International Journal of Engineering and Technology Volume 6 No.

10, October, 2016

An Assessment of Laboratory Sample Splitting Techniques

Nenuwa, Olushola B. and Adetiloye, Adeola
Department of Minerals & Petroleum Resources Engineering,
The Federal Polytechnic, P.M.B. 5351, Ado-Ekiti, Ekiti state, Nigeria

ABSTRACT

Two different laboratory sample splitting methods were used to prepared representative sub-samples. Performance evaluation tests
were conducted using the coning and quartering sampling technique and the Jones riffle sample splitter, the bulk sample was iron
chips and sand mixture. The sample splitting exercise was repeated thrice for each method and the results obtained were recorded.
On completing the performance evaluation tests, the percentage content of the iron chips and sand in the parent sample were
compared with the percentage contents in the splitted samples. It was observed that the value of the absolute difference between the
percentage content of sand in the parent sample and splitted sample for the Jones riffle sample splitter is lower compared to that of
coning and quartering technique. It was also observed that the value of the absolute difference obtained for iron content in the parent
and splitted sample is also lower for sampling with Jones riffle splitter than when coning and quartering technique was used. Jones
riffle sample splitter is therefore recommended for splitting dry granular samples whenever a cheap, efficient, accurate and less
sophisticated technique for sample splitting is desired.

Keywords: Absolute Difference, Riffler, Sampling, Splitting, Sub-Sampling

1. INTRODUCTION

Sampling is the means whereby a small amount of material is quality if scientifically supportable decisions are to be made
taken from the main bulk in such a manner that it is (Gerlach et al, 2003).
representative of that larger amount. Great responsibility rests
on a very small sample, so it is essential that samples are truly The act of sampling may introduce more uncertainty than all
representative of the bulk (Homles, 1991). Wherever possible of the subsequent steps in the measurement process (Jenkins
samples should be taken of the material when it has been et al, 1997; Crumbling, 2001; Allen, 1996). This is especially
reduced to the smallest particle size consistent with the true for heterogeneous particulate samples. The goal of most
process. For instance, the ground ore pulp will be easier to sampling procedures is to obtain a representative sample. In
sample, and will give more accurate results than the feed to terms of characterizing the level of contaminants, the ideal
the primary crusher (Wills, 2006). representative sample would have the same concentration of
contaminants as the original sample. Procedures for obtaining
Sampling procedures cover the practice of selecting representative samples from particulate matrices have been
representative quantities of test material in the field, to studied for some time in the minerals industry and a
evaluate bulk materials. Examples of the test materials are comprehensive sampling theory originated by Pierre Gy has
bulk granular solids, slurries, sludge, grains, and solid fuels. It been developed (Pitard, 1993; Gy, 1998; Myers, 1997; Smith,
is necessary to be able to sample bulk materials during 2001).
shipment and during processing operations. Taggart (1945)
defined sampling as: “The operation of removing a part The need for sample homogeneity prior to laboratory analyses
convenient in size for testing, from a whole which is of much has been long recognized by geologists, chemists, and
greater bulk, in such a way that the proportion and members of other scientific disciplines. Homogeneity is the
distribution of the quality to be tested (e.g. specific gravity, degree that the material under investigation is mixed resulting
metal content, recoverability) are the same in both the whole in the random distribution of all particles in the sample.
and the part removed (Sample).” In a laboratory before Completely homogenous materials are so rare that they may
performing a size analysis, chemical analysis, or any other be considered non-existent (Ingamells and Pitard, 1986). Yet,
determination of material properties, it is necessary to take a scientists must strive to obtain a homogenous sample in order
representative sample from the lot being analyzed. For most to obtain data exhibiting minimal error attributable to sample
metallurgical or environmental operations, the methods of heterogeneity. Care must be taken during the sub-sampling
sampling includes: Grab sampling, Coning and Quartering, phase of soil preparation, during sample transport, unpacking,
Chute-type sample splitters and Rotary riffles. Sampling can and transfer to other containers in the laboratory to avoid
be a significant source of error in the measurement process. particle sorting via different particle densities, shapes, sizes,
The characterization and cleanup of hazardous waste sites and resistance of certain minerals to mixing, such as
require data that meet site-specific levels of acceptable magnetite (Muller,1967; Ingamells and Pitard, 1986; Reeves
and Brooks, 1978). Numerous methods have been used to
ISSN: 2049-3444 © 2016 – IJET Publications UK. All rights reserved. 358
 International Journal of Engineering and Technology (IJET) – Volume 6 No. 10, October, 2016

