This is a well-organized and thorough case study, which examines how granary records may reveal the impacts of climate variability on harvests. To my knowledge, it is the first study of its kind, and the methods developed here could be useful in other regional contexts as well.

Before publication, I would like the authors to address the following:

First, since past studies have usually examined the impacts of climate variability on harvests by looking at grain prices, could the authors specifically address what the study of granaries contributes to our understanding of climate/weather and harvests that the study of prices does not? For example, does this study indicate more vulnerability or less vulnerability of farmers’ harvests and livelihoods to climate variability than the grain price data indicate? Does it indicate the same or different patterns and trends in climate/weather impacts on harvests?

Second, the study should discuss (at least briefly) how granaries might have interacted with grain markets or with farmers’ behavior. Currently, the study seems to assume that farmers used the granaries only as intended—that is, they borrowed when times were bad and paid back when times were good. But might farmers have used the granaries in other ways? Did they come to rely to some extent on the granaries and take greater risks? Were the grains that were grown from seed corn taken from the granaries used primarily for consumption, or were they sold to buy other food, pay rent, etc.?  Could farmers have tried to borrow seed to expand production in expectation of higher prices? Might farmers have willingly maintained a debt to the granary in order to plant or sell more grain and thereby improve their financial condition during average or good years? If there is no evidence that farmers did any of these things, then it would help to add a couple of sentences explaining this.

Third, the visualization of the results (i.e., presentation of the various indices) could be improved for greater clarity and utility. The way results are presented in tables 2 and 3 makes it very difficult to identify the evolution of good and bad years, to identify periods of frequent or consecutive harvest failures, and to compare the performance of different grains to one another. It is also hard to judge the reliability of good/bad harvest determinations based on data from only 1-2 domains. (E.g., how should I compare the third worst year based on only one domains’ data with the fourth worst year based on seven domains’ data?) Therefore, the ranked lists of the worst and best harvest years for different grains were not especially helpful.

It might have been helpful to see a single timeline showing all index values for all grains each year (or at least whether each index was >1, 0.5–1, -0.5–1, <-1). If such a figure would be too messy, then the authors should determine some other appropriate way to visualize the data so that the evolution of good and bad years, the periods of frequent or consecutive harvest failures, and the comparative performance of different grains is easier to see.

I would have also liked to see some brief discussion of the correlation among harvests for different grains in the same years, and whether there were lags or autocorrelation in the indices (possibly indicating persistent agricultural problems or economic hardships following especially bad harvest years).

Fourth, when discussing the relationship between climate/weather and harvests, the paper considers only those years which had a good or bad harvest and then examines the range of weather conditions for those years. This is useful, but this approach can only investigate one kind of causal relationship: how necessary were certain climate/weather conditions for a good or bad harvest [i.e., the probability of good or bad weather given a good or bad harvest, or p(weather|harvest)]. Sometimes, we might want to know how sufficient certain climate/weather conditions were for a good or bad harvest instead [i.e., the probability of a good or bad harvest given some good or bad weather, or p(harvest|weather)]. In fact, for some kinds of studies, the question of causal sufficiency will be more important than causal necessity. For example, it would be interesting to know whether or not very cold wet summers consistently brought bad harvests (a question of causal sufficiency), even if most harvests failures might have happened for other reasons (a question of causal necessity). To address causal sufficiency, we would have to start by examining the years with good or bad climate/weather conditions and then see how often they brought good or bad harvests, which is the reverse of the current discussion in section 4.4.2. This should not require a lengthy analysis, but only a test of the most likely patterns, or even some simple counting: e.g., “of x summers with below average temperature and above average precipitation, y were followed by a poor harvest, as indicated by z indices.”

General comments

Thank you for the opportunity to review this manuscript. To my knowledge, this is a first in a European (if not global) context and introduces an important new proxy for the documentary analysis of historical grain storage and hence climate variability.

The paper is beautifully written, clearly structured, contextually rich and well supported through references to previous studies. The historical context for the data is recorded clearly and succinctly, and at a level suitable for an international audience. Other than the specific point below, the methodology is clear and straightforward. The results are well presented and explained, and the statistical analysis clear and appropriate.

I really value the transparency over data availability from granaries shown in Figure 5, the corroboration of good and bad harvests from more general documentary evidence in section 4.4.1, and the comparison with wider regional studies. The discussion of the links between grain harvests and climate is strong, reflecting the multiple possible causes of a good vs. bad year. The reflections on the challenges of using grain harvest data are discussed honestly and openly. In short, aside from one relatively minor comment, I have no hesitation in recommending the manuscript for publication.

