SANDELL,-APPLICATION OF KARL FISCHER METHOD 671

APPLICATION OF THE KARL FISCHER METHOD TO THE

DETERMINATION OF WATER I N SUGAR CONFECTIONERY
MATERIALS
By D. SANDELL

The techniques are described and the suitability discussed of the Karl Fischer method
for the determination of water content of sugar confectionery and sugar confectionery raw
materials. The method is shown t o have some considerable advantages over the con-
ventional, oven-drying methods used for most control and investigational purposes. It is
rapid, tests being carried out in 7-40 min. ; it has a precision of & 1% of the water content ;
and it is productive of results in quantity. Comparable results obtained by the Karl Fischer
and oven-drying methods are presented, showing good agreement in many cases, but some
discrepancies in others.

Introduction
The difficulties associated with the determination of water in foodstuffs are well known,
and are reflected in the large number of techniques described in the literature. The most widely
used method, oven drying, is slow and is in general neither quantitative nor specific for water.*3
While many modifications of oven drying are satisfactory in a limited application, a general
method, reasonably specific for water, accurate and convenient, has been lacking. It is sug-
gested that the Karl Fischer method can be adopted to fill this gap.
The use of this method for foodstuffs is by no means new. Since the method was first
proposed by Fischer3 in 1935 (for the determination of water in sulphur dioxide) many workers
have indicated its obvious usefulness in the testing of foodstuffs. Among others, Fosnet &
Haman4 describe application to cereals and cereal products ; dairy products were tested by
Heinern~inn,~ dried vegetables by Johnson,6 sugars by Zimmermann,' and some industrial
materials by Almy et aZ.* Mitchell & Smith, in their excellent book dealing with the Karl Fischer
m e t h ~ d give
, ~ a valuable bibliography.
The Karl Fischer reagent is a solution of iodine, sulphur dioxide and pyridine in methanol.
It reacts quantitatively with water but deteriorates due to side reactions and thus requires
standardisation a t intervals. The rate of deterioration is a t a maximum when the reagent is
freshly prepared and falls to a low steady value after a few days ; with the commercial reagent,
daily standardisation has been found to be quite adequate. For a fuller description of the
reagent, the reader is referred to the volume by Mitchell & Smith.g

Experimental
The experimental work which has been carried out can be readily divided into two parts :
(I) Development of a suitable titration apparatus.
(2) Development of a suitable technique for extracting water from the material under test,
so that it may be completely titrated.
I n the following text, abbreviations have been used as follows :
K F = Karl Fischer reagent, water equivalent 5 mg./ml. when freshly prepared.
SWS = Standard solution of water in methanol, water equivalent 2.5 mg./ml.
MA = Methanol, dry, of analytical reagent quality.