obtain homogeneous soil samples. The various methods range Selecting a reduced sample mass from the quantity available
from simple grinding and sieving of the sample to a desired for analysis is one the most common procedures performed in
particle-size to various mixing and splitting devices and an analytical laboratory. However, little attention is usually
machines (Schumacher et al, 1990). given to this procedure and the effect it may have on the
quality of the data (Gy, 1998). From a laboratory analyst’s
Ingamells and Pitard (1986) expressed serious doubts as to point of view, a reported sample concentration is an estimate
the value of mechanical splitters. They stated that: ‘In of the average level of analyte in the original mass of sample
general, machines that use mechanical violence and look and received for analysis. For particulate or soil samples, this
sound as though they are efficient are most likely to cause value is often based on the analysis of an aliquot selected
segregation of heavy and light, large and small, and flat and from the soil sample. Sub-sampling practices have an
round particles’. Further concerns have been expressed important effect on the reported results. Improper sub-
concerning sample breakdown to finer particle sizes due to sampling procedures can lead to results that have significant
the violent tumbling in the machines. The riffle splitter (also biases and large imprecisions (Pitard, 1993; Myers, 1997).
called a chute splitter, Jones splitter, or sample splitter) is
perhaps the most common mechanical method for sample Several studies have been conducted comparing the
homogenization and/or sample size reduction (Ingram, 1971; effectiveness of the riffle splitter to other homogenization
Mullins and Hutchison, 1982). The riffle splitter also provides techniques. The aim of this study is to assess and compare the
one of the best general methods of sample mixing to obtain performance of a mechanical sampler (the Jones riffle
bulk sample homogeneity (Ingamells and Pitard, 1986). A splitter) and a traditional sampling method (coning and
riffle splitter is a device having an equal number of narrow quartering sampling). This work will help a mineral analyst to
sloping chutes with alternate chutes discharging the sample in choose an appropriate sample splitting technique that will
opposite directions into two collection bins (Fig. 1). Sample produce the best homogenised test sample which will
homogenization is achieved by repeated pouring of soil introduce the smallest level of inaccuracy.
through the splitter and combining the halves between passes.
The use of the riffle splitter as a sub-sampling device is done
in a similar manner with the exception that after the sample is
passed through the splitter, one collection pan is replaced
with a clean pan. The material in the ‘replaced” pan, which
contains about one-half of the original sample, is then passed
through the riffle splitter again thereby reducing the volume
in the clean pan to one-quarter of its original sample volume.
This process of sample reduction is repeated until the desired
weight or sample size is obtained (Schumacher et al, 1990).

