Magnetic-dipole two-photon absorption (MD-TPA) is introduced as a new spectroscopic technique for the study of odd-parity states. Contrary to the classical electric-dipole two-photon absorption (ED-TPA), where only spin-allowed even-parity transitions can be excited, MD-TPA allows us to excite spin-forbidden odd-parity states. This new technique is therefore very well suited to study paraexcitons in semiconductors and insulators. As examples MD-TPA measurements of RbI, NaI, and NaBr are presented.PACS numbers: 42.65.k, 71.35.+z Two-photon absorption (TPA) was first treated theoretically more than sixty years ago by Goppert-Mayer [1] using second-order perturbation theory. With respect to selection rules two-photon processes can be interpreted as two successive one-photon transitions. From Laporte's rule which states that one-photon dipole transitions are only allowed between states of diferent parity, one immediately derives that two-photon transitions are allowed only between states of the same parity. About thirty years ago Hopfield and Worlock [2] were the first to do two-photon spectroscopy on excitons. In agreement with the above-mentioned selection rule they found that in alkali halides two-photon transitions are indeed allowed to even-parity P excitons but strictly forbidden to oddparity S excitons.In order to excite odd-parity states by nonlinear spectroscopy one has to step onto three-photon spectroscopy (TPS) as was again demonstrated in alkali halides [3]. By the use of di8erent three-photon techniques one can observe the polariton dispersion and the longitudinal exciton with high accuracy [4].In this Letter we introduce a new spectroscopic technique, which allows us to excite odd-parity states by two-photon absorption. Contrary to classical two-photon absorption where both photons induce electric-dipole transitions (ED-TPA), we consider two-photon processes where one of the photons induces a magnetic-dipole transition and the other an electric-dipole transition.In the following we will briefly outline the theoretical background of magnetic-dipole two-photon absorption (MD-TPA).As in the case of ED-TPA [5] one can derive the twophoton transition probability Wg f for MD-TPA from second-order perturbation theory: -&(fl(MD)2]i)(il(ED) ilg) Ws f (x E, -Eghindi (fI (I&)i I ') (& I (s&)2 I g) ) E, -Eg -~g x 6(Ef -Egh(u)i + (d2)), where E~a nd Ef refer to the energy of the ground and final states, respectively. As intermediate states~i ) (energy E, ) one has to consider all states which are allowed for electric-dipole transitions from the ground state~g ).(ED)"and (MD) refer to the electricand magneticdipole operator, respectively. v = 1, 2 labels the incoming photons with energies hu and polarization directions e . In Eq.(1) we have assumed that the first transition (g~i) is of electric-dipole type, which is expected to be the case for many solids, where the lowest energy transitions (e.g. , 1S excitons) are dipole allowed.As first shown by Inoue and Toyozawa [5), the twophoton polarization selection rules c...