After small changes, the article will be ready for publication. The paper and the provided data set ensures sustainability for future investigations. Very good! Please refer to my upload with remarks etc. . Please eliminate my confusion about the "best of the day" g-values. 

The contribution describes a 24 years long time series of absolute gravity measurements at Brest, France. The data itself have been already published with a DOI. It is an interesting record, in particular as leveling indicates an exceptional vertical stability of the region and the site is connected to the Brest tide gauge. The manuscript is informative and well written. However, a few points should be improved: 

1) The stability of both FG5 gravimeters over the years need to be addressed more in detail. Comparisons of absolute gravimeters are mentioned (L329) but the actual results are missing, except for a general "offset dispersion" (L322) of absolute gravimeters. In particular the impact of the change of the dropping chamber of FG5-206 to FG5X-206 should be analyzed. Are there further stations, e.g. Larzac, that could be used to check the stability of both FG5?

2) I really don't understand the selection of measurements. Of course, if serious disturbances happened or the drop scatter exceeded a certain threshold a (part of) measurement should be excluded. Otherwise, an increased background noise should not affect the mean level, rather the precision of the measurement. So it is hard to understand why a significant part of the data was discarded, in order to meet the "best one-day gravity value" (L219). It would be worth to give more arguments for the data selection, e.g. an example where it becomes clearly visible why a significant part of the data was discarded. 

3) I'm confused by the mix of instrumental heights: effective height and top-of-drop. It should be clearly discriminated that the effective height is different from top-of-drop (L273). In particular if the (applied) vertical gradients have changed over the years, the gravity value in the effective instrumental height would be the best choice to document the results. Also, it is advantageous to define a common reference height close to it (as specified in the caption of Fig. 5). Table 1 provides the gravity values at top-of-drop. If all measurements were evaluated with the gradient specified in chapter 4 it is not a serious problem. Nevertheless, I propose to rework Table 1 to either use the effective instrumental height or a common reference height close to it, e.g. the values at 1.22 m used for Figure 5.

4) The site relocation is mentioned in the Conclusions with reference to Section 2. However, in Section 2, both sites are described but not the aspect of relocation. Please add some details why it was necessary to relocate and what might be the differences in the local conditions. Moreover, the trend essentially depends on this "jump", so the impact of the relative survey would be worth a critical review, also considering the overall stability documented by spirit leveling. Therefore, I suggest to additionally evaluate trends before and after changing the site. 

Please find some minor comments below: 

L121: Is there really no significant height difference between Ref01 and Ref02? What is the uncertainty of the difference of 0.010 mm? 

L154: The FG5-206 was upgraded to FG5X. Please mention this, discuss the impact on the stability of the meter and also cite [1]

L159: Please check the mass of the equipment. To my knowledge it is nowadays about 300-350 kg and was not more than 500 kg.

L186: Please define the unit microgal only once. 

L196: You are referring to a very early paper by Carter et al. (1994). It is worth to be mentioned but is there indeed no more recent publications worth to be cited addressing the requirements to detect geodynamic trends, e.g. [2] or work for Fennoscandia?

L207: You are only speculating about potential differences between both FG5. Please use comparison measurements to proof this (see my general comment 2) above) 

L218: How can it be consistent to remove measurements just because of enlarged scatter of observations? (see comment 3) above)

L246: You mention ocean tide loading up to 30 µGal. Is the tide model obtained from the CG3 record published? Otherwise, I suggest to publish it here in an appendix to allow application for future measurements or comparison with ocean tide models. 

L272: This sentence is completely wrong. The effective instrumental height is not identical with top-of-drop! Please also check [3]. Was only the gradient changed or was the whole measurement reprocessed? If only the gradient was unified, then the gravity value should be first transferred with the original gradient to the effective instrumental height, and next with the actual gradient to the common reference height (or whatever reference is used). 

L278: A reference height of 0 cm is given. It is impossible to measure directly on the floor or on the marker. Please correct. 

L278: What was the scatter of the individual gradient measurements? It would be interesting to document possible temporal changes. 

L319: Could you please also check the more recent paper [4] and the respective comparison reports to evaluate the stability of both FG5? See my general comment 2) above.

L354: The list of instrumental errors contributing to the uncertainty budget seems not applicable to absolute gravimeter in all cases: There is no "difference in vacuum condition". If the vacuum is insufficient, no measurement can be done. Also, there is no phase response/transfer function known to me that influences the absolute gravity measurement, apart from non-linearity in the fringe detection. Please update/further elaborate and provide references. 

Section 5.2.1/Figure 6: If the measurements after the 2016 would be neglected, would the trend still be significant? Please explain how the time dependent error estimate was calculated.

Code availability: For the ETERNA software, the repository [5] at KIT should be cited instead of the link to the GFZ publication. 

---

References

[1] T M Niebauer et al 2011 Metrologia 48 154, doi:10.1088/0026-1394/48/3/009

[2] Pálinkáš et al Acta Geodyn. Geomater., Vol. 7, No. 1 (157), 61–69, 2010

[3] Pálinkáš et al 2012 Metrologia 49 552 doi 10.1088/0026-1394/49/4/552

[4] Pálinkáš et al J Geod 95, 21 (2021), https://doi.org/10.1007/s00190-020-01435-y

[5] https://publikationen.bibliothek.kit.edu/1000151532 DOI: 10.35097/746