J. Sci. Food Agric. 1999 (submitted)
Contribution of Volatiles to the Flavour of Oatmeal
Meixue Zhou1, Kevin Robards1, Malcolm Glennie-Holmes2, and Stuart Helliwell1
School of Science and Technology, Charles Sturt University, PO Box 588, Wagga Wagga, NSW 2678, Australia
NSW Agriculture, Agricultural Research Institute, Wagga Wagga, NSW 2650, Australia
Sensory quality of 12 oatmeals as assessed by a trained panel was related to both variety and growing conditions with
variety being the major controlling factor. The volatile profile of the oatmeals was determined by solid phase
microextraction using headspace and gas chromatography. Volatile compounds assessed in this manner were correlated
with the sensory data and chromatographic peak areas accounted for 43 to 94% variations of sensory attributes.
Examination of groats demonstrated that most of the volatiles in oatmeals were induced during heat processing.
Key words: oat; volatiles; SPME, aroma.
Mortlock and Yarran were also grown in Cowra and
INTRODUCTION Parkes in NSW in 1996.
Oats are a significant world crop and growing recognition of
the nutritional value of the oat grain has led to an expanded Preparation of Rolled Oats
usage for food purposes. Oats are consumed as a rolled, Samples of each variety (10 kg) were dehulled and then
flaked breakfast cereal throughout the world. Flavour has stored at 8C for the two days which had to elapse before
been shown to be an important parameter for the processing could proceed. Samples (3.5 kg) were processed
acceptance of oat products to consumers (Heydanek and following normal commercial practice using small scale
McGorrin 1986). Flavour perception depends upon a steaming, kilning and rolling machines in Uncle Tobys
multifaceted series of sensory responses as described by Research & Development Centre (Wahgunyah, Victoria,
Labows and Cagan (1993). However, the most important Australia). The conditions were: steaming for 9 min, kilning
contribution to flavour arises from the volatile constituents at 100C for 45 min and 65C for 15 min, cooling to room
of a food.. In oats, Hrdlicka and Janicek (1964a, b) temperature, resteaming for 5 min and rolling to a thickness
described the presence of carbonyls, bases and an alcohol in of 55 m.
toasted oat flakes and regarded carbonyls as one of the
factors relating to the nutlike flavour. The basic fraction of Sensory testing
the volatiles had strong nutty, earthy, toasted, slightly harsh
Sensory assessment of rolled oats was performed by
and burnt flavours, whereas the neutral fraction was
the trained taste panel at Uncle Tobys R&D Centre,
described as burnt grain, browned, heavy and not oaty
Wahgunyah, Victoria. For sensory assessment, all
(Heydanek and McGorrin 1986).
samples of rolled oats were prepared as per
microwave directions for Traditional Oats with 2/3
Oats lack flavour as collected from the field and the
cup oats and 1 1/3 cup cold water (filtered), then
aroma/flavour that is associated with oat products requires
microwaved on Medium/High setting for 2.5 min
the intervention of a heat process to develop (Heydanek and
with a lid and 2 min without a lid. Mean sensory
McGorrin 1981a and b; Fors and Schlich 1989).
scores were calculated for each sample according to
Nevertheless, steam treatment alone is not effective for
attribute determined by 8 panellists (minimum two
flavour development in oats which requires the direct heat
years experience) in a round table discussion.
in the kiln without which oat products retain a flat, green
Panellists were trained in twelve (one hour) sessions
taste, i.e., raw and slightly bitter (Gansmann and Vorwerck
covering basic tastes and taste threshold levels, fruit
1995). Heydanek and McGorrin (1986) found different
flavours, aromas, savoury flavours, herbs and spices,
processing procedures resulted in different amounts and/or
essences and texture followed by a further twelve
types of volatile compounds.
(one hour) sessions of Quantitative Descriptive
Analysis (QDA) practice with on-going panel
Relatively few studies have reported varietal effects on oat
involvement evaluation. Samples were not ranked for
flavour (Molteberg et al 1996). Our previous study (Zhou et
acceptability (hedonic scale) but for definable
al 1999) reported that a new Australian oat variety, Yarran,
flavour attributes. Attributes were scored on a linear
was unacceptable to consumers because of its poor flavour.
scale of 0 - 100.
