E ANALYTICAL DETERMINATION OF REDUCING

SUGARS
By F. H. C. KELLY
Introduction.

There have been very many methods suggested from time to time
for the determination of reducing sugars, and considerable difficulty has
been experienced in obtaining one which is sufficiently foolqxoof to give
satisfactory results in the hands of all operators. In 1936 von Stieglitz
and Home 11) reviewed several methods and recommended the electro*
metric determmat~onof the end point with the EynowLane Method.
In 1937 the International Commission for Uniform Methods of Sugar
Analysis [2) received reports on three methods whi'ch appear to give
good promise of their ability to eliminate or at least greatly reduce un.
desirable features of the Soxhlett solutions used for the Eynon~Lane
For example, the effect of variations in the boiling
determination.
period and reducing effect of sucrose are at least equally as important
as the correct identification of the end-point. Also a single solution
completely non.auto,reducing can be substituted for the two separate
Soxhlett solutions and a larger volume is generally used for each titra,
tion enabling more accurate measurement of the volume of reacting
solution. T h e undesirable physical features of titrating a boiling s o h ,
tion are also eliminated. There are perhaps two aspects which are some,
times claimed in favour of the Eynon#Lane method viz. (1) relative sim,
plicity of technique and (2) in the case of materials containing a large
proportion of reducing sugars (such as final molasses) the welght of
sample which can be used for analysis is larger for the Soxhlett solution.
Each of these aspe,cts has been investigated and the results are recorded
in this paper.

A very comprehensive review of methods for the determination of

reducing sugars is contained in Browne @ Zerban's "Sugar Analysis'" {3)
which should be consulted for details of methods and interesting corn,
ments by the authors.

Methods considered by the International CO

T h e three methods referred to by the International Commission
are :Berlin Institute Method.
Ofner's Direct Method-officially adopted by the Czechoslw
vakian Beet Sugar Industry.
3. Luff-Schoorl Method-used
in Java for the determination of
invert sugar in cane sugars.

l.

2.

In each case solutions of low alkalinity are used-sodium

carbonate
(more or less buffered) being substituted for sodium hydroxide. T o
conserve space it is not intended to outline the details of technique of
the above methods as such is well set out in the report of the International
Commission which should be readily accessible to most readers.

din Institute
The' Berlin Institute Method possesses one advantage over the other
two in being conducted at a temperature below that of ebullition with,
out requiring a longer reaction time. Smolenski in reporting to the
International Commission considered this aspect to be worthy of special
note in the selection of a suitable method. H e also indicated his pre,
ference for a.method which required the determination of the preci,
pitated cuprous oxide without separation from the liquid. In this re,
spect both the Berlin Institute and Ofner's methods are similar. For
the purpose of the present investigation the incomplete nature of fundamental data available to the author militated against the selection of the
ethod for investigation as well as a weakness similar
to that responsible for the final rejection of Ofner's method for routine
purposes, ziz. the instability of the Iodine solution.

Ofner's method has recently been comprehensively investigated by

Jackson and McDonald {4f at the Washington Bureau of Standards.
In the publication of these results detalls are given which were lacklng
in the report to the Internatlonal Commission and which made lt possible
to make a more accurate assessment of the acceptability of the method.
The present writer was very favourably impressed after a trial of the
method and would give lt first preference for the determination of re,
ducing sugars in both refined and raw sugars.
But as a method for
general use m the routinc laboratory it appeared to have two disadvantages, viz. (1) the unstable nature of a standard iodine solutlon and (2)
the very small amount of sample which would be required if the method
were used for determining reducing sugar in molasses. It might be of
interest to mention here that the adaptation of this method for the determination of traces of sugar in condenser water is possible with a high
degree of precision, and concentrations as low as one part in 10 million
can become significant if a colour match is ilsed for the titration. For
investigatory purposes these advantages are considered to greatly offset
the relative simplicity of the Molisch test with alpha naphthol which at
its best only gives the order of the concentration. T h e present writer
also prefers it to the various colorimetric methods which have come into
prominence from time to time.

T h e Schoorl's iodide method with L u g s solution appealed to the

writer as one capable of general use in the routine laboratory for the de,
termination of reducing sugars in, all grades of products from " A sugar
to final molasses with a high degree of accuracy, and also for the deter,
mination of sucrose in molasses after its inversion. The method is well
described by Browne and erban in their book on Sugar Analysis (3rd
Edn.) but unfortunately the table published therein was later found to
be in error. A more correct table compiled by K. Dekker appears in

