Professional Documents
Culture Documents
IV. T h e E s t i m a t i o n of F i b r e C o n t e n t
I. INTRODUCTION
FIBRE is the dry water-insoluble matter of sugarcane and it greatly influences
the milling house efficiency, viz., extraction, milling capacity, recovery of
commercial sugar, etc. A suitable sampling technique for its accurate
estimation is thus a desideratum. That a slight increase in fibre per cent.
of cane brings about considerable reduction in recoverable 95 ~ sugar has
been shown by Kerr and Cassidy 0937) and Daubert 0939). Davis 0936)
and Ryan (1937) drew attention to the laboratory technique for the determina-
tion of fibre content. Earlier Moberly (1935) had pointed out that the
accuracy of any method of determining fibre depends primarily upon the
efficiency of sampling and that reasonably consistent results might be
obtained with samples of 200 pounds of cane. Kerr (1939) in his investiga-
tions on the sampling studies found that there was a considerable amount
of variation in fibre content within the stalk, from stalk to stalk, and be-
tween different varieties. He showed, however, that a sample of 18 stalks
for 'Badila ', l l for " Clark's seedling ", 14 for " S . J. 4 " and 43 for
" D 1135 " was sufficient for the reliable estimation of fibre content.
The determination of fibre of a complete stalk affords practical diffi-
culties and it is, therefore, imperative to see if a whole stalk could be re-
placed by some of its specific portions. The aim of the present contribution
is to present systematically the variations that exist in between rows, canes
and portions of a cane and to indicate, following a study of the variations,
the minimum sample size required for a plot of an 1/40th acre in the accurate
assessment of fibre value.
II. METHOD OF FIBRE DETERMINATION USED IN THE LABORATORY
The standard direct method (Appendix) recommends the use of calico
bags for washing the pulp but results obtained at the station (Table I) showed
269
270 K.L. K h a n n a and others
that fibre content varied appreciably when fresh and used calico bags were
employed, thus necessitating their frequent changes and maintenance of
quality. Besides such a method would be of little avail where large rounds
of analysis have to be completed during the course of the day.
TABLE I
by crushing the remaining portions of the canes. That this method gives
reliable results will be obvious from data (Table II) collected with the gwo
methods.
Sampling" S/udies in Su~arcane--I k' 271
TABLE II
Showing comparative results by two methods of fibre determination
The analysis of variance (Table III) showed that differences due to methods
were statistically not significant and that the indirect-method could be
reliably employed for studies of this nature,
272 K. L. K h a n n a and others
TABLE I I I
With the help of the formula outlined in the preceding section, the
data on fibre per cent. have been collected for three varieties--Co 299, Co 313
and Co 356 representing early, medium-early and mid-season varieties.
The material falls under three categories. In the first category rows were
at first selected at random in a plot and then random selection was made
of canes in each row in such a manner as not to include two canes from
the same d u m p . Each cane was subsequently divided into three approxi-
mately equal divisions (1) bottom, (2)middle and ( 3 ) t o p ; and each
division was further subdivided into portions, each containing a complete
nodal length (a node and an internode). Two such portions in a division
were separately analysed for fibre content. For each cane, therefore,
there were six observations, each being the value of fibre per cent. of a
portion. The statement below gives the details of the data collected
under the first category:
Under the second category, fibre per cent. was determined individually
for each cane available in two plots (1/40th acre each),--one under Co 313
Sampling Studies in SugarcanemlV 273
and the other under Co 299. The details o f data under the third category,
where the fibre per cent. was determined clumpwise are given below:
Rows* Clumps
Let
v 1 be the mean square between plots
v~ ,, ,, rows within plots
v3 ,, ,, canes within rows
v4 ,, ,, divisions within canes
v5 . . . . portions within divisions
Then
v 5 will estimate 053
V4 ,~ m ~4$Ju r
V3 ,,
(ii) The formula~ for three-fold Nested sampling in unequal case will
also be required. They are reproduced below:
m i = the total number of third order zones occurring in the ith first
order zone, and
(i) Co 313
The data (ia) mentioned in section (3) consists of 10 • 10 • 3 • 2
600 observations, each being the value of fibre per cent. of a portion in a
cane. There are 100 canes in all and for each cane there are 6 such observa-
tions. Using the formula~ corresponding to four-fold Nested sampling, the
estimates of the variances due to rows, canes, divisions and portions have
been obtained and shown in the last line of Table IV. Columns (5), (7),
(9) and (11) of the table show respectively these estimates and the adjacent
columns the percentage variations due to each of these four sources. It
will be seen that the variations due to canes is the maximum being 40.63~
and that due to rows is the minimum being 4.02~. The percentage
variations due to divisions and portions are 37.77 and 17.58 respectively.
