SAMPLING

SAMPLING
 Art of cutting a small portion of material from a large lot
and transferring it to the analyzer

 The result is not better than the sample that it is based on

 Sample must be representative

 Theory that combines both technical and statistical parts

of sampling has been developed by Pierre Gy:
Theory of sampling
 Sampling is the process of estimating the mineral content and other physical and chemical
characteristics of a mass of rock by averaging the characteristics in a number of much smaller
portions derived from this mass. The mineral content and other characteristics of the smaller
portions are obtained by assaying them individually. 
 Samples should be equal in mass and be taken at equal distances from one another in order to be
truly representative of their parent body; i.e. their zones of influence must be equal. It is not easy
to take samples from within an ore body and, unless diamond drilling is resorted to, samples are
usually taken from around the perimeter of blocks of ore, particularly in the case of tabular
bodies, as is the practise with most gold, platinum, chrome, antinomy and amosite asbestos
deposits within Zimbabwe `
Sampling
 The sample is taken for analysis of assay values, mineral composition, texture or a field specimen. 
 Economic mineral deposits are sampled to ascertain the grade of minerals, which is sometimes
decisive for the commercial evaluation of a deposit.
 The results of sampling furnish the necessary information for determining the mean thickness
of mineral bodies and the average content of the useful constituent therein; they help study
the properties of useful minerals, delineate the mineral bodies, determine correlation of
individual constituents and elements in the ore, establish the priority in mining the minerals,
their losses and dilution during exploitation. 
Cont.
Sampling may be chemical, mineralogical, technical and technological.
1. Chemical: Samples are taken for determining the content of useful and secondary components.
2. Mineralogical: It is done to ascertain the mineral and petrographical composition of the
mineral. It helps to establish the origin of the deposit, the dependences governing grade variations
and also to plan the ore dressing and beneficiation.
3. Technical: Samples are taken to study the technical properties of the raw material, which does
not require metallurgical or chemical treatment. Thus in the case of building stone, it is their
bearing capacity, in the case of asbestos – the length, strength and flexibility of fibres; mica – the
size; sand and gravel – grain size distribution, etc.
4. Technological: Samples are collected for the study of the technological properties of the raw
material in the course of its beneficiation and processing. In the case of coals, for example,
apart from their beneficiation properties, it is necessary to establish their coking qualities and
the yield of liquid products.
Cont.
 Besides these above-mentioned types of samplings, there exist geophysical sampling, which enable one to
define the physical properties of rocks and ores without collecting actual samples; and in the case of
radioactive and some other ores, to determine the metal content therein.
The process of sampling falls into several stages:
1. Taking of samples,
2. Their processing,
3. Laboratory studies of the sample (assaying), and
4. Analysis of the laboratory findings.
Purpose of sampling

In order to ascertain whether an ore body is likely to prove to be a

workable proposition, it is first necessary to ascertain its size and
value by exposing small portions of ore for sampling by means of
pitting, trenching, boring or mining. Assay values thus obtained are
combined with morphological information obtained from maps and
surveys to arrive at an overall estimate of the payable and unpayable
mineral reserves in the orebody.
 PLANNING OF SAMPLING
1. GATHERING OF INFORMATION

 What are the analytes to be determined?

 What kind of estimates are needed?
 Average (hour, day, shift, batch, shipment, etc.)
 Distribution (heterogeneity) of the determinand in the lot
 Highest or lowest values
 Is there available useful a prior information (variance estimates, unit
costs)?
 Is all the necessary personnel and equipment available?
 What is the maximum cost or uncertainty level of the investigation?
 PLANNING OF SAMPLING
2. DECISIONS TO BE MADE

