Fresenius J Anal Chem (1994) 349 : 761-767

Fresenius' Journal of

© Springer-Verlag 1994

Resolution of binary mixtures of cephalothin and clavulanic acid by using first derivative spectrophotometry J.A. Murillo, J.M. Lemus, L.F. Garcia Department of AnalyticalChemistryand Food Technology,Universityof Castilla-LaMancha,E-13071 CiudadReal, Spain Received: 6 October 1993

Abstract. First-derivative spectrophotometric method with a "zero-crossing" technique of measurement has been used for the quantification of two-components mixtures of cephalothin and clavulanic acid. As the absorption bands of these drugs overlap, both direct and derivative spectrophotometric methods have been investigated and evaluated by a rigorous statistical analysis of the experimental data. The first-derivative spectrophotometric method was found to be more accurate, direct and reproducible. Beer's law was valid over the concentration range 2.0 28.0 mg/1 for both compounds. The detection limits of cephalothin and clavulanic acid, at a 0.05 level of significance, were calculated to be 0.13 and 0.15 mg/1. The method was applied for determining these antibiotics in mixtures, some of them containing injectable dosage forms of cephalothin, and so to determine both compounds in saline and glucosed physiological sera.

Introduction Cephalothin is a semisynthetic [3-1actam antibiotic of the cephalosporin group I. These compounds were the first cephalosporins to be developed and were produced in response to a need for effective antibiotics against staphylococci. They have almost total stability to staphylococcal penicillinase and high specific activity against most Gram-positive bacteria. But, the J3-1actamases of Gram-negative bacteria include enzymes which could hydrolyse the ~-lactam ring of the early cephalosporins [1]. The J3-1actamase inhibitors have shown potential clinical usefulness based on their synergistic effects when they are combined with ~-lactamase-labile antibiotics [2-4]. The first inhibitor to be marketed for clinical use was clavulanic acid [5]. It was demonstrated that its

Correspondence to: J. A. Murillo

combined use with certain 13-1actamase-labile cephalosporins results in competitive synergism, which expands the spectrum of activity of these antibiotics [6-8].

H

H

H

$

O~N~

Cephalothin

~°0~'

O

C/~IOH

H CO~H

Clavulanic acid

In recent years, several studies have been realized about the potentiating effects of clavulanic acid on the activity of the cephalothin. The results of these studies indicate that clavulanic acid in combination with this antibiotic usually improves their activity against Mycobacterium fortuitum [9], Yersinia enterocolitica [10, 11] and Klebsiella pneumoniae [12]. Cephalothin and clavulanic acid show closely overlapping absorption spectra. Derivative spectrophotometry has proved to be useful in the multi-component resolution. Specifically, this technique has been employed for the simultaneous determination of substances of clinical interest, including cephalosporins, in mixtures, e.g., cefuroxime and cephalexin [13], cephapirin and cefuroxime [14], 7-aminocephalosporanic acid and 7aminodesacetoxycephalosporanic acid [15]. The aim of this work was to develop a simple, rapid and sensitive method for the simultaneous determination of cephalothin and clavulanic acid in mixtures by firstderivative spectrophotometry, using the "zero-crossing" technique of measurement [16].

Experimental

Apparatus All spectral measurements and treatment of data were carried out with a Beckman Instrument DU-70 Spectrophotometer connected to an IBM-PS computer with

762 Beckman Data Leader Software [17] and an Epson FX-850 printer. Among other possibilities, this system provides the capability of data manipulation for smoothing and for derivatives. Moreover, special Zoom In/Out and Trace functions allow to analyze data details and interpolation values.

Reagents Stock cephalothin solution, 0.100 g/L. Prepared by dissolving the standard sodium salt of cephalothin (Sigma Chemical Company Products) in water. Stock clavulanic acid solution, 0.100g/L. The solution was prepared by dissolving the standard lithium clavulanate (Beechman Pharmaceuticals, BRL 14151) in water. Buffer solution, 0.05 mol/L. Buffer solution of pH 6.8 was prepared by mixing sodium dihydrogen phosphate (0.1 tool/L) with sodium hydroxide (0.1 tool/L) of analytical reagent grade. Stock solutions should be stored below 5 °C in the dark. Solutions of the desired concentration were obtained by diluting the stock solutions to volume with water. All experiments were performed with analyticalreagent grade chemicals and Milli-Q water. Injectable dosage forms of Keflin Neutro (Lilly, S.A.), were labelled to contain 1 g of cephalothin sodium salt per vial. Saline physiological sera, 0.9% sodium chloride, were supplied from Apiroserum (Instituto de Biologia y Sueroterapia, S.A.)and Grifols (Laboratorios Grifols S.A.). Glucosed physiological serum, 5 % anhydro-glucose (dextrose), was supplied from Apiroserum (Instituto de Biologia y Sueroterapia, S.A.).