As part of a series of studies on oat quality for human food
use, this paper will report the effect of variety and growing
conditions on the flavour of individual oat varieties and the Extraction and measurement
correlations between volatile compounds in oatmeal and the Oatmeal (20 g) was mixed with water (100 mL) in a glass
results of sensory analysis. flask, which was then gently heated in an electric heating
mantle until boiling. Collection of volatiles was commenced
EXPERIMENTAL immediately and performed by headspace solid phase
microextraction (SPME) using a polydimethylsiloxane non-
bonded fibre (100 m film) (Supelco) for routine analyses.
Groat samples Collection time was controlled at 20 min and samples were
Eight varieties of Australian oats (Bimbil, Carralup, maintained at a temperature which kept the porridge bubbling
Cooba, Echidna, Euro, Mortlock, Pallinup & Yarran) slowly during the collection stage. A polyacrylate fibre (85
were grown in Condobolin in New South Wales in 1996; m film) (Supelco) was also examined.
(ROF). Varieties were separated into two groups in respect
Volatiles absorbed on the fibre of the microextractor were to Starch Flavour (SF) and Aftertaste Starch Flavour (ASF).
desorbed in the inlet chamber of a HP 5890 Series II Gas Bimbil, Carrolup, Echidna, Euro and Yarran lacked both
Chromatograph equipped with a flame-ionisation detector SF and ASF while other samples presented significantly
and open tubular column (SE30, 30 m x 0.25 mm, film stronger SF and ASF. The variations of Aftertaste Metallic
thickness 0.25 m) for 1.5 min. The column temperature Flavour (AMF) among varieties were similar to that of MF
program was: 50C held for 1 min, a ramp from 50 to 300C with Echidna the strongest and Cooba the weakest.
at 10C per min and then held at 300C for 4 min. Injector
temperature was set at 220C and detector temperature was Mortlock and Yarran were also grown in Parkes and Cowra
maintained at 250C. Splitless injection was employed. and, as shown in Table 1, growing site had no significant
effect on OAA, OA, SA, OAF, OF and CF of these two
For identification of these compounds, volatiles absorbed by varieties whereas MF, AMF, BF, SF and ASF were affected
the fibre of the microextractor were desorbed in the inlet significantly by growing site. Yarran was particularly
chamber of a VG Quattro GC/MS equipped with a DB-Wax sensitive to growing conditions with the sample from
Condobolin having significantly stronger MF and AMF and
(60 m x 0.32 mm) column. The ion source was 200C, 70 eV
ionisation energy. The column temperature programme was: weaker BF, SF and ASF than samples from Cowra and
Parkes. Compared to Yarran, Mortlock was less affected by
35C held for 5 min, a ramp from 35 to 220C at 3C per min
growing conditions with only the SPF of Mortlock from
and then held at 220C for 15 min. Injector and detector Parkes being significantly stronger than Mortlock from
temperatures were 220C and 250C, respectively. other growing sites.
Statistical analysis Analysis of volatiles
All analyses were performed at least in duplicate and results The analysis of thermally generated aromas by gas
recorded as the mean and standard deviation. The data were chromatography or gas chromatography-mass spectrometry
analysed by single factor or multifactor analysis of variance invariably requires a preliminary isolation of the volatile or
following Snedecor (1967). A principal components analysis aroma compounds from the food. Many techniques have
using SAS software program PROC PRINCOMP and an been reported for this step including solvent extraction,
oblique factor analysis using SAS software PROC VARCLUS steam distillation, simultaneous steam distillation-solvent
were performed to cluster the major peaks into groups of extraction, liquid-liquid extraction (dialysis) and both static
related peaks based on the correlation matrix derived from and dynamic headspace methods. The concentration of
the means. Both principal components analysis and oblique volatiles in oatmeal was very low and preliminary results
factor analysis were carried out by Dr. M. Zack in National showed that both solvent (pentane) extraction and solvent
Centre for Chronic Disease Prevention and Health collection following water distillation were not suitable for
Promotion, Centers for Disease Control and Prevention, detecting flavour compounds. With solvent extraction, the
Atlanta, Georgia, USA. large amount of co-extracted lipid complicated subsequent
analysis and/or mandated further sample clean-up.
