Journal of Oleo Science

Copyright ©2014 by Japan Oil Chemists’ Society

doi : 10.5650/jos.ess14064
J. Oleo Sci. 63, (9) 867-873 (2014)

Use of Differential Scanning Calorimetry to Detect

Canola Oil (Brassica napus L.) Adulterated with Lard
Stearin
Jalaldeen Mohammed Nazrim Marikkar1, 2＊ and Sohel Rana3
1
Halal Products Research Institute, Universiti Putra Malaysia, 43400 UPM Serdang, Selangor D.E., Malaysia
2
Department of Biochemistry, Universiti Putra Malaysia, 43400 UPM Serdang, Selangor D.E., Malaysia
3
Faculty of Science/Institute for Mathematical Research, Universiti Putra Malaysia, 43400 UPM Serdang, Selangor D.E., Malaysia

Abstract: A study was conducted to detect and quantify lard stearin (LS) content in canola oil (CaO) using
differential scanning calorimetry (DSC). Authentic samples of CaO were obtained from a reliable supplier
and the adulterant LS were obtained through a fractional crystallization procedure as reported previously.
Pure CaO samples spiked with LS in levels ranging from 5 to 15% (w/w) were analyzed using DSC to obtain
their cooling and heating profiles. The results showed that samples contaminated with LS at 5% (w/w) level
can be detected using characteristic contaminant peaks appearing in the higher temperature regions (0 to
70℃) of the cooling and heating curves. Pearson correlation analysis of LS content against individual DSC
parameters of the adulterant peak namely peak temperature, peak area, peak onset temperature indicated
that there were strong correlations between these with the LS content of the CaO admixtures. When these
three parameters were engaged as variables in the execution of the stepwise regression procedure, predictive
models for determination of LS content in CaO were obtained. The predictive models obtained with single
DSC parameter had relatively lower coefficient of determination (R2 value) and higher standard error than
the models obtained using two DSC parameters in combination. This study concluded that the predictive
models obtained with peak area and peak onset temperature of the adulteration peak would be more
accurate for prediction of LS content in CaO based on the highest coefficient of determination (R2 value)
and smallest standard error.

Key words: food adulteration, DSC, lard detection, lard stearin, thermal analysis

1 Introduction There are many reported cases of adulteration practices

Food adulteration is a common issue prevalent in many
parts of the world. Food adulterations by the commercial
sector have been mainly due to monetary benefits without
considering the ethics and social responsibility1）. Generally,
adulteration practices would become more intense, when
there is scarcity in the supply of certain food commodities.
Hence, authentication of food quality at various stages of
production, processing and distribution has become an es-
sential part of the halal food quality assurance2）. Food au-
thentication is a process by which a food is cross-checked
whether it meets its specified quality indices3）. This would
not only help consumers to get the right quality product
for the price they pay for but also safeguard them from any
adverse effects on their health.
for highly priced vegetable oils and fats such as virgin olive
oil4）, cocoa butter5）, etc. As a consequence, there has been
a great deal of interest among researchers to develop ana-
lytical methodologies to detect adulteration practices.
Owing to the need of compliance with halal and kosher
food regulations6）, many studies were undertaken to detect
animal fat contaminations in different vegetable oils using
various analytical techniques7）. Among the different instru-
mental approaches, DSC has been proven to be a useful
tool for identification of LD adulteration in oils such as
palm8）, canola9）, soybean10）, sunflower11）, etc. While DSC
cooling curves were found to be useful to detect LD adul-
teration in palm oil and palm olein, DSC heating curves
were successful in detecting LD in oils such as canola,
soybean, sunflower, etc. Although a fair amount of research

＊
Correspondence to: J. M. N. Marikkar, Halal Products Research Institute, Universiti Putra Malaysia, 43400 UPM Serdang, Selangor
D.E., Malaysia and Department of Biochemistry, Universiti Putra Malaysia, 43400 UPM Serdang, Selangor D.E., Malaysia
E-mail: nazrim_marikkar@upm.edu.my
Accepted June 4, 2014 (received for review April 10, 2014)
Journal of Oleo Science ISSN 1345-8957 print / ISSN 1347-3352 online
http://www.jstage.jst.go.jp/browse/jos/  http://mc.manusriptcentral.com/jjocs