Results and discussion

The stability of the aqueous solutions of cephalothin and clavulanic acid was studied by recording their absorption spectra. At the beginning, these spectra were performed each half an hour and then they were performed every day. The cephalothin solution was stable for at least 20 days while the clavulanic acid solution was only stable for three days, both stored below 5 °C in the dark. The changes in the absorption spectra with the pH were studied between pH 2.0 and 12.5. The cephalothin and clavulanic acid spectra were unchanged in the pH ranges 2.0-9.5 and 4.0-9.0, respectively. A medium next to neutrality was considered as suitable and a pH 6.8 was chosen. The selected pH was provided by adding phosphate buffer solution of pH 6.8, where the solutions of both compounds showed a suitable stability. Figure 1 shows the absorption spectra of cephalothin (CL), clavulanic acid (CA) and a mixture of both compounds (M). As can be seen, cephalothin could be directly determined in the presence of clavulanic acid because their absorption spectra exhibit a spectral zone where the other component does not absorb. However, the total overlap of the spectral band of clavulanic acid with the absorption spectrum of cephalothin prevents the formation, from the total zero-order spectrum, of any spectral feature utilizable to determine clavulanic acid directly. The traditional Vierordt's method, which involves the use of two simultaneous equations, and the modified Vierordt's method [18] led to results of poor accuracy and reproducibility when the absorption spectra of the components were not sufficiently separated. We found that the problem of closely overlapping spectra can be solved by derivative spectrophotometry. This technique involves the differentiation of a normal spectrum with

1.5

Sample preparation and procedure Suitable volumes of the standard compound solutions, expected to contain between 0.050 and 0.700mg of cephalothin or clavulanic acid or their binary mixture, were transferred into 25.0 mL volumetric flasks. Then, 5.0 mL of pH 6.8 buffer solution were added to each of the volumetric flasks and diluted to the mark with water. Several mixtures of both compounds in that concentration range were prepared. The absorption spectra of the samples were recorded against a reagent blank (the same as the samples without the compounds to determine), using a 1.0 cm quartz cell and stored in the IBM-PS computer. Then, the first derivative was calculated. The absolute values of the first derivative were measured at 279.00 and 236.75 nm and by using an appropriate calibration graph, the cephalothin and clavulanic acid concentrations, respectively, could be determined. These calibrations were done by varying the concentration of one antibiotic without the presence of the other compound.

1.2 ©

M

~0.9 CA

©

~0.6

0.3

o.0

200

240 2;o Wavelength (nm)

aoo

Fig. 1. Absorption spectra of (CL) cephalothin (16rag/L), (CA) clavulanic acid (16.0mg/L) and (M) a mixture (16.0mg/L of each component); reference,reagentblank

763 condition, the instrument reads 4.0 data points per nm, that is equivalent to 0.25 nm between readings. The main instrumental parameter affecting the shape of the derivative spectra is the wavelength increment over which the derivatives are obtained (A)~). This parameter need to be optimized to give a well resolved large peak, i.e., to give good selectivity and higher sensitivity of the determination. Generally, the noise decreases with an increase of A)~, thus decreasing the fluctuation in a derivative spectrum. However, if the value of A~ is too large, the spectral intensity signal of the first derivative deteriorates. Various values of A)~ were tested and a A)~ = 8 was chosen as the optimum in order to give an adequate signal-to-noise ratio. It was not necessary to use any smoothing function because of the low noise levels shown by the original and derivative spectra.

