In the ITS-90, any thermocouple figures out as an interpolation or a transfer measuring standard in its realisation, although their use were largely disseminated in previous temperature scales. National Metrology Institutes keep on calibrating some noble metal thermocouples, mainly types R and S, to be used by accredited laboratories as standards for calibration by comparison. It is known that calibration depends on the depth in which the thermocouple is settled into the furnace. Then, with this paper, we intend to present how non-homogeneity in wires interferes in the results of different calibration set-ups. Using fixed point cells, in their changing phase soaks, as ideal thermal means, Brazilian thermometry group, as the dissemination technology body in Temperature of Brazilian Accredited Laboratories for Calibration (RBC), presents some results of a study about non-homogeneity influence in wires of thermocouples, reporting their behaviour when they are moved up and down in the ingots of cells. These characteristics of thermocouples, which have alloy wires, are strengthened when we compare their results with Gold-Platinum thermocouples and Standard Platinum Resistance Thermometers. Another aim of this work is to discuss uncertainty evaluation for noble metal thermocouples. Expanded uncertainty for noble metal thermocouples depends basically on electromotive force drift along time. This effect is result of changes in alloys because of heating during calibration. We report some uncertainty budgets for thermocouple calibrations by fixed points method, discussing drift influence on our uncertainties for types R and S thermocouples (± 0,3°C (k=2) from 0°C to 1100°C). |