Table II.

Gas Chromatographic Retention Times of Alcohol-3,5-Dinitrobenzoates and Carbonyl-2,4-Dinitrophenylhydrazone from Hydrolyzed

Acetal

Compound
3,5-DNB from acetal
(Peak 1)
3,5-DNB from acetal
(Peak 2)
2,PDNH from acetal
Ethyl-3,5-DNB
Isoamyl-3,5-DNB
Acetaldehyde-2,4DNH

Retention
time,
min.
3.6
6.1
4.1
3.6
6.1
4.1

Table 111. Gas Chromatographic Retention Times of Certain Carbonyl-2,4Dinitrophenylhydrazones

Compound
Acetaldehyde
Propionaldehyde
Acetone
Isovaleraldehyde

Retention
time,
min.
4.1

4.9

5.1

6.6

the 3,5-dinitrobeneoyl chloride and the

alcohols under the conditions used, it
was observed that the conversion was
complete enough to give two large
peaks, in addition to the benzene peak,
on gas-liquid chromatography of a
portion of the reaction products from
1 111. of acetal. These peaks had
retention times corresponding to

those for ethyl-3,5-dinitrobenzoate and

isoamyl-3,5-dinitrobenzoate (Table 11).
Injection of a benzene solution of 3,5dinitrobenzoyl chloride or 3,5-dinitrobenzoic acid gave no discernible peaks
other than that of the benzene solvent.
The retention time of the 2,4-dinitrophenylhydrazone derivative of the carbonyl portion of the hydrolyzed acetal
is also given in Table 11. In Table I11
are listed, for comparison purposes, the
retention times of four known carbonyl2,4 - dinitrophenylhydrazones which
were run gas chromatographically under
conditions identical to those used for
the hydrolysis product. It is evident
that the hydrolysis products from an
acetal can be determined successfully in
the form of the derivatives by means of
gas chromatography.
In order to confirm that 2,4-dinitrophenylhydrazone derivatives do pass
through the gas chromatograph without
decomposition, samples of known isovaleraldehyde - 2,4 - dinitrophenylhydrazone were injected into the Loenco
instrument and collected after passage
through the gas chromatograph. The
derivative before and after gas chromatography showed identical behavior
on a thin layer chromatogram.
The fact that ~arbonyl-2~4-dinitrophenylhydrazones and alcohol-3,5-dinitrobenzoates can be separated easily
by gas chromatography without decomposition suggests that a number of
other derivative compounds also might
be separated and identified in this
manner. Perhaps of more significance,

however, is the fact that gas chromatography can be used as a method to

obtain the derivative in a pure state
when the quantity is too small to permit
purification by recrystallization techniques. A small sample of a mixture of
derivatives thus could be resolved into
its components and the components
tentatively identified through retention
times. After collection of the separated
components melting points can be
determined using a micro hot stage and
the identifications thus be further
confirmed. Although this report is
limited to a discussion of the successful
application of these techniques to the
analysis of trace quantities of complex
acetals, it appears that functional group
separations and analyses of very small
samples of complex aroma or flavor
extracts should be facilitated.
LITERATURE CITED

(1) Dhont, J. H., de Rooy, C., Analyst

86, 74 (1961).
(2) Lipscomb, W. N., Baker, R. H., J.
Am. Chem. SOC.64, 179 (1942).
(3) Sadtler, S. P., Catalog of Infrared

Spectrograms, Philadelphia, Pa.

(4) Shriner, R. L., Fuson, R. C., Curtin,
D. Y., The Systematic Identification
of Organic Compounds 4th ed., p.
212, Wiley, Kew York, G..Y., 1956.
( 5 ) Soukup, R. J., Scarpellino, R. J.,
Danielczik., E.., ANAL.CHEM.36. 2255
(1964).
(6) Webb, -4.D., Kepner, R. E., Food
Res. 22, 384 (1957).
RECEIVED
for review September 7, 1965.
Accepted November 15, 1965.

