A MOLECULAR beam technique has been developed which permits investigation of the primary products of reactive collisions between gas molecules or radicals. The principle of the technique can be described by reference to the schematic diagram in Fig. 1. A molecular beam of one gas, A, is formed at an aperture, 1, in the wall of the vessel, 2, and collimated by a second aperture, 3. The emergent beam from 3 crosses the reaction zone, 4, and leaves by a third aperture, 5, to enter a detector, 6. The latter may conveniently be a mass-spectrometer, so orientated that the molecular beam of A crosses the ionization chamber of the mass-spectrometer without impinging on the electrodes. With sufficient pumping speed for the collimating, reaction and mass-spectrometer compartments, the mass-spectrum peaks due to A are produced almost entirely from molecules which have experienced a collision-free transit from aperture 1 to the ionization region, and the contribution from ‘background gas’ molecules of A in these compartments can be made negligibly small. The second gas, B, is admitted into the reaction compartment at a pressure such that the mean free-path of B in B in the main body of this compartment is of the order of the molecular beam path-length in this compartment. Now let nB be the average number of molecules of B crossing unit area within this compartment in unit time, and let nA be the average number of molecules of A which cross unit area normal to the molecular beam path, in unit time, in the reaction zone. Two limiting cases are conceivable. If nA ≪ nB, then molecules of B will effuse according to the cosine distribution law into the mass-spectrometer compartment. Thus the B molecules detected there will have arrived there as a result of collisions with other B molecules in the reaction zone. On the other hand, if nA ≫ nB then any molecule of B in the reaction zone has a much greater probability of collision with a molecule of A than with another molecule of B. Thus each B molecule detected in the mass-spectrometer will have arrived there by a collision-free path from a collision with an A molecule in the reaction zone. As a first approximation the rate at which such molecules reach the detector is the same as for the case of molecular effusion in the absence of the beam of A molecules. If, therefore, A and B are molecular species which react together, it is possible to detect the primary products of the reactive collisions in the reaction zone.