Sampling mineral commoditiesthe

good, the bad, and the ugly

J
o
u
r
n
a
l

by R.J. Holmes*

and that the overall precision of the final

analysis is appropriate for the required grade
control task
The golden rule for correct sampling is
that all parts of the material being sampled
must have an equal probability of being
collected and becoming part of the final sample
for analysis. If this requirement is taken into
account at the outset in designing a sampling
system, then good progress towards obtaining
representative samples is assured. On the
other hand, if this rule is not respected, then
sample bias is easily introduced. Key design
flaws that need to be eliminated (Gy2) include
incorrect delimitation of increments, i.e.,
incorrect cutter/increment geometry,
incomplete extraction of increments, preferential exclusion of specific size fractions,
sample loss, and sample contamination. In this
regard, it is not easy to take correct samples
from some of the sampling locations listed
above and should be avoided, e.g., it is
impossible to take a representative sample in
situ from a large stockpile.

A wide range of drill holes and process streams are sampled for
resource estimation, grade control, and contractual purposes in the
minerals industry. However, despite the availability of training
courses, conferences and both national and international standards
on correct sampling practices, it is still surprising how often little
attention is given to ensuring that representative samples are
collected for analysis. The reason for this is that the responsibility
for sampling is often entrusted to personnel who do not appreciate
the significance and importance of sampling, with cost being the
main driving force rather than whether the sample is representative
of the material from which it was extracted. This seriously
undermines the precision and accuracy of the analyses
subsequently generated and can render the analysis process a total
waste of time and money and expose mining companies to serious,
potential, financial losses. Company management needs to reverse
this situation and ensure that sampling is given the attention it
deserves to generate representative samples for analysis.

Introduction
Samples are taken from a broad range of
locations in the mining industry for grade
control, including blast holes, feed and product
streams, conveyor belts, trucks, railway
wagons, and stockpiles. This process is vital to
mining companies for metallurgical
accounting, optimizing resource utilization,
and maximizing profitability (Holmes1), yet
the number of instances where poor sampling
practices are used is unbelievably large even in
this technologically advanced day and age. The
main reason for this is that sampling is often
left to personnel who do not understand its
critical importance in generating representative
samples and subsequent analyses that are
truly meaningful and can be relied upon to
make correct grade control decisions. It is not
good enough just to collect some material and
send it back to the laboratory for analysis if
the sample is not representative in the first
place. The whole exercise is simply a waste of
time and can lead to suboptimal recovery in
processing plans, reduced mine life, and loss
of sales revenue. It is, therefore, critical to
ensure that samples be free of significant bias
The Journal of The Southern African Institute of Mining and Metallurgy

Components of sampling error

It is instructive to list the key components of
sampling error (Gy2, Pitard3) to better
understand how to eliminate or minimize
them. The key components are as follows:
Fundamental, grouping, and segregation
errors
Long-range quality fluctuation error
Periodic quality fluctuation error
Weighting error
Increment delimitation error
Increment extraction error
Accessory errors.

* Carbon Steel Materials, CSIRO Minerals Down

Under National Research Flagship, Australia.
The Southern African Institute of Mining and
Metallurgy, 2010. SA ISSN 0038223X/3.00 +
0.00. This paper was first published at the SAIMM
Conference, Fourth World Conference on Sampling
& Blending, 2123 October 2009.

