A STUDY OF ANTICKIDANTS
WITH RESPECT TO VITAMIN A IN FISH OILS

By
Orson D* Bird

A THESIS
Submitted to the Graduate School of Michigan
State College of Agriculture and Applied
Science in partial fulfilment of the
requirements for the degree of

DOCTOR OF PHILOSOPHY

Department of Chemistry
1941

TABUS OF CONTENTS

Pag*

I*

Introduction.

1

XI.

Review of the Literature.

3

III.

Methods.

IV.

Experimental Results.
A.
B.
C.
D.
E.

38

Stability of Natural Vitamin
A Oils.

49

Conditions Affecting the Stability
of Vitamin A in Oils.

52

Use of Induction Test for
Evaluating Antioxidants.

69

Studies of the Nature of Natural
Antioxidants in Fish Oils.

74

Attempts at Isolation of
Antioxidant Fractions.

89

V.

Resume.

98

71.

Conclusions.

105

711.

Bibliography.

109

Till.

Acknowledgment s•

114

DC.

Biography .

115

LIST OF ILLUSTRATIONS

Page

Vitamin A Breakdown and Peroxide
Formation in Halibut Liver Oil During
the Induction Test.

44

Changes in the Ultra-violet Absorption
of Vitamin A in Halibut Liver Oil
During the Induction Test.

45

Effect of Volume of Oil Exposed
and Relative Protection from Air on
the Breakdown Rate of Vitamin A in
Halibut Liver Oil.

47

Effect of Dilution and the Presence of
Peroxides on the Breakdown Rate of
Vitamin A in Oil Solution.

55

Induction Curve for Fish Viscera Oil.
Effect of Removing and Restoring the
Phospholipid Fraction.

90

INTRODUCTION

Vitamin A is among the most labile of the
vitamins; It was this property which enabled McCollum
to differentiate it from vitamin D in cod liver oil*
This lability of vitamin A is intimately associated
with the instability of the oily vehicles, such as
cod, halibut, and tuna liver oils, in which it is
usually found.

Oils, in themselves, are notoriously

suseeptible to rancidity, and this is particularly
true of fish liver oils.
The production of oxidative rancidity in oils
is due to atmospheric oxidation, which is possible
because of the presence in the oils of unsaturated
fatty acids, such as oleic, linoleic, and linolenic,
with one, two, and three unsaturated linkages re­
spectively.

The relative rates of oxidation of these

three fatty acids have been found to be in the ratios
of 1, 12, and 100 (39).

This accounts to some extent

for the different rates of oxidation of various oils,
although other factors, such as the presence of
antioxldants, are of greater importance.
Under the influence of atmospheric oxygen
oils containing vitamin A undergo autoxidation.

This

catalytic oxidation of the oils usually does not
begin at once, even under severe conditions, but
is preceded by a period, of different lengths in
different oils, during vfcich practically no reaction
takes place.

Following this there is a period of

rapid oxidation, during which the rate of oxidation
gradually increases to a maximum, then falls off*
Plotting this reaction rate against time gives rise
to an S shaped curve, which is characteristic of
autoxidation reactions*
This period of little change at the beginning
of the autoxidation of oils is called the induction
period.

It is prolonged by the presence in the oils

of antioxidants, and shortened by oxidation catalysts.
The duration or length of the induction period of
an oil is a measure of its keeping quality; when the
induction period ends rancid odors and flavors
develop and labile constituents of the oil, such
as vitamin A, begin to break down*

It has been

assumed that the peroxides, which are formed as
primary oxidation products in oils, are the autocatalysts of this reaction (70) . Greenbank and

Holm, how ever, conclude that acids and other
products formed in the oxidation process are
responsible for the increase in rate of reaction
with time (29)•
In the present study a standardized method
has been developed for determining the induction
periods of oils containing vitamin A*

This method

has been employed in evaluating the stability of
various types of natural vitamin A oils; and also
in studying the conditions affecting the stability
of vitamin A in these oils, such as the additions
of various Inhibitors of oxidation, or antloxldantst
It has been used further to determine the nature
of natural antioxidants occurring in fish oils,
and as a guide in attempting the isolation of
these natural antioxidant s.
REVIEW OF THE LITERATURE
Mechanism of Autoxidation

Theories of Autoxidation,

There are two

theories explaining the induction period of oils
and fats*

The older explanation (19) is that some

product of the oxidation reaction, probably peroxide,

functions as a positive catalyst to promote the
reaction*

This substance is considered to

accumulate in the fat until a certain critical
concentration is reached, when a rapid oxidation
is brought about*

Early evidence for this ex­

planation was the fact, reported by Fridericia
(18), that adding rancid lard to fresh cod liver
oil soon destroyed the vitamin A in the latter*
He suggested that the effect might be due to
organic peroxides*

Rosenheim and Webster also

reported (65) that if an oil high in peroxides was
mixed with fresh cod liver oil, the growth promoting
effect on the fat as well as the chromogen responsible
for the antimony trichloride color test was destroyed*
Recently, however, it has been found that
the induction periods of oils and other substances
become progressively shorter with increased puri­
fication of the fat or oil.

Thus Mattill and

Crawford found that purifying corn oil Increased
its susceptibility to autoxidize (48).

Also the

induction period of glycerides resynthesized from
the distilled acids of a natural oil has been found
to be much less than that of the original oil (27)*
Hamilton and Olcott (24) found that purified methyl

5.

oleate ana oleic acid had practically no period of
induction, beyond that required for the diffusion
of oxygen.

This evidence supports the view then

that in natural oils the period of induction is due
to the presence of inhibitors, and not to the slow­
ness with which peroxides are Initially formed.
Products Formed During Autoxidation of Oils.
There has been much speculation concerning the
products formed during the autoxidation of the
unsaturated fatty acid components of fats and oils.
Staudinger (69) believes the steps in autoxidation
are as follows:
R.CH=»CH.R’
Autoxidlzable

R.CH--CH.R*
^

R.CjJH— CH.R’ -*• R^jH+CH.R
0

0

0

0

0
Substance

Moloxide

Aldehydes, etc.

Peroxide

The moloxides are believed to be extremely unstable
and highly reactive.

Powick (61) suggests that

peroxides oxidize the fatty acid chain with the re­
moval of two H atoms to form a new double bond,
thereby being themselves converted to oxides.

This

formation of peroxides, followed by splitting of the
new double bonds accounts for the series of saturated
aldehydes and acids found in oxidized fats.

Tschirch

(73) considers the peroxide formed to be decom­
posed by water to form an oxide, hydrogen peroxide
and ozone.

The ozone then reacts with other

unsaturated molecules to form ozonides.

These

ozonideB, in the presence of water, decompose into
aldehydes and acids.

It has been proposed by

Browne (5) that a molecule of oxygen reacts with
a double bond to produce a fatty oxide, gnfl a
liberation of an atom of active oxygen.

This

active oxygen attacks the glycerides in one of
several ways— the resulting oxide lsomerizes to a
ketone, which breaks down on hydrolysis into sub­
stances of lower molecular weight.

All these

theories agree in requiring the primary formation
of peroxides or oxides Which either break down directly
into aldehydes and acids of lower molecular weight,
or isomerize, or react with water to form hydroxy
and keto compounds.

It is known that the double bond

ruptures since azelaic and nonoic acids have been
isolated from rancid oils.
Some time after the above theories were
advanced concerning the products formed during
autoxidation of unsaturated fatty acids, Hamilton
and Olcott (21) published data which clarified
the reactions to a great extent.

They followed the

course of oxidation of higily purified oleic
acid, methyl oleate and oleyl alcohol with special
apparatus and methods which permitted the simul­
taneous measurement of the oxygen absorbed and
its distribution among the transitory and final
products of oxidation.

These included water,

carbon dioxide, and carboxyl, hydroxyl, peroxide
and aldehydic compounds.

Recently Deatherage

and Mattlll (10) after having refined Hamilton* s
and 01cott*s apparatus made a further study of
the oxidation products of several highly purified
unsaturated fatty acids and related compounds.
They found that oleic acid, oleyl aloohol, methyl
oleate, butyl oleate and cis-9-octadecene all
appear to be oxidized in a similar manner, yield­
ing the same types of products— among others,
peroxides, peracids, aldehydes, substituted
ethylene oxides, acids, alcohols, combinations of
these, and water.

They concluded that after the

addition of oxygen at the double bond to form
peroxides, these peroxides may cleave to give
aldehydes, they may either react with another
double bond to give two moles of ethylene oxide,
or they may aid in the further oxidation of the
carbon chain.

The aldehydes formed may also

autoxldize to give peracids and acids.

They

could not detect the presence of carbon monoxide
carbon dioxide.

This work therefore does

not support the theory of the mechanism of
autoxidation of fats proposed by Fowick (61),
which necessitated the production of these two
compounds during the process.

Factors Affecting Rancidity
Atmospheric Oxygen.

Oxygen, either

free or combined, is necessary for the autoxidation
reaction.

This is the basis for protecting fats

from becoming rancid by enclosing them in completely
filled containers.

However, the amount of oxygen

necessary to carry an oil to the end of its in­
duction period is rather small.

Therefore unless

the space in the container is practically entirely
filled by the oil, little advantage will result.
Peroxide Oxygen.

Oxygen in loosely

combined, or peroxide, form may play an important
part in the development of rancidity in oils.

As

noted above, Fridericia (18) found that admixture
with rancid lard slowly destroyed vitamin A in
cod liver oil and suggested thet the effect might
be due to organic peroxides in the lard.

Powick

(62) found that the inferiority of rancid lard as
a component of a ration for rats was not due to its
actual toxicity but to its ability to destroy the
vitamin A in the rations with which it was mixed.
This destruction appeared gradually over several
days and was presumably due to the oxidation of
vitamin A by the organic peroxides of the rancid
lard.

When the vitamin A was fed separately from

the lard, the rats receiving the rancid lard seemed
to require more vitamin A than did others receiving
a corresponding ration containing sweet lard.

The

latter observation agrees with that of Lease,
Steenbock, et al. (44), that rancid fats fed
coinddently or shortly after ingestion of vitamin
A reduce the storage of the latter in the liver.
The assumption is that the peroxides of the rancid
fat destroy the vitamin A after ingestion.
It has also been reported by Rosenheim and
Webster (65) that when an oil high in peroxides is
mixed with a fresh oil it gradually destroys the
growth promoting effect as well as the chromogen
responsible for the color test with antimony tri­
chloride.

Wokes and Wllllmot (80) studied the rate

of destruction of vitamin A in cod liver oil aerated
at various temperatures.

They suggested that the

destruction of vitamin A under these conditions
was due to the formation of volatile organic
peroxides.

Smith (68) found that vitamin A which

had been dissolved in oil containing a large
amount of peroxides was rapidly destroyed, even
though sealed in glass tubes under nitrogen.
Whipple (79) reported a drop in vitamin A potency
of cod liver oils as the peroxide value increased.
The same relationship was found by Lowen et al.
(46), with the vitamin A in salmon and halibut oils.
Temperature.

The rate of autoxidation of

fata and oils is influenced by temperature, as
are other chemical reactions.

In fact, most acceler<

ated tests for stability of fats are based on the
increased rate of oxidation which occurs with a
corresponding rise in temperature.

Freyer (17)

found that the rate of production of peroxides
during the aeration of oils between 100° and 115° 0
was approximately doubled for each rise in
temperature of 10° C.

Lea (40) found that cod

liver oil oxidizing at temperatures between -7
and 27° C* under the influence of weak
illumination, oxidized 1.8 times as fast
after a 10° increase in temperature.

This

accelerating effect of teuperature is lessened

in the presence of active metal catalysts (64) •
Light.

The effect of light on the rancidity

of fats and oils has been more or less confused with
temperature and oxygen effects, with which it is
often inter-related.

It was formerly claimed (76)

that light alone, in the absence of oxygen, was
capable of producing rancidity.

However, it has

since been shown that this apparent production of
rancidity by light alone was really due to traces of
peroxides in the supposedly fresh fat.

When great

care is taken that the fats are peroxide-free and
stored in nitrogen, they do not become rancid when
exposed to light (36).

A distinctive characteristic

of rancidity or oxidation initiated by exposure to
light is that once the process has begun it continues
even if the samples are transferred to darkness (40).
This is considered to be due to a stimulation of
peroxide formation by the light, which later con­
tinues to catalyze the reaction, even in the dark.
Most of the recent data indicate that light
simply acts as an accelerator of autoxidation of fats.
This is due both to the intensity of illumination
and to the wave length of the activating light.
Lea (40) found that direct sunli^xt had the greatest
effect, diffused daylight had a lesser effect, and

even a 100 watt lamp at a distance of 6 feet in
an otherwise dark room had a pronounced accelerat­
ing action.

Greeribank and Holm (21) studied the

relative oxidation-stimulating effects on cotton­
seed oil of light of different wave lengths. Wave
o
hands about 500 A wide were studied by means of
filter combinations, and the amount of energy
passed through each filter was controlled by a
thermocouple and galvanometer.

The degree of

rancidity was expressed as the amount of oxygen
absorbed per 100 ml. of oil.

The greatest effect

was in the orange yellow band, being greater than
in the orange red or the blue bands.

These results

are in opposition to those of Coe and LeClerc (8),
who found that the greatest protection against
rancidity oaused by light was afforded by a dark
green filter which transmitted only long wave­
lengths.

Davies (9) also found that light of

shorter wavelengths, especially in the region of
the ultra-violet, was most effective in promoting
rancidity.

However, the depth of color of the

protecting film was more important than its
actual tint.

Morgan (51) studied the protective

effect against rancidity of fats of various
cellophane films.

He found that the efficiency

of a particular film in protecting against
rancidity was in proportion to its absorption
of ultra-violet and blue rays, rather than its
color*

He concluded that the effect of light of

short wavelength increased rapidly with decreasing
wavelengths*
Metal Catalysts*

The role of metal

catalysts in hastening oxygenation of oils has
been known for a long time*

This is the basis

for the use of driers in paint*

These are largely

the soaps of cobalt, which have been found to be
the most active of the metallic soaps in stimulating
oxidation of oils (12).

Copper is one of the most

frequent offenders in causing rancidity in food
fats, especially butter (31).

A common method of

evaluating the effect of metals on the oxidation
of fats has been to inmerse strips of the metals
into samples of the fat and determine the rancidity
at intervals*

The results are compared with those

for fat stored under similar conditions without
metal*

Using this method of examination and

following the reaction by odor and the Kreis test,
Emery and Henley (13) found that copper was the
most active of the metals studied in promoting

rancidity in corn and cottonseed oils.

Tin and

aluminum were the least active, while lead, iron
and zinc were intermediate.
These metal contaminants in oils can he
removed by precipitation with thioglycolic acid,
which has the property of forming complexes with
metals (45).

A similar mechanism has been suggested

as being involved in the action of certain antioxidants*
For instance it is claimed (74) that aliphatic diamines
probably exert their stabilizing action by forming
extremely stable complexes with metallic ions and
thus preventing such ions from exerting their catalytic
effect*
Salts and Acids.

The rancidity accelerating

effects of metals combined as salts either in solution
in the fats and oils or dissolved in an aqueous phase
in contact with the fats, has also been studied*
Lea (42) reports that as little as 1 part per 100
million of copper salts had a detectable influence
on the rate of oxidation, which increased rapidly
at first with increasing concentration of the metal,
but later appeared to approach a limiting value*
Hydrochloric acid has been found to have an
accelerating effect on the oxidation of oils.
Hilditch and Slightholme (27) found that boiling

olive oil with dilute hydrochloric acid reduced
the induction period from 260 to 20 hours.
Watterman and Van Vlodrop (77) found that gaseous
hydrogen chloride and sulfur dioxide promoted the
polymerization of linseed oils.

