A Prequel to the Dantean Anomaly: The Water Seesaw and Droughts of 1302-1307 in Europe

The cold/wet anomaly of the 1310s («Dantean anomaly») has attracted a lot of attention from scholars, as it is commonly interpreted as a signal of the transition between the MCA and the LIA. The huge variability that can be observed during this decade, similarly with the high interannual variability observed in the 1340s, has been highlighted as a side-effect 10 of this rapid climatic transition. In this paper, we demonstrate that a multi-seasonal drought of almost two years occurred in the Mediterranean between 1302 and 1304, and respectively a series of hot and dry summers north of the Alps from 1304 to 1306. We propose to interpret this outstanding dry anomaly, unique in the 13th/14th century, combined with the 1310s and the 1340s cold anomalies, as part of the climatic shift from the MCA to the LIA. Our reconstruction of the predominant weather patterns of the first decade of the 14th century from documentary and proxy data lead to the identification of multiple European 15 water seesaw events in 1302-1307, with similarities to the seesaw conditions which prevailed in 2018 over continental Europe. It can be debated to which extent the 1302-1307 period can be compared to what is currently discussed regarding the influence of the Arctic amplification phenomenon on the increasing frequency of long-lasting stable weather patterns that occurred since the late 1980s. Additionally, this paper deals with socio-economic and cultural responses to drought risks in the Middle Ages from contemporary sources and provides evidence that there is a significant correlation between blazes that devastated cities 20 and pronounced dry seasons.


Introduction & State of the Art
While Medieval studies analyzed since decades the reconstruction (Pfister et al. 1998) and the impact of cold events on premodern societies in the context of the Little Ice Age, few papers focused on droughts, notably from economic history (Stone 2014). The so-called 'Dantean Anomaly' has been highlighted since the 2000s by Brown (2001) as a wet and fresh 25 anomaly lasting from 1315 to 1321, and leading to famine over NW Europe (Jordan 1998). This climatic anomaly has been recently described more neutrally as 'the 1310s event' (Slavin 2018). A distinctive '1300 event' has been found in proxy data even around the Pacific rim (Nunn 2007). But the focus was always on the cold and wet character of this decade and historians have clearly been fascinated by continuous rains and their impact on food security, often leading to widespread https://doi.org/10.5194/cp-2020-34 Preprint. Discussion started: 20 May 2020 c Author(s) 2020. CC BY 4.0 License. created seasonal drought indices from the above mentioned narrative sources over the longer period 1200-1400 for all three regions (IT, FR, CE) (see SI 1). As the aforementioned sources and climate historical repositories focus on extreme events, 125 we gave values of -3 (extremely dry) and -2 (very dry) if we had actual indicators for agricultural and meteorological drought (Brázdil et al 2019, 75) or a lack of precipitation over 2 months. We have, however, not applied the category of 'socio-economic drought ' (Ibid.,[75][76], as not all of its indicators are in our opinion specifically related to dry periods. To identify long-term droughts, these drought index values have been accumulated for single years (figs. 8-10); they confirm the extraordinary character of the 1302-1307 drought events. 130 If we focus on the precipitation indices only, we get for IT ( Fig. 2) a dense picture demonstrating a sustained, almost 24 months long dry period that is not at all represented by the PDSI values from the Old World Drought Atlas (OWDA, Cook et al. 2015). The case of Italian OWDA data is special, as only a handful of dendrochronological series from the Alps and The precipitation indices for FR (Fig. 3) are scarce, and yet they show a pronounced drought pattern in the growing seasons of 1304-06. In regard to IT, we lack information on the continuity of this drought over the non-growing seasons (1303/04 to 1306/7). But the general tendency is fitting with available OWDA data, except for the summers of 1293 and 1311.
The most interesting results are the precipitation indices for CE (Fig. 4). They have to be stacked, as CE covers a row of 140 quite different sub-regions, yet the tendency is the same in all of those regions. Furthermore, they are mostly consistent with OWDA data points, and if indices seem to differ (e.g. 1291/92, 1294/95, 1305/06), they provide precipitation data on the non-growing season which is not covered by tree-rings.
If we aim for a qualitative description of weather patterns, documentary data provides a clear, reliable and very dense picture (see SI 2) of meteorological conditions over the period in question. It all started with a rainy summer in 1302 with floods in 145 CE and FR, while the second half of 1302 was already without precipitation in IT. This was followed by a cold winter with freezing rivers (Rhine, Doubs, Adige) and low water levels in CE, FR and IT, while more to the East (Silesia, Russia), the winter was mild and snowless. Spring of 1303 proved cold in CE -and while we have no information on this summer from north of the Alps, IT was hit by continued meteorological, agricultural and hydrological drought the whole year 1303. The following winter 1303/04 was particularly warm in FR and split in CE: warm in its Western part, cold in Bohemia. A very 150 chilly winter with freezing rivers is reported for IT. Spring and summer 1304 were extremely dry and hot in FR and CE, with all signs of hydrological drought. IT saw strong, yet short precipitation events in late spring, interrupting the 13 months drought, and then again a dry summer until September 1304. Once more, a pronouncedly cold winter 1304/05 followed in FR and CE, strong precipitation in early 1305 in IT that continued into summer and a dry period in summer 1305 in FR. The winter 1305/06 was so chilly that the Baltic sea froze over and so did rivers in FR and CE and IT. In FR drought continued 155 into spring 1306, and in CE as in IT winter 1306/07 was again very frosty, later changing to flood conditions. In Eastern CE, drought set in in summer 1307, and a heatwave in FR and IT. https://doi.org/10.5194/cp-2020-34 Preprint. Discussion started: 20 May 2020 c Author(s) 2020. CC BY 4.0 License.