(I) Development of titration afiparatus

The requirements were : (a) the reagents should be fully protected against atmospheric
moisture ; (b) an electrical method of indicating the titration end-point, since coloured materials
mask the colour change of excess reagent (brown) to spent reagent (yellow) ; (c) a standard and
inexpensive titration vessel with rapid interchangeability, in which both the extraction and
titration could be carried out.
The apparatus as developed is centred on a standard Bakelite, screw-capped z fl. oz. jar.
J. Sci. Food Agric., 11, November, 1960
672 SANDELL-APPLICATION OF KARL FISCHER METHOD
The cap is drilled out to take a polythene plug carrying the K F and SWS burette tips and the
electrode assembly. A second jar is fitted to carry the MA burette tip. The burettes are,
respectively : left-hand K F burette, capacity 25 ml. ; right-hand SWS burette, capacity 25 ml. ;
and a standard 50-ml. burette for the MA. Each burette is fitted with two-way tap, guard bulb,
and a B7 cone on the inlet, fitting directly into a B7 socket on the reservoir head.
The three reservoirs are identical, being made up from Pyrex 2-litre conical flasks fitted
with B4o cone and socket ; each cone carries a delivery tube t o a burette, fitted with a B7
socket ; a connexion to a guard tube ; and a refill tube fitted with BIO socket and stopper.
Three guard tubes are required, filled with silica gel.
The titration vessel is supported over a suitable magnetic stirrer, which can be swivelled
about a retaining bolt so that the jar is easily unscrewed and removed.
Burette filling is carried out by the vacuum side of a Dymax Mk. I1 compressor, and the
pressure side of the same compressor provides an air stream which, after being dried, is used t o
mix the liquids in the reservoirs after the burettes have been rinsed.
The electrode system is of the ‘ dead stop ’ type, and employs a Siemens 1.5-V B5 cell,
a Clarostat type 58 ~ o o - k nwire-wound linear potentiometer, a Sifam type M35 0-25 micro-
ammeter, and two electrodes made up from 0.5 mm. diameter platinum wire.
The whole apparatus (Fig. I) is set up on a Perspex case, which has removable panels to
allow easy access to the reservoirs for filling.
Method of titration and the standardisation of the reagents
The burettes are filled and drained two or three times, and a small quantity of air is blown
through the reservoirs t o ensure homogeneity of the reagents. The burettes are then filled and
brought to the zero mark.
The first balance required is that between the K F and SWS. Ten ml. of K F are run into
the jar, which contains a magnetic float ; the speed of the float is adjusted, and SWS added until
the electrodes are immersed. The potentiometer is then adjusted until a deflection of 25 pA
is observed, a considerable excess of K F being present, as shown by the dark brown colour.
SWS is then added until the colour turns to a light brown, and finally the additions are made
in increments of 0-1ml. with an interval of 10 sec. between additions to ensure completeness
of reaction. As the end-point is approached, the meter reading falls slightly, and a t the end-
point, a marked deflection is observed from approximately 23 to 12 PA. The titration is re-
peated, and the two values should not differ by more than 0.1 ml.

I Vacuum drying tube

z Atmospheric pressure drying tube
3 Positive pressure drying tube
4 Compressor
5 Magnetic stirrer

J. Sci. Food Agric., 11, November, 1960

 SANDELL-APPLICATION OF K A R L FISCHER METHOD 673
The second balance required is that between K F and water. One drop of distilled water
is weighed into a jar whose lid is fitted with a rubber washer, and excess K F is run in. The
K F is back-titrated as described above, and a duplicate titration then carried out.
The third balance is that between MA and KF. Exactly 10 ml. of MA are treated with
excess K F and back-titrated, in duplicate.
In an actual titration to determine water content, the water-containing extract is treated
with excess K F and back-titrated with SWS as above. The water equivalent of the KF, and
the K F equivalent of the SWS and MA are known from the standardisation ; the volume of K F
actually used in titrating the water in the extract is therefore
total volume of K F added - (volume used in reacting with water in methanol and volume
used in reacting with SWS on back-titration).
Since the water equivalent of the K F is known, the quantity of water in the extract, and
hence the water content of the material under test, can be calculated.
(2) Extraction techniques
Initially, attempts were made to titrate the water in some materials directly. In most
cases, low values were obtained, and in all cases results were erratic. Some form of extraction
technique was, therefore, essential, and two methods were devised.
Method A
The material was heated with methanol in the standard jar, which was covered with a loosely
fitting Bakelite cap. It was found that the mixture of methanol plus material could be safely
heated until rapid boiling commenced, no appreciable amount of water being gained from or
lost to the atmosphere. When rapid boiling occurred, some water was lost. A hotplate was
employed for heating, set so that the mixture was just coming to the boil after heating for 5 min. ;
this 5-min. period gave complete extraction of water from a considerable range of materials,
especially those of a finely divided nature. Table I shows the effect of heating methanol/water
mixtures (plus a little dry sand to promote boiling a t the earliest possible moment) for varying
times by method A. The quantity of water present was that which would ideally be present
in a normal determination, i.e., equivalent to about 20 ml. of K F ; the quantity of methanol
was 10 ml., the standard extraction quantity.
Method B
As in method A, the material was heated with methanol in the standard jar, but the jar
was fitted with an air condenser. Here it was shown that extraction times of up to 30 min.
would be quite safe. The results are detailed in Table 11.
With either method of extraction, it is obvious that risk of loss of water depends on the rate
of boiling. Vigorous boiling, even with an air condenser fitted, carries a risk of loss of water.
The hotplate used was therefore modified by insulating half the heating surface with a hard-
board sheet ; with a thermostat setting to give a hotplate temperature of 130°, the temperature
on the insulated surface is 70°, i.e., not much higher than the boiling point of methanol (65").
The mixtures are brought to the boil on the hot surface, and then maintained a t a gentle simmer
for the rest of the extraction period on the insulated surface.
A useful guide as to the efficiency of the air condensers is that the loss of alcohol on heating
for 60 min. should not exceed 0.25 g., i.e., 2 5 % of the weight of alcohol.
Table I
Recoveraes of water z.n methanol/wuter mnzxtures heated by Method A
Time of yo Recovery of water
-~
Mean
heating, Series I Series 2 Series 3 recovery
min. %
0 100'0 99.6 99.7
2 100'0 100.7 99'7
100.3 100.3
4 100'0 100.7 100.3 100.3
6 101'0 101'0 100.8 100.9
8 101'0 99.6 99'4 100'0
10 101'0 99.6 100.3 100.3
I5 98.2 99'0 98.6 98.6