Gas Chromatographic Determination of Campesterol,

P-Sitosterol and Stigmasterol
ANDREW ROZANSKI
Research Laboratories, The Upjohn Co., Kalamazoo, Mich.
An accurate gas chromatographic
method for the determination of campesterol, p-sitosterol, and stigmasterol is
described. The sterols are separated
as trimethylsilyl ethers on a stainless
steel column, packed with 0.75-1 .O%
silicone gum rubber on an acid-washed,
silanized support, having about 4 0 0 0
theoretical plates. The quantification
is carried out by measuring ratios of
peak areas before and after addition
of a standard amount of one of the
sterols to the sample, or by using
cholesterol as an internal standard.
For the first time, the relative responses
in the flame ionization detector of the
trimethylsilyl ethers of these three
sterols and cholesterol have been
accurately determined and found to be
significantly different. The method is

36

ANALYTICAL CHEMISTRY

easily used on a routine basis with

precision and accuracy hitherto unattainable, and has been applied to
the analysis of soybean sterols.

of the commonly occurring phytosterolscampesterol (24a -methyl - cholest - 5 en-3B-o1),

0-sitosterol
(240-ethylcholest-5en-3P-o1), and stigmasterol
(24p-ethyl-cholest-5,22-dien-3~-ol) - is
important in a number of studies and
Complete
applications (7, 8, 19).
analysis of a mixture of such closely related sterols had been very difficult
before the successful application (25) of
gas-liquid chromatography (GLC) to
steroids. Their separation by paper
(18) or thin-layer (4) chromatography
XACT DETERMINATION

is difficult and a t best semiquantitative

results might be expected. It is possible to determine stigmasterol in soybean sterols by two specific methods
based on its higher degree of unsaturation-e.g., by an infrared (10) and by a
radioactive isotope dilution technique
(6).
GLC is capable of separating campesterol, @-sitosterol, and stigmasterol and
has been frequently reported (1, 2 , 7,
8, 11, IS, 15, 20, 21, 23). Derivatives such as esters (12-14, 22) and
ethers (S), including trimethylsilyl
(TMS) ethers (12, 24) have also been
separated by GLC. Although this
informative and inherently precise
method has been often used for the
separation and identification, attempts
a t quantitative analysis (1, 7 , 8, 11, 14,

20, 21), as far as can be surmised from

the often scanty information published,
have been based on the assumption of
equal specific detector responses and
usually carried out by internal normalization of peak areas. In exact analysis,
the relative responses of all the components should be established and,
unless additional information is available, the peaks under consideration
should not be assumed to constitute the
whole sample. Furthermore, the ratio
of the partition coefficients for these
sterols, especially for campesterol and
stigmasterol, is close to unity and unless
high efficiency columns are used and
the frequently occurring peak tailing is
eliminated, the overlap of the peaks is
considerable and the accuracy low,
The purpose of this paper is to describe a fully developed, easily reproducible GLC method for the determination of campesterol, 8-sitosterol, and
stigmasterol with precision and accuracy
which have not been possible so far.

EXPERIMENTAL

Apparatus. The gas chromatograph

was an F & 11 Model 810 equipped
with a hydrogen flame ionization detector. -\ Disc Integrator was used
to measure peak areas. Tlvo types of
columns were used for quantitative
analysis. Column A consisted of a 16foot X l/c-inch 0.d. stainless steel
(Type 304) coil containing 0.75y0 G E
silicone gum rubber SE-52 on GasChrom Z, 45- to 50-mesh (Applied Science
Laboratories). The column was washed
and packed in the conventional manner
(about 21 grams of the packing required), then coiled and conditioned by
heating for
hour a t 300' C. without
carrier gas flow and then a t least 24
hours a t 250' C. with carrier gas flowing.
It was finally deactivated by injecting
several 2 0 4 portions of hexamethyldisilazane while maintaining column
oven and injection port temperatures a t
120' C. and carrier gas flow rate a t
about 5 ml./minute. Column B was
purchased from F & RI Scientific Corp.
(['high efficiency column"). It consisted of an 8-foot X 1/8-inch0.d. stainless
steel coil packed with 1% silicone gum
rubber SE-30 on Diatoport S,80- to 100mesh. It was necessary to deactivate
this column with repeated injections of
microliter quantities of hexamethyldisilazane to eliminate slight degradation
of sterol TMS ethers.
Columns used during development of
this method were 6- or 8-foot X l/c-inch
0.d. glass coils packed with Gas-Chrom
Z or Anakrom ABS coated with commercially available liquid phases (Applied Science Laboratories and Analabs,
Inc., respectively).
Samples were injected with a Hamilton 701-NCH microsyringe. To deliver the desired volume, the syringe
needle was filled with a solvent before
drawing in the sample. Although, with
the quantitative technique described
later, knowledge of the sample volume

