CODEX SAMPLING PLANS FOR PREPACKAGED FOODS (AQL 6.5)
CODEX STAN 233-1969
The Sampling Plans in Appendix I of this document apply to the acceptance of defective units (defectives) in lots of prepackaged foods, as defined in individual Codex Standards, insofar as the Sampling Plans have been specifically included in such Codex Standards for the purpose of determining the acceptability or otherwise of the lot. They shall be used in accordance with the provisions dealing with the classification of defectives and lot acceptance in Codex Standards to which these Sampling Plans are stated to apply and within the limits of Section 2 of this document.
2. FIELD OF APPLICATION
2.1. TYPE OF EXAMINATION TO WHICH THE SAMPLING PLANS APPLY
The Sampling Plans in Appendix I of this document are intended primarily to cover the quality provisions of Codex Commodity Standards where an AQL of 6.5 is appropriate for the defective unit as defined in Codex Standards. For the purposes of these Sampling Plans, “quality” refers to those factors or product characteristics which are evaluated by organoleptic or physical means, such as colour, flavour, texture, defects, size and appearance. They are not intended however, to cover factors which may represent a hazard to health or which are unwholesome or otherwise highly objectionable to the consumer on the basis of which responsible authorities would reject the lot. Examples of these latter categories are pesticide residues, contaminants, blown cans, foreign material such as stones and large insects. Other criteria andsampling plans must be used in dealing with factors of this type. While these Sampling Plans are intended primarily for quality evaluation, they may be found suitable for other determinations such as net weight, Brix values and drained weight, provided an acceptance criterion with an AQL of 6.5 is appropriate for these determinations. In this case a definition of “defective” for the specific determination under consideration would be required in the respective Codex Standard.
2.2. SIZE OF LOT AND POINT OF APPLICATION
The Sampling Plans and acceptance procedures contained in this document are designed to cover lots that represent substantial portions of factory production or relatively large blocks of merchandise. The plans may also be used for small lots, but Governments may elect to use sampling procedures of their own choosing for enforcement at the retail level. This is done in recognition of the high ratio of sample size to lot size when dealing with small lots and the probability that once the production of defective or non-conforming product is no longer likely to be uniform between and within the smaller lots.
2.3. PRINCIPLES OF ACCEPTANCE SAMPLING
For detailed explanation of the statistical basis for these Sampling Plans, see Appendix II of this document.
The Sampling Plans - Appendix I of this document - are a tabular presentation appropriate for acceptance sampling of prepackaged foods where an AQL of 6.5 has been accepted for certain products characteristics. The Plans include:
1. Inspection Levels;
2. Sample Sizes in relation to lot size and container size; and
3. Acceptance Numbers.
A sample is drawn at random from the lot according to the appropriate schedule in the Sampling Plans. Each sample unit is examined according to the requirements of the individual Codex Standard and classified as either “acceptable” or as “defective”. Based on the total number of “defectives “ in the sample, the lot either “meets” or “fails” the requirements of the Codex standard, to which these Sampling Plans apply, according to the following criteria:
- Meets if the number of “defectives” is equal to, or less than, the acceptance number of the appropriate plan.
- Fails if the number of “defectives” exceeds the acceptance number of the appropriate plan.
The maximum percent defective units (defectives) permitted in a lot which will be accepted approximately 95 percent of the time. For example, a sampling plan at an AQL of 6.5 will accept a lot or production which has 6.5 percent defective approximately 95 percent of the time.
4.2. ACCEPTANCE NUMBER (C)
The number in a sampling plan which indicates the maximum number of defectives permitted in the sample in order to consider the lot as meeting the requirements of a Codex Standard.
4.3. BUYER’S RISK
The risk a buyer takes that a lot will be accepted on the basis of these Sampling Plans even though such a lot may fail to conform to the requirements of the Codex Standard.
4.4. PRODUCER’S RISK
The risk a producer takes that a lot will fail on the basis of these Sampling Plans even though such a lot in
reality may meet the requirements of the Codex Standard.
A “defective” is a sample unit which does not conform with a certain specified requirement (or requirements) of a Codex Standard (on the basis of total “demerit points”, individual tolerances for “defects”, etc.). The criteria on the basis of which a sample unit is classified as “defective” are specified in individual Codex standards to which these Sampling Plans apply (see also Sub-sections 2.1 and 2.2 of this document). Although a defective is a sample unit which fails to meet certain specified requirements in Codex standards, it does so only to an extent which is slightly below those requirements and which would not make the product objectionable to the consumer as specified in Section 2 - Field of Application, Sub-section 2.1.
