ABSTRACT1Ieasurements of the expansion of porous Vqcor glass containing fixed quantities of adsorbed water indicate that in every instance a phase transition starts belo\\ -50" C, with the main change occurring below -160" C. Two other anomalies of the length variation mere observed near -22" C and -7" C when the amount of adsorbed water exceeded the value equivalent to two monolayers. The length of the adsorbent after co~npletion of a temperature cycle was different from the initial length. This is thought to be due to damage suffered by the glass aiid t o the fact that the length of the adsorbent is different for adsorption and desorption although the quantity adsorbed may be the same.The phase transition of adsorbed xenon takes place below the normal triple point and is a function of surface concentration as shown by length variations and equilibrium pressures for fixed quantities adsorbed. All transitions are gradual aiid hysteresis is exhibited by the isosteres. The adsorption isotherms for xenon -I'ycor glass show a decreasing adsorptive capacity and a contraction of the hysteresis loop with lower temperature. The inadequacy of the capillary coildensation theory of adsorption in relation t o these results is discussed.The capillary condensation theory assumes that the free energy of the adsorbate is less than that of bulk material a t the same temperature because it is held under concave menisci in the pores of the adsorbent. The vapor pressure of the condensed substance is calculated by the relationship due to Kelvin. The theory implies that the adsorbent is inert, and only the shape and size of the pores play any role in determining the adsorptive characteristics. I t follo\vs also that the adsorbate has bulk properties except the lowered free energy arising from the presence of the concave meniscus.There is, however, ample experimental evidence to the contrary. The observed dimensional changes of the adsorbents (0.37, in the case of 17ycor glass -water system (1)) during the adsorption process indicate that the solid is perturbed by the presence of the adsorbate. Similarly, the infrared spectra of the adsorbate contain lines caused by forbidden transitions normally observable only in the Raman spectra (2,3) showing that the adsorbed liquid is in a state different from the bulk state. The crystal structure of the frozen adsorbate, too, has been found by X-ray diffraction methods to differ from the one of the bulk crystal (4). Accordingly, solidified adsorbates failed to nucleate undercooled bulk liquid of the same composition (5). These findings are to be expected in view of the fact that adsorption is the result of molecular interaction between adsorbate and adsorbent, and this interaction perturbs both.On applying the equation due to Kelvin the question arises: is it permissible to use the relationship even as an approximation when most of the quantities involved are not only unknown quantitatively but are meaningless on a molecular scale? Surface tension (or even the tern1 surface itself) of a liquid ...