摘要:

CHEMICAL SOCIETY : ANALYTICAL DIVISION 161 May, 1980 LOMBARDY SCZENTZFZC THE EXPERTS IN LABORATORY APPLICATIONS OF MICROCOMPUTERS We can supply systems based on Apple II, the most flexible microcomputer available in its price range, for the following applications: DATA ACQUISITION, MANIPULATION AND STORAGE Programs are available for GC, HPLC, ESR, etc. Area calculation, Baseline correction, Calibration curves, Statistical calculations. 0 Output to pen recorders and other analogue devices. Data sampling at long intervals, e.g. Temperature and humidity measurement in stores and warehouses. 0 Curve area measurement using the Apple Digitising Tablet. With a resolution of 200 points to the inch can be used for paper and thin layer chromatography. Colour slides can be photographed directly from colour monitor screen. SCIENTIFIC STORES CONTROL We are the only Company to market a comprehensive system to improve the efficiency of your stores: Item location, chemical name, molecular weight, supplier, cost, typical delivery time, quantity in stock, last three persons to draw the item, etc. System will print out orders, catalogues by name or MW, usage by each person, valuation, etc. WORD PROCESSING With a Daisy Wheel printer this program can be used to produce error free manuscripts. Multiple revisions will take your typist minutes instead of hours. Standard letters, reports and laboratory manuals can be updated and printed out at high speed. LO M BARDY SCIENTIFIC, 41, Lombardy Drive, Berkhamsted, Herts. H P4 2BY. Tel: (04427) 4247 A2 for further information. See page 210.

ISSN:0144-557X    

DOI:10.1039/AP98017FP005

出版商:RSC    

年代:1980

数据来源: RSC

ISSN:0144-557X    

DOI:10.1039/AP98017FX020

出版商:RSC    

年代:1980

数据来源: RSC

ISSN:0144-557X    

DOI:10.1039/AP98017BX022

出版商:RSC    

年代:1980

数据来源: RSC

摘要:

ANPRDI 17(5) 161-214 (1980) ISSN 0306-1 396 May 1980 Analytical proceedings Proceedings of the Analytical Division of the Chemical Society AD President R. Belcher Hon. Secretary P. G. W. Cobb Hon. Treasurer J. K. Foreman Hon. Assistant Secretaries D. I. Coomber, O.B.E.; D. C. M. Squirrel1 Hon. Publicity and Public Relations Officer Dr. A. Townshend, Department of Chemistry, University of Birmingham, Birmingham, B1 5 2TT Secretary Miss P. E. Hutchinson Editor, Analyst and Analytical Proceedings P. C. Weston Assistant Editors Mrs. J. Brew, R. W. Hazell, R. A. Young Publication of Analytical Proceedings is the responsi- bility of the Analyst Publications Committee: J. M. Ottaway (Chairman) W. H. C. Shaw H. J. Cluley D. Simpson ‘P. Gray A. Townshend J. N. Miller ‘P. C. Weston G.E. Penketh J. Whitehead T. B. Pierce ‘Ex officio members All editorial matter should be addressed to: The Editor, Analytical Proceedings, The Chemical Society, Burlington House, Piccadilly, London, W1 V OBN. Telephone 01 -734 9864. Telex 268001. Advertisements: Advertising Department, The Chemical Society, Burlington House, Piccadilly, London, W1 VOBN. Telephone 01 -734 9864. 0 The Chemical Society 1980 Ed itoria I Is There a Shortage of Analysts? Those long enough in the tooth to remember “The Brains Trust” will recall with feelings ranging from pleasure to irritation the habit of that famous panellist, Professor Joad, of pre- facing his remarks on any topic with “Well it all depends what you mean by . . . , ’’ and it was this nostalgic picture that sprang to mind when I started to consider this question.Prominent individuals, committees, commissions and insti- tutions have produced papers and reports that point to the shortage 01 schoolchildren qualify- ing in (or even interested in) science, the short- age of science teachers, the shortage of adequate teaching facilities for science, the shortage of university applicants in science subjects, the shortage of science graduates, and so on and on. An air of quantitative foreboding opposite future, as opposed to Current, shortages is often added by references to the demographic trend that shows that the population of eighteen year olds will decline from the mid-eighties and through the nineties. However, in practically all of these studies the “shortage” is based on the laudable assumption that in a technologic- ally advanced society we need more scientists, and “shortage” becomes the equally vague term “not enough,” with few attempts made to define even in semi-quantitative terms what “enough” would be.At the same time there are extensive and gloomy prognostications of a much more quantitative nature on the effects of microelectronics and unemployment. I sup- pose that if you try hard enough it is not im- possible to think of having scientists unemployed while there is a shortage, it all depends what you mean b y . . . Thus, if I can resist the temptation to use the clich6 that there are never enough good scientists, I regard the case for a general short- age of scientists entering the profession with some suspicion, and worry about the easy option of increasing numbers by lowering stand- ards.On the other hand I am totally convinced that Society does need many more people out- side the science professions who understand, in at least a rudimentary way, science and scien- tists. 163164 ANALYTICAL DIVISION ANNIVERSARY DINNER Anal. Proc. There is a virtually complete lack of under- standing, appreciation or sympathy shown to science by the general public and, more import- antly, by the leaders of opinion. The stereotype scientist of stage, screen and literature is either bad or mad, or both. Possible exceptions are those few “experts” (analysts!) who help to fight crime and pollution. One has simply to listen to the public debates on nuclear power, the environment, legislation on hazardous chemicals, and so on, to appreciate the point.Turning now to the word “analyst,” I support those who speak of analytical science rather than analytical chemistry, although much of the current debate seems to be taking place in chemical circles. A quick scan of the “Situa- tions Vacant” columns in the scientific press will confirm that there is a continuing strong demand for graduates to work in various fields of analy- sis. It is also true that the paucity of Chairs in analysis in UK universities results in the pro- duction of very few individuals trained to higher degree standard in analysis. Rut these two pointers to shortage are countered by the increasing number of graduates emerging not only with experience on one or more of the “big” instruments (NMR, MS, ESCA, etc.) but also with an interest in analysis as a career.I feel that it is here that the heart of the matter lies. Just as the public a t large have had a poor opinion of scientists in general, so the scientific community has shown little enthusiasm for analysis or analysts. The stereotype image here has been one of the useful but routine plodder, meticulous and pedantic. Is there any wonder that although many of the most able and cre- ative people, including Nobel Prizewinners, pursued the activity, they avoided the title like the plague. However, there are encouraging signs that attitudes are changing and it is to expedite this change and to build on it that I give niy wholehearted support to the current calls for more Chairs in analysis. The aim should be to attract the best people, with quality rather than quantity as the prime criterion in all of their actixiities. To end with the good news, I suppose that as the shortage intensifies, market forces will come into play and analysts will attract premium salaries. Employers might well respond by saying “\;C’ell it all depcnds what you mean by . . . .” The readers of this journal are probably better placed than any to contribute to the debate, and the response generated in our Correspondence column by this hopefully provocative editorial should be interesting. The views expressed are my own and are not meant to reflect those of either my Company or the Chemical Society. G. I?. PENKETH

ISSN:0144-557X    

DOI:10.1039/AP9801700163

出版商:RSC    

年代:1980

数据来源: RSC

摘要:

164 ANALYTICAL DIVISION ANNIVERSARY DINNER Anal. Proc. Analytical Division Anniversary Dinner and Presentation of Awa rds Following the Annual General Meeting of the Analytical Division and the Address of the Retiring President, Professor R. Belcher, on March 14th, the Biennial Formal Dinner was Professor and Mrs. Belcher with Professor and Mrs. T . P. Whitehead. held in the Library of The Chemical Society in Burlington House, London. The dinner was preceded by the President’s Reception, which was held in the Council Room. Dr. A . Spinks and Mrs. Spinks (L) being received by the A D President, Professor R. Belcher, and Mrs. Belcher.May. 1980 REPORTS OF MEETINGS The guests at the dinner included Dr. A. Spinks, CBE, FRS, President of The Chemical Society, and Mrs. Spinks, Professor J.H. Purnell, Professor T. P. Whitehead and Mrs. Whitehead, Dr. A. J . Amos, OBE, and Mrs. Amos, Mr. J . R. Ruck Keene, CBE, General Secretary of The Chemical Society, and Dr. C. J . Keattch. 165 Dr. R . C. Mackenzie receiving the fifteenth SAC Gold Medal from the President. The President in conversation with the two recently appointed Professors in the analytical field, G. F. Kirkbright (L) and D. Betteridge (R). The Loyal Toast was proposed by The President. Then followed the presentation of the fifteenth SAC Gold Medal to Dr. R. C. Mackenzie of the Macaulay Institute for Soil Research for his services to thermal methods of analysis. An oration on Dr. Mackenzie was given by Dr. C. J. Keattch, Honorary Secretary of the Thermal Methods Group. The President then presented the fifth AD Distinguished Service Award to Dr. D. I. Coomber, OBE, for his contributions to the SAC and AD. Dr. Coomber, formerly with the Laboratory of the Government Chemist, has been Honorary Assistant Secretary (Programmes) for about 12 years. Biographies of Dr. Mackenzie and Dr. Presentation of the fifth A D Distinguished Service Award by the President to Dr. D . I . Coomber. Coomber appeared in the March issue of Analytical Proceedings (pp. 69-71). The Toast of “The Analytical Division” was proposed by Dr. A. Spinks; Professor R. Belcher responded and also proposed the Toast of “The Guests.’’ Professor J. H. Purnell replied on behalf of the guests.

ISSN:0144-557X    

DOI:10.1039/AP9801700164

出版商:RSC    

年代:1980

数据来源: RSC

摘要:

May. 1980 REPORTS OF MEETINGS 165 Reports of Meetings North West Region The fifty-fifth Annual General Meeting of the Vice-Chairman-Dr. A. Dyer. Honorary Region was held at 6.30 p.m. on Friday, Secretary-Mr. G. Davison, Kodak Ltd., Chemi- January 18th, 1980, in the Chapman Building, cal Division, Research Department, Acornfield The University, Salford. The Chair was taken Road, Kirkby, Lancashire. Honorary Treas- by the Chairman of the Region, Mr. G. B. Crump. urer-Mr. T. Hodson. Members of Committee- The following office bearers were elected for the Dr. J. E. Cantle (co-opted), Mrs. E. Healy, forthcoming year: Chairman-Mr. G. B. Crump. Dr. A. Mathias, Mr. P. Morries, Mr. P. Platt,166 ANALYTICAL QUALITY CONTROL Anal. Proc. Mr. J. W. Ogleby (ex ojicio) and Mr. B. Taylor.Mr. G. Mi. Earnshaw and Mr. M. Green were re- appointed as Honorary Auditors. North East Region The fourteenth Annual General Meeting of the Region was held a t 7.15 p.m. on Wednesday, January 23rd, 1980, a t the Europa Lodge Hotel, Darlington. The Chair was taken by the Chairman of the Region, Mr. D. F. Griffiths. The following office bearers were elected for the forthcoming year : Chairman-Mr. D. F. Grifiths. Vice-Chnirnzan-Mr. C. L. Denton. Acting Honorary Secretary-Mr. C. L. Denton, Central Laboratories, Tioxide International, Portrack Lane, Stockton-on-Tees, Cleveland. Honorary Treasurer-Dr. J. Newham. Honor- ary A ssistant Secretary-Mr. G. Himsworth. Members of Committee-Mr. P. J . Burnill, Mr. P. G. W. Cobb, Dr. H. Hughes (ex o f i c i o ) , Dr. K. W. Jackson, Dr. J . Moss, Dr. A. A. Smales and Mr. J . Vallance. Mr. C. N. Bell and Mr. J. Whitehead were re-appointed as Honor- ary Auditors.

ISSN:0144-557X    

DOI:10.1039/AP9801700165

出版商:RSC    

年代:1980

数据来源: RSC

摘要:

