SIST-TP CEN/TR 14547:2006

SLOVENSKI STANDARD
SIST-TP CEN/TR 14547:2006
01-februar-2006
6KHPHY]RUþHQMD]DSRWUMHYDQMHVNODGQRVWLþLVWLQHL]GHONRYL]SOHPHQLWLKNRYLQSUL
WUHWMLVWUDQNL
Sampling schemes for third party conformity assessment of fineness in precious metal
articles
Probenahmeverfahren für die Konformitätsbewertung des Feingehaltes von
Gegenständen aus Edelmetall durch Dritte
Programmes d'échantillonnage pour une évaluation de la conformité du titre des
ouvrages en métaux précieux par une tierce partie
Ta slovenski standard je istoveten z: CEN/TR 14547:2005
ICS:
39.060 Nakit Jewellery
SIST-TP CEN/TR 14547:2006 en
2003-01.Slovenski inštitut za standardizacijo. Razmnoževanje celote ali delov tega standarda ni dovoljeno.

---------------------- Page: 1 ----------------------

SIST-TP CEN/TR 14547:2006

---------------------- Page: 2 ----------------------

SIST-TP CEN/TR 14547:2006
TECHNICAL REPORT
CEN/TR 14547
RAPPORT TECHNIQUE
TECHNISCHER BERICHT
November 2005
ICS 39.060

English Version
Sampling schemes for third party conformity assessment of
fineness in precious metal articles
Méthodes d'échantillonnage pour une évaluation de Probenahmeverfahren für die Konformitätsbewertung des
conformité du titre des articles en métaux précieux par une Feingehaltes von Gegenständen aus Edelmetall durch
tierce personne Dritte
This Technical Report was approved by CEN on 8 August 2005. It has been drawn up by the Technical Committee CEN/TC 283.
CEN members are the national standards bodies of Austria, Belgium, Cyprus, Czech Republic, Denmark, Estonia, Finland, France,
Germany, Greece, Hungary, Iceland, Ireland, Italy, Latvia, Lithuania, Luxembourg, Malta, Netherlands, Norway, Poland, Portugal, Slovakia,
Slovenia, Spain, Sweden, Switzerland and United Kingdom.
EUROPEAN COMMITTEE FOR STANDARDIZATION
COMITÉ EUROPÉEN DE NORMALISATION
EUROPÄISCHES KOMITEE FÜR NORMUNG
Management Centre: rue de Stassart, 36  B-1050 Brussels
© 2005 CEN All rights of exploitation in any form and by any means reserved Ref. No. CEN/TR 14547:2005: E
worldwide for CEN national Members.

---------------------- Page: 3 ----------------------

SIST-TP CEN/TR 14547:2006
CEN/TR 14547:2005 (E)
Contents
Foreword .3
Section one: General.4
1 Scope.4
2 Normative references.4
3 Terms and definitions.5
4 Symbols and abbreviations.6
5 Average outgoing quality limit.7
6 Switching between attributes and variables inspection .7
7 The credit principle.8
8 Classification by weight range.8
9 Role and use of 100% inspection of non-accepted lots.8
Section two : Choice of sampling plan .10
10 Choice between attributes and variables.10
11 Choice between the "s" method and the "σ" method .10
12 Choice of constant, c, for sampling by variables .10
13 Choice of sampling plan.11
Section three : Operation of a sampling scheme .12
14 Preliminary operations.12
15 Standard procedure for sampling by attributes .12
16 Standard procedure for the "s" method.13
17 Standard procedure for the "σσσσ" method.14
18 Procedure during continuing inspection.15
19 Normality and outliers.15
19.1 Normality.15
19.2 Outliers.15
20 Records.15
20.1 Measures of location and variation .15
20.2 Results of tests for outliers and departure from normality .16
20.3 Results of tests for heterogeneity and inconsistency of sample standard deviations .16
20.4 Significant test results.16
21 Operation of switching rules.16
21.1 The start of inspection operations.16
21.2 Switching to sampling by variables.16
21.3 Switching from variables inspection to attributes inspection.16
21.4 Switching between preferred values of c.17
21.5 Discontinuation of inspection.17
22 Switching between "s" method and "σσ" method .17
σσ
22.1 Switching from the "s" method to the "σσσσ" method.17
22.2 Switching from the "σσ" method to the "s" method.17
σσ
Section four: Tables .18
Annex A (normative) Procedures for obtaining s and σσσσ.57
A.1 Procedure for obtaining s.57
A.2 Procedure for obtaining σσ .58
σσ
Annex B (normative) Likelihood ratio tests for heterogeneity and inconsistency of standard deviations.59
Bibliography.60
2

