When you use a ruler to measure something how can you be certain that the value it indicates is correct? It may be possible that a defect occurred during manufacture when the scale was etched onto the ruler or that it has subsequently been damaged in a way that has distorted the scale? If this were the case then the scale on the ruler would not be as accurate as it should be (i.e. the measurement values obtained from it would not be as close to the true values as they should.)
The only way to have confidence that the indication from an instrument is correct is if it is regularly checked for accuracy. This is done by using the instrument to take measurements of known values. This allows us to compare the measurement value from the instrument with the actual true value. This process is called calibration.
Ideally we would check the instruments we use directly against the standards kept by the International Bureau of Weights and Measures at Sevres in France. In practice we can make this comparison in an indirect way. This is achieved by regularly comparing a measuring instrument against an instrument of higher quality which is itself regularly checked against a better instrument etc. This creates a chain of comparisons continuing on up to instruments which actually are regularly compared directly against the standards held at the International Bureau of Weights and Measures at Sevres in France.
Differences between measurement value and true value are called calibration errors (note calibration error does not mean a mistake made in the calibration process, it refers to the error between the measurement and true value that is revealed by the calibration process!). All instruments will have some calibration errors. Scientific instruments are manufactured to meet different specifications which will include specifying what the maximum error in the instrument readings should be.
When an instrument is calibrated it is not just checked at one value. Values near the two limits of the measurement range are checked as well as points in between. This is because calibration errors are not necessarily the same across the entire measurement range of the instrument. One some types of instruments there may be some kind of electrical or mechanical adjustment to adjust the instrument readings in order to minimise calibration errors, on other instruments (e.g. such as a ruler) there is no adjustment possible. The calibration errors (after adjustment if it is possible) are documented on a certificate which stays with the instrument so that the effects of the calibration errors can be taken into account when determining measurement values.
If the errors have changed since the last calibration we say that the calibration has drifted, if it has drifted so far that it exceeds the manufacturers' specifications we say that the calibration is out of specification or sometimes more simply that the calibration is out
In the period between calibrations it is possible that the calibration of an instrument may drift (e.g mechanical damage etc could occur). Care should be taken in the storage and handling of measuring instruments to minimise this risk. If the calibration is found to be out when it is next checked then measurements made since the last calibration may need to be verified. Instruments that have a good calibration history (i.e. very little calibration drift over several calibration periods) are more reliable instruments.