Proceerlings of The South African Sugar Technologists' Association -June 1970

A SOLIDS BALANCE INVESTIGATION

By E. L. MULLER
Hulett's S.A. Sugar Mills and Estates Limited

Abstract Due to the relatively small quantity of moisture

This paper is an abridged version of a thesis which in 'the chemicals added, the weights of these
was submitted to the Sugar Milling Research Insti- materials, as recorded by the factory, have been
tute in accordance with the requirements for the used.
Sugar Technology Course. The investigation was carried out at Darnall where
The objectives of the thesis were to assess the the vacuum flotation clarification process was in use.
magnitude and significance of errors inherent in the
various methods of brix determination, and to investi- The Bdx Balance
gate a method for calculating a true solids balance in In other countries, e.g. Java, an attempt was made
a raw cane sugar factory. The methods of analysis to introduce a brix balance by assuming no loss of
and the facts revealed by the literature survey have sucrose during clarification, other than in filter cake.
been omitted from this paper. Despite the shortcomings of this method as used in
,Java, it is applied in the Hulett mills and recently
Introduction there was an occurence at Empangeni which proved
its value beyond any doubt.
Brix is a parameter of major importance to fac- For a number of years, the "non-Sucrose Ratio"
tory control. This investigation was undertaken to as defined by P e r k ' V a s found a rather strong
assess the magnitude and significance of errors following.
inherent in the various methods of brix determina-
tion.
Non-Sucrose Ratio =
A literature survey which was carried out revealed
that a number of investigators throughout the world
have looked into the problems associated with brix
Tons non-sucrose in final molasses + tons
non-sucrose in sugar
determinations. The effects of suspended solids, dilu-
tion, different techniques of oven drying, etc., have Tons non-sucrose in mixed juice
been related to the densimetric and refractometric
methods of determining the dissolved solids in fac-
tory materials. Let: Tons sucrose in mixed juice = J
It might be justifiable to say that a comparison Tons sucrose in sugar = S
of methods for determining dissolved solids is un- Tons sucrose in molasses = M
necessary, since a vast amount of research has already Tons sucrose in combined losses = L
been done in this field. Also, most investigators have Tons brix in mixed juice = j
been led to the same conclusions, that is, most Tons brix in sugar = s
methods for determining brix are influenced by fac- Tons brix in molasses = m
tors such as suspended solids, non-sugars, etc., and Tons brix in combined losses = c
that these factors introduce an error into factory
control. Then:
However, the erroneous influence that these fac- Non-Sucrose Ratio =
+
(S m--5--M)
tors have on factory calculations and performances s +m + c-S-M-L
has never been examined in any great detail.
Examination in detail is therefore the aim and scope
of the investigation.
The determination of total solids in factory
materials thus enables a solids balance to be made.
The procedure which has been adopted for this Substituting W =(s+m-S-M)
balance is based on:-
W
Solids Entering Factory Solids Leaving Factory Non-Sucrose Ra,tio=
W +c-L
Total solids in Mixed Juice Solids in Filter Cake
solids in Lime added Solids in Molasses
solids in Phosphate The value of W and L are determined fairly
added Solids i n Sugar accurately and, moreover, are more or less constant
solids in Flocculant quantities. Obviously, variations in the non-sucrose
added
solids in Bagacillo ratio are caused by variations in "c" (tons brix in
added combined losses).
Proceedings of The South African Sugar Technologists' Association -June 1970 41

Assume: S =75 Normal facltory with 2.35% To obtain a representative hourly sample, an en-
s =76 loss in cake+ undetermined, tirely new sampling device would have t o be em-
M=8 89% B.H.R. ployed. However, at the time of the investigation no
L =2 40 purity final molasses such device was available. The necessity for an
m =20 hourly representative sample becomes apparent in
For a Non-Sucrose Ratio = 0.9. the solids balance when one calculates the tons of
total solids in mixed julce, i.e. Tons suspended
solids +Tons dissolved solids.
The same problem exists when weighing out the
juice for the analysis of suspended solids, i.e. the
difficulty to ensure an even di~t~ribution of insoluble
material in the 200 g taken.
For a Non-Sucrose Ratio=0.8 At the time of the investigation, there remained no
choice but to analyse the sample as received by the
available system. More attention will be drawn to
this at a later stage in this paper.

