What was done
In search of a better explanation than the two (oceanic productivity and alkalinity increases) that currently attempt to explain the strong coupling between atmospheric CO2 concentration and Antarctic air temperature that is evident over glacial-interglacial cycles, the authors come up with a mechanism that seems to satisfy all the criteria imposed by empirical observations of a number of related phenomena.

What was learned
Stephens and Keeling conclude that "by significantly limiting the sea-to-air CO2 flux in the primary region for deepwater ventilation, expanded Antarctic sea ice during glacial times may trap relatively more carbon in the deep ocean, thereby reducing atmospheric CO2 concentrations."

What it means
The mechanism that is put forth in this paper to explain the observed synchrony between Antarctic temperature and atmospheric CO2 concentration during glacial-interglacial transitions is clearly one where temperature is the independent variable that alters sea ice extent, which then alters the sea-to-air CO2 flux in the high-latitude region of the Southern Ocean and consequently changes the CO2 content of the atmosphere.  Hence, in this best explanation yet for the impressive correlation of CO2 and air temperature over glacial-interglacial cycles, atmospheric CO2 variations are the result of temperature variations and not vice versa, which is what the ice core data also tell us about this phenomenon.  Once again, therefore, we have another demonstration of the fact that it is changes in air temperature that drive changes in atmospheric CO2 concentration and not the reverse phenomenon, which figures so highly in GCM predictions of continued global warming as a result of the rising CO2 content of earth's atmosphere.