0. 646] THE SYSTEMATIC LOCATION OF GENES 407 



DEFORMED 



Triple 



Rough 



closer pairs. In general, too, we shall have to consider 

 not three genes only, but a large number, lying sufficiently 

 close together for double crossing over to be ignored, 

 the percentage observed between 

 each pair of which gives indirect 

 information as to the position of 

 all the others. 



In its general character the prob- 

 lem resembles those problems in- 

 volving errors of observation, where 

 a smaller number of unknowns are 

 determined from a larger number 

 of inconsistent equations, and which 

 are usually solved by the method 

 of least squares. The practical so- 

 lution depends on the construction 

 of a number of " normal equations " 

 for the unknowns, in which the in- 

 consistencies of the data are prop- 

 erly weighted and made to balance. 

 To make the sum of the squares of 

 the errors of the crossover percent- 

 ages a minimum would, however, 

 be wrong, and the method of least 

 squares is not directly applicabl(\ 

 It has been shown tlmt tho whole of 

 the information sn])])litMl by tho 

 data (2) is made use of liy tho 

 method of maximum likelihood, 

 and by a first approximation the 

 required normal equations may be 

 constructed. 



Beaded 



Peach 



