VIRUS PARTICLES AND THEIR FUNCTIONAL ACTIVITY 335 



where Pq is the fraction of samples containing no infective virus particles and 

 e is the base of natural logarithms. A parallel titration of the stock bacterial 

 virus preparation by the usual plaque assay technique will also yield an 

 average value, n^, for the number of infective virus particles per sample. The 

 ratio, n^jui, the "efficiency of plating" coefficient, is usually less than 1 and 

 indicates that fraction of the infected bacteria in suspension which will go 

 on to produce plaques after plating (Ellis and Delbriick, 1939). 



Absolute efficiency of plating can only be estimated after some independent 

 measurement of the concentration of characteristic physical particles has been 

 made (Luria, 1953). Methods of counting particles will be considered later in 

 Section III. 



The precision of assay by the plaque technique is readily estimated, since 

 the variance of a Poisson distribution equals the mean (Isaacs, 1957). For an 

 average plate count of n, the standard deviation equals y'n which, divided 

 by the average count, n, times 100, yields a value for the coefficient of varia- 

 tion. For example, an average plaque count of 100 per plate has a coefficient 

 of variation of 1 % , while 25 plaques per plate yields a standard deviation which 

 is 20 % of the count. The higher the count per plate, the greater the precision 

 of assay within the limits, of course, imposed by the size of the agar plate. 



3. Animal Viruses 



The first extensive use of the local lesion count as an assay method for 

 animal viruses began with the development of a procedure by Woodruff and 

 Goodpasture (1931) for counting pocks or lesions on chick embryo chorioal- 

 lantoic membranes infected with fowlpox virus. The method has since been 

 applied to the assay of many animal viruses including the various poxviruses 

 (vaccinia, ectromelia, myxoma, fowlpox), viruses of the herpes group (herpes 

 simplex, pseudorabies, B virus), the myxo group (some strains of influenza 

 virus, Newcastle disease virus) and the psittacosis-lymphogranuloma group, 

 as well as certain arthropod-borne encephalitis viruses, Rous sarcoma virus, 

 and infectious laryngotracheitis (Isaacs, 1957). 



The technique of inoculation of the dropped chorioallantoic membrane of 

 developing chick embryos is described in detail in the monograph of Beveridge 

 and BvLrnet (1946). Serial dilutions of virus preparation are inoculated on the 

 chorioallantoic membranes of groups of prepared chick embryos and the 

 average number of proliferative lesions formed per membrane is found to 

 bear an approximately linear relationship to the relative concentration of the 

 virus inocula. The above authors felt that with viruses producing well-defined 

 pocks in the range of 5-20 per membrane, assays using 4-6 membranes per 

 dilution will give results accurate to ± 50 % under standard conditions. 



While this type of assay certainly exceeds the precision of end-point assays, 

 it has been recognized that the scatter of counts around the mean for a 

 VOL. 1—23 