RESULTS and DISCUSSION Similarly, distillation/solvent extraction was also found to be
unsuitable. In this case, co-extracted materials were
eliminated but the level of volatile components in the oat
Sensory analysis of oatmeals samples (20 g) was below the detection limit using flame
Oatmeals were assessed by panellists on a non-hedonic ionisation detection. In both cases, peaks in the
scale in which individual characters such as metallic, chromatograms were attributed to solvent and the effects of
creamy or bitter flavour were assessed on a 0 - 100 scale. pre-concentration of large volumes of solvent. Problems
Results of these tests are shown in Table 1. Three aroma related to interference could not be eliminated even with
characteristics and eleven flavour characteristics were high quality solvents. In contrast, solid phase
measured and most of these characteristics differed microextraction (SPME) provided a rapid, solvent-free and
significantly between varieties with the exception of Overall economical approach to recovery of volatiles from small
Aroma (OAA), Overall Flavour (OAF), Soapy Flavour samples of oats.
(SPF) and Aftertaste Oaty Flavour (ATF).
Several variants of the technique were investigated,
Of the eight varieties grown at Condobolin, Carrolup, including immersion sampling from the slurry of oatmeal
Cooba and Bimbil had the strongest Oat Aroma (OA), and water, and collection of the headspace above the
whereas Mortlock and Yarran had the weakest OA. Starch oatmeal. With immersion sampling of heated oats, the fibre
Aroma (SA) of Yarran was the strongest, followed by became coated with starchy materials which could not be
Mortlock, Pallinup and Cooba. No SA was found in Bimbil, removed by washing of the fibre prior to injection into the
Carrolup, Echidna or Euro. GC. Headspace provided a viable alternative but even here
volatile compounds were below the limits of detection in all
The difference in Metallic Flavour (MF) between varieties cases using unheated oats. The large quantities of volatiles
was also significant with Echidna having the highest value found in uncooked oat groats by Heydanek and McGorrin
and Cooba the lowest. Oat flavour (OF) was regarded by (1981a) can be attributed to the large sample size (8 kg) and
panellists as one of the most important contributors to the complicated extraction procedure incorporating a pre-
acceptability. Cooba and Mortlock had significantly concentration step on Tenax GC. In contrast, the present
stronger OF than other varieties, while Yarran and Bimbil study used small samples (20 g) with a relatively simple
had the lowest value of OF. Most samples presented low extraction. Nevertheless, the production of an oat product
Bitter Flavour (BF) while Echidna had relatively higher suitable for cooking requires steaming, kilning and rolling
values for BF. Creamy Flavour (CF) is another important and, more importantly, oatmeal is consumed after cooking.
contributor to acceptability for human consumption with a Each step from raw groat to consumer product is expected
high CF being desired by consumers. Bimbil and Echidna to develop flavour components (Heydanek and McGorrin
had significantly lower CF than all other varieties. Bimbil 1986) and it is reasonable to examine the volatiles from a
was the only variety that had very strong Raw-Oat Flavour cooked product particularly where these are to be related to
results of sensory analysis. measured by panellists and the predicted value from
independent peak areas as shown in Table 4 which were
Two types of fibre were examined for their suitability in selected by regression analysis. Since the peak areas were
extracting volatiles from the headspace of cooked meals. It standardised, a larger value in Table 4 means a greater
was found that the non-polar polydimethylsiloxane coating contribution of the peak to the flavour measurement. For
was more suitable which agrees with previous observations example, the value of OA was given by a multiple linear
(Coleman 1997) in which a non-polar polydimethylsiloxane regression equation of the following form: OA = 37.36 +
coated fibre was found to be very effective in the 5.49 Cluster C (P2) - 3.41 Cluster B (P8) - 2.89 Cluster E
qualitative analysis of volatile and semivolatile Mailard (P14) + 1.96 Cluster D (P20) + 7.87 Cluster F (P21) - 7.72
reaction products. Many volatiles were detected in cooked Cluster A (P27) where P2, P8 ... P27 are the standard peak
oat samples (see Materials and Methods) and, as predicted, areas. From the magnitude of the multipliers (7.87 and
the number and intensity increased in going from a groat 7.72), Cluster A and F had the greatest contribution to OA,
sample to oatmeal (Figure 1). Some varietal differences followed by Clusters C, B, E and D.
were also observed as illustrated in Figure 2 and discussed
below. All six peak clusters showed significant effect on aroma.