Agricultural production in France and England
From the accounting documentation of the Bresse region (Eastern France), we have reconstructed wheat and wine yields for 160 the period 1300-1330. For each of these reconstructions, raw data of different castellanies, i.e. administrative units under the control of a steward, have been extracted and then compiled in aggregate series indexed on the year 1307. Cereal yields have been estimated from the revenue perceived on seignorial lands located in the territories of two castellanies, i.e. Jasseron and Treffort. This information is indirect, as it refers only to taxes perceived on these lands which were cultivated by tenants, and not to direct indication of cereal quantity harvested each year. Being said, if that induces a limit in the interpretation, the 165 reliability of the reconstructed series is not to dismiss entirely. Wine yield series are much less critical, as the accounters referred directly to the exact amount of wine collected in seignorial vineyards located in four castellanies. Thus, roll accounts allow for a reconstruction of a relevantly detailed chronology of the reaction of the local vineyard toward climate variations. The results show a similar pattern, namely a trend to relatively high yields before 1310 and then a downward trend reflecting 170 the deteriorating weather conditions of the 1310's anomaly. Good harvests, especially for wine, clearly stand out in 1304 and 1305, in response to the successive droughts locally described in Parisian as in Alsatian chronicles (see SI 2). Besides, the plentiful wine harvest of 1304 is confirmed by a contemporary chronicle (Jaffé 1861, 231). In the years 1306/07 vineyard production reaches average values, even if the accounts mention in both years heatwaves in June and/or July. In this case, though, temperatures reached levels so high that it prevented peasants from properly plowing the vineyard on time, which 175 can explain why production is lower than in previous years. In any case, it can be inferred from the accounting documentation of the Bresse that from 1304 to 1307, summer temperatures certainly reached above-average values.
Additionally, the link existing between good wine harvest and warm summer half-year stands out in 1313 too. Accounts mention in that year a summer dryness, which has been benevolent for the vineyard.
Experimental archeology demonstrated the impacts of drought on medieval-style agriculture (Kropp 2019): considerable 180 damage on summer crops, but stable harvests from winter crops. A mixture of different crops guaranteed altogether a sufficient harvest. This can explain the average cereal yields in figure 5, especially in 1304. In contrast, excessive humidity in 1310 and from 1314 to 1316 clearly had a negative impact on the harvest (cereals -20/-40% and wine -80/-60% respectively). Methodologically it is worth noting that cold episodes are mirrored more faithfully through agricultural proxydata than dry periods. 185 We then compared this reconstruction with well-studied and accessible English wheat yields (Campbell 2007). Fig. 6 plots the relationship between cereal productivity in Southern England and the Bresse region (FR), in comparison with East Anglia July-September precipitation indinces reconstructed from local archival sources (Pribyl et al. 2012, Pribyl 2017. From 1300 to 1320 English and French yields correlate significantly (Pearson coefficient r = 0,61). General trends are similar in the two regions, with average or above-average harvests in the 1300s. Moreover, we find a synchronous movement between 1304 and 1306 ( Fig. 6), reflecting the precipitation trend. The low level of precipitation reconstructed in East-Anglia for these three specific years then most probably applies similarly for the Bresse region, which means that this multiannual 1304-1306 drought occurred in a large part of NW Europe.

