56 A. D. HERSHEY AND RAQUEL ROTMAN 



bursts with the larger k, and that the effect of dividing each proportion by 4k 

 is to make the results homogenous. These facts justify the use of the correction. 

 Its theoretical significance is clarified in the discussion. 



Table 4 

 Single burst data for ItXr crosses 



k =a measure of disproportion between yields of parental types (Eq. 3). 



x =average yield of the r + recombinant as percent of total virus. 



y =average yield of the r recombinant as percent of total virus. 



n= average burst si?e. 



r(x, y) =coefficient of correlation between proportionate yields of the two recombinants. 



r(n, x+y) coefficient of correlation between burst size and sum of proportions of the two recombinants. 



The variations of x, y and n/100 shown are standard deviations within the sample. The standard errors of the means are 



obtained by dividing these by the square root of the number of bursts. 

 The correction referred to is described in the text. 



The data of table 4 for mixed single bursts confirm fairly well the average 

 data of table 2, except that the yields of recombinants are somewhat greater 

 owing to the exclusion of the unmixed bursts, and that the yields corrected 

 for unequal growth of parental viruses are higher still. 



The chief point of interest is the question of the correlation between yields 

 of the two recombinants in single bursts. This has been measured in terms of 

 the correlation coefficient r (Rider 1939). This measure varies between — 1 and 

 + 1, a value near indicating independence of variates, and values near unity 

 indicating negative or positive correlation, respectively. The data of table 4 

 show clearly that there is no significant correlation between the proportions 

 of the two recombinants in single bursts except for the crosses hXr7 and hr7X 

 wild type. Even for these crosses the correlation is weak and not entirely 

 convincing, especially since the data are not completely unselected (see below). 



250 



