The Provisional Low Temperature Scale of 2000, PLTS-2000, which was adopted in October 2000, uses the melting pressure of 3He to provide the basis for temperature measurement in the range from the Néel temperature of the solid, T 2000=0.902 mK, up to 1 K. The definition and derivation of the scale has been published, and the present paper now gives guidance on the practical methods by which the melting pressures can be measured in the laboratory. Various options are described, depending on the equipment available, and the uncertainties that may be achieved are considered.

An internationally-accepted ultra-low temperature scale is needed to provide the basis for reliable thermometry in the temperature range in which commercial dilution refrigerators operate, and in experiments investigating the properties of 3 He and other condensed matter. Several laboratories have developed 3 He melting-pressure scales, but there are substantial differences even between the most recent of them. These amount to about 0.3 % of T near 500 mK , rising to about 6 % of T at 0.9 mK . In 1996 a collaboration was initiated between low temperature physicists in national laboratories and elsewhere to derive an equation for the melting pressure of 3 He which could be accepted for international use from 1 K to 0.9 mK , the Néel temperature of solid 3 He. After an open workshop in Leiden in 1998, discussions took place to see if thermodynamic calculation of 3 He melting pressures could resolve the differences. In January 2000 the authors (apart from ALR and GS) met at NIST and were able to reach a compromise on the Provisional Low Temperature Scale, PLTS-2000. Its 1-sigma uncertainty is estimated to be 0.3 % of T (up to a maximum of 0.5 mK ), but this rises to about 2 % of T at 0.9 mK . The provisional status recognizes that the PLTS-2000 is a compromise, rather than a true consensus, but it is likely to be some years before it can be replaced by a more accurate scale. The scale was announced at the Quantum Fluids and Solids conference in Minnesota, USA, in June 2000, and was formally adopted by the Comité International des Poids et Mesures in October 2000.

We have measured the temperature dependence of the superconductive critical magnetic field of three samples of beryllium from three different sources. The irregular geometric shape of one sample prevented a complete analysis of the data, but the data for the other two samples were well represented by the BCS model of superconductivity. The values of T c and H c (0) are found to be sample dependent. For the purest sample studied, T c =24.34±0.02 mK, H c (0) =107.7±0.2 μ T and ( γ / V )=32.9±0.3 μ J/cm 3 K 2 . By a very small extrapolation of the data as a function of purity, it is inferred that the critical parameters for pure Be are: T c =24.38±0.02 mK, H c (0) =107.9 μ T± 0.2 μ T and ( γ / V )=33.0±0.3 μ J/cm 3 K 2 .

In order to extend the international temperature scale of 1990, ITS-90, below its lower limit of 0.65 K, we have developed a temperature scale ranging from 6 to 750 mK. Values of absolute temperature are defined on this scale by an R-SQUID noise thermometer. A review is given here of the decade of our experience in the operation of this thermometer and in modeling its systematic errors. The reproducibility of noise temperature values was assessed using superconductive fixed points and the melting curve of3He. To assess the accuracy of the noise thermometer, it was compared with another absolute thermometer (based on nuclear orientation) at the lowest temperatures and with an internationally recognized scale above 0.5 K. The noise thermometer temperatures were used to calibrate two paramagnetic salt thermometers, the result being the construction of a temperature scale that is smooth to approximately 0.01%. On the basis of these comparisons, temperature values defined by the R-SQUID are deemed to be reproducible to within 0.1% and accurate to within 0.3%.