30

Ratios of 0.10 to 10.0

20

10

0

QC

Logs of

Concentration Ratios

QA

would be inappropriate to establish acceptabil-

99

98

ity limits based on this data. In a recent article

dealing with methods for regulatory analytical

95

environmental chemistry, Kimbrough and Spin-

90

ner (1994) wrote, "If year after year certain meth-

ods produce wildly inaccurate results, then this

should be a signal that the method ought to be

70

evaluated and hopefully improved. *The accep-*

50

30

with these statements and we believe that it is

time to make concerted efforts to improve the

10

situation. If the Corps wishes to use temporary

5

limits while procedures are being brought un-

2

der better control, we suggest 0.254.00. Clearly

0.8

0.4

0

0.4

0.8

QC

this will result in many values being questioned,

Log of VOA in Soil

Concentration Ratios

QA

but wider limits will not encourage the needed

improvement.

Obviously, it is not a trivial task to signifi-

cantly improve the procedures used to estimate

VOC concentrations in soils. Before consider-

ing our suggestions for improvement, let's first

arise as a consequence of random rather than sys-

ask if there is more to learn from the archived

tematic errors. The standard deviation is larger

data. We failed to associate any specific laborato-

than expected for a normal distribution and the

ries with either good or bad results. In the ab-

tolerance intervals are excessively wide.

sence of standards and with numerous laborato-

What can we conclude from these computa-

ries participating (including some that serve either

tions? Even after excluding 24% of the ratios rep-

the QC or the QA function on different projects),

resenting the most divergent values, the agree-

ment between laboratories for the determination

it was impossible to establish a pattern. Clearly,

of VOCs on split soil samples is so poor that it

the more laboratories that submit results, the more

9