Qualitative Methods of Measuring
Oxidative Rancidity
There have been numerous methods developed
for the detection of oxidative rancidity in fats.
The earlier ones were qualitative tests only and
were usually applied to oils which had been under
natural storage conditions.

They attempted to

measure the state of rancidity of oils in contrast
with more recent accelerated tests which have been
developed to measure the susceptibility of oils to
become rancid.
Organoleptic.

The earliest and simplest

test applied to fats and oils for the detection of
rancidity was based on odor and flavor.
been called the organoleptic test.

This has

A criticism of

this test is that it is not very sensitive.

It is

also impossible to express the results of the test
on a numerical basis.
Kreis Color Test.

One of the earliest and

most used chemical tests for oxidation rancidity of
fats is the. Kreis test (38).

It is carried out “by

shaking 1 ml. of the oil with 1 ml. concentrated HC1,
followed hy further shaking 1 minute with 1 ml. 0.1^
solution phloroglucinol in ether.

The lower layer

becomes red if the oil is rancid.

Many studies of

this test have been made, one of the most thorough
being that of Fowick (61).

He concluded that the

presence of heptyllc aldehyde in rancid fats accounts
for their characteristic odor.

He also found that the

compound responsible for the Kreis test is epihydrin
aldehyde, CHgCHCHD.

This was not originally present

L_ q J
in the rancid fat in the free state but was formed
when the oxidized fat was brought in contact with
concentrated hydrochloric acid.
Several procedures have been suggested for
placing the Kreis test on a more quantitative basis.
Kerr (3*) carried out the test on samples by gradually
diluting them more and more with non-reactive oil
until no color was obtained.

Holm and Greenbank (28)

diluted the sample with petroleum ether after the test
had been carried out until the color obtained matched
a standard methyl red solution.

A still later method

was developed by Richardson (63) for the American Oil

Chemists' Society.

He measured the red color

produced in the test with the Lovibond tintometer.
Peroxide Test.

A more recent method for

detecting oxidative rancidity in oils is the
peroxide test.

This measures the loosely bound,

or active oxygen which is formed in the early
stages of rancidity.

The early technic of the test

(52) was not sufficiently sensitive to pick up the
concentrations of peroxides found in slightly
rancid oils.

Several modifications have been sug­

gested (71) t (78), one of the most successful
being that of Lea (45).

This method is rather

laborious and has been modified by Banks (2) to give
a test which is nuch simpler to carry out, yet gives
results essentially as accurate.
Schiff and Bisulfite Tests.

The Schiff

reaction has long been used as a test for oxidative
rancidity in fats and oils (6).

It consists in the

interaction of acid fuchsln with the aldehydes,
formed in the early stages of rancidity, giving a
red color.

A closely related test is the reaction

of aldehydes with bisulfite (41).
out so that the

Lea has worked this

amount of aldehydes in rancid

fat, which are considered responsible for the
objectionable odor and flavor, can be titrated directly
with bisulfite.

This method has been used by

Hamilton and Olcott (25) as one means of following

the Oxidative breakdown of oleic acid and related
compounds*
Quantitative Methods of Measuring
Rancidity

While all of these criteria give information
concerning the degree of rancidity of a fat or oil,
they are not very helpful in predicting the keeping
quality of fresh fats.

Many methods have been

developed to accomplish this latter purpose,
usually enploying as criteria one or more of the
qualitative reactions mentioned above*

The

principle involved in all such tests for sus­
ceptibility to oxidation of the fat or oil is the
measurement of the induction period.

The various

methods of doing this usually employ specific
standardized conditions.

They differ only as to

the methods used to produce these standardized
conditions and the means of following the breakdown
reaction.
Weight Increase.

Since oils take up

oxygen during autoxidation and therefore gain
weight during the reaction one of the most obvious
tests is that of determining the gain in weight of
samples of the oil when exposed in thin films to

the air, usually at elevated temperatures.

This

method has been used extensively with drying oils,
and has also been applied to edible oils (11), but
it fails to detect the early stages of rancidity
and does not measure the volatile products of autoxi­
dation.
Oxygen Absorption.

Another and more

sensitive method of measuring the oxygen involved
in the autoxidation reaction is to carry out the
observations in a closed system containing either
air or oxygen.

A manometer is connected with this

system so that any change of pressure can be observed.
This is the method used by Moureu and Dufraisse (53)
in their pioneer woric on anticocidants.

Matt ill and

Crawford (50) also used it in their earlier studies
on the stabilization of fats by antioxidants.

The

reaction has usually been accelerated by maintaining
the samples at an elevated temperature, and sometimes
using agitation.

Hew ever, Triebold, Webb and Rudy

(72) compared the results of controlled oxidation of
lard in a closed system at temperatures ranging from
70 to 95°, with and without stirring, and found that
all these variations of the method gave the same
relative results.

French, Olcott and Mattill (16)

have shown that for small amounts of fat (5 gins,
or less), the rate of oxidation is independent of
the quantity used, indicating that the oxygen
diffuses completely through this size sample.

This

indicates that stirring during the determination
of the induction period is unnecessary.

The method

has been much refined by Ifattill and coworkers (10)
(25), for the purpose of studying the reactions
taking place during the oxidation of oils.

The

method in general is too complicated, however, for
routine examination of oils.
Aeration.

One of the most practical methods

of determining the susceptibility of oils to
autoxidation is the aeration test developed by
Wheeler (?8)»

It consists in blowing air or oxygen

through the oil, which is usually maintained at an
elevated temperature, at a constant rate until
rancidity develops.

The progress of the breakdown

is followed hy making some chemical test, such as
the peroxide determination on small samples drawn
at intervals from the heated oil.

This method has

been standardized by King, Roschen and Irwin (37)
and is known as the "Swift Stability Test".

It

consists in blowing washed air at a controlled rate
through 20 ml. samples of oil in test tubes,

maintained at 100° C, until the oil has become
rancid, as determined by peroxide tests made on
small samples withdrawn periodically from the tubes.
A preliminary indication of rancidity is obtained
when the air emitted from the tubes of oil has a
rancid odor*
v
Although acceleration of the aeration
test by heating at 100° is commonly practiced, other
modifications have been proposed (66)*

In these the

test is further modified by irradiation with light or
by the introduction of metallic catalysts*

Accelerated

tests of this nature, carried out at high temperature
have been criticized by Evans (14) who claims that
certain antioxidants are inactive at temperatures above
65°, and therefore these tests do not reflect actual
conditions*
Incubation*

One of the simplest, and yet

most accurate, methods of determining the induction
period of oils is the incubation test*

This consists

in exposing samples of the oil to the air at an
elevated temperature and examining a sample at intervals
by any one of a number of methods to determine the
progress of rancidity.

In one of the first applications

of this test (23), a 100 gm* sample of oil was exposed
in an open glass vessel at 60°, and examined at

intervals for odor and flavor to detect the onset
V

of rancidity.

This method was criticized because

it was too slow and involved a large personal
error in the detection of rancidity.

These ob­

jections have been overcome in later modifications
of the test by exposing smaller samples and ex­
pressing the degree of rancidity numerically by
means of some chemical test.

The method has been

especially applicable to fish liver oils, and
other vitamin A-contalning oils, since they have
a naturally short induction period.
Various modifications of the incubation
test for stability of oils have been used.

Green-

banlc and Holm (22), exposed samples of a variety
of animal and vegetable oils (size not stated) for
10 days at 42°, then determined peroxides.

Evans

(14) employed 100-gm. amounts of cottonseed oil,
plus a catalyst, and exposed these to pure oxygen
at room temperature.

Samples were withdrawn at

intervals and peroxide determinations made.
Holmes, Corbet and Hartzler (30) exposed small
samples of several grams of fish liver oils to
an atmosphere of oxygen at 96°.

Vitamin A deter­

minations were made on test samples of the oil

removed at intervals*

The drop in vitamin A indicated

when breakdown of the oil occurred*

In the stability

test used by Lowen, Anderson and Harrison (46) 30 ml*
portions of halibut liver oils were exposed at room
temperature*

They measured both peroxide-formation

and vitamin A-breakdown and found a correlation
between the two*

Simons, Buxton and Colman (67)

exposed 2 ml. samples of fish oils in a dust-free
cabinet at 34.5°.

At intervals a sample was removed

and a vitamin A test was made on 1 gnu and a peroxide
test on the remainder.

The breakdown rate was found

to run parallel in both cases*

Control of Rancidity

Protection from Light and Air.

Since

autoxidation is dependent upon a supply of oxygen
and is catalysed by light, an obvious protection
against autoxidative rancidity of fats and oils is to
enclose them in dark glass or opaque containers*
However, it has been shown (33) that only a very small
amount of oxygen is necessary to initiate the oxidative
reaction, vhich may then proceed to the stage of
rancidity even through the concentration of peroxides
never becomes very great.

The reason for this is that

the peroxides first formed are consumed in the
progressive reactions which lead to rancidity.
Under such circumstances the peroxide test, or
other tests, Indicative of beginning stages of
rancidity, may be misleading as to the actual
degree of rancidity (35)* (60).

Use of Antioxidants

Discovery.

The inherent stability of

natural oils is due to traces of foreign sub­
stances, called antiaxidants.

This is also true

of other organic compounds which tend to autoxidize.
Moureu and Dufraisse (53) found that a trace of
hydroquinone or certain other phenols inhibited the
oxidation of acrolein and benzaldehyde.

This was

the first study of antioxidants reported.

The

authors concluded that only easily oxidlzable
substances are effective as antiaxidants.
Theories of Antioxidant Action.

The

two outstanding theories of the action of anti­
oxidants are those of Moureu and Dufraisse (53)
and of Christiansen (7)*

The former deals

exclusively with the inhibition of autoxldation,
whereas the latter is known as the chain theory.

According to the theory of Moureu and Dufraisse,
oxidant A unites with oxygen to form a peroxide,
A02. This peroxide oxidizes the antioxidant B
with the formation of a peroxide, BO, at the same
time the oxidant being transformed to a lower
oxide, A0«

AO and BO are antagonistic and mutually

react to regenerate the three original molecules
A, B and O2 in their original state.

A + C>2
A0 2 + B
AO + BO

—

■■ ■>

•

A

O

g

AO + BO
^

A + B + 02

According to this theory, the antioxidant
behaves as a true catalyst and consequently is not
used up in the reaction.

However, it has been

shown (4), (16), (24) that the antioxidant is des­
troyed in the oxidation reaction, so that this
theory is not entirely consistent with the facts.
Christiansen’s theory of antioxidants, on
the other hand, accounts for the fact that the
antioxidant is destroyed during the oxidation.

It

states that an activated molecule of the autoxidizable
substance and a molecule of oxygen unite with the
production of a peroxide and the liberation of energy.
This energy is passed on to another molecule of the

substance, thereby starting a reaction vfcich
may involve a thousand molecules, depending upon
the efficiency with which the energy is trans­
ferred, and upon the presence of foreign sub­
stances— the antioxidants.

When an activated

peroxide molecule comes in contact with a
molecule of antioxidant, the latter takes up
the energy and is itself usually oxidized in
subsequent collisions with oxygen.

The peroxide

molecule falls, however, in its turn to activate
any further molecules of the autoxidizing sub­
stance, and thus the reaction chain is broken.
The antioxidant, therefore, according to this
theory, lengthens the induction period of the
autoxidizable substance by reducing the length
of the chain reactions associated with the
oxidative process.
Mechanism of Anti oxidant Action.

It

is evident from the different types of compounds
that are effective as antioxidants, that the
mechanism of their action is not the same in all
cases.

Moureu and Dufraisse (53) believed that

the antioxidant always becomes oxidized during
the induction period of the reactant substance

and therefore that only easily ozidizable sub­
stances would be effective*

However, this is not

the case with certain Inhibitors, such as the
acid type of antiaxidants (22), (4-2)*

Additional

evidence for the belief that more than one mechanism
is involved in the protective action of antioxidants
is the fact that the combined action of certain
compounds is much greater in effect than would
be anticipated from their individual effects (30),
(58).
Evans (14) has suggested that antioxidants
increase the stability of oils by forming chemical
complexes with peroxidants, such as traces of metals,
etc., thus preventing the latter from hastening the
oxidation process.
cases.

This seems to be true in certain

For Instance, certain compounds act as

inhibitors under some conditions but not under others.
Thus, Lea (42), found that potassium cyanide acted
as an antioxidant if the oil contained a trace of
metallic catalyst, but not in a fresh oil free from
metals.
It is assumed that all antioxidants act in
the same manner to the extent that none of them is
effective after the induction period has ended (16).
This is apparently due to the fact that in partially

oxidized oils peroxides or oxidation catalysts are
present in concentrations sufficiently large to
prohibit any further protection by the antioxidant.
Early Work on Antioxidant a.

Moureu and

Dufraisse (53) studied the inhibiting effect of
hydroquinone upon the oxidation of acrolein and
benzaldehyde.

Anderegg and Nelson (1) obtained

probably the first evidence of natural inhibitors
when they found that the oxidative destruction of
vitamin A in experimental diets was greatly retarded
when animal fats, such as lard or cod liver oil,
were replaced, partially or wholly, by vegetable
oils, even though the latter were of equal or
greater saturation.

Types of Antiaxidants

Hydroquinone.

Huston and Lightbody (32)

showed that small amounts of hydroquinone incorporated
in the diets of rats increased protection against
ophthalmia.

They suggested that the hydroquinone

acted by preventing oxidation of the protective
vitamin A.

Huston, Lightbody and Ball (33) proved

that hydroquinone in a concentration of 1 in 2000
was capable of delaying the oxidation of milk fat

and of preserving the vitamin A in the fat and
in cod liver oil.

Wagner and Brief (75) studied

the effects of antiaxidants in linseed oil and
found that hydroquinone was the best of all those
tried.

Jfarcus (47) noted that vitamin A adsorbed

on finely divided solids such as lactose oxidized
in a few hours*

The presence of if. hydroquinone

in the vitamin A concentrate adsorbed on the
solids delayed this destruction up to 15 to 45
days.

Holmes, Corbet and Hartzler (50) studied

the stabilizing effect of hydroquinone on halibut
and cod liver oils.
Other Hydroxy Compounds.

Matt ill (49)

demonstrated that the development of rancidity and
the breakdown of vitamins A and E in unsaturated
animal fats was retarded by substances containing
hydroxyl groups, which he concluded might be
sterols.

He found that wheat germ oil prolonged

the induction period of cod liver oil.

Greenbank

and Holm (22) studied hydroxy compounds, such as
catechol, resorcinol, pyrogallol and phloroglucinol,
as antioxidants in cottonseed oil.

Their effect

was found to depend on the location of the hydroxyl
groups on the benzene ring.

Olcott and Mattill (55)

obtained a crystalline compound from the
unsaponifiable portion of the lipids of
lettuce v&lch contained one hydroxyl group
and was an active anti oxidant.

Later (57)

these workers suggested the name "inhlbitols"
for the antioxidant obtained from lipid
fractions of vegetables and vegetable oils.
Xnhibitols were found present in lettuce, tomatoes,
carrots, alfalfa, spinach; in the oil from wheat
germ, cottonseed, corn, sesame, palm, soybean,
and peanut but not in the oil from olive, cod
liver, palm kernel or castor bean.

