CHAPTER I

ASSESSMENT OF THE OIL ENTERING THE MILL

Processing control in the oil mill can be

carried out in two v/ays:-
a) By weighing the amount of oil produced and
calculating the ratio of that weight, to the
' • • sum of the weight of oil produced and the
weight of oil lost in the waste products.
This ratio (always lower than'l) will be
' termed "efficiency".
b) By weighing the amount of oil produced and
calculating the ratio of that weight to the
weight of oil received at the factory as
raw material and as assessed by analysis
(bunch or stripped fruit). This ratio will
be termed "'recovery" .
The first method is accurate and its use is
recommended in the case of a pilot plant. This is
because all the waste products of processing can be
•.weighed and are homogeneous enough to be sampled
•easily.
In industrial practice, this method possesses
the serious disadvantage of ignoring the accidental
losses resulting for instance from faulty plant
operation-
The second method, i.e. that based on the
analytical determination of the oil content of the
raw material, could laad to useful pointers for
.processing control but its implementation is diffi-
cult because of the pronounced heterogeneity of the
raw material.
A. Efficiency
The efficiency is given by the following
formula : -

in which
H = Weight of oil produced.
HS = Oil lost in steriliser condensate.
HR = Oil lost in bunch refuse.
Hp = Oil lost in fibre. .
HN = Oil lost on nuts.
HD = Oil lost in the clarification of
crude oil.
The coefficients a(/3 and ^represent the
efficiency of the extraction proper (»t } and that
of the purification of the extracted oil (/^).
The coefficient is affected by another one ( %*)
which takes into account the practical industrial
efficiency compared with the limit obtainable in
the laboratory.
Determination of the efficiency
The method entails the weighing or at least the
accurate estimation of the weight of waste products.
After consideration of a number of ways to obtain
these weights, the following system was proposed.
It consists in:-
a) Assuming a fixed loss in steriliser condensate.
The loss is governed solely ay the sterilisa-
tion procedure and the form ,\n which the crop
is processed (detached fruits "cr bunches).
Generally, once sterilisation procedure is
finalised, no further modification is intro-
duced. The amount of oil finding its way into
the condensate is therefore constant. The form
in which the crop is processed is practically
constant for a given factory except in the case
of/plant receiving progressively less detached
fruit and more and more bunches.
£a
The loss in steriliser condensate can therefore oe
considered as a fixed quantity -which is evaluated
once and for all subject to verification if a
modification occurs either in the sterilisation
cycle or in the fruit, particularly in the case
of marked changes in the degree of ripeness.
b) Assessing the weight of bunch refuse and titrating
the oil content therein. The percentage ratio of
bunch refuse to bunches shows consistency either
in the course of time or in relation to the type
of bunches. It is therefore generally sufficient
to weigh the bunch refuse during a certain period
of time and to carry out a few spot checks during
the year to determine this ratio with sufficient
accuracy.
c) Determining the weight of fibre from the ratio of
fibre to cake and from the weight of nuts. The
ratio of fibre or nut to cake is essentially
governed by the type of fruit which is processed.
It varies between wide limits (vide Chapter VII).
In the case of factories dealing with bunches of
one variety only, the variation is not important
and the ratio may be considered as constant.
On most estates, however, palms of various origins
and ages are encountered. In addition to differ-
ences due to the type of fruit, the stripping
procedure also affects the cake composition as a
result of the variable proportion of impurities
-mixed with the fruit at that stage.
The determination" of the cake composition is there-
fore indispensable. That operation should be done
over the whole processing time either by manual
sorting of large samples or by means of a small
scraping knives depulper. The drying of fibre
does not affect the final result since oil
content is determined on dry basis. It is how-
ever important to draw the sample of fibre during
the datsrmination of the cake composition.
 -• 3 -

d) Weighing the nuts. That operation is performed

in a tilting nut -weighing device after separation
from the fibre. The number of tilts 'of the
apparatus is recorded automatically. • This makes
it possible to know, at the end. of processing,
the"weight of nuts derived from a known weight
of bunches. Current tilting scales ippears to
provide a sufficiently high accuracy.
The oil loss is calculated directly from that
weight and the oil. content of nuts.
e) Weighing the discharged effluents. As for the
preceding weighing, the weight of effluents or
sludge originating from the clarification of
crude oil is determined in a tilting weighing
scale. The oil. .loss is calculated directly
from that weight and the oil content of sludge.
As fast methods of analysis are available (see
relevant paragraph), the oil losses can be
checked daily by this system. The quantity
of oil produced can also be accurately
assessed daily by means of a tilting scale,
the recommended method permits therefore to
calculate daily the mill efficiency.
An outline of ^tilting weighing scale is given
on the following page (figures 1 & 2).
B. Recovery
Its determination involves:-
i) The weighing of the raw material
entering the factory.
ii) The determination of the average
oil content thereof.
iii) The weighing or the volume assess-
ment of the oil produced.
The weight of the raw material can be known
with less than 0.1$ error but the determination of
the average oil content may be seriously erroneous
"since it is carried out on a small aliquot. The
determination of the oil content of the raw material
involves sampling and analysis operations.
\
In this paragraph only sampling will be con-
sidered. The analyses themselves present 'little
difficulties. .They can be carried out using con-
ventional methods. An attempt has however been made
to evolve analytical procedures particularly suitable
for application to the palm fruits or the waste pro-
ducts of processing.
 Tilting weighing
device (with
feed manifold)