---------------------- Page: 4 ----------------------

SIST-TP CEN/TR 14547:2006
CEN/TR 14547:2005 (E)


Foreword
This Technical Report (CEN/TR 14547:2005) has been prepared by Technical Committee CEN /TC 283, "Precious
metals - Applications in jewellery and associated products", the secretariat of which is held by UNI.

3

---------------------- Page: 5 ----------------------

SIST-TP CEN/TR 14547:2006
CEN/TR 14547:2005 (E)
Section one: General
1 Scope
This Technical Report specifies an acceptance sampling system of single sampling plans both for inspection by
attributes and for inspection by variables. The attributes plans are of the accept-zero form, i.e. no lot is accepted if
the sample from it contains one or more nonconforming articles. (For the purposes of this Technical Report, a
nonconforming article is a precious metal article containing less than the nominal proportion by weight of the
precious metal). The variables plans generally require smaller sample sizes than the attributes plans, but require
the precious metal content of all the sampled articles to exceed the nominal content by at least a specified amount.
The objective of this Technical Report is to provide accept-zero schemes and procedures for assuring an upper
limit to the long-term percentage of nonconforming precious metal articles in the market place (i.e. the percentage
of articles reaching the market place that contain less than the nominal proportion of precious metal) by means of
the lowest practicable sample sizes. The upper limit provided by this Technical Report is 1,47% nonconforming.
The objective is achieved in three ways:
a) the sample size reduces as the total number of articles accepted in all the lots since the last non-acceptance
increases;
b) if quality is consistently high, then subject to certain conditions it will be possible to switch from sampling by
attributes to sampling by variables;
c) under sampling by variables, further reductions in sample size may be achieved by switching from the unknown
standard deviation method (the "s" method) to the known standard deviation method (the "σσσσ" method) if there
is sufficient evidence that the process standard deviation is constant.
This Technical Report is designed for use under conditions where:
d) articles are of a similar type, weight and nominal precious metal content, all supplied by the same supplier;
NOTE Where any of these factors differ, this Technical Report should be applied to each combination of type, weight
range, nominal precious metal content and supplier separately.
e) the cost to the supplier for marking for fineness increases with the number of articles inspected and the cost of
inspection of each article;
f) for sampling by attributes, it is possible to accurately determine whether the sampled articles are conforming or
nonconforming with regard to precious metal content;
g) for sampling by variables, the precious metal content by weight for each sampled article is measurable on a
continuous scale (usually in parts per thousand, expressed as a real number), and the precious metal content from
article to article is distributed according to a normal distribution, at least to a close approximation;
h) for sampling by variables under the "σσσσ" method, the process standard deviation is stable.
2 Normative references
The following referenced documents are indispensable for the application of this document. For dated references,
only the edition cited applies. For undated references, the latest edition of the referenced document (including any
amendments) applies.
ISO 3534-1:1993, Statistics - Vocabulary and symbols - Part 1: Probability and general statistical terms
ISO 3534-2:1993, Statistics - Vocabulary and symbols - Part 2:Statistical quality control
ISO 5479, Statistical interpretation of data - Tests for departure from the normal distribution
ISO 5725-6, Accuracy (trueness and precision) of measurement methods and results - Part 6: Use in practice of
accuracy values
4

