With respect to gas transport the firn near the surface of an ice sheet can be divided into three zones. The uppermost is a ''convective zone'' located just below the surface of the ice sheet in which the air is rapidly flushed by convective exchange with the overlying atmosphere. Below the convective zone there is a ''diffusive air column'' in which diffusion is rapid but there is no convection. Between the bottom of the diffusive air column and the bubble close-off region there may be a ''nondiffusive zone'' in which diffusion is so slow that negligible gas transport occurs. The diffusive air column is characterized by progressive enrichment with depth of 15N in N2 (and heavy isotopes of gases in general) as predicted using the barometric equation. In this paper we present data on the Δ15 N of N2 in recently trapped air samples from 12 ice cores, along with numerous downcore samples from Byrd, Vostok, and Dome C. Bubble close-off depths for these cores (calculated from a densification model) ranged from 51 to 114 meters below the surface (mbs). We used these data and the barometric equation to calculate the thickness of the diffusive air column, and found that it comprised 46 to 93% of the total firm thickness at our study sites. Paleo-close-off depths calculated from the densification model for glacial sections of Byrd, Vostok, and Dome C are 15--25 m deeper than close-off depths today. Diffusive column heights, calculated from Δ15N, varied in a more complex manner. The diffusive column height at Byrd appears to have decreased from 74 m during the last glacial period to 50 m during the Holocene.

At Vostok and Dome C the diffusive column height calculated from 15N increase from about 65 m during the last glacial period to about 80 m in the Holocene. We use records of surface temperature and CO2 at Byrd and Vostok, along with their respective chronologies, to constrain the ice age-gas age difference (Δage) throughout the section of the Vostok ice core corresponding to the last glacial termination. In principle, Δage values calculated from these data can be used to discriminate whether gas in the firn mixes to the bubble close-off depth or to a depth equivalent to the diffusive column height. In practice, however, uncertainties in the chronology of Byrd and Vostok are too great to allow us to distinguish between these two possibilities. One can only say that at the time of the last termination, Δage for Vostok was between 3 and 10.5 kyr. Previous estimates fall within this range. ¿ American Geophysical Union 1992