Calibration curve intervals, comparison

The random error may be expressed as standard deviation (s, SD). relative standard deviation (RSD, formerly coefficient of variation CV or VC) or confidence interval width. In our opinion using the confidence interval is the best procedure, especially when the earlier mentioned Mandel (1964) method is used, where the error in standards as well as samples are incorporated. For comparison purposes the 95% confidence interval width of an observation will be about 4 s. If one wishes to use separate standard errors for samples and standards, e.g. if the calibration curve has a much smaller error than the standards as sometimes in the case of aqueous standards, the total error can be calculated as follows ... 

The process of calibration consists of linking an instrumental response or readout to performance. To calibrate instruments and devices, a trustworthy, traceable standard, procedure, or comparison is mandatory. Forensic laboratories calibrate equipment such as balances and pipets, as well as instruments, via calibration curves. All calibrations are transitory and must be repeated at intervals defined by standard operating procedures (SOPs) or known performance limitations. For example, an analytical balance might be calibrated weekly via NIST-traceable weights, while a calibration curve from a gas chromatograph might require recalibration or a calibration check every 12 hours. Calibration and calibration checks ensure that the device s output is reliable. Typically, the more complex the device, the more often and more detailed the calibration checking must be. 

---