Identification of meteorological patterns
The proxy and documentary data presented in 3.1 and 3.2 provide evidence for the occurrence of an alternating large-scale 195 weather pattern over large parts of Europe between 1300-1310. The found features are similar to the phenomenon of a water seesaw, as it was recently discussed by Toreti et al. (2019)  Alps. This blocking situation led to the formation of low-pressure anomalies over Northern as well as Southern Europe, with precipitation patterns associated in such a way, that Central Europe suffered a severe lack of precipitation whereas Northern and Southern Europe experienced an excess of precipitation. Thus, in the 2018 case, the water seesaw was positive over Southern Europe and negative over Central Europe.
Similar to what was reported by Toreti et al. (2019), the predominant weather patterns found for the period from 1302-1307 205 also must have originated from certain seesaw constellations and associated patterns in the geopotential height fields. A possible meteorological interpretation of the reported weather patterns (see SI 2) was found, as is illustrated in Fig. 7a-k and described in the following: In summer 1302 ( Fig. 7a), wet FR and CE and dry IT correspond to a water seesaw which was negative over IT and positive over FR and CE. Geopotential anomalies were thus positive over IT and negative over FR and CE. Winter 1302/1303 (Fig. 210 7b) was reported to have been dry and cold over IT, FR and CE, but warm over Western Russia. This situation can be explained by the presence of a blocking large-scale positive anomaly in the 500-hPa geopotential that covered whole IT, FR and Southern CE. A negative anomaly over Northern Europe would have provided such a constellation that warm (and -not reported -likely moist air) would have arrived in Silesia/Russia, as was reported for this season. Thus, a potential seesaw was tipping from the region CE, IT and FR (negative) towards Eastern Europe, i.e. here Silesia to Russia (positive). Due to 215 the lack of precipitation proxies for Silesia/Russia for the period, this likely constellation can however currently not be proven.
The 1302/1303 winter constellation likely continued throughout the whole of 1303 (Fig. 7c). Continued positive 500-hPa geopotential anomalies over IT, FR, and CE caused a long-lasting cold period in spring north of the alps, while the blockage led to a continued lack of precipitation over IT. Also in the following winter 1303/1304 (Fig. 7d), the positive 500-hPa 220 geopotential anomaly must have persisted over IT. However, the reported warm conditions in FR and Western CE indicate that the positive anomaly did extend less toward north compared to winter 1302/1303. https://doi.org/10.5194/cp-2020-34 Preprint. Discussion started: 20 May 2020 c Author(s) 2020. CC BY 4.0 License.
Spring 1304 seems to be a turning point for the water seesaw constellation (Fig. 7e). The dryness of FR and CE and the wetness reported for IT let it appear likely that the positive 500-hPa geopotential anomaly moved toward FR and CE (similar to what was reported by Toreti et al. 2019 for the 2018 drought), enabling precipitation systems to reach IT from the 225 southwest via the Western Mediterranean. The dry summer reported for IT following the wet spring could have been caused by a slight positive 500-hPa geopotential anomaly over IT during this time (Fig. 7f), but in this case, summertime precipitation is rather unlikely anyway in its subtropical climate.
The weather reported for winter 1304/1305 gives a clear indication for the presence of a large-scale positive 500-hPa geopotential anomaly over Northern Europe and a negative 500-hPa geopotential anomaly over Southern Europe (Fig. 7g). 230 Cold air masses from Eastern Europe were reported for the whole of CE and FR, while IT was reported to be wet. This constellation continued likely during summer 1305, with a slightly increased negative anomaly over eastern CE that allowed normal temperature and precipitation conditions over this region (Fig. 7 h).
Cold air masses from Eastern Europe were reported for the whole of CE, FR and IT in the winter 1305/1306, which must have been caused by yet another large-scale positive 500-hPa geopotential anomaly over Northern Europe (Fig. 7 i). A 235 negative 500-hPa geopotential anomaly can thus be expected to have been present over Northern Africa. The dry air from Eastern European masses led to the reported dry conditions over FR. That situation likely continued until the winter of 1306/1307, when drought and low temperatures were reported again for CE and IT (Fig. 7 j). The reported increasing flood conditions in IT in spring 1307 can be explained either by melting snow or by a slight movement of the positive 500-hPa geopotential anomaly towards the north, allowing precipitation systems to reach IT via the Mediterranean. 240 Finally, the CE drought and heatwave in FR and IT is likely a result of a continuing positive 500-hPa geopotential anomaly over CE (Fig. 7 k). In its center, dryness prevailed while on the western (FR) and southern (IT) borders hot air was advected from Africa.