166 ANALYTICAL QUALITY CONTROL Anal. Proc. Analytical Quality Control The following are summaries of nine of the papers presented at a Meeting of the North West, Scottish and North East Regions and the Automatic Methods and Joint Pharmaceutical Analysis Groups held on June 7th and 8th, 1979, at the University of Stirling. Acceptance Sampling Roland Caulcutt Department of Computer StudieSand Mathematics. Huddersfield Polytechnic, Queensgate, Huddersfield, HD 1 3DH Elsewhere in this issue (p. 172), Chamberlain describes the use of cumulative sum tecliniques in quality control. He illustrates the use of the V-mask to detect a change in mean level and the early detection of such changes is of great importance if a manufacturer is to prevent his process drifting out of control. In this type of application, the cusum chart and its V-mask will probably be located close to the production plant to facilitate the rapid feedback of informa- tion to plant management, enabling a tighter control of product quality to be maintained.In many companies it is essential to monitor the quality of a product received from another manufacturer. Often this product will consist of individual items which are grouped for convenience into consign- ments or batches and a decision must be reached concerning the quality of each batch. Ac- ceptance sampling is a technique that can be used to reach such decisions. It should be noted that we are unable to adjust the production process of the supplier either directly or immediately. We are therefore not inspecting the incoming batch as if it had just been produced by our own process, nor are we inspecting the batch in order to save the supplier the trouble of using cusum techniques to control his process.The purposes of acceptance sampling are (a) to pre- vent excessively “contaminated” batches from entering our production system and (b) to make the supplier aware (by the return of whole batches, if necessary) that only those batches which reach an agreed quality standard will be accepted. Quality control is not, however, confined to the scene of production. Batches and Samples Let us suppose that we are a company manufacturing pharmaceutical products, which are then packaged in screw-top glass bottles. The glass bottles are purchased from a supplier who is prepared to despatch orders for any number (he prefers orders that are multiples of 200 as the bottles are pre-packaged in cartons that contain exactly 200).Our usage of these bottles averages 1000 per week throughout the year and we have in the past placed regular orders for 7000. Many factors were taken into account when fixing the number in this order but it is doubtful if any consideration was given to the difficulties of inspecting the incoming consignments. It is very important that the bottles are compatible with the screw-tops, yet it is clearly impossible to examine each of the 7000 bottles in a consignment. We shall therefore need toMay, 1950 ANALYTICAL QUALITY CONTROL 167 take a sample, then after inspecting each item in the sample we can reach a decision about the acceptability of the whole consignment.How many bottles should we take in our sample and how will we decide whether or not the consignment is acceptable? For answers to these questions we can turn to several publica- tions, of which the two most popular are perhaps that by Dodge and Romigl and BS 6001.293 To obtain a sampling scheme from either of these publications we must specify our require- ments and the form of this specification will not be exactly the same for the two publications. One piece of information which must be supplied, whether we use Dodge and Romig or BS 6001, is the batch size or lot size (in Britain we speak of a “batch” whereas in the USA the word “lot” is more popular). It is certainly not necessary to specify a batch size of 7000.Although the consignment which has just arrived in the warehouse contains 7000 bottles we can specify a larger number or a smaller number for the batch size. Perhaps at this moment there are two consignments waiting to be inspected, in which event it might be advisable to use a batch size of 14000. On the other hand, we could regard the new consignment as consisting of seven batches, in which event the batch size would be 1000. The specification of the batch size should not be taken lightly as there is an economy of effort to be gained by having a large batch. Although a larger batch size will call for a larger sample size the latter is not directly proportional to the former. With a large batch size the sample will be a relatively small proportion of the batch.This point is discussed fully in texts such as those by Davies and Goldsmith4 and Johnson and Leone.5 It is possible that taking a random sample will be easier with a small batch. Much more important is the possibility that a smaller batch might be more homogeneous than a larger batch. If all 7000 bottles of the new consignment were produced in the same production run, whereas the bottles in the previous consignment were produced in a different run, it would be very unwise to group the two together to form one batch of 14000 bottles. This assertion is based on the assumption that most of what the supplier produces will be acceptable and that the purpose of using acceptance sampling is to detect those bottles manufactured under atypical conditions when the supplier’s process is out of control.Having specified that the batch size will be 7000 we must now supply additional information if we are to use Dodge and Romigl or BS 60012 tables. At this stage it is necessary to under- stand several technical terms if one is to use the tables in the recommended manner, and it is unfortunately true that the definition of these terms leaves room for improvement in both publications, especially BS 6001. I will therefore in this brief outline avoid all use of technical jargon and illustrate the selection and use of acceptance sampling schemes by the use of graphs. The type of graph which will be particularly useful is known as an operating characteristic. What batch size shall we use? There may also be advantages in having a smaller batch size.The Operating Characteristic Let us suppose that we are considering the possibility of using a sampling scheme which we shall call sampling scheme A and which is defined as follows: Sampling scheme A : Take a random sample of 42 bottles from the batch of 7000 bottles and accept the whole batch if the sample contains one or less defective bottles. If we put this sampling scheme into operation, what are the probable consequences? What is likely to happen to the next batch which is inspected, and does the same fate await subse- quent batches? It is wise, of course, to have asked and answered such questions before inflicting a sampling scheme upon your company. The introduction of an unsuitable sampling inspection scheme can have an adverse effect on staff morale, on relations with suppliers and on company profit. Fortunately, the operating characteristic can help us to answer such questions and the operating characteristic in Fig.1, which was obtained from Dodge and Romig,l can help us to predict the consequences of using sampling scheme A. I t can be seen from Fig. 1 that the only batches that are certain to be accepted are those with 0% defectives. The larger the percentage of defectives in a batch the smaller chance it has of being accepted and the greater is its chance of being rejected. The over-all performance of sampling scheme A is summarised in Fig. 2, which indicates the outcome of operating the168 ANALYTICAL QUALITY CONTROL A n d Proc. scheme over a period when 1000 batches are received from the supplier.The quality of these batches varies, with the distribution of percentage defective being what we might reasonably expect to receive from this supplier. 0 5 10 Proportion defective in batch, % Fig. 1. Operating characteristics for sampling scheme A. It is clear from Fig. 2 that sampling scheme A is unlikely to be satisfactory. We shall be unhappy because of the high risk of a badly contaminated batch passing through to the bottl- ing process, whilst the supplier will be very unhappy because so many high-quality batches are being returned. To simplify the discussion of the risks involved in using such a sampling scheme, let us define good and bad batches as follows : a good batch contains less than 2y0 of defective bottles, whilst a bad batch contains more than 5% of defective bottles.These definitions represent a compromise between what is desirable and what is attainable. We have confidence that a high percentage of batches produced by this supplier will contain less than2yo of defective bottles, provided he operates his cusum control scheme conscientiously. We also know from experience that a batch with 2% of defectives will cause only occasional interrup- tions to our bottling activity and the disruption of the smooth flow will be tolerable. When batches with more than 5% of defective bottles enter our system, however, the interruptions become more frequent and the disruption becomes intolerable, and therefore the number of such batches that avoid detection at the acceptance sampling stage must be reduced to a very low level.Using these definitions of good and bad batches, we can tabulate the risks as shown in Table I. TABLE I DEFIKITION OF RISKS Decision Accept the batch . . . . Reject the batch . . . . Quality of batch r----p-4Lp -- Good Bad (less than 2q/, of defectives) Marginal defectives) 7 (more than 50,; of . . Correct decision . . Less than 220/6 chance of making this error - Less than 38:; chance of making this error - Correct decision The figure of 38% in Table I is often called the consumer’s risk: with sampling scheme A we. the consumer, run a 38% risk of accepting a batch which contains 5% of defectives. A batch that was even worse (i.e., more than 5% of defectives) would have a smaller chance of being accepted. On the other hand, the figure of 22y0 in Table I is often called the producer’s risk, as this is the risk run by the supplier of a good batch being returned to him.The two risks are illustrated in Fig. 3. Clearly, it is in the long-term interests of both the consumer and the producer that these risks should be reduced. Whilst our decision is based on a sample the two risks will be present, but we can change the two risks by altering the acceptance sampling scheme. For example, we can reduce the producer’s risk by changing the acceptance number from 1 to 2 ( i e . , acceptMay, 1980 ANALYTICAL QUALITY CONTROL 169 Sampling scheme A is used to monitor the quality of 1000 incoming batches 180 batches between 0% and 0.99% defective 373 batches between 1% and 1.99% 225 batches between 2y0 and 2.99% 101 batches between 3% and 3.9976 53 batches between 4% and 4.99% 28 batches between 5% and 5.99% 40 batches with 6% or more defective defective defective defective defective defective Accept - Reject - 748 batches are accepted ~~~~~ 175 batches between 0% and 0.99% 321 batches between 1% and 1.99% 158 batches between 2% and 2.99% 57 batches between 3% and 3.99% 23 batches between 4% and 4.99% 9 batches between 5% and 5.99% 5 batches with 6% or more defective defective defective defective defective defective defective 252 batches are rejected 5 batches between 0% and 0.99% 52 batches between 1% and l.99yo 67 batches between 2y0 and 2.99% 44 batches between 3% and 3.99% 30 batches between 4% and 4.99% 19 batches between 5% and 5.99% 35 batches with 6% or more defective defectivc defective defective defective defective defective Fig.2. Effect of using sampling scheme A. the batch if the sample contains two or less defectives), but this would increase the consumer’s risk. Alternatively, we can reduce the consumer’s risk by changing the acceptance number from one to zero, but this would increase the producer’s risk. This step will, of course, increase the cost of operating the acceptance sampling procedure. Consider the two alterna- tive sampling schemes defined as follows : In order to reduce both risks, we must increase the sample size. Sampling scheme B: Take a random sample of 125 bottles from the batch of 7000 bottles and accept the whole batch if the sample contains four or less defective bottles. Sampling scheme C: Take a random sample of 280 bottles from the batch of 7000 bottles and accept the whole batch if the sample contains nine or less defective bottles.170 ANALYTICAL QUALITY CONTROL Anal.Proc. From Dodge and Romig,l we can obtain the operating characteristics of these two sampling schemes, which are illustrated in Fig. 4 together with the operating characteristic of sampling scheme A. 100 s Producers risk = 22% ,2 C \ . 100 - u 4- ,“ co 0 m GO 40 - V + + 0 6 20 2 0 5 10 Proportion defective in batch, .- 4-J 0 G. LL 5 10 Proportion defective in batch, ”/b Fig. 4. Operating characteristics for sampling schemes A, €3 and C. Fig. 3. Illustration of consumer’s risk and producer’s risk. Clearly, we get better discrimination between good and bad batches if we use a sampling This point is emphasised in Table 11, which gives the scheme with a larger sample size.consumer’s risk and the producer’s risk for the three sampling schemes. TABLE I1 COMPARISON OF CONSUMER’S AND PRODUCER’S RISKS Sampling Sample Consumer’s Producer’s scheme size risk, ?& risk, yo A 42 22 38 R 125 10 22 C 280 G 10 In selecting one of the three sampling schemes to put into operation we must balance the benefits of the larger sample size (ie., lower risks) against the extra costs involved. We have considered only three alternatives but there are available in Dodge and Romigl or BS 60012 hundreds of operating characteristics from which to choose. Each operating characteristic tells one clearly what one can expect from the acceptance sampling scheme it represents.Double Sampling The person operating the scheme would take a random sample of 280 bottles from each batch and check each bottle, paying particular attention to its ability to mate with the “standard” screw-top. Sooner or later this person would express dissatisfaction with the instructions he had been asked to follow. He would probably complain that the taking o€ such a large sample was unnecessary. With very good batches and with very bad batches it would be possible to reach a decision using a much smaller sample. If the cost of taking the sample is relatively high, it is very tempting to take a smaller sample and yet, as we have seen, a large sample is necessary to discriminate between the batch with 2% of defectives and the batch which has 5% of defectives.A solution to this problem is to use a double sampling scheme, in which we first take a small sample and, after inspecting the bottles in this sample, decide whether to accept the batch, reject the batch or take a second and larger sample. Let us consider the double scheme described below and compare its performance with the single sampling schemes A, B and C. Suppose that we decided to adopt sampling scheme C.Alay, 1980 ANALYTICAL QUALITY CONTROL Sampling scheme D : Take a random sample of 85 bottles from the batch of 7000 bottles. If the sample contains 0 or 1 defective bottles, accept the whole batch. If the sample contains 10 or more defectivz bottles, reject the whole batch. If the sample contains between 2 and 9 defective bottles, take a second random sample of 205 bottles.If the total number of defectives in the two samples is less than 10, accept the whole batch. If the total number of defxtives in the two samples is 10 or more, reject the whole batch. 171 The performance of this double sampling scheme can be described by an operating character- istic in the same way that we described the single sampling schemes. From Dodge and Romigl we can obtain the operating characteristic in Fig. 5, which is very similar to that for sampling scheme C, having a consumer's risk of 6% and a producer's risk of 13%. It would appear that the double sampling scheme (D) offers the advantage of reduced effort because, for some batches at least, we will extract and inspect only 85 bottles. Other batches will demand a total sample size of 85 + 205 = 290, which is only slightly larger than the 280 taken from every batch if we use the single sampling scheme (C).To quantify tKe effort involved in the double sampling scheme we calculate the expected sample size. Expected sample size = 85 + (205 x probability of taking second sample) The required probabilities can be calculated from the hypxgeometric distribution or less accurately from the binomial di~tribution.~~~ They can also be estimated from appropriately chosen operating characteristics in Dodge and Romigl or in BS 6001.2 Regardless of how the probabilities are obtained, they will depend upon the quality of the batch and therefore the expected sample size will in turn depend upon the quality of the batch. Fig. 6 shows how the expected sample size will vary for sampling scheme D.This is a weighted average of the two possible sample sizes, 85 and 290: $ 1 w- 0 c 0 0 .- 4 4 L a 00 80 60 40 20 300 .- v) U 5 10 G o Proportion defective in batch, % 0 5 10 Proportion defective in batch, % Fig. 6, Double sampling scheme (D). Fig. 5 . Operating characteristics for sampling scheme D. When using Fig. 6 to assess the saving in cost or effort we must not forget that a large per- centage of the batches received from the supplier will hopefully contain less than 2% of defective bottles. It is with these good batches that the double sampling scheme offers sub- stantial savings, as many such batches will be accepted without taking the second sample. Conclusion Acceptance sampling is used primarily to reach decisions about the quality of incoming batches.When these batches consist of items that can be classified as defective or satis- factory we can use the Dodge and Romigl or BS 60012 tables to select a suitable sampling scheme. The conventional method of using these publications requires the user to specify his requirements in a technical jargon that is difficult to master. In this introduction to accept- ance sampling, an alternative approach has been suggested that avoids contact with the tech- nical jargon by using the operating characteristic to describe an acceptance sampling scheme. By examining the operating characteristics presented in Dodge and Romigl or BS 60012 one172 ANALYTICAL QUALITY CONTROL Anal. Proc. can select a single or double sampling scheme that will discriminate sufficiently between “good” and “bad” batches. References 1.2. 3. 4. 5. 6. Dodge, F. H., and Romig, H. G., “Sampling Inspection Tables,” John Wiley, New York, Chichester, BS 6001 : 1972, “Sampling Procedures and Tables for Inspection by Attributes,” British Standards BS 6000 : 1972, “Guide t o the Use of BS 6001,” British Standards Institution, London. Davies, 0. L., and Goldsmith, P. L., “Statistical Methods in Research and Production,” Oliver and Johnson, N. L., and Leone, F. C., “Statistics and Experimental Design,” Volume 1, John Wiley, New Grant, E. L., and Leavenworth, R. S., “Statistical Quality Control,” McGraw-Hill, New York, 1972. 