Thermal expansion is a vital factor in yet another of the uses of glass. Our ordinary instrument for measuring temperature--the mercury thermometer--is very considerably affected by the expansive behaviour of glass. When a mercury thermometer is warmed the mercury column rises in the stem because the mercury expands upon warming to a greater extent than the glass vessel, bulb and stem, in which it is contained. The subject of the graduations and corrections of the mercury glass thermometer is a very large one and somewhat outside the scope of the present volume; but attention should be drawn in this place to the peculiarities of the behaviour of glass that have been discovered in this connection. One of these is that when first blown the bulb of a thermometer takes a very considerable time to acquire its final volume, the result being, that if a freshly made thermometer is graduated, after some time the zero of the instrument will be found considerably changed, generally in a direction which indicates that the volume of the bulb has slightly increased. By a special annealing or “ageing” process this change can be completed in a comparatively short time before the instrument is graduated. There is, however, a further peculiarity which is prominent in some thermometers, although very greatly reduced in the best modern glasses. This becomes apparent in a decided change of zero whenever the thermometer has been exposed for any length of time to a high temperature, the zero gradually returning more or less to its original position in the course of time. With thermometers made of glasses liable to these aberrations, the reading for a given temperature depended largely upon the immediate past history of the instrument; but, thanks to the Jena Works, thermometer glasses are now available which are almost entirely free from this defect. In this connection the curious fact has been observed that glass containing both the alkalies (potash and soda) shows these thermal effects much more markedly than a glass containing one of the alkalies only. _The thermal conductivity_ of glass, except in so far as it affects the thermal endurance, is not a matter of any great direct practical importance, although the fact that glass is always a comparatively poor conductor of heat is utilised in many of its applications, as, for example, the construction of conservatories and hot-houses, although even in that case the opacity of glass to thermal radiations of long wave-lengths is of more importance than its low thermal conductivity. Similar statements apply, in a still more marked degree, to the subject of the specific heat of glass.