The process of measuring, examining, testing or otherwise comparing a container or unit of product (sample unit) with the requirements of a Codex Standard.
4.7. INSPECTION LEVEL
The term used to indicate the relative amount of sampling performed on lots of a given product or class of products.
4.8. LOT OR INSPECTION LOT
Collection of primary containers, or sample units, of the same size, type and style which have been manufactured or processed under essentially the same conditions.
4.9. LOT SIZE (N)
The number of primary containers, or sample units, in the lot.
4.10. SAMPLE UNIT
The individual container (primary container), a portion of the contents of the primary container or a composite mixture of product that is examined or tested as a single unit
Any number of sample units which are used for inspection. Generally the sample comprises all of the containers or sample units drawn for examination or testing purposes from a particular lot.
The process of drawing or selecting containers or sample units from a lot or production.
4.13. SAMPLE SIZE (N)
The number of containers, or sample units comprising the total sample drawn from a lot or production.
4.14. SAMPLING PLAN
A sampling scheme which includes sample sizes, inspection levels, acceptance and/or rejection numbers so that a decision can be made to accept or reject the lot or production based on the results of inspection and testing of the sample.
5. APPLICATION OF THE SAMPLING PLANS
5.1. INFORMATION REQUIRED
In using the Sampling Plans in Appendix I of this document, the following information shall be known:
ontainer size (net weight in kg or lb)
Inspection Level (see sub-section 4.7)
Lot size (N) (see sub-section 4.9)
Requirements of the Codex Standard with respect to product quality (i.e. classification of defectives and requirements for acceptance of the lot).
The following steps are taken:
The appropriate inspection level is selected as follows:
Inspection Level I - Normal sampling
Inspection Level II - Disputes (Codex referee purposes sample size), enforcement or need for better lot estimate.
Determine the lot size (N), i.e. number of primary containers or sample units.
Determine the number of sample units (sample size (n))to be drawn from the inspection lot, consideration being giving to container size, lot size, and inspection level.
Draw at random the required number of sample units from the lot giving proper consideration to code or other identifying marks in selection of the sample.
Examine the product in accordance with the requirements of the Codex Standard. Classify any container or sample unit which fails to meet the specified quality level of the standard as a defective on the basis of the classification of defectives contained in the Codex Standard.
Refer to the appropriate Sampling Plan in Appendix I.
Consider the lot acceptable if the number of defectives is equal to or is less than the acceptance number (c) of the appropriate Sampling Plan contained in Appendix I of this document.
Consider the lot as failing if the number of defectives exceeds the acceptance number (c) of the appropriate Sampling Plan contained in Appendix I of this document.
5.3. EXAMPLES FOR THE APPLICATION OF THE SAMPLING PLANS
Inspection Level I (see sub-section 5.2 (a))
A lot consists of 1200 cases, packed 12 x 2.5 lb primary containers per case. A decision is made to use inspection Level I since the goods are not in dispute and there is no history of controversy over quality. A container is defined in the Codex Standards or is taken to be the sample unit.
Lot Size (N) = 1200 x 12 or 14,400 units
Container Size = 2.5 lb
Inspection Level = I (see Sampling Plan 1, Appendix I)
Sample size (n) = 13
Acceptance Number (c) = 2
In this example if there are no more than two (2) “defectives” in a sample size of 13 containers the lot is considered acceptable. If, however, there are three (3) or more “defectives” in the sample the lot is considered as failing to meet the requirements. A “defective” as used in the Sampling Plans is defined in the Codex Standard.
b. Inspection Level II (see sub-section 5.2 (a))
If in the foregoing example (5.3 (a)) the quality of the goods is in dispute and a referee method is required for the examination or re-examination of the lot, an increased sample size is taken at Inspection Level II, selecting at least 21 containers.