1959. Institution, London. Boyd, Edinburgh, 1972. York, Chichester, 1964. Cumulative Sum Techniques J.D. Chamberlain ICI Organics Division, Hexagoit House, Rlackley, Maizchcster The use of cumulative sum techniques (cusums) has been developed to a large extent within ICI over the past few years. It is a method of obtaining information from a time series of values on any variable of particular interest. The only essential data are the measurements of the variable, in the correct time order. Cusums are used to determine the points a t which significant changes in the average level of the particular variable have occurred. Illustration of Cusum Plotting On a particular plant, the reference value for the yield for each batch is 90. The following 50 batch yields were obtained] in the order shown, i.e., 88, 86, 90, etc. : 88 86 90 89 92 86 90 86 90 89 87 89 89 87 92 87 91 91 90 86 92 94 90 89 93 92 93 93 92 91 94 90 93 88 92 88 92 90 90 91 87 91 88 93 89 92 90 89 90 89 In order to plot the cusum graph the data in Table I is first calculated.The graphs in Fig. 1 illustrate firstly the plot of batch yield against batch number (from which little could be determined) and secondly the plot of the cusum column of the table against the batch number. Looking at this second plot, we can see that : from batch 1 to batch 20 the slope appears negative from batch 21 to batch 33 the slope appears positive from batch 34 to batch 50 the slope appears horizontal In fact, from simple arithmetic: the mean yield for batches 1-20 is 88.75 the mean yield for batches 21-33 is 92.00 the mean yield for batches 34-50 is 89.94 This example indicates that in some way the mean level of a variable corresponds to the slope of the cusum graph. If the slope is positive, then the current mean is greater than the reference value.If the slope is negative] then the current mean is less than the reference value. If the slope is horizontal, then the current mean is very near the reference value. Cusums are used mainly for two distinct purposes : post-mortem investigations and quality control.iWay, 1980 ANALYTICAL QUALITY CONTROL TABLE I CALCULATION OF CUMULATIVE SUMS Observation number 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 Batch yield 88 86 90 89 92 86 90 86 90 89 87 89 89 87 9% 87 91 91 90 Batch yield - 90 -2 -4 0 -1 2 -4 0 -4 0 -1 -3 -1 -1 -3 2 -3 1 1 0 - 86 92 94 90 89 93 92 93 93 92 91 94 90 93 88 92 88 9% 90 90 91 87 91 88 93 89 9% 90 89 90 89 -4 2 4 0 1 3 2 3 3 2 1 4 0 3 -2 2 -2 2 0 0 1 -3 1 - 2 3 - 1 2 0 -1 0 -1 173 Cusum -2 -6 -6 -7 -5 -9 -9 - 13 - 13 - 14 - 17 - 18 - 19 - 22 - 20 - 23 - 22 -21 -21 - 25 - 23 - 19 - 19 - 20 - 17 - 15 - 12 -9 -7 -6 -2 -2 1 -1 1 -1 1 1 1 2 -1 0 -2 1 0 2 1 1 0 9 - Post-mortem Investigations The aim of this type of application is to detect any significant changes that have occurred in the mean level of the variable.The follow-up to this would be to try to relate these significant changes (e.g., yield changes) with other factors, e g . , changes in the plant operating conditions. All of the data that are to be considered must already be available.Graphical Method Denote the observations by xl, x2, . . ., x,. (i) Plot the cusum graphically (as in the earlier example), but now instead of taking deviation from some reference value R, take deviations from the over-all mean Z (as we have taken deviations from 3 our cusum value must finally return to 0).174 ANALYTICAL QUALITY CONTROL Anal. Proc. (ii) Determine the point at which the cusum graph is a maximum vertical distance from the base line (in either direction). 10 20 30 40 50 -3U Observation number Fig. 1. Graphs of (a) batch yield against batch number and ( b ) cusum against batch number. (iii) Some estimate of the standard deviation of the results, under normal conditions, must If possible this is estimated from previous data; otherwise the standard deviation is be made.estimated from the data by s = This formula is only for use in this form of analysis. deviation formula. (iv) Calculate It is yzotgenerally applicable as a standard 1 maximu; cusum ~ I and then compare this value with the appropriate value The span is the number of observations over which the maximum cusum is from Table 11. sought. TABLE I1 POST-MORTEM SIGNIFICANCE TESTING 1 max. cusum I I Length of span Critical value of 1 s I 2.7 a n 6 * A 1u 12 20 30 40 50 60 70 80 90 100 110 123 ,3 . 0 4.0 5.3 6.7 8.0 9.1 10.0 10.8 11.5 12.2 12.8 13.4 14 0May, 1980 ANALYTICAL QUALITY CONTROL 175 If the calculated value of 1 is greater than the appropriate value from Table 11, this implies that there has been significant change in mean level at the point of the maximum cusum.(v) If part (iv) indicates a significant change, further tests can be applied to assess whether or not there are any other significant changes. V-mask Method A new point on the cusum graph is plotted as each individual result becomes available. The chart can then be used to indicate when the process is out of control, that is, when the process mean is away from the reference value which in this application is the desired target value. In this graphical method, it is essential that if 1 observation is represented by 1 cm on the horizontal scale, then 1 cm on the vertical scale must correspond to about 20, where 0 is the short-term standard deviation of the process. For example, if 20 = 2.08, round this to 2.0 for ease of plotting; and if 20 = 30.6, round this to 30.As each cusum point is plotted on the graph, we superimpose on this graph a V-mask, which can be used to indicate lack of control. j maximu; cusum Quality Control For use of the V-mask, see Fig. 2. V-mask - V-mask Current point I I I I t I , I I I 1 2 3 4 5 6 7 8 9 10 Observation number Fig. 2. Illustration of use of V-mask: (a) process mean close to reference value[and (b) process mean higher than reference value. The V-mask is completely defined by 8, the semi-angle, and by the parameter d. The effects of difference values of 8 and d are shown in Figs. 3-6. For example, if d = 1 and tan 8 = 0.5: (i) When the process mean is exactly on target, lack of control will be incorrectly indicated by the mask about once every 19 results.(ii) If the process mean moves lc away from target, on average it will take four or five results before lack of control is indicated. Figs. 3, 4, 5 and 6 are only applicable for determining d and 8 if the above scaling is used.176 ANALYTICAL QUALITY CONTROL If d = 1 and tan 8 = 0.8, the corresponding figures are (i) about once every 700 results; (ii) after about 30 results. 1000 500 w - g 200 m - 0 100 5 50 40 - w 30 01 10 - C 2o I 0, : 5 2 ; 2 1.5 1 .o I 1 \\ 0 U 2u 3a Displacement of current mean AnaL. Proc. loo0 t Y’ll\lx 500 200 100 50 40 30 20 10 5 4 3 2 1.5 1 .o 0 0 20 30 Displacement of current mean Fig. 3. Effect of d and 8: d = 1 ; tan 8 = (I) 0.50, Fig. 4. Effect of d and 8: d = 2; tan 8 = (I-) (11) 0.60, (111) 0.70, (IV) 0.80.0.40, (VI) 0.60, (VII) 0.55, (WIT) 0.60, (IX) 0.70. For simplicity, we have only considered a two-sided scheme, where we are interested in deviations on both sides of the target value. V-masks can also be used to consider deviations on one side of the target only, e.g., if we wish to ensure that a process was above specification. 0 (I 7a 3tJ Displacement of current mean Fig. 5. Effect of d and 8: d = 5 ; tan 8 = (X) 0.30, (XI) 0.35, (XII) 0.40, (XIII) 0.45. 0 U 20 30 I .u Displacement of current mean Fig. 6. Effect of d and 8: d = 8 ; tan 8 = (XI\.) 0.25, (XV) 0.30, (XVI) 0.35. References 1. 2 . “Cumulative Sum Techniques,” ICL Monograph No. 3, Oliver and Boyd, Edinburgh. Davies, D. L., and Goldsmith, P. L., “Statistical Methods in Research and Production,” Oliver and Boyd, Edinburgh, 1972, Chapter 10.May, 1980 ANALYTICAL QUALITY CONTROL The Impact of Automation in the Pharmaceutical Industry 177 T. E.V. Horsley” Low Wood, Woodhill A venue, Gerravds Cross, Buckinghamshire The scientific and technological age in which we live has so far been of relatively short duration. The discoveries and elucidation of the structures of complex natural products have been followed by innovative chemistry to enhance the physiological reactions. New compounds have also been developed with characteristics not shown by those found in natural materials. The pharmaceutical industry as we know it today has really grown up during the last 60-70 years, and the last 50 years is the period to be considered here because 1928 was when the major legislative changes regarding foods and drugs really began.Always by potion, blood-letting, simples, witchcraft, spell, faith and even poisonous substances, people have sought to regain a healthy mind and body from the ravages of disease. The Arabs knew that rotting dung was good for the treatment of pustulating sores. Whether incurred from endemic sources or from mis- spent youth or self-induced maltreatment in older years, the human and the animal body has needed some form of drug or chemotherapy. We have seen improvements in nutrition incorporating vitamins and dietary changes and now live in a familiar world of hormones, vaccines, antiseptics, antibacterials, antibiotics, barbiturates, tranquillisers, etc. Some of these were hardly needed, if at all, 50 years ago but our life is also changing rapidly.Whilst the Earl of Shaftesbury and Wilberforce did much to improve the life of children and slaves in the past, we in the modern world have succeeded in two achievements. Firstly, we have improved the duration and quality of life by raising the standard of general health through advances in nutrition, medicine and surgery to the extent that most of the western world and much of the third world suffers less seriously, if at all, from many of the rampant and often killer diseases of the past. Secondly, our Western civilisation has so modified our way of life as to create economic and social pressures to which many cannot easily adjust. They become prisoners of the modern world because of the pace of life. It has been said that among other things a modern executive needs about 20 tablets per day to cope with his enforced life.He eats too much irregularly, sleeps badly, crosses time barriers, attends business meetings, is unable to concentrate or take sensible decisions, and suffers from indigestion, hangovers, heart disease, constipation, etc. He may be difficult to work with and almost impossible to live with. In short, we have placed a demand on the practising pharmacist for supplies of a wide range of often highly potent dosage forms to be available with or without a prescription on a world- wide distribution. Over-the-counter and prescription products are now made and supplied in bulk to warehouses and dispensaries. The development of health services extensively avail- able on demand for the whole population further emphasises the problems-more people are a t risk ! In the 1930s, with both foods and pharmaceuticals, comprehensive batch analyses were unusual.It was often a consultant who was called in occasionally to do what seemed to be necessary. Formulations were checked more by watching the job done but some analyses to BP standards were performed. I t was an offence to supply anything which was “not of the nature or substance demanded.” Now under the Medicines Act 1968 there is legal control by licence of manufacture, importation, storage, distribution and sale of most dosage forms (see Fig. 1 ) . The industry is vulnerable because of these changes: it is a political and economic football. Many of the quality standards result from public emotion or from a fear of public reaction.Its sheer size of operation makes for difficulties in control, at least by older methods, and standardisation means large batches, long process times and large-scale bulk purchases of raw materials. Highly reactive chemicals and potent products are associated with internal and The practice of medicine is not new to mankind. He also risks coughs, colds, tropical diseases or dysentery. The Food and Drugs Act 1928 required that the customer “should not be prejudiced..’’ * Formerly of Glaxo Operations UK Limited.178 ANALYTICAL QUALITY CONTROL Anal. Proc. external safety problems and complete staff and usage records. Information may be invalu- able and present a security risk. The development costs are very high and a recovery of cost with profit to follow is essential for the future.Pharmaceuticals in 1979 ?.nust : (a) Equate to original product licence samples in respect of: formula ; source materials ; manufacturing methods of produc- tion and of control. ( b ) Give designed clinical response. (c) Have similar physical, chemical and biological characteristics to (d) Conform to packaging and labelling standards. (e) Be made and released under the control of “the responsible person,” a resident or consultant acting in accordance with the legislation and whose appointment is approved by the licensing body. original samples. Fig. 1 . Control of pharmaceuticals. There is need for accurate, comprehensive data acquisition for the industry to protect its interests.Background information on suppliers and materials, both quantitative and qualitative, is essential. Routine analysis of both a subjective and an objective nature must be reliable in spite of the fact that it is monotonous, nearly always within the desired range and often remarkably consistent in values. It requires close attention to detail, often from rela- tively unskilled staff. There are also specialised analyses to be developed and operated by skilled and experienced graduate staff. All this must be obtained, examined and recorded, and the total information giving raw material, production history, analysis, decisions and distribution records must be collated. It must be presented with selection if necessary, as required for the many interested authorities.Speed is essential, as it is also with investigation of complaints, problems or plant research. The data must be capable of rapid disentangle- ment, assessment and rapid reporting. Staff education and constant retraining for changes of techniques or simply to rejuvenate attitudes give no certainty, and recourse to automation wherever possible is now essential. Requirements for automation are given in Fig. 2. An example of an unstable reaction is that of cephalosporin C in broth samples, in which a rapid assessment is of value in guiding the fermentation controller to the proper time for extraction with maximum yield. Cephalosporin was shown by Abraham and Newton1 to hydrolyse at 100 “C in the presence of 1 N hydrochloric acid to give two compounds. Subse- quent reaction in the AutoAnalyzer system with diazotised sulphanilic acid was shown by Shaw2 to give a transient colour easily quantified in this system.Other aspects of automated process control and data processing are illustrated in Figs. 3-5. Examples were given to illustrate the need to review and check the performance of automatic equipment before acceptance and during use. (a) pH electrodes can dissolve or disintegrate in effluent systems to give an apparently acceptable value of 7.0 from liquids which are very severely off-standard. ( b ) Irregularities in the response curve in an iron(II1) - Maltol assay of streptomycin using a standard input solution were shown to be caused by the “sine-curve” operation of a thermostat operating at a temperature between 60 “C and just below 100 “C.The errors could only be contained by careful positioning of replicates of sample and standard solutions but with a loss of instrument capacity. Some of the pitfalls involved are illustrated in Fig. 6. In conclusion, it is obvious that for self-protection and in furtherance of legitimate business, a manufacturing company develops and accumulates a vast amount of data from research and179 manufacturing to which may, or should, be added any feedback of experience from the medi- cal profession. The continuous batchwise control and the interchange between the company and many authorities with differing and often opposing views is not really helping the patient. The p(ficesslty for automation is clear, and the industry could not now manage without it.The posslblllties of integrated automation lead me to believe that the analyses of samples by abstraction from stock for the company and official laboratories is fast becoming both m1s- leading and wasteful of time and money. May, 1980 ANALYTICAL QUALITY CONTROL A utomation-A nalysis Quantity -Good for numerous samples Quality -Replication good -Always attentive -Controlled errors-Masses Volumes Temperature Time -Standardised response for unstable reactions -Does not “flag” with boredom -Can be used to programme operators Escieizcy Reporting -Quick to instantaneous A ccess -To pre-selected staff Other advantages-Saving of sample size -Saving of washing-up -Can incorporate combinations of units ~~ ~ Fig. 2. Requirements for automation of analysis. Automated and Semi-automated Process Control Raw materials-Assay and detection of rogues -On-line control for quality -Batching control for quantity -Any measurable variable : time, temperature, pH, -Continuous or programmed sampling -Unit checks-tablets, capsules, ampoule clarity, etc.-On-line control including selective mechanism -Measurements of volume and quality, automatic Proccss air flow, ammonia addition, etc. Eflzrent adjustments of pH. etc. Fig. 3. Examples of automated and semi-automated process control. National pride encourages the controlling authorities in different parts of the world to insist upon being seen to be different and to require their own clerical and laboratory assessors who are frequently not as well equipped, as knowledgeable or as reliable.as the manufacturing company’s own controller and control staff.There is clearly room for international and, as far as the manufacturers are concerned, a mutual acceptance of automated systems as a means to achieving the consistency in quality demanded at as low an over-all cost as possible. After all,180 Anal. Proc. ANALYTICAL QUALITY CONTROL ~~~ ~~ ~ ~~ Automation-Data Processing I Records for raw materials-Source, quality, stock Process -Batching details -Control tests -Yield and quality -Efficiency Analytical records Distribution -Allocations, marketing Availability of data -Instant to authorised persons -Print-out if needed Fig. 4. Application of data processing to automation. Automation-Data Processing 11 (1) Operates a pass/fail system on objective data (automatically (2) Can operate with selected subjective data (fed in by senior) (3) Gives instant detection of trends-danger warnings or alarms (4) Quickly reviews data for complaints-highlights common (5) Rapid data presentation from filed records-will assess need for derived) features across batches recall and facilitate action Fig.5. Further application of data processing to automation. (1) Installation expensive (2) C.ustom-built (3) Running costs -Keep it simple-data cost money -1s it versatile or how long to convert : -Excess capacity ? -Wasteful of reagents? -What quality of staff needed? -What service facilities are available -How many VDUs ? minutes, hours or never? in emergency ? (4) Validation -How often can and should pcr- (5) I s program restrictive? -You get what your program gives I f ormance be checked ? you ! ’ ~ _ _ _ _ _ _ ~ ~ Fig.6. Illustration of some of the possible pitfalls. the customer pays eventually and with careful application of the automation available he “will not be prejudiced-he will get the quality and substance demanded.”May, 1980 ANALYTICAL QUALITY CONTROL 181 I acknowledge the kind assistance of the Directors and members of the staff of Glaxo Operations (UK) in the preparation of this paper. References, 1. 