is not required, more reproducible injections were obtained in this way,

facilitating peak area measurements.
Reagents. All melting points were
corrected. Specific rotations were determined in chloroform a t c = 0.01
gram/ml. All reference compounds
were dried by heating under vacuum
and showed the absence of solvents
by an infrared melt solvate analysis.
Stigmasterol (Upjohn), derived from
soybean sterols, was repeatedly recrystallized from a mixture of 1,2dichloroethane and n-he tane and gave
a single peak by G L 8 : m.p., 168170.5' C.;
-50.2' C.; literature
m.p. 170" C.; [ala3 -51" C. (chloroform). @-Sitosterol melted at 139140' C., literature: m.p. 140' C.
Campesterol was separated from a
sample of "sitosterols" (Upjohn), derived from soybean oil, by 16 fractional
recrystallizations from acetone. It
showed 5.0% @-sitosterol as the only
impurity determined by GLC; m.p.
160.5-161.5' C.; [a12 -32.4'
C.;
literature m.p. 158' C.; [a]%*-33'
(chloroform).
Its mass spectrum
showed the parent ion a t m/e = 400
(C2sH4s0),and a small amount of an
impurity a t m / e = 414 (CnsHsqO).
The mass spectrometric characterization of this soy sterol as campesterol
(and not "7-sitosterol") is in full agreement with the findings of Thompson
et al. (21). Cholesterol (Applied Science
Laboratories) showed m.p. 146-149'
C.; [a12 -38.5' C.; literature m.p.
149' C.; [a]? -39' C. (chloroform).
Other reference compounds used in the
development of this method were:
cholestane (puriss., Aldrich) , n-octacosane and n-dotriacontane (K & K
Laboratories).
Hexamethyldisilazane
and trimethylchlorosilane were obtained
from dpplied Science Laboratories.
Tetrahydrofuran was dried with Linde
Molecular Sieves 4-i and stored over
them for short periods of time only.
Procedure.
GLC CONDITIONS.
Operating conditions for column A
were as follows: temperatures of the
injection port, detector block, and
column oven were 320', 260', and
230' C., respectively. Helium flow
rate was 85 ml./minute (inlet pressure
30 p.s.i.), hydrogen 57 ml./minute,
air 450 ml./minute, the latter two
optimized for maximum response. Size
of the injected sample was 0.5 pl.,
containing not more than about 6 pg.
of any single component. When column
B mas used, column temperature was
changed to 250' C., and helium flow
rate to 35 ml./minute.
PREPARATIOK
O F S-4MPLES. TMS
ethers were prepared by a simplified
method of Luukkainen et al. (16).
Twenty-five to 30 mg. of total sterols in
a 2-ml. glass-stoppered test tube were
dissolved in 1 ml. of anhydrous tetrahydrofuran; 0.5 ml. of hexamethyldisilazane and 10 pl. of trimethylchlorosilane were added, the tube was stoppered tightly and heated for 40 minutes
a t 55' C. The cooled mixture was then
injected directly into the gas chromatograph.
The trifluoroacetate esters used in
comparative studies were prepared by

the method of Vanden Heuvel et al.

(26)*

METHOD
OF ANALYSIS. A portion of
the sample, whose weight need not be
known accurately, was converted to the
TRIS ethers and duplicate chromatograms obtained. The actual amount
injected was immaterial, provided it was
within the linear range of the detector.
Next, to an accurately weighed portion
of the sample a known weight (20-50%
of the sample weight) of pure stigmasterol was added, the mixture converted
to TMS ethers, and again chromatographed twice. From the values of
the mean peak area ratios (stigmasterol/
campesterol) before and after addition
of stigmasterol, the concentration of
stigmasterol was calculated (see below),
without the need for specific or relative
detector response. The concentration
of campesterol and p-sitosterol was calculated using appropriate peak area
ratios and independently determined
relative responses. If the last two
sterols could be obtained in sufficient
amounts, they could also be quantified
by the standard addition procedure
without the need for calibration.
When the amount of stigmasterol in
the sample exceeded about 75% it was
more precise (see below) to use cholesterol as an internal standard in the conventional manner.
RESULTS AND DISCUSSION

Development of GLC Procedure.