 Manual vs. automatic sampling

 Sampling frequency
 Sample sizes
 Sampling locations
 Individual vs. composite samples
 Sampling strategy
 Random selection
 Stratified random selection
 Systematic stratified selection
 Training
Sampling methods for grade control
 The following methods are generally used: point or spot, face or lump, channel or trench, drill core
shot hole, chip and bulk sampling.
 Channel sampling: The channel is usually positioned perpendicular to the structure (mineralised
zone) which is to be sampled a groove is cut across the strike of the rock formation or ore body/reef
using a diamond saw-cutter. A chisel is then used to chip out the rock material in between the
groove. The samples are collected at standard sample length as governed by the mine standard
(classical geostatistics or geological).This sampling method is most preferred for evaluation and
mineral reserves/resources estimations to chip sampling because is more effective and is auditable.
 Chip sampling: this sampling is done by a hammer and a chisel. No groove is cut, it is superficial. This
is mainly used were quick results are needed e.g. in guiding developments and quick grade
evaluations.
 Grab sampling: broken muck samples are collected by shovel or hand at certain time intervals, or
sampling intervals, or randomly so as to get the indication of the mineral value of the respective
ground. This method is applied on stockpiles, tramming, drawpoints, hoists, trucks and conveyors.
Grab sampling is often done for grade control purposes.
Cont.
 Sludge sampling: sludge or pulp from jackhammer drilling is collected during drilling at known sample
intervals as the hole deepens. The pulp is assayed and results are used to roughly indicate/delineate the ore
body limits on stopes. These results are not used for mineral reserves/resources estimation due to high
possible contamination levels- not reliable.
 Bulk sampling: large quantities than normal are collected for analyses, mainly for metallurgical investigation
or evaluation(e.g. in alluvial deposits were mineral distribution are random). Bulk sampling is an essential
part of exploration and evaluation of ores. Large samples are required to check the accuracy of the drill
samples and to obtain the needed information on the crushing and grinding characteristics of the ore as well
as for use in detailed and large-scale metallurgical tests. In natural ores, bulk samples give information
regarding the nature and quantities of various sized ore products that can be shipped
Cont.
 Point or Spot Sampling: consists of taking a number of equal portions of a mineral at points
distributed in a regular grid over a work face or a mineral exposure. Sampling may be done from a
pile of the mined mineral and from mine cars. These samples are, in the main, destined for
chemical assaying. The number of points making up an individual sample depends upon the
uniformity of mineralisation. 
 Face or Lump Sampling: This method of sample taking may be referred to the group of point
sampling. One to three lumps of rock are gathered in the face or taken from a pile of broken
mineral with the purpose of determining the mineral, and sometimes, also the chemical
composition. Face sampling is a very simple, quick and cheap procedure, but the taking sample is
often done subjectively and for this reason the accuracy of the method is rather low. This type of
sampling is employed in lump wise metallometeric surveying. Large pieces are occasionally
collected for the purpose of determining the physical properties of the mineral.
General sampling procedure
Equipment
 Base map/ survey plan with pegs, pan, clinorule, chisel, diamond saw cutter (if channel sampling),
4lb hammer, sample bags, tickets, strings, crayon, paint and brush, canvas bag, notebook, pen and
pencil 
Protective clothing
 Overall/ worksuit, goggles, gloves, hardhat, gumboots, caplamp belt
General procedure
 Having identified the area to be sampled on plan in the office, gather the necessary equipment and
PPE. Inform the Miner in charge of the section when you get into the section of your intention. This
is done so that the area is made safe and resources made available for sampling to proceed e.g.
compressed air for channel sampling / water to clean the face.
 Inspect the area to make sure the workplace is safe, bad hangings are barred down with help of
mining official (if any)
Sampling procedure
Identify the area to be sampled using the reference pegs
Mark out the sampling position and sample lengths using the crayon/chalk noting
the strike of the shear/ reef. Samples are taken at right-angle to the strike
Cut the groves and chip out the fragments with a chisel
Bag the samples and ticket them
Sketch the sampled area and record the ticket numbers
Securely tie the sample bags and put into the canvas bag and transport to the
laboratory
Fill the sample register in duplicate and submit together with samples at the
laboratory and keep one for records
 Sampling outcrops
One method of sampling is to cut grooves at right angles to the strike across
the exposed strata on surface, segregating samples according to the contrast
in appearance of the stratified minerals.
If larger samples are required, or if the outcrop is partially or completely
covered, trenching will be necessary, the overburden of top soil being put on
one side.
In extensive areas where the nature of the ground and soil are suitable and if
expenditure is warranted, a mechanical trencher may be employed to scoop
bulk samples.
Open Pit Sampling
The sampling of opencast workings varies according to the nature of the
mineral, the conformation of the deposit and the mining programme.
Usually new ore bodies of this kind will have been fully drilled by deep
prospect boreholes and foreseeable reserves thus delimited before
domestic sampling starts.
Sampling the floor of a working open pit is not popular because of debris
lying on it but should the method be used, grooves or trenches at regular
intervals may be excavated in one direction at right angles, across the
bands or if no banding is apparent along the length and breadth in a
square pattern. If the outlay on continuous trenches is not justified
samples could be taken from pits at the corners only of a square grid
pattern.
Cont.
Also as a rule, it is not easy or safe to attempt to sample the side-
