Obviously there is no fundamental basis for ei-
Ratios of archived results for VOCs in co-lo-
ther factor, but consider the dilemma when using
wider limits, such as 0.1010.0. Assume a regula-
to results for many of the same analytes in split
tory value of ten and one laboratory reports six
soil samples. This finding reinforces the conclu-
while the second reports 58. The analyses agree
sion that the major problems with VOCs in soils
within a factor of 10, but a valid regulatory deci-
occur during sampling and sample preparation
sion is impossible. Admittedly, the problem is
because the determinative step is basically the
minimized when results depart greatly from
same for each. For the 25 duplicate QC results
regulatory limits, but it is around these limits
reported, only one was outside of the limits 0.50
that disputes most often arise. Of course, this
2.00 while 5.6% of the QC/QA ratios were out-
problem is not eliminated with tighter limits,
side limits of 0.402.50. Recent results provided
but it certainly reduces the frequency of occur-
even lower percentages (3%). Only three samples
out of 153 (2.0%) had more than one VOC ratio
We reluctantly recommended temporary lim-
per sample outside of the 0.40 to 2.50 limits.
its of 0.254.00 for VOCs in soils because this cri-
We recommend that
terion may cause up to 40% of results to be unac-
a) Acceptance limits for duplicate QC ratios be
ceptable. However, we believe the results can be
set at 0.502.00, the same as for metals in soils.
dramatically improved by making fairly simple
b) Acceptance limits for QC/QA ratios be set
changes in the sampling procedures. For TPH,
at 0.402.50, the same as for metals in soils. The
where the same criterion was recommended, it is
recent data confirm the workability of those
clear that the NPD laboratories have this analysis
ranges for both QC1/QC2 and QC/QA.
under better control than the MRD laboratories.
Resolution of this difference should be the first
c) Like metals in soils, a policy must be pro-
objective, to be followed by further improvements
mulgated for dealing with samples with more
in the procedures.
than one outlier.
We believe that the U.S. Army Corps of Engi-
d) More attention needs to be given to consis-
neers should commit a significant portion of its
tency in the procedure for specifying concentra-
effort in chemical environmental measurements
tion reporting limits.
toward the realization of improved data quality.
Some specific suggestions are enumerated in this
report. Clearly, close interaction with participat-
ing laboratory personnel is also needed. To do
The objective of chemical environmental mea-
otherwise and maintain the status quo seems cer-
surements is to provide valid characterization of
tain to validate the Kimbrough and Spinner (1994)
sites under investigation so that regulatory deci-
quote, "The acceptance of very wide control lim-
its basically sends the message that if everybody
accurate analyses required to realize this objec-
is wrong, then everyone is right."
tive, the process of analysis must be considered
to start at the time of sampling. Procedures must
minimize volatilization and biodegradation losses
sample transport, and sample storage. Analytical
Grant, C.L., T.F. Jenkins, K.F. Myers, and E.F.
McCormick (1995) Comparison of fortification vs.
validation must deal with the entire procedure,
i.e., extraction, determination, and data manipu-
soils containing explosives residues. Journal of En-
lation. These requirements are doubly important
vironmental Toxology and Chemistry, 14: 18651874.
in programs where many different laboratories
Hewitt, A.D. (1994) Vapor-fortified QA/QC soil
contribute to the databank.
In our judgment, ratios of chemical analyses
subsamples for the analysis of volatile organic
on split samples that differ by more than a factor
compounds. American Environmental Laboratory,
of 4, reported by two laboratories, are useless for
Hewitt, A.D. and C.L. Grant (1995) Round-robin
decision making whenever concentrations are near
study of performance evaluation soils vapor-for-
regulatory limits. In reality, we dislike having lim-
tified with volatile organic compounds. Environ-
its greater than 2.5 and we believe that such agree-
ment is well within reach with proper sampling.
mental Science and Technology, 29: 769774.