NOW YOU SEE IT	
David L. Russell, Senior Environmental Engineer
Specialty Chemicals Division
Allied Chemical Corporation
Marcus Hook, Pennsylvania  19061
.  James J. Tiede, Statistical Scientist
Buffalo Research Laboratory
Allied Chemical Corporation
Buffalo, New York 14240
INTRODUCTION
No matter how carefully we make a measurement, there will be some inaccuracy in
it. The magnitude of those inaccuracies takes on greater importance when the measurements are used to determine compliance with a regulation or a permit.
Frequently, the sampling method used to collect data may contribute additional uncertainty to the measurement.   Occasionally,  the combined uncertainty associated with
sampling technique, flow measurement method and the chemical  analyses employed may
be larger than the value determined.   In the latter case, one wonders what is being reported—is it an actual value or accumulated error?
The identification of the variability in some water pollution analyses may appear to
be akin to looking for the pea under the walnut shells at a carnival side show—sometimes you see it, and sometimes you don't.
When decreasing pollution control permit limits are anticipated, and when the magnitude of the measurement errors cannot be proportionally reduced, the need to examine
what is measured and reported becomes much greater.
PERMITS, MONITORING AND REPORTING
In the NPDES Permit System, the USEPA does not address the inherent inaccuracies
in data measurement.    During the development of the Effluent Guidelines, industrial
efforts were directed toward the statistical evaluation of plant performance data with
respect to these Guidelines; and little attention was given to the overall accuracy of the
measurements being made.
Guideline estimates of plant performance have become "absolute" rules in NPDES
permit negotiations, and this situation has caused hardship on many industries.    In one
specific case, a plant was required to control their net discharge of a particular pollutant
at one half part per million when the analytical accuracy of the chemical test method
was around one fourth ppm.    This example was one where guidelines were strictly
followed.
The situation just described might not have existed if the limits of accuracy on the
measurements made had been examined at the appropriate time.
To comply with the existing and proposed effluent permit limitations, we need to
(a) determine and state the limits on our ability to measure effluent data, (b) use that
information as a basis for action in setting internal performance standards for our control efforts, and (c) attempt to gain recognition of the magnitude of measurement
uncertainties in current and future permit negotiations.
211