Analyst, January 1996, Vol. 121 (83-88) 83 Certified Reference Materials (CRMs 479 and 480) for the Quality Control of Nitrate Determination in Freshwater Ph. Quevauvillera, M. Valcarcelb, M. D. Luque de Castrob, J. Cosanob and R. Moselloc a European Commission, Standards, Measurements and Testing Programme (BCR), Rue de la Loi, 200, B-1049 Brussels, Belgium b Universidad de Cdrdoba, Depto. de Quimica Analitica, E-14004 Cdrdoba, Spain c Istituto Italian0 di Idrobiologia, Viale Vittorio Tonolli 50-52, I-28048 Pallanza, Italy Monitoring of freshwater and groundwater is routinely performed to control the level of contamination by a wide variety of trace and major constituents, including nitrates. In order to verify the quality of such determinations, the Standards, Measurements and Testing Programme (formerly BCR) has started a project, the first step of which was to evaluate the performance of European laboratories in an interlaboratory study and the second phase was a certification campaign of nitrate in two artificial freshwater samples.This paper presents an overview of the results of the interlaboratory evaluation and gives details of the preparation of the candidate certified reference materials (CRMs), the verification of their homogeneity and stability, the techniques used in the certification and the certified values. Two CRMs were produced: CRM 479 containing a low nitrate content (214 * 4 pmol kg-1 or 13.3 fr 0.3 mg kg-1) and CRM 480 containing a high nitrate content (885 & 13 pmol kg-1 or 54.9 fr 0.8 mg kg-1). Keywords: Interlaboratory study; certified reference material; quality control; nitrate; artificial freshwater Introduction A number of EC Directives related to water have been issued over the years, dealing with the quality of drinking water (75/440/CEE, 79/869/CEE, 80/778/CEE) or underground water (80/68/CEE) or with the protection of fish life (78/659/CEE).They prescribe the determination of a wide variety of elements, ions and compounds, including nitrate. The need to improve and to control the quality of water analysis has justified the organization of several interlaboratory studies within the Community Bureau of Reference (BCR) programme (now renamed Standards, Measurements and Testing Programme), which enabled the production of certified reference materials (CRMs) of, for example, freshwater (CRMs 398 and 399)' and artificial rainwater (CRMs 408 and 409).* Whereas the amounts of nitrate could be certified in the CRMs of rainwater, their certification in the freshwater CRMs was not possible owing to the stabilization procedure used (involving the addition of nitric acid).Consequently, it was decided to organize a separate interlaboratory study, the final aim of which was to produce freshwater CRMs certified for their contents of nitrate, one with a low nitrate content (CRM 479) and one with a high nitrate content (CRM 480). These CRMs correspond to values below and slightly above the maximum permissible nitrate content (approximately 50 mg 1-1 or 806 mol 1-1) mentioned in the drinking water Directive. The present certification follows a successful interlaboratory study, the aim of which was to improve the quality control of nitrate determination in freshwater.3 It was possible to use the group of laboratories participating in this study to certify CRMs 479 and 480 on a collaborative basis.Certification Procedure After a preparatory meeting at which all the requirements for certifying reference materials were discussed, the candidate CRMs were shipped to the participating laboratories (see acknowledgements). Each laboratory that took part in the certification exercise was requested to make a minimum of five independent replicate determinations on at least two different ampoules of each CRM on different days. The results were statistically evaluated, presented in the form of bar-graphs and discussed at a technical meeting with all the participants.In order to obtain the accuracy required for certification it is necessary to ensure that no substantial systematic error is left undetected. i. In a meeting of the laboratories participating in the certification, the sources of error and the measures taken to eliminate them were discussed. Errors related to a particular technique or analytical step could not be detected. The evidence given by the bar-graphs led to the conclusion that there was no substantial difference between the methods and that, therefore, the probability of an over-all systematic error would be low. ii. The