J. Sci. Food Agric., 11, November, 1960

674 SANDELL-APPLICATION OF KARL F I S C H E R METHOD
Table I1
Recoveries of water in nzethanollwater mixtures heated by Method B
Time of 7; Recovery of water Mean
heating, Series - recovery
min . I 2 3 4 5 6 O/
/Q

0 98.8 98.8 98.7 100'1 99'9 100.7 99'5

I0 98.7 98.7 98.8 99'7 99' I 99'7 99'1
20 98.7 98.8 98.6 100'0 100'1 100'2 99'4
98.3 100.3
30
40
99'1
100'2
100'0
99'0
100'0 101.3
100'0
++ ++ +
99'7
++
99'4
100.5
100.3
50 100'0
60 100'2 100.4
101'0
100'2 + + 100.3
+ no observation made
Extraction equipment
(a)Hotplate : Townson and Mercer model X103 hotplate operating a t 130 & 2 O , fitted

with a hardboard sheet across half the surface, with a surface temperature on the sheet of 70-75'.
( b ) Extraction vessels : standard z fl. oz. jars, transferred direct to the titration apparatus
after extraction.
(c) Air condensers : specification given in Fig. 2 .
( d ) Glass sheets: microscope slides cut t o 2.5 cm. x 1.9 cm.

FIG.z.-Air condevuev f o r extraction by method B

Made up in 6 mrn. o j d soda glass ; dimensions in cm.

A$plication to confectionery materials

The first step was to devise methods of treating the materials for extraction so as to give
a suitably large surface area. Suitable solids were ground in the usual way, in a mill, to a given
mesh size ; other, flexible solids (e.g., liquorice in sheet form) were extruded through metal
rollers to give a thin sheet of constant weight per unit area. Pastes and viscous liquids were
spread thinly on a glass sheet. Some materials, such as thin syrups, finely divided materials
(e.g., flours) and materials soluble in methanol (e.g., fats) did not need any preliminary treatment.
Extraction curves for each material were then obtained by extracting samples for varying
periods and plotting the apparent water content as ordinate against the time of extraction.
The value a t which the curve flattened out was taken as the water content. An extraction
time was then set for the particular material by taking the time required for the curve to flatten,
plus approximately 100% as a safety factor, subject to the proviso that extraction times by
method A should not exceed 6 min., and by method B, 30 min.
A further check was applied in the case of ' difficult ' materials such as cooked starch paste,
gelatin etc. The extract was titrated immediately on completion of the extraction period,
and then stirred for 3 min. with excess K F before retitrating. Since K F is much more hygro-
scopic than MA, two values of water content were thus obtained, and two curves plotted, which
theoretically should have joined a t the point of flattening out. In point of fact, with some
materials, the two curves did not actually meet, the ' immediate titration ' values remaining
0.1-0.2% below the other ' 3-min. stirring ' values. For materials which hold water tenaciously,
therefore, a 3-min. stir period with excess K F is specified before back-titration. It was estab-
lished that this 3-min. stir period was sufficient to complete removal of water from the material.
Fig. 3 illustrates typical curves obtained by method B.
J. Sci. Food Agric., 11, November, 1960
 SANDELL-APPLICATION OF K A R L FISCHER METHOD 675

