Ritual responses to drought: An examination of ritual expressions in Classic Maya written sources

Planting and rain-beckoning rituals are an extremely common way in which past and present human communities have confronted the risk of drought across a range of environments worldwide. In tropical environments, such ceremonies are particularly salient despite widespread assumptions that water supplies are unproblematic in such regions. We demonstrate for the first time that two common but previously under-appreciated Maya rituals are likely planting and rain-beckoning rituals preferentially performed at certain times of the year in close step with the rainy season and the Maya agricultural cycle. We also argue for considerable historical continuity between these Classic Maya ceremonies and later Maya community rituals still performed in times of uncertain weather conditions up to the present day across Guatemala, Belize, and eastern Mexico. During the Terminal Classic period (AD 800-900), the changing role played by ancient Maya drought-related rituals fits into a wider rhetorical shift observed in Maya texts away from the more characteristic focus on royal births, enthronements, marriages, and wars towards greater emphasis on the correct perpetuation of key ceremonies, and we argue that such changes are consistent with palaeoclimatic evidence for a period of diminished precipitation and recurrent drought.


Introduction
The Maya are one of the best-known civilisations of Mesoamerica, noted for their art, architecture, astronomy, mathematics, calendrical systems, and their hieroglyphic scriptone of the few fully developed writing systems of the pre-Columbian Americas. Maya Classic period (AD 250-900) texts are well-known for their commemoration of the passing of time and are focused especially on the deeds of kings, including royal births, enthronements, marriages, rituals, and wars. However, Terminal Classic (AD 800-900) texts are something of an exception in remaining essentially mute about the warfare and social upheavals that other archaeological evidence suggests were pronounced at this time. Instead, Terminal Classic texts constantly emphasise ritual continuity via the proper perpetuation of key ceremonies.
This narrative and rhetorical shift in the last century or so of the Classic period is not only interesting in its own right, but also implies a growing disjunction between what was actually taking place and what the texts relate. Given this dissonance, it is worth asking why this narrative change appears at precisely this time in Maya history, what was the nature of the rituals the texts record, and what these ceremonies tell us, directly or obliquely, about the preoccupations of the Terminal Classic Maya?
Although the relationship between records of royal ceremonial performance and the wider ecological and agricultural concerns of Maya society has been discussed before (e.g., Freidel and Shaw 2000;Lucero 2006;Schaafsma and Taube 2006;Dunning and Houston 2011), in this paper we explore the relationship in a novel way by combining multiple sources of evidence (epigraphic, ethnographic, palaeoclimatic, and modern rainfall data) in order to examine possible links between specific Maya rituals and periods of environmental stress. The increasing range of palaeoclimatic archives indicating diminished precipitation and even recurrent severe droughts during the Terminal Classic provides context for our discussion Leyden 2002;Rosenmeier et al. 2002;Webster et al. 2007;Wanner et al. 2008;Kennett et al. 2012, Douglas et al. 2015. Such evidence undermines the perception of people from temperate climates that the humid tropics are characterized by abundance of water. To people who live there, variable rainfall patterns-too little or too much rainfall per rainy season, enough rainfall but at the wrong time, or a series of long dry seasons-have always been critical issues.
We also provide context and greater time depth for the attention paid to food production crises in much later Maya literature, such as prophecies recorded in the Chilam Balam books (dated mostly to the seventeenth and eighteenth centuries AD; Roys 1967: 122;Edmonson 1986;Bricker and Miram 2002) or the Dresden and Paris Codices (dated to the twelfth and thirteenth centuries AD ;Love 1994;Grube 2012). Whereas earlier Maya hieroglyphic inscriptions focus mainly on royal life, texts referring directly to drought do exist, although perhaps surprisingly there are only two such references in the thousands of known Classic period texts. The first is a hieroglyphic text from the site of Comalcalco (in present-day Tabasco, Mexico), from the final resting place of a Maya priest named Ajpakal Tahn, whose burial urn was richly furnished with jade jewellery, shark's teeth, carved shell and human bone pendants, obsidian blades, a flint eccentric, and stingray spines with glyphic texts as well as iconographic scenes (Armijo Torres 1999;Armijo Torres and Zender 1999;Armijo et al. 2000;Zender 2004: 250). One of the stingray spines bears a text that says rather uncompromisingly: 'there was drought, there was famine in the thirteenth year' (Zender 2004: 257, 543), which, based on associated calendrical notations, places the drought in the latter half of the eighth century 1 (Fig. 1). The second is a prophetic rather than historical reference that is found on the Central Tablet of the Temple of Inscriptions at Palenque, which may refer to a drying out or to the 'withering' of the World Tree (Lacadena 2006;Guenter 2007: 32). These two texts confirm a deeper history of drought in the region, but raise the question of whether these two references to drought and famine are all that it is to be found in the hieroglyphic record, or whether we are overlooking some indirect evidence?
