EVOLUTIONARY MECHANISMS 287 



genes with large or small effects affect the fitness of the organism as 

 measured by the number of offspring it leaves, since we can often see 

 that the varying character is one which plays an essential part in Hfe. 

 Differences in fitness caused by genes with strong effects have been 

 occasionally measured in laboratory conditions (see p. 301), but we 

 know very little about even the order of magnitude of the variations in 

 fitness found in Nature. The laboratory measurements give quite a 

 high value for the selective disadvantage of the genes concerned, and a 

 similar value is found for the few genes which seem to be better fitted 

 than the wild type. If these measurements can be taken to give a fairly 

 accurate picture of the state of affairs in Nature, the differences found 

 are quite adequate for the rate at which evolution normally proceeds. 



3. The Measurement of Selective Advantage 



The "advantage" which one variety can have over another must, if 

 it is to be effective, be expressed in the production of offspring. The 

 simplest way of giving a measure of selective advantage is probably 

 that adopted by Haldane.^ If an organism of variety A produces i 

 offspring and one of variety B produces i — ^, then k is the coefficient 

 of selection in favour of ^. 



Fisher^ has given a fuller discussion based on considerations used in 

 life assurance statistics, and this shows more completely what is at 

 issue; the coefficient to which it leads is perhaps theoretically sounder 

 and is easier to apply when generations overlap, as they do in any real 

 case. 



Suppose that the chance of an individual surviving to age jc is /^, and 

 that its chance of reproducing itself between ages x and ;c + ^:c is 

 h^dxy then its chance of surviving and reproducing is l^b^dx. The 

 expectation of offspring from a newly-born individual is clearly the 



poo 



sum of this over his whole life from x = to jc = 00, i.e. l^b^dx. 



This is known as tiie net reproduction rate, and for bisexual species in 

 which the sex ratio is not unity it is better to calculate it for females 

 only. 



If the population is constant in numbers, the net reproduction rate 

 must be unity. If the population is increasing, the expectation of off- 

 spring is more than i. Fisher puts this the other way round. For each 

 child bom now in an increasing population, less than one parent was 



^ Cf. HaJdane 19326. '^ Fisher 1930, cf. Charles 1934. 