The author confirmed and extended
previous workers' results (12, 61) that
the use of different types of liquid
phases or conversion to common derivatives did not change the separation of
the sterols under study (Table I).
Silicone gum rubber was subsequently
used as the liquid phase in quantitative
determinations.
Even on acid-washed, silanized supports and in glass columns, tailing of the
sterol peaks could not be completely
eliminated and therefore the chromatography of free campesterol, P-sitosterol, and stigmasterol was not considered suitable for quantitative measurements. The trifluoroacetates decomposed more easily (the column
packing was well removed from the hot
injection port) than the already sensitive sterols which require careful control
of materials and conditions. Consequently, TMS ethers were used in
quantitative work. Although their retention times were somewhat longer
(Table I), extremely symmetrical peaks
were obtained and the compounds were
stable in stainless steel columns. (Free
sterols decomposed appreciably on metal
columns.)
The optimum conditions for complete
silylation of the sterols were found by
following the rate of ether formation by
GLC; unreacted sterols are separated
from the ethers. The use of elevated
temperature for ether preparation, instead of the customary overnight
standing, resulted in no undesirable
VOL. 38, NO. 1, JANUARY 1 9 6 6

37

side effects.
Tetrahydrofuran was
chosen after examining several solvents
because it showed minimum tailing
of the solvent peak and minimum
amount of the precipitate of the byproduct ammonium chloride (the slight
haze which usually formed had no
effect on the relative peak areas).
For essentially complete separation
(lye fractional impurity per peak) of
equal size peaks of the closest two
components, campesterol and stigmasterol, about 3000 theoretical plates
( N ) are required [Glueckauf's plot (9)]
using SE-52. The commercial coated
supports used in this work did not produce N higher than 200-300 per foot
of either glass or metal column IN
calculated as 16

t)
2,

where

1: =

un-

corrected retention time, y = peak

width, using @-sitosterol as reference].
Columns with the required efficiency
and giving reasonably short retention
times (<1 hour) were obtained by
using minimum liquid phase loading,
temperature, and sample size (maximum
6 pg. of one sterol; a decrease to 0.2 pg.
did not improve column efficiency),
and maximum carrier gas flow rate.
Under these conditions, column A had
3700 theoretical plates and gave the
separation illustrated in Figure 1,
with a total time per chromatogram of
about 45 minutes. Column R showed
5600 plates, even somewhat better

Table 1.

Effect

16
(3)

II;

SENSITIVITY 4x

TIME (mi.)

Figure 1.

Typical chromatogram of soy sterol TMS ethers

Column A: (1) Campesterol, (2) Stigmasterol, (3) p-Sitosterol TMS ethers, approximately 2.5 pg,, 2.5
pg., 5.0 pg., respectively

separation, and total time per chromatogram of only 22 minutes.

Column life-time is not known, but
no deterioration of column A was
observed after 5 months of almost continuous daily use. Column B, introduced near the end of this work, was
used for only a relatively short period
of time.
Relative Response and Linearity.
Since only a limited supply of campesterol and p-sitosterol was available,
the determination of relative detector
responses of the sterol TMS ethers was
carried out as follows. The response

of Liquid Phase and Derivative Formation on Sterol Separation

Liquid phase and temp., ' C.

Relative retention time (RRT)

Trifluoro- TrimethylAlcohol
acetate
silyl Ether
RRT" RRTb RRT" RRTc RRTa RRTd
2.48 0.78
2.60 1',00 2.05 1'.00 3.19 1.00
2.87 1.10 2.26 1.10 3.51 1.10
3.23 1.24 2.59 1.25 3.95 1.25
1.08
1.25
1.05
1.20
1.06
1.26

Compound
Cholesterol
cone) 230'
Campesterol
Stigmasterol
p-Sitosterol
1% QF-1 ( fluoroalkylsilicone)
Stigmast'erol
p-Sitosterol
205'
2% XE-60 (cyanoalkylsilicone) Stigmasterol
p-Sitosterol
220
1%HI-EFF-8B [poly(cycloStigmasterol
hexanedimethanol succinate)] p-Sitosterol
220'
Columns: 6- or 8-foot 1/4-inch glass coils.
Relative to cholestane ( = 1.00).
b Relative to campesterol ( = 1.00).
Relative to campesteryl trifluoroacetate ( = 1.00).
Relative to campesteryl trimethylsilyl ether ( = 1.00).
2% SE-52 (methylphenylsili-

Table II.