Perhaps the best known sample splitting method is the

classical coning and quartering technique. This technique
involves pouring the sample into a cone, flattening the cone,
dividing the flattened cone into four equal divisions
(quartering), and then removing two opposite quarters (Fig.
2). The remaining two quarters are repiled into a cone and the Figure 1: The basic components of a riffle splitter include the
process is repeated until the desired sample size is obtained. scoop, an even number of chutes, or riffles, and a pair of
collection pans (Source: Gerlach et al, 2002).
Variations on the process are possible which can enhance the
speed of sample size reduction by using just one quarter
(chosen at random) to continue the splitting process or which
allow this method to be modified to homogenize large
sample. The use of the coning and quartering method to
homogenize a sample involves the removal of the first quarter
and repiling it into a cone followed by the subsequent repiling
of the opposite quarter and then the remaining two quarters to
reform a single cone (Raab et al, 1990). This process is
repeated several times until sample homogeneity is achieved.
Several sources of error for this method have been identified.
Van Johnson and Maxwell (1981) reported that during the
cone and quarter process on large samples (several
kilograms), there is the danger of unequal segregation of
heavier materials during the flattening and coning of the
sample. Similar to the riffle splitting techniques, dusting is
also a possible source of error during cone formation. Sample Figure 2: Coning and quartering repeatedly divides a sample
loss from the inability to recollect all the soil from the into halves until the desired sample size is achieved (Source:
underlying material, the ability of the sample to “cling” to the Gerlach et al, 2002).
underlying material via static charges, and sample
embedment are all further sources of error.

ISSN: 2049-3444 © 2016 – IJET Publications UK. All rights reserved. 359
 International Journal of Engineering and Technology (IJET) – Volume 6 No. 10, October, 2016

2. METHODOLOGY repeated thrice in order to re-split the other portion of the

splitted sample. All the results obtained were recorded.
i. Materials and Equipment
vi. Determination of the Percentage Composition of the
The materials and equipment required for this study are: iron Sample Constituents
chips, sand, bar magnet, Jones riffle splitter, cutting knife and
electronic weighing balance. The percentage composition of the constituents in the parent
and splitted samples were calculated using Equation 1 and 2:
ii. Sample Collection Percent (%) of iron in the parent or splitted sample =
100% ………..…... 1
8kg quantity of sand and 4kg quantity of iron chips were Percent (%) of sand in the parent or splitted sample =
collected inside containers at Abuja boys hostel and the
Mechanical Engineering workshop respectively both located ….………………. 2
within the campus of the Federal Polytechnic, Ado-Ekiti,
Ekiti state. Where, Mi = Mass of iron in the parent or splitted sample
Ms = Mass of sand in the parent or splitted sample
iii. Sample Preparation M = Mass of the mixture of sand and iron chips in
the parent or splitted sample
1000g of iron chips and 2500g sand were weighed using the
electronic weighing balance. The two samples were
thoroughly mixed to produce a homogenous sample of 3500g 3. RESULTS AND DISCUSSION
iron chips and sand mixture. This was used as the test sample.
After conducting the performance evaluation test for both
iv. Experimental Procedure for Coning and Quartering coning and quartering technique and the Jones riffle sample
Sampling Technique splitter, the results of the tests were recorded and the data
obtained was used to calculate the percentage content of the
3500g of the test sample was poured on a flat table to form a iron chips and sand in the test samples as described earlier.
cone shaped pile, and the pile was uniformly flatten until the The absolute difference between the percentage content of
diameter is approximately equal to four times the thickness of sand and iron in the parent and splitted sample for both
the pile. The sample was divided into half by vertically sampling techniques were also obtained. The results obtained
passing the cutting knife through the center of the pile. In a were tabulated and presented graphically in fig. 3 – 9.
similar manner, the halved pile was divided into another half,
thus quartering the sample. The sample was combined Fig. 3 is a bar chart showing the percentage of materials in
diagonally with opposite quarters of the materials forming the parent sample and splitted sample for the coning and
two splitted samples. Then one of the two halves was stored. quartering sampling technique. From the graphical analysis
The other half was spread out on the table and a bar magnet shown on fig. 3, for test 1, the percentage composition of iron
was moved over it in order to separate the iron chips from the and sand in the parent sample is 28.6% and 71.4%
sand. The iron chips that were collected by the magnet were respectively. The percentage composition of iron and sand in
then placed inside a container. The two samples were the splitted sample is 71.4% and 27.8% respectively. The
thereafter weighed separately and the result was recorded. difference between the percentage content of iron and sand in
This procedure was repeated using the other splitted sample the parent sample and splitted sample for test 1 is large,
as the test sample in order to do a re-splitting. The re-splitting which indicates that this result is not reliable. For test 2, the
was done thrice in order to obtain a total of four different percentage composition of iron and sand in the parent sample
sampling tests. The results obtained were also recorded. is 72.0% and 27.3% respectively. The percentage
composition of iron and sand in the splitted sample is 61.55%
v. Experimental Procedure for the Jones Riffle Sample and 27.5% respectively. The difference between the
Splitter percentage content iron and sand in the parent and splitted
sample for test 2 is small which indicates that it is a reliable
The test sample was prepared and 3500g was weighed as result. For test 3, the percentage composition of iron and sand
earlier done on the weighing balance. The mixture of the sand in the parent sample is 65.1% and 30.9% respectively while
and iron chips was poured evenly into the hopper of the Jones the percentage composition of iron and sand in the splitted
riffle splitter. The samples then flowed through the alternately sample is 72.2% and 28.1% respectively. The difference
arranged passages in the opposite directions (chute/riffle between the percentage content of iron and sand in the parent
bank) into the two collecting pan under the dividing head and splitted sample for test 3 is also small which also
outlet. The feed sample was then divided into two indicates a reliable result. For test 4, the percentage
representative sub-samples. One of the sub-samples was then composition of iron and sand in the parent sample is 72.0%
spread on the table and the bar magnet was placed very close and 28.0% respectively, the percentage composition of iron
to it in order to separate the iron chips from the sand. The iron and sand in the splitted sample also is 75.4% and 23.4%
chips that were collected by the magnet was placed inside a respectively. Hence, the difference between the percentage
container and the sand was placed in another container. The composition of iron and sand in parent and splitted sample for
two samples were thereafter weighed separately as earlier the test 4 is low and this further confirms the efficiency of the
done and the results were recorded. This procedure was also coning and quartering technique.

