Method Development

To accurately determine the concentration of TBPs in a complex extract, it is important to have an idea of how concentrated the extract is. This gives an idea of what dilutions are needed in the sample set so that extrapolation of the results gives an accurate estimation of TBPs of the extract; the more concentrated an extract, the larger dilution factor is required to ensure the measurements are made in the linear range.

The extracts generally went through two concentration states; an initial dilute state where the TBP content was in the range of 1.5-2.0 mg/ml, and a concentrated state where the concentration was 8-10 mg/ml or higher. For the crude state the linear range was 0-10% dilution of the extract. For the concentrated state the linear range was 0- 3% dilution of the extract. A number of dilutions are made within this linear range, and then triplicate measurements of TBP are made for each dilution. As long as the results provided a linear relationship, then an extrapolation to the pure sample concentration was made, and the total biophenols determined.

For the determination of TBPs in a selected sample, the following methodology was used:

I. Depending on the concentration state of the extract, crude or concentrated, dilutions were made up covering the expected linear range of the Folin-method absorbance output, up to 3% and 10% respectively.

II. 0.1 ml of each dilution was taken in triplicate, and pipetted into a test tube.

III. 0.5 ml of Folin-Ciocalteu reagent (FCR) was pipetted into each test tube and vortexed. The tubes are then left for 5 min at room temperature, but no longer than 8 min, to allow the formation of colour.

IV. Once the colour has formed, 3.0 ml of 20% sodium carbonate was added to quench the reaction and colour of the FCR.

V. The solution was then made up to 10 ml with distilled water and centrifuged at 1500 rpm for 10 min.

VI. After 50 min, the absorbance of the solution at 725 nm was determined relative to a blank solution with no sample.

VII. The average of the absorbance of each triplicate set was determined and concentration of TBPs expressed in gallic acid equivalents (GAE) mg/ml. (Regression of the gallic acid standard curve between 0 and 0.25 mg/ml gave a linear gradient of 1.2703 abs units/mg.ml with an R2 of 0.9994.)

VIII. The results were then plotted against the dilution, and an analysis of the linearity of the results made.

IX. Regression analysis of the linear data points then allows the determination of the concentration of TBPs in the pure sample.

3.2.2.1 Repeatability and Reproducibility

Due to the requirement for extrapolation beyond the range of linearity, reproducibility for the analysis of TBPs using the Folin-method was of high importance. It must be shown that variation of experimental conditions was not magnified so much as to bring into question the validity of the analytical methodology. There are two areas of variability that needed to be addressed; the variation within a test, and the variation between tests.

Variation within the test, i.e. between test replicates, defines the variation of the triplicate samples and stems mainly from the inherent experimental uncertainty of the methodology. The variation between two identical tests analyses the repeatability of the tests and makes sure that the results achieved are consistent enough to be reliable indications of TBP content.

Table 3-3 shows the 95% confidence intervals, as a % variation from the mean, of the test triplicates, where 5 dilutions are analysed in triplicate for each run.

Table 3-3 95% confidence intervals (C.I.) for triplicates in TBP determination. % variation from mean.

Replicate Run 1 Run 2 Run 3 Run 4 Run 5 Run 6 C. I. (95%) 4.52 6.24 9.58 1.87 2.93 2.64

20-30 minutes 60 minutes

From these results it was quite clear that the variation of measurements was dependent on experimental methodology. The time dependence of the absorbance measurements indicates that it was very important, in terms of consistency of results, to maintain a defined experimental method to obtain experimental reproducibility. We can see that if the sample was not left for the full 60 min after centrifugation the variation between

the triplicates was significant. The longer the solution was left, the smaller the standard deviation, which was in agreement with Singleton (1999). If the TBP determination methodology was adhered to, the variation between replicates was smaller and the results substantially more reliable (runs 4-6).

With respect to the variation between two identical runs, the TBP content was determined through the normal methods, and the differences between the results analysed.

Table 3-4 95% confidence intervals (C.I.) for replicate determinations of TBP content.

Replicate Run 1 Run 2 Run 3 Run 4 Run 5 Run 6 Concentration (mg/ml) 9.53 9.81 10.83 9.85 10.16 10.49 C.I. (95%) 0.77 mg/ml 0.36 mg/ml 60 minutes 20-30 minutes 3.57% 7.61%

Analysis of the variability of the above replicates found that runs 1 through 3 had a mean value of 10.06 mg/ml ± 0.77 mg/ml, while runs 4 through 6 has a mean value of 10.17 mg/ml ± 0.36 mg/ml. Both sets of data gave acceptably similar results, but the first set (runs 1-3) showed an unacceptable degree of variability. It is desirable to have this technique give results within 5% of mean value at 95% confidence; the second set of results clearly show that this was possible. It also shows, once again, the importance of a consistent test methodology to this analytical technique.