Measuring and maximizing precision in analyses based on use of calibration graphs.

Confidence-band statistics are used as a basis for measuring and improving precision in chemical analysis. Currently used calibration procedures often result in loss of precision and are therefore replaced by weighted least-squares and multiple-curve calibration techniques. Measures of precision such as correlation coefficient, relative standard deviation, and detection limit are replaced by more meaningful parameters based on confidence-band statistics. These new techniques are incorporated into a general scheme for developing methods for routine analysis.