Conspicuous investment in ritual practice is a widespread human response to periods of climatic stress, with rainbeckoning during episodes of drought being especially common cross-culturally. While also considering the wider character of the Classic Maya glyphic corpus, we place considerable emphasis on two particular rituals, one involving the 'scattering' of precious substances and the other a 'bathing' ritual involving a particular pair of deities. These are in fact the two most commonly-recorded rites in the Terminal Classic period, and we argue below that they were both closely related to the yearly agrarian cycle, respectively symbolizing the act of sowing and the invocation of rain-bearing clouds. Beyond this general agricultural and ecological link, we examine the temporal and spatial distribution of these rituals, which suggest a close association with periodic as well as protracted droughts.

Rain-beckoning rituals
Ritual activity as a mechanism for dealing with environmental stress has been discussed in many ethnographic and archaeological studies worldwide (e.g., Frazer 1911;Butree 1930), but also specifically in the Maya area. For instance, Nash (1970: 45) discussed cave rituals to beckon rain performed in Chiapas during the times of drought. Similar rituals have been reported also among the Tzotzil of Zinacantan (Gossen 1999: 185) as well as Nahua and Otomi people in Veracruz (Sandstrom 2005), and the use of ritual as a coping strategy to anticipate and mediate risk among the Lowland Maya has been discussed by Freidel and Shaw (2000). Various ritual obligations related to agriculture in the Maya area are also documented (e.g., Wilk 1991;Tzul 1993;Flores and Balam 1997;Hatse and De Ceuster 2001;Grandia 2004). Recent ethnographic fieldwork at six different villages in the Cayo Fig. 1 Inscribed Stingray Spine 3, of Urn 26 from Comalcalco, which records an event of drought and famine in AD 783 (after Zender 2004: 543, Fig. 73) and Toledo districts of Belize (Downey and Jobbová 2011) gathered first-hand information from local informants about recent historical climate variability, experiences of drought, and short-term responses to changing weather patterns 2 and provides important modern context for such Classic Maya practices. A further goal of this research was to identify modern-day ceremonies related to drought or other types of climate stress, to establish their time-depth and determine whether such practices persisted in Belize into the twentieth century. The results indicated a variety of coping strategies with regard to environmental disaster or stress (especially drought, but also locusts, and hurricanes), including a surprising variety of rituals that could be enacted during periods of drought. While this ethnographic study found many local differences among accounts of general-purpose and droughtrelated rituals across the two study regions of Belize, one or two documented modern rituals exhibited greater coherence, of which perhaps the most interesting is a rain-beckoning ceremony called Ch'a-cháak known especially in the Yucatan, but also in parts of northern and central Belize. Clearly, this ritual is related to Chaahk, the ancient Maya deity of rain, the personification of thunder and associated with rain and clouds (see Stone and Zender 2011: 41;Wrem Anderson and Helmke 2013). One ritual, however, was described in nearly identical ways in both study regions: if it did not rain, the village's saint was taken out from the church and 'bathed' in a spring, or left out in the sun until the stone started to sweat at which point the water was poured over it. Interestingly, the eminent Mayanist Sir J. Eric S. Thompson documented a similar ritual during his ethnological research in the 1930s at San José Succotz (Cayo district) and San Antonio (Toledo district). In describing the rain-beckoning ceremonies he writes: If the prayers for rain are not effective, the Mayas call the attention of the saints to the drought. Any saint from the church is taken outside and placed well in the sun, so that he or she may be convinced how parching are the hot rays of the sun. Undoubtedly in earlier times a statue of one of the rain gods was the victim of this irreverent treatment. At present no statues of the old gods survive and the Christian saints have to suffer in their place. Sometimes the saint is taken out of the church and marched around the building, while prayers are offered to Huitz-Hok and Santa U [the moon]. The previous night is passed in vigil (1930: 53).