laboratories participating in the certification exercise applied their methods correctly, i.e., the determinations were performed only when the method was under statistical con- trol.The simplest way of producing synthetic freshwater CRMs would have been to certify the nitrate contents on a gravimetric basis. It was, however, decided to proceed with a certification on a collaborative basis in order to obtain detailed information on the techniques routinely used for nitrate determinations, in particular on the within- and between-laboratory relative standard deviations [coefficients of variation (CV)]. This procedure logically followed the improvement scheme initiated by the first interlaboratory study (see below). Preliminary Investigations A feasibility study was carried out to investigate the optimum conditions for the preparation of the candidate CRMs of artificial freshwater to be certified for their nitrate contents.3 Solutions were prepared at three levels of concentration, viz., 0.5, 8 and 53 mg kg-1, and their stability was verified at 20 and 40 "C over a period of 6 months.84 Analyst, January 1996, Vol.121 ~~ Possible procedures for achieving a good stability of natural freshwater samples were discussed in a preparatory meeting with the participants in the project. Freezing was not considered to be a suitable procedure for the long-term storage of CRMs of natural water as it may lead to irreversible physico-chemical changes (e.g., formation of insoluble Ca salts) and would necessitate the use of special containers and ways of dispatch- ing. Freeze-drying would also be susceptible to physico- chemical difficulties.Consequently, in view of the risks of instability of natural waters and of the high variability in composition, it was decided to prepare artificial solutions containing known concentrations of nitrate and other major constituents, if necessary with addition of stabilizing agents. The pH of the solution was considered to be critical for the stability and subject to changes when the sample is in contact with CO2 (e.g., during bottling). The evaporation of C02 as well as the precipitation of CaC03 were considered to be possible sources of inhomogeneity and instability. Problems were considered to be less if samples could be equilibrated with air and if a carbonate buffer would be added in order to avoid major changes. On the basis of the chosen procedure, Na2C03 was added up to a concentration of 530 mg kg-l to maintain the pH against air; the resulting pH ranged between 9 and 9.5, which is higher than the pH observed in natural water.The pH was therefore lowered by adding acid (0.5 ml of HCl, 10 mol kg-1 in a litre of solution) in order to achieve a pH value of 6.8. Implicitly, Ca could be present to such an extent only that precipitation would not occur and all other ions could be added with the exception of, for example, phosphate and ammonium that might provoke microbiological growth. The addition of UV absorbing organic matter was also discussed to match the presence of humic acids; as it was found difficult to prepare water samples containing humic acids, the use of lauryl sulfate having a wide absorption range in the UV and known to be stable at concentrations in the range 1-3 mg 1-l was recommended; 2 mg I-' of lauryl sulfate was therefore added to the samples.Finally, phenylmercury acetate (C8H8Hg02) was added to avoid the development of moulds. Different concentrations were studied and an undesirable white precipitate was observed in the samples for concentrations of 100 mg 1-1 of phenylmercury acetate a well as deleterious effects on the redox copper-coated Cd column at the determina- tion step in the photometric method based on reduction to nitrite; this effect was not noticeable at the 10 mg 1-1 level which was therefore used for the sample preparation. The levels of nitrate concentrations were chosen to be in the range of levels found in the environment, considering the maximum value specified in the EC Directive (50 mg 1-1 or 806 moll-I).For the first interlaboratory study, the participants were instructed to prepare samples with concentrations of about 1, 10 and 50 mg kg-1 (respectively solutions A, B and C). The concentra- tions of the other compounds were CaC12.4Hz0 (9 1.7 mg 1-I), MgS04.7H20 (123 mg 1-1) and Na2C03 (530 mg 1-l) to simulate the hardness of a natural water. Nitrates were added in the form of high-purity salts (KNO3) dissolved in water obtained from a Millipore water-purification system (Milli-Q water). The water was boiled, homogenized by mechanical shaking using a magnetic stirrer and filtered with sterile filters to 0.2 pm. The different compounds were added by pumping. In order to verify the long-term stability of the samples on storage, two types of ampoules of 200 ml were selected, in white and brown glass, respectively.The short-term stability (3 months) of the three sets of samples was verified at 20 and 40 OC. The homogeneity was verified prior to the stability experiments by performing five measurements in 15 ampoules (for each of the three concentrations considered) randomly selected during the filling procedure. No instability was demonstrated over a period of 3 months for the three sets of solutions, both in white and brown glass ampoules. Conse- quently, the white glass ampoules were selected for the storage of the solutions and a temperature of 20 "C was adopted for the storage conditions. Interlaboratory Study The three above-described solutions were subsequently shipped to a group of about 30 laboratories for an interlaboratory study which enabled variations in standard deviations for ion chromatography due to, for example, the application of different columns, different eluents, and the use of chemical or electronic suppression (conductivity detection) to be explained.This intercomparison allowed a group of experts to be constituted and prepared them for the certification campaign. In addition, this study enabled the suitability of the procedure used for the preparation of candidate CRMs of verified homogeneity and stability to be confirmed. The results of this exercise are described in detail elsewhere.3 Preparation of the Candidate Reference Materials Two 150 1 PVC containers (one for each solution) to be used for the preparation and homogenization of the candidate reference materials were cleaned with a detergent, rinsed with distilled water and further rinsed with ultrapure water.For each of the reference materials, the ampoules were cleaned in a similar way. The ampoules were (air) dried for 2 d and conditioned for at least 24 h with the solution that they would contain. The preliminary investigations had shown that this procedure was adequate to bring the walls into adsorption equilibrium with the solution. The sample preparation followed the procedure used in the preliminary investigations. The two reference materials were prepared from ultrapure water to which freshly prepared solutions of the different substances described in Table 1 were added (amounts in grams added to approximately 150 1 of ultrapure water).All reagents were of pro analysi quality. The final pH of the solutions was about 6.8. The homogenization of the solutions was achieved by maintaining a constant agitation (with a mechanical shaker) during the addition of the solutions. The (conditioned) ampoules were filled with the CRM solutions and immediately heat-sealed. The ampoules were then stored at ambient temperature in the dark. Precautions were taken to avoid contamination during the ampouling procedure. Homogeneity Study One ampoule was selected out of each 50 ampoules prepared to verify the between-ampoule homogeneity of each CRM. Thus, a total of 25 ampoules was analysed per CRM, each ampoule being analysed in triplicate. The method repeatability was determined by ten replicate analyses of one sample of each candidate CRM.In order to establish the homogeneity and stability of candidate CRMs, a highly repeatable FI method was optimized.' Photometric detection was found to give the most repeatable results. The method was based on the reduction of nitrate to nitrite and the formation of a diazo compound which Table 1 Amounts of substances (g) added to 150 1 of water for the preparation of the two candidate CRMs KN03 3.670 and 12.475 Na2C03 79.500 CaC12.4H20 13.755 MgS04.7H20 18.450 Potassium phthalate 0.750 Lauryl sulfate 0.300 Phenylmercury acetate 1 S O 0Analyst, January 1996, Vol. 121 1.12 ..................................................................................... 