Relative Response of Sterol TMS Ethers in the Flame Ionization Detector

Compound
Relative wt. response (RWR)"
Campesterol TMS ether
0.94, 0.89
Stigmasterol TMS ether
0.92 , 0.91, 0.97, 0.98 , 0.97
0.76, 0.75, 0.79 , 0.77
p-Sitosterol TMS ether
Relative to cholesterol TMS ether ( = 1.00)

38

ANALYTICAL CHEMISTRY

Mean
RWR
0.91
0.95
0.77

Mean
relative
molar
response"
0.94
1.00
0.82

of the abundant stigmasterol TMS

ether relative to cholesterol TMS ether
was measured by injecting a series of
standard mixtures containing from 0.3
to 6 pg. of the former and an approximately constant amount of 4 pg. of the
latter, chromatographing each solution
four times, plotting the mean ratios of
peak areas us. ratios of weights (Figure
2), and calculating the equation of the
line by the method of least squares. The
relative weight response, given by the
slope, had a relative standard deviation
of l.5yoand was constant up to a load
of 6 pg. of stigmasterol, The limit of
detector linearity was not determined
because column overloading, evidenced
by nonsymmetrical peaks (and the
accompanying loss in column efficiency),
was apparent with larger amounts.
The plot passed exactly through the
origin (intercept of -0.003 with a
standard deviation of 0.014).
The responses of @-sitosterol and
campesterol TMS ethers were then
determined relative to cholesterol ThlS
ether by using only one to two standard
solutions and assuming that the calibration line passes through the origin.
Since campesterol contained some psitosterol, the latter impurity was determined and corrected for by using cholesterol as an internal standard and the
appropriate relative response. The relative responses of these three phytosterols were determined several times
in this manner over a period of three
months, using two F & M Model 810
gas chromatographs, two columns of
type A prepared with different batches
of packing, and column B. The results are shown in Table 11. The overall relative standard deviation for these
factors was 3.5%. It was concluded
that the relative detector responses of
these closely related compounds are
reasonably independent of operating
conditions and time, and need not be
redetermined frequently.
It is reasonable to assume that the
free sterols also show differences ic

relative responses in the flame ionization detector. Thus, if uncorrected

peak areas are used in quantitative
measurements, errors will arise, especially serious for p-sitosterol.
It is interesting that in the series
cholesterol, campesterol, and p-sitosterol TLIS ethers the relative response decreases with the carbon number or retention time and that the introduction of a double bond a t C-22
in @-sitosterol (to give stigmasterol)
increases the relative molar response as
much as the removal of the C-24-ethyl
group. Furthermore, these relative molar responses could not have been predicted from the flame ionization deMol. Wt.
tector C-factors
(No. c,atoms X 12 ( 5 ) .
The relative C-fa'ctors, using cholesterol
TILTS ether as 1.00, for the three phytosterols were in the range 0.98-1.00,
irrespective of whether the total number of carbons was used in the calculation, or if those bearing heteroatoms
were omitted.
There are good indications that the
relative responses shown in Table I1
represent actual flame ionization detector responses and not differences in
component losses during chromatography.
Thus in Figure 2, where
varying weights of stigmasterol TMS
ether are compared with approximately
constant weight of cholesterol T M S
ether, the good linearity and zero
intercept would not be expected if
component loss occurred, as usually

Table 111.

Precision of Peak Area Ratios

Compounds compared (i/j)

Relative
retention
time

ratio
(A,

(ti/tj)"

Stigmasterol TMS-cholestane

3.36
2.96
1.44
1.27c

n-Dotriacontane-cholestane

Stigmasterol TMS-cholesterol TMS

Stigmasterol TMS-@-sitosterolTMS

100
ub

Ai

Aj)

1.27
0.93
1.45

0.1-0.7
0.7-1.4

VAT

A,

0.120
0.061
0.043
0.010
0.016

9.4
6.6
3.0

3.8

1.7

Column A :
Retention time stiemasterol TRIS ether. 37 min.
A , = mean for theset when a range is shown in column 3.
c

tjlti.

per cent lost is inversely related to the

amount injected.
It should be noted that a preliminary
report by Miettinen et al. (17) indicates
that cholesterol, campesterol, and 8sitosterol TASS ethers have identical
responses (per unit mass of unsubstituted parent sterol) in the flame
ionization detector. It is difficult to
rationalize such conflicting results, a t
least until the data are published.
Standard Addition Method.
The
use of an internal standard, required
in exact quantitative GLC t o allow
for variations in the volume injected,
met with two difficulties. First, in