Lot Size (N) = 1200 x 12 or 14,400 units
Inspection level = II (see Sampling Plan 2, Appendix I)
Sample Size (n) = 21
Acceptance Number (c) = 3
5.4. NOTES ON SAMPLE SIZE
It is not necessary to restrict the sample size to the minimum corresponding to the appropriate lot size and Inspection Level. In all cases a larger sample may be drawn. In the example at 5.3 (b) an even more reliable estimate of lot quality could be made by taking a sample of 29 or even 48 and applying the corresponding acceptance numbers of 4 and 6 respectively.
d. LÊy ngÉu nhiªn ®ñ sè ®¬n vÞ mÉu theo yªu cÇu tõ l« cã xem xÐt thÝch hîp vÒ m· sè hay c¸c dÊu hiÖu nhËn d¹ng kh¸c khi chän mÉu.
e. KiÓm tra s¶n phÈm theo yªu cÇu cña Tiªu chuÈn Codex. XÕp bÊt cø hép hay ®¬n vÞ mÉu nµo kh«ng ®¹t møc chÊt lîng vµo lo¹i bÞ lçi dùa trªn sù xÕp lo¹i lçi trong Tiªu chuÈn Codex.
f. Tham chiÕu KÕ ho¹ch LÊy mÉu thÝch hîp trong Phô lôc I
g. KÕt luËn l« cã thÓ chÊp nhËn nÕu sè lçi b»ng hoÆc thÊp h¬n sè lçi cã thÓ chÊp nhËn (c) cña KÕ ho¹ch LÊy mÉu trong Phô lôc I cña tµi liÖu nµy.
h. KÕt luËn l« kh«ng ®¹t nÕu sè lçi vît qu¸ sè lçi cã thÓ chÊp nhËn theo KÕ ho¹ch LÊy mÉu trong Phô lôc I cña tµi liÖu nµy.
5.3. VÝ dô vÒ ¸p dông KÕ ho¹ch lÊy mÉu
a. KiÓm tra cÊp I (xem môc 5.2 (a))
Mét l« s¶n phÈm cã 1200 thïng, mçi thïng cã 12 hép nÆng 2,5 pound. QuyÕt ®Þnh ¸p dông KiÓm tra cÊp I v× kh«ng cã tranh chÊp hay cã tiÒn sö tranh c·i vÒ chÊt lîng hµng. Hép ®ùng s¶n phÈm, theo ®Þnh nghÜa trong Tiªu chuÈn Codex, hay mét hép bÊt kú ®îc lÊy ra tõ l« gäi lµ mét ®¬n vÞ mÉu.
Quy m« l« s¶n phÈm (N) = 1200 x 12 hay 14.400 hép (®¬n vÞ)
Quy m« hép = 2,5 pound
CÊp kiÓm tra = I (xem KÕ ho¹ch lÊy mÉu 1,Phô lôc I)
Quy m« mÉu (n) = 13
Sè hép bÞ lçi cã thÓ chÊp nhËn = 2
Trong vÝ dô nµy, nÕu sè hép bÞ lçi kh«ng nhiÒu h¬n hai (2) trong mét quy m« mÉu cã 13 hép th× l« hµng ®îc coi lµ cã thÓ chÊp nhËn. Tuy nhiªn, nÕu sè lçi trong mÉu tõ ba (3) trë lªn, l« hµng bÞ coi lµ kh«ng ®¹t tiªu chuÈn. Kh¸i niÖm “lçi” sö dông trong KÕ ho¹ch LÊy mÉu ®îc ®Þnh nghÜa trong Tiªu chuÈn Codex.
KiÓm tra cÊp II (xem môc 5.2 (a))
NÕu trong vÝ dô trªn (5.3 (a)), cã tranh chÊp vÒ chÊt lîng hµng vµ ph¬ng ph¸p träng tµi ®îc yªu cÇu ®Ó kiÓm tra hay t¸i kiÓm tra l« th× ¸p dông KiÓm tra cÊp II vµ cÇn quy m« mÉu lín h¬n, Ýt nhÊt lµ 21 hép.
Quy m« l« s¶n phÈm (N) = 1200 x 12 hay 14.400 ®¬n vÞ
CÊp kiÓm tra = II (xem KÕ ho¹ch lÊy mÉu 2,Phô lôc I)
Khèi lîng tÞnh trªn 1 kg (2,2 pound) nhng kh«ng qu¸ 4,5 kg (10 pound)
NET WEGHT IS GREATER THAN 1 KG (2.2 LB) BUT NOT MORE THAN 4.5 KG (10 LB)
Quy m« l« s¶n phÈm (N)
Lot Size (N)
Quy m« mÉu (n)
Sample Size (n)
Sè lçi cã thÓ chÊp nhËn (c)
Acceptance Number (c)
2.401 - 15.000
15.001 - 24.000
24.001 - 42.000
42.001 - 72.000
72.001 - 120.000
Khèi lîng tÞnh trªn 4.5 kg (10 pound)
NET WEIGHT GREATER THAN 4.5 KG (10 LB)
Quy m« l« s¶n phÈm (N)
Lot Size (N)
Quy m« mÉu (n)
Sample Size (n)
Sè lçi cã thÓ chÊp nhËn (c)
Acceptance Number (c)
601 - 2.000
2.001 - 7.200
7.201 - 15.000
15.001 - 24.000
24.001 - 42.000
EXPLANATORY NOTES ON ACCEPTANCE SAMPLING
Sampling is the process of drawing or selecting containers or sample units from a lot or production. As a result of sampling, information is obtained by which an estimate can be made to accept, reject or negotiate the merchandise in question. Sampling procedures which contain both sample size and acceptance criteria are commonly referred to as “acceptance sampling”.