0.04

0.03

CA

0.02

M

-4-3

~" 0.01

o.oo

-0.01

-0.02 200

2;o

220

aoo

Wavelength ( n m ) Fig. 2. First-derivative spectra of (CL) cephalothin (16.0 mg/L), (CA) clavulanic acid (16.0 mg/L) and (M) a mixture (16.0 mg/L of each component); reference, reagent blank

respect to the wavelength. Peak-to-peak and baseline measurements (generally referred to as graphical measurements) and zero-crossing measurements are the most common techniques used to prepare analytical working curves. The zero-crossing method was used in this work with satisfactory results. This involves the measurement of the absolute value of the total derivative spectrum at an abscissa value corresponding to the zero-crossing wavelengths of the derivative spectra of the individual components. Measurements made at the zero-crossing of the derivative spectrum of one of the two components would be a function only of the concentration of the other component. Figure 2 shows the first derivative spectra of cephalothin (CL), clavulanic acid (CA) and their mixture (M). The zero-crossing values of cephalothin appear at 214.50 and 236.75 nm, while the first derivative values of the clavulanic acid are zero at wavelengths above 260 nm. From among these wavelengths, 279.00 and 236.75 nm were selected as optima to determine cephalothin and clavulanic acid, respectively. At the other wavelengths, worse results were obtained: the calibration curves were poorly linear, the scattering of experimental points was somewhat unacceptable and the intercept on the y-axis was significantly different from zero.

Selection of optimum instrumental conditions The scan speed of the monochromator has virtually no effect on the derivative signal because the differentiation is obtained digitally. The scanning speed only determines the distance between each data point collected every 0.05. A scan speed of 300 nm/min was selected. In this

Calibration graphs and statistical analysis of experimental results The mutual independence of the analytical signals of cephalothin and clavulanic acid measured at 279.00 (h279.oo) and 236.75 nm (h236.75), respectively, was confirmed by carrying out the following experiments. Four calibration graphs were constructed from the first-derivative signals by measuring at 279.00 nm standard samples containing between 2.0 and 28.0 mg/L of cephalothin, in the absence of clavulanic acid (Po) and in the presence of 8.0 (Pl), 16.0 (P2) and 28.0 mg/L (P3) of clavulanic acid. Figure 3 shows these four series of first derivative spectra where the concentration of cephalothin is increased (from 2.0 to 28.0 rag/L) in all cases and the concentration of clavulanic acid is 8.0 (b), 16.0 (c) and 28.0 mg/L (d) and the first series (a) without the presence of clavulanic acid. The experiments showed that the height at 279.00nm (h279.oo) was proportional to the cephalothin concentration. Similarly, four calibration graphs were prepared from the first-derivative signals by measuring at 236.75nm standard samples containing between 2.0 and 28.0 mg/L of clavulanic acid, in the absence of cephalothin (%) and in the presence of 8.0 (ql), 16.0 (q2) and 28.0mg/L (q3) of cephalothin. Figure 4 shows these four series of first derivative spectra where the concentration of clavulanic acid is increased (from 2.0 to 28.0rag/L) in all cases and the concentration of cephalothin is 8.0 (b), 16.0 (c) and 28.0 mg/L (d) and the first series (a) does not contain cephalothin. Similarly, the experiments showed that the height at 236.75 nm (h236.75) was proportional to the clavulanic acid concentration. It can be verified in Fig. 3 that all curves containing the same concentration of clavulanic acid converge at an abscissa value corresponding to the zero-crossing wavelength of cephalothin (236.75 nm). Similarly, Fig. 4 shows that all curves containing the same concentration of cephalothin converge at the abscissa value selected to determine this antibiotic (279.00 nm), where the derivative signal of clavulanic acid is zero.

764 0.050

CL(rng/I)

('~)

0.040

cz(mg/O

o.o02o

8,0 16.0 24.0

0.030

(b)

0.0078 0.0152 0.0229

4o

'

'

',

~

,,

i

0.05

,

236.75 nrn

2 7 9 . 0 0 "rim

!

CL(rr~g/I) (e)

0.04

---r

i

~ff

-0.020 /

0.0114 0.0192 0.0268

~h

~

h2..... o.oo~0

12.0 20.0 28.0 4

oo,o

"~'

h,~.....

2.0

4.0 12.0 20.0 28.0

0.03

h ......

CL(rr~g/l)

0.0040 0.0115 0.0192 0.0271

;::o

h2re.oo

i:°°i °

,+J .~

0.02

'~

0.01

4

- - - r ii

.