ISSN: 2049-3444 © 2016 – IJET Publications UK. All rights reserved. 360
 International Journal of Engineering and Technology (IJET) – Volume 6 No. 10, October, 2016

Fig. 4 is a bar chart showing the absolute difference of the Fig. 5 is a bar chart showing the absolute difference of iron
percentage content of sand in the parent sample and splitted content in the parent sample and splitted sample for the
sample for coning and quartering splitting technique, from the coning and quartering splitting technique, from the chart, test
chart, test 2 gives the lowest absolute difference of sand 4 gives the lowest absolute difference of iron content in the
content in the parent and splitted sample (0.22%) which parent and splitted sample (3.39%) which means that test 4
means that test 2 is the most reliable result. The absolute results are very reliable. The value obtained for test 3 is also
difference obtained for test 3 and 4 are also low, 2.78% and low (7.04%). The value obtained for test 1 and 2 are high
6.58% respectively, this further assures the efficiency of this (42.8% and 34.5% respectively). The parent sample must
technique. The value obtained for test 1 is high (44.2%), thus, thus, be re-splitted at least twice in order to obtain sub-
the result of test 1 is erroneous and so re-splitting the sub- samples that will adequately represent the content of the
sample is necessary. parent samples.

Fig. 3: Bar chart showing the percentage of materials in the parent sample and splitted sample for coning and quartering technique

Fig. 4: Bar chart showing absolute difference of sand content in the parent sample and splitted sample for Coning and Quartering splitting
technique

Fig. 5: Bar chart showing absolute difference of iron content in the parent sample and splitted sample for Coning and Quartering splitting
technique

ISSN: 2049-3444 © 2016 – IJET Publications UK. All rights reserved. 361
 International Journal of Engineering and Technology (IJET) – Volume 6 No. 10, October, 2016