85 0 0.08 z 0.80 is coupled to sulfanilamide (SPA) and N-( 1 -naphthyl)ethylene- diamine (NED) to yield an azo derivative with a maximum absorption at 540 nm.After mixing with the carrier (N&C1- Na2B407-EDTA), the sample reaches the redox Cd column (copper-coated with EDTA-CuS04) where nitrate is reduced to nitrite. Then, the reduced analyte solution is sequentially mixed with the SPA and NED solutions to yield the azo compound, which is monitored at 540 nm as it passes through the flow cell of the photometric detector. The measured signal is the height of the FI peak. The CVs obtained for nitrate are listed in Table 2. As shown, the overlap was within the uncertainty U,, of the CV (an estimation of the uncertainty U,, of the CV is calculated as follows: U,, =: CV/v2n, with n = number of replicates).Consequently, it was concluded that both batches of CRMs were homogeneous, within the uncertainty of the method. - I I I Stability Study The stability tests were performed by analysing randomly selected samples after 1, 3, 6 and 12 months of storage at 4, 20 and 40 "C. One replicate analysis of each of three ampoules stored at different temperatures was performed at each occasion of analysis. In this way, the long-term variability included both the analytical uncertainty and the between-bottle variability. The same analytical procedure as for the study of homogeneity was used. Any change in the content of an analyte with time indicates an instability provided that a long-term analytical reproducibility is obtained. Instability would be detected by comparing the contents of nitrate in samples stored at different temperatures with those stored at low temperature at the various occasions of analysis.The solutions stored at 4 "C were used as reference for the samples stored at 20 and at 40 "C, respectively. Fig?. 1 and 2 are plots of the ratios (R,) of the mean value (X,) of three measurements made after a period t at, respectively, 20 and 40 "C, and the mean value (X40c) from three determinations made at each occasion of analysis on samples stored at a temperature of 4 "C: R, = yt/%40c Table 2 Results of the homogeneity tests. CV method * CV between t CRM cv f u,, cv f u,, CRM 479 2.33 k 0.52 3.03 f 0.43 CRM 480 2.54 f 0.57 3.22 k 0.46 *Ten replicate determinations. t One determination in each of 25 different ampoules 120 l- v) a 3 cd 1.04 > m a .- N 0.96 cd rr 1.12 - - g 088 z z .................................................................................. 0.80 I I 1 I 1 I 0.00 2.60 5.20 7.80 10.40 13.00 Time (months) Fig. 1 lines: 1.1, 1.0 and 0.9. Stability study of CRM 479. Circles, 20; squares 40 "C. Dotted the uncertainty U, on the ratio R, was obtained from the CV of three measurements obtained at each temperature: U, = (CV? + CV40~2)''~ . R, In the case of ideal stability or absence of additional changes, R, should be 1. In practice, however, there are some random variations owing to the uncertainty on the measurement. On the basis of these results, it was concluded that no instability could be demonstrated and the materials were considered suitable for certification.The materials will be monitored further at regular intervals. Analytical Methods Used in the Certification Precautions taken by the laboratories participating in the certification are summarized in Table 3 which was used as a check list to avoid sources of error. A summary of the techniques used in the certification is given in Table 4. Technical Scrutiny All the results submitted for certification by the participating laboratories were discussed in a technical meeting to confirm the performance of the methods of analysis and their respective values. All data submitted for certification were obtained from laboratories that had demonstrated good analytical quality control and had fully implemented the analytical precautions outlined in the previous sections.The results accepted after the technical scrutiny were then statistically evaluated. Figs. 3 and 4 show the bar-graphs of the accepted results; each set of results is identified by the code number of the laboratory. Variations of 5% in the efficiency of the Cd column are not uncommon owing to the inhomogeneity of the column; higher variations were, however, found unacceptable for certification. Doubts were expressed on the use of hydrogencarbonate- carbonate buffer in comparison with borate buffer, the latter giving more reliable results. Discussions arose on the apparent discrepancy within the different SPEC (spectrophotometric determination of the diazo compound) results for the CRM 480. No explanation could, however, be found.Since no systematic error was suspected the data were accepted for certification. A further assessment of the sets of data was carried out using a Youden's plot4,5 which enables the results obtained on two solutions with different analyte concentrations to be compared, in order to detect possible systematic bias. Results are plotted in a scatter