Preamble

The theoretical bases of thermometry are discussed in Chapter 1. It is rare for experimentalists to use methods which relate directly to a Carnot cycle, though there are certainly users of secondary methods which have respectable theoretical support of another sort (e.g. the Boltzmann factor in 60Co y-ray anisotropy). Some properties of some materials have useful dependences on temperature in the desired range (e.g. the electrical resistance of carbon composites) but are best described by polynomial fits rather than fundamental theory. The fact is that there are many ways of measuring temperatures in the millikelvin range and experimentalists make choices which best fit their needs. The extents to which these techniques give temperatures which are accurate (i.e. equal to the true Kelvin temperatures) or consistent when compared with temperatures determined in other ways are often not fully certain. Users frequently rest their cases on widely accepted results obtained and published by respected workers in the past (e.g. for the melting pressure of 3He (Greywall (1985)), or they take strength from demonstrations by others that two methods do in fact agree within acceptable limits. Examples of comparisons include platinum NMR versus nuclear orientation (Berglund, Collan et al. (1972)), platinum NMR versus CLMN susceptibility (Alvesalo et al (1980)), noise versus 60Co y-ray anisotropy (Soulen and Marshak (1980)), and 3He melting pressure versus CLMN susceptibility (Parpia et al. (1985)). Disputes can and do arise about whether small temperature-dependent effects are truly properties of the system being studied or whether they are artifacts of an imperfect scale. A recent example of this has been the debate about whether the low temperature heat capacity of liquid 3He does or does not contain a term proportional to T3 ln T.