result if Pu concentrations in BM and sieved aliquots from the same
soil sample were equal over the wide range of concentrations encountered.
If the Pu in BM and sieved fractions were equal, then in the
linear model
log y = Bg + By log x ,
where y and x are the BM and sieved Pu concentrations,
respectively,
Bg = 0 and By = 1 so that log y = log x, or equivalently, y = x.
However,
By = 0,
region for Bo:
B, = l are not within the computed 95% confidence
By for the mean data
Draper and Smith,
1967,
page 64,
(labeled A and B)
in Table 4
for computational details).
(see
Hence,
the null hypothesis that By = 0 and 6, = 1 for these data was rejected.
Additional information relative to Question 1 is given by the data in
Table 5.
These are “39-240 Pu concentrations (nCi/g dry) in BM and
sieved soil fractions of surface (0-5 cm) samples analyzed by LFE and
LASL.
As was the case in Table 4,
the mean LFE and LASL results are
paired since for a given stake number the various aliquots (from BM
or sieved fractions) are from the same soil sample.
The logs of the
mean data (Columns A and B) are plotted in Fig.
The fitted
10.
line is parallel to the line expected when BM = sieved, i.e., when By
= 1,
but the sieved concentrations tend to be higher on the average
than do the BM concentrations.
however, the points By = 0,
region for Bo:
By.
Contrary to the case in Fig. 9,
8, = 1 are within the joint 95% confidence
Hence, for these data, the differences in eleva-
tion and slope between the two lines in Figure 10 are not statistically significant.
Looking at Figs. 9 and 10, one might ask why there is a parallel
situation between the fitted and BM = sieved lines in Fig.
not in Fig. 9.
10 but
One difference between these two data sets is that
the Pu concentrations in Fig. 9 are on profile samples, while those
in Fig.
10 are restricted to surface (0-5 cm) samples.
Hence,
the
data in Fig. 9 include much lower concentration levels than occur
in Fig.
10.
Could it be that a parallel situation exists only for
higher-level concentrations?
To investigate this possibility, only
those data pairs from Table 4 for which the LASL Pu concentrations
were approximately in the range of the LASL data used in Fig.
139
10