In order to investigate the temperature dependence of thermoelectric inhomogeneity in base-metal thermocouples, thermoelectric scan tests were performed at several temperatures using mineral-insulated metal-sheathed thermocouples of types K, N, E and J. To induce a severe inhomogeneity in the thermoelements, the thermocouples were heat-treated for 1000 h: types K and N at 950 °C, E and J at 750 °C. After the heat treatments, type K and N thermocouples were scanned at 180 °C in a stirred-liquid bath, and at 550 °C, 750 °C and 1000 °C in a sodium heat-pipe furnace. The measurement for the type E and J thermocouples was done only up to 750 °C. From the scan results, we found that the thermoelectric inhomogeneity, expressed as a ratio of maximum emf change within the working segment to the total emf generated, decreased at high temperature. We concluded that some base-metal thermocouples might have temperature-dependent inhomogeneity; therefore, applying the inhomogeneity value obtained at low temperature to assess the calibration uncertainty at higher temperatures may result in overestimation of the uncertainty.

A system was fabricated to realize the triple point of argon for the calibration of long-stem standard platinum resistance thermometers. A cryostat was constructed so that the temperature could be controlled quasi-adiabatically, and the melting was realized using the continuous-heating method. The combined uncertainty of the realization of the triple point of argon for a confidence level of 95% was 0.6 mK.

A new calibration method that provides first-order compensation for the time-lag error of dynamic temperature comparisons is suggested. The calibration uses the ramp response of the first-order system, assuming that the temperature change of the system can be approximately linearized in a narrow region near the calibration point. The calibration is done by measuring the rate of temperature change and comparing the temperature readings of a reference thermometer and test thermometers while the temperature of the system is decreased and then increased. This calibration is useful for industrial thermometers at low temperatures without using a sophisticated temperature-control system. We demonstrate in this work that the deviation of the result of time-lag compensation from the calibration in a stable thermal enclosure was less than 20 mK. The result indicated that the time-lag compensation can be used for calibrations where the required uncertainties are around 40 mK, and the calibration procedure is much simpler than that for comparison in a stable enclosure. Furthermore, the compensation suggested here has wider applicability at higher temperature ranges where thermometers are calibrated dynamically in liquid baths or furnaces.

The long-term stability of standard platinum resistance thermometers (SPRTs) in a temperature range between 0.01 °C and 419.527 °C was investigated from 352 samples of data accumulated from fixed-point calibration conducted during a period of eleven years. The rates of change of resistance RTPW at the triple point of water and the resistance ratio at the freezing point of Sn (WSn) and Zn (WZn) were analysed. For 75% of the samples investigated, the rate of change of RTPW was within 5.17 mK/year in magnitude. For the same portion of SPRTs, the rates of change in WSn and WZn were within 0.82 mK/year and 1.69 mK/year, respectively. Further statistical investigation showed that measurement at the triple point of water only was not a sufficient test to identify the change in WSn or WZn. However, due to the strong correlation between the changes in WSn and WZn, measurement of the resistance ratio at one of the fixed points in the range of usage is adequate to identify possible change in the entire range of usage.

A cryostat for fixed-point calibration of capsule-type SPRTs (standard platinum resistance thermometers) was developed. Using this system, cryogenic fixed points defined on the International Temperature Scale of 1990 (ITS-90) were realized. The cryogenic cells were argon, oxygen, neon, and two equilibrium-hydrogen (e-H 2 ) cells, made by INRiM, Italy. The uncertainty of the realization of each fixed point was estimated to range from 0.53 mK to 0.43 mK (k = 2). The realizations of the triple point of e-H2 using two sealed cells coincided within 0.1 mK. Therefore, we are able to calibrate capsule-type SPRTs down to 24.5561 K within an uncertainty of 1 mK (k = 2) by this system. A closed-cycle helium gas refrigerator was used for the cryostat. Each sealed cell was designed so that it could accommodate three sealed cells in the thermometer wells made within the cell. Therefore, the cryostat was designed to accommodate only one sealed cell at a time. The base temperature of this liquid-free cryostat, when one sealed cell and three capsule-type SPRTs were attached for calibration, was ∼17 K. For the realization of the triple point of e-H 2 , we used liquid helium for additional cooling. Adiabatic melting of the triple point was realized by controlling the inner-most radiation shield at a temperature very close to that of the triple point, and by applying a heat pulse by a heater directly wound to the cell. The amount of the heater power and the waiting time for the thermal equilibrium after each heat pulse were chosen in a way that the adiabatic melting could be finished within ∼6 h for each cell. The triple point of each cryogenic fixed point was deduced from the equilibrium temperatures between the heat pulses and subsequent extrapolation to the liquidus point. For the oxygen cell, temperatures of two solid-solid transitions (α-β and β-γ transitions) were also measured, and the results were consistent with values reported in the literature within the designated uncertainty.