The manuscript presents an interesting new dataset and related discussion concerning CL36 and REY analyses on a limestone bedrock fault scarp near Sparta, Central Greece. The Authors describe in details the new advance in methodology for Cl36 cosmogenic dating in the last years. Their results indeed allows to better constrain earthquake slip for recent seismicity along the Sparta Fault, including the historical earthquake that destroyed Sparta in 464 B.C. In fact, this is relevant in terms of exposure dating methodology. The Sparta Fault was already sampled for 36Cl at two sites, but previous papers were not able to obtain the resolution achieved by the Authors of the manuscript. Interestingly, the Authors clearly show that REY analyses are of little use for the Sparta Faultbedrock scarp, due to the presence of quartz and other silicate components in the breccia matrix included in the fault damage zone. This is a major contribution from the presented investigations.

My main comment concerns the general use of exposure dating for understanding the paleoearthquake history on a limestone bedrock fault scarp. In fact, the literature shows that defining individual earthquake ruptures using 36Cl on bedrock fault planes might be quite difficult. For instance, during the 2016 earthquake sequence in Central Italy, the main limestone scarp along the Vettore Fault ruptured following both the August 24 and the October 30 mainshocks, with a maximum slip of 20 cm and 210 cm, respectively. Clearly, exposure dating will not be able to discriminate among these two events, as already discussed by Bubeck et al. 2015 (Bubeck, A., Wilkinson, M., Roberts, G. P., Cowie, P. A., McCaffrey, K. J. W., Phillips, R., & Sammonds, P. (2015). The tectonic geomorphology of bedrock scarps on active normal faults in the Italian Apennines mapped using combined ground penetrating radar and terrestrial laser scanning. Geomorphology, 237, 38-51) and Cowie et al. 2017, for instance. This point should be clearly discussed in the Introduction, and taken into account in the Conclusions. The series of 4 strong, M7 paleoevents interpreted by the Authors is therefore affected by intrinsic problems of resolution; the 4 identified strong events might include several smaller , M6 to 6.3 (for instance, the Mw 6.3 L'Aquila eq in 2009 generated max displacement of ca. 10 cm), seismic events. M6 to 6.3 is a very severe earthquake for an ancient town like Sparta in 464 B.C., but also for the modern town of Sparta today. Therefore, conclusions in terms of seismic hazard based on the results collected by the Authors of this manuscript must be treated with care. 

Some specific comment are included in the attached annotated manuscript.

Goodfellow et al. present a study on the Sparta fault in Greece. The Sparta fault is a normal fault that has been studied previously by several authors that Goodfellow et al. cite, that have shown evidences that this fault is the source of the M~7 earthquake occurred in 464 BC. Goodfellow et al. use 36Cl cosmogenic dating and rare earth elements to study the paleoearthquakes on this fault. They collect samples on the exact same site that was previously studied by Benedetti et al. in 2002 using also 36Cl cosmogenic dating. In 2002, this study was pioneering the use of 36Cl in situ cosmogenic nuclide as a tool to unravel a fault seismic history by dating the exposure of the fault plane. Since then, many papers and groups have applied this approach (e.g. Schlagenhauf et al. 2011, Benedetti et al. 2013, Tesson et al. 2018, Cowie et al. 2017, Goodall et al. 2020, Iezzi et al. 2023)

I don’t think the results of this paper are worse publishing at this stage and I think it needs a serious rewriting and reinterpretation of the datasets with a rigorous treatment specially concerning the modelling of the 36Cl dataset.

Here are the main concerns and comments that I have on the paper.

