The condensation coefficient (C.C.) of a liquid is defined as the ratio of the observed rate of evaporation under given conditions to the 'expected' rate calculated from the known saturation vapour pressure, for the same conditions. Values of the C.C. less than 1 have been reported in the past, but the accuracy of the experiments has been criticized on various grounds. Experimental work is described which establishes that the C.C. of glycerol at 18 degrees C is 0$\cdot $052 $\pm $ 0$\cdot $005. Previous experimental determinations are reviewed in the light of this result, and it is found that for various liquids the C.C. is nearly equal to the free-angle ratio defined by Kincaid and Eyring on the basis of a simple model of liquid structure. This equality is not found for chloroform. The theory of evaporation is discussed on a model of the liquid surface consisting of a compact ordered layer with a mobile adsorbed layer of molecules above it. On the assumption that evaporating and condensing molecules pass through the adsorbed condition, it is shown that the C.C. is determined by the rates of exchange between substrate and adsorbed layer, and adsorbed layer and vapour, respectively. In particular, a C.C. near unity results if the rate of exchange between substrate and mobile layer is very much greater than that between mobile layer and vapour. The above assumption does not give a satisfactory explanation of the equality of C.C. and free-angle ratio for alcohol molecules. It is shown that for polar molecules which form an oriented surface layer in the liquid state it is to be expected that evaporation will take place mainly by direct exchange of molecules between the ordered layer and the vapour, although an adsorbed layer will be present. In this case an argument given by Herzfeld shows the condition for equality of C.C. and free-angle ratio to be that there should be no exchange of energy between rotational and translational co-ordinates in the process of evaporation. It is further concluded from a rather qualitative discussion that a major contribution to C.C. and free-angle ratio for the simple alcohols arises from restriction of rotation of the hydroxyl group about the carbon-oxygen bond.