5'? . ~ 0.00 236.75 n m

1 236.75 ~ m

279,00 "am

279.00 nm

-0.01

-0.02

210

230

250

2~0

290

Wavelength ( n m )

310

210

z~o

2go 2}o 2~o W a v e l e n g t h (rim)

Table 1 shows the results of the statistical analysis of the experimental data: the regression equations calculated from the calibration graphs, along with standard deviations of the slope and the intercept on the y-axis. The linearity of the calibration graphs and the conformity of the systems to Beer's law is proved by the high value of correlation coefficients of the regression equations. The significance of the intercept of the y-axis of all regression lines was evaluated by applying Student's "t" test at 95% confidence level and seven degrees of freedom [19]. If the intercept on the y-axis for the regression lines calculated by the least square method is negligible, it is necessary to again perform the fitting of the data according to a function whose intercept on the y-axis is zero and therefore the value of the slope (too) may be calculated. The results of this study for all the calibration graphs are reported in Table 2. It can be seen that the "t" value calculated does not exceed the theoretical value and hence the intercept on the y-axis is negligible in all of them. Consequently, the new values of the slope are calculated (Table 2). The precision of the proposed method was evaluated by applying the statistical technique of "Analysis of

Fig. 3a-d. Four series of first-derivative spectra of mixtures of cephalothin and clavulanic acid. The concentration of cephalothin is varied (from 2.0 to 28.0 mg/L) in all cases. The concentration of clavulanic acid is: a does not contain, b 8.0 mg/L, c 16.0mg/L, 31o d 28.0mg/L

Variance", which enables the total variation to be brokeninto their components: the variation between samples and the calibration graphs-sample interaction. The validity of the analysis of variance assumes that the residual error variance does not change from one sample to another or from one calibration graph to another. To carry out an analysis of variance, the variance ratio (Fexperiment,0 must be calculated and compared to the theoretical value of "F" [20] for adequate degrees of freedom at 95% confidence level. Table 2 shows the results obtained in this study. As can be seen, in both cephalothin calibration graphs and clavulanic acid calibration graphs the experimental value "F" is smaller than the theoretical value of "F" and thus at 95% confidence the source of variation is not significant. Therefore, it can be deduced that the amplitude of the derivative signal of the mixture, measured at the zero-crossing point of the derivative spectrum of one of the two components, is a function only of the concentration of the other component, in accordance with the theoretical predictions, i.e., the variation of both h279.oo and h236.75 was not affected by the presence of clavulanic acid and cephalothin, respectively, for any ratio of concentrations of the two components in the full range tested.

765 0.05 (a)~

CA(rag~t)

, .,...~

h ......

(b) ,

--5To-- -o-Eo-~;

0.04 0.03 -

-

rI

t2.0

0.0114

20,0 28,0

0.0190 0.0269

c2__(2£/'_)__bz'_.'e_ -

-

r

-

4.0 12.0 20.0

0.0190

2 8 . 0

0 . 0 2 6 6

0.0036 0.0115

-

I

[> 0.02

75

"~ O.Ol rn

.~ 0.00

i

" 236.75

~'r~

,279.00

236.75 r,zn,

~:m,

279.00 ~,rr~

-0.01 I 1 -0.02

-

r

J

T

~"

0.050

(c)

CA ( m g / g )

~o.o~0~ 0.040

hzs~.r5

2.0 8.0

0.0018 0.0074

I6.0

0.0149

24.0

(a)

h, .....

2.0 8.0 16.0 24.0

0.0019 0.0O74

0.0226

---i-

0.020

CA(rag/O

0.0150

O. 0 2 2 5

1-

'~ 0.010 ,7 .~

~h 4

0.000

ff 236.76 n m

279.00 n m

75

~ZT~

279.00

n~,n

-0,010

-0.020

-

-

-

210

2~o

2go

do

2~o

3,0 ~1o

~o

Wavelength (nm)

2~'o 2"}o 2~o Wavelength (nm)

Fig. 4a~t. Four series of first-derivative spectra of mixtures of cephalothin and clavulanic acid. The concentration of clavulanic acid is varied (from 2.0 to 28.0 rag/L) in all cases. The concentration of cephalothin is: a does not contain, b 8.0 mg/L, e 16.0 rag~L, ~1o d 28.0 mg/L

Table 1. Statistical analysis of calibration graphs for the determination of cephalothin (2.0-28.0 mg/L) and clavulanic acid (2.0-28.0 mg/L) by first-derivative spectrophotometry for n = 8 standard specimens Other antibiotic present Antibiotic determined

Antibiotic

Cephalothin (X = 279.00 nm)

--

Clavulanic acid Clavulanic acid (X = 236.75 nm)

Concentration (rag/L)