The use of effigies in agricultural and/or specifically rainmaking rituals has a long tradition and is geographically widespread. In ancient Egypt, Mesopotamia, Sumer, Hittite Anatolia, Classical Greece and Rome, such ceremonies involved god effigies being carried to a river and washed, often followed by a sacrificial offering and a communal meal (Başgöz 1967: 305). Many aspects of these rituals are still used in rainmaking-ceremonies today, or have persisted until recently. For example, in Turkey children make a doll, carry it around the village, and at each house water is poured over the head of the doll. They are offered food, which is then cooked and eaten (Başgöz 1967: 304).
Emphasis on rainmaking rituals is also obvious in past and present activity in other tropical environments such as southeast Asia. One widely celebrated Southeast Asian holiday is the Songkran festival (from the Sanskrit word saṃkrānti describing astrological passage, and marking the beginning of a new Solar year; Monier-Williams 1899: 1127). This New Year festival, under different names, is celebrated, for example, in Burma, Cambodia, and Laos ( Fig. 2a), but can be traced back to India and Hindu rituals. It is performed in the middle of April (Fig. 2b), corresponding to the hot and dry period of the year, before the start of the monsoon, when people are praying for good rainfall and abundant harvest in the upcoming season. During the Songkran festival people cleanse Buddha images by pouring scented water over them and smaller effigies are often taken out of the monasteries and carried through the streets while people sprinkle water at them (Milne 1924;Rajadhon 1956Rajadhon , 1958Rajadhon , 1961Ashley 2005). Similar rituals, with slight variations, are performed during the T'ngai Leang Saka (the third day of a Cambodian New Year) and Boun Pi Mai (Laotian New Year) (Rajadhon 1961;Chiebriekao 2008).
In Thailand and Cambodia, a royal ploughing ceremony is called Phra Rat Cha Phithi Charot Phra Nangkhan Raek Na Khwan (literally the 'royal ploughing ceremony marking the auspicious beginning of the rice growing season;' Royal Institute Dictionary 1999) (Fig. 2c). It is an ancient ritual of Hindu origin, dating back to the Sukhothai period (AD 1238-1438) and usually taking place in May, June, or July, the exact date being set by Brahmin astrologers of the royal household. During this ceremony the king, as lord of the harvest, tills the ground with a plough pulled by sacred bulls. At the end of the ceremony, the king scatters rice over the ploughed furrows, which is then quickly gathered by people who believe that it will ensure a good harvest. More than a religious ritual, the ploughing ceremony is a state event that has both political and economic significance, functioning as a reminder of a bond between the king and farmers (Crawfurd 1830). This same ceremony has also been recorded in Burma as one of the rain invoking rites, which the king himself is obliged to perform in order to prove his nobility and illustriousness, with such royal actions emphasising his role as a 'Peasant King', 'one of them,' and theoretically inspiring peasants to work hard for a plentiful harvest (Maung Nyunt 1997). It is worth noting the 2 Approval for human subjects research was received from the University of Arizona Human Subjects Protection Program (No. 09-0418-02) and the Institute for Social and Cultural Research (Permit No. ISRC/H/2/5). differences between Songkran rain festivities and the southeast Asian royal scattering/ploughing ceremony in terms of the greater agency given to ordinary people in the first case, but the more hierarchical, royal interventions involved in the second. One reason may be the interest the king and state might have in rice as a taxable commodity, much as Maya kings may have had with respect to maize.

Possible Classic Maya parallels
The ceremony described by Thompson (1930) above provides a link between contemporary rain calling rituals and those he recorded in the 1920s, but it is very likely that at least some of these rituals or certain aspects of them might have survived from pre-Columbian times. Several anthropologists and archaeologists focusing on contemporary Maya rituals have commented on the extent that pre-Columbian beliefs can still be detected (e.g., Thompson 1930;Vogt 1976Vogt , 1998Schuster 1997). Ethnologist Evon Vogt said that 'considering that 500 years have elapsed since the Spanish Conquest, I am impressed with the enduring nature of Classic Maya concepts and beliefs ' (cited in Schuster 1997: 50). Furthermore, ritual theorist Pierre Smith (1982) made a distinction between 'periodic' and 'occasional' rites. Whereas periodic rituals are performed cyclically, occasional rituals are performed on an ad hoc basis, thereby increasing ritual frequency. For example, increased frequency of ritual activity in times of environmental stress among the Maya has been documented by several ethnographic studies (e.g., Girard 1949Girard , 1995Freidel and Shaw 2000). More important for this particular study are recent studies of material remains in caves of Western Belize (Moyes 2006;Helmke 2009;Helmke et al. in press), which have shown that there is an evidence of increased ritual activity in caves during the latter part of the Late Classic period (ca AD 680-960) coincident with climatic drying. Based on the evidence, authors of these studies argue that this ritual activity can be associated with rain-making and agricultural security. This emphasis on ritual activity in the archaeological record corresponds well with increased ritual focus of Terminal Classic Maya texts.
