Data Collection and Handling - Mineral Resource Estimation - page 69

( 1995 ). The sampling theory and practice with application to
the mineral industry was developed by P. Gy, ( Sampling of
Particulate Materials, Theory and Practice 1982), with later
significant additions by F. Pitard ( 1993 ) and D. F. Bongarçon
( 1998a , b ), among others. A summary of sampling theory is
presented below in Sect. 5.2. The application of sampling
theory is to develop deposit-specific sampling protocols to
minimize sampling variances.
The analytical procedures used to analyze samples are
generally well known and controlled, but may still be a
source of error. It is necessary to develop a strict, compre-
hensive, and enforceable quality assurance and quality con-
trol (QA/QC) program. This should be independent of the
laboratory and should include analyzing duplicate samples
of pulp and coarse material, blanks, and known standards. A
good QA/QC program should reduce the laboratory errors to
2-5 % relative error, which is small relative to other errors in
resource estimation.

error rate or less on the checked information is considered
acceptable. More than a 2 or 3 % error rate generally triggers
a line-by-line check of the entire database.

5.2 Basics of Sampling Theory

The discussion presented here has been based on the
Centre for Compuational Geostatistics (CCG) Guidebook 2
(Neufeld 2005 ). Perfect measurements are not possible. The
relatively large mass of a sample must be reduced to a small
subsample of a few grams for the final chemical analysis.
There will always be a discrepancy between the content of
the lot, the original sample, and the assay sample. This dis-
crepancy is termed the sample error.
In sampling there are two forms of error: one that is
present due to the intrinsic properties of the material being
sampled and one that arises from improper sampling pro-
cedures and preparation. This section presents a brief re-
view of the concepts and guidelines that are used to design
sampling protocols that will minimize the errors introduced
through improper procedures. The goal is to estimate and
use the “fundamental error” that is always present. The
reader interested in this topic should refer to more detailed
discussions of P. Gy's Sampling Theory, for example in
Pitard ( 1993 ).

5.1.9 Sampling Database Construction

A computerized database is required for resource estima-
tion and presents another potential source of errors. There
may be transcription errors (more so if done manually) and
sometimes a lack of record-keeping. Inconsistencies in the
geologic database compared to the information originally
mapped can be consequential. They may be due to errors or
a decision to re-code certain drill hole intervals.
A quality control program on data input should be im-
plemented and should also include procedures to provide
an estimate of error rates in the database. There should be
safeguards in place against gross errors, such as in a per-
centage of grade in rock not allowed to be less than 0 % or
greater than 100 %, or greater than the maximum percentage
that can exist in the rock according to its mineralogy. Other
checks include the consistency of the sampled intervals
and the location of the drill holes within the project area.
Manual checks of original assay certificates, geologic logs,
and other information should be done on a routine basis and
as part of the quality control of the database. These checks
should include all relevant information, such as grades,
down-the-hole surveys and the surveyed drill hole collar
locations.
Databases, when audited either externally or internally,
are checked against the original information available, in-
cluding laboratory assay certificates; checked and signed
off geologic logs; and down the hole deviations and collar
information properly checked and signed off as well. It is
customary for auditors to check line by line and manually
verify about 10 % of the total information available in the
database; although actual practice varies, generally a 1 %

5.2.1 Definitions and Basic Concepts

The Fragment Size , d α (cm) , is the actual size of the frag-
ment, or average size of the fragments, in the increment
α. The Nominal Fragment Size , d (cm) , is defined as the
square mesh size that retains no more than 5 % of the over-
size material.
The Lot , L , is the amount of material from which
increments and samples are selected. A lot of material should
have well-defined boundaries: the content of a bag, truck,
railroad car, ship, etc. A lot is also referred to as a batch of
material. An Increment , I , is a group of fragments extracted
from a lot in a single operation of the sampling device.
The Sample is a part of a lot obtained by the reunion
of several increments and meant to represent the lot in fur-
ther calculations or operations. A sample must respect cer-
tain guidelines that Sampling Theory lays out. Sampling is
often carried out by progressive stages: a primary sample is
extracted from the lot, and then a secondary sample is ex-
tracted from the primary sample, and so on.
The Component is the constituent of the lot that can
be quantified by analysis. It may be a chemical or physical
component such as: a mineral content, water content, percent
fines, sulphur content, hardness, etc.