VOLUME 110

JUNE 2010

269

Synopsis

P
a
p
e
r

Sampling mineral commoditiesthe good, the bad, and the ugly

Of these errors, the weighting, increment delimitation,
increment extraction, and accessory errors lead to bias and
need to be eliminated, while the others including the
fundamental, grouping and segregation, long-range quality
fluctuation, and periodic quality fluctuation errors need to be
reduced to achieve acceptable precision. Minimizing or even
better eliminating bias is critical, because bias cannot be
eliminated once it is present. There is no point in being
precisely incorrect. Sources of bias that can be eliminated
include incorrect delimitation and extraction of increments,
sample spillage and sample contamination, while sources of
bias that need to be minimized include changes in moisture
content and dust losses.
Elimination of increment delimitation and extraction
errors is particularly important, because, together, they are
responsible for a large proportion of incorrectly designed
sampling systems. These two errors arise from incorrect,
increment delineation due to bad sample cutter design and
subsequent extraction of the increment that is collected. The
delimitation error is eliminated if the cutter geometry is
correct, e.g., selection of a parallel section of ore on a
conveyor belt or radial cutter lips on a rotating Vezin cutter
(see Figure 1), whereas the extraction error is eliminated if
increments are completely extracted without any sample loss
(e.g., no reflux or loss of sample from the cutter aperture).
Accessory errors are also common in badly designed and
maintained sampling systems and where operator training is
poor and procedures are not regularly checked. Such errors
include sample contamination, loss of sample such as dust
(see Figure 2), alteration of the chemical composition of
samples, loss of moisture (see Figure 3), particle degradation,
operator mistakes, such as mixing up sample labels, fraud
and sabotage. An example of fraud is collecting a timed
sample well in advance of the time at which it is scheduled to
be taken.

(a)

(b)

Figure 2Examples of unacceptable sample loss due to (a) a hole in the

side of a chute and (b) windage underneath a jaw crusher

Blast hole sampling

Blast hole sampling is quite problematic for a range of
reasons, including segregation of the drill cuttings (see
Figure 4), non-uniform thickness of cuttings in the cone and
correctly allowing for sub drill cuttings generated by drilling

Figure 1Correctly designed rotating Vezin cutter with radial cutter lips

270

JUNE 2010

VOLUME 110

Figure 3Moisture loss from an open slow moving conveyor transferring

primary increments from the primary cutter to the secondary cutter

Figure 4Cone of blast hole cuttings showing the large variability in the
cuttings and hence the difficulty in extracting representative samples
The Journal of The Southern African Institute of Mining and Metallurgy

Sampling mineral commoditiesthe good, the bad, and the ugly

Taking vertical slices from one of more slots dug into

the side of the cone of cuttings
Taking two complete channel cuts through the whole
cone (see Figure 5(a))
Extracting one or a number of complete radial sectors
from the cone of cuttings using sector cutters (see
Figure 5(b))
Automatic collection and division of the cuttings on the
drill rig using compressed air/suction and a cyclone to
separate out the cuttings.
Of the above options, whereas the last method is in
principle the best approach, a substantial proportion of the
cuttings is not picked up by the sample collection system
(usually the coarser particles) and the cyclone and sample
division system tends to clog up when wet cuttings are
encountered. In addition, the first method is very subjective
about where the vertical cuts are taken, leaving channel
sampling and use of sector cutters as the best approach for
sampling the cuttings from blast holes.

(a)

(b)

Figure 5Sampling of blast hole cuttings by (a) manual extraction of

two channel cuts 180 apart and (b) using a sector cutter during drilling
The Journal of The Southern African Institute of Mining and Metallurgy

Plant sampling
Sampling particulate material
The best location for sampling a process stream in a mineral
processing plant is at the discharge point of a conveyor belt
or chute where the complete stream can be intersected at
regular intervals. The design rules for such a sampling
system are as follows:
The sample cutter must take a complete cross-section
of the process stream
The cutting time at each point in the stream must be
equal
The cutter should intersect the stream in a plane
normal to the stream trajectory
The plane of the cutter aperture must not be vertical or
near vertical, because particles that strike the inside
edge of the cutter lips (and hence which should appear
in the sample) are deflected away from the cutter
aperture into the reject chute
The cutter speed must be uniform duting its traverse
through the stream
The cutter aperture must be at least three times the
nominal top size (d) of the particles being sampled, i.e.,
3d
The cutter speed must not exceed 0.6 m/s unless the
cutter aperture exceeds 3d
There must be no contamination of the sample or
change in its quality
Bucket-type cutters must have sufficient capacity to
accommodate the entire increment without any
overflow or loss of sample
The sample cutter must be non-restrictive, self-clearing
and discharge completely each increment, and for high
capacity streams have a large cutter body and
streamline design to eliminate sample reflux
Belt scrapers need to be located so that the scrapings
are intersected by the sample cutter.
A well designed cross-stream sample cutter with a large
body to accommodate high capacity streams is shown in
Figure 6(a). The back of the cutter is designed to direct
incoming material downwards towards the exit chute at the
bottom of the cutter, thereby minimizing build-up of material
in the cutter throat during sampling and hence sample reflux.
On the other hand, the cutter shown in Figure 6(b) has
inadequate capacity for high capacity streams and sample
reflux from the cutter aperture is evident.
Other examples of poorly designed sample cutters are
shown in Figure 7. The primary cutter in Figure 7(a) has
vertical cutter lips and very limited capacity to accommodate
the sample being collected; the cutter in Figure 7(b) is a
bucket-type secondary cutter with a number of problems. In
the latter case, the gap between the dump gate at the bottom
of the cutter and the cutter body is excessive, resulting in
sample loss from the cutter during its traverse. In addition,
there is substantial build-up on top of the cutter that can
potentially contaminate the sample when it is dumped at the
end of the cutter traverse.
The cutters, discussed so far, have been cross-stream
cutters where the cutter passes through a falling stream of
material, and, hence, it is reasonably straightforward to
VOLUME 110