Among ceremonies commemorated on Classic Maya monuments, there are two that prevail during the Terminal Classic and that we examine in detail below. The first involves the 'scattering' of precious substances and the other the 'bathing' of a particular pair of deities, known as the BPaddler Gods ( Schele and Miller 1986: 52, 183;Freidel et al. 1993: 91-94;Stone and Zender 2011: 51, 69;Stuart 2016). A good place to start is with the stelae from the sites of Ixlu and Jimbal where we can see depictions of the Paddler deities in the upper portions of the scene, amidst dotted-scroll motifs that represent clouds, floating above the king, who performs a 'scattering' ceremony (Schele and Miller 1986: 52, 183;Stuart et al. 1999: 169-70) (Fig. 3a). The Paddlers are an important pair of Maya deities whose names remain undecipherable. One is nicknamed the Old Jaguar paddler, recognizable by his jaguar spots and the ear of a feline, whereas the other, the Stingray Paddler has a prominent stingray spine or sharpened bone piercing his septum (Fig. 3d). Their names are often represented in the glyphs as signs that resemble diminutive and stylised paddles, wherein the one is qualified by a sign for k'in ('sun, day,' perhaps 'light') and the other by ak'bal ('night') or ahk'ab ('darkness') (Stuart 1984: 13-15) (Fig. 3c). These deities are often depicted paddling a long dugout canoe, as for example in the scene incised on a human bone found in Burial 116 at Tikal, Guatemala, where the Paddlers ferry the deceased Maize God across the waters of the underworld to a place of resurrection (Freidel et al. 1993: 92;Stone and Zender 2011: 51) (Fig. 3b). From other texts, we know that this pair of Maya deities is associated not only with the creation of the world (Freidel et al. 1993: 92) but also with rain. Stela 1 from the site of Jimbal, for instance, specifically mentions Chaahk, the Maya deity of thunder and rain in connection with the names of the Paddler Gods.
In the ritual depicted on Stela 2 of Ixlu, the Paddler Gods are depicted floating within dotted-scrolls (Fig. 3a). Earlier studies identified this scroll motif as representing blood (Schele and Miller 1986: 52, 183;Stuart 1988: 184) and thus the Paddlers were thought to be born from the blood of the king's auto-sacrifice (Stuart 1984: 14-15;Schele and Miller 1986: 52, 183). Nevertheless, with the decipherment of the glyph for muyal 'cloud' in Classic Maya, the dotted-scroll motif is now understood as representing clouds, both of rain and incense, the two being symbolically equivalent (Houston and Stuart 1990). More recently, Stuart and Houston have suggested that these scenes depict the Paddler Gods undergoing a 'bathing' ritual, as a kind of rite of purification, possibly related to 'rainmaking rituals' (Stuart et al. 1999: 169-171).
Examination of the texts that accompany these scenes led Stuart and Houston to note the close affinity between the Paddler Gods and a particular verbal statement ( Fig. 4b-c) that is usually written as ya-AT-ji or ya-ti-ji (the latter phonetic spelling has enabled the decipherment of the more common logographic spelling, Stuart et al. 1999: 169;David Stuart, pers. comm. 2000). Analysed as y-at-ij, the root of this verbal expression is at 'to bathe,' which is interpreted as a nominalised construction (marked by the suffix) with a possessive prefix. This would prompt the translation of 'it is the bathing of the Paddler Gods' (Prager 2013: 261), in the passive mood, assuming that at is intransitive, based on Ch'olan cognates (Stuart et al. 1999: 169). Nevertheless, the constructions involving this expression and the Paddlers may also be interpreted as transitive constructions, involving a perfective suffix -Vj, not least since the cognate atih in Ch'orti' is the transitive form of 'bathe, wash' (Wisdom 1950: 454). On this basis, the clause would be translated as 'the Paddler Gods bathe(d) ' (MacLeod 2004: 294;Alfonso Lacadena, pers. comm. 2013).