walls of an open pit, and domestic sampling for grade purposes is
often therefore carried out by catching the chips and sludge from
blasting bore-holes drilled around the periphery of the pit. The
taking of grab samples from the broken ore lying on the ore body, on
conveyor belts or from cars and railway trucks is common practise.
In recent years sampling has been facilitated by the use of electronic
instruments which measure the concentration of certain minerals at
the pit face, or in the assay office in the place of laborious and more
expensive assay procedures.
Sampling a coal mine
Underground sampling practise in coal mines is less standardised
than that of gold mines. It may vary as conditions differ from mine to
mine, at greenside colliery of Apex Mines, samplings are made about
65m apart , a rubber sheet is placed on the floor to catch the
fragments that drop. The chipper starts sampling from the floor
upwards, the whole mining height being treated as one sample.
Sampling dishes are 20cm in width to match the width of groove.
Samples are placed in 25kg bags.
Sampling a Chrome mine
The initial sampling is usually done by bore-holes drilling and
subsequent analysis of cores. Where possible, the sampling of seams
exposed in prospect trenches, tunnels and adits is also undertaken.
For this information, the economic seam width and value decided.
Thereafter because of the general consistency of grade, very little
routine underground sampling is done. End product sampling is
however, carried out judiciously by taking regular samples daily from
the end – product after it has been loaded onto lorries or into railway
trucks.
Stope sampling
Before the sampling gang goes underground, clear and concise instructions
must be given regarding the locality of the stope face which has to be sampled.
A pencil tracing should be made on the 1:200 stope plan which shows all
numbered survey pegs in the vicinity and which portion of the face is to be
sampled.
The general sampling procedure has already been outlined in sampling
procedure. In stope sampling the location of the first section to be sampled is
determined by measuring half the sampling interval from the top of the stope
face, which must be located from a numbered peg, or, if this is not possible,
from another established position such as the top or bottom of a raise/winze
connection in the drive.
Quality control
In order to control the quality of the assay results several practices are done which include but not limited to the following:
 Standard samples
 Repeats
 Blank
 Duplicate
 Core Sample splitting
 Sampling standard procedure enforcement and supervision
 Training
 Statistical Data analyses and recommendations
 Spiked samples consist of base material into which spiking grains were voluntarily introduced. They are used to quantitatively monitor
the effectiveness of a processing laboratory at recovering and identifying specific indicator miner
 After assaying the values are plotted on the assay level plan on their rightful positions on space. Hence x,y,z of the sample
must be known. It is this assay information together with geology model that are used for ore reserves estimations
and ,mine planning and scheduling.  
• Analogously, the homogeneity of a pulp can be checked by taking a number of
replicate portions for analysis from the same pulp and calculating the precision
of replication.
• Analytical precision and accuracy are best established and maintained through
the use of reference standard samples and analytical control samples. Certified
standard samples of mineral materials are available from mineral industry
suppliers and some governmental agencies and professional or trade groups
in several countries, but such standard materials are expensive and limited in
quantity. Nevertheless, such standard materials may help to provide initial
calibration
• Matrix-matched control samples to provide batch-to-batch and day-to-day
calibration and quality control in the laboratory can be prepared and blended
from the reject samples from the project. The analytical development of these in-
house control samples can be calibrated to the certified standards. Ideally, every
set of analyses produced by the laboratory should have one or more control
samples and duplicates included for quality control and assurance
Sources of error
Contamination- either during sampling, transportation or assaying.
 Fatigue
 Human error
 Laboratory- mix up or machine calibration
 Bias toward softer material during chipping or fine material during grab sampling
 Salting
 Errors in sampling mislead judgement of the information end user, the geologist, and hence the
entire management (mining/ exploration).Hence quality control and quality analyses measures
should be crafted , implemented and enforced to guard against making catastrophic decisions.
Gy Pierre
SAMPLING ERRORS
SAMPLING ERRORS
Pierre Gy’s theory addresses seven types of sampling error
and offers proven techniques for their minimization.
The seven major categories of sampling error cover
differences within samples. Other differences can exist, such
as,within space (covered by geostatistics) and within
time(covered by chronostatistics).
The internal sample errors are:
Fundamental Error:
Fundamental Error: This is loss of precision inherent in the
sample due to chemical and physical composition and includes
particle size distribution.
Incurred due to constitutional heteregonity(particle size
distribution)
It can be reduced by decreasing the diameter of the largest
particles or by increasing the sample mass.
Grouping and Segregation
Error:
Error due to non-random distribution of particles, usually by
gravity.
Results from distributional heteroginity(clustering)
It can be minimized by compositing an analytical sample from
many randomly selected increments or by properly
homogenizing and splitting the sample.
Long-range Heterogeneity
Error:
This is fluctuating and non-random.
There is correlation for samples that are in close range
It is spatial and maybe identified by variographic experiments
and can be reduced by taking many increments(portion of a
sample) to form the sample.
.
Periodic Heterogeneity Error:
This fluctuation error is temporal or spatial in character and
can be minimized by compositing samples correctly. Cyclic
temporal/ spatial distribution