The proper­

ties of these inhlbitols and those of vitamin E
were so similar that it was found impossible to
separate the two.
hydroxyl groups.

Both were found to contain
The activity was destroyed

by reagents which react with the hydroxyl group.
Strangly enough, the isolated inhlbitols were
found to protect lard and purified fatty acids
and esters but not the vegetable oils from which
they came.
Evans, Emerson and Emerson (15) isolated
three individual compounds possessing vitamin E
activity from wheat germ and cottonseed oils.
The se were designated o< ,

/S

and

Y

tocopherol.

The relative effectiveness of these three
compounds as anti oxidants for lard at 75° C
was investigated by Olcott and Emerson (59)*
who found them to be increasingly effective in
the following order:

o<

Rtospholiplds.

,g

and

Y

•

Evans (14) found vegetable

lecithin to be a good antioxidant at 0*05 to 0 *1%
concentration in vegetable oils.

Holmes, Corbet

and Hartzler (50) also found crude lecithin to
be an effective antioxidant for vitamin A.

They

used lecithin prepared from soy beans as an anti­
oxidant for halibut and cod liver oils.

On the

other hand, Olcott and Mattill (56) claimed that
commercial lecithins had only moderate antioxlgenic action on cottonseed oil, little on
lard and none on a mixture of lard and cod liver
oil.

They found these materials contained little

true lecithin, the little effect there was being
due to their content of cephalin.
Acids. Salts and Amines.

A variety of

acids and amines have been found effeotive as
antioxidants.

Hilditch and Sleightholme (27)

found treatment of olive oil with concentrated
sulfuric acid increased its induction period.
Gresnbank and Holm (22) studied the antioxidative

effect of various organic acids on cottonseed
oil*

The most active acid found was maleic, while

its isomer, fumaric, was practically inert*

Other

acids having a fair degree of activity were
citraconic, itaconic and citric.

The effectiveness

of maleic acid as an antioxidant for a number of
oils was investigated*

Its effectiveness was

about the same in all cases except corn oil, where
it was considerably greater.
The antioxidative effect of various watersoluble substances was investigated by Lea (42).
The antioxidants were dissolved in water, which
was in contact with lard.

The aliphatic hydroxy

acids, such as lactic and gLycollic, the ethanolamines, and maleic acid were moderate, while
polybasic hydroxy acids, such as tartaric and
citric acids, were powerful antioxidants*. The
aliphatic amino-acids were all powerful antioxidants.

Proteins also had considerable

protective ability.

It was noted that these

substances were effective antiaxidants even when
the water content of the fat was only 0.25
Also, the salts of acids, such as citrates,
lactates, malonates, etc., were active as
antioxidants, as well as the acids themselves*

Olcott and Mattill observed (58) that the
crude esters of vegetable oils were stabilized by
oxalic and maleic acids and by sulfuric and
phosphoric acids and their salts.

Citric and

pyruvic acids were also protective.

Salts and

esters of dicarboxylic acids were inactive, in
contrast to the findings of Lea, indicating that
the carboxyl groups must be free.
Complexes.

In addition to the individual

chemical compounds mentioned above, many complex
substances have been used as antloxidants.

Musher

(54) patented the use of a great number of finely
divided vegetable materials, such as oat, barley
or soy bean flour, as antiaxidants.

Lowen,

Anderson and Harrison (46) have also studied the
use of oat flour as an antioxidant*

They found it

to be effective for lard and vegetable oils, but
rather ineffective for fish oils.
Olcott and Mattill have made a preliminary
classification of antloxidants (58), dividing them
into three types:

the acid type inhibitors in

the first type, inhlbitols and hydroquinone in
the second, and phenolic inhibitors other than
hydroquinone, in the third.

The acid type was found

active in vegetable oils and crude vegetable oil

esters, but inactive in lard and lard esters,
and in purified fatty acids and esters*

The

second type of Inhibitors was active in the latter
group, but not in the former*

The phenolic type

was active in all the vehicles mentioned*
Synergism of Antioxidants.

A further

observation noted in this work was that any type I
inhibitor, when used with any type II or III com­
pound prolonged the induction period of certain
fats and oils longer than the summation of their
effects would be expected to do*

This synergism

of antioxidants was also noticed by Holmes, Corbet
and Hartzler (30)*

The latter workers found that

while either hydroquinone or lecithin afforded
protection alone to halibut or cod liver oil, a
combination of the two was better than would be
expected from their computed additive effects*
Attempts at Isolation of Antioxidants.
Several attempts have been made to isolate anti­
oxidant compounds from natural products*

Royce

(66) suggested that gossypol, a polyhydroxy phenolic
compound was the chief antioxidant in cottonseed
oil.

Olcott and Mattill (55) prepared a crystalline

antioxidant from the unsaponifiable fraction of
the lipids from lettuce*

This substance seemed

to be distinct from vitamin E, since it was

soluble In 92f. methyl alcohol, while the
latter was preferentially soluble in petroleum
ether*
Hilditch and co-workers in England (27),
(3), (20) have attempted to isolate the natural
antioxidants of olive, linseed, tung, and other
oils.

They found that saponification practically

eliminated the Induction periods of these oils,
that is, alkalies were detrimental.

Dilute

hydrochloric acid also destroyed the antioxidant.
They found that the natural antioxidants of olive
and linseed oils were apparently removed by boil­
ing with water; at least the induction period of
the oils disappeared.

However, they were unsuccess­

ful in recovering the antioxidants from the water
extracts.

Working with linseed oil and linseed

press cake, they found the latter much richer in
antioxidants than the former.
with soybean oil.

This was also true

They found that the inhibitor

present in soybean press cake could be extracted
with acidulated methyl alcohol and was soluble
in acetone.

The final concentrate obtained in this

manner was a viscous oil which readily reduced
aimnoniacal silver nitrate and Fehling's solution,
but not iodine.

It did not give a coloration

with ferric chloride*

This concentrate therefore

corresponded In properties to the inhlbltol
fractions of Mattill and coworkers*
Olcott and Mattill (57) Isolated anti­
oxidant fractions in the form of viscous oils from
the unsaponifi able portions of the lipids of a
number of substances including lettuce, tomatoes,
wheat germ oil, cottonseed oil, corn oil, etc*
They suggested the name, inhibitol, to designate
this type of natural antioxidant*

With the ex­

ception of the inhibitol from lettuce, which was a
crystalline solid distinguishable from vitamin E
fractions by its solubility, all were inseparable
from vitamin E concentrates prepared from the same
sources*

This fact became understandable when

three tocopherols, all having vitamin E activity
but in different degree, were isolated from wheat
germ oil by Evans and coworkers (15)*

All were

later shown to have antioxidant activity (59)*
although not in direct ratio to their vitamin E
activities*
Thus it appears that the large group of
inhlbitols are in reality mixtures of the naturally
occurring tocopherols.

From recent work of Hickman

(26), using a newly developed method of molecular

distillation of oils, it appears that the
natural antioxidants present in vegetable oils
are largely of this type.

METHODS

Development of the Induction Test

Although many tests have been suggested
for evaluating the breakdown periods of oilst it
was felt that since vitamin A-containing oils were
to be studied a test employing the actual break­
down of vitamin A as a criterion would be most
satisfactory.

The first stability studies were

carried out by blowing oxygen through 25-gram
sano>les of fish liver oils contained in deep test
tubes heated in a steam bath at 100° C.

At in­

tervals a small sample was withdrawn and its
vitamin A content determined by the antimony
trichloride color test*

Plotting these vitamin

A values against time gave a curve representing
the breakdown of the vitamin A in the oil*

The

effect of adding various inhibitors could be
evaluated easily in this manner by comparing the
breakdown curves for an oil before and after
addition of a particular inhibitor*
This method had several disadvantages*
It was difficult to keep the flow of oxygen
through the oil constant, only a very few oil

samples could be studied at ozxce without
complicated equipment, and oils containing
phospholipids could not be handled this way at
all beoause of excessive foaming*
It was found that a much simpler method
of determining the relative stability of vitamin
A-containing oils consisted in exposing small,
fairly uniform samples of the oils to air at
elevated temperatures*

At intervals one sample

of the serieB was removed and a determination
made of the vitamin A remaining in it*

These

determinations were made in the earlier studies
by means of the antimony trichloride color method
and later with the Hilger Vitameter— the latter
method being considerably more accurate.

Since

this method of determining stability offered many
possibilities, studies were made of its charac­
teristics and limitations*
Effect of Surface Area and Temperature*
Samples of equal weight of halibut liver oil were
placed in open, cylindrical containers of different
cross sections and exposed to the air, in the dark,
at 27*5° and 37° 0 respectively*

At intervals a

sample was removed, mixed thoroughly and its
vitamin A content determined.

The induction period

of a particular sample of oil was designated as the

40*

time required for its vitamin A content to
drop to 80^ of the original value.

The data

are given in table 1.

TABLE 1.
Effect of Surface Area and Temperature on the
Induction Period of Halibut Liver Oil.
Sample

(a)
(b)
(o)

Surface Area
(Sq. Cm.)

12.6
7.07
3.14

Wt. Sample
(Goa.)

0.82
0.82
0.82

Induction Period,
Hours
37°
27.5°

8
12
18

Ratio
27.5°/37

15
17
40
Average

1.9
1.42
2.22
1.85

This study indicates that the in­
duction period varies inversely with the area of
the exposed sample, and a decrease in temperature
of 10° approximately doubles the length of the
induction period.
Effect of Depth of Sample.

A series of

4.6 gm. (3 ml.) amounts of halibut liver oil was
weighed into cylindrical containers of the same
size so that the column of oil in each was 2 cm.
in diameter and 2.8 cm. deep.

These were exposed

to the air, in the dark, at 37°.

At proper time

intervals one of the containers was removed and a
sample carefully taken from the top surface of

the oil.

Next a sample was removed from the

bottom.

This was done by submerging the tip

of a pipet (on the other end of which was a
rubber bulb) to the bottom of the oil, gently
blowing out the drop of oil which had risen into
the pipet, then drawing up the sample.

The

vitamin A was then determined in these two samples.

TABLE 2.
Effect of Depth of Samples on the Induction
Period of Halibut Liver Oil.

Hours at 37°
___________
284
330
379
450
501
672

Original Vitamin A
Top
Bottom
84.0
82.5

81.1
73.4
69.3
61.0

86.0
85.1
82.9
78.5
77-0
66.4

These results indicate that diffusion of
air into oil is sufficiently rapid that, even in
deep samples the rate of breakdown is approximately
the same throughout*
Agitated vs. Stationary Samples.

Some

workers have suggested that in induction tests of
oils the samples must be agitated during exposure.
Therefore the following comparison of methods was

made.

Vitamin A-breakdown tests were run on a

group of various types of oils hy two methods*
One method was that mentioned above, consisting of
exposing approximately 0.2 gm* samples in Brlenrasyer flasks at 37°*

The other involved agitating

3 ml* samples of the oils in T tubes in a specially
constructed rocking device, according to a method
suggested by Dr* K. Hickman.

The arms of the T

tub os were each two inches in length so that the
oil traveled over a distance of four inches and
back during each revolution of the driving device,
which operated at four revolutions per minute.
The whole apparatus was operated in a 37° room.
The open ends of the T tubes were loosely plugged
with cotton during operation, to permit free
access of air*

At intervals samples of about

0.25 gm* of oil were withdrawn, dropped into
tared flasks to be weighed and diluted for
vitamin A determinations by the Vitameter.

The

endpoint adopted for the induction test by both
methods was the same— namely length of exposure
required to reduce the vitamin A content to 80%
of the original value*

TABLE 3*
Comparative Induction Tests with Agitated
and Stationary Samples.
No.

32089
52099
32109
32H9
32129
32139
32149

Type of Oil
__________

Induction Period (Hours)
Agitated
Stationary

Distillate
59
Distillate
U5
Fish Liver Oil
57
Fish Liver Oil
54
4
Concentrate
4
Concentrate
Reference Cod Liver OillOO

60
124
51
49
4.5
5.5

100

The two methods of measuring stability of
oils gave closely agreeing results in this series.
The method employing stationary samples was much
simpler to carry out, and therefore preferable.
Peroxide Formation in Relation to Vitamin
A-Breakdown.

A series of approximately 0.25 gm.

samples of halibut liver oil in 50 ml. Erlenmeyer
flasks was exposed at 37°.

At intervals a sample

was removed and its vitamin A and peroxide content
determined.

The data are shown in table 4, where

the vitamin A determinations are expressed as per
cent original vitamin A remaining and the peroxides
as peroxide numbers.

By peroxide number is meant

the number of ml. of 0.002 N thiosulfate per gm.,
or millimoles of peroxide oxygen per kilogram of oil.

TABLE 4
Relation of Peroxides to Vitamin AEreakdown.
Hr s. at 37°
0
16
18,2

% A Remaining
—

Peroxide Number
3.4

100,0

28.0

89*5

55-0

20

73-0

80,0

22
24

30.4
38.3

131.0
162.0

These data are presented in graphic form
in Figure 1, where the upper curve represents vitamin
A-breakdown and the lower one, the increase of
peroxides.

The data indicate that peroxides develop

in the samples at about the same rate at which the
vitamin A breaks down.

That is, both the vitamin A-

breakdown and the increase of the production of
peroxides make suitable criteria for the induction
test8 on fish liver oils.
Change of XJltra-violet Absorption of
Vitamin A During Breakdown,

In order to study

the induction test more thoroughly, the following
experiment was performed.

A series of approximately

0,25 gm. samples of halibut liver oil in open
ml. flasks was exposed to the air at 37°.

$0

At in­

tervals, beginning at the end of 12 hours, one of
the samples was removed and an absorption spectrogram

1
a

u

a

WAWW

in the ultra-violet region was made on the medium
quartz spectrophotometer.*

These absorption curves

are presented in Figure 2.

The numbers on the

curves represent the respective number of hours
exposure at 37°*

It will be seen that as the

peak of absorption drops with time, there is also
a gradual shift of the peak to the left, beginning
at about the twentieth hour.

The small curve at

the left is the induction curve of the same halibut
liver oil, obtained by plotting the extinction at
various periods of exposure, as determined on the
spectrophotometer, against time*
Outline of Induction Test Method.

The

method finally adopted for induction tests of oils,
and used throughout the eaperiments to be reported,
was carried out as follows:
A series of 0.2 to 0.25 gm* samples of the
vitamin A-containing oil was accurately weighed into

50 ml* Erlenmeyer flasks*

The flasks chosen were

of uniform design with bottoms as flat and smooth
as possible*

Each sample of oil was introduced

into the flask from a pipet, allowing it to drop
squarely in the center of the bottom so as to
permit it to spread out in a uniform, thin layer.
*Spectrograms made by Dr. J. M. Vandenbelt in the
Section of Physical Chemistry at the Kedzie
Chemical Laboratory.

ff m
m

'• > -

x /

A

The flasks were then placed In an incubator at
37°C, care being taken to see that they were all
on the same level so as to insure uniform temperature.

The flasks were left open for free

circulation of air.

At intervals one flask from

the series was removed from the incubator.

If the

vitamin A, only, was to be determined the sample
was dissolved in isopropyl alcohol and the assay
made on the Vitameter.

In case the peroxides were

to be determined simultaneously, the sample was
dissolved in chloroform and diluted to 10 ml. with
the same solvent.