the report of the International Commission and also in the Washington

Bureau of Standards Circular C440--"Polarimetry and Saccharimetry
of Sugars" 17). The present writer was also unable to agree with the
original table, the use of which proved misleading until the existence
of a more recent table was recognised.
The original Luff's solution, made up of 17.3 g. CuSO4 .5'H20, 11Y g.
citric acid and 185.3 g. Na2C03 per litre, was found during the course
of these investigations to be very sensitive in copper reducing power to
small variations in the carbonate concentration. I t was found essmtzal
to analyse the sodium carbonate used for the solution in every case, exr
cept when a freshly opened bottle of A.R. grade material could be used.
I t did not appear clear whether the quantity of citric acid used referred
to the monohydrate (in which form it is invariably marketed) or to the
anhydrous equivalent. This differentiation was important for this particular formula because of its effect on the copper reducing power of
the solution. For these investigations l l f g. of citric acid monohydrate
C
(&&$&.
H2O) was used for each litre of solution. It might be men,
tioned here that A.R. grade sodium citrate dihydrate ( N a & & % & . 2H20)
is available on the market and could be used to replace the citric acid and
some of the sodium carbonate (e.g. 161 g. sodium citrate dihydrate plus
98.3 g. anhydrous sodium carbonate per litre).
The effect of variations in alkalinity was tested and it was found
that if an accuracy of f 1.0 per cent. was desired then it was necessary
to use the correct amount of Na2C03 to within f 0.5 or
1.0 per cent.
if using sodium citrate $us Na2C03. Presumably the weight of citric
acid or sodium citrate would require to be within the same limits. Had
the original weight of Na4CO3 been increased by about 10 to 20 per cent.
with the same weight of CuS04. f H 2 0 and citric acid it appears that a
solution much less susceptible to this source of error would have been
obtained.
A five minute boiling period was specified for the method. This
is open to error due to variations in time of bringing to boil, correct
identification of time of commencement of boiling and time between
cessation of boiling and effect of cooling. A series of tests was made
varying the time of boiling and maintaining the other factors as nearly
constant as possible. Care was also taken to standardise the rate of boiling within practical limits. It was found that with a five minute boiling
period an error of

i- 15' seconds =
4 parts per 1000
3- 30
,, = -i- 8 ,, ,,
+4$
,, = + l 1
,, ,,
+ 60 ,, = + l 3 ,, ,,
-- l$ seconds = - T parts per 1000
- - 10
- 30
,, ,, ,,
- 4s
,
- - 21 ,, ,,
- 60
,, - - 24
,,
,,
I,

.I,

9,

9,

'Thus a slight error in overboiling is leqs serious than a similar error

in underboiling. For the average routine determinati~n~the

error would

1.0 per cent. It was found that had the

probably be not less than
boiling period been selected as 10 minutes instead of five the probable
errors
from this cause would have been reduced tenfold. In order to
.-take advantage of this fact it would have been necessary to prepare an
entirely new table. T h e errors liable to arise during preparation of the
solution were of such magnitude that the work involved in preparing
a new table did not appear to be justified.
Other difficulties experienced with this method were d c c ~ m ~ o s i t i o n
of the potassium iodide solution, and the presence of oxides of nitrogen
in the sulphuric acid used in the last stage for the acidification of the
solution. Each of these effects produced low results. In the Case of
the K1 solution, decomposition may havc been retarded by increasing the
amount of alkali or by keeping in a dark bottle or both. Keeping in a
dark bottle did not commend itself favourably because decomposition
would probably not be detected if it occurred for some rcason other
than photo,reaction. After several unfavourable experiences with K1
solution i t was declded to weigh out K1 as required. T h e only sulphuric
acid available was "chamber acid" and this produced distinct "afterblueing" and erroneous results. T h e substitution of hydrochloric acid
was completely successful in overcoming this difficulty.
The modification which Schoorl found advantageous for the Luff's
solution was then tested. In t h s case more copper and less citric acid
were found by him to give a solution which with a 10 minute boiling
period had exactly the same reducing power for both dextrose and levu,
lose. This is an advantage in analyses of sugar mill products in cases
where the dextrose~levuloseratio is not constant for seasonal or other
reasons. Such advantage applies also to any comparison with the Soxh.
lett solution.
The modified Luff's solution is made up from 25 g.
G u S O ~ . SH20, 70 g. citric acid and 388 g. Na&03. 10H20. The
specification of the decahydrate instead of the anhydrous form has some
advantages, but under conditions of low humidity in a hot climate efflor,
escence is likely to be quite significant, and here again an analysis should
be performed unless using a new bottle of A.R. grade. The effect of
errors in the composition of the sodium carbonate was found to be very
much less than with the original Luff's solution. A variation of h T per
cent. did not have any significant influence on the copper reducing power,
'whereas an extra 10 per cent. only increased the copper reducing power
by 1 per cent.
T h e effect of variations in boiling time was significantly less than
was the case with the original Luff's solution at the five minute mark,
but slightly greater when each was compared after 10 minutes boiling.
e latter difference was however scarcely significant being forf 15 second = & 1 part per 1000
f 30
,, = f 2 ,, ,,
,,
f 45
,, = i 3 ,, ,,
60
,, = f 4 ,, ,,
9,

9,

I t was noted that Schoorl had substituted potassium iodide crystals

and hydrochloric acid in his modification, which changes are considered
desirable after experien'ce \in this laboratory. T h e advantages of the

ified Luff,Schoorl method over the original method appear to be

ently significant to justify preference in its use. Also the realisation of these improvements makes possible the substitution of the modi.
ff,Schoorl method for the Eynon~Lanedetermination using Soxh,
lett solutions.
lication of the Luff-Schoorl Met