Specific comments

Lines 168-173. I’m interested to hear more about how well this methodology deals with multi-year (as opposed to single-year) poor harvests. Did multi-year runs of crop failure occur during the study period and, if so, what happened as a result in granaries? I notice at least one instance of this mentioned in the results. There is also mention in Lines 444-446 of granaries being empty – an obvious limitation of the proxy – due to factors in addition to harvest levels.

Line 525. Anthropogenic would be better than man-made.

General Comments

  This original study is very interesting because it shows how grain records (rye, barley, and oats in this case) from public granaries may be used as proxies for evaluating grain harvests in relation to weather and climatic patterns. It is a (first) and promising attempt using granary data for historical-climatological research.

  This paper is based on a set of unpublished historical granary data coupled with long-term series reconstructed for three basic climatic variables already available (temperature and precipitation series + scPDSI) and well-studied for the Czech Lands territory. In addition, the few existing biases for their use are also explained (ligne 481 to 487).

  The choice of data processing (selection of four annual grain variables) statistical analyses (Spearman's rank correlation coefficient) is wise, easy to understand and clearly describe. There's no excess statistical analysis, which is a very good thing.

  The use of four different types of weighted grain indices is a good way of getting around the problem of missing data. This methodology could be used as an example when analyzing similar data available throughout Europe, and for other crop types.

  In the section "4.4.1 Extreme harvest years and documentary data", for the early 19^th century, documentary weather data for the selected years of poor grain harvests & of good grain harvests are very impressive and precise! The role of climatic factors in a good or bad harvest is thus easy to identify.

  Grain data from public granaries therefore appear interesting to identifies bad or good grain harvests in relation to weather and climate patterns, but could they also be used to identify other specific environmental factors such as pest insect attacks (e.g. locusts or beetles) in crops? Of course, these phenomena are often influenced by climatic conditions but this aspect is not mentioned in the article.

Ligne 540 to ligne 552 : This paragraph about other important non-climatic factors for years of bad and good harvests, especially conflicts and wars at that time, is particularly welcome, as it avoids the (potential) criticism of an overly deterministic vision.

  The change in crop type (here the increasing importance of potato growing) as a factor influencing cereal production should be further explored, in connection with the evolution of cultivation methods in the early 19^th century. The Industrial Revolution in the 19^th century brought technical and technological advances which had an impact on the development of arable farming. Scientific advances, such as mechanisation and artificial fertilizer improved yields.

  Ligne 555 : "Specifically, it identifies bad or good grain harvests in relation to weather and climate patterns, situating them within the broader context of the Czech Lands in the late 18th and the first half of the 19^th century", it’s ok ! But, as the dataset in the article concerns only the Czech Republic, the data is not combined comparatively across wider areas, I'd suggest changing the title slightly from "for historical climatology" to "for historical climatology in Czech Republic".

  Maps, charts and graphs are clear, well presented and easy to interpret. The English is very good, as is the style.

  The article is perfectly suited for Climate of the Past and deserves to be published with just a few minor revisions.

Specific comments (about the references used):

  Various works are cited for different European and Asian countries (China), but for France, only the work of Kaplan in 1977 is cited.

The question of grain harvests and grain management has, however, been discussed at length from various angles in the masterpiece of the French historian Jean Meuvret, published in 1977, "Le problème des subsistances à l’époque de Louis XIV" I. La production des céréales dans la France du XVII et du XVIII siècle. & II. Le commerce des grains et la conjoncture (J. Meuvret, Mouton & Cie and École des Hautes Études en Sciences Sociales, Paris, 6 vol.). These works cover much of the 17^th and early 18^th centuries, and regularly refer to granaries, so they could have been cited. However, the authors may not be familiar with these works, which are unfortunately only available in French, not widely distributed and not easily accessible.

  There's also Abbott P. Usher's book : "The history of the grain trade in France 1400-1710, Cambridge Harvard University Press, 1913" (book in open access), which refers to the granaries, but this French-centric study is now a little outdated, and we prefer to use J. Meuvret.

Technical corrections about the bibliography: 

For France, the work of "Gast, M. and Sigaut, F. 1979" appears in the references but not in the text. A correction is therefore necessary.