FIG. 3.-Extvactioia cusves by method B for some

matevials
A Fudge, immediate titration
B Fudge, titration aftcr 3-min. stir with excess K F
C Gelatinjsugar syrup, inirncdiate titration
D Gelatinisugar syrup, titration after 3-inin. stir with excess KF
E Pectin 60'80 mesh immediate titration
F Pectin: Go,"So mesh,' titration after 3-min stir with excess IIF

0 5 10 I5 20 25 30
EXTRACTION TIME, min.
Results
Tables I11 and IV give the experimental procedures for a range of raw materials, inter-
mediate products and confectionery, based on extraction curves obtained by the methods
detailed in the preceding paragraph.
Table V gives an indication of the significance of results obtained by the Karl Fischer
method, with confidence limits for single and triplicate tests a t 95% probability. The sample
number in all cases was 10.
Table 111
Methods fov confectionesy saw matevials
Material Preparation Extraction Time, 3 min. Approx. sample
method min. stir KF weight, g.
Sugar, granulated None A 5 No 5'0
,, raw A 5 5'0
icing 4 5 5' 0
Maitose A 5 2'0
Dextrose A 5 2'0
Thin sugar syrups A 5 0'5
Glucose syrup Tgin filma B 30 Yes 0'5
Caramel >>
a
, I B I5 NO 0'5
Treacles a
B 15 0'5
+- 4 o C
, I I ,

Gelatin 30 B 30 Yes 0'5

Gum arabic +60 B 30 0.5
-80h
Liquorice extract +- 4300 b B 30 0'5
Agar +- 430o C B 15 0'5

Pectin +-8oC60 B I5 0'5

Wheat flour None A 5 0'5
Starch maize A 5 0.5
Starch, acid treated A 5 0.5
,, moulding A 5 0'5
,, farina A 5 0.5
Cocoa powder B I5 2'0
Coconut, medium B 20 2'0
,, fine B I0 2'0
flour B I0 2'0
Coidensed milk &'in filtna B I5 0'5
Citric acid None A 5 I '0
Fats soluble in methanol A 5 2'0

Qliquid spread as thin film on glass sheet

ground by mill t o give mesh size
Csample graded t o give correct mesh size

J. Sci. Food Agric., 11, November, 1960

676 SANDELL-APPLICATION OF KARL FISCHER METHOD
Table IV
Methods f o r confectaonery in pyocess and finished confectionery
Material Preparation Extraction Time, 3 min. Approx. sample
method min. stir K F weight, g.
Liquorice paste (undried) Thin filma B 20 Yes 0'3
Liquorice (dried) Extruded into B 30 0.5
thin sheet
Cream paste Small pieces* A 5 No 1'0
Fondant b B I5 Yes I '0
(until
dissolved)
Gelatin/sucrose syrup Thin films B 15 Yes 0'5
Gelatin-based sweets before
depositing ,, I > B '5 Yes 0.5
Starch-based sweets before
depositing 2, I, €3 '5 0'5
Gum arabic-based sweets
before depositing Thin filma B 30 Yes 0'5
Boiled sweets Broken into* None Dissolved in excess K F 2'0
small pieces before back-titration
Fudge Small pieces B 30 Yes I '0
Chocolate Shavings by h 5 2'0
razor blade
Fondants Chopped by B 15 Stirred 1'0
razor blade until
dissolved
Marshmallows Chopped by B 15 No 0.5
razor blade
Pontefract cakes Extruded into B 30 Yes 0'5
thin sheet
Lozenges Ground by mill, B 10 NO 2'0
+30 -40
a liquid spread as thin film on glass sheet
' * broken into small pieces by hand

Table V
Signi3cance of results (n = 10 in all cases)
Material Average Range Std. Limits a t 959/0
"/u water deviation ~~~
probability
~ ~