Calibration graph

Slope ( x 104)

Po

9.52

8.0 16.0 28.0

Pl Pz P3 qo

8.0 16.0 28.0

ql q2 q3

--

Cephalothin

Standard deviation

A similar study was applied to the absorbance values of the both compounds obtained by conventional spectrophotometry. A mathematical treatment of the absorbance data obtained at two wavelengths led to a calibration graph for" each one of the series mentioned before. Thus, the absorbance of cephalothin was directly measured in the zero-order spectrum at 261 nm, where the other

Intercept ( x 104)

Correlation coefficient

Slope ( × 105)

Intercept ( x 104)

1.33

0.999

0.39

0.65

9.55 9.58 9.58 9.51

0.53 0.50 0.93 0.60

0.999 0.999 0.999 0.999

0.23 0.57 0.26 0.64

0.38 0.95 0.44 1.08

9.54 9.59 9.47

-- 0.16 - 2.81 -- 0.41

0.999 0.999 0.999

0.82 1.21 0.81

1.37 2.03 1.35

component does not show absorbance. The other wavelength of measurement was 220 nm, where both cephalothin and clavulanic acid absorb. As Beer's law is obeyed for both compounds over the whole wavelength range used, an equation system was obtained that allowed to know the absorbance of clavulanic acid in the mixture at the selected wavelength. Thus, the four calibration

766 Table 2. Statistical parameters corresponding tO the estimate of the intercept on the y-axis,significanceby applying Student's test and evaluation of the residual error variance among the calibration graphs of cephalothin and clavulanic acid, respectively,by applying the Analysis of Variance. (95% confidence; 8 degrees of freedom) Statistical Parameters texperimental

ttheoretieaI m0 ( x 104)

Calibration graphs of cephalothin (h279.0)

Calibration graphs of clavulanic acid (h236.75)

Po

Pl

P2

P3

qo

ql

q2

qa

2.050 2.365 9.59

1.381 2.365 9.58

0.528 2.365 9.60

2.126 2.365 9.63

0.556 2.365 9.54

0.166 2.365 9.53

1.383 2.365 9.45

0.303 2.365 9.45

FexperimentaI FtheoreficaI

Table 3. Resolution of synthetic mixtures cephalothin and clavulanic acid by using the first-derivative spectrophotometry

1.91 3.07

2.66 3.07

[CL]

[CA]

[CL] (h279.oo)

Theoretic (mg/L)

Theoretic (mg/L)

Found (rag/L)

Recovery (%)

Found (mg/L)

Recovery (%)

2.0 4.0 8.0 12.0 16.0 16.0 24.0 28.0 28.0

28.0 16.0 24.0 8.0 16.0 20.0 28.0 2.0 12.0

1.96 4.06 7.95 11.84 15.83 15.95 24.13 28.22 28.12

98.1 101.5 99.4 98.6 98.9 99.7 100.5 100.8 100.4

28.12 15,92 24,12 8,14 15,71 19,62 27,70 1,94 11.93

100.4 99.5 100.5 101.7 98.2 98.1 98.9 96.8 99.4

graphs obtained for the absorbance measurements of cephalothin and clavulanic acid were subjected to the "Analysis of Variance". The experimental values of "F" were 4.90 and 40.59 for cephalothin and clavulanic acid calibrations, respectively, and thus in both cases higher than the theoretical value of "F" at 95% confidence level. In conclusion, direct absorbance measurements are not adequate to determine cephalothin and clavulanic acid simultaneously in their mixtures; however first derivative spectrophotometry permits that resolution. The detection limits were 0.13 and 0.15 mg/L for cephalothin and clavulanic acid, respectively, when defined as the analyte concentration leading to a signal three times the blank standard deviation [21]. The determination limits were 0.44 and 0.51 mg/L for cephalothin and clavulanic acid, respectively, when defined as the analyte concentration leading to a signal ten times the blank standard deviation [22].

