My experience with using 'secondary standards' as calibrated references is that you must prove them first. That is you need to select several items and repeatably check them over a period of time and select the one with the best stability. you then need to back this one piece of equipment up with another stable reference to be used when you suspect that the first of giving an unexpected result. Even then you can not be absoluteley sure of your calibration. I would only normally recommend doing this when direct calibration against standards is not possible.