Cluster B (P6), C (P7) and F (P13) presented significant but
Relationships between volatiles and flavour attributes minor contributions to OAA. Cluster F (P21) and A (P27)
There were more than 50 volatile compounds detected in were found to be most important contributors to OA and
cooked oat porridge (Figure 2). The 46 most abundant and SA. Cluster C (P2), D (P20) and F (P21) positively
easily separated peaks were correlated (as raw peak areas contributed to OA but negatively to SA. In contrast, Cluster
without any correction for detector response) with the A (P27), B (P8) and E (P14) was negatively contributed to
flavour data as evaluated by taste panellists. Table 2 shows OA but positively to SA.
the ANOVA (F values) of all the selected peaks. F values
larger than 2.8 or 4.4 indicate a significant (P < 0.05) or Both aroma and flavour of the porridge were also attributed
very significant (P < 0.01) difference between varieties. to all the six clusters with cluster A (mostly P27) and cluster
Varieties differed significantly in all peak areas except F (mostly P21) having the greatest contributions. Cluster A
peaks 5 and 22 perhaps due to lower signal to noise ratios. contributed positively to OF, SF, SPF, and ASF and
There were strong co-correlations between the peaks. For negatively to ROF, AMF and AOF while cluster F showed
example, peak 1 (P1) was significantly correlated with P4, the opposite effect on these measurements. Cluster B (P1,
P5, P6, P8, P10, P11 and P16; P2 was significantly P5 or P8) showed a positive effect on CF, SF and ASF and a
correlated with P3, P7, P9 and P31 (Table 3). Thus, negative effect on MF, OF, AMF and AOF while cluster C
regression analysis was inappropriate to identify the peaks (P2, P7 or P15) had the opposite contribution to these
significantly affecting flavour because of multicolinearity flavours. Cluster D effected MF, SPF positively and OF, SF,
between the peaks themselves. From the correlation matrix AMF and ASF negatively. Similar to cluster D, cluster E
(selected data shown in Table 3) based on the means, contributed positively to MF and negatively to OF, SF and
principal components and oblique factor analyses were AOF.
performed to cluster the 46 peaks into six groups of related
peaks that might then be used as independent variables in Identification of volatile compounds
developing a multiple regression correlation between peak The identity of peaks is not essential to an oat breeding
area and flavour. This procedure yielded six peak clusters as program since most flavour attributes were predictable from
follows: Cluster A: P23 - 32, P34 - 38, P41 and P46; Cluster peak areas alone. However, identification of the volatiles is,
B: P1, P4 - 6, P8, P10 - 12 and P16; Cluster C: P2, P3, P7, nevertheless, desirable and essential to studying changes
P9 and P15 where P15 is negatively associated with other during processing and storage. Kovàts Indices of the 46
peaks in the cluster; Cluster D: P19, P20 and P22 where peaks seen in chromatograms of oatmeal (Figure 2) on SE30
P22 is negatively associated with other peaks in the cluster; are given in Table 5.