---------------------- Page: 6 ----------------------

SIST-TP CEN/TR 14547:2006
CEN/TR 14547:2005 (E)
3 Terms and definitions
For the purposes of this document, the terms and definitions given in ISO 3534-1:1993 and ISO 3534-2:1993 and
the following apply. For ease of reference, the definitions of some of these terms are quoted from these standards
while others are re-defined or newly defined.
3.1
inspection
activity such as measuring or gauging the precious metal content of an article, and comparing the results with
specified requirements in order to establish whether conformity is achieved
3.2
lot
definite collection of precious metal articles of the same type, the same approximate weight, the same nominal
precious metal content and from the same supplier
3.3
sample
a set of one or more articles taken from a lot and intended to provide information on the lot
3.4
sample size
number of articles in the sample
3.5
sampling plan
combination of sample size to be used and associated lot acceptability criteria
NOTE 1 A sampling plan does not contain the rules on how to draw the sample.
NOTE 2 There is a distinction between the terms "sampling plan" (3.5), "sampling scheme" (3.6) and "sampling
system" (3.7).
3.6
sampling scheme
combination of sampling plans with rules for changing from one plan to another
3.7
sampling system
collection of sampling plans, or of sampling schemes, each with its own rules for changing plans, together with
sampling procedures including criteria by which appropriate plans or schemes may be chosen
NOTE This Technical Report is a sampling system indexed by lot size and credit.
3.8
test portion
a quantity of material taken from an article or articles for analysis
3.9
lower specification limit, L
the nominal minimum precious metal content of an article
NOTE L will normally be expressed in terms of parts per thousand by weight.
3.10
inspection by attributes
inspection by means of which each article is classified simply as conforming or nonconforming with respect to the
lower specification limit
3.11
(acceptance) sampling by attributes
lot acceptance procedure based on the number of nonconforming articles in a sample
NOTE In this Technical Report a lot is automatically non-accepted if one or more articles in the sample are classified
as nonconforming.
5

---------------------- Page: 7 ----------------------

SIST-TP CEN/TR 14547:2006
CEN/TR 14547:2005 (E)
3.12
inspection by variables
inspection by means of which a single measurement is obtained for the precious metal content of each article in a
sample
3.13
(acceptance) sampling by variables
lot acceptance procedure based on the measurements of the precious metal content of the articles in the sample
3.14
process average
process level averaged over a defined time period or quantity of production
NOTE In this Technical Report the process average is the quality level in percent nonconforming during a period
when the process is in a state of statistical control (see ISO 3534-2:1993, 3.9)
3.15
credit K
total number of articles accepted in a sub-sequence of lots since a lot in the sequence was non-accepted
3.16
average outgoing quality (AOQ)
expected average quality level of outgoing product for a given value of incoming product quality
NOTE Different definitions of AOQ are used depending on whether or not nonconforming articles removed in the
100% inspection of non-accepted lots are replaced by conforming articles. This Technical Report is based on the
assumptions that every lot that fails to satisfy the acceptance criterion when the credit is zero is 100% inspected and that
every article found to be nonconforming is scrapped and not replaced.
3.17
average outgoing quality limit (AOQL)
maximum AOQ over all possible values of incoming product quality level for a given acceptance sampling scheme
NOTE For the sampling schemes in this Technical Report the AOQL is 1,47%.
3.18
quality
the totality of features and characteristics of a product or service that bear on its ability to satisfy stated or implied
needs
NOTE In this Technical Report, quality will be interpreted in the narrow sense of whether or not the precious metal
content of an article exceeds the lower specification limit. Quality is high if it does and low if it does not.
4 Symbols and abbreviations
The symbols and abbreviations used in this Technical Report are as follows:
AOQ average outgoing quality, in percent nonconforming articles
AOQL average outgoing quality limit, in percent nonconforming articles
b  constant used in the development of the likelihood ratio tests of Annex B
c  constant used in the acceptance criteria for sampling by variables
NOTE In this Technical Report, c takes the values ¼, ½, ¾, 1 and 1¼
K  credit (an integer)
L  lower specification limit for the precious metal content, expressed in parts per thousand by weight
n  sample size
6