Correlation of drought periods and city fires
Medieval city fires are a topic touched upon mainly by cultural history (Jenkrift 2003, 83-100;Riegg 2003;Wolf 2015;245 Wozniak 2011245 Wozniak , 2015. But a close connection between drought and fire has been made plausible for extreme years like 1540 Wetter et al. 2014;Mauelshagen 2010, 127-129). Nevertheless, as a general phenomenon this has been put into question (Zwierlein 2011, 102-110), although the latent fire risk of wood-based pre-modern buildings with open fires to heat, cook and provide light is more than obvious (Bitterli 2015;Contessa 2000, 16-18). Already contemporaries saw a close connection between drought and blaze: "Many cities were consumed by domestic blazes because of the drought and sterility 250 that prevailed in this year" (Wattenbach 1851b, 641).
We have been comparing accumulated drought indices, yet distinguishable by seasons, with the number of blazes we could take from the archives of societies (see sections 2.1, 2.3 and 2.4). A peak in blazes around the 1302-05 drought is visible for IT (Fig. 8), with a significant correlation of droughts and blazes (r = 0,346) over the whole period. The same peak is visible for FR in 1306 ( Fig. 9) with an even higher correlation of fires and droughts (r = 0,657) over the two centuries. The correlation of drought years and blazes in CE (Fig. 10) is still significant (r = 0,379). Furthermore, we have suspected from these results (Figs. 8-10) that the probability of a blaze might lag by one year the drought event, as wooden structures had dried over long periods and might ignite more easily even with a temporal distance to the drought. We have then tested in a cross-correlation if this assumed connection existed in our data and we found a very significant correlation regarding a one year lag for FR (r=0,83) and a signifcant one for IT (r=0,59), but none for CE (r=-0,167). The differing results in the case of 260 CE can probably be explained by the non-critical use of documentary data in the ‚Deutsche Städtebücher', while relevant blaze information was validated by historical source-criticism for IT and FR.
As the OWDA information for CE (here broadly defined as a rectangle from the Rhine to the Alps to Novgorod in the East and all of Denmark plus Southern Scandinavia to the North, i.e. 47,34° N-58.69° N, 7.52 ° E -30.88°E) is relatively dense and reliable, it has been combined with available information on blazes from documentary data. Yet, the correlation of low 265 PDSI values and blazes in the same years ( Fig. 11) is very weak, if not nonexistent (r = -0,06).
What is surprising, is the totally different outcome of the correlation between droughts and blazes for CE, calculated one time from the OWDA-data -with a very bad correlation -and the other time from the drought-indices based on written sources, where the correlation is much better (respectively r = -0,06 and r = 0,379). The data for blazes stayed the same in both calculations. Here, the discrepancy between the reconstruction from the OWDA on the one hand and the drought 270 indices, on the other hand, comes into play: The OWDA (even more so the maps provided with it) shows a lot more dry periods than documentary data, that mirror mainly outstanding ones. Regarding the misleading picture the OWDA gives for Italy (in comparison with a reconstruction on the basis of written sources, see 3.1.), a general comparison with reconstructions from written sources would be advisable, especially for those regions and periods where there is a good basis of written sources and where the amount of dendrological data is at the same time rather low (like for Italy). 275

Discussion
Finally, we want to highlight aspects of societal impacts and adaptation measures to drought beyond the classical rogation ceremonies and other religious processions (overview Brázdil 2020), especially as we have only one example for them from IT: In May 1303 there were processions for rain in Parma, that indeed 'provoked' a one-day rain (Bonazzi 1902-04, 84). But there is more to find about how droughts impacted medieval societies, how they were perceived and how contemporaries 280 reacted.

