Griffiths effects in random Heisenberg antiferromagnetic S=1 chains

I consider the effects of enforced dimerization on random Heisenberg antiferromagnetic S=1 chains. I argue for the existence of novel Griffiths phases characterized by two independent dynamical exponents that vary continuously in these phases; one of the exponents controls the density of spin-1/2 degrees of freedom in the low-energy effective Hamiltonian, while the other controls the corresponding density of spin-1 degrees of freedom. Moreover, in one of these Griffiths phases, the system has very different low temperature behavior in two different parts of the phase which are separated from each other by a sharply defined crossover line; on one side of this crossover line, the system ``looks'' like a S=1 chain at low energies, while on the other side, it is best thought of as a S=1/2 chain. A strong-disorder renormalization group analysis makes it possible to analytically obtain detailed information about the low temperature behavior of physical observables such as the susceptibility and the specific heat, as well as identify an experimentally accessible signature of this crossover.