We take into account three-body anisotropic forces between molecules to calculate the energy of the S 0 ͑0͒ triplet in solid hexagonal close packed hydrogen under pressure. Three-body contributions result in one term depending on the orientation of only one molecule ͑crystal field term͒ and two others that couple rotations of different molecules ͑roton terms͒. Three-body interactions contribute, to a large extent, to the roton frequencies. Their inclusion in the calculation increases the calculated average frequency of the triplet, even at relatively low density, changing substantially the estimate of the internuclear distance. By contrast, the triplet splitting is substantially unaffected by three-body terms, resulting therefore a good candidate to test anisotropic two-body potential models against experiment.The structure and the dynamical properties of solid hydrogen under pressure are exciting subjects of active research. [1][2][3][4][5][6] In the low-pressure hexagonal close-packed ͑hcp͒ phase, the molecules rotate in their lattice sites and J is a good quantum number. In an ordered crystal ͑with only para-H 2 molecules͒, the rotational excitation acquires a collective character, due to the anisotropic interaction between molecules. The transition from the ground state ͑all molecules in the J = 0 state͒ to the state with one J = 2 excitation gives rise, in the Raman spectrum, to the S 0 ͑0͒ triplet, that has been observed experimentally at low pressure 7 and up to about 110 GPa. 4,[8][9][10] Above this pressure, solid para-hydrogen transforms to an orientationally ordered phase ͓broken symmetry phase ͑BSP͔͒, whose crystal structure has been investigated theoretically, 11-13 spectroscopically, 3 and, recently, with more direct methods as x-rays and neutron scattering. 6 Values of the S 0 ͑0͒ Raman frequencies and of their pressure evolution contain information on the same anisotropic intermolecular potential components which drive the transition to the BSP.The quantitative analysis of these data by theories based on known anisotropic pair potentials is extremely problematical and is far from being satisfactory. 4,14 One problem concerns the splitting of the triplet which is always overestimated, as underestimated is the transition pressure to the BSP, if calculated analogously. The other quantity of importance is the average position of the triplet, which depends on the rotational constant and therefore gives information on the intramolecular distance and on its pressure dependence. The information extracted with the analysis of the Raman data depends, however, on our capability to calculate the contribution of the anisotropic potential to the roton energies ͑average value and splitting͒ in the solid. The main contribution to the splitting comes from the electric quadrupolequadrupole interaction, and is generally calculated within first-order perturbation theory. 1 Three other contributions may have importance at high pressure: One comes from different components of the anisotropic pair-wise potential, one fr...