These are laser transition photons. In a laser, there is initially spontaneous emission of photons and many of these are "lost" in terms of their contribution to the beam. Some are on the main axis of the mirrors and therefore can participate in stimulated emission. The purpose of the mirrors around the outside of the cavity (not the main axis mirrors, but mirrors placed around the laser tube where spontaneous photons would be directed) is not to reflect the photons back into the lasing medium, but to prevent them from ever being emitted in the direction of the outside mirrors in the first place. If photons really have a source and a sink as Wheeler-Feynman suggested in 1940s, then without the sink part of the equation, the photons will cease to be generated. That is, the highly reflective mirrors around the cavity (not on the ends, again, a new set around the cavity), reduce the sink to a very low level so that any absorption of spontaneous photons is extremely minimal. Then the photon does not emit in that direction, but rather in a direction where a sink is more likely - on the main axis where it can participated in stimulated emission (increasing the gain) and then out the laser tube on the main axis. The power goes up, assuming the photons behave this way. The random nature of spontaneous emission in QM has always been a curious item, but the source-sink nature of photons goes back even to Einstein's mention of the ideas of Tetrode, who we communicated with frequently on the source-sink possibility of photons. This model explains a lot of aspects of photon entanglement, non-locality, etc. The experiment then, is to put highly reflective mirrors around the laser tube cavity, in addition to the mirrors on the end. The loss from spontaneous emissions is high, and being able to force them to emit originally down the main beam axis adds to the stimulated emissions already taking place.