A 1 ml. sample was taken from

this solution and diluted further for the vitamin A
determination.

To the remaining 9 ml. of solution

were added 18 ml. glacial acetic acid and then two
drops saturated potassium iodide solution.

The

mixture was shaken and allowed to stand 10 minutes
in the dark.

Fifteen ml.

15%

potassium iodide were

than added and the liberated iodine titrated with
freshly diluted 0.002 N thiosulfate.

A blank

was carried out under the same conditions.

This

is based on the Banks modification of the Lea
Peroxide test.
In carrying out the induction test, the
first few vitamin A determinations generally checked
those of the unexposed oil.

As succeeding samples

were tested, however, they showed a slight
falling off, and finally a sudden drop.

The

peroxide content rose at the same time.

The

point where the vitamin A-breakdown curve
reached 8of« of the original vitamin A activity
was arbitrarily taken as indicating the end of
the induction period.

This was chosen since it

was considered that a difference of

20%

was

beyond the experimental error of the method and
represented a significant drop in vitamin A,

Relation of Induction Test to Natural
Conditions

In order to establish a relationship
between the above described induction test and
breakdown of vitamin A under actual conditions, an
experiment was carried out to determine the effect
of the amount of oil sample on the induction
period.

Samples ranging from 0,3 to 5,0 ml, of

oil were placed in 7 ml, bottles, one series
tightly capped and the other open.

At intervals

one of the bottles of each series was removed,
the whole sample mixed and the vitamin A
determined.

The data are presented in table 5

and illustrated in Figure 4-.

.t i l

TABLE 5.
Effect of Volume of Sample on the Induction
Period of Halibut Liver Oil.

Sample
(Ml.)

Induction Period,
Hours______
Capped
Open
10
38
160

0.313
1.23
5.0

11
40
230

From these data it is evident that the in­
duction period is directly proportional to the volume
of oil in the open containers; also that there is no
appreciable difference between the induction periods
of the oil in open or closed containers, except in
the case of the 5 ml. samples.
The induction period of this halibut liver
oil was

kours as determined by the standardized

method described above.

Since the induction period

of a 5 ml. sample in a capped bottle at 37° was 230
hours, it would be approximately twice this, or about
500 hours, at room temperature.

Thus there is a

factor of approximately 100 which represents the ratio
between the value obtained by the induction test and
the expected breakdown time of the same oil stored in
5 ml. amounts in practically full, capped bottles
at room temperature.

EXPERIMENTAL RESULTS

Stability of Various Types of Natural
Vitamin A-Containing Oils

Using the induction test described above,
data on the stability of several types of fish oils
containing vitamin A were collected over a period of
several years.

Some typical data are recorded below

in tables 6 and 7*

The halibut liver oils, and

mixed fish liver oils, were prepared hy alkali-digestion of the livers, followed by gravity separation of
the oil.

The cod liver oil was prepared by the ordinary

commercial process consisting of steam treatment of
the livers to separate the oil, which rises to the
top and is drawn off.

The tuna liver oils were prepared

by extraction from the livers with oils, etc.,
following a preliminary heating to coagulate the liver
protein.

The solvent-extracted oils were prepared

by coagulating the livers by heat, followed by
exhaustive treatment of the residue with either ethyl
ether or petroleum ether.

Part of the viscera oils

were prepared by alkali digestion and part by
solvent-extraction.

TABLE 6

Induction Periods of Fish Oils Prepared by
Alkali Digestion or Steam Rendering*

Halibut Liver Oils
Sample Number

Induction Period, Hours

76405
79365
79375
81036
82246

7.5
10
30
16
10

83896
88726
88756

19

25

93587

14

1878

18

16

2028
2978
10388
10398
10438_________________
Average (15)
Tuna Liver Oils

12
25
18
12
21
16*9

(Oil-Extracted)

80006
80016
80066

80
70
120
75
90
40
50
70
42

80076

80166
80266
80276
80756
80876
80886
60
82206
50
82216
70
82406
32
82436
60
82446__________________ 72
Average (15)

65.4

51.
TABLE 6 - Continued
Cod Liver Oils
Sample Number

Induction Period. Hours

82506
82516

100
100

99777
32149

200
100
Average (4)

125

id Fish Liver Oils
24
51
16
10
20
20
16
48
12
17
52
17
18
21
19

79255
81576

82226
82236

82796
82806
83896
84026
88706
88766
11918
11948
11968
13818
13888
Average (15)

20.7

1 Viscera Oils
13088
13808
15198
15208
15618
15748
15758
16899
Average (8)

12
4.5
7.5
12
12
4.5
7.5
5
8*1

TABLE 7
Induction Periods of Fish Oils Prepared
by Solvent Extraction.
Induction Period,
Hours

Type

Number
82596

6058
5268
5498
5598
5648
6758
10948

Halibut Liver Oil
Halibut Liver Oil

168

Halibut Viscera Oil
Halibut Viscera Oil
Halibut Viscera Oil
Halibut Viscera Oil
Halibut Viscera Oil
Black Cod Viscera Oil

100
125

81

82
100
65
164

Average (8)

111

The data indicate that cod liver oils and solventextracted oils are the most stable, while alkali-digested
oils, as a class, are least stable.
the two is time liver oil.

Intermediate between

TCiis can be accounted for by

the method used in the extraction of the tuna livers,
even though the latter are first digested with alkali.

Conditions Affecting Stability of Vitamin
A in Oils

Free Fatty Acids.

In table 8 are presented

certain data collected in connection with experiments
on the oils obtained from fish viscera.

They indicate

the entire lack of correlation between free fatty acids
and stability.

Although these solvent-extracted viscera

oils, with a few exceptions, are extremely
stable, they are also high In free fatty adds.
The fatty acid content aeems to be a function of
the particular lot of viscera, rather than a fac­
tor Influencing the stability of the oil.

This

is indicated by the fact that each pair of oils,
representing oils from the caecum and intestines
respectively of a particular lot of viscera,
are similar in free fatty acid content.
TABLE 8
Relation of Tree Fatty Acids to Stability of
Vitamin A in Fish Viscera Oils.

Viscera Oil
Vitamin A
No.______ Source___________U.S.F. u/gm.
0127
0187
0757
0777
0997
1007
1247
1257
1387
1397

1608
1818

Halibut Caecum
Halibut Intestine
Ling Cod Caecum
Ling Cod Intestine
Black Cod Caecum
Black Cod Intestine
Tuna Caecum
Tuna Intestine
Mackerel Caecum
Mackerel Intestine
Halibut Caecum
Halibut Intestine

750,000
600,000
65,000
21,100
153,000
120,500
46,200
37,400
27,000

15,700
67,500
15,660

Per Gent
Free Acid #

25.6
27.3
10.9
9.9
18.5
18.9

62.6
61.5
24.5
25.1

50
7.5

100
180
115
168+
70

66
8

56.6

7
140

54.7

120+

Calculated as oleic acid.

Effect of Peroxides and Dilution of Oil
Medium.

Induction
Period, Hours

Other workers have found that when

peroxides are added to vitamin A-containing oils

the vitamin A breaks down at a rate proportional
to the concentration of peroxides present*

In a

preliminary experiment a high-potency vitamin A
concentrate was diluted at different concentra­
tions in corn oil.

The vitamin A-breakdown as

well as peroxide-formation were followed in
these dilutions of vitamin A, while only the
peroxide-formation was followed in the c o m oil
diluent*

It was found that the rate of peroxide-

formation and vitamin A-breakdown were propor­
tional to the concentration of vitamin A, but in
all cases were much greater than the rate of
peroxide-formation in the c o m oil diluent alone.
It was thought this might possibly be due to the
peroxide content of the vitamin A concentrate,
which unfortunately was rather high*
Another experiment was therefore carried
out in which the source of vitamin A was a vitamin
A distillate of practically negligible peroxide
content (kindly supplied by Dr* K. Hickman).

This

had an E?-^
of 264- at 528 Mu. Induction tests.
1 cm.
involving both vitamin A-breakdown and peroxideformation, were carried out on this distillate
and on 5» 10 and 25-fold dilutions of the same
in corn oil.

Peroxide tests were made on the

diluent c o m oil and also on a peroxidized cod
liver oil, having a peroxide number of 280, diluted
1:80, 1:400 and 1:800 with this diluent oil.

The

vitamin A induction curves and peroxide curves are
shown in Figure 3.

They indicate, as in the pre­

liminary experiment above that the rates of vitamin
A-breakdown and peroxide-formation are inversely
proportional to the dilution factor.

The addition

of peroxides to the c o m oil in fairly large
amounts did not hasten the peroxidation of the latter.
Therefore it is evident that the autoxidation of
these dilutions of vitamin A was affected more by
the vitamin A itself than by the peroxides
introduced with the vitamin A.
Effect of Different Oil Media.

In table 9

are presented data on the induction periods obtained
for 1:80 dilutions of a highly purified vitamin A
concentrate in different oil media.

This illustrates

the inherent difference in the protective effect of
various vegetable oils for vitamin A; also the
Increased protective effect of one of these oils
when a small amount of hydroquinone is added to it.
This method provides a means for measuring the
relative antioxidant content of a natural oil by
dissolving a vitamin A concentrate in it and

&3br 3&/X \

%

«

mm */*&V

K>ivJ

56.

comparing its breakdown period with a standard.

TABIE 9
Effect of Type of Oil Used as Diluent on
the Stability of Vitamin A.
Diluent

Induction Period f Hours

Corn Oil with 0.05% Hydroquinone
Cottonseed Oil
Linseed Oil
Wheat Germ Oil
C o m Oil
Peanut Oil
Coooanut Oil

Effect of Oxidation Catalysts,

140
545°
50
27.5
15.5
7

(a) Salts.

In an experiment (to be described later) aimed at
separating the phospholipid fraction from a halibut
viscera oil a procedure was employed in which an
alooholio solution of calcium chloride came in con­
tact with the oil.

When an induction test was made

on this calcium chloride-treated oil it was found to
be extremely unstable.

Next, the effects of various

salts on the stability of the same viscera oil were
studied.

This was an attempt to find whether the

calcium ion or chloride ion caused the effect noted
above, although the series included many other
compounds, such as the salts of cobalt, copper and
iron.

The procedure consisted in allowing ten gram
samples of the oil to stand overnight in contact with
10% of their weight of a particular salt, which had
been finely powdered in a mortar*

The next morning

the oil was filtered to remove the salt and an in­
duction test carried out as described above*

The

data are presented in table 10*

TABLE 10
Effect of Various Salts on the Induction
Period of Pish Viscera Oil*
Treatment
Cobalt Nitrate (6 H20)
Copper Nitrate (3 H jjO)
Calcium Chloride (Anhyd.)
Magnesium Chloride (6 H 2O)
Stannous Chloride (2 H 2O)
Manganous Sulfate (4 H 2O)
Ferric Chloride (6 H 2O)
Barium Chloride (2 H 2O)
Sodium Chloride
Cadmium Chloride (2 H 2O)
Magnesium Carbonate
Ferric Citrate
Calcium Carbonate
Magnesium Sulfate (Anhyd*)
Sodium Acetate (Anhyd*)
Ammonium Chloride
Ammonium Phosphate, Monobasic
Sodium Nitrate
Calcium Phosphate, Monobasic
Calcium Citrate
Ammonium Nitrate
Sodium Phosphate, Dibasic
Magnesium Citrate
Sodium Sulfate (Anhyd*)
Control Oil Alone

Induction Period,
Hours
2*5
2*5
3
3
4*5
15
16
23
30
32
39
42
48
60
62
62

63
63
64

65
70
73
75
82
82

Per Cent
of Control
3*0
3*0
3*7
3*7
5*5
18*3
19*5
28.0
37*0
39*0
47*5
51*3
58*5
73*0
75*6
75*6
76*8
76*8
78*0
79*3
85*4
$9*0
91*4
100*0
100*0

The data show a great range of effect;
from almost complete elimination of the induction
period with the cobalt and copper nitrates to no
effect at all with sodium sulfate.

It is hard to

detect any trend of effectiveness, although the
chlorides seem to have the strongest action as a
group.

This is interesting in view of the later

findings with hydrochloric acid.
(b)

Acids.

Following this study of

salts, an experiment to detexmine the effect of
acids on the induction period of vitamin A oils
was carried out.

In order to obtain maxi mum and

reproducible contact between the oil and the acid,
the oil was dissolved in petroleum ether and
washed with 8o£ methyl alcohol in which 1% of the
acid was dissolved.

The method of treatment

was well standardized and was used later in
studies of the antioxidant content of various oils.
A control test had to be made with 8 0 methyl
alcohol alone, since this had some effect.

The

results obtained by treatment with common acids
are presented in table 11.

TABLE 11

Effect of Acids on the Induction Period
of Halibut Liver Oil.

Treatment

Nunfcer

6058
19899
24139
27509
28169
24129

Induction Period,
Hours

Control Halibut Liver Oil
(6058) Washed with 8o£ Methyl Alcohol
Same as 19899 plus if. Acetic Acid
Same as 19899 plus 1> Phosphoric Acid
Same as 19899 plus 1> Hydrochloric Acid
Same as 15899 plus 1> Sulfuric Acid

81
50
50
55
5

126

It is evident from these data that hydro­
chloric acid greatly accelerates the breakdown of
vitamin A in halibut liver oil.

On the other hand,

sulfuric acid has a protective effect.

Acetic and

phosphoric acids have very little effect.
The effect of hydrochloric acid was studied
further.

In table 12 are data showing the effect of

treatment of the oil with various amounts of this
acid.

In this case the acid was added directly to

the oil; the 0.5^ level being added undiluted, while
for the O.O^f. and 0.005% levels the acid was first
diluted 10-fold and 100-fold respectively with
absolute ethyl alcohol.

TABUS 12

Effect of Various Amounts of Hydrochloric Acid
on the Stability of Vitamin A In Halibut
Liver Oil.

Induction Period,
Hours ____
37020
Control Oil
37020-A With 0.5% HC1
37020-B With 0.05% HC1
37020-C With 0.005% HG1

60
1*5
29
45

These data indicate that treatment of halibut
liver oil with as little as 0 .005% hydrochloric acid
has an appreciable detrimental effect on its stability.
An induction test, in which both vitamin Abreakdown and peroxide-formation were measured, was
carried out an a sample of halibut liver oil to which
0.5% hydrochloric acid had been added.

This indicated

that the presence of the acid not only hastened the
destruction of vitamin A, but also increased the
production of peroxides.

The data are in table 13,

TABLE 13
Effect of Hydrochloric Acid on the Development
of Peroxides in Halibut Liver Oil.
Treatment
____________________
Control
Control
Same as
Same as
Same as

Halibut Liver Oil 37020
Halibut Liver Oil plus 0.5% HC1
(2) after 1 hr. at 37°
(2) after 2 hrs. at 37°
(2) after 3 hrs. at 37°

Peroxide
5.49.2
14.4
15•5
33.4

% Vitamin A
NumberRemainin
100
100
71

67
58

61.

To find If hydrochloric acid also
accelerates the autoxldation of non-vitamin A
oils, an experiment was carried out in which
samples of c o m oil were exposed in the incubator
at 37° in exactly the same manner in which the
vitamin A induction tests were oarried out.

Part

of the samples had 0.5^ hydrochloric acid added*
Peroxide tests were made on these samples at in­
tervals*

The data, presented in table 14, indicate

that hydrochloric acid has a decided accelerating
effect on the accumulation of peroxides in c o m oil*

TABLE 14
Effect of Hydrochloric Acid on the Develop­
ment of Peroxides in C o m Oil.