Drought and infrastructural responses
Italy provides a number of infrastructural responses to the drought experience of 1302-04. In 1303, the city of Parma built a new, larger and deeper fountain (Chron. Parm., 86). Something comparable happened in the Tuscan city of Siena, settled away from any larger watercourses and traditionally struggling with water scarcity. Not surprisingly, the century-old myth of an 285 underground river below the city, the so-called Diana, was still alive in Siena. In April 1305, one year after the end of the https://doi.org/10.5194/cp-2020-34 Preprint. Discussion started: 20 May 2020 c Author(s) 2020. CC BY 4.0 License. drought, documentary evidence reveals that the city council actively searched for this underground river by digging in a local church (Bargagli Petrucci 1903, II, 20). An even more real-time response can be found for 1303: In spring, when a dearth of grain already strangled Siena, the city council decided to import grain via the small port of Talamone, 100 km South-West of Siena (ASS, CG 62, 1303 March 26, c. 99). They had done so already the year before. But in September 1303 -after the 290 successful grain imports that even had created an abundance of food in the city -the city council decided to buy the port for the Republic and to invest heavily in its refurbishment and expansion in the following years (Sordini 2000, 73-112). Sienese citizens were settled in the newly designed city, with infrastructures and military fortifications (Fig. 12). Although Siena had discussed buying the port years before, it was the drought experience and connected food scarcity that let the plans become real and lead to a long-term infrastructure investment, possibly the most expensive project the Republic of Siena financed 295 outside the capital's walls.

Cultural aspects of drought
The aforementioned infrastructural responses by the city of Siena to the drought experience provoked a satirical response by Dante Alighieri. In his 'Divine comedy'. the Florentine poet mocked his Sienese neighbors with a famous verse: "You will find them [some Florentines] amongst the foolish crowd [the Sienese] who put their trust in Talamone, and will lose more 300 hope there than in their search for the Diana" (Dante, Divine Comedy, Purgatory, canto XIII). The Purgatory, finished during the early 1310s, makes clear that the futile Sienese efforts to search for their underground river did not go unnoticed by their neighbors. Nor did the Florentines ignore the acquisition of Talamone -they had experienced in 1303/04 how vital the port was for their food security, too. So it was just consequent to sign a trade agreement with Siena in August 1311 that guaranteed Florentine access to maritime grain trade via Talamone, although under conditions very beneficial for Siena 305 (Banchi 1871, 126-127).
In another famous medieval text, the De regimine principum, a kind of manual on good governance by Thomas Aquinas and Ptolemy of Lucca, the latter starts writing from 1302 onwards and reflects on the importance of food security: "Food that is sold is not as effective for nourishment as it should be, since it is often adulterated. As Solomon says in Proverbs (Prov. 2,5): 'Drink water from your own cistern', which includes all nourishment, but especially drink, because it can more easily be 310 adulterated.
[…] There is greater security in using one's own food, since outsiders can easily poison something not kept in its proper storehouse or pantry, and it is more likely to be harmful." (Blythe 1997, 114). The whole reflection about safe access to food and drink is framed with a most striking biblical proverb, traditionally attributed to Solomon himself: "Drink waters out of thine own cistern, and running waters out of thine own well. Let thy fountains be dispersed abroad, and rivers of waters in the streets." (Prov. 5,(14)(15)(16). That Ptolemy of Lucca wrote about the drought in his chronicle (Clavuot 2009, 652) exactly 315 the time he worked on the mirror of princes makes a connection even more plausible.

Societal responses to water scarcity
1922), and when two blazes hit the city in November 1302, more than 200 people fought the fire in an organized way and 350 used more than 1800 water charges (Agnolo di Tura, 265). These are strong indicators of a proto fire-guard system in Italian city-states around 1300, before official fire guards were established, e.g. 1344-48 in Florence (Contessa 2000. One week after the fire, and again in early August 1304, the blaze was even a topic for homilies of the already mentioned Dominican Giordano da Pisa who reminded his fellow citizens that the fire only did what it should do according to God's will: bring warmth-and when it burned the city's houses, this was no sin, but God's will (Varanini, Baldassari 1993, 314). 355 Another strategy to cope with fire threat was to oblige citizens, in case of droughts or strong winds, to arrange full water buckets at the doors of their houses, ready to be immediately used by anyone in case of fire emergency. A Parisian chronicler described this coping strategy already in 1305, when high temperatures combined with strong wind made authorities worry about a potential disaster (Buchon 1827, 116-117). This is a shared concern of other sources, too (Wattenbach 1851a, 676).
That wind was crucial can also be derived from the fact that the fire was able to cross water bodies like rivers (Wozniak 360 2015).

