Realization and dissemination of units in a revised SI

The General Conference of Weight and Measures (CGPM) with its Resolution 12 has officially prospected the adoption, possibly at the next General Conference of 2011, of new definitions for kilogram, ampere, kelvin and mole using fundamental constants as reference quantities. Thus, the outlines and features of the SI after that date are worthy to be considered and a general afterthought of the metrological activity in the new situation to be attempted.The future realisations of units will greatly benefit from the accelerated endeavour to determine the relevant fundamental constants with the minimum possible uncertainty before their values are fixed by definition. In fact, many of those experiments will become realisations of units when the new definitions are adopted.To define the units by reference to fundamental constants implies to abandon the identification of the unit with its primary standard, as in the old metrological tradition. To realise a unit will definitely consist in assigning a value to a primary standard, consistent with the fixed values of the reference constants, by means of an experimental procedure, independent of a specific definition, which could even not exist. The primary standard should be suitable to dissemination by direct comparison, thus essentially stable and accessible with the highest precision, while the role of the realisation experiment would be mainly related to indirect measurements, typical of scientific activity, which involves the coherence of the unit system. The two distinct roles, of unit realisation and primary standard, correspond to different uncertainty components, of which only one is implied in dissemination activity, just aiming at compatibility of measurements of a specific quantity. Each of the two uncertainty components has a different evolution from the time of the unit redefinition.These last considerations could validly contribute in critical issues, such as deciding whether the time for a unit redefinition has come or it should be preferable to wait for new and better experiments before fixing the value of a constant. This could be the case for the kilogram redefinition.