867
 J. M. N. Marikkar and S. Rana
has been dedicated to detect LD in oils and fats, the
amount of effort taken to detect LS in these oils is still
scanty. In a preliminary study, Yanty et al.12）compared the
common characteristics features of LD and its fractionated
form LS. Although there were significant differences
between LD and LS with regard to their physical character-
istics, some of the original chemical characteristics of LD
were still retained in LS12）. This preliminary finding has
urged us to investigate whether the DSC approaches
adopted for detection of LD in some of the above men-
tioned oils could still be valid for the detection of LS in
them. Hence, the objective of this study was to investigate
the possibility of detection of LS in CaO using DSC cooling
and heating curves.
2 Material and Methods
2.1 Materials
LD was extracted using adipose tissues of swine collect-
ed from local slaughter houses according to the method re-
ported previously by Marikkar et al.8）. Three authentic
samples of CaO were supplied by Sime Darby Food Market-
ing Sdn Bhd, Selangor, Malaysia. All chemicals used in this
study were of either analytical or HPLC grade.
2.2 Methods
2.2.1 Fractional crystallization of LD
Fractional crystallization of LD was carried out using
acetone as solvent medium according to the method re-
ported by Yanty et al.12）. LD was melted at 60℃ and mixed
Fig 1　DSC heating curves of lard stearin (A) and lard (B), and DSC cooling curves of lard stearin (C) and lard (D).
868
J. Oleo Sci. 63, (9) 867-873 (2014)
with acetone in 1:2（w/v） ratios. The solution was boiled at
60℃ until become uniformly dissolved and left at 5±1℃
for 24 h to crystallize. The precipitated fat was filtered off
to give a high melting fat fraction（LS）followed by evapora-
tion of the mother-liquor under reduced pressure to yield a
liquid called low-melting fraction.
2.2.2 Preparation of binary blends
The extracted LS were blended with CaO at different
ratios. A total of twelve samples were prepared:（A）100:0,
（B）95:5,（C）94:6,（D）93:7,（E）92:8,（F）91:9,（G）90:10,（H）
89:11,（I）88:12,（J）87:13,（K）86:14,（L）85:15（ w/w）, and
identified by the mass ratio of CaO to LS. After prepara-
tion, all samples were kept under cold storage at 4℃. Prior
to analyses, the samples were removed from the cold
storage, and left static at room temperature for 0.5 h
before being warmed at 60℃ until they became completely
molten.
2.2.3 Triacylglycerol compositional analysis
The TAG compositions of samples were determined ac-
cording to the method described by Yanty et al.12） using
Waters Model 510 liquid chromatography equipped with a
differential refractometer Model 410 as the detector
（Waters Associates, Milford, MA）. The analysis of TAG was
performed on a Merck Lichrosphere RP-18 column（5 μm）
（12.5 cm×4 mm i.d.; Merck, Darmstadt, Germany）which
was maintained at 30°. The mobile phase was a mixture of
acetone:acetonitrile（63.5:36.5）and the flow rate was 1.5
mL/min. The injector volume was 10 μL of 5％（w/w）oil in
chloroform. Each sample was chromatographed three
times, and the data were reported as peak area percentag-
es. The identification of the TAG peak of the samples was
 Detection of lard stearin in canola oil

