Dear authors,
many thanks to your responses to two reviewers. Please prepare a revised version of your manuscript in which you will address all suggestions and queries of both reviewers. Moreover, I would like to see in your Introduction also an additional paragraph in which you will present rogation data (ceremonies) as one of documentary data/sources used in the Mediterranean in a broader context of other documentary data related to drought/documentary-based drought reconstructions in Europe (including quotations of some relevant papers).
With many thanks for your effort I am looking forward to your revised manuscript,
best regards R. Brazdil

Dear author,
your responses to referees and your revised manuscript were sent for re-review to both referees. While the one referee was already satisfied with your changes, the second referee expressed still serious doubts related to the physical meaning of the clusters and validation of the series in your paper (see below). Please be so kind to work with these comments and prepare corresponding changes in your manuscript to incorporate them into revised version of your manuscript. Besides it, please respond separately to every points raised in the corresponding review. I hope you will be able to use these comments for further improvement of your article to create conditions for a subsequent acceptation of your revised paper for publication in Climate of the Past.
With many thanks for your kindly effort and best regards,
Rudolf Brazdil

Referee comments:
I think that the authors have addressed some of the questions raised in my previous review and I recognize this effort. However, I think that some of my concerns have not been properly addressed.
- Physical meaning of the clusters. I think that the authors fail in providing a convincing explanation for their proposed clusters. In their reply they acknowledge that the differences between Lleida and Cervera are not due to climatic factors, but rather to the higher resilience of Lleida to precipitation shortage due to the irrigation system. Thus, rogations in Lleida should only detect the most extreme events and their sensitivity to drought is different. However this different sensibility to water shortage is not taken into account in the clustering, when all the series are treated equally. A physically consistent explanation of the inclusion of the Teruel series in the Pyrenees cluster is not provided in the reply. According to figure 4, Teruel is only significantly correlated with Huesca, but the explained variance is lower that 4%. No other significant correlations are found for Teruel within this cluster. Consequently I still think that the clusters have no climatological support, which rests validity to the rest of the analysis.
- Validation of the series. I acknowledge the effort in validating the series with the inclusion of the Barcelona instrumental series, which shows an overlapping period with the rogation series. However, I do not think that the authors have provided a fair comparison since they are comparing a local series (Barcelona), with a Regional average (DIMED). This has several problems whihc have been treated in model verification (see for instance Gilleland et al 2009 and references therein). Since the Barcelona rogation series is available, they should provide the comparison of the two local series during the overlapping period if they want to provide a convincing validation.
So, according to these points, the clustering, which is the most original part of the paper, should be reconsidered to get a partition based on the climatic signal in a convincing way. On the other hand a fair validation should be provided.
Minor point
Language is still an issue and should be checked carefully. For example line 193 of the new version: ‘461 years of continues data’
Ref
Gilleland et al (2009)Intercomparison of Spatial Forecast Verification Methods . Weather and Forecasting. DOI: 10.1175/2009WAF2222269.1

Dear authors,
as I have informed you in mean time, due to contradictory opinion of two original reviewers about your paper, I decided to invite two additional reviewers. In attachment you will find now three reviews - the first one is from one original reviewer (as a response to your revised paper based on former reviewer comments), two following are from the new reviewers. As you will see, in reviews are addressed several very important comments which should be - besides others - particularly reflected in your responses to reviewers and in the new revised version of your article. I hope you will work very carefully with these comments to prepare a new revised version of your article, in which comments of reviewers will be well reflected. Please do not hesitate to contact me if you need any additional explanations
With expectation of your responses to comments of reviewers and best regards,
Rudolf

Dear Dr. Tejedor,
we obtained two reviewes of your revised paper. While referee #4 ask only for minor revision based on her/his expression, referee #3 ask for major revision. Please be so kind to revise your paper based on comments in both reviews for further consideration. I believe you will be able to work with all critical comments and further improve your paper.
With many thanks for your effort and best regards,
Rudolf Brazdil