First the message convey by the manuscript is confusing and the scope of the paper is not clear. Goodfellow et al. have sampled for cosmogenic dating the exact same site as in Benedetti et al. 2002, they argue that their aim was to understand why the 464 BC was not found in the 36Cl profile made and analysed by Benedetti et al. in 2002, they also want to “redate the paleoseismicity” and finally complement by rare earth elements the paleoseismic history. However in their conclusion, while they use the 36Cl record to derive seismic events, they argue that the 36Cl concentration in the profile might vary with mineralogical variations and thus interpretation in terms of exposure duration might be difficult…this is contradictory, and I would suggest the authors to better explained what they mean and strengthen the scope of the paper to either a methodological paper or a paper on the paleoseismicity of the Sparta fault. As it is, none of their conclusions appear convincing to me (see details below). Also according to previous studies using 36Cl as a paleoseismological tool, mineralogical variations in the fault scarp are taken into account in previous papers with the chemical composition of each sample (see details in Schlagenhauf et al. 2010). In the paper I could not understand why they use a mean composition and not the chemical composition of each sample to avoid this problem.

My second concern relates to the dataset treatment and the associated modeling. Goodfellow et al. have sample for cosmogenic dating the exact same site, but they sample only half the profile that was previously sampled by Benedetti et al. 2002 and have mixed several types of samples with different thickness which affect the 36Cl concentration  in each sample and could affect the comparison in between samples. This is not discussed. Moreover, they made several unexplained adjustments that are not justified, such as increasing arbitrarily the concentration by 5% of the depth core profile. More importantly, the modeling of the dataset, which is also based on the Schlagenhauf et al. 2010, is made without explaining several assumptions that are crucial to the results. The 36Cl modeling is not well explained and treated in the paper as straightforward. In the model used by Schlahgenhauf et al. the choice of the discontinuities in the 36Cl profile is crucial and allows determining the number of events. Here, the authors do not explained why they chose to model their datasets with 5 events, what results would yield others models with less events ? would the RMS or AIC be better ? how do they define the position of their events? Others models to unravel the seismic history with 36Cl dataset on a fault plane have been published so far (e.g. Tesson and Benedetti 2019, Tikhomirov et al. 2019, Beck et al. 2018, see also Iezzi et al. 2021) but the authors do not cite them and does not explained why they chose the one published in 2010.

Besides, the production rate they chose for Ca spallation is arbitrary, on which publication is it based ? They cite Lifton et al. 2005 but this is not a publication related to the production rate of 36Cl from Ca. This production rate is almost 18% higher than the one used by Benedetti et al. 2002, this has obviously an effet on the age of the yielded earthquakes. The authors do not discuss this aspect, but argue that their results allow finding the 464 BC event, but would a different production rate yield the same result for the Benedetti et al. record ?

Concerning the rare earth elements and Yttrium treatment, the authors fail to cite the most recent papers, and in particular Moroaetis et al. 2023 that have discussed specifically the mechanism of REE-Y impregnation on active carbonate normal fault scarps, moreover the data treatment is not well explained and the discussion thus poor and not up to date. See also Bello et al. 2023.

Finally, the presented fault geometry is oversimplified compared to the previous publication of Armijo et al. 1991. It is a pity that considering the means we have now to map the fault trace the authors did not take the opportunity to refine the initial map made by Armijo et al. also because in Benedetti et al. paper the authors claim that the 464 BC could not be seen at Anogia because it might have bypassed the main scarp. It would have been good that the authors explore this explanation, especially since they state it is one of their aim.

Thus the yielded results and conclusions are not convincing, I don’t see what the paper brings in terms of new results or approach, since the paleoseismicity results appears similar to the conclusions of Benedetti et al. 2002. At this stage, one interesting aspect is that they allow a unique comparison of 36Cl dataset acquired at the same site with almost 20 years of difference. The comparison is outstanding since the difference in the 36Cl concentrations is of at most 19%. This appears exceptional considering that the chemistry extraction and the measurements were made in different labs, with different methods and measured in two different AMS (see Merchel et al. 2011 for an interlaboratory comparison of 36Cl). I am not sure many Quaternary geochronological dating techniques would yield such result. For the cosmogenic nuclides community this might be an interesting result.

Review of “Paleoearthquake reconstruction on an impure limestone fault scarp at Sparta, Greece “by Bradley W. Goodfellow, Marc W. Caffee, Greg Chmiel, Ruben Fritzon, Alasdair Skelton and Arjen P. Stroeven (egusphere-2023-1585).

