194 TRANSURANIC ELEMENTS IN THE ENVIRONMENT 



150 — 



o 

 a. 



< 



z 

 o 

 o 



100 — 



50 



Y/X = 50.1-6.2 X 



• * * 





Y/X = 74.7-52.8 X 



J_ 



0.2 0.4 0.6 0.8 



SAMPLE WEIGHT, g 



1.0 



1.2 



Fig. 4 Relationship between ' ^'' ' ^ ''Cs concentration and sample weight in Hypericum 

 walteri from a floodplain receiving reactor effluents. 



In this case we do not have the luxury of taking a larger size sample to reduce 

 variability as we do for soil samples. Assuming that the concentration is constant, what is 

 the best way to estimate it? The problem can be viewed as twofold: to determine (I) the 

 minimum amount of sample required to yield a consistent estimate of the true 

 concentration and (2) the method for combining the data to give a best estimate of the 

 concentration. 



A method of estimating the minimum sample weight required for an analysis 

 consistent with the homogeneous dispersal assumption is illustrated in Figs. 5 and 6. The 

 graph in Fig. 5 was obtained by arranging the samples in random order and computing the 

 variance for an increasing number of samples starting at 10. The initially higlily 

 fluctuating variance that then decreased to a final value of about 1950 is typical of the 

 plots obtained when this procedure is applied to positively skewed and leptokurtotic 

 (sharp peak) data. A similar procedure was used to generate the graph in Fig. 6 except 

 that the samples were arranged in order of decreasing weiglit. The form of the initial 

 phase of Fig. 6 (sharp increase followed by a steady decrease) is similar to that of Fig. 5, 

 and the variance appears to stabiUze at approximately 750 for samples weigliing >0.2 g; 

 however, when samples weigliing <0.2 g are added, the variance then increases 

 continually until the same tinal value as that in Fig. 5 is obtained. This suggests that 