Tilting weighing
device for palm
oil

Figure 2
Sampling #
In the case of processing of bunches, the amount
of oil entering the factory can be assessed from the
weight of frash bunches or from the weight of sterilised
fruits (as collected from, the strippar).

a) StgrilisQd fruit
The second method is, "a priori", more attrac-
tive because of the higher homogeneity of the steri-
lised fruit compared with that of the bunches. It
iapliss however the weighing of a material not
particularly suited for that operation. It was
triad in the pilot plant and it has been possible to
establish that with homogeneous batches of fruit
without impurities the oil content is determined
with an error of less than 1% if the sample is made
up of the following number of fruits:-
13 x P material — 2,500 fruits
T x T material — 4,000 fruits
P.G-. material -- 5,700 fruits
Mixed material — 5,700 fruits
On account cf the pronounced variability of
the material, the determination of the percentage
of trash (x) of the fruits requires the analysis
of 8,500 x 500 g samples to obtain an accuracy of
lyis or of 340 x 500 g for an accuracy of 57- Hew-
ever, by using 5 kg samples, the percentage of
trash in sterilised fruits can be evaluated with
a 5% error if 52 samples of fruit are examined.
The variations in oil content of the trash
are smaller than those of the trash content of
fruit. The determination, with an accuracy of
10$, of the oil content of trash requires the
soxhlet extraction cf 27 samples (the average oil
content is 20% on dry matter in trash).
Although the above remarks still permit to
envisage the determination cf the cil content of
sterilised fruit, the method vvas definitely
discarded when it was observed that after stripp-
ing the sterilised fruit sustains an important
loss of weight. (vide Chapter II, Effect of
Sterilisation).
After 10 minutes storage the apparent
, increase in oil content of the fruit is approxi-
mately 2%. It can reach 4 to 5^ after one hour.
Variations of such magnitude constitute
too serious an obstacle to the accurate deter-
mination of the oil content of sterilised fruit.
# Unless mentioned to the contrary the number
of bunches or fruit to be sampled is
given for P = 0.05.
(x) Vide appendix 2.
b) Fresh bunches
The analysis of fresh bunches involves:-
- the determination of the ratio of
fruit to bunch.
- the determination of the ratio of
oil to fruit.
The ratio of fruit to bunch
This is generally established by chopping
the spikelets from the bunch then cutting the
fruit from its socket. The method is extremely
time consuming and can only be used if a limited
number of determinations are to be carried out.
A worker can only deal with a few bunches per
day.
To speed up the work, the bunch or the
spikelets are sometime allowed to remain in a
heap for a day or two. The removal of the fruit
is then carried out. The resulting desiccation
of the fruit and the stalk is detrimental to the
accuracy of the analysis (v:ie the analysis of
fruit).
In order to achieve easy fruit stripping
without the need for maturation storage of
bunches or spikelets, the use of steam at'
atmospheric pressure was tried. It has been
observed that under these conditions, a period
of 20 minutes steaming practically does not
affect The weight of the fruits or spikelets
and releases all the fruits from the sockets.
A trained worker using suitable equipment can
in this manner strip 20 bunches per hour. It
has been established that it is necessary to
strip 1,200 bunches drawn from homogeneous
batches to achieve an accuracy of 1% in the
determination of the ratio of fruit to bunch.
- Ratio of oil to fruit
The determination of this ratio requires
two operations: the assessment of the per-
centage of pulp and the oil content of that
pulp.
- Single bunch analysis
The object of the experimental work was
to establish what weight or -what number X5f
fruits should be drawn from a bunch to obtain
an estimate of the proportion of pulp with a
given accuracy.
 The statistical analysis of 36 samples of
fruits per bunch (1/3 external fruits, 1/3 middle
fruits, 1/3 inner fruits) drawn from a number of
Dura .and Tartars "bunches of various sizas showed
that approximately 3 to 6% of the weight Qf fruits
must "be taken in order to attain a relative error
of 1$ on the determination of the ratio of pulp
to fruit.
Table 1 gives the relevant data for Dura
"bunche s' -
Table 1
Number of fruits to "be drawn from a "bunch to
determine the percentage ratio of pericarp "to
fruit 7/ith an accuracy of 1$

i'Bunch" i Bunch ! Bunch Bunch j Bunch Average

i 1 ! 2 .! J_ 4 j 5 i 6
'
Number of fruits 75 106 ; 166 200 i 90 127
$ of the total weight i i .1
of fruits 5 « o$ i 5 • 6$ ; 6. 4$ 6.6$ I 5.1$ i 5.9$

In the case of fruit from Tenera bunches issued

from T x T crossings, the ratio of pulp to fruit can
be determined with an accuracy of 1% using samples of
half the size indicated above.
Vanderweyen and Alias (l) have established that
in a sample consisting of 50 fruits of average weigh"?,
the percentage of pulp is ascertained with an accur-
acy of 2.5$ to 4.5^. The determination of pulp on
fruits of average weight is perfectly justified for
single bunch analyses performed in a research station.
It is impractical for industrial analyses carried out
in control laboratories.
The determination of the moisture content of
pulp requires a larger number of fruits because of
the higher variability of that characteristic. A
few results are given in Table 2.
Table 2
Number of fruits to be drawn from a bunch to
determine the percentage ratio of water to
pericarp with an accuracyof 1$

onchi Bunch iBuncn iauncn Average ;