JUNE 2010

271

below the design depth of the bench. In addition, because the

top of the bench is fractured due to prior blasting, relatively
few cuttings are generated from the top few metres of the
blast hole, so the cone of cuttings is biased towards the
material in the middle and lower sections of the bench.
Separate reverse circulation (RC) drilling for grade control
purposes is currently considered to be the best solution
(Pitard4), although this adds considerably to the overall
costs. The drill cuttings are easier to collect using this
approach due to the smaller sample masses involved and
multiple benches can be drilled at the same time.
Confining the discussion to sampling of blast hole cones,
a number of methods are used as follows:

J
o
u
r
n
a
l

P
a
p
e
r

Sampling mineral commoditiesthe good, the bad, and the ugly

(a)

(b)

Figure 6Examples of cross-stream primary cutters that are (a) well

designed for sampling high capacity streams and (b) of insufficient
capacity resulting in sample reflux from the cutter aperture

ensure that the cutter intercepts the complete stream. In

addition, it is reasonably easy to check visually that crossstream cutters are operating correctly, (i.e., that increment
delimitation and extraction are correct), thereby reducing the
need for expensive bias tests. In contrast, cross-belt cutters
take samples directly off conveyor belts, but it is very difficult
to check visually that they are operating correctly and remove
a complete and correctly delimited cross-section of material
from the conveyor belt. Often such cutters leave a layer of
material on the conveyor belt due to wear/incorrect
adjustment of the skirts on the bottom of the cutter or the
deliberate action of maintenance staff who leave a gap
between the bottom of the cutter and the conveyor to
eliminate possible damage to the conveyor belt.
Consequently, expensive bias tests are required to assess
their performance. For the above reasons, cross-stream
cutters are recommended in preference to cross-belt cutters,
particularly for high capacity streams.
Examples of cross-belt cutters are shown in Figures 8 and
9. In the cross-belt cutter installation, shown in Figure 8, the
cutter is fully enclosed, so it is impossible to check its
operation. Figure 9 shows two poorly designed cross-belt
cutters. In Figure 9(a), the conveyor belt profile does not
match the trajectory of the cutter, so there are gaps where
material on the conveyor is not collected. The design of the
cross-belt cutter shown in Figure 9(b) is much worse and is
no more than a paddle that removes some material from the
conveyor belt. Analysing samples collected by a device such
as this is really a complete waste of time because there is no
way that the samples will be representative.
Manual sampling from the top of conveyor belts (see
Figure 10) is also totally unacceptable for a number of
reasons. Firstly, there are serious safety issues with this
practice and, secondly, it is not possible to take a complete
cross-section of the material on the conveyor belt.
Consequently, the sample collected will not be representative.