Increment Delimitation Error:

Error tied to inappropriate sampling design and the wrong
choice of equipment.
.
Increment Extraction Error:
This error occurs when the sampling procedure fails to
precisely extract the intended increment. Incorrect material
collection
Well-designed sampling equipment and good protocols are
crucial.
Preparation Error:
This error is the expression of loss, contamination,
and alteration of a sample or subsample. Field and
laboratory techniques exist to address this problem,
Human error
 Incorrect Increment and Sample Preparation
Errors
Contamination (extraneous material in sample)
Losses
Alteration of chemical composition
Alteration of physical composition (agglomeration, breaking of
particles, moisture, etc.)
Involuntary mistakes (mixed sample numbers, lack of knowledge,
negligence)
Deliberate faults (salting of gold ores, deliberate errors in increment
delimitation, forgery, etc.)
.
 Global
Global Estimation
Estimation Error
Error
GEE
GEE

Total Sampling Error Total Analytical Error

TSE TAE

Point
PointMaterialization
Materialization Error
Error Weighting Error
PME
PME SWE

Increment
Increment Delimi-
Delimi- Increment
IncrementExtraction
Extraction Increment
Incrementand
andSample
Sample
tation
tationError
Error Error
Error Preparation
PreparationError
Error
IDE
IDE IXE
IXE IPE
IPE
Point Selection Error
PSE

