Thermoelements are usually given a high-tem- perature anneal before use to remove internal stresses (from wire drawing) and to stabilize metallurgical/crystallographic changes (grain growth, etc.) and the effects of chemical impurities (volatilization, oxidation, diffusion, etc.). For Au wire from two sources, both 99.999% pure, significant changes occured on annealing at temperatures from 700°C to 1000°C [2]. Interestingly, the changes in the two samples were of opposite sign: when isothermally annealed at 1000°C for 10h the change at the Sb point (,,~ 631 °C) for one sample was + 1.7 IxV (equivalent to 85 mK for Au/Pt), and -0.8 ~tV (40 mK) for the other. This difference highlights the significant effect of low levels of chemical impurity, such as Fe and Pd, and crystallographic defects in even the "best" of Au thermoelements. Moreover, it would seem that a 10 h uniform anneal at 1000°C is required to stabilize these changes. Further annealing of 99.999% Au wire at 1000°C, to 100 h, had little effect for both wire sources, less than 0.4 ~tV (equivalent to 20 mK at 631°C: for Au/Pt). On the other hand, the emf at 639°C in a sample of 99.99% wire changed by 1.4 ~tV when given