Sampling dry concentrates

(a)

The requirements for sampling dry concentrates are

essentially the same as those for sampling particulate
material, except that the particle size is much smaller. Hence,
a cutter that takes a complete cross-section of the concentrate

(b)

Figure 7Examples of poorly designed and maintained primary and

secondary cross-stream cutters. The primary cutter in (a) has vertical
cutter lips and very limited capacity; there is sample loss from the gate
at the bottom of the secondary cutter in (b) as well as build-up on top of
the cutter

272

JUNE 2010

VOLUME 110

Figure 8Fully enclosed cross-belt sampler installation

The Journal of The Southern African Institute of Mining and Metallurgy

Sampling mineral commoditiesthe good, the bad, and the ugly

(a)

stream at a transfer point is required, the recommended

minimum cutter aperture being 10 mm. However, in situ dip
sampling from holding tanks and vessels is carried out as
shown in Figure 11, but this is not recommended because the
concentrate will almost certainly segregate in the vessel,
resulting in a non representative sample.
The only acceptable way of extracting representative
samples in situ from vessels is full depth sampling where a
number of full vertical columns are extracted from the vessel
or sampling from a moving stream as the vessel is filled or
emptied. In the latter case, dry concentrates are occasionally
sampled while they are being conveyed via feeder tubes into
vessels, but, once again, this is satisfactory only if the cutter
extracts a full cross-section of the concentrate stream. Using
a pneumatic sampling device attached to the side of the
feeder tube (as shown in Figure 12) is not satisfactory,
because the full cross-section of the stream is not sampled

(b)
Figure 9Examples of poorly designed cross-belt cutters. In (a), the
conveyor profile does not match the cutter trajectory; in (b) the cutter is
simply a paddle

Figure 11Manual sampling of vessels using a ladle is not

recommended due to segregation of the material in the vessel

The Journal of The Southern African Institute of Mining and Metallurgy

Figure 12Sampling dry concentrates pneumatically from the side of a

feeder tube does not extract a full cross-section of the concentrate
stream and generates a biased sample
VOLUME 110

JUNE 2010

273

Figure 10Manual sampling from the top of a conveyor belt raises

serious safety concerns and in any event does not provide a representative sample

J
o
u
r
n
a
l

P
a
p
e
r

Sampling mineral commoditiesthe good, the bad, and the ugly

and the suction action of the sampler will preferentially collect
the smaller, less dense, particles, resulting in a seriously
biased sample.

Sampling slurries
As for sampling particulate materials and dry concentrates,
the best location for sampling a slurry is at a transfer point
where a cross-stream cutter can gain access to the full slurry
stream and take a complete cross-section of the stream (see
Figure 13(a)), thereby providing a representative sample.
Taking samples using a ladle that does not intercept the full
slurry stream as shown in Figure 13(b) is not acceptable. One
key difference with slurries is the need to ensure that
dribbles from underneath pipes and launders are also
intercepted by the sample cutter, as illustrated in Figure 14.
Although it is common practice in industry, sampling of
slurries via taps on the side of pipes is not satisfactory for
extracting representative samples because segregation and
laminar flow of slurries in pipes are common. Examples of
this unsatisfacory practice are shown in Figure 15. In
addition, pressure pipe samplers used for extracting samples
for on-line analysers (see Figure 16) do not extract a full
cross-section of the slurry stream, so they are prone to bias
as well.

(a)

Sampling trucks, railway wagons, and stockpiles

Sampling of material that is stationary (e.g., in trucks,
railway wagons and stockpiles) is particularly problematic
because in many cases it is not possible to ensure that all
parts of the material being sampled have an equal probability
of being collected and becoming part of the final sample for
analysis, particularly for large stockpiles. Other than
sampling the material from a moving stream, when the
material is transferred to or from the truck, railway wagon or

(b)
Figure 13Examples of (a) a full cross-stream slurry sampler and (b)
manual sampling of a slurry stream using a ladle that is too small for
the stream being sampled

Spray guard

(a) Correct

(b) Incorrect

Stre
am

Stream
Increment
(c) Correct
Figure 14Examples of correct and incorrect cutter designs for sampling slurries

274

JUNE 2010

VOLUME 110

The Journal of The Southern African Institute of Mining and Metallurgy

Sampling mineral commoditiesthe good, the bad, and the ugly

(a)

(b)

preparation laboratory prior to chemical analysis is as

important as the primary and subsequent sampling steps in
ensuring that the final sample submitted for analysis is
representative. In fact, there are many instances where the
veracity of the sample delivered to the sample preparation
laboratory is totally destroyed in the preparation process by
dividing the sample down to a sample mass that is too small
for the nominal top size of the material being prepared, (i.e.,
the minimum sample mass requirements are not observed,
(Gy2 and Pitard3)). This is illustrated in Figure 18, where the
sample mass has been divided down to about 300 g for
pulverization, but the sample is far too coarse(~10 mm) for
division down to this sample mass, resulting in poor division
precision. The sample should have been crushed to a much
smaller particle size (~3 mm) prior to division.
Other sample preparation problems that need to be
eliminated include:

Discarding part of the sample because it is too heavy

Incorrect use of sample dividers, e.g., riffles
Sample loss in dust extraction systems
Cross contamination between samples.