Conclusion
The wet anomaly of the 1310s has attracted a lot of attention from scholars these last years (Slavin 2019), as it is commonly interpreted as a signal of the transition between the MCA and the LIA (Campbell 2016). The huge variability that can be observed during this decade, similarly with the high interannual variability observed in the 1340s, have been highlighted as 365 side-effects of this rapid climate change. In the context of global warming, specialists now agree that periods of rapid climate change are accompanied by a probabilistic higher frequency of extreme events (Sippel and Otto 2014). To date, in the field of Medieval climate history, no efforts have been made to underline the outstanding period of drought of the first decade of the 14th century. However, we have demonstrated that two exceptional series of warm and dry summer half-year occurred respectively in the mediterranean italian regions between 1302 and 1304, and north of the Alps from 1304 to 1306. 370 Following our indices reconstruction from narrative sources, such multi-seasonal and supraregional scale droughts did not occur during the 13th century and do not find any similar case before the 1360-1362 drought that struck all across Europe, more specifically in 1360-61 in Central Europe (Brázdil et al 2020, 82-83;Kiss 2017, 44-45;Bauch 2017Bauch , 1102Bauch -1104 and England (Pribyl 2017, 102-104), while in 1362 on the Balkan peninsula, around the Black Sea and the Aegean drought conditions prevailed (Kiss, Nikolic 2015, 13-14). As administrative data gives indicators of drought in Catalonia in 1361/62 375 (Fynn-Paul 2016, 137), one might even connect a major city fire in Urgell (Battle 1999, 79-82) with this continuous lack of precipitation. There are even indicators for a global dimension of the event: The years 1360 and 1362 were characterized in Japan by major droughts causing famines (Farris 2006, 109), also on the Korean peninsula (Robinson 2009, 163) and also 1362/63 in Western Rajasthan (Rao 2009, 19), which is a considerable difference to the 1302-07 event that is not traceable outside of Europe.
We propose then to interpret these remarkable conditions, combined with the 1310s and the 1340s cold anomalies, as part of the climatic transition from the MCA to the LIA. More than a last glow of the MCA, the dry anomaly of the 1300s combined with the wet-cold anomaly of the 1310s allow us to speak of a 'long' Dantean Anomaly: Not a new, but now a much more Future studies should investigate whether such a scenario was also present in the early 14th century when the transition from 395 MCA and LIA occurred. It appears plausible that such a climatological transition is temporally associated with a reduced latitudinal temperature gradient and consequences similar to the currently ongoing Arctic amplification.
Historical sources, coupled with acute source criticism, are deemed to be useful to refine the chronology of extreme events in combination with natural proxy-data. As we have demonstrated, the OWDA tree-ring reconstruction sometimes missed information due to scarce raw data, especially for southern Europe. An in-depth analysis of narrative and administrative 400 sources, which are sufficiently numerous from the 14th century onward, permits to draw a more accurate picture of this epoch's climate, including winter conditions that must be taken into consideration to get a full image of the droughts' extent.
Finally, droughts are phenomena that offer more for a cultural history of climate than just the analysis of religious mitigation strategies. They show both in contemporary perception as in historical analysis a connection to blazes, a major threat to medieval cities. We could demonstrate for the first time a correlation of droughts and blazes over 200 years, including a one-405 year lag in these phenomena. Furthermore, drought provokes reflections on thirst and the use of water that we otherwise hardly ever find in medieval texts.

Author contribution
Martin Bauch provided general conceptualization, curation & analysis of the Italian historical sources, the creation, evaluation, and visualization of drought indices and city blazes, writing of the original draft and funding acquisition. Thomas Labbé 410 provided conceptual input, curation & analysis of the French historical sources and visualization of other agricultural proxies.
Patric Seifert provided conceptual input on climatological backgrounds, meteorological analysis, and interpretation of documentary data for weather patterns and geopotential maps.