THE SYSTEMATIC LOCATION OF GENES BY 

 MEANS OF CKOSSOVER OBSEEVATIONS 



E. A. FISHER 



Inteoductoky 



In tlie construction of a chromosome map, the dis- 

 tances between neighboring genes are equated to the per- 

 centage of crossovers which have been observed between 

 them. Owing to errors of random sampling, and some- 

 times to other disturbing causes, inconsistencies always 

 arise between the distances so determined. For example, 

 in the important data given by Lancefield and Metz for 

 the sex chromosome of Drosophila willistoni [1, p. 241] 

 we have the following values : 



Within such a small range, double crossing over may 

 be ignored; yet it would be wrong to use such inconsist- 

 encies as an argument against the linear arrangement of 

 the genes. For although the true crossover values may 

 be accurately additive, errors of random sampling will 

 certainly disturb the observed percentages. The practi- 

 cal problem is to assign to the distances between the 

 genes values which shall be as far as possible in accord 

 with the whole of the observations available. In other 

 words, we have to make use of as much as practicable, 

 ideally the whole, of the information supplied by the 

 data; giving due weight (i) to the greater accuracy of 

 the values obtained from the larger number of observa- 

 tions, (ii) to the greater accuracy of values obtained from 

 406 