There are many types of acceptance sampling systems in use today. A plan that is suitable for one product or type of inspection may be entirely unsuitable for another product or inspection system. The plan selected is determined to a large extent by the degree to which it satifies the needs of the user.
In developing these acceptance sampling plans, initial consideration has been given to quality evaluation of the end product. This requires opening of containers with resultant loss of products. This type of inspection is referred to as “destructive sampling”. Not only is the loss of product an important consideration, but also destructive sampling is generally quite time consuming, Consequently, both inspection time and economic loss of product through destructive inspection are significant limiting factors in developing sampling plans for quality evaluation of processed foods. Sample size must necessarily be relatively small in order to make the plan practical in application.
The aim of any sampling plan should be to accept more “good” lots and reject more “bad”lots. Since probability and chance are involved, decisions will, of necessity, involve an element of risk. This risk factor has to be accepted as a part of any sampling procedure. One method of reducing the buyer’s risk of accepting deliveries of non-conforming quality is to increase sample size. In other words, the larger the sample, the less risk involved in accepting “bad” lots. Inspection level is the term indicating the relative amount of sampling and inspection performed on lots of a given product or class of products. If the inspection lot is packed under close control and meets the requirements of the Codex Standard, changing inspection levels do not appreciably change the buyer-seller risk. In other words, this would be a “good” lot and should be passed practically all of the time by a good sampling plan. The effectiveness of a sampling plan in discriminating between “good” and “bad” lots can be estimated by examination of the OC curves (see Appendix III) for the various sample sizes. For example, if a lot is produced so that it does not contain more than 6.5 percent defectives, such lot will be passed at least 95 percent of the time by the sampling plans applicable for an AQL of 6.5. On the other hand, if the production contains an appreciable amount of defective material, a higher inspection level (i.e. a larger sample size) will reduce the risk of accepting these non-conforming lots. The effect of increased sample size is explained in greater detail under the discussion of OC curves.
One of the initial considerations in the development of a statistical acceptance sampling plan is the selection of an appropriate AQL or Acceptable Quality Level. This characteristic is defined as the maximum percent defective units in lots that will be accepted most of the time (approximately 95 percent of the time). Lots or production containing more defective material will be accepted less often - the ratio of rejection to acceptance increasing as the sample size increases and as the percent defective material in the lot increases.
In developing these sampling plans, an AQL of 6.5 was selected for lot acceptance with respect to quality evaluation. In other words, an AQL of 6.5 is used in these sampling plans (Appendix I) to determine whether or not the inspection lot meets minimum quality requirements of the Codex Standard. This value was selected on the basis of years of experience and the capability of industry to produce preserved fruits and vegetables and certain other processed foods at this level under good commercial practice. For other factors (such as Brix value and net weight) other AQLs may be selected. Sampling plans can be drawn up for a full range of AQLs from a very strict value of 0.10 to a rather lenient value of 25.0 and higher,
depending either on the type of product and/or on the criteria involved.
These sampling plans provide for two inspection levels, I and II. These two levels provide some discretion in the application of the Sampling Plans to the inspection of a commodity, depending upon circumstances. For normal trading purposes Level I is recommended. In the case of dispute or controversy, i.e. for Codex referee purposes, Level II is recommended. Smaller sample sizes than those provided by Levels I and II may be justified, e.g. when a delivery is being checked for labelling or for detection of non-permitted additives. However, the acceptance sampling criteria of the Plans, which permit 6.5 percent “defectives”, do not apply to such an inspection.
The problem of buyer’s and seller’s risks in relation to sample size and lot quality is illustrated through the use of Operating Characteristic Curves (OC Curves). Appendix III contains OC Curves for the sampling plans contained in Appendix I of this document. For Purposes of destructive inspection sample sizes in excess of 84 are not practical, since any further inspection beyond this point will not generally provide sufficient additional data to warrant the time and expense of testing.