Hours at 37°

Peroxide Nunfcer

Control C o m Oil
73

150
195
245
264
313
384
435

600

2
9.5
17.8
15.8
20.7
38.4
28.8
32.4
40.5

Corn Oil plus 0.5f. Hydrochloric Acid

0

16
24
70
122
213

6.1
8.9
17.6
63.7
234
310

Certain experiments were next carried out
to study the mechanism of the action of hydro­
chloric acid on oils.

Some representative data

from these experiments are presented in table 15*
It was found that the effect was the same .whether
the oils were washed with methyl alcohol contain­
ing hydrochloric acidf or were treated directly
with the concentrated acid.

Also the result was

the same when the oil was treated with dry
hydrogen chloride gas.

The action proved to be

a reversible one, since the stability of oils
which had been treated with hydrochloric acid was
restored practically to the original value when
they were treated with potassium hydroxide,
either in the form of alcoholic potash or powdered,
solid potassium hydroxide.

It was found that a

certain optimum amount of alkali was necessary to
restore the stability of such oils; less than this
amount apparently reacted only with the free
fatty acids of the oils.

TABLE 15

Effect of Hydrochloric Acid and Potassium
Hydroxide on the Induction Period of
Halibut Liver Oil,
Number

6058
19899
29179
31769
33589
33599
38990
42670

Treatment

Induction Period,
Hours

Control Halibut Liver Oil
(6058) Washed with 8of. Methyl Alcohol
Same as 19899 plus 1% Hydrochloric Acid
(29179) Washed with 8of- Methyl Alcohol plus
1 % KOH
(6058) Treated with 0.5^ I^ydrochloricAcid
(33589) Treated with Dry KOH
Control Halibut Liver Oil j^2
(38990) Treated with Dry Hydrogen Chloride Gas

8l
50
1*5
53
2
70
28
5

Since a majority of the salts which were
active as oxidation catalysts for vitamin A oils were
chlorides, it was thought this might be related to
the similar activity of hydrochloric acid.
ments showed this to be the case.

Experi­

The data from

some typical experiments are presented in table 16.
When the control halibut liver oil stood in contact
with dry calcium chloride its induction period dropped
from 60 hours to less than one hour.

However, when

this extremely unstable oil was treated with dry
sodium hydroxide in an amount slightly in excess of
that required to neutralize the free fatty acids
present the induction period was restored to 36 hours.
The similarity of action of hydrochloric acid

and calcium chloride in reducing the induction
period of oils would indicate that the action of
the latter migrt be due to interaction with the
free fatty acids of the oil to produce free
hydrochloric acid.

There is evidence for this

supposition in the data presented in the latter
part of table 16.

The control viscera oil, 5598,

was first treated with aqueous sodium hydroxide
to remove free fatty acids*

This t reatment

reduced its induction period from 82 to 30 hours,
due to the solvent effect of the water on the
antioxidant.

Five-gram amounts of the control

oil and the acid-free oil were dissolved in 25
ml. amounts of ether and 0.5 gm. anhydrous calcium
chloride added to each.

After 24 hours at room

temperature the ether solutions were filtered free
from calcium chloride and the ether removed by
distillation.

The effect of -She calcium chloride

on the induction period of the acid-free oil was
nuch less than on the control oil containing
free fatty acids.

TABLE 16

Relation of Calcium Chloride and Free Fatty
Acids to the Induction Periods of
Fish Oils.

Number
37020
39220
39230
5598

8658
8668

8508

Treatment

Induction Period,
Hours

Control Halibut Liver Oil
(37020) Treated with Calcium Chloride
(39220) Treated with Sodium Hydroxide
Control Halibut Viscera Oil
(5598) Treated with Calcium Chloride
(5598) Treated to Remove Free Fatty
Acids
(8668) Neutral Oil Treated with Calcium
Chloride

(c) Metallic Soaps.

60
0.5

36
82
2
30
12

Among the commonest

of the oxidation catalysts for oils are the metallic
soaps, such as those of cohalt which are used as
driers in paints.

This type of oxidation catalyst

may be introduced in the production of fish oils
and concentrates due to their contact with metal
during processing.

Experiments along this line

were carried out in which fish oils were purposely
contaminated with cobalt linoleate and other
similar catalysts.

Also, means were investigated

for eliminating such catalysts from oils in which
they had been incorporated.

Data from these

experiments are presented in table 17.

TABLE 17
Effect of Thioglycolic Acid and Ammonia on Fish
Oils Containing Metallic Oxidation Catalysts.

Nunfeer

Treatment

Induction Period,
Hours

6058
19149
19159
14458

Control Halibut Liver Oil
(6098) plus 0.1% Cobalt Linoleate
(19149) plus 1% Thioglycolic Acid
Fish Viscera Oil plus 1% Cobalt
Linoleate
14458-A (14458) Treated with Ammonia Gas
2028-11 Control Halibut Liver Oil #2
22289
(2028-11) plus 1.2% added Tuna Liver
Oil Concentrate (Containing
Trace of Copper)
25539
(22289) Plus 1% Thioglycolic Acid

81
24
74
1*5
11
18
1.5

19

Among a number of substances used in
attempting to stabilize oils containing metallic
oxidation catalysts, the most successful were
thioglycolic acid and ammonia, especially the former.
Halibut liver oil, 19149, had its induction period
reduced from 81 to 24 hours by the addition of 0.1%
cobalt linoleate.

Treatment with 1% thioglycolic

acid restored the induction period of the oil
nearly to its original value.

Fish viscera oil,

14458, which had had its induction period reduced
to 1.5 hours by the addition of if. cobalt linoleate,
was stabilized to the extent of an 11 hour induction
period by bubbling ammonia gas through the oil.

The second control halibut liver oil,
2028-11, was a typical alkali-digested oil*

Its

Induction period was reduced from 18 hours to 1*5
by mixing with it 1.2f. of a certain tuna liver
oil concentrate*

Upon spectrographlc examination

of this tuna liver oil concentrate, it was found
to contain a trace of copper.

This oil and con­

centrate mixture was stabilized to more than its
original induction period value by the addition of
if* thioglycolic acid*
(d)

Peroxides.

Even more important than

metallic soaps as oxidation catalysts in fish oils
are peroxides*

These are always formed rapidly at

the time the vitamin A content of an oil begins
to break down, and are even present to some extent
in comparatively fresh oils*

It is probably this

small initial amount of peroxides which initiates
and autoxidizes the vitamin A content of fish
oils when they break down.

Therefore it would b e

quite important to be able to stabilize a vitamin
A-contalning oil which already had a substantial
peroxide content*

Thioglycolic acid and amines

were the only compounds found in this study which
were effective in stabilizing fish oils containing
appreciable amounts of peroxides.

Data showing the

stabilizing effect of thioglycolic acid and ethylene
diamine are presented in table 18.

TABUS 18
Effect of Thioglycolic Acid and Ethylene Diamine
on Per oxidized Halibut Liver Oil.

Number
38990
47691
48211
50941-D
52291

Treatment

Induction Period,
Hours

Control Halibut Liver Oil
(38990) Plus Peroxides (Per.No., 18.7)
(47691) Plus 1$ Thioglycolic Acid
(38990) Plus Peroxides (Per.No., 20)
(50941-D) Plus 0.1$ Ethylene Diamine

28
13
70
13
38

The action of the thioglycolic acid in these
experiments is worthy of note.

The peroxidized oil

to which it had been added (48221) underwent a
slight, rather abrupt loss of about 10 to 12$ of
its vitamin A content during the first 20 hours*
exposure of samples in the incubator.

Following

this there was a continuous but very slow drop
with time.

Ihls drop continued as nearly a straight

line as long as the test was continued, namely
until the vitamin A level reached 43$ at 264 hours.
The nominal induction period (80$ of original
vitamin A) was 70 hours, although there was not
a true induction curve for this oil.

Aside from

the preliminary slight drop at 20 hours, the main

reaction was not In the nature of an autoxidation with its characteristic increase in
rate of oxidation with time, but rather a
simple oxidation at a constant rate*

This fact

might indicate that the thioglycolic acid was
destroying the peroxides as rapidly as they
were formed*

Probably this is not the general

mechanism, however, since peroxidized oil,

52291, containing added ethylene diamine and
with an original peroxide nunfeer of 20, still
had a peroxide number of 17 after standing a
month*
Effect of Anti oxidant s.

The relative

effectiveness of various compounds as anti­
oxidants for vitamin A was determined by in­
corporating them in halibut liver oil.

The

induction period of the oil was determined
before and after the addition of the antioxidants
so as to give an estimation of the improvement
in stability.

Some typical results of these

studies are given in table 19*

The compounds

listed in the upper part of the table were
dissolved in the respective halibut liver oils
in the concentrations shown*

The salts and acids

were powdered in a mortar, from one to three

per cent was added to the oil and then mixed well*
After several hours* standing, with occasional
shaking, the solids were filtered off and induction
tests made on the clear oils*

The acetic acid and

acetic anhydride were added directly to the oils
dispersed by shaking*

The clear oil was

decanted off for testing*
TABLE 19
Effect of Various Antioxidants on the Induction
Periods of Halibut Liver Oils*
Treatment

Induction Period, Hours
After
Before

0*09% Hydroquinone
0*1 % Hydroquinone
0*05> Hydroquinone
0*5 % Lexinol*
0.5
Lexinol
0*5 /• Lexinol
0.5 7* Phospholipid from Fish Viscera Oil
0*1 7. Ethanolamine
0.1 > Ethanolamine
0 *01% Ethylenediamine
0*1 f. Ethylenediamine
0*1 >
Tocopherol
0.5 > <y Tocopherol
1*0 % o( Tocopherol
1*0 7. Corn Oil Non-Sap. Fraction
0.1 > Condensate of NH3 and Acetone
Maleic Acid
Maleic Acid
Citric Acid
Citric Acid
Tartaric Acid
Succinic Anhydride
Fumarie Acid
Succinic Acid
Ascorbic Acid
Benzoic Acid
Acetic Acid
Acetic Anhydride
* Vegetable lecithin prepared from soybean oil.

5.5
5-5
19
5*5

12
28
12
12
81
12
12
19
19
19
19
19
5.5

12
5*5

12
5.5

12
12
12
5.5
5.5
5.5
5.5

54

120
169
50
24
47
30

60
122
75

112
32
65
104
33
53
35
36
34

26
22
20.5
19
18
8.5
4.5
4.5
3.5

These results Indicate that hydroquinone
is the most active antioxidant of those studied,
v4iile the amines are nearly as active.

Alpha

tocopherol is also a very good antioxidant for
vitamin A hut has to be used in relatively large
amounts.

Among the organic acids maleic acid is

the most effective, while citric and tartaric
acids are only a little less active.

It is

interesting that fumarlc acid, the naturally
occurring isomer of maleic acid, is considerably
less active than the latter.

Benzoic and acetic

acid, and acetic anhydride had a detrimental
effect on the stability of halibut liver oil.
In the earlier experiments with organic
acids as antioxidants the oils were triturated
with rather large amounts of the powdered acids—
usually about three per cent.

Since it was

obvious that very little if any of these dry
acids actually dissolved in the oil, it seemed
desirable to find a lower limit of concentration
of acid which would still be active as an anti­
oxidant.

Powdered citric acid was added to 10 gm.

lots of halibut liver oil in amounts equivalent to
1, 0*1, and 0.01^.

These were shaken occasionally

during a period of several hours and then filtered.

Even in the case of 0*0lf. acid there seemed to
be practically as much remaining as was
originally added*

The results of induction

tests on these oils are shown in table 20*

They

show that the stabilizing effect is accomplished
with a very low concentration of added acid,
and that larger amounts are not much more
effective*
TABLE 20
Effect of Various Amounts of Citric Acid on
The Induction Period of Halibut Liver
Oil.

Induction Period,
Hours______
Control Halibut Liver Oil
H.L.0* plus1 Citric Acid
H*L*0* plus0*1^ Citric Acid
H.L.0* plusO.Olf. Citric Acid

Synergism of Anti oxidant s.

5,5
26
26
21

A syner­

gistic action of hydroquinone and vegetable
lecithin as antioxidants has been reported by
Holmes, Corbet and Hartzler

[JO),

A study was

made here of the synergistic relations of
hydroquinone, lexinol (vegetable lecithin)
and citric acid.
21.

The data are given In table

They show that the combination of

hydroquinone and citric acid have almost as
great a synergistic effect as does the combination
of hydroquinone and lexinol— that is, when paired
together a much greater protective effect results
than would be expected from their additive effects.
Also, viien a combination of hydroquinone, citric
acid and lexinol is used an even greater effect
is shown.

On the other hand, a combination of

lexinol and citric acid shows no synergistic
effect.
TABLE 21
Synergism of Antioxidants.

Treatment

Induction Period,
Hours

Control Halibut Liver Oil
H.L.0. plus Citric Acid
H.L.0. plus 0.05% Hydroquinone
H.L.0. plus 0.05% Hydroquinone plus Citric
Acid
H.L.0. plus 0.5% Lexinol
H.L.0. plus 0.5/* Lexinol plus Citric Acid
H.L.0. plus 0.5> Lexinol plus 0.05f.
Hydroquinone
H.L.0. plus 0.5% Lexinol plus 0.05%
Hydroquinone plus Citric Acid

5
34
34
275
30
36
340
400

Studies on the Nature of Natural
Antioxidants in Fish Liver.
Oils

During the course of studies of the
stability of various fish oils it beoame apparent
that different chemical treatments of the oils
greatly affected their stability.

Therefore

experiments were set up to investigate the chemical
nature of the natural anti oxidants present in
these oils by applying various chemical treatments
and determining the fate of the antioxidants by
means of the induction test*

In most cases the

oils studied were solvent-extracted halibut liver
oils*
Effect of Acids*

In table 11 were shewn

data which indicate the effect of several common
acids on the stability of vitamin A in a halibut
liver oil*

Acetic acid had no effect,while

phosphoric and hydrochloric acids were detrimental—
especially the latter*

On the other hand, sulfuric

acid enhanced the stability of the oil*
These results were obtained when the
halibut liver oil, which was dissolved in
petroleum ether, was washed with 80% methyl alcohol
containing if. of the acid in question.

The effect

of aqueous sulfuric acid was next Investigated,
as was also the effect of the acid salt, sodium
acid sulfate.

The results of these experiments

are summarized in table 22.

They show that in

all cases where sulfuric acid comes in contact
with halibut liver oil the induction period is
increased*

This effect is only slight in the

case of aqueous sulfuric acid, probably due
(as later experiments show) to the solvent effect
of water for the antioxidant fraction of the oil.

TABLE 22

Effect of Sulfuric Acid and Sodium Acid
Sulfate on the Induction Period of
Halibut Liver Oil.

Nunfcer

6058
19899
24129
25559

58990
40930

Treatment

Induction Period,
Hours

Control Halibut Liver Oil
(6058) Washed with 8of. Methyl Alcohol
Same as 19899 plus if* Sulfuric Acid
{6058) Washed with Aqueous Sulfuric Acid
Control Halibut Liver Oil $2
(38990) Treated with Sodium Acid Sulfate

Effect of Alkalies,

81
50
126
95
27.5
40

(a) Aqueous Alkali.