Cluster E: P14, P33 and P42 - 45; and Cluster F: P13, P17,
P18, P21, P39 and P40. In groats, relatively few peaks were detected (Figure 1) with
the major compounds being identified from retention data as
This indicates that the original 46 peaks ca be represented palmitic acid and eicosane. Smaller peaks included hexanal
by the 6 clusters as independent variables in the multiple and nonanal and these assignments were supported by mass
regression. One of the peaks from each of these clusters, spectral data. After processing of the groats into oatmeal,
which had a relatively high correlation or high variation more volatile compounds were produced (Figure 1) and
between varieties, was then selected as an independent from the retention times, the major compounds included
variable. In order to get comparable coefficients (the larger hydrocarbons (alkanes) (decane, hexadecane, heptadecane,
the coefficients, the more important the variables), octadecane, nonadecane and eicosane), alcohols (1-
regression analysis was performed after standardising all the heptanol, 3-octanol, 2-octanol and 1-octanol), acids
peak areas to a mean of zero and a variance of 1.00 by (heptanoic acid, octanoic acid, nonanoic acid, dodecanoic
subtracting the mean of the twelve means from each acid and palmitic acid) and aldehydes (hexanal, nonanal and
individual mean and dividing by the standard deviation of decanal) (Table 5) and these assignments were also
these twelve means. supported by mass spectral data. The most pronounced
flavour effect is expected to occur during the processing of
Regression analysis showed that except for OAF and BF, oat products due to heat treatment. Most of the aldehydes
other flavour measurements such as OA, SA and SF could are products of reactions logically expected to take place in
be predicted by a suitable linear combination of the area of the boiling water heat treatment (Heydanek and McGorrin
the representative peak from each group with from 43% in 1986). The hydrocarbons are derived from oxidative
the case of CF to 94% in the case of OA (adjusted R- decomposition of lipids which are believed to make no
squared statistic, percent variation determined by significant significant flavour contribution to foods (Shahidi et al
peaks, Table 4) of the variations being explained by the 1997). Similarly, some alcohols and aldehydes could also be
independent variables (uncorrelated peak areas). Figure 3 derived from the oxidative decomposition of unsaturated
illustrates the correlations between the actual value fatty acids such as oleic, linoleic and linolenic acids
(Heydanek and McGorrin 1986). mass spectroscopy identification of volatiles from rancid
oat groats. J Agric Food Chem 29 1093-1095.
Heydanek M G, McGorrin R J 1986 Oat flavor chemistry:
CONCLUSIONS principles and prospects. In: Oats: Chemistry and
Sensory testing requires well-trained panellists and is Technology, ed Webster F H. American Association of
expensive, time consuming and difficult. Limited samples Cereal Chemists, St. Paul, MINN.
can be tested which makes sensory testing of progeny from Hrdlicka J, Janicek G 1964a Carbonyl compounds in toasted
oat breeders an inappropriate method for selecting a good oat flakes. Nature 201 1223.
quality variety. This study showed that 43 to 94% of the Hrdlicka J, Janicek G 1964b Volatile amines as components
variations in most flavour parameters could be predicted of toasted oat flakes. Nature 204 1201.
from less than six peak areas from six independent peak Labows J N, Cagan R H 1993 Complexity of flavor
groups in the headspace chromatograms. This makes it recognition and transduction. In: Food Flavor and
possible for a breeder to identify the flavour profiles of a Safety. eds Spanier A M, Okai H, Tamura M. American
large number of lines in earlier generations since the Chemical Society, Washington, DC.
measurement of the volatiles can be easily performed on Molteberg E L, Solheim R, Dimberg L H, Frølich W 1996
relatively small samples. Variation in oat groats due to variety, storage and heat
treatment. II. Sensory Quality. J Cereal Sci 24 273-282.
Shahidi F, Aishima T, Abou-Gharbia H A, Youssef M,
Shehata A A Y 1997 Effect of processing on flavor
precursor amino acids and volatiles of sesame paste
Acknowledgements (Tehina). J Am Oil Chem Soc 74 667-678.
The authors gratefully acknowledge the assistance of Uncle Zhou M X, Robards K, Glennie-Holmes M, Helliwell S 1999
Tobys Ltd, and the support of a grant from the Grains Effects of oat lipids on groat flour pasting properties. J
Research and Development Corporation of Australia. Sci Food Agric (In press).
REFERENCES Captions for Figures
Coleman W M 1997 A study of the behavior of polar and
nonpolar solid-phase microextraction fibers for the Figure 1 Chromatograms comparing the volatile compounds
extraction of Maillard reaction products. J Chromatogr in the headspace of Yarran groats (a) and Yarran oatmeal
Sci 35 245-258. (b). Vertical scale is the same in (a) and (b)
Fors S M, Schlich P 1989 Flavour composition of oil obtained
from crude and roasted oats. ACS Symposium Series 409 Figure 2 Chromatograms comparing the volatile profile of
121-131. cooked oatmeals of cv. Mortlock (a), Yarran (b) and Euro
Gansmann W, Vorwerck K 1995 Oat milling, processing (c). The major peaks are identified by number as peak 1 to
and storage. In: The Oat Crop, ed Welch R W. 46.
Chapman and Hall, London, pp 369-408.
