The calibration curve is the most common technique, but it's only useful when a large sample is available. As with other calibration methods, some information about the samples is necessary ahead of time to calibrate your instruments. For instance, if you have a large number of tools that each weigh one kilogram, you could use them to calibrate a scale by measuring the weight of each and then taking the average. Adjusting the scale so that the average equals one kilogram calibrates the scale. This method's accuracy depends on the sample, so naturally occurring phenomena are used when possible. When calibrating a frequency monitor, it's normal to use crystal oscillators because their frequency is well-known and very regular.
The standard addition calibration procedure works well with a small sample size or when you're not measuring a standard object repeatedly, as in the calibration curve method. If the sample were a number of different-sized weights, they could each be weighed separately and the weights plotted on a graph. This technique assumes a linear curve that should pass through the graph's origin. Adjusting the instrument so the graph passes through the origin calibrates it. Alternatively, if there's a small population of similar objects, a technician can weigh one alone, then two together and then three and so forth to create a similar graph for calibration.
If one instrument has been calibrated, you can calibrate other instruments with it. The process is straightforward--take a measurement with the calibrated instrument and then repeat with the uncalibrated instrument. Adjust the second instrument until its measurements agree with the first and then it's also calibrated. The accuracy of this method depends on the first instrument's accuracy. This method is common when only a single highly accurate instrument is available, and then cheaper instruments are recalibrated on a regular basis. Many people use this method when they set their wrist watches according to the atomic clock.