Number of fruits 350 ; 300 11,069il,342j 150 638

% of the total weight | j
of fruit i26.9$ il5-S$|41.1$;43. 3$ ; 8.6$ 27.2$
 In the case of Tenera bunches, the sample must
also amount to about 25$ of the total fruit to achieve
an accuracy of 1%.
Finally, to determine the cil c'ontent of fruit
of a bunch, whether Dura or Tenera, with !,"-> accuracy,
'it is necessary to analyse from 15 "co ICOfo of its
fruit;. That kind accuracy is seldom required. It
shows however the considerable heterogeneity of the
bunches. If a 5% relative error is deemed acceptable,
it is necessary to take approximately 2.3^ of the fruit.
That degree of accuracy is sufficient in certain cas'as
but it is of no interest if the object is to assess the
extraction efficiency which must have an accuracy of
approximately ~L%.
As a palliative against the hsterogensity of bunches
the possibility of building up samples from spikelets
instead of from fruits was envisaged. The idea being
that a spikelet of 5 fruits might be more representa-
tive of the bunch than just 5 fruits -oaken at random
from it. The analysis carried out on a number of
bunches showed that for an equal number of fruits,
the error is larger for the spikelets than for the
fruits.
- Analysis of a batch of bunches
•j
Practically in the industry the need for single
bunch analysis does not arise. The problem is to
determine the oil content of a batch of bunches.
To that end, a study has been made of the number
of fruits that it is necessary to draw from each
bunch in order to arrive at a predetermined degree of
accuracy.
.The number of fruits to be taken depends on the
manner in v/hich the sample is drawn. For instance,
10 fruits drawn at random lead to the same relative
error as 3 selected respectively from outer, middle
and inner fruits.
Moreover, increasing the sample to 9 and 24
fruits per bunch, still distributed as 1/3 each from
outer, middle and inner layers respectively does not
decrease the relative error appreciably. The fact
that 10 fruits drawn at random -are just as represen-
tative of a bunch as 3 selected from outer, middle
and inner layers respectively was established only
when the research was about to be completed. Con-
sequently, all experiments aimed at ascertaining the
number of bunches to be sampled from a batch were
carried out with the,.-3 fruits per bunch sampling
technique.
 The number of bunches required to achieve a
predetermined degree of accuracy is given by the
general formula:-
N = A () 2

where
N = number of bunches
A = a constant which has been
determined in each case .

x = The required relative error

(in percentage)
Table 3 gives the value of "A" as determined
for the various components of the Dura and Tenera
bunches subjected to test.
Table 3
Value of "A"

Particulars Dura i Tenera

Bun c he s ; Bunches
Ratio of pericarp to fruit ; 26.5 i 3-6
Ratio of moisture to wet
pericarp 13.0 ; 9-1
Ratio of oil to fruit 28.2 12.3
Ratio of fruit to bunch 3-1 : 12.6

The following remarks may be made regarding the

data of the above table:-
The heterogeneity of the fruit is more
pronounced for Dura than for Tenera.
On the contrary, Dura bunches are more
homogeneous than Tenera's in respect
of the fruit to bunch ratio.
c) Loose fruit
The sampling has shown that the analysis of approxi-
mately 1% of the fruits in batches of ?.(J. crop leads to
a relative error of 1$ on the oil content. In practice,
no more than 0.01$ if the fruit is subjected -to analysis:
Factories dealing with loose fruit assess the oil_,_con-
tent on fruit with" an error of 5 to 1C$ (20$ oil - 1%
under suitable conditions). The above results were
obtained on batches of fruit derived from completely
stripped bunches: that is containing the whole of the
inner fruits, parthenocarpic fruits, etc... Normally,
the batches of fruit bought from the local farmers are
more homogeneous because the inner fruits are not
included.
 - 10 -

The method, of sampling described above should

therefore lead to a more accurate determination
of the oil to fruit ratio. The impurities con-
tent of loose fruit subjected to test was fairly
constant, approximately '!.% (sand, calyx leaves
etc....)- The sampling of estate loose fruit
was not studied.
2. ANALYTICAL PROCEDURE ?CR THE PHUIT AM) THE 7/ASTE
PRODUCTS OF PROCESSING "
A. Desiccation of the fruit and of the fresh
bunches
This point was studied with a view to
establishing if the stripping made possible by
ageing of the bunches modified the ratio of
fruits to bunch. A few results are given in
Table 4 which shows the loss in weight of freshly
stripped fruit (cut from the spikelets) and of
bunches as a function of time.
Table 4
Effect of time of storage on the loss of weight
(in percentage units)

. — _
Hours of Prxiits Bunches
storage Dura Tenera Dura Tsnera
i " 1
0 0 0 0 c
1 0.1 0.2 0.2
1 0.7 I - ! 0.9 -
15 - 1.7 - 1.7
20 -2 ^
C • JL 1.9 2.1
25 2.1 2.5 2.3 2.4
68 4.8 5.7 6.1 6.0
90 6.0 7.1 7.5

