We define 'comparative references' as anything that you look at (compare to) to decide if a part or product is good. Some examples include samples of:
Surface Finish (visual)
Plastic color chips
The main thing is that each have a unique identifier and be controlled so the user is assured that it is the current and correct sample. Depending on your situation, control may be by including in your calibration software (to establish a time period for certification of correctness), keeping a local list of the sub-assemblies showing their revisions, etc. An important part of this is that an authorized, qualified person(s) approve the use of each, and that the authorization is documented.
There may be better ways but this has worked well for us and accepted by more than one registrar.
Now that I've had a little time to think about my 'shoot-from-the-hip' answer, let me add another example.
We also test some of our devices for air flow by attaching them to a test fixture with a flow meter. This is test hardware that must be calibrated at pre-determined intervals. The same reasoning would apply if you were checking any type of liquid or current flow.
I believe another type of fixture would be the type that physically tells you the part is good, e.g.; it slides under a bar, fits between pins, falls between the marks on a go/no-go gauge, etc. How does an operator know that the gauge isn't worn or slightly bent without a regular (documented) inspection of the gauge itself? Just be sure to tailor the effort involved to the nature of the measurement. No sense spending a lot of time and money if the spec being measured isn't critical to the function, but that's a company decision.
Hope this is of some use.
[This message has been edited by Alf Gulford (edited 09 October 2000).]