The second, 'scattering' ritual is more clearly part of a longer tradition, with a relatively wide geographical distribution across Mesoamerica. It is often represented both in iconographic and glyphic form. In iconographic form, it appears probably as early as 900 BC on the Humboldt Celt, where Justeson (1986: 443) interpreted it as the ceremonial casting of maize kernels. The scene usually involves a ruler with outstretched arms and open hands throwing or scattering small round objects (Fig. 5a). In written form, this action is represented by a glyph depicting an open hand with small dots falling from it ( Fig. 5b-c). The early glyphic form may be recorded in Isthmian writing on La Mojarra Stela 1 in Veracruz dating to second century AD (Justeson and Kaufman 1993: Fig. 6). Many Early Classic examples are also known from Teotihuacan, the great metropolis in the central Mexican highlands (e.g., Helmke and Nielsen 2014: 89-91, 93, Figs 9a-b, 11). In Maya writing the glyph in question is read chok, meaning 'to scatter, sprinkle' (Stuart 1984: 9;Schele and Grube 1995: 40). While this reading is clear, there remains on-going debate about what the bead-like objects falling from the hand represent, with suggestions that they  are droplets of water (Kelley 1962: 40;Dütting 1974: 50;Thompson 1962: 300f), grain (Thompson 1962: 300f;Justeson 1986: 443;Proskuriakoff 1993), blood (Stuart 1984(Stuart : 9, 1988Schele and Miller 1986: 181-182), incense pellets (Love 1987: 11-14) or a combination of these (see Landa in Tozzer 1941: 140-144).
On the basis of these suggestions and juxtaposing the 'scattering' and the 'bathing' rituals with the observations from the cited ethnohistorical studies, we propose the following hypotheses: 1.) As acts symbolising the sowing of crop seed, and the invocation of rain-bearing clouds, these rituals were closely related to the agrarian cycle. As such, the 'bathing' rituals may be the precursors, or proto-forms, of some of the later rain-beckoning ceremonies described above, such as the 'bathing' of the village's saint, or the ceremonies known among the Yukatek Maya as ch'acháak, attested in both the ethnohistoric and ethnographic sources (i.e., Rejón García 1905;Gann 1917;Irigoyen 1976;Love 1984Love , 2011Freidel et al. 1993). They are rain-beckoning rituals and at present are performed annually at the end of the dry season, immediately preceding planting and sowing. Similarly, the 'scattering' rituals might be also interpreted as related to agricultural ceremonies, celebrating in their emulation, the cycles of planting and sowing of grains on the open field. 2.) These rituals may have originated not just in response to general fears about water scarcity, but also to particular episodes of drought and, without pre-supposing this conclusion, it should be considered whether they might not therefore provide historical markers of time periods with diminished precipitation.

Analytical Approach and Data
In order to test the above hypotheses, we have reviewed the entire corpus of hieroglyphic texts, noting all known occurrences of these two particular kinds of rituals: 'the scattering of drops' or uchoko'w ch'aaj in Classic Maya and 'bathing' or yatij. As there is some diversity in the manner in which these two expressions are recorded in the glyphic texts we refer to each in the remainder of this text by the verbal root of the action: chok and at respectively. The focus on these two rituals is also advantageous since  at is best known for the Terminal Classic, whereas chok is documented for the entirety of the Classic period but also with a high relative incidence in the Terminal Classic. Thus, rather than focusing on a single ritual action, we are able to compare and contrast the spatial and temporal incidence of these two distinct but symbolically-related rituals, seemingly relevant to the semantics of agrarian practices. Our working database consists of information about the site from which the chok and at statement originates, the date of the statement (or the particular historical iterance) and also the latest date of the text (to assess the degree of overlap between the written source and the event recorded, or whether the historical iterance is highly retrospective). Altogether, 23 at ('bathing') statements can be recognized from inscribed monuments at seven different sites across southern Mexico, Guatemala, and western Honduras. Of these, 19 can be securely dated (Table 1). We also have added four further monuments that are missing explicit written references to 'bathing' but which have scenes clearly depicting this ritual iconographically (e.g., Ixlu, Stela 2; Jimbal, Stela 1) (  (Table 3). Below, we first explore the monthly distribution of chok and at statements, in order to see if there was a preference for performing these rituals during certain seasons (especially given that the Maya solar calendar did not account for annual drift, and as such the emic temporal intervals are not inherently locked to a given seasonality). Ideally, we would explore the seasonality of only those statements with non-period ending dates, which are more likely to represent explicitly special events such as droughts.  Table S7) or royal accessions (using a database compiled by Martin 2014). To further elucidate the timing of chok and at occurrences, we compare these to the seasonal distribution of total rainfall in Maya area. We also compare the longer-term temporal distribution of chok and at statements throughout the Classic period with palaeoclimatological evidence to explore whether any relationship between the occurrence of these statements and drier periods can be substantiated. Again, to control for possible biases in the overall textual sample, the temporal distribution of chok and at statements is assessed against the temporal distribution of known dated Maya texts from the vantage of the entire glyphic corpus (using a database          compiled by Prager 2008). In terms of temporal distribution we look at statement distribution both throughout the time of their existence (Early Classic to Post-classic) to explore their diachronic change, and also separately for Early/Late Classic and Terminal Classic to explore the ways in which their occurrences differ between these two periods.