---------------------- Page: 8 ----------------------

SIST-TP CEN/TR 14547:2006
CEN/TR 14547:2005 (E)
N  lot size
p  process average, expressed as the proportion of articles that are nonconforming
p  process average at which the probability of acceptance of a lot is a
a
s  standard deviation of the precious metal content of the articles in the sample, expressed in terms of
parts per thousand (see Annex A)
X the smallest of the measured precious metal contents for the articles in a sample, expressed in terms
min
of parts per thousand
σ  standard deviation of the precious metal content of the articles generated by the production process,
expressed in terms of parts per thousand (see Annex A)
5 Average outgoing quality limit
When submitted quality is very high, almost all the lots will be accepted for marking for fineness and enter the
market place unchanged. The average outgoing quality (AOQ) of articles to the market place will therefore be a low
percent nonconforming.
When submitted quality is very low, most of the lots will be non-accepted and 100% inspected. In this case, most of
the articles will be subjected to inspection and only the conforming articles marked for fineness, so the AOQ of
articles to the market place will again be a low percent nonconforming, albeit at a relatively high average inspection
cost per article marked for fineness.
As quality moves from one of these extremes to the other, the AOQ will increase, reach a limit, and thereafter
decrease. The upper limit to the average outgoing percent nonconforming is called the average outgoing quality
limit (AOQL). The plans in this Technical Report have been devised in such a way that no sampling scheme
constructed from the plans will give an AOQL of more than 1,47% nonconforming, subject to the normality
assumption in the case of sampling by variables.
As sample sizes by their nature are constrained to be integers, the sampling plans have invariably been chosen to
err on the conservative side, i.e. to provide an AOQL of no more than 1,47% nonconforming. As a consequence,
some schemes constructed from plans in this Technical Report will provide an AOQL that is substantially less than
1,47% nonconforming.
The average outgoing quality after a sequence of lot submissions will only approach the AOQL under the most
adverse sequence of lot qualities. In practice this is unlikely to occur, so the AOQ will generally be very much better
than the nominal AOQL of 1,47% nonconforming. In particular, if submitted lot quality in percent nonconforming is
consistently better than the AOQL, then the AOQ will tend to be even better than the AOQL; this is because the
sampling procedure can only result in the lot either being unchanged or in the removal of its nonconforming articles.
6 Switching between attributes and variables inspection
The attributes schemes in this Technical Report provide a firm guarantee that the long-term percentage of
nonconforming articles entering the market place will not exceed 1,47%, whatever the quality of submitted lots.
Attributes inspection is therefore an integral and obligatory procedure of this Technical Report if the consumer
protection implied by the AOQL is to be maintained.
This Technical Report also provides the possibility of switching to variables inspection when inspection results
indicate that the process average is stable and reliable at a level well below 1,47%, and that the precious metal
content is distributed from article to article in accordance with a normal distribution. However, under variables
inspection the AOQL of 1,47% is only guaranteed if the assumption of normality is valid, at least approximately. The
use of sampling by variables is, therefore, optional (at the request of the supplier and at the discretion of the Assay
Office).
If measurements have been made of the precious metal content of the articles in at least 10 previous samples and
there is sufficient evidence from a statistical test on these data that the within-lot article-to-article variability in the
precious metal content is stable, consideration should be given to using the "σσσσ" method. If inspection is switched to
the "σσ" method, the square root of the weighted average value of s² over the previous 10 lots shall be taken as σσ
σσ σσ
7

