The comparison involved the six NMIs previously involved in CCT-K3 (BNM-INM/CNAM, SMU, IMGC, NMi-VSL, NPL, PTB) and 18 European national laboratories. The comparison was divided into five different loops coordinated by a co-pilot chosen among the laboratories having participated in the CCT-K3 comparison. BNM-INM/CNAM played the role of pilot in establishing the link between the five loops. In each loop, an artefact in the form of a standard platinum resistance thermometer (SPRT) was circulated among the participant laboratories.
In order to have sufficient information about the possible drift of the SPRTs, the co-pilots have performed a calibration over the full temperature range at the beginning and at the end of the loop.
Each participant laboratory carried out the measurements in agreement with the protocol. After initial measurements of R(tpw) and successful annealing treatment, each calibration consisted of three measurements of R(t) at each fixed point, in separate realizations.
Each measurement was followed by a measurement of R(tpw). The order of fixed points was Zn, Sn, In, Ga, Hg, Ar, but if a fixed point was not available, it was omitted. Values of W(t), and their average value, were calculated.
The results were reported to the co-pilot and BNM-INM/CNAM, with uncertainty budgets and other specified data.
The results of the comparison were analysed by BNM-INM/CNAM. For the sake of clarity, the results are firstly presented loop by loop. In a second step a EUROMET Reference Value (ERV) taking into account the whole comparison was defined. In order to calculate this ERV advantage was taken of the presence of BNM-INM/CNAM in the five loops; the differences between each laboratory's results and BNM-INM's were considered.
The participant laboratories decided during a Workshop organized by the pilot in Vienna in April 2005 to use as ERV the weighted mean rather than the simple mean.
The differences (TLab - TERV) are presented with the associated uncertainties. Given that the protocol of the comparison contains a detailed description of how the uncertainties are to be calculated, the uncertainty budgets established by the participants seem consistent or, at least, homogeneous.
The degrees of equivalence between the participating laboratories (in the same loop or in different loops) are given with their associated uncertainty.
Finally, a method for establishing the bilateral equivalence between the participants in CCT-K3 and in EUROMET'T-K3 is proposed.
It is recommend to use a group of 'linking laboratories' composed by the pilot and the co-pilots in order to link EUROMET.T-K3 to CCT-K3.
The hypothesis is that the mean temperature of the pilot and co-pilot laboratories is the same in EUROMET.T-K3 as it was in CCT-K3.
A specific analysis of the pilot's and co-pilot's results shows that even when the results given by one or several members of the 'linking laboratories' are not positioned in the same place as in EUROMET.T-K3 and CCT-K3 the mean calculated for the group is really reproducible.