PLANT VIRUS ASSAY 2 7 



where numbers of lesions are proportional to virus concentration. This 

 appears the simplest method of obtaining direct estimates of the relative 

 concentration of infective virus particles. For such low virus concentra- 

 tions there are established statistical procedures for estimating concen- 

 trations and experimental error. These have been used in association 

 with the egg membrane technique for the bioassay of animal viruses. 

 At present for the reduction of local lesion data and the estimate of 

 experimental errors the most reliable method, introduced by Klecz- 

 kowski, involves preliminary conversion of values to log lesions plus a 

 constant. The constant is estimated in each instance from the local 

 lesion counts themselves. 



A serious cause of variability in the local lesion method is the great 

 range of susceptibility to infection between plants and between leaves 

 on the same plant. This variability is mainly physiological; it is found 

 amongst plants from carefully selected clones as well as amongst plants 

 grown from seed. Methods of designing experiments similar to those 

 used in agricultural field trials overcome much of this variability. The 

 half-leaf method first suggested by Samuel and Bald has been widely 

 used to reduce the variability of lesion counts. Samples of virus to be 

 compared are inoculated on the opposite halves of the same leaves. 

 Comparisons of each sample may be made with a standard sample of 

 virus, or samples may be paired in all possible combinations for inocu- 

 lation on opposite half leaves. 



The limits of accuracy attainable by the local lesion method vary 

 widely according to the details of experimental arrangement, technique, 

 environment, virus, and host plant. It is possible under favorable con- 

 ditions to prove the significance of differences between lesion counts as 

 small as 10 per cent. Where host tissues are variable and techniques 

 inefficient, differences of 50 per cent may not be significant. A range 

 between 15 and 25 per cent is a fair estimate of the extent of differ- 

 ences necessary for significance in careful and accurate work. In the 

 series of papers by Spencer and Price already cited, significant differ- 

 ences were generally found at this level. 



In the ordinary use of the local lesion method many of its theoreti- 

 cal shortcomings are ignored. In effect, competent workers adopt meas- 

 ures to reduce the chances of variation in the constants TV and K, 

 although they don't generally describe their precautions in that fashion. 

 Virus samples as far as possible are suspended in uniform media and 

 inoculated on uniform test plants in a uniform way. When care is 

 taken to eliminate variables it seems, in practice, that comparisons of 

 lesion counts give reasonable estimates of relative virus concentration. 

 It is seldom, however, that any real attempt is made to give these esti- 



