96 



Life and Letters of Francis Galton 



there will be a great bias in the returns owing to the limited number of 

 gametic constitutions in the sires. The following table gives the results as 



Table of Colour Inheritance in Horses. 



Percentages taken only to whole numbers. 



published by Galton. In the first line of the series of rows, A (i), we see 

 that for 68 cases of chestnut mated with chestnut all the offspring were 

 chestnut. Galton does not comment on this, but it was the source of con- 

 siderable later controversy. A certain number of matings of chestnut with 

 chestnut taken from Wetherby's Thoroughbred Studbook gave the same result 

 as the first row of Galton's matings ; but a longer series, wherein it was pointed 

 out that the Studbook did contain some instances of chestnut mated with 

 chestnut not producing chestnut, was rejected on the ground that these 

 instances must be due to error of record, a most circular process of reasoning. 

 It is clear from B where we are dealing with a fairly adequate number 

 of crossings both ways that (i) there is not a predominance of sire or dam 

 for chestnut with bay matings, and (ii) bay may contain a factor of chest- 

 nut. If we work out from B the number of bays with a factor for 

 chestnut we find them to be 3T5 °/ o , while 68 '5 °/ o lack that factor. Applying 

 these percentages to the 1900 bay and bay matings in A (ii) we should 