The seasonality of chok and at statements
Given that chok and at rituals appear to be related to agricultural practices, we examined the frequency of such statements over different months of the year (Fig. 6). Direct comparison with modern monthly rainfall totals does not demonstrate a significant association; however, the relationship between seasonal variability and occurrence of at statements (p=0.107) deserves further examination (Fig. 6a-b). The wettest periods are typically between May-June and September-October, with a short drier spell in August (known as the 'meagre season' in Belize or canícula elsewhere in Maya area) and a true dry season from February to April (Hastenrath 1967;Magaña et al. 1999). The high points of precipitation roughly correspond with high occurrences of chok and especially at statements: more precisely, there is a significantly higher frequency of at statements in June, October, and November (X 2 , p=0.016) and likewise significantly higher chok occurrences in June and October (p=0.041), even if the overall seasonality of chok statements seems less pronounced than for at (albeit in part due to differing sample sizes). A closer connection still is with present-day planting seasons in the Maya area. The first and primary planting occurs in May, at the end of the dry and beginning of the rainy season, and the second planting in October-November, especially prevalent in the humid central lowlands (i.e., Peten, Belize, Chiapas, and Tabasco) and often involves fast-ripening varieties of maize (Brewbaker 1979: 107;Nations and Nigh 1980: 10-13;Downey and Jobbová 2011: 179). Assuming there has been no dramatic change in rainfall seasonality since the Classic Maya period, it follows that local populations would rely on these months to bring the rain, especially in May before the main planting season. This is substantiated by modern practices among traditional Maya communities, where they often plant a week or two before the onslaught of the expected rains. Our suggestion is therefore that the high number of occurrences of at statements in June reflects situations where planting had occurred, but the expected rains had not yet arrived, and as a result rituals petitioning for rain were performed. A modern example of this behaviour was observed in Crique Sarco village in southern Belize in 2011, where locals had already planted by the end of May, but the expected rains were delayed. People were concerned and said that they would wait another few weeks, but if the rains still did not come, they would have to perform rainmaking ceremonies. Another example is a festival performed by contemporary Yucatec Maya,  known as Pa'puul, or 'breaking pots,' which serves as a petition for rain in Yucatan, Mexico, and is performed on the 24 th of June. The festival involves frogs associated with water, with the breaking of pots producing a sound thought to evoke the clap of thunder and rain (Smithsonian National Museum of the American Indian 2012). Many aspects of the Pa'puul festival most likely originate in ancient Maya tradition, not least since particular period ending ceremonies involved the discard of pottery and kitchen utensils at the turn of the calendrical phase (Tozzer 1941: 151-152; see also Pendergast 1971: 9). It is clear that the occurrence of at statements is highly seasonal, supporting the hypothesis that these rituals were somehow  1950-2000 (www.worldclim.org) and are used as a proxy for the distribution of rainfall throughout the year (for distribution of rainfall in Maya lowlands see Fig. 11b). a) Seasonality of chok statements. b) Seasonality of at statements. c) chok and d) at statements in comparison to seasonality of e) warfare statements and f) the incidence of known dated texts, contrasting the Late Classic incidences against those of the Terminal Classic involved in, or represent an early form of rain-beckoning rituals. There is less observable seasonality for the chok rituals, and while this may partly reflect the latter's wider geographical and temporal distribution, it is probably more due to the fact that chok or 'scattering' rituals were more general purpose ceremonies possibly associated with annual sowing and fertility, and initially less closely related to rainfall. We can further assess the seasonality of chok and at statements by comparing them with other kinds of events mentioned in Maya texts, such as royal accession, warfare, or indeed the overall background sample constituted by all known and dated Maya texts. Maya texts in general exhibit a more random distribution across the year (Fig. 6f), whereas texts relating warfare events show a slightly greater but statistically insignificant prevalence during the dry season (Fig. 6e), which is similarly the case for accession events (Martin 2014: chart 3). Martin has suggested that the Classic Maya planned public ceremonies at times when it would be easiest to travel and when it was least likely that the ceremony would be spoiled by rain, and similarly that warfare would be more likely to take place during the dry season than the wet season when people are occupied by planting (see also Schele and Freidel 1990: 62;Martin 2014: 100, 170-174). In summary, chok and at statements exhibit higher seasonality than other kinds of texts and stronger congruence with the start of the rains and planting seasons.