diagram, in comparison with the expected values or, alternatively, the means of the laboratory means. The diagram is divided into four quadrants in which two straight lines represent the expected values (or the means) for the two samples. If the results are affected by random errors only, they will be spread randomly over the four quadrants. If the results are located in I I86 Analyst, January 1996, Vol. 121 the lower left and upper right quadrants, forming a characteristic elliptical pattern along the 45" line passing through the expected values, one may conclude that systematic errors occurred in the measurements, underestimating or overestimating the concen- trations in both samples.The Youden graph for nitrate in CRMs 479 and 480 (Fig. 5), obtained using the mean values of each laboratory, shows clearly the prevalence of systematic errors since 14 out of 15 data are located in the upper right and lower left quadrants. However, as discussed below, these systematic errors are not statistically significant and no differences could be observed, on statistical grounds, between the different techniques used. With respect to the per cent. CV of the mean of Table 3 Some possible sources of error and measures taken to prevent them Analytical step Systematic error by Preparation Weighing Volumetric manipulation Sample preparation Adsorption/desorption Reagent contamination Contamination by toolslvials Contamination from laboratory air Adsorption/irreversible precipitation Incomplete conversion Calibration Contribution +/- +/- +I- + + + - - +/- Minimized by Calibrated balance Dilutions, etc., carried out with calibrated glassware, Use of distilled water washed non-metallic containers; Reagents of appropriate purity were chosen; verification Distilled water washing; verification with blank Use of clean benches or clean room; care in performing temperature control blank determinations used where appropriate with blank determinations determinations methods under cover or in closed systems; verification with blank determinations pH control Excess of reagents; methods a priori verified Reagents of suitable purity and stoichiometry were chosen Table 4 Summary of techniques and sample masses as applied in the determination of nitrate Sample mass = l o g 100 mg 200 mg 100 mg 50 mg 4 g 10 g 50 Pg 4 g 160 Pg 50 Pg 250 Pi3 Autosampler FI 25 Pg FI Sample pre-treatment and calibration Reduction on Cd column; addition of sulfanilamide and N-( 1- naphthy1)ethylenediamine dihydrochloride.Calibrant: KN03 (purity 399.99%) in water; calibration graph No pre-treatment. Calibrant: KN03 (purity > 99%); calibration graph Reduction on Cd column; addition of sulfanilamide and N-( 1 - naphthy1)ethylenediamine dihydrochloride.Calibrant: KN03 (purity > 99%); calibration graph No pre-treatment. Calibrant: KN03 (purity > 99%); calibration graph No pre-treatment. Calibrant: NaN03 (purity > 99.99%) verified Addition of NH4C1 buffer (pH 8.2); reduction on Cd column; against KN03 (purity 99.99%); calibration graph addition of sulfanilamide and N-( 1 -naphthyl)ethylenediamine. Calibrant: NaN03 (purity > 99.99%) verified against KN03 (purity 99.99%); calibration graph No pre-treatment. Calibrant: NaN03 (purity 299.5%) in water; calibration graph No pre-treatment. Calibrant: KN03 (purity 299.5%) in water; calibration graph Addition of NH4C1 buffer (pH 8.2); reduction on Cd column; addition of sulfanilamide and N-( 1 -naphthyl)ethylenediamine. Calibrant: KN03 (purity > 99%) in water; calibration graph Addition of NH4C1 buffer; reduction on Cd column; addition of sulfanilamide and N-( 1-naphthy1)ethylenediamine.Calibrant: NaN03 (purity 399.99%) in water; calibration graph No pre-treatment for CRM 479; dilution for CRM 480. Calibrant: KN03 (purity > 99%) in water; calibration graph No pre-treatment. Calibrant: NaN03 (purity 299%) in water; calibration graph Reduction by hydrazine addition with a copper catalyst; addition of sulfanilamide and N-( 1-naphthy1)ethylenediamine. Calibrant: KN03 (purity > 99.99%) in water; calibration graph Addition of N h C l buffer: reduction on Cd column; addition of sulfanilamide and N-( 1 -naphthyl)ethy lenediamine dihydrochloride. Calibrant: KN03 (purity 399.5%) in water; calibration graph No pre-treatment.Calibrant: NaN03 (purity > 99.5%) in water; standard additions Addition of NH4Cl buffer; reduction on Cd column: addition of sulfanilamide and N- 1 -( 1 -naphthyl)ethylenediamine dihydrochloride. Calibrant: KN03 (purity 399.5%) in water: calibration