In regions of active oceanic convection, such as the Labrador Sea, small- and mesoscale spatial variability is observed in the temperature and salinity fields (T and S). Often T and S structures are “density-compensated,” with the density contribution of the S anomaly nearly equal and opposite to the contribution from the T anomaly;this is manifest as variability in the “spice” field, τ = αT + βS, where α and −β are the local expansion coefficients for T and S. Here the mechanisms for generating T and S variability by convection around a preexisting mesoscale eddy, with particular attention to τ variability, are investigated. The authors perform several numerical experiments with identical density stratification, mesoscale circulation, and surface buoyancy forcing, but with different combinations of T and S in the stratification and surface flux. In all cases with both T and S variations present, it is found that spice variability exceeds that of density. In particular, there are substantial heat... Abstract In regions of active oceanic convection, such as the Labrador Sea, small- and mesoscale spatial variability is observed in the temperature and salinity fields (T and S). Often T and S structures are “density-compensated,” with the density contribution of the S anomaly nearly equal and opposite to the contribution from the T anomaly;this is manifest as variability in the “spice” field, τ = αT + βS, where α and −β are the local expansion coefficients for T and S. Here the mechanisms for generating T and S variability by convection around a preexisting mesoscale eddy, with particular attention to τ variability, are investigated. The authors perform several numerical experiments with identical density stratification, mesoscale circulation, and surface buoyancy forcing, but with different combinations of T and S in the stratification and surface flux. In all cases with both T and S variations present, it is found that spice variability exceeds that of density. In particular, there are substantial heat...