The F value of 27.22 is significant at the 97.5% confidence level. The mean TRU concentration
in the plowed blocks was 62.8 pCi/g before plowing and 8.2 pCi/g after plowing, an 87%
reduction.
A comparison of the original with the repeat IMP readings on the unplowed blocks shows that
the treated blocks may legitimately be used as self-controls. The original TRU concentrations
averaged 77.2 pCi/g, and the repeat values averaged 75.2 pCi/g. This is well within the
measurement error of the IMP detector, and shows that the untreated concentrations did not

change between the measurements.

To test whether the pattern of contamination in the soil was altered by plowing, a multivariate
analysis of variance was performed on the soil profile data. The null hypothesis was that the
vector of mean concentrations by depth was not changed by plowing, and the alternative was
that the vector of means was significantly altered. The maximum likelihood estimator was

used, yielding a chi-square (seven degrees of freedom) statistic of 16.7. The null hypothesis can

be rejected at the 97.5% confidence level; i.e., plowing did significantly alter the vector of
mean concentrations.
The last part of the statistical analysis was an attempt to describe the after-plowing
distribution mathematically. If the plow mixed the soil, and hence the contamination, the
concentrations would be fairly uniform with depth. To check this, a linear regression of mean
4lam concentration as a funetion of depth was performed for each of the four plowed
blocks. The slopes of the lines were then tested for significant deviations from zero. The null
hypothesis was that the slope was zero;i.e., there was no gradient with depth. The results are:

Plot No.

Equation of Line

Test of Hg:B, = 0 vs, H;:B,40

1

Y = 3.6 - 0.06X

Accept

Hg

at 90%

2

Y = 0.7 + 0.007K

Accept

Hg

at 80%

3

¥ = 0.89 + 0.03X

Accept

Hg

at 80%

4

Y = 0.47 + 0.02X

Accept

Hg

at 80%

In all four cases, the slope

apparently did take place.

did not significantly differ from

zero, so that some mixing

However, in each block there were at least two subsurface observations of concentrations much

higher than the bulk of the depth samples. This indicates that some of the surface
contamination is deposited by the plow at depth without being mixed. Of the ten such "hot"
Spots, two were near the surface (0 to 10 cm), two were at 10 to 15 em, and the remainder were
30 cm or deeper. The TRU concentrations in the 10 spots ranged from 25% to 100% of the

original (before-plowing) TRU from IMP value, with a median of 35%. There was a weak trend

of less contamination (as percent of original) being deposited with increasing depth for the "hot"

spots. Other than these "hot" spots, observed TRU values rarely exceeded 6.6 pCi/g, regardless

of the original surface concentration.

Conclusions

The plowing experiment has clearly demonstrated that surface contamination can be reduced
substantially by plowing in Enewetak-type conditions. The multivariate analysis confirmed that the
distribution of contamination across the entire profile is altered significantly. Contamination is
mixed throughout the plowed profile; however, some proportion is deposited at depth with little
mixing. In mixed areas, the contamination is highly diluted, regardless of the original
concentration. "Hot" spots are inevitable and can be expected to result in concentrations of 25-50%
of the original surface levels. These "hot" spots were observed to occuratall depths sampled, but
most were observed at 30 cm or deeper.

B-9-20

Select target paragraph3