In studying the OC Curves for AQL 6.5 several conclusions can be drawn, namely:
All of the Curves have the same general slope although the curve for sample size 6 is flatter.
All curves intersect at a point represented by the coordinates of “6.5 percent defective”and approximately “95 percent probability of acceptance”.
As the sample size increases, the curves become steeper and more discriminating, i.e. lots having “defectives” in excess of 6.5 percent are rejected with greater frequency.
The reliability of the larger sample size is not in direct proportion to the increased sample. For example, for a lot that is 20 percent defective a sample size of 6 (curve E) will accept such lot 65 percent of the time; whereas a sample size of 48 (curve L) will accept the same lot 22 percent of the time. In this example the ratio between probabilities of acceptance is only 3 to 1.
To illustrate the use of the OC Curves (AQL 6.5) let it be assumed that a lot is 10 percent defective. A lot
with 6.5 percent defectives will be accepted approximately 95 percent of the time, the frequency of
acceptance increasing as the percent defective decreases. However, the 10 percent defective lot fails to measure up to requirements, and while it may be a marginal lot, it may not be acceptable. An examination
of the OC curves shows that a sample size of 6 (curve E) will accept this marginal lot 88 percent of the
time; a sample size of 84 (curve M) is somewhat better, accepting the lot 65 percent of the time.
If, on the other hand, the lot is 30 percent defective, a sample size of 6 (curve E) will accept the lot only 42
percent of the time, whereas a sample size of 21 (curve J) will accept such a lot only 8 percent of the time and a sample size of 84 (curve M) will always fail such a lot.
Phô lôc II
Chó gi¶i vÒ lÊy mÉu chÊp nhËn
LÊy mÉu lµ c«ng ®o¹n lÊy hay chän hép ®ùng s¶n phÈm hoÆc ®¬n vÞ mÉu tõ mét l« hay mét qu¸ tr×nh s¶n xuÊt. LÊy mÉu cung cÊp nh÷ng th«ng tin mµ dùa vµo ®ã ngêi ta cã thÓ ®¸nh gi¸ nh»m chÊp nhËn, tõ chèi hay th¬ng th¶o l« hµng cã vÊn ®Ò. Thñ tôc lÊy mÉu bao gåm c¶ quy m« mÉu vµ c¸c tiªu chÝ chÊp nhËn thêng ®îc gäi lµ “lÊy mÉu chÊp nhËn”.
Cã nhiÒu kiÓu hÖ thèng lÊy mÉu chÊp nhËn hiÖn hµnh. Mét kÕ ho¹ch cã thÓ phï hîp víi mét s¶n phÈm hay mét h×nh thøc kiÓm tra nµy nhng l¹i cã thÓ kh«ng hoµn toµn phï hîp víi mét s¶n phÈm hay hÖ thèng kiÓm tra kh¸c. Ph¹m vi cña kÕ ho¹ch ®îc x¸c ®Þnh nh»m tho¶ m·n c¸c yªu cÇu cña ngêi sö dông.
Cã 2 cÊp kiÓm tra trong KÕ ho¹ch lÊy mÉu, cÊp I vµ II. Hai cÊp kiÓm tra nµy cho phÐp lùa chän c¸ch ¸p dông KÕ ho¹ch LÊy mÉu ®Ó kiÓm tra hµng ho¸ tuú thuéc vµo hoµn c¶nh cô thÓ. KiÓm tra cÊp I nªn ¸p dông ®èi víi nh÷ng môc ®Ých thêng m¹i th«ng thêng. Trong trêng hîp cã tranh chÊp hay tranh c·i, nghÜa lµ cÇn cã träng tµi Codex, nªn ¸p dông cÊp kiÓm tra II. Ngoµi ra, cã thÓ ¸p dông c¸c quy m« mÉu nhá h¬n quy m« ë cÊp I, II trong mét sè trêng hîp, vÝ dô nh kiÓm tra vÒ ghi nh·n l« hµng hay dß t×m c¸c phô gia kh«ng ®îc phÐp. Tuy nhiªn, c¸c tiªu chÝ chÊp nhËn mÉu cña KÕ ho¹ch cho phÐp møc lçi 6,5% kh«ng hiÖu qu¶ ®èi víi h×nh thøc kiÓm tra nµy.