The sensory properties of 16 samples of hard cheese encompassing the m
ain types on sale in the United Kingdom have been studied. A panel of
16 assessors rated the cheese according to five odour, tell flavour an
d five textural attributes. The data were analysed by the Residual Max
imum Likelihood technique and estimates of effects of sample, order of
tasting, carryover and assessor were computed. Significant sample dif
ferences were found for 19 of the 20 attributes. Generalized Procruste
s Analysis was also applied to the individual assessor matrices for od
our, flavour and texture as a method of data simplification that allow
s for differences between assessors. Consensus models were fitted whic
h accounted for 41-58% of the total variation after rotation and scali
ng. These models were of low dimensionality and were used to construct
perceptual space maps. Interpretation of the individual dimensions in
terms of the original attributes was achieved by correlation and proj
ection methods. This allowed the key elements of the sensory propertie
s of hard cheese to be identified.