Argon isotope data indicate retained argon in bulk amber (matrix gas) is radiogenic [40Ar/39Ar ???32o] than the much more abundant surface absorbed argon [40Ar/39Ar ???295.5]. Neutron-induced 39Ar is retained in amber during heating experiments to 150?? -250??C, with no evidence of recoiled 39Ar found after irradiation. A maximum permissible volume diffusion coefficient of argon in amber (at ambient temperature) D???1.5 x 10-17 cm2S-1 is calculated from 39Ar retention. 40Ar/39Ar age calculations indicate Dominican Republic amber is ??? 45 Ma and North Dakota amber is ??? 89 Ma, both at least reasonable ages for the amber based upon stratigraphic and paleontological constraints and upon the small amount of radiogenic 40Ar. To date, over 300 gas analyses of ambers and resins of Cretaceous to Recent age that are geographically distributed among fifteen noted world locations identify mixtures of gases in different sites within amber (Berner and Landis, 1988). The presence of multiple mixing trends between compositionally distinct end-members gases within the same sample and evidence for retained radiogenic argon within the amber argue persuasivley against rapid exchange by diffusion of amber-contained gases with moder air. Only gas in primary bubbles entrapped between successive flows of tree resin has been interpreted as original "ancient air", which is an O2-rich end-member gas with air-like N2/Ar ratios. Gas analyses of these primary bubbles indicate atmospheric O2 levels in the Late Cretaceous of ??? 35%, and that atmospheric O2 dropped by early Tertiary time to near a present atmospheric level of 21% O2. A very low argon diffusion coefficient in amber persuasively argues for a gas in primary bubbles trapped in amber being ancient air (possibly modified only by O2 reaction with amber). ?? 1991.