Second Report of Referee A -- LF16076/Nguyen ---------------------------------------------------------------------- I would like to thank the authors for their answer to my questions. It seems to me, though, that in addition to [9,10], reference [8] already reached g0~\gamma. Although I understand that the results presented here are an improvement over the previous achievements, I'm not sure it's enough to justify a publication in PRL. ---------------------------------------------------------------------- Second Report of Referee B -- LF16076/Nguyen ---------------------------------------------------------------------- I thank the authors for their careful consideration of my comments, and I believe that their changes strengthen the manuscript. Concerning the larger question of whether this result represents a significant advance: This question is always difficult and subjective, but I feel that to satisfy this criterion, it would be important to demonstrate that this new device can access a regime that is not dominated by cavity losses. I understand the authors’ arguments that it should work with different mirrors. The fact is, however, that from their original proposal (Ref. 13) one could already argue that it should work, and the role of an experimental paper is to demonstrate that it does work in the face of myriad practical difficulties. Reply 1: It is helpful that the authors have clarified in the manuscript that the in-vacuum realignment is automated. More generally, I was concerned with the integration of in-situ positioning stages, which adds complexity to the vacuum setup and may introduce problems with the vibration stability and cavity lock (problems that might first arise with a higher finesse cavity). It seems that perhaps we are just trading one technical challenge (small, high-finesse cavities) for another. Whether this new challenge is worth it depends on the system parameters that can be observed. Reply 2: The authors state that their losses are dominated by absorption from mirror contamination, but whatever the source of the losses, the average number of round-trips for a cavity photon is reduced by the losses. Therefore, it is still not clear to me that an increased finesse would not result in more sensitivity. In my opinion, all other points have been resolved. In reading the revised manuscript, I noticed two points that I missed earlier: - In Figs. 3 and 4, the x axis is labeled simply as “detuning.” It should be made clear in either the axis label or the caption (or both) that this is the detuning of the probe laser frequency from the cavity. - In the conclusion, it is stated that only in very short fiber cavities have ratios of g to gamma significantly larger than 1.7 been demonstrated (by Reichel and colleagues). In Hood et al., Science 487, 1447 (2000), for a very short cavity which is not a fiber cavity, this ratio is 110 to 2.6.