The desiccation of bunches appears faster than

that of fruits. Prom a large number of tests
carried out over several months by Vanderweyen &
Alias (2) it appears that the loss in weight of
Tenera and Dura loose fruits amounts respectively
to 3-45$ and 2.79$ after 1 day and to 6.23?' and
5.08$ after two days. The loss recorded for" the
Tenera material is statistically higher than that
for Dura's (P = 0'. 01) over a period of 4 days, but
the relative rate of evaporation of the pulp
moisture is higher for Dura.
 The loss in weight of bunches amounts to 1.7
to djf after 1 day and 3-1 - 3.4$ after two days.
Vanderweyen i Alias ascribe the loss of -.'/eight
of the bunch chiefly to the desiccation of the stall*
and that of the fruit tc loss of moisture through the
area of attachment to the bunch and hot through the
exocarp. They suggest to allow for a 1 to 2> correc-
tion for bunches stored for 1 or 2 days before analysis
or, bettar still, if the weight of bunches as harvested
ia known to add the loss of weight to the weight of the
bunch stalk.
This can be arranged on research stations but not
on commercial estates where it is not possible either
to weigh the bunches immediately after harvesting or
even to know with any degree of certainty how many
hours have elapsed between harvesting and analysis or
processing.
3. Desiccation of the pulp
In the standard analysis of the fruit of the
oil palm, the fresh pulp is peeled off and weighed.
Does this depulping entail a certain amount of
desiccation liable to induce an increase in oil
content? The loss of weight of fresh pulp immediately
after depulping was assessed. Depulping was performed-
as quickly as possible. The time elapsing between the
beginning of the operation and the first weighing was
•2 minutes. Figure 3 gives the loss of weight against
time.
Figure 3
Effect of time on the desiccation o: :ulp
Loss of
Time weight (f)

0.70 2 0.02
'' I Loss of weight 5 0.05
0.50 (in #)
8 0.09
10 0.12
0.50
s 15 0.19
0.40 24 0.29
l
30 0.36
0.30
41 0.48
0.20 51 0.60
61 0.71
0.10
75 0.86
0
10 20 30 40 50 60 70
Time in minutes
 - 12 -

It may be observed, that up to 75 ninutes, the

desiccation is dirsctly proportional to time and
•that a normal time of de pulping, 15 to 20 mjnutes,
does not practically affect the oil content. The
depulping operation entails very low losses of
weight if it is carried out with suitable care.
It appears from ten of thousands of analyses that
the difference between the initial weight and the
sum of the weights of nuts and pulp averages 0.4$.
About 1 to 2% of the analyses show a depulping loss
of more than ifo (for a sample of 100 to 150 g of
fruit). This is in fact a loss of matter rather
than the effect of desiccation. As a general rule
the results of such analyses are discarded in control
laboratories.
Recently, an investigation into the determina-
tion of the oil content of pulp .(3»4) has shown that
the loss of weight due to depulping and pounding was
approximately 1%. It seems that 95$ of the loss is
due to pounding which is not used in the Belgian
Congo. The extraction being carried out on pulp as
peeled off from the fruit.
Determination of the oil content
a) Combined determination of oil-, moisture and
non-oily solids (N.O.S.)
The standard analytical procedure for the
assessment of fatty matter requires drying of
the pulp o^ the fibre before solvent extraction.
The drying is generally carried out • in an oven
and is truly the most time consuming step of
the analysis. It usually requires 12 to 14 hours.
The possibility of substituting azeotropic dis-
. tillation to drying. ha.s been .envisaged by several
research workers but, to our knowledge, no
apparatus suitable for the fruit of the palm or
the waste products of processing has yet been
evolved.
The proposed apparatus is a Kumagawa
extractor (high temperature Soxhlet extraction)
in which the solvent is, before percolation,
separated from the water by a separator of the
Dean & Stark or the Sundin type depending on
whether the solvent used is heavier or lighter
water.
Moreover, the method of oil extraction,
from the effluents was evolved by a firm belong
ing to the Cooperative (5). The description
and the specifications of the equipment are
given in Appendix.
The method can be used for the combined
determination of moisture, oil and IT. 0.5. of
the pulp of the fruit , the- cake , the bunch
refuse .and the mill effluents.
 - 13 -

The time required 'for" the analysis varies

with the solvent "used. It is shorter for toluene
.(or xylsne) than with -richlcrethylene. The
latter offers the advantage of being uninflammable.
The completa analysis with toluene as solvent,
that is the weighings, the extraction and drying
talces 2-3C hours for pulp and 1.3C hours for fibre
on buiich refuse.
te) Determination of oil by measurement of the refrac-
tive index
In the' preceding method, the solvent is usually
distilled off and the residue is dried to constant
weight. In order to eliminate the drying step, the
possibility has been investigated of titrating oil
directly in the solvent through measurement of the
refractive index'of the solution.
*
Mention is made in technical publications
(6,7) of several procedures for the determination .
Of oil in various organic media based on the varia-
tions of the refractive index according tc oil-
content.
The following factors which affect the
refractive index were studied:-
i) The oil concentration,
ii) The temperature of the solution.
iii) The ?.F.A. content.
iv) The nature of the oil.
The effect of moisture content needs not be
considered since the solution of oil in xyiene is
always dehydrated by azeotrcpic distillation prior'
to testing.
The time of ebullition of the solution of oil •
in xyiene does not affect the refractive index.
i) Effect of concentration and temperature
The results regarding the effect of .
concentration and temperature plotted on
Figure 4 were obtained for a 5-76>: F.F.A.'-
oil.
The refractive index of the solution^
: of oil in xyiene fellows a linear relation-
ship until 6C5t. After that point, the line
- deflects slightly into a curve. The follow-
ing equation gives the concentration of oil
for 100 ml of solution according to the
refractive index.
 - 14 -