2. Newton, G. G. F., and Abraham, E. P., Biochem. J., 1956, 62, 651. Shaw, W. H. C., Internal Confidential Report, Glaxo, Greenford, May 1960. Practical Considerations in the Analytical Use of Microprocessors D. Betteridge Department of Chemistry, University College of Swansea, Singleton Park, Swansea, SA2 8PP The microprocessor-based microcomputer is compact, cheap and robust.It is therefore practical and economic to use microprocessors to computerise small-scale instrumentation such as titrators, polarographs and balances. At present, it is less profitable to use them for large instruments such as mass spectrometers. Firstly, they can be used to replace mechanical or electronic parts in an instrument. Thus, a mechanical programmer may be replaced by a chip, with consequent saving in manufacturing cost and increased relia- bility. Such changes may benefit the analytical chemist as a consumer, and they do account for the proliferation of advertising along the lines “instrument X, now with microprocessors.” Secondly, they can be incorporated into instruments as controlling devices, and thirdly, they are capable of data processing. The analyst looking at his own problems of computer applications has to make the choice of micro or mini and in doing so he has to take into account the technical problems of program- ming and interfacing.The options open and practical difficulties are well illustrated by con- sidering the design and construction of a microprocessor controlled automatic titrator. Between 1975 and 1979, several groups have devised microcomputer-controlled titrators.l-’ Because of differences in their philosophy and the available state of microprocessor develop- ment, the instruments proposed differ from one another in several respects. The first developments were carried out simultaneously but independently in our own laboratories1 and in those of Radiometer2 and Mettler.Ours was basically a three-cornered collaboration between an electrical engineer (E. L. Dagless) and industrial chemist (D. R. Deans) and an academic chemist (D. Betteridge), with the research student (P. David) working in whichever laboratory was appropriate at the time. The problem as posed was to see whether the titration of a mixture of organic acids and boric acid as found in the ICI nylon process plant could be titrated sequentially without addition of mannitol, and with an accuracy of 0.5% or better. The existing procedure was manual and consisted in first titrating potentiometrically the organic acids, and then the boric acid after addition of mannitol, to a pre-determined potential.In analytical instrumentation they have three distinct functions. It is these last two roles which are discussed in detail here. Laboratory Prototype The first stage was to find a solution to the chemical problem, given the capabilities of the microprocessor. Three possibilities were considered : (i) to mimic a conventional automatic titrator, by getting feedback to the microprocessor, which would then control the rate of addition of titrant until a pre-set potential was reached and taken as the equivalence point; (iz) Gran’s method; and (iii) to use a differential method to locate the. equivalence point. The first of these is straightforward and can lead to a practical improvement in existing instrumentation. However, we rejected it on the grounds that there was little novelty in such an adaptation and that it did not make full use of the potential of the microprocessor. The use of Gran plots seemed to have many attractions,s but in practice it was found that the curves deviated from linearity near the equivalence point and even refinements of Gran’s equations such as those due to Johanssong and Pehrsson et aZ.lO,ll were not much more success- ful.It has the disadvant- ages that the peak in the differential curve coincides with the equivalence point only when the The use of a differential of the titration curve was then investigated.182 ANALYTICAL QUALITY CONTROL Anal. Proc. titration curve is symmetrical and it makes most use of the data close to the equivalence point. These disadvantages were offset by the ease of computation and the applicability of the @or- ithm to many systems, no knowledge of any equilibrium constants being required.The development of the program was performed via a terminal on a main frame computer, with a high-level language. Consequently, the time involved in the development of the best solution to the problem was used economically. For use by the microprocessor, the program had (in 1976) to be converted into assembly code, a compelling reason for using the simplest adequate program. The automatic burette, pH meter and electrode assembly were taken from a conventional autotitrator. The control functions were exerted as a pulse to a relay, and the analogue to digital conversions and vice versa were carried out by cheap solid-state devices.Programming the microprocessor in assembly code took some time and was facilitated by the availability of fast punched-tape readers and simulated PROM. Two titrations could be per- formed in sequence in 1 min with a coefficient of variation of 1%. If 4 min were taken the precision improved to 0.25%. At the higher rate there was a slight systematic error, but at the slower rate the accuracy was as good as with a manual titration. The elimination of mannitol from the procedure resulted in sufficient savings in the course of 1 year to pay for the material cost of the apparatus. The approach of Christiansen et a1.2 was similar, except that they used the titration data to slow the rate of addition of titrant as the equivalence point is approached.If one were to carry out a similar exercise now, the major change in procedure would be to do the program development work on a PET or similar microcomputer, with a suitable interface. This would automatically carry out the tedious step of converting an acceptable high level com- puter language (BASIC) to machine code and also reduce the cost of the development system that is needed to support the microprocessor. However, even a PET linked to the basic coni- ponents of an automatic titrator represents a laboratory “lash-up.” The microprocessor was an Jntel8008. As a laboratory instrument, it performed very satisfactorily. Control Laboratory Model The conversion of the laboratory prototype to a robust control laboratory instrument re- (i) There should be as few control knobs or switches as possible; ideally the operator should (ii) it should be as compact and robust as possible, thus taking full advantage of solid-state (iii) it should be applicable to more than one titration; (iv) it should have error-detecting routines that indicate common malfunctions such as broken connections, defective electrodes, etc.Further work was necessary to meet these objectives, but the basic program has been expanded to incorporate control and error-checking routines and modified so that the equival- ence point is detected “on the wing.’’ The pH meter was replaced by a voltmeter on a chip costing about L25. The end result is a robust, compact titrator which is functioning well under industrial conditions. quired a further definition of objectives. only have to press a run button and read off the result; electronics ; We held the following to be important: Other Approaches The strategy outlined above in both stages of development places emphasis on simplicity and implies a willingness to dedicate a microprocessor to a single function.We believe that this emphasises the quality of a microcomputer that distinguishes it from a minicomputer, i.e., its smallness. At the outset of the work there was a severe limit on the capacity of a microproces- sor, and this reinforced our philosophy in a practical way. Large-scale integrated (LSI) circuits have increased so dramatically that with a PET or similar microcomputer one can obtain many of the facilities of a minicomputer and have, by the standards of Intel 8008, immense computing power.Even with more advanced microcomputer kits, such as the SDK 80, more complicated programs can be used, and interaction between the operative and the computer is p0ssible.3-~ That of Wu and Malmstadt4 permits the operator to select virtually any type of titration and any appropriate program for processing the titration data. Both linear and logarithmic However, this is no longer true.;Way, 1980 ANALYTICAL QUALITY CONTROL 183 titrations are permitted. However, although the choice of procedure is left completelv to the user, the implementation is computer controlled and all of the results reported are satisfactory for most purposes. The flexibility is gained at the expense of simplicity and cost. Conclusions The only thing certain about microprocessor-controlled titrations is that most workers prefer a differential procedure for the determination of the end-point.There are completely different viewpoints as to whether it is preferable to have fixed or interactive programs. To some extent this depends on whether one is opting for a versatile laboratory instrument or one for process control. By now most of the possibilities have been tried and the consumer should feel confident that any commercial microprocessor-based titrator should be reliable. In view of the development cost if he were to make it himself, he may also feel it is good value for money. As in all computer applications, the really difficult part is getting the chemistry right. 1. 2. 3. 4. 5. 6. 7. 8. I). 10. 11.References BAteridge, D., Dagless, E. L., David, P., Deans, D. R., Penketh, G. E., and Shawcross, P., Analyst, Christiansen, T. F., Basch, J . E., and Krogh, S. C., Anal. Chem., 1976, 48, 1051. Leggett, D. J., Anal. Chem., 1978, 50, 718. Wu, A. H. B., and Malmstadt, H. V., Anal. Chem., 1978, 50, 2090. Avdeef, A., and Bucher, J . J., Anal. Chem., 1978, 50, 2137. Busch, N., Freyer, P., and Szameit, H., Anal. Chem., 1978, 50, 2166. Martin, C. R., and Freiser, H., Anal. Chem., 1979, 51, 803. Gran, G., Analyst, 1952, 77, 661. Johansson, A., Analyst, 1970, 95, 535. Pehrsson, L., Ingman, F., and Johansson, A., Talanta, 1976, 23, 760. Pehrsson, L., Ingram, I?., and Johansson, A., Talanta, 1976, 23, 781. 1976, 101, 409. The Analyst and the Statistician A. R. Rogers Department of Pharmacy, Heriot- Watt University, 79 Grassmarket, Edinburgh Most chemists receive little training in statistics and a course of ten lectures appears to be typical.lJ A recent survey3 of analytical chemists in industry and public service showed that those in practice are aware of the importance of statistical methods, especially in relation to sampling.I hope that those of us in educational establishments will take note of these needs. Titrations When assaying some material by titration or when standardising a volumetric solution, we may well be able to choose the amount (mass or volume) of sample taken so as to give a pre-determined approximate titre. Should we aim at (say) 25 ml from a 50-ml burette or (say) 8 ml from a 10-ml burette? The latter may bring economies, but we may fear that the precision will be poorer.In statistical terms, will there be a significant fall in the coefficient of variation (relative standard deviation) ? We can presumably carry out replicate titrations by both methods and measure the standard deviations of the results. We then divide the larger standard deviation by the smaller and square the quotient, which gives what is called F. If the calculated F is smaller than the F value in tables (for given numbers of replicates and a pre-determined probability level, such as P = 5%), there is no significant difference in the precision. Suppose that we are happy to change over to 10-ml burettes provided that the relative standard deviation does not rise by more than 5%. How many replicates would be necessary to establish the position? The answer is “several thousand.” Even to detect an increase of 18% in the standard deviation requires at least 100 determinations by each method.It is understandable but perhaps not excusable that analysts are reluctant to provide statistical evidence in support of claims of good precision. Let us start with something apparently very simple. Why has the experiment not been done?184 ANALYTICAL QUALITY CONTROL Anal. Pvoc. Luckily, the other check we should wish to make, namely whether the two methods on aver- age give the same results, is not likely to require many replicates. If the relative standard deviation is O.l%, the t-test will reveal a difference of 0.1% in the means with as few as eight determinations by each method.Glenn4 has shown that objective statistical tests do not necessarily give the answer predicted by a chemist’s professional judgment and experience with three examples of t-tests, so care is necessary. Spectrophotometry Suppose that I wish to assay a drug by preparing a solution and measuring the absorbance at an appropriate wavelength. I can choose, within reason, how dilute to make the solution. What is the optimum absorbance at which to aim? Again, this seems a simple question, but it is very difficult to answer. If the spectrophotometeris such that the random errors in the transmittance measurement are independent of the trans- mittance and if only one measurement is made, the relative standard deviation is smallest if the absorbance is 0.434.Presumably because comparisons of standard deviations require very large numbers of measurements, the Twyman and Lothian theory apparently remained untested for over 30 years, until IsmaiP showed that it was approximately correct for the Hilger and Watts Uvispek spectrophotometer. This has recently been confirmed in my own lab~ratories,~ not only for the Uvispek but also for the Unicam SP 500 and SP 600 and the transmittance scales of the Cecil 272. However, the absorbance scales of the Cecil 272 and the Unicam SP 30 and the Perkin-Elmer 200 show constant random errors of the absorbance, and the relative stand- ard deviation is smaller the greater the absorbance. From the point of view of relative stand- ard deviation alone, with these three instruments the absorbance should be made as large as possible.Of course, analysts are interested in accuracy in addition to precision, and with the low levels of intensity of radiation reaching the detector at high absorbances, it is unlikely that a spectrophotometer will measure high absorbances as accurately as low absorbances. Let us return to the spectrophotometers with constant random errors of transmittance measurement. Is it not customary to make two or three measurements on the same solution and to take the average, or those which agree with one another? Also, a good analyst will check that the two cuvettes (cells) match and are properly cleaned, and will measure and subtract a “cuvette blank” (or “cell blank”), which may be small. The optimum absorbance depends not only on the design of the instrument but also on the number of replicate absorbance measurements, whether a cuvette blank is subtracted and the extent of correlation of the “actual” and “blank” measure- ments.’ One theoretical answer was given by Twyman and L ~ t h i a n .~ How often does the analyst make a single absorbance measurement ? Quantity Control The Weights and Measures Act 1979 requires that from January 1st 1980 (a) packaged goods when tested by an authorised inspector shall contain on average not less than the nominal amount, (b) the number of packages containing less than a specified fraction of the nominal amount shall not exceed a specified number (in relation to the number tested) and (c) no packages shall contain less than a specified (smaller) fraction of the nominal amount.The Weights and Measures Act 1963 required that the minimum amount should be stated on the label, although certain relaxations of this requirement are allowed.8 What is our response to this change, as producers and as customers? The change is consequent upon an EEC Direct- ive. As analysts and quality controllers, do we under- stand the implications? The change in approach goes hand in hand with a change from attempting to control quality at the point of sale to building in quality in the manufacturing and packaging process. Similar encouragement to the building in of quality has followed the implementation of the Medicines Act 1968. Those working in the pharmaceutical industry will be well aware of the importance of the controls exercised by the licensing authority and the work of the medicines inspectorate.There is still a British Pharmacopoeia and a European Pharmacopoeia, which provide public standards of quality for medicines at the point of sale or supply, but the manu- facturer must build in higher quality within the factory. It is difficult for the quality con- Do we welcome or regret the change?May, 1980 ANALYTICAL QUALITY CONTROL 185 troller to ensure that this happens unless he has considerable statistical knowledge and ability or access to statistical advice. The consumer may wish to ensure that his tin of beans contains not less than 100 g or his aspirin tablet not less than 285 mg, but this is best achieved by en- suring that the goods are made and packed so as to produce an appropriate average with a small standard deviation and a distribution more leptokurtic than normal. Calibration Graphs Considerable expertise in statistics exists among those involved in biological and micro- biological analysis.The designer of a biological assay may start by exploring the shape of the dose-response graph. Is the biological response a straight-line function of the amount of the preparation administered? Over what range is the line straight ? Is there an intercept? I suggest that a chemist considering titrimetric determination is likely to take it for granted that the titre is directly proportional to the amount of substance taken for assay. In a colori- metric procedure, a calibration graph of absorbance versw amount of substance may be pre- pared, but when direct ultraviolet measurements are being considered for the assay of a drug, argument is likely to centre on the exact value of A:: rather than on whether the value varies with the concentration.When a calibration graph is prepared, an inappropriate statistical test (the significance of the correlation coefficient) may be applied in an attempt to examine the straightness of the line. Imag- ine a colorimetric procedure in which the absorbance at a specified wavelength is a rectilinear function of the concentration, but there is an intercept (the “blank” absorbance) when the concentration is zero. Should one measure the blank directly and subtract it from each actual absorbance, should one estimate it by extrapolation, or should one put the blank solution into the reference cuvette when measuring the absorbances ? These matters have been d i s c u s ~ e d ~ - ~ ~ but unfortunately there is still no clear answer. Of course, the correct treatment depends on whether the standard deviation of the absorbance or the transmittance (or indeed neither) is constant ; the “actual” and “blank” values are positively correlated’ and this must be taken into account; also, the slope and the intercept of the calibration graph are negatively corre- latedl* and this adds further complications. Tablet Assays Many tablet preparations in the British Pharmacopoeia have limits for the content of the active ingredient in a sample of 20 tablets of 95.0-105.0~0 of the prescribed or stated amount. The range is intended to accommodate difficulties of mixing and tablet-to-tablet variation as well as the uncertainty of the assay. For many assays the relative standard deviation is around 1 yo. Suppose that for some new tablet preparation the relative standard deviation of the assay is much worse, say 6%. The answer is “no,” because these figures have to be added and subtracted as their squares. If & 5% is correct with a relative standard deviation of the assay of 1%, then the allowance for everything else is (tj2-l2)$ = 4.9%. We must then combine the values 4.9 and the 6.0 as their squares and take the square root, thus (4.92 + 62)t = 7.75%. That is, appropriate limits would be about 92.5-107.5y0. Let us look also at some of the possible ways of dealing with “blank” measurements. Should the limits be widened by 5% to 90-110%? Conclusion Analysts should learn enough statistics to enable them to ask statisticians the right questions and to understand the answers they receive. I hope that papers such as this will encourage those who educate analysts to pay more attention to statistics and those who are analysts to make more use of statistics and statisticians. Statisticians are used to having to advise specialists in a range of disciplines. References 1. 2. 3. 4. 5. 6. Education and Training Committee, Proc. SOG. Anal. Chem., 1972, 9, 173. Education and Training Group Committee, Proc. Anal. Div. Chem. Soc., 1977, 14, 1. Education and Training Group Committee, Proc. Anal. Div. Chem. Soc.. 1979, 16, 107. Glenn, A. L., Proc. SOC. Anal. Chem., 1967, 4, 116. Twyman, F., and Lothian, G. F., Proc. Phys. Soc., 1933, 45, 643. Ismail, M., PhD Thesis, University of London, 1964.Anal. PYOC. Hamdi, M. A., and Rogers, A. R., Paper presented a t XXI Colloquium Spectroscopicum Internation- Ward, D. H., .J. R. Stat. Soc., 1975, A130, 170. Agterdenbos, J., 2. Anal. Chem., 1957, 154, 401; 1957, 157, 161. Higgins, J., Analyst, 1964, 89, 211. Doerffel, K., 2. Chem., 1968, 8, 236. Riandel, J., and Linnig, F. J . , Anal. Chem., 1957, 29, 743. 186 ANALYTICAL QUALITY CONTROL 7. 8. 9. 10. 11. 12. ale, Cambridge, July 1979.

