Two forecast integrations started from slightly different (or possibly identical) initial conditions, using two different NWP models, aimed at studying forecast error growth. While identical twin integrations differ only due to initial value differences, fraternal twins also diverge due to differences in model formulation.