Comments of referee #3:

Rogations are a ‘cultural’ proxy. Therefore, they are affected by a certain degree of subjectivity, due to the perception of people about hydroclimate events. In consequence, the analysis must be cautious, taking into account their historical and sociological nature. My main criticism to studies based on rogations is that they often ignore this problem, perhaps because attempt to reconstruct long and continuous series of droughts (and floods). Authors have solved some of my doubts, but, in my opinion, certain problems persist, in particular in relation to the calibration using the overlapping period 1786-1899. I can accept that in 1787 “rogations were still deeply established in the society”, but authors say that historical process after 1834 “affected the occurrence of rogations and respective records on documentary series of public institutions”. If this is the case, it would be more appropriate to use the period 1787-1834 for calibration purposes. If early instrumental series are not sufficient, the best reliability test is comparing the occurrence and severity of droughts with evidences from other independent documentary data or proxies.
I suggested a seasonal study, including autumn months (rainfalls in October and November may be important for sowing), but it seems that authors have not considered this idea. Well, I can accept the definition of an annual index. However, the seasonal treatment would may shed light on the ordinal scale introduced (levels 0 to 3). For instance, when authors affirm that “a dry winter can give any ‘preventive’ rogation of level 1 in early spring… if drought is persistent during the following spring, rogation ceremonies convoked by institutions are increased to higher levels”. In their letter they say that “if there was a drought of level 2 it is because those types of ceremonies of level 1 did not work”. Must we infer that a level 2 rogation was always preceded by a level 1 rogation? The interpretation of levels seems be influenced by the cumulative character of droughts. In that case, the ‘annual’ DI values obtained as the weighted average of the number of level 1, 2, and 3 rogations recorded, are misleading. It would be more appropriate to consider for each year the maximum level recorded, because minor levels are in some sense subsumed in a level 3 rogation.
Finally, a comment on the hydroclimate responses to volcanic forcing. According to Fischer et al (2007) there is a tendency to anomalously dry conditions over the Iberian Peninsula in the winter of years 0 and 1 following volcanic eruptions, resembling a positive phase of the North Atlantic Oscillation. But the behaviour in spring is different, showing significant wetting over the western Mediterranean 2 years following an eruption (Rao et al., 2017). These authors explain that volcanic forcing may modulate spring and summer climate by stimulating a negative East Atlantic Pattern response. Therefore, we have a different behaviour in winter (dry, positive NAO) and spring (wet, negative EAP). This seasonal detail is not commented by authors. These results support my idea that a seasonal study is more convenient, not only to convert qualitative data to ordinal indices (levels), but also to interpret results in terms of natural radiative forcings.

Comments of referee #4:

I find that the authors have done an excellent job in addressing my points. The paper is an interesting and important contribution that shows the value of and challenges of using documentary sources – in this case information from rogation ceremonies. The paper now reads well and I very much like the discussion. It will be an informative resource for future work and I commend the authors for so fully discussing the pros and cons of their work. I recommend accept with some very minor points below that I picked up upon reading. The main one is that the abstract could be shortened a little, there is perhaps too much detail in there that distracts from the key results. The final conclusion paragraph could also be strengthened by removing the first sentence.
Line 26, standardized precipitation
Shorten abstract a little.
Line 408 – in this paper
Line 430, don’t start with Finally, instead start with We found…
Throughout – drought indices rather than Drought indices
Line 539, it is because
Line 549, in this paper
Lines 647 – 650 – I would recommend delete this sentence and start with ‘Finally, the historical data from rogations….

Dear author,
I would like to inform you that after the last revision the paper is acceptable for the publication in the Climate of the Past. Congratulations! But please take in account in the final version of your manuscript several technical changes which are specified in the attached pdf file.
I would like to express you our thanks, that you have contributed with your paper to the drought special issue.
With best regards,
Rudolf Brazdil