ISSN:0144-557X    

DOI:10.1039/AP9801700166

出版商:RSC    

年代:1980

数据来源: RSC

摘要:

186 ANALYTICAL QUALITY CONTROL Anal. Froc. EEC Legislation-Is There a Need for Quality Control? R. Sawyer Laboratory of the Government Chemist, Cornwall House, Stamford Stvect, London, SE1 9NQ The need for quality control of legislation is just as important as the need for quality control arising out of legislation. Both aspects are dealt with in this paper, together with some con- sideration of enforcement activities and quality control applied to methods proposed for intro- duction into legislation. Quality Control of Legislation Two major principles underlie the food legislation of all Member States of the Community, namely protection of the health of the consumer and the prevention of fraud. The latter principle aims to protect both purchaser and honest trader. Why is the community attempt- ing to harmonise legislation? If there is to be a Common Market, goods such as food must be able to move freely as within a single national market.The movement of goods manufactured or marketed in one country of the Community must not be prevented from sale in another by technicalities that govern labelling, composition or variably assessed safety factors. Individual countries have developed legal systems and ways of enforcement by virtue of differing economic and social systems, differing consumer tastes and dietary patterns. Be- cause at present there is no possibility of abolishing nine sets (and in the future twelve sets) of national law, the community has attempted to make arrangements for harmonisation of the provisions upon which law is based.The ways and means of harmonisation is by negotiation of Directives and Regulations made under the provisions of Articles 100 and 43 of the Treaty of Rome. A full description of the development of Directives and the consultation process leading to the development of UK Regulations under the Food and Drugs Act 1955 was given earlier.1 The important feature of the development of Directives is that they lay down objectives to be gained but that they leave the means of attaining the objectives and their enforcement to the provisions of law and the legal systems of the various countries. Directives, at first sight, may appear to be simple and straightforward, but detailed examina- tion frequently results in the exposure of ambiguities and uncertainties.It may be that the presence of ambiguities and uncertainties is essential to the unanimous acceptance by the nine Member States. The consultative process provides the ways and means of establishing a quality control procedure in the development of Directives and the resulting UK Regulations. The Directives and Regulations adopted and in preparation are summarised in Tables I and 11. It is important to bear in mind that it is not the intention of the Community or of indi- vidual Member States that the harmonisation of food law should lead to the disappearance of foods that are regarded as traditional to individual states, but that it should help food nianu- facturers to sell products in other. Member States. The view in the UK is that there is merit in the harmonisation of general principles, especially those of horizontal application, and that detailed compositional control of specific classes of foods should be adopted only as necessarv to protect the interests of consumers and good manufacturers.I t may be argued that some of the commodity Directives developed so far relate to less important products such as chocolate, fruit juice and honey. No real attempt has been made to develop harmonised law in respect of the staple items of diet. However, it is important to note that significant horizontal issues have been tackled and that some measure of agreement has been reached on important classes of additives and, more recently, on labelling, weights and measures and a general Directive on materials and articles in contact with food.May, 1980 ANALYTICAL QUALITY CONTROL TABLE I TITLES ADOPTED IN EEC LEGISLATION 187 Vertical Issues *Oils and fats: erucic acid *Fresh meat: trade in (health) ‘Poultry meat fresh: trade in (health) *tMilk liquid : marketing regulation *Sugars *?Eggs marketing standards *Milk : dehydrated and preserved *?Frozen poultry: water content *Coffee and coffee products *Cocoa and cocoa products *Honey *Fruit juices Preserved meat products Surface water quality toil seeds: sampling, etc.* UK Regulations implemented. t EEC Regulation. $ UK Regulations proposed. Horizontal Issues *Colours *Preservatives *An t i oxidants *Emulsifiers, stabilisers, etc. Pre-packed liquids : prescribed volume *Bottles used as measures *Pre-packed products : weight or volume $Materials and articles contact food Materials and articles : vinyl chloride Labelling, presenting and advertising Particular nutritional uses Enforcement Practices As indicated earlier, the structure and philosophy of Food law andenforcement practices varies considerably between Member States.Differences in practice impact on the operation of en- forcement systems, one of the most obvious being the concept of compliance with a regulation “on average” or by all units of a population. Another difference is the concept of using legally prescribed methods of sampling and analysis for enforcement purposes. Because of the structure of the enforcement systems in many European States and the use of semi-official or approved laboratories, the concept of prescribed methodology finds favour in a number of European Countries.So far, few Directives contain detailed provisions relating to sampling and analysis; a device that has been adopted has been to include an Article which provides for “elaboration of a Directive to cover sampling and analysis procedures to be used to establish compliance with the requirements of the Directive.” In the UK the methods of sampling and analysis are left to the competence of enforcement officers and the Public Analysts, respectively. Under the terms of the UK Acts, the sampling officer is required to divide the TABLE I1 TITLES PROPOSED IN EEC LEGISLATION Pro flosals Draft Proposals 1. Water quality 1. Materials and articles: 2. Pre-packed food : prescribed quantity (a) Glass 3. Product liability (b) Cellulose film 4.Fair competition 2. Extraction solvents 5 . Materials and articles: 3. Frozen foods (a) Plastics 4. Flavourings (b) Ceramics 5 . Acids, bases and salts 6. Meat extracts 6. Liquid milk (health hygiene) 7. Natural mineral waters 8. Jams, marmalades 8. Oils and fats 9. Casein, caseinates 9. Starches and modified starches 7.* Community trade mark lo.* Fresh meat and poultry meat: 10. Tomato products minced and chopped 11.* Flour and milled products 12. Mustard 13. Surface-treatment agents 14. Soft drinks 15. Canned fruits 16. Canned vegetables 17. Infant food 18. Instant mashed potato 19. Chips (frozen or chilled) 20. Bakery products * EEC Regulation.188 ANALYTICAL QUALITY CONTROL Anal. Proc. sample into three parts, one to be given to the vendor, one to be analysed by the enforcement authority and one to be held for possible future action on instruction of the Court.Much energy is now being directed to the elaboration of methods of analysis for inclusion into Directives. Quality control procedures are being adopted in the sense that attempts have been made to require that the methods proposed for inclusion in Directives are subjected to collaborative study before they are declared acceptable for adoption. The collaborative test philosophy has been accepted by the Commission as a result of pressure by the UK and some of the other Member States. Currently, the consultative machinery is also invoked in respect of the choice and approval of methods and collaborative studies have been organised on a National and a Community basis.The other substantial difference in philosophy of enforcement relates to the degree of emphasis placed on in-plant or retail inspection. Except for imported foods and for public health considerations, UK food law has been almost universally enforced at the point of sale. The Weights and Measures Act 1979 calls for a fundamental change of philosophy so far as the UK, Ireland and The Netherlands are concerned. Until the present time these countries have used a minimum weight system which is enforceable at the retail level; the Directive calls for the introduction of an average system and in-plant inspection. A summary of the different enforcement practices in seven of the other Member States is shown in Table 111. This indicates firstly the degree to which the veterinary profession is in TABLE I11 SUMMARY OF ENFORCEMENT SYSTEMS Country France ..Germany Italy n e Belgium. . h’etherlands Denmark Ireland . . .. .. .. .. .. .. .. Animal foods . .Veterinary inspection to retail level, central organisation (health), local and national laboratories .Veterinary control of abattoirs a t Lande . .Veterinary inspection at abattoirs. State vets (health) inspect butchers . .Veterinary inspection to retail level, municipality organised . . State veterinary service act as inspectors, central organisation operated a t district level . .Veterinary inspection at abattoirs, central organisation also covers meat shops. Separate inspectorates for dairy and fish produce .. Other foods Regionally organised inspection and laboratories.State control, 3-part sampling Police action at Lande. Sample formalities vary Carabinieri, surveillance use informers. Spot fines, little sampling District inspection, local laboratory (appointed), 3-part sampling, tribunal District inspectors, municipal laboratories, 2-part samples District inspection aided by city vets. Approved local laboratory service Local health inspectors, local health laboratories. Two regional analytical laboratories, 3-part sampling Metrology Centralised inspection service, all levels As other foods As for other foods Central service, economic affairs As for other foods Central service Police activity control of the meat processing industries in other Member States, and secondly the different functions afforded to Central Government and Local Government agencies.In the UK the division of responsibility between Central and Local Government and between Environ- mental Health Officers, Trading Standards Officers and Public Analysts are more distinct than in other Member states. It is clear that moves away from point of sale control towards in- factory control will have repercussions both on enforcement practices and on quality con- trollers within the food processers.May, 1980 ANALYTICAL QUALITY CONTROL 189 Consideration of Selected Directives The group of horizontal directives dealing with food additives include purity criteria for the additives, but so far only a limited number of methods for determination of the purity criteria have been developed. General limits have been included for cadmium, mercury, arsenic, lead and the sum of copper and zinc.A more general requirement has now been included within specifications that is worded “shall not contain a toxicologically dangerous amount of any ele- ment, in particular in heavy metals.” This statement is thus open to interpretation and obviously could lead to the introduction of a wide armoury of testing procedures involving identification, quantitation and speciation of a range of elements, which in the opinion of the interested party might be regarded as falling within the terms defined. Agreement on methods of analysis to be applied to the determination of additives in food has been achieved in a limited number of cases but much work is still to be carried out in this area if the objectives of the Commission are to be reached.A general Directive has been elaborated on Materials and Articles in Contact with Food. This is cast in such general terms that any enforcement action is based on matters of opinion. The directive covers “all materials and articles which in their finished state are intended to come into contact with foodstuffs or which are in contact with foodstuffs and are intended for that purpose’’ and, in addition, “materials and articles in contact with water intended for human consumption, but not fixed public or private water supply equipment.” The scope of the Directive is far-reaching as it embraces all aspects of manufacture and storage of foodstuffs. The key requirements are : “materials and articles must be manufactured in accordance with good manufacturing practice so that under normal or foreseeable conditions of use, they do not transfer their constituents to foodstuffs in quantities which could : (a) endanger human health; (b) bring about an unacceptable change in the composition of the foodstuffs or a Such basic philosophy is sound in theory but difficult to apply in practice, since the key words of the preamble “normal or forseeable conditions of use” cover a wide range of eventualities.Examples of subtle and not so subtle changes in organoleptic quality of foods may be demon- strated by storing canned foods of differing types in different temperature environments and in different canning systems. The General Directive on Materials and Articles in Contact with Food is one of a suite of Directives that are in the process of elaboration on this topic.A limit for vinyl chloride monomer in poly(viny1 chloride) and in foodstuffs in contact with the plastic has been laid down in an adopted directive. This sets out the limit for foods in terms of specified criteria for the method of analysis, viz., the detection is to be based on gas - liquid chromatography using the headspace method and the limit of detection is to be 0.01 mg kg-l. Further Directives have been proposed on plastics in general and on glass and ceramics. The provisions can include : deterioration in the organoleptic characteristics thereof.’’ 1. a list of authorised substances for manufacture; 2. purity standards for these; 3. special provisions relation to usage; 4.specific limits on migration of certain constituents or groups of constituents into foods ; 5. an over-all migration limit, into or on to foods; 6. provisions relating to health or hazards that may arise from oral contact; 7. any other appropriate rules; 8. rules to establish compliance. For example, in respect of the general range of plastics for use with foods, detailed discus- sions are taking place on migration limits, choice of appropriate test solvent in relation to the class of the food to be placed in contact and the methods of analysis to be employed to check migration limits. The results of a Community-wide collaborative study have been reported by RossL2190 ANALYTICAL QUALITY CONTROL Anal. Proc. Another horizontal directive that will have a wide ranging impact on the activities of quality control staff is that on Food Labelling, which has recently been adopted.I n the UK, the Food Standards Committee is currently reviewing the Labelling of Food Regulations. The general principles on which labelling law is based is that the consumer needs sufficient basic information about food being offered for sale in order to make an informed choice when purchasing. That this information is now more important than hitherto is recognised as a consequence of the growth of self-service shopping. Labels must therefore contain specific and basic information, presented in such a way that it can be readily understood by the purchaser. Apart from the weight of contents, the basic information consists essentially of an informa- tive, specific name, the name and address of the organisation responsible for the food and, for most pre-packed foods, a declaration of ingredients and additives.Labelling law also contains elements relating to claims on labels and in advertising, the general law being that in controlled areas claims may only be made subject to certain qualification and subject to a proper justifica- tion. The specific issues relate to energy value, protein claims, specific vitamins and minerals, slimming, diabetic claims and other medicinal value. A general Community directive has been adopted on foods for particular nutritional uses. Nutritional labelling will be considered by the Food Standards Committee as part of the general labelling review. As compositional standards for particular foods are introduced interactions with labelling law occur since each compositional regulation includes specified designations of the products controlled.The scope of the adopted directive is very close to the 1970 labelling regulations and Article 2 reads: “1. ( a ) be such as could mislead the purchaser to a material degree, particularly The labelling and methods used must not : (i) as to characteristics of the foodstuff and, in particular, as to its nature, identity, properties, composition, quantity, durability, origin or provenance, method of manu- facture or production; (ii) by attributing to the food effects or properties it does not possess; (iii) by suggesting that the food possesses special characteristics when all similar foodstuffs possess such characteristics.” Article 3 details the following compulsory requirements : “ 1 .the name under which the product is to be sold; 2. the list of ingredients; 3. in the case of prepackaged foods, the net quantity; -2. the date of minimum durability; 5. special storage conditions or conditions of use; G. name and address