The mean fibre per cent. is 14.09 and the error expressed as its percentage
has come out to be 1.35.
Fibre per cent. for this plot was determined during the month of
November, whereas the fibre per cent. for each of the 6 plots shown in the
first 6 rows of Table IV was determined during February. For the purpose
of comparing the errors in these two sets of plots, account has to be taken
of the fact that the sample sizes were different. The average error ~ in
the first 6 plots is 2.91, whereas the adjusted error per cent. for the 7th plot,
in the table, comes out to be 2.21. For all practical purposes, these two
percentages may be considered to be of the same order. The average fibre
per cent. for the first 6 plots is 13- 67, whereas the fibre per cent. for the 7th
plot is 14.09. These values also are not significantly different from one
another. The variations due to the different sources in these 6 plots show
more or less a similar trend with those of the last plot.
When, however, a pooled analysis is made of the first 6 plots and the
estimates of the variances obtained by the formula~ given in Section (4) it
is seen that variation due to canes still remains the maximum and that the
276 K . L . Khanna and others
X X
X o o ~ ~ o,
X X
9
o o o o ~
6 6 6 6
r.
II II
b
b b
. 9 : : :
Sampling Studies in Sug,arcanemIV 277
variation due to plots now becomes the minimum. The following table
gives the estimated variances and their values expressed as percentages of
the total variation:
Variance Variance
of total
(ii) Co 299
The similar data available for Co 299 have been subjected to the same
type of analysis and the results have been set out in Table V. It has been
found that the mean fibre per cent. of Co 299 is significantly higher than that
of Co 313. The variations in the fibre per cent. shown by the different sources
are in general high and have reflected in the high standard errors. In one
plot, the standard error has been as high as 9"02~o. These plots have
shown a divergence amongst themselves in respect of the variances due to
the different sources, but the variation due to canes has none the less been
high in all the plots except in one.
This set of four plots has also been subjected to a pooled analysis. The
estimates of the variances are given below:
Variance V a ri a nc e %
of total
It will be noticed in the above table that the variation due to rows has
been the highest and that due to divisions has been second in order of
magnitude. The pl~ts under Co 299, therefore, show a marked divergence
278 K . L . Khanna and others
~3
0 X
X o 0 o
X
o
0 0 0
.
,--t
b
C~ OO
r~
b
~.o
(..)~
o =
03
N N N
b p.c~
& o . ~I
]I ]I
b vo
b
0
b
o =
03
Jl II
I
b
Sampling Sludies in Sugarcane--lU 279
from the plots under Co 313, analysed before, in the distribution of the
percentage varations.
VI. AGREEMENT IN BETWEEN THE FIBRE PER CENT. VALUES OBTAINED FROM
A COMPOSITE SAMPLE OF A CANE AND THE M E A N OF THE FIBRE
PER CENT. VALUES OBTAINED FROM SUB-SAMPLES OF THE C A N E
An attempt has then been made to find out whether or not the fibre
per cent. obtained from a composite sample of a number of sub-samples
taken from a cane agrees with the mean fibre per cent. of the sub-samples.
For this purpose, 12 random canes of Co 313 were taken and each was
divided into 6 portions. The fibre per cent. was then determined for each
composite sample as well as individually for each sub-sample. Another
set of 12 random canes was taken and each cane was divided into 3 portions
for similar analysis. Table VI shows these details. Student's t-tests reveal
TABLE V I
Showing the agreement in between the mean of individual values and the
value of the composite sample
Set i Set II
that the agreement in between the mean of the individual values and the
value of the composite sample is very close in both the sets.
VII. DATACOLLECTEOUNDERTHE SECOND CAaXGORYCO 313 ANt) CO 299
After having arrived at the conclusion that the fibre per cent. of a cane
determined from a composite sample can be relied upon, it was decided
B4
280 K.L. K h a n n a and others
to determine on this basis, the fibre per cent. of all the canes in an 1/40th
acre plot under each of the varieties, Co 313 and Co 299. There were
7 rows, 242 clumps and 567 canes in the plot under Co 313, and 7 rows,
220 clumps and 671 canes in the plot under Co 299. The estimation of the
variances due to rows (a12), clumps (a22) and canes (e32) obtained according
to the formulae (unequal numbers) outlined in Section 4 (ii) are given below:
Co. 299 .. 18.32 0.47 2.57 1.39 12.29 3.50 30-67 6.52 57.14
Here also, it is found that the mean fibre per cent. of Co 299 is significantly
higher than that of Co 313. It will be noticed further that the percentage
variation due to canes (%2) has been the highest. The striking feature in
the above results appears to be the high magnitudes of the standard errors.
The error in respect of Co 313 has been as high as 3-6570, which seems to
be rather too high for the data of complete enumeration.
