One common mistake that scale users make it to confuse checking a scale with a precision weight to see if is within its accuracy limits with actually calibrating the scale.  Many times you will hear technicians say “Oh, I checked the scale and it was calibrated” in response to the question “Did you calibrate the scale?”  Using a precision weight to verify that the scale is within its accuracy limits is certainly better than doing nothing but it allows the scale to be more inaccurate than it would be if it were re-calibrated.Why do these modern marvels that we call digital scales require calibration?  Let’s look at the high accuracy scales where force restoration load cells are used to convert the weight into a proportional current that can then be converted to a digital equivalent and measured using computer techniques.  The load cell uses a very strong permanent magnet and an iron based magnetic keeper path to generate a high magnetic field across a small air gap.  A circular coil attached to the weight restoration mechanism ( a Roberval structure with many flexures used to eliminate any side loading effects) is used to generate the restorative force to bring the weigh pan back to a vertical center null position (as detected by an electro optical sensor) by passing an electrical current generated by a servo amplifier through the coil. 

This current is measured by an Analog to Digital converter and from that point on the computer has control.
Unfortunately nature does not always follow our nice simple linear equations and throws in a few complications.  This is especially true in magnetic circuits which contain iron.  They have problems associated with temperature, problems associated with magnetic flux density, problems with air gaps, and others.  The computer does its best to compensate for the linearity and drift problems.  The plot of force versus current shows that the motor gain (force generated per ampere of current) slumps as the current is increased.  The same plot run at different ambient temperatures show that motor gain is changed by temperature.  Things such as mechanical vibration can cause changes in motor gain.  Do anything to affect the level position of the scale and it causes an apparent change in motor gain.  The computer knows the temperature (usually measured in the magnetic core) and the current flowing in the force coil and tries to compensate for the non-linearity.  Calibration gives the computer a known anchor point, usually at full scale, which allows the computer to reset its various parameters based upon a certified mass.  

Calibrate often, or better yet, buy a scale with automatic internal calibration.  These scales have a known reference weight built in, and periodically, based upon time and temperature, perform an automatic calibration.