Solids i~ Other Material Added

It is therefore submi,tted that an evaluation of the a. Lime
brix loss will give a better understanding of factory It is difficult to assess, from one week to the next,
performance than does the present non-sucrose ratio, the quantity of lime consumed by the process. How-
although it is not suggested that the N-S ratio be ever a fairly accurate estimation can be made of the
discarded. weight of lime used over a period of one month.
In the following report, one must differentiate be- Also, the quantity of solids in lime added is rela-
tween a brix balance and a solids balance as the latter tively small compared to the total solids in mixed
includes all suspended (insoluble) solids. juice. (Solids in Lime % Solids in Mixed Juice=
0.4%).
Solids in Mixed Juice Thus one can use the monthly estimated weight of
. Solids in mixed juice is the s!tarting point of the lime without introducing any significant error in the
solids balance and the accuracy of the balance there- solids balance.
fore depends to a great extent on the determination
of this factor. b. Mono-Calcium Phosphate
This material is delivered in bags of known weight,
Sampling of Mixed Juice and an accurate account of the number of bags used
per shift is kept.
The determination of total solids in mixed juice Solids in M.C.P. % Solids in Mixed Juice amounts
entails two separate analyses. Firstly, dissolved solids to 0.17%. With an accurate monthly weight of
on filtered juice, which can be found by a number M.C.P. available, the quantity of solids added in this
of methods and, secondly, suspended solids. The material is easily arrived at.
accuracy of the latter determination is more depen-
dent upon the sampling procedure than the actual c. Flacculant
method of analysis. The addition of this material is likewise care-
The mixed juice sampling device recommended for fully controlled, and amounts only to 0.004% of the
use in South African factories1 does not give a repre- solids in mixed juice.
sentative distribution of sand and other insoluble Again the consumption of flocculant is known to
material in the final sample. Two reasons can be the pound and this facilitates its use in the solids
given for this: balance.
(a) If there is not sufficient turbulence in the scale
tank before it tips, most d the heavy d. Bagacillo
suspended particles will settle to the bottom This material is added to the muds leaving the
of the tank. When the tank empties the juice clarifiers 'and although it was possible to measure
which reaches the sampling tube first, will the flow rate of the bagacillo in pounds per minute,
contain the greater proportion of sand, yet a more reliable method of determining the quantity
will represent only an insignificant portion of added had to be found.
the total sample. The method finally decided upon is as follows:
(b) The tube itself can be compared to a D.S.M. The dry bagacillo content of the mixed juice varies
screen, especially if 'the holes are canted to one only slightly from hour to hour. By analysing the
side to regulate the quantity of juice obtained mixed juice for dry bagacillo per cent a fairly
per tip. It must be remembered that the recom- accurate average can be found.
mended tubeL consists of 15 holes, each The same applies to the bagacillo content of the
3/32" diameter x 1" pitch. Thus, a large pro- filter cake. Assuming that the bagacillo content of
portion of the sand is deflected away from the clear juice is negligible, the difference between baga-
tube. cillo in filter cake and bagacillo in mixed juice will
42 Proceedings of The South African Sugar Technologists' Association - June 1970

give the amount of bagacillo added to the muds. 2. The average suspended solids per cent mixed
This method is dependent on a reliable weight of juice is also applied to the factory recorded
filter cake. weight of mixed juice. (This analysis has since
The solids in the filtrate do not have to be taken become part of the daily routine carried out at
into account as they are re-cycled continuously. all Hulett factories.)
3. The average dry bagacillo per cent filter cake
Solids in Material Leaving the Factory is used to estimate the quantity of bagacillo
a. Solids in Filter Cake added to the muds from the factory recorded
weight of filter cake.
The determination of solids in filter cake necessi- 4. An average figure for dry bagacillo per cent
tates both weighing the material and analysing for mixed juice enables one to apply this figure to
moisture. Unfortunately, no factories weigh all their the factory recorded weight of mixed juice.
filter cake but estimate the total output by weighing
an occasional S.A.R. truck and road transport
vehicle. The method in use at Darnall is as follows: Analysis of Mixed Juice
An individual truck or lorry is weighed every
'