of manufacturer, packager or seller; 7 . place of origin or provenance; 8. instructions for use.” An issue that has been debated at length is the need for a date of minimum durability. The UK attitude to date marking has been that open date marking is of interest to the consumer for relatively short-life foods for which a “sell by” date is appropriate; for foods of longer shelf life the view has been that a date is not necessary other than for stock control purposes.This is normally related to foods with an estimated life in excess of 1 year. Voluntary action by food packers has been to use “sell-by” date marking. The directive demands “minimum dura- bility” dating; the exact meaning of this phrase is open to interpretation and is usually under- stood as “best consumed before.” However, the life of any commodity after it leaves the control of the distribution system will depend on the storage treatment given in the household. As indicated earlier, canned commodities can deteriorate drastically and variably according to the storage conditions, nature of product and the type of construction materials used. UK legislation has so far specific- ally excluded water froni the list of ingredients; the directive now requires water to be listed, the position to be determined by calculation : The other contentious issue is that of ingredient listing.May, 2980 ANALYTICAL QUALITY CONTROL 191 “added water and volatile products shall be listed in order of their weight in thejnished produce; the amount of water added as an ingredient shall be calculated by deducting from the total amount of finished product the total amount of the other ingredients used.This amount need not be taken into consideration if it does not exceed 5% by weight of the finished product .” The position of dehydrated and concentrated ingredients may be determined in relation to the amount present in a reconstituted form. Exercises in calculation of ingredients are awaited ; average allowable water contents for ingredients will have to be established.Fat content of products such as meat and fish has a marked effect on the moisture content. The ingredients listing provision also includes the problem of designation of additives and the question of emphasis to be given to presence or low content of one or more ingredients essential to the specific properties of the food. The question of water content of frozen poultry is the subject of a specific Regulation, which lays down test procedures. However, the question of water contents of other products, such as ham and cured meats, calls for more consideration. It should be noted also that Directives on water quality have been elaborated and these contain stringent test procedures for evalua- tion of quality.The interaction with food process industry cannot be avoided, as the proposed directive encompasses water “regardless of origin” and used in “preparation or preservation of foodstuffs.” Many canneries operate on private bore-hole supplies and in view of the pro- visions of the Directive the extensive test procedures and standards will apply. The interactions of quality control staff with all aspects of the requirements of the labelling and the other directives cited may not be obvious at first sight, but many quality control managers act as advisors on legislation, compositional law and serve to modify the excesses of the advertising men and not least to advise on durability, packaging and storage of food pro- ducts. Finally, the weights and measures Directive clearly states that the average concept is to be invoked.It may be assumed by parallel arguments that “average” compositional requirements may follow. The expression of results of analysis may then require some interpretation; the basic question is whether the average composition is based on actual content of a single package or on the de- clared average net content. The weights and measures Directive contains detailed sampling and tolerance provisions, which indicate that the above problems could be acute with certain styles of product. This resum6 has attempted to give some indication of the impact of the EEC on quality control. It is evident that legislators, administrators, quality controllers, enforcement authorities and analysts alike will be heavily engaged.In all events the cost will have to be calculated and there can be no doubt where the bill will be served. The intention so far as compositional requirements is not yet clear. References 1. 2. Sawyer, K., Pvoc. Anal. Div. Chew. Soc., 1976, 13, 238. Rossi, L., J . Assoc. 0-v. Anal. Chem., 1977, 60, 1282. Our Results are Right, Give or Take a Little G. A. Best Freshwater Survey Ojicer, Clyde River Puri$cation Board, Murray Road, East Kilbride, Glasgow In any scheme devised for the collection and collation of data from different laboratories, the accuracy of the results supplied is of fundamental importance. Those examining the data need to know that any differences between results from different areas are real and not caused by analytical errors. Apart from comparison of results between laboratories, analysts need to know that the results produced by their work are accurate. This is important in a river purification board laboratory not only for the comparison of the quality of rivers from different parts of the area but also for looking at long-term trends in water quality.It is also sometimes necessary to take legal action against polluters and analytical results may be used in evidence against transgressors.192 ANALYTICAL QUALITY CONTROL Anal. Proc. Until fairly recently, most water pollution control authority laboratories carried out analyti- cal work for “in-house” use only-to determine the quality of river waters, polluting load of discharges and chemical analysis associated with pollution incidents. Little was done to con- firm that the analytical method being used to determine, say, ammonia in a laboratory in East Kilbride produces the same accuracy and precision as that in another laboratory in Aberdeen. And yet analytical quality control should be a routine and integral part of any programme of analysis of samples.It has been suggested that it should be 10-20y0 of the analytical effort. In 1972 the chemists of the then nine river Boards in Scotland started a collaborative testing programme. Samples of sewage effluent and synthetic samples were analysed for a variety of determinands by the methods of analysis used routinely in each laboratory. The results shown in Table I are for one of the synthetic samples distributed and demonstrate the wide variation of values for some parameters but for others there was close agreement.The BOD results were surprisingly good, considering the nature of the test, whereas the synthetic detergent results were particularly bad. TABLE I INTER-LABORATORY TESTING OF A SYNTHETIC SAMPLE FOR VARIOUS DETERMINANDS Laboratory Actual I A 7 Parameter v a l u e 1 2 3 4 5 6 7 8 9 Suspendedsolids 62 52 49 54 50 49 - 52 48 48 BOD . . . . 38 37 36.5 39 35.5 35.0 35.1 37.4 37.6 36.1 NH,-N . . . . 16.5 11.7 14.5 17.5 13.2 15.5 15.0 16.8 13.6 15.7 NO,-N . . . . 13.4 11.2 7.8 17.6 5.0 11.7 12.4 13.6 - 13.3 Syntheticdetergent 0.45 0.1 0.41 0.29 <0.3 0.23 0.6 0.2 0.62 0.28 Coefficient of variation, 5 3.7 11.6 32 48 % The impetus towards a further inter-laboratory testing programme for the water industry came with the setting up of the Harmonised Monitoring Programme. The purpose of the programme is to assess the polluting load entering the marine environment from all inland sources in the UK, i.e., rivers and direct discharges.As part of the programme, the Clyde River Purification Board are supplying analytical results of monthly samples taken from the lowest reach of twelve of the major rivers in the Board’s area. The scheme requires that the results from all the participating laboratories-there are nine in Scotland-are comparable and of defined standards of accuracy. To satisfy this, a systematic approach to analytical quality control proposed by the Water Research Centre was accepted by the Department of Environ- ment. The flow chart for the approach towards achieving good inter-laboratory agreement is shown in Fig.1. In the national analytical quality control scheme, a large number of water authority labora- tories are taking part. In order to carry out the programme effectively, the participating laboratories have been divided into two groups. The first group of laboratories consist of laboratories from different parts of the UK ; the Forth River Purification Board represents the River Purification Boards of Scotland. These laboratories carry out the initial testing and method development. Once a method has been satisfactorily completed by the first group, it is then carried out by the remainder of the laboratories. In Scotland, the second group con- sists of all the laboratories of the River Purification Boards.Once a method has been completed within the prescribed standards of accuracy there then has to be a constant check that the accuracy is maintained. This is done by the use of control charts and of regular follow-up tests. This is usually 520% of the result but at small concentrations it is expressed in absolute terms as, e.g., milligrams per litre-the greater error is chosen. The error has been divided equally between inter- laboratory bias and random errors. Once the target of accuracy has been set, it is then necessary to find analytical methods which are free of bias and to check that satisfactory precision is achievable by those methods. The idea of all laboratories using the one clearly defined method was rejected because it was One of the first tasks was to set a maximum value for tolerable error.May, 1980 ANALYTICAL QUALITY CONTROL 193 unrealistic to have one set of samples being carried out by one method whilst the routine analysis is done by a method that may be subject to bias. It is also very difficult to define a method precisely because, try as one may, deviations will occur.There are tests for non- specific determinands such as suspended solids, BOD, etc., and for these determinands the prescribed method should be followed precisely. Set up working group V Decide objectives for quality control Choose analytical methods Ensure precise description of tests I ! V I 1 Obtain intra-laborat ory I precision V Compare standard solutions I I V I 1 I Set up control I charts I J V I I I Obtain in ter-laboratory bias Follow-up tests I Fig.1. Flow chart for achieving good inter-laboratory agreement. The following are probably the most common reasons for the bias of results: (i) sample instability; (ii) method of analysis does not determine all forms of the determinand; (iii) blank interference ; (iv) interferences ; and (v) non-linearity of calibration graph. These aspects have to be checked out on any method to be used for an inter-laboratory exer- cise before the precision testing proper starts. The greater the number of replicate determina- tions, the better is the accuracy of the information on the precision. However, river authorities routinely analyse over 20 determinands and if each method is to be rigorously tested, then the precision testing has to be carried out as economically as possible.The system that has been developed requires that on each of 10 days the following analyses are carried out in duplicate.194 ANALYTICAL QUALITY CONTROL 1. Blank. 2. Standard solution 1-near lower limit of range found. 3. Standard solution 2-near upper limit of range found. 4. Local river water. 5. Same river water spiked with a known amount of determinand. Anal. Pvoc. The results of the tests are analysed statistically to provide estimates of the standard devia- tion of individual results in each batch of analyses. Table I1 shows the results for the analysis of chloride for Scottish River Purification Board laboratories. Once the laboratories have demonstrated that they have achieved the necessary within-laboratory precision, it is then necessary to compare standards.This is to reveal any bias that may be present because of inaccuracy of the calibration standards. TABLE I1 PRECISION TESTING FOR CHLORIDE Target values = 5% relative S.D. or 0.25 mg l-l, whichever is greater. Spiked Standard solution Standard solution River water river water Concen- Relative Concen- Relative Concen- Relative Concen- Relative tration S.D., tration S.D., tration S.D., tration S.D., f - - - - - - * 7 f - - 7 7 ~ - 77-7 Yo % % % A . . . . . . 10 5.5 90 1.1 14.1 4.9 73.7 1.1 B . . . . . . 20 4.2 180 1.0 65.3 2.1 145.2 1.1 c . . .. . . 10 3.9 90 0.6 22.3 3.3 61.1 0.5 D . . . . . . 7 6.1* 63 0.6 10.0 4.1 60.4 0.5 E .. .. .. 5 5.0 45 1.9 6.1 4.1 36.0 2.4 F .. .. .. 11 2.7 110 0.4 13.4 4.1 59.9 0.6 G .. .. . . 20 3.6 180 0.9 41.8 2.3 124.3 1.9 Spiked recovery, / O 99.1 100.8 99.1 101.8 103.4 98.9 95.0 0’ * Significantly greater than 5%.Each laboratory analyses in one batch a suitable number of replicates of both the laboratories standard solution and a solution distributed by the co-ordinating laboratory-the standards should not differ by more than 5%. After establishing that each laboratory meets the required accuracy for its determinand and that standards are comparable, the next step is to check for inter-laboratory bias. This test provides direct evidence of the inter-comparability of results. The participating laboratories are provided with the following: a standard solution; a river water containing the determinand near the lower limit of the method; and a river water containing the determinand near the upper limit of the method.The samples are analysed once on each of five days and from the results is obtained the mean value with a 90% confidence limit. Laboratories that yield a bias greater than 10% of the true value are deemed to have failed the test and the reason for the bias is investigated. Once a laboratory has been “harmonised” for a particular determinand, the next one is studied. In addition, it is necessary to have regular checks to ensure that there is no appreci- able drift away from accuracy. This is achieved by (a) having control charts-a control standard solution for each determinand is analysed with every batch of sample whether these be daily or weekly. The result, whether it be in milligrams per litre or a colorimetric reading, is compared on the control chart with the results of many previous analyses with limitations drawn for warning and action limits.This gives a quick and frequent check for drift or for errors. ( b ) Follow-up tests consist of distributing a sample to each laboratory for analysis, which is then analysed on each of five days to check for bias. It is desirable to use a different natural sample for each follow-up so that in time each method will be checked for as many different types of real sample as possible. Once the whole scheme has been followed through for a particular determinand, a laboratory can say with confidence “our results are right, give or take a little.”M a y , 1980 ANALYTICAL QUALITY CONTROL 195 Analysis and Quality Control of Beers and Lagers During and After Processing P.M. Carpenter Quality Control Department, Whitbread West Pennines Limited, Samlesbury, Preston, Lancashire, P R 5 OXD Criteria by which beer is judged are : presentation; retention of head formation ; clarity, which obviously depends upon the type of beer, i.e., naturally matured in bottle or traditional draught will have a degree of turbidity, whereas other beers will be sparkling; and palate. The last criterion is associated with aroma and will be affected by the temperature at which beer is served. With the expansion of markets, specific brands are now available throughout the whole of the UK and even further afield. This has been brought about mainly by the supermarkets and the sale of canned beer for home consumption, whilst at the same time, local trade either in the form of naturally matured, conditioned draught or tank beer is available.This movement has resulted in the need for beer palate and clarity to remain far more stable for longer periods whilst at the same time the conditions to which certainly canned beer is subjected are far more demanding than hitherto. A customer with a favourite brand will expect the beer to be consistent no matter where it is purchased in this expanded market. The process of beer production can be divided into the following: 1 . The production of raw sugar/protein liquid called wort, which is produced from the malt with additional ingredients such as hops and sugar. 