332 C. E, SCHWERDT 



point is desired. De Beer (1945) lias published a method for the graphic 

 estimation of the standard error (SE) of individual ID50 values from dose- 

 response curves where the per cent incidence of positive responses is plotted 

 on a probit scale versus log dilution of virus. A simpler method requiring no 

 elaborate scales and nomographs has been devised by Pizzi (1950), who 

 developed a simple formula for the determination of the SE of an individual 

 ID50 estimated by the Reed and Muench method. Estimates of SEidjo by 

 this formula have shown excellent agreement with estimates of standard 

 deviations from replicate titrations (Schwerdt and Merrell, 1952). From the 

 SE of two individual titration end points, the SE of the difference between the 

 titers can be computed and thus a test of significance applied to this difference. 



In a titration based upon an aU-or-none response where average host 

 susceptibility varies significantly and the incidence of positive responses is 

 highly irregular, particularly in the neighborhood of the ID50, some measure- 

 ment other than frequency of response must be made and correlated with dose 

 in order to quantitate the relative infectivity of a virus preparation. The most 

 useful relationship has usually been found to be the incubation period and log 

 dilution inoculated, i.e., a correlation of the length of time necessary for the 

 detection of symptoms with log dose. This method has not found general 

 application but has proved successful for the assay of mouse encephalomye- 

 litis viruses (Gard, 1940), various tumor viruses such as rabbit papilloma 

 virus (Bryan and Beard, 1939), avian erythromyeloblastic leukosis (Eckert 

 et al., 1954), and Rous sarcoma virus (Bryan, 1956), and several viruses of the 

 psittacosis-lymphogranuloma venereum group (Golub, 1948; Gogolak, 1953; 

 Crocker, 1954). Some of these viruses exhibit a linear relationship between 

 incubation period and log dilution of virus. With others, a linear correlation 

 is found between log virus concentration and the reciprocal of the incubation 

 period. 



It is sometimes possible by a great expenditure of time and animals to 

 determine an ID50 for such viruses. Then, if a plot of the log ID50 inoculated 

 against incubation period is linear, one can estimate an ID50 value for a virus 

 preparation from the average incubation period observed after inoculation of 

 a single dilution (Golub, 1948). Generally, however, the incubation period or 

 its reciprocal is plotted as a function of log dilution of virus for both the 

 unknown and a standard sample and a relative potency estimated from the 

 ratio, at a given response level, of the log dilution unknown/log dilution 

 standard. The antilog of this ratio, then, expresses relative potency in arith- 

 metic terms. Relative potencies are most readily estimated if the response 

 (incubation period or its reciprocal) is a linear function of log dose and the 

 slopes of the two curves are the same. The application of this indirect method 

 of bioassay of viruses has been presented with numerous examples in an 

 excellent review by Bryan (1957). 



