To perform this work, two sets of solar water heaters, each set consisting of two flat plate collectors and a storage tank, were tested. The collectors in one system consist of aluminium absorbers painted matte black, and those in the other system consist of copper-aluminium composite absorbers with anode oxidized coating. For each of the systems, one collector is glazed and the other is unglazed. The experimental results showed that, if thermosyphonic reverse flow in the solar systems was not allowed, the stagnant absorber temperatures of all the collectors were 6-8 degrees C and about 1 degrees C lower than the ambient temperature at clear and overcast nights, respectively, the glazing and absorber coating of a collector had insignificant effects on the stagnant temperature depression of the collector absorbers (defined as the temperature difference between ambient air and absorbers), but the weather conditions had considerable effects. These results implied that the collector might be damaged by freezing at clear nights even when the air temperature was above 0 degrees C, such as 2-3 degrees C, and the possibility of freeze damage at clear nights was much higher than that at overcast nights for a given ambient air temperature slightly above the freezing temperature. Experimental results also indicated that if reverse flow in the solar systems were allowed, the absorber temperature of the collectors was stable all night at both clear and overcast nights and even higher than the ambient air temperature at overcast night as a result of the fact that the heat lost by the collectors at night was offset by the hot water inside the storage tank of the systems through the thermosyphonic reverse flow. This indicated that the reverse flow was very effective for preventing freezing of the collectors, and the freeze damage could be, theoretically, avoided by keeping the water temperature inside the storage tank of a solar thermosyphonic system at a certain level. (C) 2008 Elsevier Ltd. All rights reserved.