As the data of tables 6 and 7 show, induction
periods of alkali-digested fish oils are much

lower than those of solvent-extracted oils.
Since alkali-digested oils are treated during
preparation with large volumes of aqueous
alkali it was thought this might be a factor in
their reduced stability*

Therefore the effect

of aqueous alkali on a solvent-extracted halibut
liver oil was investigated.
in table 23*

The data are shown

Ten gram amounts of halibut

liver oil, 6058, were dissolved in minimum
amounts of a mixture of equal parts ethyl al­
cohol and ether and to these solutions were
added aqueous sodium hydroxide equivalent to
half, equal, and double the amount of free
fatty acid (12*8 per cent estimated as oleic)
present in the oil*

These mixtures were

warmed and shaken for about an hour, whan the
ether layer was separated and the ether re­
moved from the oil by distillation*

The same

treatment was followed for 17769* the last
sample in the table, except that the
phospholipids had been removed from the oil,
prior to the alkali treatment, by precipitation
with cold acetone.

The aqueous ammonia treat­

ment was carried out in the same manner as the
treatment with aqueous sodium hydroxide*

TABLE 23

Effect of Aqueous Alkali on the Induction
Period of Halibut Liver Oil.

Number

6058
18399
18179
18409
18839
17709
17769

Treatment

Induction Period,
Hours

Control Halibut Liver Oil
(6058) plus Half Theoretical Amount NaOH
(6058) plus Theoretical Amount NaOH
(6058) plus Twice Theoretical Amount
NaOH
(6058) plus Aqueous NH4OH
(6058) After Removing Phospholipids
(17709) plus Theoretical Amount NaOH

81
23
25
17
33

78
4

It is evident from these results that there
is little correlation between the amount of alkali
used and the amount of reduction of the induction
period.

However, when the phospholipid fraction is

removed from the oil the alkali has a much more
severe effect on the antioxidant.

Treatment with

aqueous ammonia, representing an amount of ammonia
theoretically required to neutralize the free fatty
acids of the oil, had a slightly less detrimental
effect than treatment with sodium hydroxide.
(b)

Dry Alkali.

The effect of dry alkali

on the stability of fidi oils was determined by
treating halibut liver oil with ammonia gas and
sodium ethylate respectively.

In the case of the

former, an excess of ammonia gas was passed
through 25 gm. halibut liver oil, 6058*
Three volumes acetone were then added, which
brought down a voluminous precipitate*

This

precipitate was filtered off and the acetone
removed from the filtrate by distillation.

There

was no reduction in free fatty acids— the ammonia
therefore had not neutralized any of them.
The treatment with sodium ethylate was
as follows:

To 12 gm. halibut liver oil, 6058,

was added 37 ml. 0.79N sodium ethylate (about
half the theoretical amount necessary to com­
pletely saponify the oil), causing an immediate
precipitation of soap.

After standing overnight,

the soap was extracted with acetone, filtered
and the acetone distilled under vacuum.

More

soap separated so the acetone extraction was
repeated.

The resulting limpid oil had an E^*
1 cm.

value of IO8.5 , compared to 72 for the original
oil.

The data from stability tests on these dry

alkali-treated oils are presented in table 24.

former, an excess of ammonia gas was passed
through 25 gm. halibut liver oil, 6058*
Three volumes acetone were then added, which
brought down a voluminous precipitate*

This

precipitate was filtered off and the acetone
removed from the filtrate by distillation.

There

was no reduction in free fatty acids— the ammonia
therefore had not neutralized any of them*
The treatment with sodium ethylate was
as follows:

To 12 gm* halibut liver oil, 6058,

was added 37 ml. 0 *79N sodium ethylate (about
half the theoretical amount necessary to com­
pletely saponify the oil), causing an immediate
precipitation of soap*

After standing overnight,

the soap was extracted with acetone, filtered
and the acetone distilled under vacuum.

More

soap separated so the acetone extraction was
repeated.

The resulting limpid oil had an

1 cm.

value of 108*3 , compared to 72 for the original
oil.

The data from stability tests on these dry

alkali-treated oils are presented in table 24.

TABLE 24

Effect of Dry Alkali on the Induction Period
of Halibut Liver Oil.

Nunfcer

Treatment

6058

Control Halibut Liver Oil
(6058) plus Ammonia Gas
{6038) plus Sodium Ethylate

18309
24169

Induction Period,
Hours
81
81+
8l+

These results indicate that treatment with
dry alkali has no destructive effect on the natural
antioxidants in halibut liver oil.

Apparently

losses of antioxidant content from treatment with
aqueous alkali has been due to the water content
of the alkali solutions.
(c) Alcoholic Alkali.

The first treatment

of halibut liver oil with alcoholic alkali was made
by shaking a solution of the oil in petroleum ether
in a separatory funnel with 80% methyl alcohol to
which potassium hydroxide had been added.

This

combination of petroleum ether and 8of. methyl
alcohol was used since the two phases could be in­
timately mixed by shaking but would separate at once
on standing.

A control treatment of the petroleum

ether solution of the oil with 8of. methyl alcohol

had to be made, since this alone had some effect*
The data, shown in table 25* indicate practically
no effect of alcoholic potassium hydroxide*
Treatment of halibut liver oil with al­
coholic ammonia was as follows*

A 2*86 N solution

of ammonia was prepared by bubbling the gas through
absolute ethyl alcohol*

Three and a half ml. of

this solution were added to 10 gm* halibut liver
oil, 6058, shaken, and allowed to stand overnight
at room temperature*

The oil was then taken up

in ten volumes of acetone and the precipitate that
formed filtered off*

The solvents were then dis­

tilled off and a stability test made on the
resulting oil*

The results were so outstanding

that a further similar treatment was given an
alkali-digested halibut liver oil, both with
and without the supplementary acetone-treatment
given the ether-extracted halibut liver oil,

6058, used above*
It i s evident from the data that treat­
ment of halibut liver oil with alcoholic
ammonia greatly enhances its stability.

This is

especially true in the case of an oil prepared by
solvent extraction such as 6058*

The effect is

even more pronounced when acetone is added to

the ammonia-treated, oil*

TABLE 25
Effect of Alcoholic Alkali on the Induction
Period of Halibut Liver Oil*

Number

Treatment

Induction Period,
Hours

6058

Control Halibut Liver Oil
(6058) Washed with 80^ Methyl Alcohol
28179
(6058) Washed with 8o> Methyl Alcohol
with Added KGH
18779
(6058) plus Alcoholic Ammonia with
Acetone
2028-11 Control Halibut Liver Oil #2
23489
(2028-11) plus Alcoholic Ammonia Alone
23499
(2028-11) plus Alcoholic Ammonia with
Acetone
19899

Effect of Water.

8l
50
53
550
18
48
73

The above experiments

indicated that it was the water content of aqueous
alkali which had the solvent effect on the antioxidants of halibut liver oil.

It is difficult

to extract oil with pure water, due to emulsions
which form.

However, it was possible to extract

halibut liver oil, 6058, with a hot solution of

33% ethyl alcohol and get a separation of the
aqueous and oily layers.

The oil was then dis­

solved in ether, dried over sodium sulfate and
the ether distilled.
A second experiment dealing with water

solubility of the antioxidant was also carried
out.

Ton grams of halibut liver oil, 6058,

were dissolved in 40 ml. petroleum ether and
this solution was extracted twice with 40 ml*
amounts of a 10% aqueous solution of sodium
sulfate*

The sodium sulfate has been shown to

have no effect on the antioxidant and was simply
added to prevent emulsions*

The petroleum ether

was then distilled from the oil.

Induction tests

were carried out on the resulting oils from
these two experiments*

The data are presented

in table 26*

table: 26
The Solvent Effect of Water on the Antioxidant
of Halibut Liver Oil*

Number
______

6058
19139
31709

Treatment

Induction Period,
Hours

Control Halibut Liver Oil
(6058) Washed with Aqueous Alcohol
(6058) Washed with 10% Sodium Sulfate

81
40
16

This data indicates that the antioxidant
is more soluble in water then in alcohol since
washing the oil with a water solution contemning
sodium sulfate reduces its stability more than
washing it with 33% alcohol.

Effect of 80 Per Cent Methyl Aloohol.
Many of the chemical treatments given the fish
oils whose antiaxidant content was being studied
were carried out by washing the oil first dis­
solved in petroleum ether, with an 8of* solution
of methyl alcohol to which the particular chemical
was added*

This necessitated knowing the solvent

effect of the methyl alcohol alone for the anti­
oxidant.

The solvent effect was detexmlned by

dissolving 10 gins, of the oil in 40 ml. petroleum
ether; this solution being given two thorough
extractions in a separatory funnel with 40 ml.
portions of 80^ methyl alcohol.

The petroleum

ether layer was than carefully separated and
the solvent evaporated.
To find the effect of additional
extraction, another experiment was carried out
the same as above except the oil was extracted
with aloohol a total of five times.

These two

alcohol-extracted oils were then put on induction
tests.

The data are shown in table 27.

They

show that although a portion of the antioxidant
is removed by extraction with 8of. methyl alcohol,
no significantly greater amount is removed by
continued extraction.

TABLE 27

The Solvent Effect of 8 of. Methyl Alcohol on
The Antioxidant of Halibut Liver Oil.

Nuniber

6O58
19899
20409

Treatment

Induetion Period,
Hours

Control Halibut Liver Oil
(6058) After Two Washes with 8of.
Methyl Alcohol
(6058) After Five Washes with 80%
Methyl Alcohol

Effect of Various Chemicals*
benzoyl Chloride.

8l
50
46

(a) Dinitro-

Halibut liver oil was treated

with various chemicals which react characteristically
with certain chemical groupings.

This was an attempt

to establish which chemical groupings are important
in the action of the natural antioxidants of these
oils.

First, 1 gm. finely powdered 5,5-dinitrobenzoyl

chloride was added to 10 gm, halibut liver oil, 6058.
This mixture stood overnight, was than shaken
occasionally for 2 hours and filtered by suction.
(b)

Formaldehyde.

Ten grams halibut liver

oil, 6058 , were dissolved in 40 ml. petroleum ether
and washed twice with 40 ml. portions of a solution
composed of 80 ml. methyl alcohol and 20 ml. 40f.
formaldehyde.

The petroleum ether layer was

carefully separated and the solvent distilled from it.

(c)

Phthallc Aiihydride.

Ten grams

halibut liver oil, 6058, were dissolved in 40 ml*
of a saturated solution of phthalic anhydride in
petroleum ether (very slightly soluble) and al­
lowed to stand in this solution for half an hour
at room temperature*

The petroleum ether was

then removed by distillation*
A second experiment was carried out in
which 5 gm* halibut liver oil, 6058, were dis­
solved in 45 ml* ethyl ether vhich contained 50
mg* phthalic anhydride*

This solution stood in

the dark at room temperature for 23 hours, when
the ether was removed by distillation.
Induction tests were carried out on all
the chemically-treated oils described under (a) to
(c).

The data from these tests are presented in

table 28*

TABLE 28
The Effect of Various Chemicals on the Anti­
oxidant of Halibut Liver Oil.
Nunber
6058
28269
19899
27979
27969
28199

Treatment

Induction Period,
Hours

Control Halibut Liver Oil
(6058 ) Plus Dinitrobenzoyl Chloride
(6058 ) Washed with 8of. Methyl Alcohol
(6058 ) Washed with 8O7. Methyl Alcohol
Containing 8f. Formaldehyde
(6058 ) Plus Phthalic Anhydride in
Petroleum Ether
(6058 ) Plus Phthalic Anhydride in
Ethyl Ether

81

11
50
29
22
6

The action of dlnltrobanzoyl chloride
on the activity of the antioxidant would tend
to indicate the activity of hydroxyl groups in
the latter*

However, the possible breakdown

of benzoyl ohloride to form hydrochloric acid
oust not be overlooked*

It has been shown

that only a very small amount of hydrochloric
acid is necessary to effectively reduce the
stability of halibut liver oil*
Formaldehyde appears to be moderately
reactive toward the antioxidant under the con­
ditions employed*

This would indicate that

the antioxidant may be a phenolic type compound*
This reasoning is also in harmony with the
observed action of phthalic anhydride on the
antioxidant activity of oils*

This compound,

which the data shows to be so readily des­
tructive of the antioxidant in halibut liver
oil, characteristically reacts with phenols and
amines*
Relation of Chemical Treatment to Added
Antioxidants.

Various types of anti oxidants

were added to a given halibut liver oil and the
effects of different chemical treatments on these
added antioxidants were investigated.

This was

to gain some information concerning the nature of
the natural antioxidants in halibut liver oils by
noting the effect of known chemical treatments on
the added antioxidants.
To 30 gm. amounts of halibut liver oil were
added 0 .1^ ethanolamine, l.of. o( tocopherol, and
0.05^ hydroquinone respectively.

First, the induction

periods of these three samples were determined to
serve as a baseline, then part of each sample was
treated as follows:
(a) Extracted with 8o£ methyl alcohol.
Ten grams of the oil were dissolved in
40 ml. petroleum ether and washed twice
with 40 ml. portions of the 8of. methyl
alcohol, after which the petroleum ether
was evaporated from the oil layer.
(b) Extracted with 80^ methyl alcohol coin
taining hydrochlorio acid.
Same procedure as (a) except the 8of.
methyl alcohol contained an added 1%
hydrochloric acid.
(o) Treated with phthalic anhydride.
Five grams were dissolved in 45 ml.
ethyl ether in which had been dissolved
50 mg. phthalic anhydride. After
standing overnight the ether was
evaporated. (The sample containing
ethanolamine gave a fine white pre­
cipitate when the ether solution of
phthalic anhydride was added).
The induction periods of all these samples were
determined.

The data are presented in table 29.

TABLE 29

Effect of Chemical Reagents on Different
Types of Antioxidants Added to Halibut
Liver Oil.

Nunfcer

I

Induction Period,
Hours

2028 II

Control Halibut Liver Oil

19

28379
28819
28409

0.1^ Ethanolamine
(28379) Plus 8of. Methyl Alcohol
(28379) plus 8 0 Methyl Alcohol
with HOI
(28379) plus Phthalic Anhydride

56

105

28419

l.of.
Tocopherol
(28389) plus 8of. Methyl Alcohol
(28389 ) plus 8 of» Methyl Alcohol

28539

(28389 ) plus Phthalic Anhydride

105

Hydroquinone
(28399) plus 8 0 Methyl Alcohol
(28399) plus 8ofo Methyl Alcohol
with HC1
(28399) plus Phthalic Anhydride

169

28529
II

Treatment

28389
28829

17

15i
57
90
110

with HC1
III 28399

28839
28429
28549

0,03$

11

170

These results indicate that the induction
periods of halibut liver oils stabilized by the
addition of ethanolamine or hydroquinone were greatly
reduced by extraction with 8o£ methyl alcohol, either
alone or with added hydrochloric acid.

However, as

might be expected, this treatment has no effect on an
oil stabilized by the addition of alpha tocopherol.
Phthalic anhydride had no effect on oils stabilized

by any of the above treatments, although it had
been found very destructive of the natural
antioxidant of halibut liver oil*

Attempts at Isolation of Active Antioxidant Fractions of Fish Oils

Experiments were carried out whereby
various fractions of stable fish oils were
separated by appropriate treatments.

The object

was to find in tfdich of these fractions the
stabilizing factor, or antioxidant, was to be found.
This was studied by two methods.

The first was to

add these separated fractions to another fieh oil
of known degree of stability and determine whether
it was made more stable by the addition of these
fractions.