Palaeoclimate and chok and at statements
Our second hypothesis is that these rituals may have originated in response to particular episodes of drought and as such could serve as markers of diminished precipitation. To address this, we compare the distribution through time of chok and at statements against palaeoclimatic evidence for periods of greater or lesser rainfall (Fig. 7). However, since there is considerable geographical variability in climatological records, the central Peten-roughly a geographic median of the Maya world-was chosen as a case study area. This region provides palaeoclimatic data from the Macal Chasm speleothem (Akers et al. 2016), the sites located in the area mention both chok and at rituals, and in fact produced the majority of chok and at statements during the Terminal Classic period (Fig. 8). Nevertheless, our finds suggest that the marked increase in the number of the at statements in the period AD 652-751 closely follows an overall increase in the incidence of hieroglyphic texts in general (Fig. 7). There is, however, a small second peak in at statements between AD 850 and 900 by which time the number of hieroglyphic texts in general has decreased considerably. The peak in occurrence of chok statements appears slightly but significantly later than the The temporal distribution of chok and at dates was compared to the temporal distribution of all known Maya dates known from the glyphic corpus to determine whether the correspondence between the appearance of the respective statements and periods with diminished precipitation is significant overall increase in hieroglyphic texts (KS-test, p=0.05). The palaeoclimatic data from different regions of Maya lowlands match this increased incidence of chok statements well (Table 4). To summarise, during the Early/Late Classic period, it seems that incidences of chok rituals closely follow the incidence of texts in general (Fig. 9a), suggesting that we should not interpret chok rituals as responses to unusual events but rather as general-purpose ceremonies performed at periodendings in the Maya calendar and/or as part of other important ceremonies, such as accessions. It is further noteworthy that chok rituals appear as highly hierarchical, with a focus on the ruler as the principal officiator, underscoring his key role in the social structure as a bringer of agrarian fertility. During the Terminal Classic period, however, the incidence of chok  rituals becomes more seasonal and more closely matches that of the at statements (Fig. 9b), suggesting that chok rituals were being repurposed in this period to focus on ensuring agricultural security at risky periods of the year. Indeed, many Terminal Classic monuments suggest a pattern where chok and at rituals were performed conjointly ( Fig. 10a-b, such as Jimbal Stela 1 where Jaguar Paddler is shown performing the scattering, Fig. 4a).
We also see other changes in the character of the Maya textual sources in the Terminal Classic. By AD 800 the obvious war narratives such as the texts from Naranjo, Yaxchilan, Piedras Negras, and Bonampak disappear almost altogether (Helmke et al. 2010: 120-121) and the period between AD 800-850 involves attempts at re-establishing and maintaining the 'old order' with former adversaries conducting joint rituals and visiting each other (e.g., Caracol and Ucanal, Tikal and Calakmul at Seibal) (Grube 1994: 95-97;Helmke et al. 2010). Between AD 850-910, there are more references to period endings and rituals (e.g., Ixlu, Jimbal, Xunantunich, Machaquila, Uaxactun, Tonina) (Schele and Grube 1995). These shifting emphases in the texts were further accompanied by changes in the composition and iconography of stelae, with a de-emphasis on the king as sole autocratic ruler, and an increased emphasis on so-called confrontation scenes, wherein Fig. 9 The Seasonality of chok statements during the Early/Late Classic and Terminal Classic periods in comparison to seasonality of dated texts during the same periods power-sharing and decentralisation are apparent between multiple actors (especially pairs, Chase 1983: 105 -110) and the appearance of emblem glyphs and other royal titles at formerly secondary centres (Martin and Grube 2000: 98-99;Rice and Rice 2004: 133-134;Valdés and Fahsen 2004;Zrałka 2008: 200;Helmke et al. 2010: 109-110; see also Murphy et al. (2016) who argue that the increased frequency of agriculture related rituals and diminished accession and ruler-focused rituals are related to change in political organisation between Maya Classic and Postclassic period). In addition, there are also changes in settlement patterns and in many cases a decrease in overall population (e.g., for the Belize River Valley see Willey et al. 1965;Ford 1990;LeCount and Yaeger 2010;Hoggarth 2012; for the Peten Lakes see Rice andRice 1980, 2004;Rice 1986).

