Chain of custody record, Westchester County Department of Labs

NYC 9/11 Public Portal Document

long-term trend more than on specific test events. No one analyst should always be lower or higher than the group. If so, a review of procedures is to be performed. 10.3.3 Round Robin testing 10.3.3.1 A round robin test is the preparation of one analysis, yielding one resulting plate, which is read by all analysts at a location. 10.3.3.2Round robin testing is required monthly for coliform (total coliform sheen and non-sheen and fecal coliform blue and non-blue), HPC, fecal streptococcus and phytoplankton (using one microscope field). 10.3.3.3Record the date, analysts, sample used, analyst counts, and Supervisor comments. 10.3.3.4 The intent of this test is to provide training of the analysts. Plates selected are intended to be challenging. Therefore, the record of the results are not to be used to describe method precision.

11 Preparation and Usage of Quality Control Charts

11.1 Refer to ELAP Manual, item no. 261 for instructions on constructing a control chait and computing limits. 11.2 The DWQC Upstate Laboratories use raw value, percent recovery, or for the case of solids, regressed predicted raw values in the creation of control charts and control limits.

11.3 A new control chart with freshly computed control limits is to be started annually. 11.3.1 The last twenty quality control data points for the previous period should be used to compute the new control limits. 11.3.2 If twenty points are not available, data from a larger period of time may be used, assuming that there have been no method changes during that time. 11.3.3 If twenty points are still not available, default limits of 80-120% for references and matrix spikes and +/- 20% for duplicates may be used. Note that the reference method may state default limits that supercede these.

11.4 The Grtibbs outlier test (see J.K. Taylor's Quality Assurance of Chemical Measurements, p. 35), the Rule of Huge Enor, or another published outlier test, is used to verify that the QC data is free of outliers. See section 12.

11.5 All data is to be included in the calculation of new limits (including out of control data) unless the following applies. 11.5.1 There is a definite technical reason for the point to be removed or, 11.5.2 An outlier test finds that the point is acceptable to remove.

11.6 Control charts ai'e to be updated and reviewed before release of data.

11.7 Trending should be evaluated each time new data is entered onto the chart. The rule of sevens is to be used as a guide;

QUAL1020D02.doc Page 19 of 21

NYC-WTC_000072058