ANNUAL REPORT, 1937 25 



grandsires and the transmitting indexes of the two granddams. This sum is called 

 the Prediction or Pedigree index. It is built on indexes which are the result of 

 comparing daughters' production with that of their dams. 



Some investigators have claimed that the daughters' records alone are a better 

 measure of the transmitting ability of their sire than is an index. This seems 

 questionable since both sire and dam supposedly contribute about equally to 

 their offsprings' capabilities. 



The Jersey breed publishes a tested sire list which gives the average production 

 of the daughters of all bulls with ten or more daughters with records. A study 

 was made of this material to try to determine its prediction value for bulls of the 

 Jersey breed. Thirty-five bulls (the same number as used in the previous Guern- 

 sey study) whose pedigree could be made complete for two generations were 

 chosen at random. The sires and grandsires of these bulls were in the tested sire 

 list and the average production of their daughters known. The dams had records 

 and the granddams also had records and their sires were in the tested sire list. 

 The material was used as indicated above to obtain the Prediction or Pedigree 

 index. The records as published in the tested sire list are not corrected to a 

 standard number of milkings per day. The daughters of one bull might have made 

 their records on four-times-a-day milking and those of another bull on twice-a-day 

 milking. This is obviously unfair, so the material was used both in its original 

 printed form and also corrected to a standard three-times-a-day milking. 



On the uncorrected basis the composite pedigree index of these thirty-five bulls 

 was 11,684 pounds of milk testing 5.41 percent fat. This is what the daughters 

 of these bulls should have produced if the method is sound genetically and the 

 records complete enough. They actually averaged 10,684 pounds of milk testing 

 5.37 percent. This is a 10 percent error. The individual errors ranged from 

 — 52 to +18 percent. When corrected to standard three-times-a-day milking, 

 the daughters were expected to produce 12,232 pounds of milk testing 5.41 percent 

 fat. They actually produced 11,409 pounds of milk testing 5.37 percent fat; an 8 

 percent discrepancy with the individual errors ranging from —41 to +13 percent. 



In the Guernsey breed, using indexes instead of daughters' records only, the 

 error on the composite thirty-five bulls had been 0.7 percent and the range of 

 individual errors —16 to +14 percent. 



The writer then tried various combinations of parts of these Jersey Pedigrees 

 in an effort to find a practical way of using this material to predict what the 

 daughters of a young bull would produce. The results are indicated in the accom- 

 panying table. 



The task of the breeder is to ascertain, beforehand if possible, the potentialities 

 contained in the germ cells, particularly of breeding bulls. Only in this way does 

 it seem possible to take some of the guess out of breeding operations. From this 

 study it appears that the daughters' records alone are not nearly as efficient for 

 predicting production of the daughters of bulls as are indexes based on a compari- 

 son of daughters' records with those of their dams. 



