A primitive equation model is used to study the sensitivity of baroclinic w
ave life cycles to the initial latitude-height distribution of humidity. Di
abatic heating is parametrized only as a consequence of condensation in reg
ions of large-scale ascent. Experiments are performed in which the initial
relative humidity is a simple function of model level, and in some cases la
titude bands are specified which are initially relatively dry. It is found
that the presence of moisture can either increase or decrease the peak eddy
kinetic energy of the developing wave, depending on the initial moisture d
istribution. A relative abundance of moisture at midlatitudes tends to weak
en the wave, while a relative abundance at low latitudes tends to strengthe
n it. This sensitivity exists because competing processes are at work. Thes
e processes are described in terms of energy box diagnostics. The most real
istic case lies on the cusp of this sensitivity. Further physical parametri
zations are then added, including surface fluxes and upright moist convecti
on. These have the effect of increasing wave amplitude, but the sensitivity
to initial conditions of relative humidity remains. Finally,'control' and
'doubled CO2' life cycles are performed, with initial conditions taken from
the time-mean zonal-mean output of equilibrium GCM experiments. The attenu
ation of the wave resulting from reduced baroclinicity is more pronounced t
han any effect due to changes in initial moisture.