 

 well-developed and standardized methods this error term is reasonably 

 constant. It has been sho^\^l ^^ that one handicap of small tests — that of 

 inadequate knowledge of the standard deviation — may be overcome if it 

 can be demonstrated that error variance is homogeneous ^^'ith that of the 

 kno^^'n method. A table has been presented to facilitate this homogeneity 

 test. Thus advantage may be taken of past experience with its cumulated 

 error terms and larger degrees of freedom, so that smaller significant differ- 

 ences are required in a given test or comparison. Cumulative error terms 

 for the methods developed and studied in this laboratory have been worked 

 out." 



Correlations Between Laboratory and Greenhouse Methods 



OF Testing Fungicides 



Correlations between the slide-germination method, the three tomato 

 diseases, and wheat smut technique have given interesting results.^" The 

 relative sensitivity of different fungi or diseases depends on the chemical 

 nature of the toxicant. Among tomato foliage diseases a closer correlation 

 exists when only nitrogen, nitrogen plus sulfur on the same carbon atom, 