Spatial patterning of chok and at statements
A further important feature of a number of chok and at statements is that they do not occur evenly across the whole Maya world with certain sites producing such statements well in excess of the number of hieroglyphic texts at these sites in general (X 2 , p=5.9e-8, p=2.8e-24). If we further compare the distribution of at statements against the map of modern precipitation values (Fig.  11b), all sites with at statements are predominantly located in areas with moderate totals of annual rainfall, especially the sites in Peten (i.e., Tikal, Ixlu, Jimbal and Naranjo), but also Copan in Honduras. More precisely, they can be characterised as falling into something of a 'goldilocks' zone between high and low annual precipitation regimes, with variation in rainfall likely to have  PM# 2004.15.6.16.28). b) Ixlu Altar 1 (drawing © Linda Schele, Los Angeles County Museum of Art). c) Depiction of Ajpakal Tahn of Comalcalco and texts commemorating some of the rituals that he performed at regular intervals, especially on the day 10 Sip in the Haab calendar. The dates in parentheses represent the proleptic Gregorian calendar (drawings by Marc Zender, after Zender 2004: Figs. 71-76) considerable effect both inter-annually and over longer time periods. Furthermore, sites with at rituals are located in areas without easy access to groundwater and in the case of Tikal and Jimbal also without surface water sources. It is therefore possible that 'bathing' rituals were perceived of as more important in this region with unpredictable but highly consequential environmental stress than either in the rain-poor, but groundwater rich northwest Yucatan, or the rain-rich southern highlands. If we consider only chok ('scattering') statements, the Early/ Late Classic period distribution (Fig. 11a) is much more regionally variable, but more closely matches the clustering of the at statements in a core geographical region during the Terminal Classic (Fig. 11b).
Beyond this general geographic patterning, it is worth emphasising that the incidence of these rituals was likely further conditioned by individual actors at specific places on specific occasions. For example, Tonina contributes 9 of the 23 known at expressions, and the six datable examples from this site occur within a span of only 24 years. The Tonina examples are also performed at regular intervals, corresponding to the celebration of so-called hotun intervals of about five years (i.e., AD 697, 702*, 706*, 711, 716, 721), 3 suggesting that a single ritual specialist might have been responsible for the entire set in much the same way as Comalcalco texts commemorate the rites performed by a single priest named Ajpakal Tahn over a period of 12 years (Fig. 10c, Zender 2004: 250).

Conclusions
There is a close correspondence between the occurrence of at and chok statements and the onset of rainy seasons. This strongly supports our hypothesis that these rituals were closely related to the agrarian cycle, symbolizing the act of sowing, and the invocation of rain-bearing clouds. The frequency of at ('bathing') statements through time matches the frequency of surviving Maya texts overall, indicating that Classic period examples of 'bathing' rituals were an already well-established tradition that perhaps was even performed annually (in a wider agricultural context) at the end of the dry season and before the second planting. There is a small second peak in at statements during the Terminal Classic, corresponding with a statistically-significant increase of chok rituals. This also coincides with other changes in the Terminal Classic, such as shifts in narrative form and content, as well as changes in settlement patterns, all of which can now be linked in various ways to palaeoclimatic records suggesting a period of more frequent droughts. More precisely, the increased seasonality of 'scattering' statements and their appearance alongside 'bathing' rituals in Terminal Classic texts suggest that the focus of these rituals became more narrowlyfocused on food security and rain-making, particularly for sites lying in the Maya heartland where diminished or delayed precipitation would have had considerable adverse consequences.
Funding The ethnographic research briefly mentioned in the article was funded by National Science Foundation and by Alphawood foundation (funding was awarded to the Principal Investigator of the project Dr. Sean Downey). However, the research that forms the main part of this article was conducted separately from the ethnographic project (as a part of a PhD research of the first author), without any specific funding apart from AHRC scholarship received for the PhD research in general.

Compliance with Ethical Standards
Conflict of Interest The authors declare that they have no conflict of interest.
Informed Consent Approval for human subjects research was received from the University of Arizona Human Subjects Protection Program (No. 09-0418-02) and the Institute for Social and Cultural Research (Permit No. ISRC/H/2/5).
Open Access This article is distributed under the terms of the Creative Comm ons Attribution 4.0 International License (http:// creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made.