tenth time it collects filter cake and a record of the

number of loads is kept. The average weight of every Analyses ----- -
tenth load is taken as the weight 'per individual load.
Although it is not accurate, this method had to be Factory
used for calculating the total solids balance. If
this weight is overestimated then the total solids lost
in undetermined will, of course, be underestimated,
and vice versa.
b. Solids in Sugar These figures are the arithmetic means obtained
Due to the fact that the sugar is weighed accurately during the period of investigation. The last column
and that moisture ,determinations are done on samples shows the weighted mean brix recorded by the fac-
from each truckload no difficulties are found in cal- tory during the same period.
culating tons of solids in sugar leaving the factory. The disparity between the factolq figure (13.08)
c. Solids in Molosses and that recorded by the author (12.96) is due to
The determination of solids in final molasses is two factors:
done on a weekly composite sample using the Karl 1. The hydrometer used by Darnall was checked
FiSchers9% 1 0 , 1 1 titrimetric moisture method. The against an S.M.R.I. calibrated spindle and
analyses are done weekly by the staff of the Hulett's found to be 0.1" out.
Central Laboratory. 2. The technique used by the testers was found
to be incorrect. They immersed the spindle into
Solids ina Stock the jar as soon as the vessel was filled. This
The impossibility of determining accurately the procedure leads to air bubbles adhering to the
hydrometer.
quantity of solids in stock, necessitates the use of
"Made and Estimated" figures for sugar and This difference of 0.12 was consistent with each
molasses. hourly sample and is therefore not due to the fact
Errors are frequently made in estimating the that the one figure is an arithmetic mean, while
weekly stock and these will therefore give erroneous the other is a weighted mean.
figures for solids in sugar and molasses made and
estimated. However, by doing a solids balance on Analysis of Filter Cake
the monthly or to-date figures one can minimise the
effect of these stock errors. Moreover they are, Bagacillo Filter Cake
eventually, self-correcting.
Rate of
Moisture %
Results of Investigation
65 Ib/min.
The methods used in this investigation for calculat-
ing the total solids balance relies on the application
of factors to the monthly or to-date figures supplied
by the factory. Filter cake was analysed for moisture in parallel
These factors aie as follows: with the laboratory and no difference was found. A
1. The average difference between factory re- check was made on the amount of evaporation occur-
corded mixed juice brix (spindle) and dry sub- ring from the filter cake between the sampling point
stance per cent filtered mixed juice. This enables and the weighbridge. This was found to be negligible
one to arrive at an estimated dry substance per and only occurred from the top layer of cake in the
cent filtered mixed juice from the factory truck. Thus, the mill figures for moisture per cent
figures. cake are used in the solids balance.
Proceedings of The South Afvican Sugar Technologists' Association -June 1970 43