Robotic sample preparation systems are increasingly

being used to overcome occupational health and safety
restrictions on sample masses and the vagaries of manual
sample preparation, but they still need to be carefully

Figure 15Taps from the side of slurry pipes do not extract a full crosssection of the slurry stream, so the samples will not be representative

(a)

Figure 16Pressure pipe samplers do not extract a full cross-section of

the slurry stream, so the samples taken are unreliable

Sample preparation
The preparation of mine and/or plant samples in the sample
The Journal of The Southern African Institute of Mining and Metallurgy

(b)
Figure 17Sampling from (a) the top of railway wagons and (b) the side
of stockpiles will not provide representative samples
VOLUME 110

JUNE 2010

275

stockpile, the only way of obtaining representative samples is

to drive an auger or spear down to the bottom of the truck,
railway wagon or stockpile and then extract the full vertical
column of material without any sample loss. Collecting
material from the tops of railway wagons or from the side of
a stockpile (as shown in Figure 17) will not provide
representative samples.

J
o
u
r
n
a
l

P
a
p
e
r

Sampling mineral commoditiesthe good, the bad, and the ugly

designed to meet all the requirements for correct sample
preparation, in particular:

Particle size after crushing

Minimum sample mass requirements
Minimal sample loss in rotary driers
Minimal cross contamination of samples.

System verification

Figure 18Example of a sample ready for pulverization prior to

analysis, but the sample mass is far too small for the nominal top size
of the material

A well designed sampling system will facilitate inspection to

verify conformance to correct design principles. For this
purpose, large and easily accessible inspection hatches are
required (with safety mesh installed to prevent accidental
injury) for inspection of key items of equipment, including
cutters, feeders, crushers, dividers, etc. The ability to monitor
sampling ratio and increment mass extraction ratio also
provides valuable information. The key items that need to be
checked when verifying system performance include:

Cutter speed
Uniformity of cutter speed while cutting the ore stream
Number of cuts
Size and geometry of cutter apertures
Worn and/or missing cutter lips
Build-up and/or blockages in cutter apertures and
chutes (see Figure 19)
Reflux from cutter apertures
Ingress of extraneous material when the cutter is
parked
Holes in chutes and bins resulting in sample loss
Increment/sample mass
Particle size.

Conclusion
Despite the availability of training courses, conferences, and
standards on correct sampling practices, many examples of
poor sampling practices can still be found in industry, usually
because the responsibility for sampling is entrusted to
personnel who do not fully appreciate the significance and
importance of sampling. Cost is often the main driving force
rather than whether the samples collected are meaningful,
which seriously undermines the reliability of the final
analyses. Company management needs to recognize this and
act accordingly to ensure that sampling systems are well
designed and provide representative samples for analysis,
thereby maximizing resource utilization and minimizing
financial risks.

(a)

(b)

References
1. HOLMES, R.J. Correct Sampling and Measurement The Foundation of
Metallurgical Accounting, Chemometrics and Intelligent Laboratory
Systems, vol. 74, 2004. pp. 7183.
2. GY, P.M. Sampling of Particulate Materials. Theory and Practice, 2nd
Edition, Elsevier, Amsterdam, 1982.
3. PITARD, F.F. Pierre Gys Sampling Theory and Sampling Practice, 2nd
Edition, CRC Press Inc, Florida, 1993.

Figure 19Examples of blocked sample cutters identified while

verifying conformance with correct sampling principles for (a) a crossstream cutter and (b) a Vezin divider

276

JUNE 2010

VOLUME 110

4. PITARD, F.F. Blasthole sampling for grade controlThe many problems and
solutions, Proceedings Sampling 2008, 2729 May 2008, Perth, Australia
The Australasian Institute of Mining and Metallurgy, Melbourne.

The Journal of The Southern African Institute of Mining and Metallurgy