C = NY - at - Ns
A - bt

where

% = refractive index of
xylene at OOC.
N
s =
refractive index of the
solution.
t = temperature.
A, a and b = constants which have
been established.

Figure 4

Refractive index

\ r\ x x x
\\\\

Oil
10 20 30 40 50 60 70 80 90 100
Variation of the refractive index according
to the oil concentration and the temperature
of the solution
 - 15 -

ii) Effect of acidity

This has "been studied on oil acidified
by addition of palm oil fatty acids. The
refractive index of artificially acidified
oil is not, of course, identical to that of
a spontaneously acidified oil which, contains
a higher proportion of mono and diglycerides.
Although the refractive index of fatty acids
was found, .to "be. 1.439-8 at 600Q .against
1.4585 at 40°C for most oils, the difference
in refractive index between oil of normal
acidity (between 1 and 6$) is of the same
magnitude as the testing error, that is
0.0002. It may be ignored. The variation
in refractive index resulting from variations
in P.P.A. is negligible if the average index
is determined daily on the oil produced at
the mill.
iii) Effect of the nature of the oil
.Twelve samples.of oil extracted from
fruits of various origin and from different
parts of the bunch were found to have a
nD40°C index ranging from 1.4583 to 1.4590
(average 1.4584).
In that case again, the recorded
differences may be assimilated to the
testing error. It may therefore be postu-
lated that the refractive index is indepen-
dent of the nature and the origin of the
oil. The oil concentration can be deter-
mined with an accuracy of ± 1.5$', compared
with the solvent evaporation method. The
procedure avoids the evaporation of the
solvent after extraction and the frequent
ovsrhaating «suiting Z The determination
proper is very fast but aelicate volume
adjustments are required. ^therefrom.
The method was successfully used at
Mongana but it is not recommended for a
laboratory where supervision of personnel
is not permanent.' Checking .is easy. It
simply involves the evaporation of the
solvent from the 100. ml of solution from
which one drop has been drawn for the
refractometer reading. Zylene was
selected in prefer.ence to benzene or
toluene on account of its lower vapour
pressure.
 - 16 -

The table "below gives a few comparative

results obtained by weighing of the oil and by
refractometry.

Table 4

Origin \ Oil as : Oil

of | determined by : determine.d .
i the oil • refract cine try ;• by weighing i
! Fruits 21.60 g 21.20 g I
i Fruits 26.60 g 25-20 g
!

j Fruits 30.40 g 29-60 g i

j Fruits 20.50 g ; 19-75 g '.
!
; Fruits j
28.80 g 27-30 g ;
1 Fruits 42.10 g 41.20 g |
| Fruits 18.20 g 19-10 g 1
| Fruits 22.90 g 23-40 g -
i ;
; Crude oil : 21-30 g 21.60 g
; Crude oil i 30.30 g 29OO g
• Fibre i 32.40 g 32.55 g :
: Fibre i 31-60 g 31-30 g !
;
i1 Fruits •
25.40 g i 25.30 g
; Fruits 26.80 g 26.70 g

Assessment of the oil content of p through its

moisture content
It is known that F.M. Dyke of the Research Depart-
ment of HCB (Huile-rie-s du Congo 3elge ) established more
than 30 years ago tha~ in fresh fruit the sum of oil
and moisture of pulp is practically constant and equal
to 84$.
It had been observed that in the case of loose
fruit derived from bunches stored long enough to make
manual stripping possible, the total oil + moisture
fell -to 82.5 (± 0.25) except for fruit with very low
pulp content (30$) for which the sum remained below
&0$. The oil content was therefore described by the
following equation:-

Y = 82.5 - X
where
Y is the oil content
X _is the moisture content of pulp.
 - 17 -

As a result of this observation a table was

prepared giving the sum of oil and moisture content
of fresh pulp over a wide range of moisture content.
The tabl-e- could.,be used. to determine the oil content
of fresh pulp in the following manner: The moisture
^en-tent as determined by analysis, was deducted .from
the corresponding sum of oil and moisture. The method
has been adopted by all important mills of the Belgian