The following gives an indication of the degree of accuracy obtained

under the conditions examined :(l) Composition of solution . .,
.. ..
0.1 per cent.
(2) Boiling time
..
..
..
..
0.2 ,, ,,
(3) Measurement of 25 ml. of modified Luff's
Solution
..
. . . . . . .. f 0.1 ,, ,,
(4) Measurement of 25 ml. of sample
. . + 0.1 ,, ,,
( 5 ) Accuracy of titration (say 10.0 ml.) . .
+ 0.5 ,, ,,
this gives
0.3 per cent. for 15.0 ml. of solution consumed.
Thus the overyall probable error would be-

+
-

J0.1"

0.2'

+ 0.1"

+.l2

+ 0.3"

= f 0.4 per cent.

The effect of such on the following analyses would be.001 g. per 100 g. sample
(a) Reducing sugar in raw sugar =
(b) Reducing sugar in final molasses = f .O$ g. per 100 g. sample
(c) Sucrose in final molasses = _f 0.2 g. per 100 g. sample
(d) True purity of final molasses = 1- 0.3 g. per 100 g. sample

The* two aspects in which the Eynon,Lane analysis may be com*

petitive can now be discussed:-practice and the cor.
(a) Relative simplicity of technique-with
rect apparatus suitably arranged it has been found possible t o complete
an analysis or a series of analyses with approximately equal expedition
by either method. I t is considered that differentiation between the
Eynon-Lane and the Luff,Schoorl techniques on these grounds is difficult,
and is largely influenced by the familiarity of the operator with one or
the other method.

(b) The effect of differences in weight of sample is not as serious

as is sometimes reported to be the case and actually the modified method

is slightly better than the original method. The final volume of solu,
tion should contain about 40 mg. of reducing sugar compared with 50 mg.
in 25 ml. with the Eynon,Lane method. The processes of weighing,
to volume and measuring 25 ml. volumes should all be at least
as 2 1 part per 1000; thus a 20 per cent. difference in weight
sample is really not significant, even from the point of view of sample
Unfortunately the table quoted for the modified Luff's solution in

Browne 9 Zerban does not indicate the corrections necessary for the
nation of reducing sugar in the presence of sucrose. A table
refare prepared for use in this laboratory using "AnalaR

Milligrams R.S. per 27 ml. solution when concentration of sucrose

per 27 mc aliquot isl-

. dextrose and dried refined sugar of polarisation 99.9"+ and cont s n i n g 0.043 per cent. reducing sugars as determined by Ofner's method.
For the determination by Ofner's method the standard of comparison
between solutions used in this laboratory and those used for the preparation of the table by the Jackson E3 McDonald was the " A n a l a R
B.D.H. potassium dichromate used for standardising the sodium thio.
The results obtained for this investigation are set out in Table

It might be mentioned that in order to obtain thc end,point more ac.

curately the sodium thiosulphate titration was only carried to within
0.7 ml. of the end,point with 0.1N. Na2S203and completed with 0.01N.

Na2S203.
'

T h e matter of clarification of samples before conducting the re.

ducing sugar analysis is apparently an open question among analysts. If
the only purpose of clarification is to render the end~pointmore exactly
identifiable then it does not appear to the present writer to be warranted
even with such dark products as final molasses when the modified Luff.
Schoorl method is used. So far tests in this laboratory have not indi,
cated that clarification removes reducing nonmgars in quantities which
might be considered significant in general analytical work.
T h e re.
moval of calcium is however important if such is present in large amounts.
If this is not removed then it will precipitate as CaCOx from the copper
carbonate solution and affect the copper reducing power of the latter.
T h e general run of samples of sugar mill products met with in this
laboratory have not indicated justification for its remo$al. However,
should such be desired calcium is better precipitated as carbonate i n
solutions made just alkaline to phenolzphthalein.
Solutions of such
alkalinity have not been found to affect the copper reducing power of
the modified Luff's solution. T h e use of oxalate as a precipitating radicle
has been found disadvantageous in that any excess present reacts with
potassium iodide in the presence of the hydrochloric acid used for the
final acidfication, and iodine is liberated by the oxalate as well as the
copper.

n Stieglitz, C . R., and Hame, L. C.

"Report on Analytical
ethods used in Factory Control." Proc. Q.S.S.C.T. 1936. p. l 01.
2 ) Report of International Commission for Uniform Methods of Sugar

Analysis.

Int. Sugar Journal, Vol. 39. 1937. pp. 1st-40s (appendix).

rowne, C. A., an
1941.

Zerban, F. W.

Jackson, R. F. an
Sugar in Refined Sugars."

"Sugar Analysis."

3rd. Ed.

ald, E. J. "Determination of Invert

Int. Sugar Journal. 1944. p. 41.

ates, F. J. and Associates. 'Polarimetry Saccharimetry and the

Sugars." Circular 440 of the National Bureau of Standards, Washington. May 1. 1942.

Kalamia Estate,