Three types of water, unprocessed water and two membrane-processed water, w
ere employed to produce Oolong tea-drink samples. A commercial sample of Oo
long tea-drinks using the same materials was also included for testing. A m
odified quantitative descriptive analysis was conducted on a line scale by
15 trained panelists and a hedonic test was conducted by 50 naive panelists
. There were significant treatment effects on 13 out of 19 sensory descript
ive attributes. The commercial sample was higher in red-brown color and alk
aline-flavor, but lower in yellow-brown color, oolong-tea aroma, astringenc
y, bitterness, Oolong-tea flavor, fermented-tea flavor, and astringent afte
rtaste. There were also membrane effects on color, aromas, and flavors of t
ea drinks. The most interesting phenomenon was that the unprocessed sample
was found to be the lowest in the intensity of clearness by trained panelis
ts. This implied that membrane process could improve the clearness of simil
ar tea drinks. It was also found that 12 attributes, which could separate u
nprocessed, commercial and membrane-processed samples well, were important
to the sensory properties of unsweetened Oolong tea drinks in principal com
ponent analysis. Concerning the preference test, the naive panelists liked
the unprocessed sample the least. It seemed that this was affected by the a
ppearance (low clearness) of the unprocessed sample. (C) 2000 Published by
Elsevier Science Ltd. All rights reserved.