Heydanek M G, McGorrin R J 1981 Gas chromatography- Figure 3 Correlations of flavour measurement between the
mass spectroscopy investigations on the flavor chemistry value scored by panellists and the value predicted from
of oat groats. J Agric Food Chem 29 950-954. chromatographic peak areas of volatiles.
Heydanek M G, McGorrin R J 1981 Gas chromatography-
Table 1 Sensory scores of oatmeal samples as determined by a trained panel
OAA OA SA OAF MF OF BF CF ROF SF SPF AMF A
Bimbil 46.7 41.8a 0.0d 53.7 20.8cd 34.6bc 8.125abc 5.6d 41a 0.0d 0.0b 14.8b 25
Carrolup 47.1 43.2a 0.0d 51.8 26.7bc 37.0abc 11.1abc 25.5abc 0.0b 0.0d 0.0b 19.1ab 20
Cooba 48.5 41.2a 15.7c 54.2 16.5d 40.1ab 3.3c 26.5ab 6.2b 20.5c 0.8b 11.5b 22
Echidna 44.0 38.8ab 0.0d 55.0 36.2a 37.0abc 17.7a 3.1d 2.8b 0.0d 0.0b 24.0a 19
Euro 43.8 40.6ab 0.1d 52.0 18.6cd 37.4abc 6.1bc 17.3bc 1.1b 0.0d 0.0b 13.9b 21
Mortlock 44.8 34.3bc 22.5b 52.8 19.7cd 36.8abc 10.1abc 22.9abc 7.6b 25.3bc 5.5b 14.1b 25
Pallinup 48.8 38.7ab 22.5b 52.7 22.8bcd 35.8abc 3.7c 19.8abc 6.6b 26.5b 0.0b 19.5ab 21
Yarran 48.6 34.2bc 33.0a 54.0 29.3ab 35.0abc 4.7bc 16.9c 3.1b 2.6d 0.0b 23.5a 18
Mortlock 46.5 34.3bc 22.6b 54.1 18.6cd 39.5ab 11.7abc 28.6a 5.6b 24.2bc 0.1b 17.1ab 24
Yarran 47.5 35.0bc 28.8a 56.8 21.0cd 32.4c 14.3ab 19.0bc 6.3b 29.5b 0.0b 16.0ab 23
Mortlock 45.0 34.2bc 22.2b 55.1 20.0cd 41.4a 5.5bc 23.9abc 6.5b 20.1c 24.8a 14.0b 25
Yarran 47.0 31.3c 32.6a 52.5 19.6cd 35.5abc 14.3ab 22.0abc 5.1b 36.1a 0.0b 14.9b 20
P-value 0.196 0.000 0.000 0.307 0.000 0.039 0.003 0.000 0.000 0.000 0.000 0.008 0.
*Values followed by the same letter are not significantly different from each other, SSR test at the 5% probability
OAA: Overall aroma; OA: Oat aroma; SA: Starch aroma; OAF: Overall flavour; MF: Metallic flavour; OF: Oat flavour;
BF: Bitter flavour; CF: Creamy flavour; ROF: Raw oat flavour; SF: Starch flavour; SPF: Soapy flavour; AMF: Aftertaste
metallic flavour; AOF Aftertaste oat flavour; ASF: Aftertaste starch flavour.
Table 2 ANOVA of the peak area (F value)
Peak P1 P2 P3 P4 P5 P6 P7 P8 P9 P10 P11 P12 P13 P14 P15 P16 P17
F value 11.23 12.47 17.58 6.17 1.51 10.8 6.66 7.22 4.65 5.32 5.40 2.80 3.04 6.46 3.23 4.69 4.90
Peak P24 P25 P26 P27 P28 P29 P30 P31 P32 P33 P34 P35 P36 P37 P38 P39 P40
F value 9.96 17.36 47.09 31.96 18.48 29.66 4.06 44.30 53.22 3.12 15.97 25.63 27.83 24.86 14.81 13.67 3.49
*F0.05(11, 11) = 2.82; F0.01(11, 11) = 4.46)
Table 3 An example of correlations between the various peaks, shown here between the first two peak
areas and other peak areas
Peak 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
1 1.00 -0.24 -0.35 0.97 0.80 0.86 -0.11 0.93 0.20 0.93 0.79 0.49 0.28 -0.10 0.26
2 -0.24 1.00 0.97 -0.19 0.15 -0.06 0.80 -0.46 0.79 -0.18 -0.18 -0.48 0.31 0.18 -0.41
Peak 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38
1 0.17 -0.23 -0.06 -0.23 -0.11 -0.10 0.08 -0.25 -0.19 -0.18 0.01 -0.12 -0.32 -0.25 -0.15
2 0.42 0.47 0.11 0.49 0.43 0.52 0.54 0.70 0.53 0.47 0.43 0.50 0.21 0.67 0.60
*r (0.05) = 0.58; r (0.01) = 0.71.