The second method was to determine the

stability of an oil after a certain fraction had
been removed, then find if the original stability
of the oil could be attained by adding the isolated
fraction back to the residue from which it had
been separated.
Cold Acetone Precipitation.

One method

of fractionation employed was the separation of
an acetone-insoluble fraction.

Typical of several

such experiments is the following*
Thirty grams of a solvent-extracted ling
cod viscera oil were dissolved in 30 ml. ethyl
ether and poured slowly into a liter of ice-cold
acetone with continuous stirring.

A small amount

of flocculent precipitate which separated was filw
tered off quickly on a Buchner funnel.

The yield

after drying in a vacuum was 0.54 gm., or 1.8%.
The acetone was removed from the filtrate by
distillation.

Fart of the isolated acetone-insoluble

fraction was redissolved at a concentration of 1.8%
in some of the residual acetone-soluble oil.

The

Induction periods of the original ling cod viscera
oil, of the residual oil after removal of the
acetone-insoluble fraction, and of the reconstituted
oil were determined.

The data are in table 30, and

are presented in graphic form in Figure 5.

TABLE 30

Effect of Cold Acetone Precipitation on the
Antioxidant of Fish Viscera Oil.
Number
0757
2268
2448

Treatment
Ling Cod Viscera Oil
Acetone-soluble Fraction
(2268) plus Phospholipid Fraction

Induction Period,
Hours
100
36
80

f* 7 f/v

Lii
FH:

t

m
Hi

The Induction period data, and especially
the curves of Figure 5, illustrate what a large
proportion of the stabilizing fraction of this oil
was present in the acetone-insoluble, or phospholipid
fraction*

There is no doubt that this acetone-

insoluble fraction was a phospholipid since a
phosphorus determination on fractions of a similar
oil, prepared in an identical manner, indicated

1*58% phosphorus in the acetone-insoluble fraction,
and only 0*012% in the acetone-soluble oil*
However, this type of antioxidant is not found in
fish oils prepared by alkali-dlgestion since these
are entirely soluble in acetone*
Extraction with an Aqueous-Alcoholic
Solution of Alkali.

Ten grams of a solvent-

extracted halibut liver oil, which had been freed
from its phospholipid content by precipitation with
cold acetone, was treated with the theoretical
amount of aqueous N /10 sodium hydroxide to neutralize
the free fatty acids present*

After warming on the

steam bath the resulting neutral oil separated as
an upper layer*

The aqueous layer was drawn off from

below in a separatory funnel and washed with ether
to remove all traces of the neutral oil.

It was then

acidified to pH 5*0 with acetic acid and the fatty

The induction period data, and especially
the curves of Figure 5, Illustrate what a large
proportion of the stabilizing fraction of this oil
was present in the acetone-insoluble, or phospholipid
fraction.

There is no doubt that this acetone-

insoluble fraction was a phospholipid since a
phosphorus determination on fractions of a similar
oil, prepared in an identical manner, indicated
1.58% phosphorus in the acetone-insoluble fraction,
and only 0.012% in the acetone-soluble oil.
However, this type of antioxidant is not found in
fish oils prepared by alkali-digestion since these
are entirely soluble in acetone*
Extraction with an Aqueous-Alcoholic
Solution of Alkali.

Ten grams of a solvent-

extracted halibut liver oil, which had been freed
from its phospholipid content by precipitation with
cold acetone, was t reated with the theoretical
amount of aqueous N /10 sodium hydroxide to neutralize
the free fatty acids present.

After wanning on the

steam bath the resulting neutral oil separated as
an upper layer.

The aqueous layer was drawn off from

below in a separatory funnel and washed with ether
to remove all traces of the neutral oil.

It was then

acidified to pH 5*0 with acetic acid and the fatty

acids thereby released were extracted with ether*
This ether extract was dried over anhydrous
sodium sulfate and the ether removed by dis­
tillation*

The yield was 1*05 gm. of fatty

acids which were partially solid at room
temperature*
To 4*5 gms* of halibut liver oil prepared
by the method cf alkali-digestion was added 0*502
gm* of the fatty acid fraction described above*
The data from induotion tests on this composite
oil and the fractions mentioned above are presented
in table 51*

TABLE 31
Stabilizing Effect of a Fraction of Halibut
Liver Oil Extracted by Aqueous Alkali*

Number
17709

18369
2028

18389

Treatment
Control H*L*0* (Solvent-extracted)
(17709) Extracted with Aqueous NaOH
Control H*L*0» (Alkali-digested)
(2028) plus Fraction from 17709

Induction Period,
Hours
78
4
12
l6£

It is evident that dilute aqueous alkali
either destroys or extracts the antioxidant fraction
of halibut liver oil very completely*

The fact that

the aqueous alkali extract contained a sub­
stance rtiich when acidified possessed some
anticxidant effect Indicates that it was partly
a matter of extraction.
Another experiment was carried out in
which 10 gm. of the same halibut liver oil as
above were dissolved in 20 ml. of an equal
mixture of alcohol and ether.

To this were

added 50 ml. of N/10 aqueous sodium hydroxide
and the mixture shaken occasionally for a half
hour.

Then 20 ml. additional ether were added

to separate the two layers.

To the water layer

were added 60 ml. N/5 sulfuric acid and the
solution was extracted with ether.

This ether

extract was washed once with water and dried
over anhydrous sodium sulfate.

The ether was

evaporated, yielding 1.62 gm. of a fatty acid
fraction.

The ether was evaporated from the

solution of neutral oil, yielding 8.05 gras* of
oil.

Part of this isolated fatty acid fraction

was re-confcined with some of the neutral oil
in the proportions in which the two fractions
were originally present.
were run on these samples.
shown in table J2.

Induction tests
The results are

TABLE 32

Partial Recovery of the Antioxidant Fraction
of Halibut Liver Oil after Extraction
with Dilute Alkali.

Nunfcer
17709
27499
27539

Treatment

Induction Period,
Hours

78

Control H.L.O. (Solvent-extracted)
(17709) Less Fatty Acid Fraction
(27499) Plus Fatty Acid Fraction

6
14

These data show that part of the antioxidant removed from halibut liver oil by
aqueous alkali extraction can be restored to
the residual oil by returning the fatty acid
portion removed by the alkali extraction.
Partition Between 80 per cent Methyl
Alcohol and Petroleum Ether.

Ten grams solvent-

extracted halibut liver oil, 6058, were dissolved
in 40 ml. petroleum ether and thoroughly extracted
with two 40 ml. portions of 8of. methyl alcohol.
To the combined methyl alcohol extracts were
added 75 ml. ether, and then water until a separa­
tion of layers took place.

The aqueous layer

was extracted twice more with ether.

The ether

was distilled off from the combined ether extracts,

yielding 0.36 gm. of an oily extract.

Half

of this oily residue was re-combined with half
of the petroleum ether-soluble fraction, after
the petroleum ether had been evaporated.

The

induction period of the alcohol-extracted oil
was determined before and after the addition of
the fraction soluble in 8 of. methyl alcohol.

The

data are shown in table 33*

TABLE 33
Separation of Antioxidant from Halibut Liver
Oil by Methyl Alcohol Extraction.

Number

6058
19899
20179

Treatment

Induction Period,
Hours

Control Halibut Liver Oil
(6058 ) Extracted with 8of. Methyl Alcohol
(19899 ) Plus Ether-Soluble Portion of
Methyl Alcohol Extract

8l
50
64

The data from this experiment show that
at least part of the antioxidant removed from
halibut liver oil by extraction with 80^ methyl
alcohol is soluble in ether and can be separated
in a definite fraction.

Restoring this to the

fraction from which it came restores a share of
the original stability.

Extraction with Aqueous Sodium Sulfate.
Past experiments have indicated that the anti­
oxidant fraction of halibut liver oil is more
soluble in water than in alcohol.

Therefore an

experiment was carried out in which an extractant
of as nearly pure water as possible was used*
Ten grams halibut liver oil, 6058, were dissolved
in 4-0 ml. petroleum ether and extracted thorough­
ly with two 40 ml. portions of a 10% solution of
sodium salfate (pure water caused an unbreakable
emulsion).

The solvent was evaporated from the

petroleum ether layer, yielding 8*8 gm. residual
oil.

The aqueous sodium sulfate solution was

extracted three times with ether.

This ether

extract was dried and the ether evaporated.
resulted 0*69 gm* of a sharp smelling oil.

There
To

0.35 gni« of this were added 4*65 gm* of the
residual oil.

Table 34 gives data on the induction

tests of the extracted and reconstituted oils*

TABLE 34
Attempted Separation of the Antioxidant of
Halibut Liver Oil by Extraction with
Aqueous Sodium Sulfate Solution*
Nunfcer
6058

52669
32679

Treatment

Induction Period,
Hours

Control Halibut Liver Oil
{6058) After Washing with Na 2S04
(32669) Plus Na 2S04 Extract

81
14
16

These data indicate that the ant iocsidant
fraction is much more completely soluble in 10%
aqueous sodium sulfate than in 80% methyl alcohol.
However, none of the antioxidant removed by the
sodium sulfate solution could be recovered from
it by extraction with ether.

This indicates

either that it vias destroyed by the sodium sulfate
solution or rendered ether-insoluble by it.

RESUME

Development of Induction Teat.

A

method has been developed for studying the
stability of oils containing vitamin A*

This

consisted of exposing small, uniform samples to
the air at a temperature of 37° C* under
standardized conditions*

The vitamin A-breakdown

in these samples was measured either by frequent
vitamin A determinations using the Vitameter,
or by peroxide determinations*
Studies were made of the effect of
temperature, area, and depth of exposed samples
on this induction test*

It was found that the

induction period varied inversely with the area
of the sample exposed, and a decrease in tempera­
ture of 10° approximately doubled the length of
the induction period*

The depth of sample was

relatively unimportant, although the weight of
oil exposed largely determined the rate of
vitamin A-breakdown*

Agitation of samples

proved unnecessary, since almost identical
results were obtained with small, unagitated
samples as with larger samples rtiich were con­
stantly agitated during exposure•

When titrations of peroxides as well as
vitamin A determinations were made on samples of
oils exposed in the induction tests, it was found
that as vitamin A began to break down peroxides
always increased at the same time.

Thus either

determination was satisfactory as an index in the
test, but vitamin A determinations usually proved
easier*
A critical examination of the breakdown of
vitamin A during the induction test was also made
by means of spectrophotometric measurements of
exposed oil samples*

A complete absorption curve

in the ultra-violet was made for each sample of
a particular halibut liver oil as it was removed
from the constant temperature room during the in­
duction test*

This study indicated a drop in the

peak of absorption with time, following the end
of the induction period, and eventually a shift
of the peak toward the lcwer ultra-violet.
An attempt was made to correlate results
obtained by this accelerated induction test and
actual breakdown of vitamin A under certain
conditions of storage*

These correlation studies

showed that a factor could be used to translate
the induction period obtained by the above method

into time required for the beginning of vitamin
A-breakdown in storage.

The studies also

emphasized the importance of well-filled con­
tainers for storage, and the fact that breakdown
time is largely a matter of the bulk of oil
being considered.
Stability of Natural Vitamin A Oils.
This standardized induction test was used to
investigate the stability of a number of vitamin A
oils of various sources and methods of preparation.
It was found that cod liver oils and solventextracted oils were most stable, having average
induction periods of 125 hours and H I hours
respectively.

Tuna liver oils were intermediate

in stability, with an average of 65 hours.

Oils

prepared by aqueous alkali digestion were much
less stable as a class.

These ranged from 21

hours and 17 hours, respectively, for averages of
mixed fish liver oils and halibut liver oils, to
8 hours for alkali-digested viscera oils*
Conditions Affecting the Stability of
Vitamin A in Oils.

A study was made of the

conditions affecting the stability of vitamin A
in oils.

There was found to be no correlation

between the free fatty acid content of a group

of oils and their inherent stability.

Peroxides,

however, had an important effect on the induction
periods of oils.

With varying dilutions of

vitamin A in a given vegetable oil, induction
periods were in a direct ratio to the dilution
factor.

But regardless of the length of the in­

duction period vitamin A dropped and peroxides
rose at the same time.

Addition of peroxides

alone to the vegetable oil used for a diluent
caused no such accelerated induction period.
Therefore it was concluded that vitamin A is of
the nature of an oxidation catalyst and stimulates
autoxidation of oils in which it is dissolved.
The effect of addii^ various oxidation
catalysts to vitamin A oils was studied.

Among

these were salts, acids, metallic soaps and
peroxides.

There was a great range of effect

among the salts, being greatest with those which
are known to be strong oxidation catalysts, namely
salts of cobalt and copper.

However, it was found

that calcium and magnesium chlorides were just as
active, and a number of other chlorides were also
very active.
Among the acids, none was very active as
an oxidation catalyst except hydrochloric, which

was extremely active*

It was found that this

proxidant effect of hydrochloric acid and chloride
salts could be counteracted by the action of
strong alkalis*

This made it seem that the

mechanism of action of the chloride salts might
be their reaction with the free fatty acids of
the oils in which the salts were plaoed, liber­
ating free hydrochloric acid*
Metallic soaps were found to be extremely
active oxidation catalysts, but their effect
could be neutralized by thioglyeolic acid or
ammonia.

The strong proxidant effect of added

peroxides could be neutralized by only two
methods— the addition of thioglyeolic acid or amines.
Use of Induction Test for Evaluating
Antioxidants*

One of the greatest uses of the

induction test was in evaluating the effect of the
addition of various inhibitors of autoxidation,
or antioxidants, to vitamin A oils.

A great

number of these was studied, among them being
hydroquinone, phospholipids, organic acids,
amines, tocopherols, various complexes, and ammonia*
Of these, hydroquinone was found to be the most
active when acting alone.

However, certain

combinations of two or more acting together showed
outstanding synergistic effects.

Studies of the Nature of Natural Antioxidants in Fish Oils,

Numerous studies were

carried out to determine the nature of the natural
antioxidants present in fish oils.

These were made

by subjecting the oils to various chemical treat­
ments and evaluating the effects on the antioxidants
in the oils by means of induction tests.

Aqueous

alcohol and aqueous alkalis extracted the anti­
oxidants from oils.
pronounced effect.

Water had an even more
On the other hand, dry alkalis

or nearly anhydrous alcohol had no effect.

Dinitro-

benzoyl chloride, formaldehyde and phthalic anhy­
dride had a pronounced destructive effect on the
antioxidants, indicating perhaps their phenolic
or amine nature.
Attempts at Isolation of Antioxidant
Fractions.

Several attempts were made at iso­

lation of active antioxidant fractions from fish
oils.

The most successful was the separation of

a phospholipid fraction by cold acetone precipita­
tion.

The fraction separated in this way restored

almost completely the stability of the residual
oil from vixich it was separated.
Extraction of oils with aqueous alkali or

80% methyl alcohol removed a good share of the

antioxidant fraction but evidently a part was
destroyed in the process, since the original
stability of the oils were not restored by
returning these fractions*
Extraction of oils with water, to which
sodium sulfate was added to avoid emulsions, was
even more destructive of the antioxidant since
restoring the extract to the water-extracted
residual oil did not increase its stability
significantly*

CONCLUSIONS

Studies leading to the development of
an induction test for vitamin A oils demonstrated
that oxygen diffuses through oils quickly and to
a considerable depth.

Thus it is unnecessary to

agitate the samples of oil, as many workers have
done, during exposure to the air.