This manuscript presents a reappraisal of the earthquake history along the exhumed Sparta fault scarp using Cl36 data profiles and rare-earth elements and yttrium (REY) data. The study attempts to combine the Cl36 dataset of Benedetti et al. (2002) with new Cl36 data including some collected in the colluvium wedges at Anogia, with added REY data to refine the composition & help the discussion of the paleosismology along that fault segment.

First, it’s important to present the input of the original study which presented the analysis of two sets of continuous exposure history using 36Cl profiles sampled on the ~10m high fault scarp on 2 fault segments at Anogia (64 samples) et Parori (65 samples) and their model allow for the identification of the 464 B.C.E. earthquake that destroyed Sparta at their Parori site together with four additional earthquakes that ruptured the Sparta fault in the last 13 ka with similar co-seismic slip of ~2 m and with time intervals ranging from 500 yr to 4500 yr (Benedetti et al., 2002).  The 464 B.C.E. earthquake was not resolved from the modelling of the original 36Cl dataset of the Anogia site by Benedetti et al. (2002) and several parameters such as the inheritance and erosion were neglected in the original analysis, the geometry was simplified and the production rate of Cl36 in Calcium was actively debated at the time. Therefore, there are room for a reappraisal of older dataset to help to better assess the seismic history of normal fault using 36Cl data.

The present study completed the dataset (new 36Cl and REY data) and refined the 36Cl modelling using the codes of Schlagenhauf et al. (2010) that can consider all the production pathways, a complete geometry & composition of the fault scarp and colluvium wedge and define the level of inheritance and/or erosion. The reappraisal of the Cl36 modelling at Anogia allows now for the elusive 464 B.C.E. earthquake to be also resolve together with 4 to 5 earthquakes, co-seismic slip ranging from 60 to 240 cm.

Overall, the present manuscript more less concurs with the conclusions of the original paper. Yet, at this stage, it is difficult to assess the reappraisal and improvement made on the 36Cl analysis at Anogia and I would therefore recommend major corrections to be done before considering any publication. It is unfortunate that the REY data has not help much as with other studies (i.e Manighetti et al., 2010), so I will focus my remarks and questions on the cosmogenic data analysis.

            The revised 36Cl modelling is only apply to the one of original profile, it might help to revised both sites for the discussion using the same production rate & codes??

            Scaling samples that have different geometry, thickness and therefore attenuation must be discussed and scaled properly.

            Combining samples from different profiles that have different fault geometry, erosion, inheritance and potential shielding must be discussed.

            There is no clear comparison of the results of the different models, a figure would help clarify (height versus time, using the co-seismic slip of each earthquake estimated age).

            The modelling of the 36Cl data does not appear to include the contribution of all the pathways despite being integrated in the codes of Schlagenhauf et al. (2010). 36Cl is produced by spallation of K, Ca, Ti and Fe; slow negative muon captures by K and Ca; and low-energy (thermal and epithermal) neutron capture by 35Cl and also not integrated in the modelling, composition data are available for the original dataset (see appendix of Benedetti et al. 2002) and the new 36Cl. That will affect the model ages of the different earthquakes and there is no need to average over the profile if the data exist for each sample.

            The production rates of Cl36 in Ca, K, Fe and Ti used in the study need a proper discussion and be better justified. Several aspects are typically discussed, the production rates of the different targets, the scaling factors used, the atmospheric model, and the geomagnetic database used to correct for the temporal variation of the production rates. The paper should be clearer on the topic, the scaling of solar modulation and long-term uncertainties defined by Lifton et al. (2005) is not a production rate paper. The production rates of the main targets producing Cl36 have also been scaled for the different scaling scheme in the CRONUS-Earth effort (see Marrero et al., 2015, even if the abstract only present the LSDn solutions). It seems strange to work on a reappraisal of a dataset without using the up-to-date production rates or at the very least present a comparison of the modelling results using different production rates.

Additional remarks are made in the pdf.