Long Range Periodic

Point Selection Error Point Selection Error
PSE1 PSE2

Fundamental Grouping and

Sampling Error Segregation Error
FSE GSE
GEE=TSE +TAE
TSE= (PSE+FSE+GSE)+(IDE+IXE+IPE)+SWE

Error components of analytical determination according to P.Gy

 LOT
Primary Secondary Analysis Result
sample sample
m
s1 s2 s3 sx
x
(s)
Propagation of errors: s x  i
s 2

Example: s x  (5%) 2  (2%) 2  (1%) 2  30(%) 2  5.5%

GOAL: x = m

Analytical process usually contains several sampling and

sample preparation steps
 Weighting error
Sample Concentration Volume c·V
No. 3
mg/l m g
1 6.25 4.58 28.63
2 4.36 3.71 16.18
3 5.58 5.20 29.02
4 4.64 5.71 26.49
5 4.86 4 .54 22.06
6 3.65 6.78 24.75
7 3.73 7.12 26.56
8 5.98 5.81 34.74
9 4.96 5.86 29.06
Mean 4.89 5.48 26.39
Sum 44.01 49.3 237.49

å c iV i
Weighted mean concentration: = 4.82 mg/l
å Vi
Total emission estimate (unweighted):
M = 9×c ×V = 9×4 .89 × g / m ×5 .48 ×m = 241.13g
3 3

Total emission estimate (weighted):

M w = 9×c w ×V = 9×4 .82 × g / m ×5 .48 ×m = 237.72g
3 3

Weighting error (in concentration): 0.07 mg/l

Weighting error (in total emission): 3.41 g
List seven (7) different sampling methods.
What are the characteristics of good samples?
What is difference between valuation and evaluation?
What is the purpose of sampling?
What decisions are to be made when planning for sampling?
A recap
List any 5 quality control measures.
5 sources of error which results in contamination either during
sampling, transportation or assaying.
7 Gy Pierre sampling errors
The equation for propagation of errors
 Weighting error
Sample Concentration Volume c·V
No. 3
mg/l m g
1 6.25 4.58 28.63
2 4.36 3.71 16.18
3 5.58 5.20 29.02
4 4.64 5.71 26.49
5 4.86 4 .54 22.06
6 3.65 6.78 24.75
7 3.73 7.12 26.56
8 5.98 5.81 34.74
9 4.96 5.86 29.06
Mean 4.89 5.48 26.39
Sum 44.01 49.3 237.49

å c iV i
Weighted mean concentration: = 4.82 mg/l
å Vi
Total emission estimate (unweighted):
M = 9×c ×V = 9×4 .89 × g / m ×5 .48 ×m = 241.13g
3 3

Total emission estimate (weighted):

M w = 9×c w ×V = 9×4 .82 × g / m ×5 .48 ×m = 237.72g
3 3

Weighting error (in concentration): 0.07 mg/l

Weighting error (in total emission): 3.41 g
Estimation of Fundamental Sampling
Error by Using Poisson Distribution
• Poisson distribution describes the random distribution of rare
events in a given interval.
• If ms is the number of critical particles in sample, the standard
Deviation expressed as the number of particles is

𝜎 𝑛 = √ 𝜇𝑛
 
(1)
• The relative standard deviation is

  1
𝜎 𝑟= (2)
√ 𝜇𝑛
The Fundamental Error (FE) variance σFE2 identified by Gy
(1982) is the ‘irreducible minimum’ of sampling errors, is the only error
that can be estimated before performing the sampling and arises from
the inherent variability of the material being sampled.
 According to François-Bongarçon (1998), FE is ‘the smallest
achievable residual average error’, a loss of precision inherent in the
sample due to physical and chemical composition as well as particle
size distribution. It arises because of two characteristics of broken ore
materials, namely the compositional heterogeneity and the
distributional heterogeneity.
 Example
Plant Manager: I am producing fine-ground limestone that is used
in paper mills for coating printing paper. According to their speci-
fication my product must not contain more than 5 particles/tonne
particles larger than 5 mm. How should I sample my product?