Vanderweyen and Alias (3) confirmed the correla-

tion. From 136 'analyses they derived the following
regression equation: -
y = 37.33 - 1.08 X
from which they computed a table showing the oil
content of pulp according to moisture content (see
Table 6). These authors do not specify the time
elapsed between harvesting and analysis. It must be
assumed however that the time is short sines the work
was. carried out. on a research station. There was a
discrepancy of a few percentage units between the
results of Dyke • and those of Vanderweyen, which we
assumed originated from a more important desiccation
of the fruit in the case of Dyke's work than in the
case of
We wanted to check whether the latter ' s results
were applicable' to fruit stripped by the local farmer,
as. it is delivered to mills dealing with loose fruit,
•that -is- without information or control over the age of _
the fruit.
To that end, 150 analyses wers carried out on-
loose' frui't "delivered, on the average, 4. days after
harvesting of the bunches. The results confirm the
INEAC data. The slope coefficient is smaller than 3#-
Table 5 gives the average result of 10 analyses listed
in order of increasing moisture content.
 - 18 -

Table 5
Analysis of fresh pericarp
Fruits plucked routinely "by the small holders

| % Moisture/7» Oil/fresh % Oil + fo Oil/dry % Oil/fresh

fresh pulp pulp moisture pulp pulp
(calculated)
24.47 60.90 85-37 80.60 60.81
26.86 58.95 85.81 80.12 58.30
30.00 55.08 85.08 78.37 55/00
31-57 52.08 83.60 76.06 53.55
32.58 52.18 84.76 77.40 52.29
33-20 51.35 84.55 76.87 51.64
37-71 51.32 85.03 77.46 51.11
34-57 49-52 84.09 75.69 50.20
35.62 50.33 85.95 78.15 49.10
36.23 48.50 84.73 76.04 48.46
37.24 47.39 84.63 75.55 47.40
38.08 47.83 85-91 77.12 46.52
40.50 44.36 84.86 74-53 43-98
| . 46.87 37.67 84.54 70.43 37.29
i
51.80 31-59 33.39 65.39 32.11
Average :
35.55 49-27 84.82 76.01 49.17

It is therefore confirmed that the IITEAC' table

can be used for loose fruit as received in processing
mills. We assume that the discrepancy recorded
between the above figures and those of Dyke is due to
the fact that desiccation is now less pronounced than
it was 30 years ago. Transport conditions have
improved considerably and the time elapsed between
harvesting and delivery to the mill certainly exceeded
«taBK the present average of 4 days (Numerous markets
were open only weekly or even fortnightly).
 - 19 -

We give below (Table 6) the table established

by Vandarweyen for the analysis of fresh pulp
together with the data collected at Mongana.
Table 6
Anal.ysis of fresh ta ult)

i Sun of ?» mois- % Oi 1 % c5il

! % Moisture/ j turs ancL oil to to ;o
fresh MulD : ^~v -
Pulp f r e sh T3UlT3 gaJt ^ mis
i Mongana INEAC Mcngana INEAC Montana ISEAC
25 83. 3 85.4 60.3 6C.4 80. 4 80.5
; 80.1 80.1
26- 85-3 85.3 59-3 59-3
27 85.2 - 85.2 58.2 56.2 79. 7 79-7
28 85-2 85.1 57.2 57-1 79.4 79-3
29 ! 85.2 85.1 56.1 56.1 79.0 79.0 1
30- | 85.1 85-1 55.1 55-0 78.1 78.6 i
31 85.0 84.9 54.0 53-9 73.3 78.1
32 85.0 S4.-8 53-0 52.6 77.9 T~> ,C '
i l . D
]
33- 84.9 84.7 51-9 51.7 77.5 77.2 .
34- 84.9 84.7 50.9 50.7 77.1 76.8
35- 84.0 34.5 49-8 49-6 76.6 76.3
36 84.8 34.5 48.8 48.5 76.2 75.8
37 84.7 84.4 47.7 .47. 4 75.7 75.2
! 38 84.7 84.3 46.7 46.3 75.3 74.7
39 84. 6 84.3 45- 6 45-3 74.7 74.3
40 84.6 34.2 44.6 44.2 74.3 73-7
41 84.5 84.1 43-5 43-1 73-7 73-1
42 84.5 84.0 42.5 42.0 73-3 72.4
43 84.4 83-9 41.4 < r\ f^
i*y , y 72.6 71.8
44 84.4 83-9 40.4 39-9 72.1 71.2
: 45 84.3 83.8 39-3 38.3 ! 71.4 70.5
46 84.3 83.7 38.3 37.7 70.9 69-8
1
47 84.3 83.6 37.3 36.3 70.4 69-0
48 84.2 83o 36.2 35-5 69. 6 68.3
49 84.1 83. 5 35-1 34.5 68.8 £.1/ . Si
O O

50 84.0 33.4 34-0 .33-4 66.0 66.8 ;

Case of the sterilised fruit

It was deemed necessary to ascertain whether a
regression equation between the oil content and the
moisture content of pulp of sterilised fruit could be
established as it was for fresh fruit.
 - 20 -

To that end 300 samples of sterilised fruit

were subjected ~o analysis. The 300 results are
given in Table 7 in 15 groups of 15 each.
Table 7
Analysis of sterilised pulp

1
ivloi attire/ Oil/ Sum of oil ; Oil/
Fresh "oul'D and moisture ; Dry T2uiD
16-58 65.44 32.02 78.44
18.4-5 64.00 82.45 J 78.48
19-64 63.27 32.91 78.73
20.70 62,67 83-37 79-03
21.45 61.68 83.13 78.53
22.68 60.55 83.23 78.32
23.85 59-98 83-32 73.77
24.81 59.78 84.59 79.51
25.95 56.84 32.79 76.76
' 27.27 56.64 33-91 77-88
28.83 55.39 84.22 77.83
29.82 54.03 33.35 | 76.04
31.75 51-90 33.55 76.04
33-01 '50.85 83.86 ! 75-91
39.42
______ , 45-22
_____ 34.62 ________
74.65
•
25.67 57.88 83-55 77.38
( o) Each ressuit is the avei'a*ce of 20 anal.ys es.