Fig. 6 is a bar chart showing the percentage of materials in the bulk sample. Test 1 gives absolute difference of 1.92%
the parent sample and splitted sample for Jones riffle sample while test 3 and 4 gives absolute difference of 0.65% and
splitter, from the chart, test 1 shows that the percentage 0.72% respectively. These results confirmed that Jones riffle
content of iron and sand in the parent sample is 28.6% and sample splitter is an efficient sampling device.
71.4% respectively while that of iron and sand in splitted
sample is 26.1% and 73.4% respectively, hence the difference Fig. 8 is a bar chart showing the absolute difference of the
between the percentage composition of iron and sand in percentage iron content in the parent sample and splitted
parent and splitted sample for test 1 is small. For test 2, the sample for the Jones riffle sample splitter, from the chart, test
percentage content of iron and sand in parent sample is 26 2 gives the lowest absolute difference (0.09%). The value of
.2% and 73.8% respectively and that of the percentage the absolute difference for test 1 is 2.47% while that of test 3
content of iron and sand in splitted sample is 26.2% and and 4 are 0.66% and 0.72% respectively. These results further
73.6% respectively. For test 3, the percentage content of iron confirmed the efficiency of the Jones riffle sample splitter.
and sand in parent sample is 26.2% and 73.8% respectively,
while that of the splitted sample is 26.9% and 73.2% Fig. 9 is a bar chart comparing the absolute difference of the
respectively. For test 4, the percentage composition of iron sample contents in both coning and quartering technique and
and sand in parent sample is 26.9% and 73.2% respectively, the Jones riffle sample splitter, from the chart, test 1, 2, 3 and
while for the splitted sample, it is 26.1% and 73.9% 4 gives lower value of the absolute difference of sand content
respectively for both iron and sand. The difference in the in the parent sample and splitted sample for Jones riffle
percentage composition of iron and sand in parent and sample splitter when compared to that of coning and
splitted sample for all the four tests are small and these results quartering sampling technique. The absolute difference of
showed that the sample splitting method is efficient for all the iron content in the parent sample and splitted sample is also
four tests conducted. lower when Jones riffle sample splitter was used than when
coning and quartering technique was used for sampling.
Fig. 7 is a bar chart showing the absolute difference of the These implied that sampling using the Jones riffle sample
percentage sand content in the parent sample and splitted splitter gives more accurate and reliable results than sampling
sample for Jones riffle sample splitter, from the chart, test 2 using the coning and quartering sampling technique.
gives the lowest absolute difference of the percentage sand
content in the parent sample and splitted sample (0.15%)
which means that test 2 sub-sample is accurately representing

Fig. 6: Bar chart showing the percentage of materials in the parent sample and splitted sample for Jones Riffle Sample Splitter

ISSN: 2049-3444 © 2016 – IJET Publications UK. All rights reserved. 362
 International Journal of Engineering and Technology (IJET) – Volume 6 No. 10, October, 2016

Fig. 7: Bar chart showing absolute difference of sand content in the parent sample and splitted sample for Jones riffle sample splitter

Fig. 8: Bar chart showing absolute difference of iron content in the parent sample and splitted sample for Jones riffle sample splitter

Fig 9: Bar chart comparing the absolute difference of sample contents in both the coning and quartering technique and the
Jones riffle sample splitter.

ISSN: 2049-3444 © 2016 – IJET Publications UK. All rights reserved. 363
 International Journal of Engineering and Technology (IJET) – Volume 6 No. 10, October, 2016