2. The fermentation of this solution to produce the rough beer and its maturation.3. Processing, including filtration and packaging. Quality control is divided into two sections : biological and chemical/physical. Biological Control These are initially selected from single-cell isolates and screened to give the desired flavour in the final beer and behaviour pattern during the fermentation. The yeast crop that is obtained from one fermentation is used in subsequent ones up to approximately 10 generations. Be- fore repitching, it is examined for non-culture yeast and bacterial contamination. The purity of this pitching yeast is probably one of the most critical points in the biological control in breweries, as sterility of plant will be wasted if the yeast used for pitching is itself contaminated. Infection by Saccharomyces types other than the selected strain can have far-reaching effects on palate, filterability or behaviour of fining in traditional draught beer.The main problem with biological examinations is the time that is required for growth of contaminants, especially those of an anaerobic/microaerophilic type, which will be the most harmful in breweries. A rapid 3-day forcing method using modification of de Man, Rogosa and Sharpe media with the pH adjusted to 5.3 is now used during the second day of fermenta- tion and competition by the primary yeast present is overcome by inhibition with cyclohexi- mide (,4ctidione). The groups of bacteria that have been found to cause problems in beer are Lnctobacillus, Zymomonas, Pediococcus, M~C~OCOCCZ~S, Hafwia and Acetobacter/Acetomozzas. Unfermented beer and its subsequent stages, including final package, are microbiologically examined by aerobic plates and anaerobic forcings, and the level of infection in the forcing is assessed by direct microscopical examination.The volume of beer examined on processed beer is much greater than at the earlier stages and the technique of membrane filtration is used at this stage. In addition to examining the beer, plant is also examined by means of swabbing where pos- sible. To ensure effectiveness of cleaning, strengths of detergent/sterilants are checked before and after use by chemical means. I t is also necessary to examine water, compressed air and carbon dioxide supplies which are used extensively throughout the brewery. Chemical/Physical Control moisture, extractable matter, colour and nitrogen.One of the most important factors is the yeasts that are used to ferment the wort. All raw materials are purchased to a certain specification. In malts, checks are made for In water, which is a very important raw196 ANALYTICAL QUALITY CONTROL Anal. Proc. material, the mineral salt content is tested for consistency, calcium being important in its removal of oxalate in the early stages of the brewing process, whilst the presence of metals such as copper and iron, if in excess of 0.2 p.p.m., can give rise to protein haze formation in the final package. A system has now been developed whereby pre-delivery samples of certain raw materials are examined by the larger laboratories within the company for acceptance or otherwise. This scheme is a far more positive form of quality control and, where possible, it is recommended.Oxygen plays a very important role in the brewing industry. Too much in the hot wort will result in variability in the final beer colour, whilst in the early stages of fermentation, levels of oxygen in the region of 8 p.p.m. are required for yeast growth. Current work in this field, however, is showing that there is a wide range of oxygen requirements by yeast used in the brewing industry. At the end of fermentation, the oxygen content will be virtually nil and all efforts are made from this stage on to prevent any increase that will be detrimental to the palate of the final package, especially in beer that is subjected to a long shelf-life, i.e., canned and bottled beers.All beers are brewed to a certain recipe, and at the end of fermentation of each batch, checks are made on the present gravity, colour, pH and bitterness to ensure that the correct procedures have been carried out. Further checks are also made at the end of the maturation period, especially with regard to oxygen and carbon dioxide content, again, the latter having a set specification for each brand of beer, depending on the final package. Container lagers normally have about 2.00 volumes, whilst container ales being served at slightly higher temperatures have approximately 1.50 volumes. Beers in bottle and can are in the region of 2.70 volumes. Cask-conditioned beers have an exceedingly low carbon dioxide content of approximately 1 .OO volumes.Possibly the most important quality control checks take place when the beer is filtered and placed into large tanks prior to packaging. Here, a complete check is made on every tank for original gravity, present gravity, colour, pH, oxygen, carbon dioxide, palate, head retention and turbidity, and unless all parameters measured are within specification, then packaging does not take place. Further checks are made on bottled and canned beers for the possibility of oxygen pick-up during final processing and all efforts are made to keep this to an absolute minimum. The shelf-life of all beers is measured by storage at approximately 60 O F , bottles and cans being examined on a weekly basis up to 12 weeks and container beers up to 25 days. Throughout the whole of the process, tasting is a very critical quality parameter, which relies entirely on the human palate.Here, training is essential for all operators who are carrying out this test as they need to know what the major off-flavours are likely to be. Quality monitoring is extended outside the brewery with strict control kept on storage periods in warehouses and routine examinations at trade outlets for the main criteria men- tioned at the start of this paper. These results are co-ordinated weekly and, if necessary, action is taken. On occasions throughout the year, samples are bought from as wide a range of selling points as possible and audits carried out. The frequency of sampling has been built up over the years from experience, the emphasis being placed on prevention rather than cure, and at points in the process that give least financial loss as well as preventing lower quality products from reaching the trade.Standard methods are used throughout the whole of the Whitbread organisation, most of which are based on the recommended methods of the Institute of Brewing and the European Brewing Convention. I t is essential that results emanating from the laboratories can be relied upon. A system has been developed whereby monthly samples of identical beers are distributed throughout all laboratories where specific parameters are checked and results compared statistically. Equally as important as the measurements themselves is the system of response to results outside specification. Reporting by exception is a way in which information is rapidly communicated to the Departmental Manager concerned.Finally, “Who does quality control, and where does it begin?’’ is a question that is often asked and the simple answer is that it lies in the hands of every individual employed within the brewery, but the final measure of how effective this has been will be judged by the consumer.May, 1980 ANALYTICAL QUALITY CONTROL 197 Systematic Approach to the Control of Accuracy and Precision in Routine Analysis G. Ghersini M. Ghiri and F. Salghetti Laboratori C I S E , Casella Postale 3986, 20100 Milan, Italy ENEL, Latina Nuclear Power Station, 04100 Latina, Italy Laboratori C I S E , Casella Postale 3986, 20100 Milan, Italy Routine analyses should always involve suitable control of the accuracy and precision of the results.Among the different procedures usually applied for this purpose, only a few possess the required reliability, and generally consist of systematic checks on reference samples or on duplicates. In some instances, however, such usual procedures are by no means easily applicable, either because of the limited availability of samples or because of the work load placed on personnel or instrumentation by the need for frequent duplications. In such instances, the usual procedures can be advantageously replaced by a different approach to the control of routine results, especially developed far the measurement of low- level radionuclides in environmental samples,l*2 but which can easily be applied to more tradi- tional analytical problems. The approach is based on the fact that, for most analytical pro- cedures, the analytical result, R, is related to the experimental (instrumental) measure, Af, through a general relation of the type R = k(M-B)/EP containing a conversion factor, k (to take into account physical constants, initial to final volume ratio, etc.) and a number of experimental parameters that can be generalised in a blank value B (measured under the same conditions as M ) , a calibration or efficiency term, E (peculiar to the instrument used for the final measure), and a chemical recovery, P , of the analyte throughout the analytical procedure.In many instances, the accuracy and precision of the final results depend almost exclusively on the variability of one or more of these parameters.Therefore, control of their variability can provide an indirect effective control of the reliability of the final result itself. A control procedure based on this approach involves the following steps : identification of the parameters that are most relevant to the accuracy and precision of the final result; pre- evaluation of their mean values and ranges; and a systematic check of their values during routine analyses. If the procedure also includes a periodic statistical analysis of all collected values for each parameter, then, in addition to greater reliability of the control process, better estimates of the relevant parameters will be obtained. The routine determination of 89Sr in fish edibles can be used as an example to illustrate the control procedure in more detail.The analytical method involves the dissolution of the sample, the separation and isolation of strontium by nitrate precipitation in an aqueous alcoholic medium, and low-background 19-counting of the solid nitrate source. The counting time is generally 20 h, so that not more than one analysis per day can be performed. Devia- tions of the results depend almost entirely on the variability of the chemical recovery of strontium and of the background of the /3-detector. Before routine analyses start, a set of eight values are collected for chemical recoveries (with s5Sr tracer) and background values, and their average values and standard deviations are estimated. During routine analyses, such values are controlled with control charts reporting the results of systematic check measurements of both chemical recovery and background, carried out on a random basis but with a fixed mean frequency (generally 10 days). Two control charts are maintained for each parameter, one for its actual value and the other for the variations between two consecutive measurements.In both charts, “warning” and “action”198 ANALYTICAL QUALITY CONTROL Anal. Proc. limits are reported at 5 and 1% confidence levels, rgspectively. Upper h i t s for variations are calculated directly from the average variation, R, through the expressions Action Warning a t 1% confidence Warning limit = Action limit = (1 + 0.756 tUeu5) and (1 + 0.756 t,,,) Control charts revised where to.o5 and are Student’s coefficients at 5 and 1% confidence levels, respectively; obviously the lower limits for variations are zero.Periodically (in general after eight new measurements), the average value and standard deviation of the new set of data are compared with the preceding values through suitable statistical tests. The hypothesis is tested of the identity between the average value f , and standard deviation sp of the new set, and those of the set of all cumulated preceding values (Zl, sl), by means of the usual t and F tests. A “runs test”3 follows on all collected nieasure- inents, discriminating between values higher and lower than f l , in order to control possible drifts. If all tests give positive results at the 5% confidence level, then the new set of values is cumulated with the preceding values, a new average and standard deviation are evaluated, and control charts are up-dated.The same is done when one or more tests are negative at the 5% but positive at the 1% confidence level; this case, however, is a “warning” condition, and the analytical procedure and the routine results will be looked at with particular care. One or more tests negative at the 1% confidence level imply “action” : the routine is stopped and, after checking and correction of the analytical method, a new control procedure is started. The flow diagram of the above revision procedure is illustrated as case A in Fig. 1. Case B refers to particular parameters, for which periodic variations are likely to occur : in the example of 89Sr determination, it concerns the background values, which can be affected by seasonal New control charts Case A F-test for s: ,.., s: ,.., s; Runs test 0 Fig.1. Flow diagram of the periodical revision of control charts.May, 1980 ANALYTICAL QUALITY CONTROL 199 variations of natural radioactivity in the environment. In these instances revision still in- volves the cumulation of all collected values but each set of (eight) measurements is kept distinct from the other throughout the routine control. Each revision therefore consists in checking the hypothesis of identity of a set of average values and of a set of standard devia- tions. For the sake of simplicity, it involves only a t test between minimum and maximum average values. Depending on whether the degrees of freedom of the standard deviations are the same or not, a Cochran or a Bartlett test is then performed. In the determination of y-emitting radionuclides in environmental samples by direct y-spectrome- try, the pre-treatment of the sample consists of a single drying step, which generally does not affect the accuracy and precision of the final result.Also in this instance, counting times are of the order of 15-20 h. The experimental result consists of a spectrum, in which the charac- teristic peaks of the different radionuclides are superimposed on a continuous background, mainly due to cosmic and other environmental radiation. The concentrations of radio- nuclides are derived from the areas of their peaks, corrected for the corresponding blank values, through a conversion factor that includes the efficiency of the y-detector, derived from a graph of efficiency nersus energy of the peak, experimentally determined.In this instance, deviations of the final results depend almost entirely on the variability of the blank value and of the efficiency, so that the control procedure is applied only to these two parameters. However, while the blank value is controlled for all radionuclides that present a significant peak in the blank spectrum, the efficiency graph is controlled only at two energies, in the lowest and in the highest energy ranges of the spectrum. An additional control of the instrument performance is also carried out, by applying the control procedure tothevalues of a large area of spectrum so as to detect possible variations of the continuum background, and to the width of a reference peak, to control possible variations of the resolution of the instrument.Two examples of control charts are shown in Fig. 2, and refer to blank values for direct y- spectrometry. With 40K, the values appear to remain self-consistent throughout the time period of control: a first revision, carried out in January 1979, gave a better estimate of the average value (1.414 0.030 c.p.m., compared with 1.406 & 0.051 c.p.m.), and appreciable narrowing of the control limits. With G°Co, after a warning situation detected by the control chart of the average value in October 1978, an action situation was pointed out in November by the chart of variations; after checking and correction of the analytical method, a new con- trol chart was prepared. A further example will illustrate possible variations and/or extensions of the procedure. 1.6 IZ 1.4 1.2 0.2 Q c.1 0 Fig. k 0 - I 0.2 0 U & - 0 0 . 0 n U M J J A S O N D J F M A M J 1978 1979 2. Examples of control charts. Blank values dire2 y-spectrometry. L I F, I I M J J A S 0 N O J F M A M J 1978 1979 for (a) 40K and (b) %o in fish edibles by As already indicated, the above control procedure was developed for environmental radio- activity measurements, but can find useful application in the control of precision and accuracy200 EQUIPMENT NEWS Anal. Proc. of results of routine analyses in many other fields, especially when the major sources of errors and deviations of the final results can be identified in one or more parameters (chemical recovery, blank value, calibration, etc.) used t o convert the instrumental reading into the analytical result. It can advantageously replace a control based on duplicates when the over- all analytical procedure is particularly lengthy and tedious, or when for some reason a frequent performance of duplicate analyses is difficult or uneconomic. Its main advantages lie in its simplicity and economy, as the control involves the systematic measurement of parameters that almost always should be measured or checked during routine operation, such as blank values or chemical recoveries. In comparison with duplicates, its disadvantage is the risk of having underestimated or overlooked sources of error that are not included in those put under control: however, a careful choice of the relevant parameters, made by an experienced analyst, or an occasional check of the over-all procedure with a standard sample should lower such risks to acceptable levels. References 1 . 2 . 3. Bazzano, E., and Ghersini, G., Energ. Nucl. (Milan), 1978, 25, 556. Ghersini, G., Ghiri, M., and Salghetti, F., CISE Report, RAMA-1, No. 6, 1978 (classified). Walsh, J. E., “Handbook of Non-parametric Statistics,” Van Nostrand, New York, 1965.

