2002 Volume 71 Issue 5 Pages 702-709
The floral scents of 26 samples from 13 freesia (Freesia hybrida hort.) cultivars were evaluated by sensory evaluation of untrained panelists (30-64 persons). A seven-point bipolar scale with 32 descriptors was used for the evaluation. Uniformity and reproducibility of ratings on these scales had been confirmed by the evaluation of 30 different types of flower scent by an untrained panel. A five-point monopolar scale with 8 descriptors was also used. These are newly designed scales for evaluating freesia flower scent. As a first step in statistical analyses of sensory data, the panelists (their evaluations) were investigated by principal components analysis (PCA) and the outliers were eliminated. Then the uniformity of ratings among assessors was evaluated by using split-half coefficients. The detectability for differences among freesia samples was confirmed by 1-way ANOVA. As a result, 29 (12 bipolar scales and 5 monopolar scales) descriptors having high concordance and resolution were selected. PCA was carried out on the matrix of mean ratings for the selected sensory attributes. The PC score plots showed that the scent profiles of the freesia samples of the same cultivar were similar to each other, indicating that untrained panelists can be taught to evaluate floral scent of freesia cultivars using original scales.