1.6

1.4

GLC of phytosterols, such as soybean

sterols, a variety of minor impurities
are found and it would be necessary to
correct for any overlap with the
internal standard by a second chromatogram without the standard. Second, it
was found that the standard deviation
of the peak area ratio greatly increased
with the separation of the peaks, considering approximately equal size peaks
with only negligible overlap. This is
illustrated in Table 111, column 5 , for
sterol TMS ethers and some hydrocarbons.
A modified standard addition procedure was therefore developed for
mixtures containing two or more closely
occurring but well resolved components.
I n general, peak area for the i component relative to a second component
j is independent of the amount injected
and the dilution of the sample, and is
given by :

1.2

2 1.0

Where K

VI

:0.8
U

LL

2K 0.6

1
0

Figure 2.
ether

specific weight response,

w/w per cent of given component

in the sample.
Let 2 be the weight of component
i added per unit weight of sample under
study (in the present work i = stigmasterol, j = campesterol) and the new
chromatogram give peak areas -4%'and
A>'. Then it can be shown that the new
area ratio A,' is again independent of
the amount injected and sample dilution
and is given by:

0.2
0.4

Hence,
0.2

0.4

0.6

0.8
1 .o
RATIO OF WEIGHTS

1.2

1.4

1.6

.a

Relative weight response of stigmasterol TMS ether-cholesterol TMS

Peak area ratios are means of four runs.

Approximately 4 pg. of cholesterol used every time

Equation 1 holds true if detector response is linear and if K i / K j (relative

VOL. 38,

NO. 1,

JANUARY 1966

39

Table IV.

Wt. stigmasterol
added per unit
wt. sample

Stigmasterol

40.2
40.0
41.6
41.2
41.6
41.2
40.6
Mean 40.9
Rel. std. dev. % 1 . 6

0,2831
0.1735
0.2101
0.1699
0.2849
0.2638
0.2766

Found, yo
Campesterol

@-Sitosterol

34.4
34.3
35.5
35.1
35.7
35.3
35.0
35.0
1.6

16.9
16.8
17.7
17.5
17.7
17.5
17.6
17.4
2.2

Each result based on two chromatograms.

2.882, Art

In order to test whether the very

slight peak overlap of campesterol and
stigmasterol TMS ethers was giving
rise to a systematic error in this method,
three mixtures of the reference compounds were made up and analyzed
by the standard addition of stigmasterol
procedure. As shown in Table V, the
mean error of these determinations is of
the same order of magnitude as the
precision.

Precision of the Sterol Analysis

(s)

(cs)(2)
= 1.194,

b,

= 1.903.

Column A
Table V.

Accuracy of the Sterol Analysis

ACKNOWLEDGMENT

The author thanks W. E. Robbins,

U. S. Department of Agriculture, Beltsville, Md., for a gift of @-sitosterol,R.
T. Rapala, Eli Lilly and Co., for a sample of campesterol, and M. F. Grostic,
The Upjohn Co., for the measurement
of the mass spectrum of campesterol.

Error (found-true)
True concn., %
Found, yo
Campesterol Stigmasterol Campesterol Stigmasterol Campesterol Stigmasterol
37.6
46.7
66.6

60.5
50.8
29.9

37.4
45.5
65.9

61.3
50.5
30.3

Mean error
Column A.
From the actual weights of reference compounds.

-0.2
-1.2
-0.7
-07

0.8
-0.3
0.4
0.3

detector response) is constant during

the measurements. The concentration
of any other component X, is given by:

This method takes advantage of the

high reproducibility of area ratios of the
closely occurring sterol peaks (for which
peak overlap has been reduced to an
insignificant amount), it does not require
strict control of volumes injected, and,
in the case of the component which is
added, it does not require the determination of its specific or relative
detector response. Its only drawback
is that two chromatograms are required,
but as stated before this is often also
necessary when an internal standard is
used with a mixture of natural origin.
The soundness of this method is
illustrated by the constant results obtained on a mixture of campesterol,
0-sitosterol, and stigmasterol, derived
from soybean oil, to which varying
amounts of stigmasterol were added
(Table IV).
Accuracy and Precision. The relative standard deviation of the phytosterol peak area ratio (A,) depended
on the magnitude of A , and the separation factor. This is illustrated in Table
111,line 4,5. Each u for stigmasterol0-sitosterol and stigmasterol-campesterol TMS ethers was based on duplicate results for a t least 15 samples derived from soybeans with A, in the range
shown in column 3, and was computed
by an analysis of variance. Highest
40