Sampling Expert: That is a bit too general a question. Let’s first

define our goal. Would 20 % relative standard deviation for the
coarse particles be sufficient?

Plant Manager: Yes.

Sampling Expert: Well, let’s consider the problem. We could use the
Poisson distribution to estimate the required sample size. Let’s see:
The maximum relative standard deviation sr = 20 % = 0.2. From
equation 2 we can estimate how many coarse particles there should be
in the sample to have this standard deviation
  1 1
𝑛= 2 = 2
=25
𝑠 𝑟 0. 2
If 1 tonne contains 5 coarse particles this result means that the primary
sample should be 25 tonnes. This is a good example of an impossible
sampling problem. Even though you could take a 25 tonne sample there
is no feasible technology to separate and count the coarse particles
from it. You shouldn’t try the traditional analytical approach in con-
trolling the quality of your product. Instead, if the specification is really
sensible, you forget the particle size analyzers and maintain the quality
of your product by process technological means, that is, you take care
that all equipment are regularly serviced and their high performance
maintained to guarantee the product quality.
 P. Gy’s Fundamental sampling error model

 
According to Gy (1982), if the mass of the pile of broken
ore (100s of kg) is very large compared to the sample mass
(a few kg), the variance of the fundamental error may be
expressed as:
 1 1
 

2 3
𝜎 =𝐾 𝑑 (
𝐹𝐸 − ) 𝑁
Where:
 
𝑀𝑆 𝑀 𝐿
Ms =Mass of sample measured in grams.
ML = Is the mass of material from which Ms is taken, measured in
grams
= Nominal size of fragments in the sample
K = Is a constant for any given ore
 SAMPLING CONSTANT, K

 K

Mineralogical composition factor

liberation factor
Granulometric factor
shape factor
SIMPLIFIED FORMULA
f is a shape factor that relates the volume and the diameter of the
particles of broken ore to one another.
 
g is defined as the average fragment volume divided by the
nominal fragment volume .
c is a factor for a material consisting of two components,
approximately equal to the ratio of metal density (p) to the
dimensionless grade of the lot.
 
is a dimensionless number between 0 (no liberation) and 1
(complete liberation), which varies with the size of the
fragments and also depends on the nominal size of the fully
liberated metal grains.
 d d d d

f= 1 f= 0,524 f= 0,5 f= 0,1

default in most cases

Estimation of shape factor
d

L=d
L

𝐿
 
𝛽=
√ 𝑑
𝛽=1
 
 𝛽 max

Estimation of liberation factor for unliberated and

liberated particles
Estimation of size distribution factor, g

 
g=

 
Wide size distribution g = 0.25

Medium distribution g = 0.50

Narrow distribution g = 0.75
Identical particles g = 1.00
 Mineralogical composition factor,c

 
C=
 CONCLUSIONS
 Sampling uncertainty can be, and should be
estimated
 If the sampling uncertainty is not known it is questionable
whether the sample should be analyzed at all
 Sampling nearly always takes a significant part of the
total uncertainty budget
 Optimization of sampling and analytical procedures may
result significant savings, or better results, including
scales from laboratory procedures and process sampling
to large national surveys
Exercise:
Problem No. 1—What error is introduced when a
sample of given weight, MS, is taken from a pile of
broken ore? This problem is simply answered by
substituting the known factors into the original Gy’s
equation. Assume that dN= 1.25 cm. For
the sake of this exercise assume the following sampling conditions for a gold
bearing ore crushed to about 0.93 cm, with c= 16 000 000 for a gold-silver
amalgam, f= 0.5, g=0.25 and l= 0.000035.
QUESTIONS
How do you calculate the standard deviation and relative standard
deviation using Poisson distribution?
Write down the formula used to determine the constant on P.Gy’s
fundamental sampling error model and explain all the variables
present.