graph Final determination* SPEC of diazo compound at 540 nm IC; conductivity SPEC of diazo compound at 540 nm IC; conductivity IC; conductivity SPEC (SFA) of diazo compound at 540 nm IC; conductivity IC; conductivity SPEC (SFA) of diazo compound at 540 nm SPEC (SFA) of diazo compound at 540 nm IC; conductivity IC SPEC of diazo compound at 540 nm SPEC (SFA) of diazo compound at 540 nm IC SPEC (SFA) of diazo compound at 540 nm Series 01 02 03 04 05 05 06 07 08 09 10 11 12 13 14 15 * IC = Ion chromatography; SPEC = visible light or UV spectrometry; SFA = segmented flow analysis.Analyst, January 1996, Vol.121 87 0 N92b n N92a laboratory means, the values obtained for both CRMs were in good agreement with precisions usually obtained in a single laboratory for this range of concentrations. The CVs are plotted in Fig. 6 against the values obtained in the interlaboratory study 845- 01 SPEC 03 SPEC 05 SPEC ' I I 190 200 210 220 230 240 +......+......+......+......+.....+.......~.*....+.*....+......+......+ I +-----.-----> < ---- ---, 841 00 WEC 12 SPEC 13 SPEC 15 SPEC 02 l C 04 1c 05 I C 06 IC 07 I t 10 I C 11 IC 14 1c :MEANS I 1 <-*-, I ! Fig. 3 Laboratory means and 95% CI Nitrate in CRM 479 in pmol kg-1. 01 SPEC 03 SPEC 05 SPEC 09 SPEC 12 SPEC 13 SPEC 15 SPEC 02 IC 00 IC 05 I C 06 IC 07 I C 08 IC 10 I C 11 IC 14 IC ;MEANS: Fig.4 630 850 870 890 910 930 +.....+......+.....+.....+...... +......+......1.....+.....+......+.*...+ i g ....... ........ > I (-*-> I < -_-__-_- t ----.---, I I I <-*I--> ! g-------t------> I 1 <-**-, ! Laboratory means and 95% CI Nitrate in CRM 480 in pmol kg- 1. C R M Nit rate/pmol kg-' g8fi7 9351 Fig. 5 horizontal and vertical lines are the means of the laboratory means. Youden plot. Nitrate in CRM 479 versus nitrate in CRM 480. The (three samples, respectively, BCR-ma, BCR-FWb and BCR- FWc). An improvement in the precision was clearly achieved for concentrations of the order of 200 pmol kg-l (CRM 479 versus BCR-FWb), whereas the difference was lower for samples with concentrations of about 800 pmol kg-1 (CRM 480 and BCR-FWc) since a good precision had already been obtained in the interlaboratory study.Fig. 6 also compares the CVs obtained in other interlaboratory studies: the first group of results (indicated by N followed by the respective year) was obtained in the framework of an international collaborative programme for the assessment and monitoring of the acid- ification of rivers and lakes;6,7 the second group of results (referred to as R1 to R4) corresponds to interlaboratory studies organized by the Italian network for the study of atmospheric deposition chemistry from 1989 to 1992.8.9 The diagram shows a regular decrease in CV values from 2625% for concentra- tions of about 10 pmol kg-1 to 5-10% for concentrations of about 100 pmol kg-l.Statistical Evaluation The sets of accepted results were submitted to the following statistical tests : Kolmogorov-Smirnov-Lilliefors tests to assess the con- formity of the distributions of individual results and of laboratory means to normal distributions; Nalimov test to detect 'outlying' values in the population of individual results and in the population of laboratory means; Bartlett test to assess the over-all consistency of the variance values obtained in the participating laboratories; Cochran test to detect 'outlying' values in the laboratory variances (si*); One-way analysis of variance (F-test) to compare and estimate the between- and within-laboratory components of the over-all variance of all the individual results. The purpose of this statistical examination is essentially to ensure that the population of results accepted for certification has a normal distribution before the 95% confidence interval of the mean of means is calculated.This was true in all the cases (Kolmogorov-Smirnov-Lilliefors tests). In addition, no out- lying mean values were detected (Nalimov test). The set of variances was not homogeneous for both CRMs. As two different methods were used, each having a different repeatabil- ity and reproducibility, this is not surprising and fully acceptable (it is also for the reason that the s, calculated is not really applicable to any particular method; sw is a composite value for the methods used in this particular certification). Comparison of the Methods In the certification exercise laboratories applied two different techniques, namely spectrophotometric determination of the 20 - 8 > 0 v .10 - R3a P R4a R2b BCR-Was 10 100 NO;/pmol kg-' 1000 Fig. 6 frame of