Single Mean of
test triplicates
Sugar, raw 2'12 0.09 0'035 *0.07 k0.04
,, syrup
Treacle
Glucose syrup
26.0
18.5
I 8.4
0'5
0'3
0.6
0'3.5
0.09
0.23
*
i0.7
0
k0.45
' 2
s0.4
-10.1
10.26
Gelatin 16.7 0'4 0.13 iT0.27 h0.16
Pectin 12'2 0'4 0.16 *0.32 k0.19
Wheat flour '4'4 0'25 0.07 &0.14 +0.08
Starch 9'3 0.6 0.16 kO.33 to.19
Coconut 3'2 0.16 0.05 j0.10 i0.06

Cream paste 4'9 0.15 0.04 k0.03 Lto.02

Lozenges 1.85 0.03 0'02 h0.05 &o.o3
Boiled sweets 3.49 0.18 0.05 j0.10 sJ0.06
Chocolate 1.35 0.08 0.03 k0.06 t0.04
Marshmallows 18.5 0'4 0'22 40.44 &0.25

Table VI shows the relationship between values obtained by the Karl Fischer method
and by oven drying ; the oven drying procedure in section I of the table was direct drying for
5 h. a t 100' ; in section 2 , the material was dispersed on sand before drying for 16 h. a t 100'.
All values quoted are means of duplicate determinations.
Discussion and conclusions
The obvious advantages of the method a r e :
(a) Rapidity of test. Tests can be carried out on all the materials listed in this paper in
40 min. or less. Many materials take only 7 or 8 min.
( b ) Reproducibility of results. The degree of precision indicated in Table V is adequate
J. Sci. Food Agric., 11, November, 1960
 SANDELL-APPLICATION OF K A R L FISCHER METHOD 677

for most purposes ; greater sensitivity could be obtained (at the expense of some of the con-
venience of the method) if required. Generally speaking, a precision of &I% of the water
content, based on the mean of triplicate determinations, is quite possible.
(c) Specific nature. The method is not absolutely specific for water, but interfering sub-
stances (e.g., bases) are not common constituents of foodstuffs. There is a special advantage
over oven drying when heat-sensitive materials, or materials with volatile components, are tested.

Table VI
Relationship between Karl Fischer and oven drying values of watev content
Section I Section 2
Material Karl Oven Material Karl Oven
Fischer drying Fischer drying
76 "/o Yo 76
Raw sugar 2'1 2'1 Condensed milk 27'3 27.2
Coconut, fine 2.8 2'7 Fondants* 9.2 9'1
,, medium 3'0 3'0 Marshmallows* 18.5 I 6.4
,, flour 3'2 3'0 Boiled sweets* 4'0 3'0
Cocoa powder 4.6 4'3 Chocolate* 1'4 1'4
Starch (i) 6.5 6.0 Lozenges 1'9 4.9t
,, (4 9'7 8.7 Liquorice Allsorts 9'9 9'9
,, (iii) 10.6 9.6 Treacles (mean of 26) 20:9 22.7
Flour (i) 11.8 11'0 Cream paste (mean of 12) 6.0 5'7
,, (4 16.8 14'4
Agar 18.2 I 6.4

* Samples purchased from confectionery retailers

t This high value is due t o the presence o f volatile matter other than water

?he question of whether the results obtained with the reagent are absolute is a difficult
one, because of the lack of a general reference method ; however, the evidence indicates that
the results are a t least as genuine as those obtained with oven drying. In this connexion, an
examination of the results in Table VI will reveal the difference between comparative values
by Karl Fischer and oven-drying methods. These fall into two sections:
(I) Where the oven drying was for 5 h. a t IOO', a method applied to finely divided solids.
(2) Oven drying a t for 16 h., after dispersion of the material on sand.
IOOO