done in accordance with the TAG profiles of LD and LS re- Table 1　‌Comparative triacylglycerol (TAG) compo-
ported previously by Yanty et al.12）. sitions of lard stearin (LS) and lard (LD)1.
2.2.4 Thermal analysis
TAG LS LD
Thermal analysis was carried out using a Mettler Toledo
b
differential scanning calorimeter（ DSC 823 Model）, LLLn 0.20±0.00 0.61±0.01a
equipped with a thermal analysis data station （STARe soft- LLL 0.23±0.00 b
0.65±0.04a
ware, Version 9.0x, Schwerzenbach, Switzerland）. Before OLL 1.80±0.03b 5.85±0.02a
the analysis, calibration of the instrument was done using
PLL 2.59±0.01b 8.29±0.00a
Indium as the metallic standard, based on the onset tem-
b
perature of fusion and the heat of the indium’s fusion. OOL 2.44±0.01 8.32±0.01a
Nitrogen（99.99％ purity）was used as the purge gas at a LPO 7.60±0.05b 24.67±0.01a
rate of 20 mL/min. Approximately 4-8 mg of molten sample PPL 4.36±0.03a 2.90±0.01b
was placed in a standard DSC aluminum pan and then her-
OOO 2.32±0.03b 4.76±0.02a
metically sealed. An empty and hermetically sealed DSC b
aluminum pan was used as the control. The oil/fat samples OPO 8.28±0.05 21.06±0.13a
were subjected to the following temperature program: 70℃ PPO 26.93±0.16a 3.31±0.00b
isotherm for 1 min, cooled at 5℃/min to −70℃. The PPP 0.48±0.01a 0.51±0.01a
samples were then held at −70℃ isotherm for 1 min, and b
OOSt 1.10±0.00 3.87±0.08a
heated at 5℃/min to reach 70℃12）.
2.2.5 Statistical analysis StPO 35.29±0.21a 13.09±0.09b
Data were statistically analyzed using Minitab（version PPSt 1.67±0.01a 0.45±0.01b
14）statistical package. The t-test was applied to determine StStO 2.17±0.00a 0.87±0.00b
the significant differences between the means of individual StStSt 2.54±0.04 a
0.79±0.00b
TAG molecular species of LD and LS at a level of p＜0.05.
UUU 6.99 20.19
The relationships between LS content（％）of the samples
and each of the individual DSC parameter of adulteration UUS 19.57 57.89
peak were determined using Pearson’ s correlation analysis. USS 68.75 20.17
In the stepwise procedure, the DSC parameter that was in- SSS 4.69 1.75
cluded in the model was the one which had the highest 1
Each value in the table represents the mean±standard
Pearson’ s correlation with the response variable［LS（％）］.
deviation of three replicates. Means within each row
The significance level of an independent variable for entry
having different superscripts are significantly (p<0.05)
and stay in the calibration model was set to 0.15 during ex-
different. Abbreviations: O, oleic; P, palmitic; L, linoleic;
ecution of the stepwise variable selection procedure in the
Ln, linolenic; S, saturated; St, stearic; U, unsaturated.
Minitab.

tion of low-melting TAG group in it was significantly lower

（Table 1）. This would have led to the situation that LS had
3 Results and Discussion
3.1 Comparative thermal profiles of LD and LS
The DSC cooling and heating curves of LD and LS are
compared as shown in Fig. 1. The profiles of the thermal
curves of LS were distinctly different from those of the
native LD sample with regard to a number of features.
During both cooling and heating processes, LS was found
to display single thermal transitions in the high-tempera-
ture region above 18℃ while LD showed two well-separat-
ed thermal transitions in both low- and high- melting
regions of the curves. For native LD, the transition in the
region below 10℃ was due to low-melting TAG group,
while the transition in region above 10℃ was due to high-
melting TAG group12）. LS not displaying any significant
thermal transition in the temperature region below 18℃
can be assumed to be due to the fact that it had a greater
proportion of high-melting TAG group, while the contribu-
J. Oleo Sci. 63, (9) 867-873 (2014)
（39.3℃）than LD
a higher Tendset （34.0℃）. Tendset is a DSC pa-
rameter which is usually designated to denote the comple-
tion of the melting of all TAG molecular components
present in a fat sample.
3.2 Monitoring of CaO adulterated with LS by cooling
curves
The DSC cooling curves of canola oil and that adulterat-
ed with LS are compared as shown in Fig. 2. The curve of
CaO without adulteration was found to display a major
exothermic peak at −56.3℃ and a minor exothermic peak
at −19.5℃. Further, the onset of crystallization of pure
sample of CaO was found to be at −16.35℃ and there was
no significant thermal transition above −10℃. These ob-
servations were generally attributed to the presence of
predominantly low-melting triunsaturated TAG groups in
CaO and were in accordance with the findings reported
869
 J. M. N. Marikkar and S. Rana

Fig 2　DSC cooling curves of (A) canola oil (CaO) and (B) CaO adulterated with 5% lard stearin (LS), (C) 6% LS, (D)
7% LS, (E) 8% LS, (F) 9% LS, (G) 10% LS, (H) 11% LS, (I) 12% LS, (J) 13% LS, (K) 14% LS, and (L) 15%
LS.

