intensity of the tensile stress set up in the colder layers of glass will depend upon the temperature gradient which exists in the glass; thus if glass were a good conductor of heat it would never be possible to set up a sufficient difference of temperature between adjacent layers to produce fracture; for the same reason, vessels of very thin glass are less apt to break under temperature changes than those having thick walls, since the greatest difference of temperature that can be set up between the inner and outer layers of a thin-walled vessel can never be very considerable. It also follows from these considerations, that if a cold glass vessel be simultaneously heated or cooled from both sides, it can be safely exposed to a much more sudden change of temperature than it could withstand if heated from one side alone; on the other hand, when very thick masses of glass have to be heated, this must be done very gradually, as a considerable time will necessarily elapse before an increment of temperature applied to the outside will penetrate to the centre of the mass. It should also be noted here, that in addition to the thermal conductivity of the glass, its heat capacity or specific heat also enters into this question, since heat will obviously penetrate more slowly through a glass whose own rise of temperature absorbs a greater quantity of heat. It will thus be seen that “thermal endurance” is a somewhat complicated property, depending upon the factors named above, viz.: coefficient of expansion, thermal conductivity, specific heat, Young’s modulus of elasticity, and tensile strength. The coefficient of thermal expansion varies considerably in different glasses, and we can here only state the limiting values between which these coefficients usually lie; these are 37 × 10^{-7} as the lower, and 122 × 10^{-7} as the upper limit. These figures express the cubical expansion of the glass per degree Centigrade, the corresponding figures for steel and brass respectively being about 360 × 10^{-7} and 648 × 10^{-7} respectively. It should be noted that vitreous bodies of extremely low expansibility are obtainable by the suitable choice of ingredients, but in some cases these “glasses” are white opaque bodies, and in all cases they present great difficulty in manufacture, owing to the fact that alkalies and lime must be avoided in their composition.