214 



HOROWITZ AND LEUPOLD 



per cent of the total, it becomes pos- 

 sible to estimate the intensity of the 

 selection which operates against the 

 detection of multifunctional genes. 

 With a random distribution of func- 

 tions, one-half of genes with a single 

 function will be detectable by the 

 usual methods, one-fourth of bifunc- 

 tional genes, and, in general, (Yi)^ of 

 72-functional genes. The original mini- 

 mal estimate of 84 per cent of unifunc- 

 tional genes, based on the observation 

 that this fraction of the mutants re- 

 sponds to single growth substances can 

 now be corrected. A sufficiently close 

 approximation is given by neglecting 

 genes with more than two functions, 

 and we obtain 73 per cent as the cor- 

 rected frequency of unifunctional 

 genes: 



Observed frequency = 



^"^ = 0.84 



84 + 16 

 Corrected frequency = 



84 X 2 



0.73 



84 X 2 + 16 X 4 

 The exact value is given by the first 

 term of a Poisson distribution, and is 

 equal to 0.71 (see Appendix). 



This value is so high, that in spite of 

 the uncertainties in its determination 

 it may be regarded as strongly sup- 

 porting the conclusion that at least the 

 majority of genes controlling biosyn- 

 thetic reactions in Neiirospora are uni- 

 functional. There are several obvious 

 sources of uncertainty in the calcula- 

 tions. First, they should be based on 

 the number of genetically different 

 mutations, rather than on the total 

 number of occurrences; this cannot be 

 done at the present time. Second, the 

 assumption was made that all of the 

 unanalysed mutants, 16 per cent of 

 the total, represent multi-functional 

 genes; this is almost certainly incorrect 

 and biases the calculations against the 

 one gene-one enzyme theory. Finally, 

 the number of temperature mutants is 



too small to give an accurate estimate 

 of the frequency of indispensable 

 functions. It is to the last point that 

 we now turn. 



THE FREQUENCY OF INDISPENSABLE 

 FUNCTIONS IN E. COLI 



It was clearly desirable to obtain a 

 more reliable estimate of the fre- 

 quency of indispensable functions, but 

 to even double the existing number 

 of temperature mutants in Neurospora 

 would be a formidable operation. We 

 therefore turned to E. coli K-12, with 

 the expectation of recovering large 

 numbers of temperature mutants by a 

 modified penicillin technique (Davis, 

 1948; Lederberg and Zinder, 1948). 

 Providentially, this method proved to 

 be unsuited to our purpose: al- 

 though temperature-independent mu- 

 tants were obtained, the yield of tem- 

 perature mutants was zero. This was a 

 fortunate circumstance, since it forced 

 us to adopt a more direct method, one 

 which introduces fewer uncontrolled 

 selective variables into the experiment 

 than would the penicillin technique. 

 The method is simply that of plating 

 out U.V.-treated cells on minimal me- 

 dium and incubating them for 48 hours 

 at 40°. The plates are transferred to 

 25° for an additional 5 days, and the 

 colonies which come up during this 

 second period— so-called secondary 

 colonies— are picked off and tested. 

 This procedure was made feasible by 

 a visual method devised by Dr. Leu- 

 pold which makes it easier to detect a 

 few secondary colonies on a plate con- 

 taining hundreds of primary colonies. 

 Altogether 161 temperature mutants 

 were obtained by this method. Of 

 these, only 37, or 23 per cent, were un- 

 able to grow on the Neurospora com- 

 plete medium at 40° and therefore 

 represent losses of indispensable func- 

 tions. The statistics are shown in Ta- 

 ble 1. 