Table 4 Peaks significantly contributing flavour components (the larger the number the more important the contribution of
Peak OAA OA SA OAF MF OF BF CF ROF SF SPF AMF AOF ASF
I 46.5 37.4 16.7 22.5 36.9 19.3 7.6 15.4 2.6 16.9 22.4 11.1
A -7.77 42.67 4.15 -9.69 21.79 5.82 -7.24 -10.5 17.63
(27) (P27) (P41) (P27) (P27) (P30) (P27) (P27) (P27)
B 1.00 -3.41 14.91 -2.10 -1.95 3.34 8.05 -3.49 -1.14 7.37
(P6) (P8) (P8) (P5) (P8) (P1) (P1) (P5) (P1) (P8)
C 1.29 5.49 -22.7 2.50 6.61 -16.4 4.05 4.71 -11.2
(P7) (P2) (P2) (P15) (P2) (P2) (P7) (P2) (P2)
D 1.96 -9.37 1.64 -0.95 -11.0 8.08 -1.88 -8.89
(P20) (P20) (P20) (P19) (P20) (P12) (P22) (P20)
E -2.89 12.14 4.08 -4.42 -4.22 -3.21
(P14) (P14) (P14) (P45) (P33) (P14)
F -2.44 7.87 -44.1 -5.18 7.72 -17.9 -7.93 5.62 9.79 -14.6
(P13) (P21) (P21) (P39) (P39) (P21) (P17) (P21) (P21) (P21)
R2** 0.67 0.94 0.74 0.78 0.85 0.43 0.64 0.78 0.68 0.92 0.85 0.80
* I = Intercept of the regression equation.
** R2 = adjusted R-squared statistic.
The predicted value of each sensory measurement could be calculated by a regression equation. For example, OAA = 46.5
+ 1.00 P6 + 1.29 P9 - 2.44 P13. A blank value in the table signifies that the related peak groups had no significant
contribution to the flavour measurement.
Table 5 Kovàts retention indices on SE30 of the 46 peaks detected by flame ionisation detection from the headspace of
Peak Retention KI Possible Compounds Peak Retention KI Possible Compounds
No. time No. time
1 3.78 822 Hexanal * 24 15.54 1458
2 5.18 877 25 15.76 1477
3 5.66 897 26 16.98 1583
4 5.78 902 2-Nonane ?? 27 17.07 1591 Dodecanoic acid ??
5 6.92 951 1-Heptanol ?? 28 17.18 1601 Hexadecane *
6 7.59 981 Octanal ?? 29 17.69 1645
7 7.69 985 3-Octanol ?? 30 17.88 1662
8 7.80 990 2-Octanol ?? 31 18.32 1700 Heptadecane *
9 8.06 1003 Decane * 32 18.40 1707
10 8.87 1041 33 18.56 1721
11 9.11 1053 Octanol * 34 18.85 1746
12 9.67 1081 Nonanal * 35 19.18 1775
13 10.67 1132 36 19.43 1776
14 11.29 1165 Heptanoic acid * 37 19.50 1802 Octadecane *
15 11.42 1172 Decanal * 38 19.63 1814
16 12.37 1225 39 19.79 1827
17 12.89 1255 Octanoic acid * 40 20.08 1844
18 13.26 1277 41 20.63 1891 Nonadecane *
19 13.90 1316 42 20.96 1917
20 14.16 1336 Nonanoic acid ?? 43 21.33 1953 Palmitic acid *
21 14.37 1355 44 22.47 2071
22 14.54 1370 45 22.74 2101 Eicosane
23 15.25 1432 46 22.84 2112
* Retention time match the standard; ?? deducted from the Kovàts indices.