that a 3-minute extraction time was adequate for
from different depths rather than from discrete
the sandy soils present at the inland ranges. This
samples. Thus, we conducted more analyses than
study was conducted with the standard 20-g/100-
were envisioned in the original plan. Nonetheless,
mL soil-to-solvent ratio and should be tested with
depth profile data were acquired. The compositing
the 40-g/50-mL ratio if that is selected. Our expe-
procedure was validated using the data in Table
rience says that a 3-minute extraction time will be
9. A reference method of analysis (EPA Method
adequate.
8330) was compared with an on-site colorimetric
No currently available on-site method will pro-
method and RP-HPLC analysis of the extracts used
vide reliable estimates for RDX in soils from the
for the colorimetric analysis. The geometric lay-
inland ranges at Fort Ord (Crockett et al. 1996).
out of grids and the sample positions within each
Use of HPLC, the method of choice for laboratory
grid allowed us to evaluate spatial heterogeneity
analysis of explosives, in a mobile laboratory or
over various distances as intended in the draft pro-
in an expedient laboratory set up at Fort Ord,
tocol.
would provide estimates for RDX as well as HMX,
To determine an action level for the inland
TNT, 2-AmDNT, and 4-AmDNT. If HPLC is
ranges, it is necessary to specify the depth inter-
selected, either for on-site or off-site use, we rec-
val over which the action level applies. This is
ommend the LC-CN column for obtaining quan-
particularly important for impact ranges that have
titative results. This method can use the same
a significant concentration gradient with depth,
acetone extracts of the on-site methods, but they
such as we have demonstrated both at the inland
must be diluted 1:4 with water prior to injection.
ranges at Fort Ord and at CFB-Valcartier. Our sur-
Usually, an LC-18 column is used for quantitation
face samples were vertical composites from 015
and the LC-CN column is used for confirmation
cm, and this interval may be a reasonable choice
only in Method 8330, but for the five analytes
for an environmental risk-based criterion.
present at the inland ranges, the LC-CN provides
better resolution and more reliable quantitation.
LITERATURE CITED
around time of the on-site colorimetric method for
HMX, duplicate analyses should be conducted
Bauer, C.F., T.F. Jenkins, S.M. Koza, P.W .
using this method on each composite after they
Schumacher, P.H. Miyares, and M.E. Walsh (1989)
are formed for the first rings. Sampling and analy-
Development of an analytical method for the
sis would continue for increasingly distant rings
determination of explosives residues in soil: Part
until two successive rings yield results having all
III. Collaborative test results and final performance
four composite analyses below the action level. If
evaluation. USA Cold Regions Research and
desired, an upper 95 or 99% confidence limit could
Engineering Laboratory, Special Report 89-9.
be calculated for the mean of the four results.
Bauer, C.F., S.M. Koza, and T.F. Jenkins (1990)
When the upper confidence limit is still above the
Liquid chromatographic method for determina-
action level, a further ring could be sampled and
tion of explosives residues in soil: Collaborative
study. Journal of the Association of Official Analytical
six results. While the distribution of results from
Chemists, 73: 541552.
discrete sample analysis at the inland ranges tends
Crockett, A.B., H.D. Craig, T.F. Jenkins, and W.E.
to be log-normal, we do not expect that the distri-
Sisk (1996) Field sampling and selecting on-site
bution of composites from within a ring will devi-
analytical methods for explosives in soil. EPA Fed-
ate greatly from a normal distribution. Conse-
eral Facilities Forum Issue Report 540/R-97/501.
quently, we believe it would be appropriate to use
EPA (1994) Nitroaromatics and nitramines by
normal distribution statistics in this regard. If the
HPLC. Second Update SW846 Method 8330, Sep-
data indicate otherwise, log-normal distributions
tember, 1994.
could be used.
EPA (1996) Drinking water regulations and health
We tried to follow a draft protocol (App. A) for
advisories. U.S. Environmental Protection Agency
conducting preliminary site assessments wherein
Office of Water Report EPA 822-B-96-002.
we optimized the information available from rela-
Gilbert, R.O. (1987) Statistical Methods for Environ-
tively few measurements. This plan worked quite
well, although it was not possible to follow every
Nostrand Reinhold.
aspect of the protocol. For example, the depth
Gore, S.D., and G.P. Patil (1994) Identifying
study should be done with composite samples
extremely large values using composite sample
24