It was stated previously that in the solids balance, Undetermined Solids Lost: 78 1 tons.
the quantity of dry bagacillo added to the muds is Undetermined solids lost per cent solids in clarified
found by difference, i.e.: juice=3.41%.
Tons dqy bagacillo added = Tons dry bagacillo in The factory reported the following losses for the
filter cake - Tons dry bagacillo in mixed juice. same periold:
To check the accuracy of this assumption, one can Per cent in
compare the quantity of bagacillo found by measur- Tons Clarified Juice
ing the rate of, addition (65 lb. per minute), with that Brix . . . . . . . . . 848 3.65
found by analysing the filter cake for dry bagacillo Sucrose . . . . . . . . . 398 2.01
per cent (8.66%). The latter figure should be greater
than the former because of the bagacillo already con- Discussion of Results
tained in the mixed juice. At the outset of this investigation it was realised
In fact, the difference between the two figures that it would be difficult to draw concrete conclusions
should equal the quantity of bagacillo in mixed juice from the results obtained. However, it was felt that
(0.06% on juice). an investigation would at least reveal the short-
The following figures will illustrate the above check comings in more detail and thus provide sufficient
on the accuracy: information to enable us to orientate our way of
thinking towards establishing a more reliable system
Darnall-August 1968: for tracing losses in the factory.
Tons mixed juice= 184 776 The results of the balances are difficult to evaluate
Tons filter cake= 7 866 and lead one to a choice of two conclusions:
By Analysis (a) The solids balance is as inaccurate as the
simple brix balance as applied in all the
Tons dry bagacillo in filter= Hulett mills.
8.66
- x 7 866 = 681 tons (b) The solids balance, despite being more
100 accurate analytically (because it is based on
Tons d,ry bagacillo in mixed juice= dry substance), has nevertheless given answers
0.06 which are, for all practical purposes, the same
- x 184776 = I l l tons as the brix balance. This, therefore, seems to
100 indicate that both methods have the same
degree of reliability.
By difference tons bagacillo added ... = 570 tons Due to the following reasons, the former conclu-
sion seems the more likely one:
1. It was mentioned earlier that the determination
By Measuring the Rate of Addition of suspended solids per cent mixed juice was
65 lb. per minute at 56.0% moisture known to be inaccurate due to the sampling
is equivalent to . . . . . . . . . . . . . . . 590 tons procedure employed.
Difference . . . . . . . . . . . . . . . . . . 20 tons 2. The weight of filter cake is more of an estimate
A similar calculation for September gives a dif- than a true weight and can influence the final
ference of 16 tons. results considerably.
We can therefore accept the analisis figures to be
sufficiently accurate for the purposes of a monthly 3. The application of "factors" to the Mill Report
solids balance. Sheets can only lead to reliable results if they
are determined over a long period. That is, they
Calculation of the Solids Balance should result from routine analyses so as to
obtain weighted averages for the period under
The complete method of calculation is given in consideration.
Annexure I. The results of this typical example are
as follows: 4. The brix balance as applied in the Hulett mills
assumes no loss of sucrose during clarification,
Solids In Solids Out other than in filter cake:
Mixed Juice: Then:
Suspended Filter Cake . 1 862 Tons sucrose in clear juice=tons sucrose in
Solids 684 Molasses ... 4 209 mixed juice - tons sucrose in filter cake.
Dry Substance 23 343 Sugar ... 17 883 Tons brix in clear juice=
chemicals:
Lime . . . . . . 95 tons sucrose in clear juice
x 100
Mono-calcium
Phosphate 42 purity of clear juice
Separan ... 1 These assumptions clearly give rise to the fol-
Bagacillo . . . . . . 570 lowing inaccuracies:
(a) Losses of sucrose do occur during clarifica-
Total (Tons) 24 735 tion and the magnitude of these losses can
differ widely from factory to factory.
44 Proceedings of Tlze South African Sugar Technologists' Association -June 1970