4. CONCLUSIONS Ingamells, C.O. and Pitard, F.F. (1986): Applied

Geochemical Analyses. J. Wiley and Sons, Inc. New
The coning and quartering sampling technique and sampling York. 733 pp.
using the Jones riffle sample splitter was actually conducted
by observing all necessary precautions. From the performance Ingram, R.L (1971): Sieve Analyses. &I R.E. Carver (ed.)
evaluation results obtained, the value of the absolute Procedures in Sedimentary Petrology. J. Wiley and Sons, Inc.
difference between the percentage content of sand in the New York. pg. 4967.
parent sample and splitted sample for the Jones riffle sample
splitter is lower compared to that of coning and quartering Jenkins T.F., Grant C.L., Brar G.S., Thorne P.G., Schumacher
technique. It was also observed that the value of the absolute P.W., Ranney T.A. (1997): Field Anal. Chem. Technol. 1
difference obtained for iron content in the parent and splitted (3)151–163.
sample is also lower for sampling with Jones riffle splitter
than when coning and quartering technique was used. It can Muller, G. (1967): Methods in Sedimentary Petrology. b W.
therefore be concluded that using Jones riffle sample splitter Van Engelhart et al., Sedimentary Petrology, Part 1. Hafner
for sampling procedure will give more accurate results with Publishing Co., New York. 283 pp.
higher efficiency than when the coning and quartering
sampling technique is being used, the sub-samples that will Mullins. C.E. and Hutchison, B.T. (1982): The Variability
be obtained will also be more homogeneous and will more Introduced by Various Subsampling Techniques. J. Soil Sci.
closely represent the bulk or parent sample than when coning 33:547-561.
and quartering technique is adopted. Although, the coning
and quartering technique is more economical, it does not Myers J.C. (1997): Geostatistical Error Management:
assure accuracy. The Jones riffle sample splitter is therefore Quantifying Uncertainty for Environmental Sampling and
recommended to be used for mineral sample splitting Mapping. Van Nostrand Reinhold, New York.
whenever a cheap, mechanized and less sophisticated
technique for sample splitting is desired. Further researches Pitard F.F. (1993): Pierre Gy’s Sampling Theory and
should however be encouraged in order to evaluate the Sampling Practice (2nd Edition). CRC Press: Boca Raton, FL.
performance of other sample splitting techniques.
Raab, G.A, .Bartling, M.H, Stapanian, M.A, Cole, W.H.,
Tidwell, R.L. and Cappo, K.A (1990): The Homogenization
of Environmental Soil Samples in Bulk. In M. Simmons (ed.)
REFERENCES Hazardous Waste Measurements. Lewis Pub. In press.

Allen T. (1996): Particle Size Measurement, Vol. 1, Chapman Reeves, R.D. and Brooks, R.R. (1978): Trace Element
& Hall, London. Analysis of Geological Materials. J. Wiley and Sons, Inc.
New York. 421 pp.
Crumbling D.M. (2001): Applying the Concept of Effective
Data to Environmental Analyses for Contaminated Sites, EPA Schumacher B.A., Shines K.C., Burton J.V., and Papp M.L.
542-R-01-013, US Environmental Protection (1990): A Comparison of Soil Sample Homogenization
Agency,Washington, DC. Techniques. Lockheed Engineering and Sciences Company,
Inc. Las Vegas, Nevada 89119. Pp.1-8.
Gerlach, R.W., Dobb, D.E., Raab, G.A., Nocerino, J.M.
(2002): Gy sampling theory in environmental studies. Smith P.L. (2001): A Primer for Sampling Solids, Liquids,
Assessing soil splitting protocols. Journal of Chemometrics, and Gases, Society for Industrial and Applied Mathematics,
16: 321 – 328. Philadelphia, PA.

Gerlach, R.W., Nocerino, J.M., Ramsey, C.A., Venner, B. C. Taggart (1945): Handbook of Mineral Dressing. Wiley,
(2003): Gy sampling theory in environmental studies, Section 19.
Sampling error estimates. Elsevier. Pp. 159 – 168.
[5] Gy P. (1998): Sampling for Analytical Purposes. Van Johnson, W.M. and Maxwell, J.A (1981): Rock and
Wiley, New York. Mineral Analysis. 2nd. ed. J. Wiley and Sons, Inc. New York.
489 pp.
Holmes, R.J. (1991): Sampling methods: Problems and
solutions, in Evaluation and Optimization of Metallurgical Wills, B.A. (2006): Will’s Mineral Processing Technology:
Performance, ed. D. Malhotra et al., SME Inc., Littleton, An introduction to the practical aspects of ore treatment and
Chapter 16. mineral recovery, 7th edition, Elsevier Ltd, pg. 40.

ISSN: 2049-3444 © 2016 – IJET Publications UK. All rights reserved. 364