Although, a relationship may be observed, as in

the case of fresh fruit, between the oil and moisture
contents of the pulp, important deviations occur
around 25% moisture. At that level, oil on dry pulp
falls by almost 3$ for a difference of slightly more
than ~L% in the moisture content of pulp.
If we compare the results obtained on fresh and
sterilised fruit, we observe that sterilisation not
only induces a certain amount of desiccation but also
increases the oil content on dry matter.
Three hypotheses can be formulated to acco'unt
for the increase in .oil/dry matters from 76$ for
fresh pulp to 1$% for sterilised pulp. (The difference
exists for 2/3 of the results, discarding those at
both ends of the range).
 - 21 -

- The oil content en dry matter

loose fruit purchased from the local
farmer is different iron that bought
in the form of bunches. A large
proportion of the fruit with high
oil/dry basis might not be delivered
by the local farmer.
The absolute oil content increases
during sterilisation.
- Part of the N.O.S., that is part of the
extracted dry pulp is solubilised and
washed away thus increasing the ratio
of oil to dry matter.
In all probability the second hypothesis should
be rejected whilst the other two should be retained.
The local farmer, tends in fact, to keep the large
fruits,those with low moisture content and high
moisture + oil content, for his personal consumption.
The loss also may occur of loose fruits which are
relatively drier (8) and richer in oil on dry matter.
Moreover, it may be observed that a certain
amount of dry matter finds its way into the conden-
sate resulting from sterilisation. It is difficult
to ascertain which part of this loss comes from the
bunch stalk and which originates from the fruit.
CONCLUSIONS
The calculation of the percentage recovery in a
mill implies necessarily the determination of the
amount of oil entering the factory. That determina-
tion requires the analysis cf the bunches or the
loose fruit. The analysis of bunches provides the
oil content with an accuracy of 1% only if mere than
3,000 bunches of a batch of standard homogeneity are
subjected to test. That degree of uniformity pro-
bably prevails in some sectrrs, such as blocks or
divisions of one estate during a certain period of
time. The number of bunches that must be drawn
daily for the computation of the monthly recovery is
about 100. The relative error must not exceed 1%, .
failing which the calculated percentage recovery
loses its usefulness and, particularly, the faculty
of detecting any abnormal loss occurring during
processing.
.-. - <
The analysis of the sterilised fruit requires
smaller quantities of fruit to achieve the same
degree of accuracy but it has the disadvantage of
requiring the accurate weighing of the fruit
precisely when the sample for the oil determination
is being drawn. Storage of the fruit after stripp-
ing entails an apparent increase in oil content as
a result of desiccation which may reach a few percen-
tage units. Moreover, the weighing of the sterilised
fruit must be performed automatically. That operation
lias not "been investigated, but it seems according to
constructors of automatic weighing scales that serious
difficulties would be encountered.
 - 22 -

In the case of loose fruit from, palm groves,

the analysis of sterilised fruit leads to an
accuracy of 10% on the oil content. A certain
amount of sorting (removal of the parthenocarpic '
fruits,....) probably lowers -he relative error.
The sampling of sterilised fruit from plantation
crop has not been investigated.
The determination of the oil content of the
crop must-be carried out by solvent extraction to
obtain a satisfactory degree of accuracy excspt in
the case of freshly stripped fruit. The latter
to disappear as a large proportion of the crop iq-.. sow
processed in the form of bunches.
To sum up, the procedure now recommended-for
process control as well as for the determination of
the oil content of bunches relies on the assessment
with as high an. accuracy as possible of the process-
ing losses and the oil production. The percentage
recovery is no longer determined but rather the
efficiency. The procedure makes use of tilting
weighing scales for the nuts and the oil. The oil
loss on fibre is calculated from the composition of
the cake and the weight of nuts. Past analytical
procedures may be used for the determination of oil
in waste products discharged at various stages of
processing.
The losses can be computed daily whilst the
oil production can also be ascertained on a daily
basis. The efficiency can therefore be worked out
daily.
An additional advantage of the control proce-
dure described above lies in the accurate determina-
tion of the kernel content of fruit or bunches
simply through the weight of nuts and the ratio of
dry kernel to nut.
STANDARDS OF RIPENESS'
Vanderweyen (19) has written: "In an isolated
fruit, the oil content reaches maximum level at full
maturity, that is when the fruit drops from the
bunch still attached to the palm". It is known that
the ripening of the 500 to 3,000 fruits of a bunch
does not occur simultaneously.
Theoretically, to extract the maximum of oil,
only the detachable fruits should be processed. In
practice, a criterion of ripeness is selected. It
is based on the number of fruits already detached
from the bunch and not on the number that can be
detached. The criterion is sometime 5 or 10 or 20
detached fruits.
Although it is relatively easy tc determine
perhaps not the optimum degree of ripeness but one
very close to it, it is difficult to do so on a
large batch of bunches to be harvested. It should "be
made clear that we are not particularly interested
in ripeness; we are more concerned with the absolute
oil content of the bunches. The disadvantage of
harvesting at less than full ripeness is the concomitant
drop in oil content whilst harvesting at toe advanced
a degree of ripeness increases the number of _detached
fruits which may be bruised or may ferment (increase
in the acidity of the oil).
The conventional chemical analysis of a batch of
bunches does not provide useful information on the
average degree cf ripeness. In seme mills in the
?ar East, checking of ripeness is carried out by
assessing the number of detached fruits from every
bunch of a control batch. The classification adopted
in a specific case is given in the table below.
The table specifies the proportion of bunches having
reached maximum oil content. The distribution fre-
quencies were calculated for a 7 days harvesting7
cycle and for an average total ripening period of
10.6 days. This is the time elapsed between the
fall of the first fruit to the time when all fruits
have been detached.
Table 8
Proportion of the bunches having
reached maximum oil content

