Fróðskaparrit - 01.01.2005, Síða 152

150
NEWS AND PROGRESS 2004
variations coincided with the cycle of feed-
ing and spawning. During the spawning
period the texture was soft and the mus-
cle more susceptible for lipid oxidation.
Texture became firmer and fatter during
the feeding period and odour and flavour
characteristics less rancid. These effects
were explained by the chemical composi-
tion and biological parameters. The senso-
ry profile of the products in this part of the
study was influenced by herring size and
age, but not by sex or gonad maturity.
Although the sensory results in this
project were obtained during a long-term
period, they were not subjected to any
assessor “drifting” that could not be re-
moved by mathematic corrections. This
was due to the intensive training and the
use of reference material. By serving the
same product as a reference before each
session assessors could recapitulate the
descriptors easily and recalibrate their
evaluations both qualitatively and quanti-
tatively. Serving the reference as an un-
known sample made it possible to monitor
the performance of the panel. Multivari-
ate data analytical techniques allowed for
quick calculations and results were easily
interpreted with the visual layout. Asses-
sors differing from the rest of the panel
could be identified quickly as well as the
descriptors involved in the deviations.
The lipid content was one of the primary
factors influencing the sensory quality. A
broad variation was found within catches
and some of this variation could be as-
cribed to differences in size. However,
neither herring size nor maturity could be
used to sort herring according to lipid con-
tent, due to very broad distributions with-
in especially maturity classes. As sorting
the raw material according to lipid content
could help the processors to differentiate
the raw material and obtain a better utili-
sation of the resources, efforts were made
to investigate the possibilities of develop-
ing a quick-and-easy non-destructive lipid
measuring method. The lipid content was
measured by chloroform/methanol/water
extraction and correlated to Fatmeter-,
NIR- and NMR-measurements. The re-
sults showed the Fatmeter, NIR and NMR
to have different suitability and applicabil-
ity. The Fatmeter readings rely on the cor-
relation between water and lipid contents.
This relation varied during the maturation
cycle, which altered the Fatmeter values
especially from whole herring. Fatmeter
measurements can therefore only be used
in a broad primary sorting of fillets. High
correlation was found between NIR and
solvent extraction, but measuring point in-
fluenced the prediction results. Measure-
ments performed by NIR and Fatmeter on
meat side of fillets could not predict the
lipid content in herring fillets. The average
lipid content in a fillet or whole herring
could be predicted from single NIR spec-
tra, if measurements were performed in
the middle section on the skin side of the
sample. A high correlation was also found
with the NMR measurements on mince.
The NIR technique showed highest poten-
tiality as a production line measurement
for sorting whole herring or fillets into
more homogenous batches.
Two other important quality parameters
liquid holding capacity and texture were