(b) It is known that the presence of large quan- Acknowledgements

tities of insoluble solids cause an error in I wish to place on record my thanks to the Direc-
the determination of sucrose in mixed tors of Hulett's S.A. Sugar Mills and Estates Limited
juice, and this error is carried forward into for permission to publish process records and other
the brix balance. data used in this paper.
(c) An apparent purity of clear juice (i.e. p d / I am particularly indebted to Mr. J. B. Alexander,
brix ratio) is used to calculate the tons Research and Develolpment Consultant, and to Mr.
of brix in clear juice. R P. Jennings, Chief Research Chemist, for their
It would appear that our measuring tools have be- valued counsel on many occasions.
come too crude for this investigation and that the
increase in factory efficiency has gone beyond the References
point of reliable information. Although all these references are not used in this abridged
report, they are given for the convenience of those wishing
to make use of them.
1. Laboratory Manual for South African Sugar Factories,
Conclusions S.A.S.T.A., 1962.
Although a step in the right direction has been 2. Clayton, J. L. "Brix and Purity of Raw Juices." Proc.
taken by using dry substance instead of specific 26th Conference, Q.S.S.C.T., 1959, p. 131.
gravities and refractive indices, the solids balance, as 3. Deicke, R. "Preliminary Investigations into the Tnfluence
presented in this report, requires refinement. of Suspended Solids on the Brix of First Expressed
Juice." Proc. 25th Conference, Q.S.S.C.T., 1958, p. 159.
Empangeni Mill now lends itself to a more reliable
4. Laboratory Manual for Queensland Sugar Mills, 4th
investigation due to the following: Edition, Bur Sug. Expt. Stns., 1961, p. 57.
1 . A sampling device has been designed and 5. Clayton, J. L. "Refractometer Brix for Factory Con-
installed at Empangeni to give a representative trol." Proc. 29th Conference, Q.S.S.C.T., 1962, u. 235.
sample of mixed juice for the determination of 6. Jensen; E. D. "A Comparison of Methods for the De-
suspended solids. termination of Dissolved Solids." Proc. 29th Conference,
2. A rifler apparatus has been designed and tested Q.S.S.C.T., 1962, p. 227.
to enable sub-sampling of the mixed juice 7. Graham, W. S., and MacGillivray. A. W. "The Deter-
mination of Brix." Proc. 43rd Congress, S.A.S.T.A.,
sample. 1969.
3. Empangeni has installed a Servo-Balans to 8. Honig, P., and Sattler, L. "Determination of Water in
weigh syrup. Molasses." J.S.J., Vol. LVIII, 1956, p. 92.
A point to bear in mind is the possible introduc- 9. Douwes Dekker, K. ''Tl~e Composition of South
tion of the new cane payment system in South Africa. African Final Molasses. Proc. 31st Congress,
This system would make the water scales available S.A.S.T.A., 1957, p. 92.
for syrup weighing and once the syrup sampling 10. Zerban, F. W. "Determination of Moisture in Sugar
problems have been solved, one would know what Solutions with the Karl Fischer Reagent." Ind. Eng.
Chem., Anal. Ed., Feb. 1946, p. 138.
quantity of material is entering the boiling house. 11. Oosthuizen, D. M. "The Determination of Water in
If any further work is envisaged on this project, Raw Sugar Using the Karl Fischer Method." Proc. 43rd
the following recommendations are tentatively put Congress, S.A.S.T.A. 1969.
forward: 12. Relf, E. T. ,"Determination of Dry Substance in
1. All analyses must be done on a routine basis Molasses and Massecuites." Proc. 25th Conference,
so as to obtain weighted averages. Q.S.S.C.T., 1958, p. 189.
13. Smart, S. G., and Mitchell, J. J. I.C.U.M.S.A. Report,
2. Filter cake must be weighed. I.S.J. 52, p. 244.
3. Correct sampling procedures must be adopted. 14. Vandenberg, H. "The Application of a 'Dirt-Correction'
4. Syrup scales must be installed. Factor when using Juice Scales." Proc. 20th Conference,
5. At present, it is doubtful whether any existing Q.S.S.C.T., 1953, p. 95.
laboratory staff could handle the extra work 15. D o ~ ~ w eDekker,
s K., and Laubscber, P. J. "The Effect
of Removing Bagacillo from Mixed Juice on Sugar
load entailed and one would therefore require , Manufacture." Proc. 25th Congress, S.A.S.T.A., 1951,
extra staff. p. 57.
In view of these last-mentioned factors, an analogy 16. Anon. ''Comparison of the Results obtained on
to Lost Absolute Juice % Fibre is being sought 'Muddy' juice samples by the Official Jackson and Gillis
for application in the boiling house. That is, one method No. 4 for the determination of sucrose in Mixed
Juice with the Normal Weight Method." Annual Re-
must attempt to express the sucrose losses in terms port, Sugar Milling Research Institute (S.A.), 1955,
of a substance which remains a constant quantity p. 15.
(other than that which is actually lost) such as fibre 17. Smith, N. "1936 Seasonal Investigation on Hot and
in the milling tandem. At present, an investigation Cold Liming." Bur. Sug. Expt. Stns. Tech. Comm., No.
is being conducted by Hulett's Research and Develop- 3, 1937.
ment Department into the feasibility of a chloride '18. Wright, P. G. "Removal of Suspended Solids from
ion balance. Mixed Juice." Proc. 26th Conference, Q.S.S.C.T., 1959,
p. 177.
If this proves successful, it will be possible to ex-
press sucrose as a ratio to chloride in all factory 19. Perk, Charles G. M. "Forty-first Annual Summary of
Laboratory Results." Proc. 40th Congress, S.A.S.T.A..
materials thus enabling one to pinpoint any losses, 1966.
Proceedings of The South African Sugar Technologists' Association -June 1970

