Results from the various laboratories are summarized in Table 19. All datasets except the one from Argentina are internally relatively consistent and also relatively close to the nominal values.

Figure 81 compares the country results for each of the samples and Fig. 82 compares the results for each country with the nominal values of the tracer concentration in radar plots with a logarithmic concentration axis. In Fig. 82, a horizontal bar has been inserted showing a symmetric realistic 2ct error of ±5% around the nominal value. This value is derived mainly from a realistic assessment of the uncertainty in the accuracy of the secondary standard.

ю

-p-

FIG. 81. Bar plot of HTO laboratory tests for different participating countries compared with the nominal value of each sample (yellow bar). Width of bar represents uncertainty in nominal value (cont. on p. 127).

FIG 82. Radar plots of results for each country with the nominal values of the tracer concentration.

A few difficulties prevail:

• The high values reported from Argentina may possibly be explained by contamination of the samples. When the samples were received in Argentina, they were erroneously stored in a room with about 80 Ci of HTO for tracer injection. Samples were stored for about 14 d before being analysed.

• For higher HTO concentrations, China and Vietnam are consistently reporting values that are too low. This problem may be due to erroneous calibration at higher concentrations. If that is the case, the problem may be rectified by making a new calibration with new certified standards.

• Indonesia, although having reported the lowest background counting rate of all the laboratories, seems to have a problem with the low content samples of brine and produced formation water.

The round robin test conducted has been a useful exercise. The main conclusion from the test is that all laboratories seem to handle this kind of analysis in a satisfactory manner.