The measure­

ment of either vitamin A or peroxides is a
satisfactory method for following the course of
the induction test*
As a general rule, fish oils prepared by
solvent extraction, derived either from livers
or viscera, are much more stable than those
prepared by alkali-digestion*
reasons.

This is for two

First, because solvent-extraction

carries the phospholipid content of the livers
into the oil, thus tending greatly to increase
its stability.

Second, large quantities of water

are used in the preparation of alkali-digested
oils, thus tending to extract the antioxidants
from the oils.

This lowered stability of alkali-

digested oils can be remedied by inclusion of
antioxidants, and by improvements in methods of
digestion.

Among the components of fish oils which
may affect their stability adversely, oxidation
catalysts such as peroxides and metallic soaps
are the most important.

Peroxides develop in the

oils during processing.

Metallic soaps may also

be introduced during processing by the interaction
of free fatty acids in the oils with metallic
containers.

Certain acids— especially hydro­

chloric acid— have been found to cause rapid
breakdown of vitamin A oils.

Calcium chloride,

sometimes used as a dehydrating agent during the
processing of fish oils, is a very active oxidation
catalyst.

The data indicate that in all cases

where chloride salts cause a rapid production of
rancidity in oils, the probable cause is the
development of free hydrochloric acid by the
reaction of the chlorides with the free fatty
acids in the oils.
Vitamin A itself may act as an oxidation
catalyst.

This is concluded from the fact that

when a very pure vitamin A distillate, free of
peroxides, is diluted in corn oil the rate of
breakdown is always proportional to the concentra­
tion of vitamin A.

On the other hand, adding

only a peroxidized oil to the same corn oil
causes no such Increase in rate of breakdown
with increasing concentration.
Many types of compounds are active anti­
oxidants for fresh fish oils, which are relatively
free from peroxides and other oxidation catalysts.
Among these, hydroquinone is the most active,
while amines are almost as effective.

Toco-

pherols and various phospholipids are quite
effective.

Certain organic acids, especially

citric, malaic and tartaric, are active anti­
oxidants in fish oils, although they have been
stated to be inactive in animal oils when used
alone, their only effect being to enhance the
effect of tocopherols and inhibitols in such
oils.
The only antioxidants studied vrhich are
effective in oils containing peroxides are thioglycolic acid and amines.

Their action is not

due to a reduction of the peroxides in the oils,
since the peroxides are practically the same
after the addition of these agents.

Oils

containing metallic oxidation catalysts are
stabilized only by thioglyeolic acid and

ammonia) among the compounds studied.

There

Is evidence that the stabilizing effect of
these two compounds is due to the formation of
complexes with the metallic catalysts in the oils*
From studies of the nature of the natural
antioxidants present in fish oils it is concluded
that several types of compounds are involved*
With regard to oils prepared by solvent-extraction,
phospholipids are of major importance in viscera
oils but not in liver oils.

Another fraction,

active as an antioxidant, is removed from halibut
liver oil by extraction with 8of. methyl alcohol;
however, continued extraction with this solvent
does not remove all the anti oxidant.

On the

other hand, water removes practically all the
antioxidant.

Chemical reactions indicate that

both amines and phenolic compounds may be
involved*
Separation of relatively pure antioxidant fractions was only possible in the case
of phospholipids.

In other cases a good deal

of destruction accompanied the separation of the
fractions, since they could not be re-combined
with the residual oils from which they came*
This indicates the extremely labile character
of the compounds involved*

109.

BIBLIOGRAPHY

(1)

Anderegg, L. T., and Nelson,
Chem., 18» 620 (1926)*

(2 )

Banks, Adam, J. Soc. Cham. Ind., 56. 13T (1937).

(3)

Banks, Adam, and Hilditch, T. P., Ibid.,
J5L, 411T (1932).

(4 )

Branch, G. E. K., Almquist, H. J., and Goldworthy,
E. C., J. Amar. Chem. Soc., 55. 4052 (1933).

(5)

Browne, 0. A., Ind. Eng. Chem., 17. 44 (1925).

(6 )

Browne, C. A., J* Amsr. Chem. Soc,, 21, 975 (1899).

(7)

Christiansen, J. A., J. Phys. Chem., 28, 145 (1924).

(8)

Coe, M. R., and Le Clerc, J. A., Ind. Eng. Chem.,
26, 245 (1934).

(9)

Davies, W. L., J. Soc. Chem. Ind., ££, 148T (1934).

Y.

E., Ind. Eng*

(10)

Deatherage, P. E., and Mattill, H. A., Ind. Eng.
Chem., J51, 1425 (1939).

(11)

Delore, P., Bull. Soc. Chim. Biol., 11, 74 (1929).

(12)

Eibner, A., and Pallauf, F., Chem. Umschau. Fette,
Ole, Wachse, Harze, J>2, 97 (1925).

(13)

Emery, J. A., and Henley, R. R., Ind. Eng. Chem.,
]£, 937 (1922).

(14)

Evans, E. I., Ind. Eng. Chem., 27, 329 (1935).

(15)

Evans, H. M., Emerson, 0. H., and Emerson, Gladys A.,
J. Biol. Chem., 113. 319 (1936).

(16)

French, R. B., Olcott, H. S., and Mattill, H. A.,
Ind. Eng. Chem., 2£, 724 (1935).

(17)

Freyer, E., Oil and Soap, 12, 139 (1935).

110.

(18)

Fridericia, L. S., J. Biol* Cham., 62. 471 (1924).

(19)

Genthe, A., Z. Angew. Chem*, 19* 2087 (1906).

(20)

Green, T. G., and Hilditch, T. P., J. Soc. Chem.
Ind., £6, 23T (1937).

(21)

Greenbank, G. R., and Holm, G. E., Ind. Eng. Chem.,
2£, 167 (1933).

(22)

Greenbank, G. R., and Holm, G. E., Ind. Eng. Chem.,
26. 243 (1934).

(23)

Grettie, D. P., and Newton, R. C., J. Oil and
Fat Ind., 8, 291 (1931).

(24)

Hamilton, L. A., and Olcott, H. S., Oil and Soap,
l£, 127 (1936).

(25)

Hamilton, L. A., and Olcott, H. S., Ind. Eng.
Chem., 2£, 217 (1937).

(26)

Hickman, K. C. D., Ibid., 32, 1451 (1940).

(27)

Hilditch, T. P., and Sleightholme, J. J.,
J. Soc. Chem. Ind., J51, 39T (1932).

(28)

Holme, G. E., and Greenbank, G. R., Ind. Eng.
Chem., 1£, 1051 (1923).

(29)

Holme, G. E., Greenbank, G. R., and Deyser, E. F.,
Ibid., 19., 156 (1927).

(30)

Holmes, Harry N., Corbet, Ruth E., and Eartzler,
Eva R., Ibid., 28, 133 (1936).

(31)

Hunziker, 0. F., and Hosrnan, D. F., J. Diary Sci.,
1, 320 (1918).

(52)

Huston, R. C., and Lightbody, H. D., J. Biol.
Chem., 2£» 547 (1928).

(55)

Huston, R. C., Lightbody, H. D., and Ball, C. D.,
Ibid., 2 2 * 507 (1928).

(54)

Kerr, R. H., Ind. Eng* Chem., 10. 471 (1918).

Ill

(35)

Kilgore, L. B., Oil and Soap,

269 (1933).

(36)

Kilgore, L. B., Ibid., 10, 66 (1933).

(37)

King, A. 3., Roschen, H. L., and Irwin, W. 11.,
Ibid., 10, 105 (1933).

(38)

Kreis, H., Chem. Z., 26, 897 (1902).

(39)

Kuhn, R., and Meyer, K., Zeit. phyaiol. Chem.,
185. 193 (1929).

(40)

Lea, C. H., Proc. Roy. Soc., (London), 108B,
175 (1931).

(41)

Lea, C. H., Ind. Eng. Chem. Anal. Ed., 6_,
241 (1934).

(42)

Lea, C. H., J. Soc. Chem. Ind.,

(43)

Lea, C. H., Dept. Sci. Ind. Research (Brit.),
Food Invest., Special Report, 46, 119 (1938).

(44)

Lease, E. J., Lease, J. G., Wever, Janet, and
Steenboek, H., J. Nutrition, 16, 571 (1938).

(45)

Long, J. S., and Egge, W. S., Ind. Eng. Chem.,
20, 809 (1928).

(46)

Lowen, L., Anderson, L., and Harrison, R. V/.,
Ibid., 29., 151 (1937).

(47)

Marcus, J. K., J. Biol. Chem., 90, 507 (1931).

(48)

Mattill, H. A., and Crawford, B., Ind. Eng.
Chem., 22_, 341 (1930).

(49)

Mattill, H. A., J. Am. Med. Assoc., 89, 1505 (1927)

(50)

Mattill, H. A., and Crawford, B., Ind. Eng. Chem.,
96, 387 (1932).

(51)

Morgan, W. L., Ibid., 2£, 1287 (1935).

(52)

Morrell, R. S., and Marks, S., J. Soc. Chem. Ind.,
JO, 27T (1931).

(53)

Moureu, C., and Dufraisse, C., Compt. Rend.,
174. 258 (1922); Chem. Rev.,
113 (192 6).

295T (1936).

112.

Musher, S., U. S. Patent 2,026,697*
Olcott, H. S., and Mattill, H. A., J. Biol.
Chem., 25, 65 (1931).
Olcott, H. S., and Mattill, H. A., Oil and Soap,
12. 98 (1936).
Olcott, H. S., and Mattill, H. A., J. Amer.
Chem. Soc., J>£» 1*27 (1936).
Olcott, H. S., and Mattill, H. A., Ibid, £8,
2204 (1936).
Olcott, H. S., and Emerson, 0. H., Ibid.,
22, 1008 (1937).
Pool, W. 0., J. Oil and Fat Ind., £, 331 (1931).
Powick, W. C., J. Agr. Research, £6, 323 (1923).
Powick, W. C.,

Ibid., 21. 1017(1925).

Richardson, A. S., J. Oil and Fat Ind., 8, 269
(1931).
Rogers, W., and Taylor, H. S., J. Phys. Chem.,
20, 1334 (1926).
Rosenheim, 0., and Webster, T. A., Lancet, 2_»

806

(1926).

Royce, H. D., Oil and Soap, 10, 123 (1933).
Simons, E. J., Buxton, L. 0., and Colman, H.
Ind. Eng. Chem., 32. 706 (1940).

B.,

Smith, Ernest L., Biochem. J., 33. 201 (1939).
Staudinger, H., Ber., 58b , 1075 (1925).
Stevens, H. N., Ind. Eng. Chem., 24. 918 (1932).
Taffel, A., and Revis, C., J. Soc. Chem. Ind.,
50. 87T (1931).
Triebold, H. 0., Webb, R. E., and Rudy, W. J.,
Cereal Chem., 10, 263 (1933).

113.

(73)

Tschirch, A., Chem. Umsohau Fette, Ole,
Wachse, Harze, 32, 29 (1925),

(74)

U, S. Patent 2,130,322.

(75)

Wagner, A. M,, and Brief, J* C., Ind. Eng.
Chem., 22, 40 (1931).

(76)

Wagner, H., Walker, R., and Oestermann, H.,
Z. Untersuch. Nahr. u. Genussm., 25,
704 (1913).

(77)

Waterman, H. J., and van Vlodrop, C., J.
Soc. Chem. Ind., £L» 333T (1936).

(78)

Wheeler, D. H., Oil and Soap,

(79)

Whipple, D. V., Ibid., 1^, 2?1 (1936).

(80)

Wokes, F., and Willimott, S. G., Biochem.
J., 21, 419 (1927).

(1932).

ACKNOWLEDGMENTS

The writer wishes to make the following acknowledgments
To Dr. L. T. Clark, Managing Director of Research at
Parke, Davis and Company, for allowing data collected at the
Parke, Davis Laboratories to be used in this thesis.
To Dean R. C. Huston, Prof. C. A. Hoppert, and Prof.
C. D. Ball of the Chemistry Department at the Michigan State
College, for their cooperation in outlining and directing
the course of the problem.
To Dr. A. D. Emmett, Assistant Director of Research
at Parke, Davis and Company, for his kind personal inter­
est and helpful counsel.
To Beatrice, without whose patient cooperation this
thesis could not have been written.

115

BIOGRAPHY
Orson David Bird was born at Romulus, Michigan, March
18, 1905.

His elementary school education was obtained at

Hayti district school, Romulus township, and he was grad­
uated from Wayne High School in 1921.

After earning a

teacher's cretificate at Michigan State Normal College, he
entered Michigan State College in 1924-•
degree in 1926.

He received a B. S.

Prom June, 1926, to June, 1928, he held a

half-time Calumet Baking Powder Co. fellowship at the Agri­
cultural Experiment Station, East Lansing, working on the
determination of Aluminum in plants.

During this time he

did graduate work at Michigan State College, obtaining in
June, 1928, the M. S. Degree in Chemistry with a minor in
Bacteriology.

The title of his thesis was

"A Study of

Carbohydrate Derivatives."
In June, 1928, he accepted a position in the Research
Department of Parke, Davis and Co., where he is located
at the present time.
In June, 1941, he presented to the Graduate School of
Michigan State College, a thesis entitled:

"A Study of

Antioxidants with Respect to Vitamin A in Fish Oils.”
PUBLICATIONS
A Proximate Chemical Analysis of the Timothy Bacillus, by
R. D. Coghill and 0. D. Bird, J. Biol. Chem. 81.
115 (1929).

116

The Determination of Aluminum in Plants. I. A Study of the
Use of Aurintricarboxylic Acid for the Colorimetric Deter­
mination of Aluminum, by
0. B. Winter, W. R. Thrun and 0. D. Bird, J. Amer. Chem.
Soc.,
2721 (1929)
The Determination of Aluminum in Plants.
Plant Materials, by

II. Aluminum in

0. B. Winter and 0. D. Bird, J. Amer. Chem. Soc. 51.
2964 (1929).
A Study .of Halibut Liver Oil. I. With Respect to its Vit­
amin Potency, Physical Constants and Tolerance, by
A. D. Emmett, 0. D. Bird, C. Nielsen and H. J. Cannon,
Ind. Eng. Chem.,
1073 (1932)
Preparation of Seibert's Tuberculin (T. P. T.) for Diagnos­
tic Purposes, by
Lawrence T. Clark, Arthur D. Emmett and Orson D. Bird,
Amer. Rev. of Tuberculosis, 30 . 4-71 (1934-).
The Comparative Bi&logical Potency of Vitamin A as an Alcohol
and as an Ester, by
A. D. Emmett and 0. D. Bird, J. Biol. Chem., Proc. XXXI,
119 (1937).
Vitamin K Potencies of Synthetic Compounds, by
S. A. Thayer, S. B. Binkley, D. W. MacCorquodale, E. A.
Doisy, A. D. Emmett, Raymond A. Brown and Orson D. Bird,
J. Amer. Chem. Soc., 61. 2563 (1939).
Spectrophotometrie Determination of Vitamin A.
to Fish Liver Oils, by

Application

D. T. Ewing, J. M. Vandenbelt, A. D. Emmett and 0. D.
Bird, Ind. Eng. Chem. Anal. Ed., 12, 639 (1940).
Determination of Vitamin B^ (Riboflavin) . A Comparison of
Bioassay, Microbiological, and Fluorometric Methods, by
A. D. Emmett, 0. D. Bird, R. A. Brown, Gail Peacock and
J. NL Vandenbelt, Ind. Eng. Chem., Anal. Ed., 13. 219
(1941).