6 CIRCULAR 95 2, U. S. DEPARTMENT OF AGRICULTURE 



sented each day as a known sample and was included as a coded 

 control at least once in the samples given to each judge. Each judge 

 rated the flavor of 3 samples at each judging session. A variance 

 analysis of the scores was made, mean scores being adjusted for block 

 differences since removal of such differences contributed to the pre- 

 cision of the results. 



For the palatability study of the 11 commercial samples and their 

 control collected in 1951, a randomized design based on 3 replications 

 of a 3 x 4 triple rectangular lattice was used. Six judges rated 

 3 samples per day for 6 days, giving 9 scores per sample. 



The experimental designs for the palatability studies on the 1952 

 samples were selected to provide for analysis of the data by the in- 

 complete block method of Cochran and Cox (4) as well as by the 

 ranked-pair method of Terry, Bradley, and Davis {15). Samples 

 from each field block were evaluated separately because tests indi- 

 cated heterogeneity among data for blocks from the same location. 

 The design used with the samples from the experiment station 

 provided 4 replications of scores per sample per judge; that used with 

 samples from the Hill farm, 3 replications per sample per judge; and 

 from the Turner farm, 4 replications of each sample in each of 3 scor- 

 ing sessions. 



RATINGS BY PANEL B 



Two volunteer tasting panels of 20 persons each from the staff of 

 BPISAE judged 72 samples of peanut butter from the 1950 peanut 

 collections. The samples were rated in groups according to variety 

 and geographic origin. For the rating of each variety, control 

 samples of the same variety and roast were furnished. Each sample 

 was tested by 20 persons, unless otherwise stated. Samples were 

 presented to panel members in pans, 1 pair in the morning, 1 in the 

 afternoon. Of each pair, 1 sample was a commercial sample and the 

 other was a control, but the identity of neither was known to the 

 testers. Panel members tested the 2 samples at random, with an 

 interval of 15 to 20 minutes between samples to allow for detection 

 of aftertaste. 



Panel members were instructed to rate the samples for flavor on a 

 scale of 1 to 10; 10 was described as a perfect score, 5 very strong off- 

 flavor, and 1 inedible. 



Separate analyses of variance were performed on the scores for 

 each variety and geographic region of origin according to a paired 

 plot design in order to avoid, insofar as feasible, differences associated 

 with type or region. 



In the palatability tests on the 1951 samples each taster was 

 given 3 samples of peanut butter at each test period. One of these 

 samples (labeled "S") was identified to the tasters as having been 

 made from nuts known to have been grown in soil free of any insecti- 

 cide. Of the 2 samples given as unknowns, 1 was known to be iden- 

 tical to "S" and the other was from peanuts which may or may not 

 have been grown on insecticide-treated soil. These 3 kinds of 

 samples were recorded as the "identified standard," the "unidentified 

 standard," and the "treated sample." Each taster was asked to 

 indicate which of the 2 unknowns was the unidentified standard. 

 He was also asked to rate all 3 samples on the 10-point scale used for 

 rating the 1950 samples. 



The scores given the samples were used in an analysis of variance. 