various interlaboratory studies on nitrate. Coefficients of variation (%) between laboratories, obtained in the88 Analyst, January 1996, Vol. 121 diazo compound (SPEC) and ion chromatography (IC), which made it possible to compare the results per technique. A grand mean of the means of all the eight laboratories applying either SPEC or IC was calculated. The obtained grand means were then compared in order to investigate whether a particular bias could be attributed to any method. As shown in Table 5, the CVs within SPEC or IC are systematically larger than those between the two techniques.Consequently, it cannot be inferred that the results of SPEC do not agree with those of IC for the certified content of nitrate. It could therefore be concluded that both SPEC and IC gave values that were good approximations of the true value for nitrate and could hence be applied in the certification. Certified Values The certified values as amounts of substance content and as mass fractions of nitrate are presented in Tables 6 and 7. The certified value is the unweighted mean of the accepted sets of Table 5 Results of the evaluation of consistency of the methods used for CRMs 479 and 480 cv (96) between means of Technique laboratories cv (%) of final with the between deter- same No. of sets means of CRM mination* technique of results techniques CRM479 SPEC 3.12 8 1.23 CRM480 SPEC 3.25 8 0.61 IC 3.67 8 IC 1.94 8 * SPEC = Visible light or UV spectrometry; IC = ion chromato- graphy * Table 6 Certified values of nitrate (as amounts of substance content) Certified CRM value Uncertainty Unit P* CRM 479 214 4 pmol kg-1 16 CRM480 885 13 pmol kg-I 16 * Number of data sets.Table 7 Certified values of nitrate (as mass fractions) Certified CRM value Uncertainty Unit P* CRM479 13.3 0.3 mg kg-* 16 CRM480 54.9 0.8 mgkg-* 16 * Number of data sets. results (p). The half-width of the 95% confidence interval of the mean is used as the estimate of the uncertainty. Availability The CRMs are available from the Institute for Reference Materials and Measurements, Retieseweg, B-2440 Geel, Bel- gium. Each bottle is accompanied by a certificate and a report describing the certification campaign.10 The authors thank all the laboratories that participated in the certification campaign: Preparation, homogeneity and stability studies. University of Cbrdoba, Department of Analytical Chemistry (Cbrdoba, Spain). Analyses. Anglian Water (Colchester, UK); Centre d’Estudits AvanCats de Blanes (Girona, Spain); C.N.R., Istituto Italian0 de Idrobiologia (Pallanza, Italy); C.N.R., Istituto di Ricerca sulle Acque (Brugherio, Italy); C.N.R.S., Service Central d’ Analyse (Vernaison, France); Compagnie Gknkrales des Eaux (Maisons- Lafitte, France); Dansk Teknologisk Institut (khus, Denmark); E.C.N., Energieonderzoek Centrum Nederland (Petten, The Netherlands); EPAL, Laboratorios Centrais (Lisboa, Portugal); GSF, Inst. fur Okologische Chemie (Neuherberg, Germany); Instituto Hidrogrhfico (Lisboa, Portugal); K.I.W.A. (Nieuwe- gein, The Netherlands); Ministhe des Affaires Economiques (Brussels, Belgium); Presidio Multizonale di Prevenzione (Venezia, Italy); University of Plymouth (Plymouth, UK). References I 2 3 4 5 9 10 Quevauviller, Ph., Vercoutere, K., and Griepink, B., Mikrochim. Acta, 1992, 108, 195. Reijinders, H. F. R., Quevauviller, Ph., van Renterghem, D., Griepink, B., and van der Jagt, H., Fresenius’ J . Anal. Chem., 1994, 348, 439. Quevauviller, Ph., van Renterghem, D., Valcfircel, M., Luque de Castro, M. D., Cosano, J., and Griepink, B., Anal. Chim. Acta, 1993, 283, 600. Youden, W. J., Ind. Qual. Control, 1959, 15, 24. Youden, W. J., and Steiner, E. H., in Statistical Manual of the Association of Oficial Analytical Chemists, Association of Official Analytical Chemists, Arlington, VA, 1975. Hovind, H., Norwegian Institute for Water Research (“A) Report, Oslo, 1988,40 pp. Hovind, H., Norwegian Institute for Water Research (“A) Report, Oslo, 1993, 50 pp. Mosello, R., Bianchi, M., Geiss, H., Marchetoo, A., Morselli, L., Muntau, H., Serrini, G., Serrini Lanza, G., and Tartari, G. A., Doc. Ist. Ital. Idrobiol, Annual Report, 1992, 35, 49 pp. Mosello, R., Bianchi, M., Geiss, H., Marchetto, A., Morselli, L., Muntau, H., Serrini, G., Semni Lanza, G., and Tartati, G. A., Doc. Ist. Ital. Idrobiol, Annual Report, 1993, 40, 49 pp. Quevauviller, Ph., and Valcfircel, M., EUR Report, European Commission, Brussels, 1995, EUR 16137 EN, p. 27. Paper 5103408F Received May 30,1995 Accepted September 14,1995