previously9, 13）. The adulterant LS, on the other hand, was
found to possess greater proportions of high-melting TAG
groups when compared to either native LD or CaO 9）. As
shown in Table 1, LS was found to possess excessive
amount of di-saturated TAG molecules （68.75％） . Owing to
this fact, LS was found to display the most dominant
thermal transitions in the higher temperature regions（＞
20℃）of the thermal curves when compared to either LD
（Fig. 1）or CaO9）. As a consequence, a series of changes oc-
curred in the thermal profile of the cooling curve of CaO
once it was adulterated with LS, mainly due to dissimilari-
ties between CaO and LS with respect to their fatty acid
and TAG compositions. For instance, the major exothermic
peak of CaO at −53.6℃ was tended to become larger in
size while its peak position and onset temperature were
shifting to a higher temperature region as shown in the Fig.
2. Meanwhile, the minor exothermic transition appearing
at −19.5℃ was also found to increase in size with the in-
creasing level of adulteration. It was broadened gradually
870
J. Oleo Sci. 63, (9) 867-873 (2014)
while its peak position shifted toward the higher tempera-
ture regions. This trend was usually attributed to binary
mixture formation in CaO after adulteration with LS, which
contained a significant proportion of higher melting TAG
molecular species. According to statistical analysis data
presented in Table 2, changes in the DSC parameters of
this minor peak namely peak temperature（T）, peak area
（A） , and peak onset temperature（ON） showed good corre-
lations（r＝＋0.999; p＜0.0001）with the increasing level of
adulteration by LS. Although dramatic variations in the
DSC parameters of this endothermic peak could be expect-
ed at higher levels of adulteration of CaO with LS, the
minimum level of detection is a matter of importance from
an analytical point of view. As this exothermic peak started
to undergo changes only when the adulteration level has
reached 5％, it would be considered as the detection limit
of this approach.
 Detection of lard stearin in canola oil

Table 2 Pearson correlation coefficient between lard stearin content (%) and each of the
individual DSC parameter of the adulteration peak of the cooling and heating
curves1.
Curve type DSC parameter DSC value range Correlation coefficient
A  22.55 to 77.45 ＋0.994 (p < 0.0001)
Cooling T − 4.45 to 2.60 ＋0.998 (p < 0.0001)
ON − 2.92 to 4.92 ＋0.999 (p < 0.0001)
A   9.99 to 46.66 ＋0.976 (p < 0.0001)
Heating T  14.3 to 25.1 ＋0.955 (p < 0.0001)
ON  10.9 to 19.3 ＋0.971 (p < 0.0001)
1
Abbreviations: LS, lard stearin; A, peak area; T, peak temperature; ON, peak onset.

Fig 3　‌D
 SC heating curves of (A) canola oil (CaO) and (B) CaO adulterated with 5% lard stearin (LS), (C) 6% LS, (D)
7% LS, (E) 8% LS, (F) 9% LS, (G) 10% LS, (H) 11% LS, (I) 12% LS, (J) 13% LS, (K) 14% LS, and (L) 15%
LS.

3.3 Monitoring of CaO adulterated with LS by Heating two overlapping endotherms: a large - higher temperature
Curves transition at −17.86℃ and a small -lower temperature
The DSC heating curves of CaO and that adulterated transition at −28.50℃. The endset of melting of CaO was
with LS are compared as shown in Fig. 3. The curve of CaO found at −6.79℃ since there was hardly any significant
without adulteration is characterized by the occurrence of thermal transition above this point. As pointed out earlier
871
J. Oleo Sci. 63, (9) 867-873 (2014)
 J. M. N. Marikkar and S. Rana

Table 3　‌Summary of the stepwise regression analysis carried for predic-

tive model of LS content using the minor transition of the cooling
curve1.
Step Regression equation R2 SE
1 LS (%) = 11.1 + 1.43 T 0.996 (p < 0.0001) 0.22
2 LS (%) = 1.58 + 0.18 A 0.989 (p < 0.0001) 0.36
3 LS (%) = 8.47 + 1.27 ON 0.997 (p < 0.0001) 0.18
4 LS (%) = 7.97 + 0.958 T + 0.0596 A 0.998 (p < 0.0001) 0.16
5 LS (%) = 9.23 + 0.406 T + 0.907 ON 0.997 (p < 0.0001) 0.19
6 LS (%) = 6.18 + 0.0594 A + 0.854 ON 1.00 (p < 0.0001) 0.05
1
Abbreviations: LS, lard stearin; A, peak area; T, peak temperature; ON, peak onset;
R2, coefficient of determination; SE, standard error.