A random sample taken from these data has also been analysed. Ten
d u m p s were selected at random in each of the seven rows and the fibre per
cent. values of two canes taken at random from each clump were subjected
to similar analysis. The estimatrs of the variances due to rows, clumps
and canes are given below:
Co 313 .. 14.18 0.40 2.82 1.09 !2.51 4.41 50.84 3.18 36.65
Co 299 .. 19.03 0.54 2.84 2.54 20.00 5.10 40.00 5.09 40.00
The differences in the mean fibre per cent. values obtained from the
complete data and the sample sets are not statistically significant and both
these sample sets have also revealed that canes contribute a considerable
part of the total variation. The error in respect of Co 313 in the sample
set has been less than that in the data of complete enumeration; and it is
mainly due to the fact that ~2 came out to be zero in the latter.
Fibre per cent. was determined during January and March for a set o f
4 plots each, under each of the varieties, Co 313, Co 299 and Co 356, where
Sampling Studies in S u g a r c a n e - - I V 281
the fibre per cent. of a clump was found from a composite sample. The
estimated variance with other details have been set out in Table VII.
TABLE VII
Estimates of the variances due to the different zones and S.E. expressed as
percentage of the mean
Mean ~m of r TotalS1,
o r Totalr Plot size No. of observations
mean
Plot 1 13-13 0.24 1.83 0.00 0.00 3.5422 I00.00 1/40 acre I 6 rows X I0 clumps
2 13.16 0.17 1.29 0.00 0.00 1.7572 100.00 do ] do
3 13.51 0.20 1.48 0.00 0.00 2.4098 100.00 do l do
4 13.62 0.25 1.84 O. 1522 6.04 2.3681 93.96 do l do
Analysis reveals that the mean fibre per cent. in January was significantly
less than the mean fibre per cent. in March, whereas in between the different
B5
282 K.L. K h a n n a and others
varieties themselves, the mean fibre per cent. shows the following order of
magnitude : -
J a n u a r y - C o 299 > Co 356 > Co 313
March--Co 299 = Co 356 > Co 313.
The standard errors expressed as percentages have been very low for
all the varieties. The table will indicate that in majority of the plots the
standard errors have been well within 2~o. It will further be noticed that
the variation between rows is but insignificant compared to the variation
between clumps.
IX. S A M P I I N G REQUIREMENTS
Approx. Approx. i
Co. 299 15 15 9 , Rows = 3
I
Divisions ~ 3 in a
cane
Portions = 6 in a
cane
* Indicates "Impossible"
Sampling Studies in Sugarcane--IV 283
plots for Co 313, Co 356 and Co 299 analysed in Tables IV and V have
been less than 5~o of the mean except in three plots under Co 299 (vide
Table V). In the data referred to in these two tables, the fibre per cent. of
a cane was represented by the fibre per cent. values of six portions taken
equally from its three divisions, whereas in the data collected subsequently
the fibre per cent. of a cane or of a clump was represented by the fibre per
cent. value of a composite sample. For the sake of comparison with the
results obtained in the subsequent plots it has been considered necessary
to keep in the former plots the number of divisions and portions as they
are in working out the sample sizes.
Since only three rows were taken at random from each of the plots
analysed in Tables IV and V except for plot (7) in Table IV, the number
of rows has not been increased beyond " t h r e e " in working out the sample
sizes for these plots. In Table VIII have been shown the required samples
sizes in terms of the number of canes per row worked out on the basis of
289 and 570 error margin. In some plots, the contribution to the error
per cent. made by rows has been too high to admit of the required reduction
in error by an increase in the number of canes alone. Such cases have
been marked with an asterisk(*). This contingency has occurred mostly
under the error margin of 289 and brings out the necessity for a larger
representation of rows in the random sample.
Similar combinations of rows, clumps and canes have also been worked
out in respect of the random samples taken from the data of complete
enumeration of Co 313 and Co 299. The table given below shows the
combination worked out on the basis of 570 error. Combinations on the
basis of 289 error are not possible on account of the reasons furnished in
the preceding paragraph.