---------------------- Page: 9 ----------------------

SIST-TP CEN/TR 14547:2006
CEN/TR 14547:2005 (E)
(see Annex A). The value of σσσσ shall be continually updated on the basis of the previous 10 lots as long as the "σσσσ"
method is in effect.
If measurements have not been made under sampling by attributes, or the tests given in Annex B strongly indicate
that there is a lack of homogeneity of the sample standard deviations or a lack of conformity of the most recent
standard deviation with the supposedly known value σσ, then there is no choice under sampling by variables except
σσ
to use the "s" method.
Both the "s" method and the "σσσσ" method tend to progress more rapidly to very small sample sizes, thus saving on
inspection effort. For each of these methods there are five options, distinguished by a parameter c that takes
values from 3 to 13 by intervals of 3. Larger values of c require smaller sample sizes, but are only appropriate when
quality is an order of magnitude better than the AOQL.
7 The credit principle
The unifying idea for the sampling plans in this Technical Report is the credit principle. It is assumed that a supplier
will submit for marking for fineness a continuing sequence of lots of articles of a similar precious metal content.
Each lot may be of any size.
NOTE The AOQL protection for short sequences or for an isolated lot will still be valid, but the supplier will not benefit
to the same extent from an accumulation of credit and the resulting reduction in inspection costs.
At the start of inspection, credit is set to zero. The appropriate sample size for each lot is determined from the
relevant table by entering the table with the lot size and the credit. A sample of this size is randomly selected from
the lot and inspected. If the first lot fails to meet the acceptance criterion, credit remains at zero. Otherwise credit is
increased by the lot size. The process is repeated with successive lots, with credit increasing by the size of
accepted lots until a lot is non-accepted, at which point the credit is reset to zero and a new sequence is started.
By this means, a supplier who submits lots of consistently high quality is rewarded by smaller sample sizes and
therefore lower inspection costs, while the required AOQL is still guaranteed.
8 Classification by weight range
It was assumed in Clause 7 that the supplier was submitting for marking for fineness a continuing sequence of lots
of articles of a similar precious metal content. Indeed, the methods in this Technical Report were essentially
developed for articles of identical value submitted in lots of possibly varying size. In practice, of course, the supplier
may submit lots of different types of article with a widely varying nominal weight of precious metal per article. In
order to ensure that credit accumulated on articles of low value is not misused to provide artificially small sample
sizes when sampling from articles of high value, lots are classified according to which weight range the average
nominal weight of precious metal per article in the lot belongs. See Table 4 for the preferred weight ranges for gold,
silver, platinum and palladium. A sequence of lots for the purposes of this Technical Report is then considered to be
a sequence of lots from the same supplier of articles of a similar type and the same precious metal, with the
nominal precious metal content within the same weight range. Suppliers may have a number of sequences of lots
running concurrently.
9 Role and use of 100% inspection of non-accepted lots
If a supplier continually submits lots with in excess of 1,47% articles nonconforming, then eventually a sample
could be selected that satisfies the acceptance criterion. If the previous non-accepted lots had all been scrapped
without any further inspection, then this would be the first lot to enter the market place and the required AOQL of
1,47% nonconforming would at once be exceeded. Clearly such a procedure does not provide the required level of
consumer protection.
If, instead, the first lot in a sub-sequence (i.e. when the credit is zero) is always 100% inspected if it is non-
accepted, and the conforming articles found in the lot are allowed to enter the market place, then these conforming
articles dilute the effect of erroneously accepted lots and it becomes possible to devise plans, such as those in this
Technical Report, that provide the required consumer protection.
100% inspection may be relatively expensive if the articles are of low value with little workmanship and the lot is
large. For such articles the supplier may be advised to submit a small lot at zero credit in order to avoid the
8

---------------------- Page: 10 ----------------------

SIST-TP CEN/TR 14547:2006
CEN/TR 14547:2005 (E)
possibility of an expensive 100% inspection of a large lot. Consultation of the relevant table from among Tables 1, 2
and 3 will help in determining an appropriate lot submission strategy.
On the other hand, articles may have had a considerable investment of workmanship that the supplier does not
wish to be destroyed, or the re-refining costs of scrapped articles may be relatively high. In this case, the supplier
may prefer all non-accepted lots to be 100% inspected, even when the credit exceeds zero, so that all the
conforming articles in the lot can be placed in the market. The supplier may demand this option, as the AOQL
consumer protection is not thereby compromised.
9

---------------------- Page: 11 ----------------------

SIST-TP CEN/TR 14547:2006
CEN/TR 14547:2005 (E)

Section two : Choice of sampling plan

10 Choice between attributes and variables
At the start of inspection operations, when there is no evidence on which to judge the normality or otherwise of the
distribution from article to article of the precious metal content, there is no option but to use sampling by attributes.
Sampling by attributes does have certain advantages over sampling by variables:
a) The AOQL is guaranteed under sampling by attributes, regardless of the distribution from article to article of
the precious metal content.
a) No tests for outliers or for heterogeneity or inconsistency of the standard deviations are necessary, so
attributes inspection is administratively easier.
b) Inspection by attributes is generally cheaper and quicker, as it is only necessary to determine whether or not
the precious metal content of each sampled article exceeds the level specified, not the precise content for
each article.
c) An attributes scheme can be more readily understood and accepted; for example, it may at first be difficult to
accept that, when inspecting by variables, a lot can be non-accepted on measurements taken of a sample that
does not contain any nonconforming articles.
d) If quality is consistently excellent so that lots are continually accepted, eventually a point will be reached under
sampling by attributes where the required sample size is only 1. For sampling by variables, it is necessary to
take a sample of at least two articles from each lot in order to be able to determine the standard deviation,
either for use under the "s" method or for monitoring the variability under the "σσσσ" method.
The advantage of sampling by variables is that sample sizes are generally smaller and the lower sample sizes (with
the exception of sample size 1) are reached more quickly. This is an important consideration if inspection is time-
consuming or expensive.
11 Choice between the "s" method and the "σ" method
If it has been decided to switch to sampling by variables and the conditions for switching are satisfied (see Clau
...