ISSN:0144-557X    

DOI:10.1039/AP9801700186

出版商:RSC    

年代:1980

数据来源: RSC

摘要:

200 EQUIPMENT NEWS Anal. Proc. Equipment News Gamma Counters The Philips PW 4800 series of gamma counters now incorporate microprocessors. Data are produced as counts per minute, yo CPM, and, for radioimmunoassay, yo bound. Samples are carried in trays, programmable by a hole- punched card. There are three sample size versions ; general purpose standard, 3 1 trays with ten samples; high capacity, with 31 trays of 20 samples; and an RIA dedicated version, with 40 trays of 12 standard RIA vials. The detector is of the through-hole type, which has lower volume dependence, and a choice of sizes is offered. A 2 x 2 in detector is used with low enerzv isotoDes onlv. such as RIA applications, whereas for applications where high energy isotopes are to be used a 3 x 3 in detector is supplied, together with a 200 kg low activity lead shielding.Insert El on the Reader Enquiry Service form for further information. Pye Unicam Ltd. Liquid Scintillation Counters Liquid scintillation counters, series PW 4700, with programmable tray sample handling, possess extensive measurement correction and data processing abilities. The extent of quenching for single labelled samples can be measured by using either the sample channels ratio (SCR), where two channels are used to compare different parts of the same spectrum, or by the external standard channels ratio (ESR), where a comparison is made between two different parts of the Compton spectrum produced by a gamma source (radium-226). The latter method is applicable to both single and dual label samples.Each tray of samples can carry a hole-punched programme card, which specifies one of 15 available programmes. The first seven programmes have pre-set windows matched to the most commonly used isotopes; five for single labelled samples, 3H, lac, 32P, 3H (strongly quenched) and 1251 (RIA), and two for dual labelled samples, 3H/14C and 1aC/32P. The eighth programme is dedicated to single photon counting and the remaining seven offer a free choice of window settings, five for single label and two for dual label samples. Insert E2 on the Reader Enauirv Service form for further information. Pye-Unicam Ltd.201 :\fa., 1980 CHEMICAL SOCIETY ANALYTICAL DIVISION I I Solvents for I R and UV Spectroscopy from Fluka I Reference Standards for Mass Spec. including PF K and triazines I I i Complete range of NMR sample tubes from Wilmad Glass Reference Standards A Complete for NMR including 0 1 , 4 Il)enlce lor NMR Accessories including Coaxial Cells and Microcell Assemblies from Wilmad Glass Deuterated Solvents including '100%' purity for NMR Quartz tubes ESR Spectroscopy Precision bore glass and quartz tubing from Wilmad Glass FOR FULL SET OF CATALOGUES: APPL Y TO: FLUOROCHEM LIMITED DINTING VALE TRADING ESTATE, DINTING LANE, GLOSSOP, DERBYSHIRE, SKI3 9NU.Tel: Glossop 4917 or 2855 (STD code 04574) Telex: 669960 A4 for further information. See page 21 0.202 CHEMICAL SOCIETY ANALYTICAL DIVISIOK Anal. Proc. DIONEX IC ANALYSIS:- SIMULTANEOUS ANALYSIS OFANIONS AND CATIONS Almost any ion with pK < 7 can be analyzed by ion chromatography (IC). Matric effects are minimized in IC by using sample dilution as the the only pretreatment. The extraordinary sensitivity of the IC has been applied to the analysis of inorganic anions such as F-, CI-, SO,*-, aliphatic carboxylic acids such as formate and acetate, phosphonic, and small aliphatic sulfonic acids. Monovalent cation or divalent cation analyses are easily performed with separate sample injection onto different col urn ns.The unique dual system IC 16 enables the analyst t o simultaneously determine a wide variety of ions with a single sample injection. The IC 16 contains two complete systems, each capable of operating independently or in a coupled mode. Use of different column sets in the two systems allows numerous possibilities for simultaneous ion analysis. Among these are: = anion and monovalent cation analysis aliphatic carboxylic acids and inorganic anion analysis aliphatic carboxylic acids and monovalent cations - monovalent and divalent cations coupled analysis for increased specificity in anion analysis There are over 20 Application Notes available, call or write for a complete list.DIONEX Corporation 1228 Titan Way, Sunnyvale CA 94086 U.S.A. In the U.K. Contact:- Roger George 4, The Buchan, Camberley, Surrey.GU 15 3x6. Tel: 0276 26373 Telex: 896691 D) Malic A) Strong Acids (CI-. NO;. C D E ) Lactic SO,2-, etc ) F I Succinc El Unknown or Formic ,-7---i 1 1 I --- 0 4 8 12 16 20 24 28 Mlnutes K' Application Note 21 A5 for further information. See page 210.May, 1980 EQUIPMENT NEWS 203 High Sensitivity Monitor The detection of low level iodine-125 and carbon- 14 contamination on laboratory benches, floors and technicians’ clothes and hands by a high sensitivity beta and gamma contamination monitor, the LB1210B, offers a detection limit of 10-5 pCi cm-2.With a 100-cm2 xenon-filled detector, the unit can be operated by battery or direct from a 220-240-V mains supply. A plug-in, external, 200 cm2, xenon-filled detector is available for the extended surveying of hands and clothes. Laboratory Impex Limited. Insert E3 on the Reader Enquiry Service form for further information. Recorder The 100 MSR recorder is designed in modular form, with easily interchangeable range selection, chart type and speed, single or multiple channels with ink or dry-writing options.The Record Electrical Co. Ltd. Insert E4 on the Reader Enquiry Service form for further information. Ultraviolet - Visible Light Source An ultraviolet - visible light source using pulsed light is announced. Oriel. Insert E5 on the Reader Enquiry Service form for further information. Photomacroscope The Techniscope is available in monocular or binocular form, and offers basic zoom magnifica- tion from 7x to 45x, extendable with an auxiliary lens to 3.5 x to 90 x . Fibre optics, 35 mm or Polaroid camera backs and other accessories are available. Finlay Microvision Co. Ltd. Insert E6 on the Reader Enquiry Service form for further information. X-ray Microanalyser The basic PGT System I11 analyser has a computer with a 32K memory, a single dual- density floppy disc capable of storing up to 100 spectra and a flicker-free colour display.Up to four spectra can be shown simultaneously with elements of interest defined on the spectrum by colour-coded windows and their peak intensities automaticaliy displayed. Inset E7 on the Reader Enquiry Service form for further information. PGT International. Microscope Multi-viewing Systems A series of adaptors and optical relay tubes are used in combination with either the 110 or 120 Microstar and clear crisp images (field of view) can be seen by up to nine viewers. Twenty three inches of space is provided between viewing stations. Multi-viewing systems can be fitted to the current 110/120 Microstar and Biostar ranges of microscopes and also to earlier Model 10 and 20 microscopes.Reichert- Jung UK. Insert E8 on the Reader Enquiry Service form for further information. Nitrogen Oxide Analyser The Meloy NA530R analyser is designed to meet the United States EPA “Equivalency” requirement for nitrogen oxide monitoring. The analyser has a minimum detection limit of 2p.p.b. The response time is typically less than 5 s, with a rise time to 95% of better than 30 s. Five scale ranges are selectable, from 0 to 5 p.p.m. to 0 to 0.1 p.p.m., with output to a panel meter or digital display, recorder and optional 4-20 mA line transmitter. The analyser has two independent simul- taneous photometric measurement systems, one system containing direct sample air and ozone and the other sample air in which all of the NO, has been converted to nitric oxide.The two signals are subtracted to give the nitrogen dioxide reading. Techmation Ltd. Insert E9 on the Reader Enquiry Service form for further information. Digital pH Meter The CD330 digital pH meter, reading to 0.01 pH unit or 0.1 mV, is mains operated, but with a built-in battery option for “in field” applica- tions. The ranges covered are 0-13.99 pH units, 0 to f 199.9 mV and 0 to j, 1.999 V. Walden Precision Apparatus Ltd. Insert El0 on the Reader Enquiry Service form for further information. Safety Vacuum Oven A vacuum oven suitable for accelerated drying, controlled atmosphere work, solvent removal,204 EQUIPMENT NEWS Anal. Proc. impregnation and embedding is announced. The 12.5mm thick toughened glass inner plate door is enclosed by a 4 mm thick trans- parent polycarbonate sheet, with temperature controlled by a direct-reading, hydraulic thermostat, scaled for 30-200 "C with sub- divisions of 10 "C.A rectangular shaped chamber has three shelf spaces 14 x 11 in, with a working volume of 31 1. The oven will operate at any pressure less than 1 000 mbar (atmospheric pressure). Insert E l l on the Reader Enquiry Service form for further information. A. Gallenkamp & Co. Ltd. Literature Literature describing the Carlo Erba Automatic Carbon-in-Water Analyser, TCM 400, which measures volatile organic carbon (TIC) and residual organic carbon (ROC) by a direct method to give more accurate results, is now available from Erba Science (UK) Ltd. Insert El2 on the Reader Enquiry Service form for further information. Technical Applications notes now available are : Packed Column Chromatography SN 2 Pyrolysis Gas Chromatography for the Determination of the Structure of Detergent Stocks SN 3 Pyrolysis Gas Chromatography for the Determination of the Structure of Rubber SN 4 Pyrolysis Gas Chromatography of Wood and Bark Capillary Column Chromatography SN 5 Tobacco Smoke Analysis SN 6 Fatty Acid Analysis SN 7 Essential Oil Analysis SN 8 Steroid Analysis SN 9 Elemental Organic Microanalysis SN 10 The Determination of Nitrogen in Elemental Analysis Yesterday and Today Tobacco using Automatic Analyser Environmental A nalysis SN 1 BOD, COD, TOC, Three parameters for the Determination of Organic Substances Content in Water and their Correlation iWicvostructure Analysis SN 11 Determination of the Microstructure of Porous Substances by means of Automatic Instrumentation Selective GC Analysis of Sulphztr Compounds SN 12 Flame Photometric Detection Standard Chromatograms TR 1 Steroid Profile-Clinical Research TR 2 Oils-Polycyclic Aromatic Hydro- TK 3 Oils-BP Exploration FV 2151 TR 4 Water Pollution Control (FV 2900) TR 5 a Micro Volume ECD with FV 2900 Centre carbons b Micro Volume ECD c Micro Volume ECD TR 6 Analysis of Rubber by Pyrolysis and TR 7 Analysis of Polyethylene R 001 Chromatography Discussion Group Bulletin April 1975. R 002 Study of Carbon, Hydrogen and Nitrogen Determination by Combustion - Gas Chromatography.E. Pella and B. Columbo. 1972 R 003 Micro and Trace Determination of Oxygen in Organic Compounds. Wolfgang J. Kirsten. 1978 R 004 Gas Chromatography of Nitrogen and Phosphorus Containing Compounds.G. R. Verga and F. Poy. R 005 On-column Injection on to Glass Capillary Columns. Grob R 006 Use of an Automatic CHN Analyser to Determine Organic and Inorganic Carbon in Soils. G. Viovannini, G. Poggio and P. Sequi. 1975 R 007 Gas Chromatographic Headspace Determination of Residual Acrylonitrile in Acrylonitrile - Butadiene - Styrene Resins and Migration into a Simulated Fatty Foodstuffs Liquid. G. Di Pasquale, G. Di Iorio, T. Capaccioli, P. Gagliardi and G. R. Verga. Erba Science (UK) Ltd. Insert E l 3 on the Reader Enquiry Service form for further information. Cryogenic Unit/FV 2900 Technical Reprints C. E. Roland-Jones 1975 A new cataloguc scientific glass apparatus and instrumentation has been produced by R.B. Radley & Co. Ltd. Insert E l 4 on the Reader Enquiry Service form for further information. Information Sheet No. 5, July 1979, describes an improved Struers Petrimat petri dish filler and stacker. Camlab Ltd. Insert E l 5 on the Reader Enquiry Service form for further information, The Series-S AGC laboratory process analyser is describcd in an illustrated leaflet b y Carle Instruments. A comprehensive catalogueMay, 1980 EQUIPMENT NEWS 205 describes the recently announced range of digital panel instrumentation by the Inter- national Microtronics Corporation. The use of a fluorimeter to measure the aromatic compo- nents of oil in the field is described in a mono- graph, which, with details of the Model-10 fluorimeter, can also be obtained from Techma- tion Ltd.Insert E l 6 on the Reader Enquiry Service form for further information. X 12-page book ST FV4200E gives a full description of the Fractovap 4200 Series dual- column gas chromatograph with various options and many accessories. Erba Science (UK) Ltd. Insert E l 7 on the Reader Enquiry Service form for further information. A new brochure gives details of the range of laboratory furnaces. Anderman & Co. Ltd. Insert E l 8 on the Reader Enquiry Service form for further information. Full details of the Metronic range of electronic weighers is contained in a colour leaflet available from Oertling Ltd. Insert E l 9 on the Reader Enquiry Service form for further information. Two leaflets published by Lauda detail their range of thermostats available in the UK.Roth Scientific Ltd. Insert E20 on the Reader Enquiry Service form for further information. A comprehensive laboratory products guide of recorder consumables has been published by Graphic Controls Ltd. Insert E21 on the Reader Enquiry Service form for further information. The complete range of electrophoresis systems centred on the GOO Chamber is fully described in Shandon’s latest chromatography - electro- phoresis catalogue. Shandon Southern Products. Insert E22 on the Reader Enquiry Service form for further information. Two leaflets, published in August and September 1979, feature faster HPLC and practical experience with the analysis of effluents using an AutoAnalyzer SMA 12/60 System, respectively. Techmation Instruments Co. Ltd. Insert E23 on the Reader Enquiry Service form for further information.A series of application studies are now available, with the following titles. “Low level nitrogen elemental analysis of crude oils, coals and fuel oil, lubricating oils, textile fabrics and sur- factants with the model 240B Elemental Analyser,” “Use of the System 4 microprocessor controller in thermogravimetric analysis of plastics, additives and fillers,” and “The ASTM method for screening potentially hazardous materials.” Perlzin-Elmer Ltd. Insert E24 on the Reader Enquiry Service form for further information. A monograph “Circulation, Dispersion and Plume Studies” describes fluorimetry techniques to determine and monitor the movement of water. Estuarine flushing for determination of the build-up and release pattern of pollutants, hydraulic model studies for inland and coastal waters to predict the effect of proposed drains and barriers and study of the dispersion of material from marine outfalls are among the topics discussed.Thermal plume studies, chlorine contact chamber and settling basin efficiency determination, tidal transport and underground water system studies ate also detailed. An eight page booklet on the Model-1 0 field fluorimeter is also available. Techmation Ltd. Insert E25 on the Reader Enquiry Service form for further information. A leaflet illustrating a simple method of checking detector tube hand pumps, which is just as accurate as the usual water displacement method of calibration, but is much more portable, can be obtained from Jencons (Scientific) Ltd.Insert E26 on the Reader Enquiry Service form for further information. The July, 1979, edition of “Kent Technical Review” features the P4000 integrated control system, a programmable gas sample selector and a range of intrinsically safe level controllers. George Kent Ltd. Insert E27 on the Reader Enquiry Service form for further information. Application note No. 252 on the Measurement of Oxidative Stability of Polyethylene by Differen- tial Thermal Analysis and Application Note No. 250 on Derivative Thermogravimetry are206 CHEMICAL SOCIETY : AKALYTICAL DIVISION Anal. Proc. AG for further information. See page 21 0.May, 1980 CONFERENCES AND MEETIYGS 207 availablc. Stanton Redcroft. Insert E28 on the lieadzr Enquiry Service form for further information. A procedure using liquid chromatography for testing the quality of pure water that is suitable for use in reverse-phase liquid chromatography is described in a leaflet available from J . T. Baker Chemicals B.V. Insert E29 on the Reader Enquiry Service form for further information. Model 1415 multi-purpose electrophoresis cell, together with the use of the reversible cooling stage, and of isoelectrofocusing techniques can be obtained from Bio-Rad Laboratories Ltd. Insert E30 on the Reader Enquiry Service form for further information. A booklet detailing the range of Wilks infrared spectroscopy accessories is available from Foxboro Analytical Ltd. Insert E31 on the Reader Enquiry Service form for further information. Literature describing applications of the

ISSN:0144-557X    

DOI:10.1039/AP9801700200

出版商:RSC    

年代:1980

数据来源: RSC

摘要:

May, 1980 CONFERENCES AND MEETIYGS 207 Conferences and Meetings Gordon Research Conferences Research Conference on Analytical Cliemistry,” August 11-15, 1980, New HaPnpton, New by F’ Hampshire, U S A Requests for additional information should be addressed to Di-. Alexander J‘I. Cruikshank. The analytical section of this year’s Conferences will be held a t New Hampton School. The papers presented on August 11 will be: “Trace Organic Analysis by Mass Spectrometry : New Instruments and New Techniques,” by R. G. Cooks ; “Chemical and Electrochemical Reac- tions and Surface Analysis a t Mcdified Elect- rodes,” by R. Murray; and “Californium-252 Plasma Desorption Mass Spectrometry of Large Molecules,” by R. D. Macfarlane. Those presented on August 12 will be: “Taking Ad- vantage of Noise in Analytical Measurements,” by G.M. Hieftje; “New Developments in Vibrational Spectroscopy,” by J. S. Grasselli ; and “High Molecular Weight Mass Spectro- metry,” by L. Friedmann. Those on August 13 will be : “Self-adaptive Experimentation as Applied to the Characterisation of Complex Systems,” by J. Frazer ; “Physico-chemical Measurements by High-performance Liquid Chromatography,” by C. Horvath; “The Determinations of Trace Amounts of Neuro- transmitters,” by R. M. Wightman; and “Spectrometric Applications of Multi-channel Image Detectors,” by Y. Talmi. Those on August 14 will be “Bioassay Chemistry, Ana- lytical Chemistry in Biological Testing,” by M. Guerin ; “Theoretical and Experimental Ap- proaches to High Resolution Chromatography, ” by S.P. Cram; “Computerised Gas Chromato- graphy - Mass Spectrometry and High Resolu- tion Two-dimensional Electrophoresis in Studies on Human Diseases,” by E. Jellum; and “Matrix-isolation Spectroscopy,” by E. Wehry. Those on August 15 will be “Chemistry in Molecular Thin-film Assemblies,” by L. R. Faulkner ; and “Discussion of 1981 Gordon Gordon Rcsearch Conferences, Pastore Chemical Laboratory, University of Rhode Island, Kingston, Rhode Island 02881, USA. International Environment and Safety Exhibition and Conference Srptewzber 1-4 1980, Wembley This Conference is being staged a t the Wenibley Conference Centre this year and will focus attention on the latest technical and scientific developments for the safeguarding of health and protection in our working and social environ- ment.Thesc compatible themes will attract an international audience consisting of environ- mentalists and safety specialists to the Exhibi- tion and Conference. The Exhibition will feature displ-ays of equip- ment and services fmm 3 major international companies, including .those involved in occupa- tional hygiene, health and safety, ,environmental ponitoring and pollution control. Companies ‘are thus provided with a prime and important market-building opportunity a t the exhibition to promote key products and instrumentation to major potential customers. The Conference will feature six parallel environment and safety seminars. Papers for the conference have already been received from the USA, USSR, Germany, Spain, Mexico, Poland, Hungary and the UK, and it is expected that there will be a heavy demand for places during the four-day conference sessions. For detailed information regarding participa- tion in the Exhibition and Conference contact Ernest H.Vincent, International Environment and Safety, Newgate, Sandpit Lane, St. Albans, Herts, AL4 OBS. Tel. St. Albans 51993.208 COURSES Anal. Proc. Expochem '$0 Octobev 6-9, 1980, Houston, Texas, USA This international exposition of analytical instrumentation in the industrial and bio- medical fields will be held a t the Astrohall in Houston, Texas, on October 6-9, 1980. The technical programme will consist of presenta- tions by the leading authorities in most areas of analytical chemistry from various parts of the world. Topics to be included are emission spectroscopy, automation, environmental analysis, new instrumentation, surface analysis, pharmaceutical analysis and data treatment.The chromatography section will be represented by the 15th International Symposium on Advances in Chromatography, which will be included as part of Expochem '80. Current developments in gas chromatography, high- performance liquid chromatography, high- performance thin-layer chromatography and gas chromatography - mass spectrometry will be presented. Most of the lecture presentations will be on an invited basis ; however, poster sessions will also be available. Prospective speakers must sub- mit 200-word abstracts by May 1, 1980. Inten- sive short courses on a variety of subjects will be given on the weekend prior to the exposition.Enquiries regarding exhibition space, the technical programme or short courses should be directed to Dr. Albert Zlatkis, Chemistry Department, University of Houston, Houston, Texas 77004. USA. First Asian and Pacific Chemistry Congress April 26-May 1, 1981, Singapore The Singapore National Institute of Chemistry, in conjunction with the Singapore National Academy of Science, is organising the Congress, which is to be held in the Shangri-la Hotel. It will be a major congress with international participation, including several Nobel Laureates in Chemistry, to discuss Chemistry and Health, Energy and Environment, as well as Analytical Chemistry and Chemical Education. The Con- gress is co-sponsored by the two Universities and various national organisations in Singapore, and also by several international bodies such as UNESCO, the Federation of Asian Chemical Societies (FACS) and the International Council of Scientific Unions Committee on Science and Technology in Developing Countries (COSTED) . For further information on the Congress contact the Congress Secretary, 1st ASPAC Congress, Singapore Professional Centre, 129B Block 23, Outram Park, Singapore 0316, Republic of Singapore.

ISSN:0144-557X    

DOI:10.1039/AP9801700207

出版商:RSC    

年代:1980

数据来源: RSC