Estimates of stratospheric age from observations of long-lived trace g
ases with increasing tropospheric concentrations invoke the implicit a
ssumption that an air parcel has been transported intact from the trop
opical tropopause. However, because of rapid and irreversible mixing i
n the stratosphere, a particular air parcel cannot be identified with
one that left the troposphere at some prior time. The parcel contains
a mix of air with a range of transit times, and the mean value over th
is range is the most appropriate definition of age. The measured trace
r concentration is also a mean over the parcel, but its value depends
both on the transit time distribution and the past history of the trac
er in the troposphere. In principle, only if the tropospheric concentr
ation is increasing linearly can the age be directly inferred. We illu
strate these points by employing both a one-dimensional diffusive anal
og of stratospheric transport, and the general circulation model (GCM)
of the Goddard Institute for Space Studies (GISS). Within the limits
of the GCM, we estimate the time over which tropospheric tracer concen
trations must be approximately linear in order to determine stratosphe
ric age unambiguously; the concentration of an exponentially increasin
g tracer is a function only of age if the growth time constant is grea
ter than about 7 years, which is true for all the chlorofluorocarbons.
More rapid source variations (for example, the annual cycle in CO2) h
ave no such direct relationship with age.