00 No detached fruit: green bunch 0$

0 1 to 10 detached fruits 15%
10 fruits to 1/4 of all fruits
detached 30$
2 25 to 50/c of all fruits detached 30$
3 50 tc 75$ of all fruits detached 25$
4 75 to 100$ of all fruits detached 0$
5 All fruits detached 0$
 - 24 -

A similar method was triad 'with a slight modi-

fication of the classification. Sampling was standar-
dised.
\

The following provisional classification was

adopted:-

Table 9
1 No detached fruit
2 I to 10 detached fruits
3 II to 20 detached fruits
4 21 to 40 detached fruits
5 41 to 70 detached fruits
6 71 to 100 detached fruits
7 100 to half of all fruits
detached
8 50 to 75$ of all fruits
detached
9 75 to 100% of all fruits
detached.
The checking consists in examining the first
10 bunches from every lorry load being unloaded.
Presently, the trend is to draw the optimum
frequency curve. It is governed by the harvesting
cycle, the speed of ripening, the type of handling
between harvesting and bunch checking, the age of
palms,etc... Some IFEAC specialists consider that
the palm reaches "stability" from the point of view
of speed of ripening and oil content of fruit only
after 7 years.
The specifications set up for the optimum curve
are:- Maximum absolute oil content and ainimum
average acidity. As an illustration 5 actual fre-
quency curves are given below. The curves are being
tried experimentally in a large number cf estates in
the Congo.
A. Oil content of bunches
It is assumed that maximum oil content in
the fruit is attained-only when that fruit
detaches itself, then it must be implicity acknow-
ledged that bunches are, inpractice, harvested
below that maximum. The following question there-
fore arises: What is the effect of the average
ripeness on the oil content of bunches?
 - 25 -

It is extremely difficult to study the evolution

of the absolute oil content of bunches during ripening.
It is however easy to assess the relative oil content
of the pulp or of the dry matter or of the fruit.
The oil content of fruit can be used in the assessment
of the absolute oil con-cent only if the evolution of
the weight of the fruit during ripening is known.
The study of the increase in absolute oil content
is in progress. It requires a very large number of
tests. Moreover, the increase in the oil content of
dry matter in the course of ripening is also being
investigated on a large scale.
It may be pointed out that the oil shortage
resulting from insufficient ripeness as calculated
from the Kehren data (20) is as follows:-
Table 10

Loss of oil
expressed as
Ripeness percentage Gil content
-of total oil of bunches
Ripe fruit Ofc 20%
2 days before 2.6/c 19. 5#
5 days before 7.7/= 2.8.5*
3 days before 23.1?° 15.4*

The figures of the 3rd column of the table show

the drop in the oil content of bunches calculated on
the basis of 20$ at optimum ripeness.
The data show how serious the problem is; also
that it might be easier in the present conditions of
oil extraction to increase oil production through
close control of bunch ripeness than through process
control in the mill. It oust however be taken account
of the fact that too advanced a degree of ripeness may
trigger an increase in P.?.A. which may wipe off the
profit derived frcm the extraction of a larger amount
of oil. In addition to the loss in value inherent in
the acidity, the difficulties of disposing on the
market of oil with high acidity must be reckoned with.
 - 26 -

5<rs u8nc
50 | 1 y in 1° Crop with high
proportion of
green bunches.
40 -Hiper crop still
' \ containing too
high a proportion
30 of "0" bunches.
t!

20
\
\
10

0 1-0 11-20 21-404-70 7W.OO

Classes
Figure i
Rroeness curves

\ Frequency in
po
\CroT) with wide range of
^^ _^ - \
^^T r*-i
* - nC^XQ
A £l p r j aSO.
oo T*o 1">A"h*i
i fcj -t» w UC "^w he
40 ^ ^ proportion of green
\ bunches and that
over-ripe bunches.
30 ' ~r-r~.-Experimental harvest.
; \
' \
20 / i
' - . /\ t^-^\
X
10 ! ,''
'^1—^.^^x
\>N^ x
S , / N '?^<^
-
-^ , ^^-^
ft ' x \ / ^\ ° N.
x ^*.,' X
0
0 1-0 11-20 21-40 41-70 71-100 lOOi
*
Classes
Figure 6
Ripeness curves