for cooling curve, the heating curve of adulterant LS was
also found to be completely different from the heating
profile of CaO. Once CaO was adulterated with LS, there
were many changes in the heating curve of CaO due to the
TAG compositional changes. According to a number of pre-
vious reports, TAG compositional changes caused by adul-
terations were influential on the thermal profiles of oils and
fats8−10）. After the addition of LS, the major thermal transi-
tion of CaO occurring at −17.86℃ was started to show
changes as shown in Fig. 3. Clearly, the overlap between
the major（−17.86℃）and minor（at −28.50℃）transitions
of CaO was increased with the increment of level of LS（％）
in CaO admixtures（Fig. 3）. However, this change alone
could not be suffcient to make a confirmation of the detec-
tion of LS adulteration in CaO. It was mainly because the
changes at lower levels of adulterations would go unnoticed
or become difficult to detect with certainty. Alternatively,
an emerging thermal transition in the higher temperature
region between 0.0 to 70.0℃ would be more appropriate as
it would show a better reflection of the adulteration with
LS. The particular endothermic transition could be taken
as a reference peak for LS since it was also a reference
peak for detection of LD adulteration in CaO 9） and sun-
flower oil（SFO） 11）
as reported previously. Although LD was
detected in CaO and SFO using this peak only at 8％ level
of adulteration, the results of the current investigation
showed that this peak could be used to detect LS in CaO as
low as 5％. As the past and the present investigations
showed that the adulteration peak for both LD and LS have
been detected in the same temperature region of the
heating curve, the particular endothermic peak could be
probably attributed to the higher melting TAG components
common to both LS and LD. According to the data pre-
sented in Table 1, both LS and LD were found to have a
relatively higher proportions of SPO while other higher
melting TAG species such as PPSt, StStO, and StStSt were
found to be present roughly in similar amounts.
3.4 Predictive models for quantification of LS in CaO
For establishing predictive models for quantification of
LS in CaO, DSC parameters showing significant variations
at different levels of adulteration are required to be select-
ed. As mentioned earlier, DSC parameters of the adultera-
tion peak appearing in between -4.45 to 2.60 of the cooling
curve and DSC parameters of the adulteration peak ap-
pearing in between 14.3 to 25.1 of the heating curves were
found to display good correlations with the increasing
levels of LS in CaO（Table 2） . When the DSC parameters of
the minor thermal transition of the cooling curve were
used as response variables for the execution of the step-
wise procedure, predictive models for the determination of
LS content（％）of the samples were obtained as shown in
Table 3. While the first three models were based on indi-
vidual DSC parameter, the remaining three were obtained
using parameters （peak temperature/ peak area/ peak onset
temperature）in combination of two. According to Table 3,
all models showed a higher coefficient of determination
with good confidence limits. However, the model obtained
in the sixth step would be more accurate for the prediction
of LS content of CaO based on the highest coefficient of
determination（ R2 value）and smallest standard error of
prediction. Likewise, the outcome of the execution of step-
wise procedure for the DSC parameters of the adulteration
peak of the heating curve is shown in Table 4. In this case
also, the first three models were based on individual DSC
parameter, while the remaining three were obtained using
two parameters in combination. Although all models
showed a higher coefficient of determination with good
confidence limits, models obtained with singular DSC vari-
able displayed relatively lower coefficient of determination
（R2 value）with higher standard error of prediction. Out of
the three models obtained using parameters in combina-
tion of two, the model obtained in the sixth step would be
more accurate for the prediction of LS content of CaO
based on the highest coefficient of determination （R2 value）
and smallest standard error of prediction.

872
J. Oleo Sci. 63, (9) 867-873 (2014)
 Detection of lard stearin in canola oil

Table 4　Summary of the stepwise regression analysis carried for predic-

tive model of LS content using the minor transition of the heating
curve1.
Step Regression equation R2 SE
1 LS (%) = -5.56 + 0.754 T 0.912 (p < 0.0001) 1.04
2 LS (%) = 1.32 + 0.286 A 0.953 (p < 0.0001) 0.76
3 LS (%) = -4.78 + 0.974 ON 0.943 (p < 0.0001) 0.83
4 LS (%) = -1.41 + 0.269 T + 0.193 A 0.968 (p < 0.0001) 0.66
5 LS (%) = -5.73 + 0.321 T + 0.6 ON 0.969 (p < 0.0001) 0.65
6 LS (%) = -1.79 + 0.16 A + 0.456 ON 0.976 (p < 0.0001) 0.55
1
Abbreviations: LS, lard stearin; A, peak area; T, peak temperature; ON, peak onset;
R2, coefficient of determination; SE, standard error.

