Laserfiche WebLink
Tare -7-- <br /> VII. Fre uencv of Analysis of Method Blanks <br /> Method blanks are analyzed for each type of matrix, and the Method Blanks <br /> are analyzed with each batch, or every 10 samples, for each type of <br /> matrix, The Method Blanks, prepared from clean sample matrix using the <br /> same procedure used for the samples, allow laboratory contamination to be <br /> detected. <br /> VIII. Fre uencv of Analysis of Du licate Samples <br /> Duplicate samples are analyzed for each type of matrix, and the duplicate <br /> samples are analyzed with each batch, or every 10 samples, for each type <br /> of matrix. As these analyses are a measure of analytical precision, <br /> precision data can also be obtained by analyzing duplicate spike samples. <br /> This method for accessing precision is recommended for those samples <br /> which do not usually contain reportable levels of the specified analyte. <br /> When metal analysis of drinking water exca ;3s MCT, levels, the analysis <br /> should be repeated and/or checked by an alternative method. For organic <br /> analyses performed by GC, all positive results should be confirmed either <br /> by using a second, dissimilar column or by using GC/MS. <br /> IX. Frequency of Analysis of Spiked Samples <br /> Spiked samples are analyzed for each type of matrix, and the spiked <br /> samples are analyzed with each batch, or every 10 samples, for each type <br /> of matrix. As these analyses are a measure of laboratory accuracy, <br /> samples are spiked with all of the analytes to a concentration of five <br /> to ten times the reporting limit. For samples which usually contain a <br /> higher concentration of the analyte, the samples are spiked at a level <br /> similar to the concentration of the analyte. <br /> X. Establishment of Acceptance Limits for Precision and Accuracv <br /> Acceptance limits for precision and accuracv are based on analysis of <br /> duplicate samples analyses and of matrix spike samples. In general, the <br /> acceptable control limits are defined as the mean value plus or minus <br /> three standard deviations. The percent recovery for spiked samples is <br /> calculated as follows: <br /> 9 R 1002 (Theoretical Value -� Value Found) x 1002 <br /> (Theoretical Value) <br /> The percent variation in duplicate samples is calculated by <br /> 7V a ( Value(!) - Value(2) ) x 1001 <br /> (Value(!) 4- Value(2)) 2 <br /> and standard deviation by <br /> S (Sample Variance) <br /> `J f(n/n-1) x Variancel� <br /> RECEIVED MAR 1 3 1990 <br />