Co 313 C o 299
3 19 I 3 64 1
3 15 2 3 44 9~
6 4 1 6 4 1
6 4 2 6 3 2
TABLE IX
Number of clumps to be selected from each row so that S.E. may not exceed
289 and 5% of the mean
Results for plots analysed in Table VII
Data collected during January
s% 389
! |
C o 313 Q @
3 2 2 2 12 6 I 7 11
* $
Co 299 Q D i 3 3 3 Very Rows ~
large
$
~o 356 O @ 3 3 3 19 19 J
Co 313 .. Zl 3 zl Zl 10 4 7
Co 299
Co 356
Zl
Zl
Zl
5
Zl Zl
/1
4
11
4 4
2
~ Rows--
Plot I Plot II Plot III Plot IV Plot I PlotII Plot III Plot IV Remarks
I
Co 313 ~ 9 2 zl /1 /1 6 3 4 5 ]
I
Co 299 I 9 Zl Zl zl /1 6 10 10 10 ~- Rows = r
i I
Co 356 Q 6 3 2 2 ! 2 14 5 7 5 i
Co 313 g 9 Zl 2 zl zl 5 )
I
Co 299 I g tl /1 Zl /1 2 ~- Rows = r
I
Co 356 I 6 Jl 2 2 ,'1 10 J
* Indicates Impossible
Sampling Studies in Sugarcane--IV 285
Therefore, the results given in the first line of the above table might be
compared with those set out in Table VIII.
In view of the large variations in individual canes due to differences
in their ages it is necessary to take a clump, which represents all age groups
so far as the progeny of a planted set is concerned, as a basis for recom-
mending suitable sample size for fibre determination; and for this purpose
the plots analysed in Table VII have been selected. Combinations of rows
and clump have been worked out on the basis of 289 and 5~o errors and
shown in Table IX where all the three varieties, viz., early, medium-early
and mid-season have been represented.
For an 1/40th acre plot, of which the number of rows would not exceed
6 or 7, the following sampling sizes may be recommended. Range o f
variation in the sizes has been indicated for each variety. The number of
rows has been kept constant at 6, as it was pointed out before that rows
should be given as large a representation as possible.
..~ 6 x 1 = 6 1 6x 2 =12
C o 313
6x2 =12 6X 6 =36
] fix 2 •lg
Co 299 .. 6 x 1 = 6
J 6x10 =60
"~ 6 x l =6 6 x 2 =12
Co 356 .. }"
J 6x2 =12 I 6 x 14 --- 84
The size of the sample has been shown in terms of the number of clumps,
and it may be recalled here that the mean fibre per cent. value of a clump
is to be determined from a composite sample taken from the different divi-
sions and portions of a cane in the manner as described in Sub-Section VI.
X. SUMMARY OF C O N C L U S I O N S
(iii) Analysis revealed that the fibre per cent. obtained from a composite
sample taken from a cane agrees very closely with the mean of the fibre
per cent. values of the different sub-samples comprising the composite sample.
(iv) Of the four sources of variation, namely rows, canes, divisions,
and portions, the variation due to canes has been the maximum. This is
understandable in view of the difference in the ages of the daughter shoots.
(v) A clump represents all age groups so far as the progeny of a planted
set is concerned and therefore, sampling for fibre has to be confined to whole
clumps drawn from different rows. For an 1/40th acre plot, the rows of
which will not exceed 6 or 7, the required sammpling sizes with the rang of
variation in each variety are indicated below. The fiber per cent. value of
a clump has to be determined from a composite sample as outlined in Sub-
Section VI.
l 6 x 1 = 6 1 6 x 2 =12
Co 313 Q 9 (
J 6• =12 J 6 x 6 =36
1 6x 2 = 12
Co 299 .. 6 x 1 = 6
J 6 x 10 = 60
..l 6 x 1 = 1 l 6x 2 =12
Co 356
J 6x2 =12 6 x 14 = 84
ACKNOWLEDGEMENTS
The work was carried out as part of the Sugarcane Research Scheme
being financed jointly by the Bihar Government and the Indian Central
Sugarcane Committee to whom grateful thanks are due. The assistance
rendered by Mr. M. Farooque, Senior Chemical Assistant in Estimation
of Fibre Content, is also acknowledged.
REFERENCES
1. Daubert .. Facts About Sugar, 1939, 34, 11, 38-39.
2, Davis, M . B . .. Proc. 2nd Ann. Conf. Q.S.S.C.T., 1936, 54-57.
3. Kerr, H . W . .. I . S . J . , 1939, 41, 486, 227-30.
4. ---- and Cassidy, N . G . .. Bull. Bur. Sug. Expt. Sta., Queensland, 1937, 2.
5. Mahalanobis, P . C . .. Phil. Trans. Roy. Soc., London, 1944, 231, 327-451.
6. Moberley, G . S . .. Proc. 9th Cong. S. Aft. Sug. Technol. Assn., 1935, 61-64.
7. Deerr, Noel .. Handbook Sug. TechnoL Assn., India, 1928.
8. Ryan, W . J . .. Proc. 3rd Ann. Conf. Q.S.S.C.T., 1937, 26-27,
9. Roy, S. N. and Banerjee, K. S . . , Science and Culture, 1940, 6, 3, 189.
APPENDIX
DETERMINATION OF FIBRE CONTENT