In section (I) results for both methods are in good agreement for materials of low water
content, but an increasing discrepancy is noted for materials of higher water content. As a
matter of interest, materials dried in this manner can be titrated with the reagent, and an
apparent residual water content, equal to the discrepancy, found. This suggests that the
discrepancy is due to incomplete driving off of water in the oven.
In section (2)there does not appear to be any pattern in the discrepancies ; good agreement
is obtained with some materials but substantial differences with others. In the case of lozenges,
the large difference is due to the presence of volatile material other than water, i.e., chloroform.
Values for treacle by oven drying are much higher than Karl Fischer values, which suggests
that some decomposition occurs on heating ; and with the marshmallows, where the water is
tenaciously held by the colloid components, oven drying appears to give low values.
Finally, some indication of possible sources of error in the light of experience, might be
fitting. Possibly the greatest weakness of the method lies in its sensitivity, which means that
only small sample weights can be tested of materials with a high water content. Precautions
must be taken to prevent loss of water, especially when the material is to be spread as a thin
film ; these include the use of rubber-sealed jars and the weighing of material when cold or only
just warm. Care in extraction is also necessary ; the material must be prepared to the correct
specification for surface area, and rapid boiling must be avoided. The reagents used are all
hygroscopic, and must not be exposed to air for longer than is necessary. All equipment in
contact with methanol must be dried by heating before re-use.
If any doubt exists about completeness of extraction, a useful check is to leave the extract
for a few minutes in contact with excess K F after titration, and then retitrate.
J. Sci. Food Agric., 11, November, 1960
678 COLEBY et a1.-IONISING RADIATIONS A N D MEAT. IV
Acknowledgments
Thanks are due to Dr. J. W. Hughes for assistance in the preparation of the paper and read-
ing of proofs ; to J. Preston Ltd., 208 West Street, Sheffield, I, for assistance in the construction
of the apparatus, and to the directors of Geo. Bassett & Co. Ltd., for permission to publish this
paper.

Process & Product Research Section

Geo. Bassett & Co. Ltd.
Sheffield, 6
Received 15 February, 1960

References
1 Lindsay, W. N., & Mansfield, T., Iizdustv. Engng Johnson, C. M., Industv. Engng Chem. (Anal.),
Chem. (Anal.), 1944, 16, 628 1945, 17, 312
Makower, B., Chastain, S. Jl., & Neilson, E., Zimmerman, A., Fette Seijen, 1939, 46, 446
Industv. Engng Chenz , 1946, 38, 725 Almy, E. G., Griffin, W. C., & Wilcox, C. S.,
Fischer, K., Z . aizgew. Chevlz , 1935, 48, 394
Fosnet, R . H., & Haman, R. \T'., Ceveal Chem., Industv. Engng Chem. (Anal.), 1940. 12, 392
1945, 22, 41 Mitchell, J . , & Smith, D. M., ' Aquametry ', 1948
Heinemann, B , J . Dazvy Scz., 1945, 28, 845 (London : Interscience Publishers Ltd.)

TREATMENT OF MEATS WITH IONISING RADIATIONS.

1V.-Comparison of the Deterioration in Quality during Storage of
Eviscerated Chicken Carcasses Treated with Chlortetracycline or
Radiation*
By B. COLEBY, M. INGRAM, H. J. SHEPHERD and M. J. THORNLEY

Eviscerated chicken carcasses were treated with chlortetracycline (CTC) or irradiated

with 0.3 or 0.6 Mrad, and then stored at oo. Changes in quality during storage up t o 26 days
were studied, and microbial counts made on the carcasses. 0.6 Mrad caused a fairly rapid
decline in quality, but with 0.3 Mrad or CTC deterioration of quality was only noticed when
the carcasses had been stored a t oo for 19 days or longer. This loss of quality was not due t o
the growth of micro-organisms.

Introduction
In earlier papers it was shown that pasteurising doses of ionising radiation (up t o I Mrad)
considerably delay the microbial spoilage of eviscerated chicken carcasses stored a t chill tem-
peraturesl but that deterioration of eating quality is apparent before that type of spoilage
occurs.2 The present experiments were made t o determine if this deterioration is due t o slow
growth of the micro-organisms present, or is a consequence of irradiation, or whether i t is a
natural occurrence during storage of unfrozen chicken carcasses.
Eviscerated chicken carcasses can be stored a t chill temperatures for only a short period
(usually not more than about 12 days a t 0') before putrid smells develop, indicating microbial
spoilage. Treatment with tetracycline antibiotics delays this spoilage, and enables quality
* Part I11 : J . Sci, Fd Agvic., 1960, 11 61
J. Sci. Food Agric., 11, November, 1960