When it started operation, the 3.9 m AAT suffered from poor matching of air temperatures inside and outside the dome. There was little natural ventilation and the forced ventilation was limited, being originally designed only to supplement active control of the dome air temperature (which had been cut from the installation to save funds). Obviously, improvements were essential and reliable measurements of local seeing were needed to decide the best way to proceed. Comparison of routinely observed seeing with inside/outside temperature differences indicated local degradation of about 0.5 arc sec diameter for each 1°C. Fast response thermometers showed that the worst air temperature fluctuations were generally confined within about 5 m from the dome aperture. A Hartmann technique, devised to allow direct comparison of the seeing of the 3.9 m telescope with that of a small telescope, mounted either inside or outside the dome, confirmed the magnitude of the dome seeing. After a careful review of heat flows within the building, the forced ventilation was substantially upgraded and improved seeing has resulted. However, a variant of the Hartmann technique, using the 3.9 m telescope in autocollimation, has shown that local degradation is still often 0.5 arc sec diameter or worse. Work to quantify the remaining causes of seeing degradation will continue, including the development of a system for routine microthermal analysis. The optical and microthermal techniques used at the AAT are widely applicable to the seeing studies vital for the large telescopes of the future.