ANNEXURE I
CALCULATION OF THE SOLIDS BALANCE

The following factors are applied in the calculation :

1. The difference between factory recorded spindle brix and dry substance % filtered mixed juice '= 13.08 - 12.79 = 0.29
2. Suspended Solids % Mixed Juice . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 0.37
3. Dry Bagacillo % Filter Cake . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8.66
4. Dry Bagacillo % Mixed Juice . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 0.06
Although many balances were calculated, the following is given as a typical example.

BALANCE FOR DARNALL - AUGUST 1968

The complete method of calculation is given in this example:
(a) Factory figures from Monthly Report Sheet
Tons Mixed Juice weighed . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Tonsfiltercake . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Tons lime . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Tons ~ono-caidium'Phosphate . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Tons Separan . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Tons Sugar made and estimated . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Tons Molasses made and estimated . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Mixed juice spindle brix . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Moisture % sugar . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Karl Fischer solids % Molasses . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Moisture % filter cake . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
(b) Estimated Figures
Dry substance % filteied mixed juice = 12.97 - 0.29 . . . . . . . . . . . . . . . . . . . . . . . . . . . .
(c) Calculation of Balance
0.37
(1) Tons suspended solids in mixed juice = -x 184 776 . . . . . . . . . . . . . . . . . . . . . . . . -
100
(2) Tons actual mixed juice = tons mixed jujce weighed - tons of suspended solids = 184 776 - 684 ...... - --
12.68
(3) Tons dry substance = estimated dry substance % mixed juice x (2) = -x 184 092 -
. . . . . . . . . . -
100
(4) + +
Tons total solids in mixed juice = tons dry substance tons suspended solids = (3) (1) = 23 343 684 + =
8.66
(5) Tons dry bagacillo in filter cake = dry bagacillo % filter cake x tons filter cake =
100
-
x 7 866 ...... - -

0.06
(6) Tons dry bagacillo in mixed j~iice= dry bagacillo % mixed juice x tons mixed juice weighed = -x 184 776 =
100

(8) Tons total solids into clarifiers = total solids in mixed juice + tons lime f tons mono-calcium phosphate + tons-
+ + +1
Separan = 24 027 95 42 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . -
76.33
(9) Tons solids in filter cake = tons cake - tons moisture in cake 7 866 - -x 7 866
= . . . . . . . . . . -
100
(LO) Tons solids in mud leaving the clarifiers = Tons solids in filter cake- tons dry bagacillo added to mud = 1 862 - 570 ='

(11) Tons solids in clear juice tons solids into clarifiers (8) - tons solids in mud (10) = 24 165 - 1 292
= .... -
80.47
(12) Tons solids in Molasses made and estimated = -x 5 231 . . . . . . . . . . . . . . . . . . . . -
100
0.423
(13) Tons solids in sugar made and estimated = 17 959 - (-- x 17 959) . . . . . . . . . . . . . . . . . . = 17 883
100
(14) Solids In Solids Out
Mixed Juice : Suspended solids 684 Filter Cake 1 862
Dry Substance 23 343 Molasses 4 209
Chemicals : Lime 95 Sugar 17 883
M.C.P. 42
Separan 1
Bagacillo * 570
,
Total 24 735 23 954
Undetermined Solids lost = 781 tons.

The factory reported the following losses for this period

Tons
Brix 848
Sucrose 398
For discussion on this Paper, see page 50.