Water Triple Point Comparison: Plateau averaging or Peak Value?

With a certain regularity, national metrology institutes conduct comparisons of water triple-point (WTP) cells. The WTP is the most important fixed point for the International Temperature Scale of 1990 (ITS-90). In such comparisons, it is common practice to simply average all the single measured temperature points obtained on a single ice mantle. This practice is quite reasonable whenever the measurements show no time dependence in the results. Ever since the first Supplementary Information for the International Temperature Scale of 1990, published by the Bureau International des Poids et Mesures in Sèvres, it was strongly suggested to wait at least 1 day before taking measurements (now up to 10 days), in order for a newly created ice mantle to stabilize. This stabilization is accompanied by a change in temperature with time. A recent improvement in the sensitivity of resistance measurement enabled the Istituto Nazionale di Ricerca Metrologica to detect more clearly the (possible) change in temperature with time of the WTP on a single ice mantle, as for old borosilicate cells. A limited investigation was performed where the temperature of two cells was monitored day-by-day, from the moment of mantle creation, where it was found that with (old) borosilicate cells it may be counterproductive to wait the usual week before starting measurements. The results are presented and discussed, and it is suggested to adapt the standard procedure for comparisons of WTP cells allowing for a different data treatment with (old) borosilicate cells, because taking the temperature dependence into account will surely reduce the reported differences between cells.