Conclusion 4）Li-Chan, E. Developments in the detection of adultera-

This study demonstrated the use of DSC cooling and
heating curves for detection of CaO adulterated with LS.
According to the results, DSC cooling curve of CaO was
found to have a minor exothermic transition at −19.5℃,
which was greatly influenced by LS and showed significant
variations when the adulteration levels reached 5％ level.
The DSC thermodynamic parameters of this peak namely
peak area, peak onset and peak temperature were useful as
variables to obtain a prediction model to estimate content
of LS in CaO. Similarly, the DSC heating curve of CaO was
found to have emerging endothermic peak in between 0.0
to 70.0℃, which was greatly influenced by LS and showed
significant variations when the adulteration levels reached
5％ level. The DSC thermodynamic parameters of this peak
were also found to be useful as variables to obtain a predic-
tion model to estimate content of LS in CaO.
Acknowledgement
Authors gratefully acknowledge financial assistance re-
ceived under Putra Grants Scheme（GP-IBT/2013/9409200）
of Universiti Putra Malaysia.
Reference
1）Moore, J. C.; Spink, J.; Lipp, M. Development and ap-
plication of a database of food ingredient fraud and
economically motivated adulteration from 1980 to
2010. J. Food Sci. 77, 118-126（2012） .
2）Aida, A. A.; Che Man, Y. B.; Wong, C. M. V. L.; Raha, A.
R.; Son, R. Analysis of raw meats and fats of pig using
polymerase chain reaction for halal authentication.
Meat Science 69, 47-52 （2005） .
3）Dennis, M. J. Recent developments in food authentica-
tion. Analyst 123, 151R-156R （1998） .
tion of olive oil. Trends Food Sci. Technol. 5, 3-11
（1994）.
5）Lipp, M.; Anklam, E. Review of cocoa butter and alter-
native fats for use in chocolate-Part B. analytical ap-
proaches for identification and determination. Food
Chem. 62, 99-108（1998）.
6）Regenstein J. M.; Chaudry, M. M.; Regenstein, C. E.
The kosher and halal food laws. Compr. Rev Food Sci
Food Safety 2, 111-127（2003）.
7）Nina Naqiyah, A. N.; Marikkar, J. M. N.; Dzulkifly, M. H.
Differentiation of lard, chicken fat, beef fat, and mut-
ton fat by GCMS and EA-IRMS techniques. J. Oleo
Sci. 63, 459-464（2013）.
8）Marikkar J. M. N.; Lai O. M.; Ghazali H. M.; Che Man Y.
B. Detection of lard and randomized lard as adulter-
ants in RBD palm oil by differential scanning calorime-
try. J. Am. Oil Chem. Soc. 78, 1113-1119 （2001）.
9）Marikkar, J. M. N., Ghazali, H. M.; Che Man, Y. B.; Lai, O.
M. The use of cooling and heating thermograms for
monitoring of tallow, lard and chicken fat adulterations
in canola oil. Food Res. Int. 35, 1007-1014（2002）.
10）Marikkar, J. M. N., Ghazali, H. M.; Long, K.; Lai, O. M.
Lard uptake and its detection in selected food prod-
ucts deep-fried in lard. Food Res. Int. 36, 1047-1060
（2003）.
11）Marikkar, J. M. N., Dzulkifly, M. H.; Nor Nadiha, M. Z.;
Che Man, Y. B. Detection of animal fat contaminations
in sunflower oil by differential scanning calorimetry.
Int. J. Food Prop. 15, 683-690 （2012）.
12）Yanty, N. A. M.; Marikkar, J. M. N.; Che Man, Y. B.;
Long, K. Composition and thermal analysis of lard
stearin and lard olein. J. Oleo Sci. 60, 333-338（2011）.
13）Przybylski, R.; Mag, T.; Eskin, N. A. M.; McDonald, B.
E.（2005） （Ed.）
Canola Oil. In F. Shahidi , Bailey’s indus-
trial oil and fat products, 6th edn., New Jersey: John
Wiley & Sons, Inc., vol 2, pp. 61-120.

873
J. Oleo Sci. 63, (9) 867-873 (2014)