Applications The validity of the proposed method was tested by successive determinations of cephalothin and clavulanic acid in synthetic mixtures. The results are shown in Table 3. As can be seen, satisfactory results were obtained for the percentage recovery of both compounds, indicating that the method is effective for the simultaneous determination of cephalothin and clavulanic acid by applying this technique. Cephalothin and clavulanic acid are not simultaneously commercialized. Because of this, the proposed

[CA] (h236.75)

method was applied to the determination of both compounds in mixtures of standard clavulanic acid and commercial injections containing cephalothin (Keflin Neutro, 1 g of cephalothin sodium per vial). Since cephalothin sodium is usually administered intravenously, the method was also applied to the determination of this cephalosporin and clavulanic acid in physiological serum. Working mixtures were prepared by dissolving the appropriate amounts of the standard clavulanic acid and cephalothin pharmaceutical vials in different media: water, saline physiological serum (Apiroserum and Grifols, 0.9% sodium chloride) and glucosed physiological serum (Apiroserum, 5% glucose). The mean recovery percentage expressed as percentage of the contents resulting from the average of three determinations of three different vials of cephalothin pharmaceutical products was: 98.3% for water, 98.9% for Apiroserum physiological serum, 99.0% for Grifols physiological serum and 98.7% for Apiroserum glucosed physiological serum. These results were in good agreement with the stated content of the cephalosporin. In the same way, the mean recovery percentage obtained for standard clavulanic acid was: 101.3% for water, 100.5% for Apiroserum physiological serum, 99.5% for Grifols physiological serum and 101.2% for Apiroserum glucosed physiological serum. Other components of the pharmaceutical preparation tested do not absorb in the wavelength range of interest. This makes it possible to apply the proposed method to the determination of cephalothin in the indicated pharmaceutical formulations and also to determine clavulanic acid when both are present in a mixture.

767

Conclusion Derivative spectrophotometry permits the simple, rapid, sensitive and direct determination of mixtures of drugs having closely overlapping spectra. In this work, the simultaneous determination of cephalothin and clavulanic acid in binary mixtures has been achieved by first-derivative spectrophotometry. The measurement wavelengths were 279.00 and 236.75 nm for cephalothin and clavulanic acid, respectively. The statistical analysis of the results indicated that the presence of one of the components does not interfere with the determination of the other. The method was applied to the determination of pharmaceutical dosages of cephalothin and clavulanic acid mixtures, in both aqueous medium and physiological serum (saline and glucosed). Satisfactory recovery was found in all cases.

Acknowledgement. The authors gratefully acknowledge the financial support from the "Direcci6n General de Investigaci6n Cientifica y T~cnica" (Project NO PB 88-0365),

References 1. Sabath LD, Jago M, Abraham EP (1965) Biochem J 96:739 2. Sabath LD, Abraham EP (1964) Nature 204:1066 3. Hamilton-Miller JMT, Smith JT (eds) (1979) Cole M in Beta lactamases, Academic Press, London 4. Hamilton-Miller JMT (1977) J Antimicrob Chemother 3:195

5. Cole M (1980) Philosophical transactions of the Royal Society of London B 289:207 6. O'Callaghan CH, Muggleton PW (1967) J Gen Microbiol 48:449 7. Miller JM, Baker CN, Thornsberry C (1978) Antimicrob Agents Chemother 14:794 8. Reading C, Cole M (1977) Antimicrob Agents Chemother 11:852 9. Cynamon MH, Palmer GS (1983) Antimicrob Agents Chemother 23:935 10. Jim~nez Valera M, Ruiz Bravo A, Ramos Cormenzana A (1986) Antimicrob Agents Chemother 29:104 11. Jimbnez Valera M, Ruiz Bravo A, Gdmez C, Ramos Cormenzara A (1988) J Antimicrob Chemother 22:707 12. Jackson RT, Harris LF, Alford RH (1978) Antimicrob Agents Chemother 14:118 13. Morelli B (1988) J Pharm Biomed Anal 6:199 14. Morelli B (1988) Analyst 113:1077 15. Murillo JA, Lemus JM, Garcia LF (1991) Anal Lett 24(4): 683 16. O'Haver TC, Green GL (1976) Anal Chem 48:312 17. Beckman Instruments (1987) Spectroscopy 2:16 18. Glenn AL (1960) J Pbarm Pharmacol 12:595 19. Lark PD, Craven BR, Bosworth RCL (1968) The relationship between two variables-simple linear relationship. In: The Handling of Chemical Data, Pergamon Press 20. Box GEP, Hunter WG, Hunter JS (1988) Estadistica para investigadores, ed Revert6, Barcelona, p 658 21. Nomenclature, symbols, units and their usage in spectrochemical analysis-II (1978) Spectrochim Acta Part B 33:242 22. Guidelines for data acquisition and data quality evaluation in environmental chemistry (1980) Anal Chem 52:2242