ANALYTICAL CHEMISTRY

precision was obtained when the peaks

were of about equal size. Estimates of
the relative standard deviations of X
(see Equations 1 and 2) for stigmasterol,
campesterol, and 0-sitosterol were calculated by considering the relationship
between X , A,, and A, (where the area
ratios are mean values, each based on
two chromatograms, see Experimental).
For peaks not differing in size by more
than about 40% these estimates were calculated to be 1.0%, 1.2%, and 1.6%,
respectively. The larger value for
0-sitosterol reflects the loss in precision
due to larger peak separation for the
stigmasterol-/%sitosterol pair than for
stigmasterol-campesterol (Table 111).
These values were confirmed by repeatedly analyzing a mixture of the
three sterols (derived from soybeans)
by the modified standard addition
method described. Reasonable agreement with the above prediction was
obtained (Table IV), if the magnitudes
of peak area ratios are taken into
account.
When the level of the compound used
for the standard addition exceeds about
75% of the sample, necessitating comparison of very dissimilar peak areas,
it is more accurate to use cholesterol as
an internal standard (cholesterol is
particularly suitable because of its
chemical similarity to the phytosterols
and small separation factor). An estimate of the precision of the procedure
was then obtained from Table 111,
giving relative standard deviation for
equal size peaks of about 3.0%.

LITERATURE CITED

(1) Bloomfield, D. K., J. Chromatog. 9,

411 (1962).
(2) Castle, M., Bloudin, G., Nes, W. R.,
J.Am. Chem. SOC.85,3306 (1963).
(3) Clayton, R. B., Biochem. 1, 357
(1982).
\ - - - -

(4) Copius-Peereboom, J. W., Beaks,

H. W., J. Chromatog. 17,99 (1965).
(5) Dal Nogare, S., Juvet, R. S., Jr.,

Gas-Liquid Chromatography, Interscience, New York, 1962.

(6) Donia, R. A., Ott, A. C., Drake, N.,
ANAL.CHEM.29,464 (1957).
(7) Eisner, J., Wong, N. P., Firestone, D.,
Bond, J., J. Assoc. Oflc. Agr. Chemists

45,337 (1962).
(8) Eneroth, P., Hellstrom, K., Ryhage,
R., J. Lipid Res. 5,245 (1964).
(9) Glueckauf, E., Trans. Faraday SOC.
51,34 (1955).
(10) Johnson, J. L., Grostic, M. F.,
Jensen, A. O., ANAL. CHEM.29, 468
(1957).
(11) Kallianos, A. G., Shelburne, F. A.,
Means, R. E., Stevens, R. K., Lax,
R. E.. Mold. J. D.. Biochem. J. 87. 596
(12)-Knights, B. A., J. Gas Chromatog. 2,
160 (1964).
(13) Ibid., p. 338.
(14) Kuksis, A., Can. J. Biochem. 42,
407 (1964).
(15) Kuksis, A., Huang, T. C., Can. J.
Biochem. Physiol. 40, 1493 (1962).
(16) Luukkainen, T., Vanden Heuvel, W.
J. A., Haahti, E. 0. A., Horning, E. C.,
Biochim. Biophys. Acta 52, 599 (1961).
(17) Miettinen, T. A., Ahrens, E. H.,
Grundy, S. M., J. Lipid Res. 6 , 411
(1965).
(18) N e k , R., Steroid Chromatography, Elsevier, New York, 1964.
(19) Poulos, A., Greiner, J. W., Fevig,
G. A., Ind. Eng. Chem. 53,949 (1961).
(20) Thompson, M. J., Louloudes, S. J.,
Robbins, W. E., Waters, J . A., Steele,
J. A., Mossettig, E., Biochem. Biophys.
Res. Commun. 9, 113 (1962).
(21) Thompson, M. J., Robbins, W. E.,
Baker, G. L., Steroids 2,505 (1962).
(22) Vanden Heuvel, W. J. A., Sjovall, J.,
Horning, E. C., Biochim. Biophys.
Acta 48, 596 (1961).
(23) Vanden Heuvel, W. J. A., Sweeley,
C. C., Horning, E. C., J. Am. Chem.
SOC.82,3481 (1960).
(24) Wells, W. W., Makita, M., Anal.
Biochem. 4,204 (1962).

RECEIVEDfor review July 16, 1965.

Accepted November 15, 1965.