Flora in the Latin East: Archaeobotanical Remains from Crusader Arsur

Abstract

Archaeobotanical analysis in Crusader-period sites in the Southern Levant is rare, and the plant finds from Arsur (Apollonia-Arsuf ) add new data to the scarce Crusader botanical material. The fills of refuse from one cesspit from the 12th and 13th centuries CE have revealed mineralised plant finds seldom uncovered in the Southern Levant. Together with the carbonised plant material from other loci of Crusader-period contexts, this plant assemblage forms a comprehensive picture of the diet of the site inhabitants. The analysis of Crusader crop material shows a diet of typical Southern Levantine character that was based on the cultivation and use of indigenous crops also known from other contemporaneous sites.

Keywords:

• Palestine
• Latin East
• Crusader
• Apollonia
• Arsuf
• Arsur
• Cesspit
• Archaeobotanical remains

Introduction

Archaeobotanical processing of sediment samples is standard practice at most sites of the Bronze and Iron Ages in the Southern Levant (Orendi 2020). At sites of the Islamic or Crusader periods, however, botanical processing of sediment is rarely practiced, and reports on archaeobotanical data are few—e.g., Caesarea Maritima (Ramsay 2008: 262–265; Ramsay and Holum 2015), Khirbet Qana (Ramsay 2008: 269), Jerusalem (Amichay et al. 2019; Amichay and Weiss 2020), al-Burj al-Ahmar (Hubbard and McKay 1986), Ashkelon (Forste 2023; Forste and Marston 2019) and Hisban (Hansen, Walker and Heinrich 2017). Major sites of the Crusader period, namely Jerusalem, Akko and Jaffa, lack (published) archaeobotanical investigations despite the intensive excavations at these sites. The plant finds from Arsur add new data to the scarce Crusader botanical material. Due to different preservation conditions, the archaeobotanical macro remains from the 12th and 13^th centuries CE have been preserved, mineralised and carbonised, and they present a holistic picture of the diet of the Crusader town of Arsur (Fig. 1).¹

Fig. 1: Arsur: location map (drawn by Itamar Ben-Ezra)

Historical and archaeological background

The site of Arsur was continuously occupied from the late 6th century BCE to the 13th century CE with architectural remains exposed from the Persian, Hellenistic, Roman, Byzantine and Islamic periods and with the name of the town changing from Apollonia to Arsuf and Arsur. According to Albert of Aix (1923: 55, Chapter 54), the site was conquered by the Crusaders in 1101. Towards the mid-12th century, ownership was transferred to a Crusader noble family and the site became the centre of a feudal seigneury. By the end of a Mamluk siege in 1265, the town and castle had been destroyed, never to be resettled (Tal and Roll 2011; Kenzler and Zeischka-Kenzler 2021).

In 1187, Arsur was briefly occupied by Saladin’s forces (Lyons and Jackson 1982: 268, 317). Archaeologically, the destruction is well documented in Area T by thick layers of debris with scorch marks, as well as weapons and ammunition. The site was reconquered by the Crusaders during the Third Crusade, after the battle of Arsuf, in which Richard I (‘the Lionhearted’), king of England, personally directed the victory over the army of Saladin on September 7, 1191 (Kedar 2015). Following this event, a peace treaty was signed between the Christians and the Muslims in September 1192 CE, formally restoring the seigneury of Arsur to the Crusaders. It seems that resettlement at the site took place in 1236 at the latest, and when John II of Ibelin took over the governance, a new construction of the castle took place, along with a renewed settlement in parts of the town. New dwelling complexes with commercial areas were built on the rubble of the former residential area, some of them still oriented to the former structures.

Excavations at Apollonia/Arsuf/Arsur over the past four decades have uncovered crucial material evidence for the site’s medieval occupation, which was mostly confined to the walled town. The collaborative research project between the University of Tübingen and Tel Aviv University focused on the Crusader-period town and its urban–rural relationships. During a geomagnetic survey of the town in May 2012, three areas (U, T and W) were found to be suitable for archaeological investigations because of their different structures (Fig. 2). The project continued from 2012 to 2016 (Zeischka-Kenzler et al. 2018; 2021).

Fig. 2: Site plan indicating excavation areas (survey and drawing by Slava Pirsky)

In Area U, located in the southwestern area of the walled town, on the slope, excavations were conducted in 2012 and 2013. An area totalling ca. 220 m² was excavated. The southern half of the area is occupied by a larger building, which dates back to the 13th century CE. In the northern half, some structures were preserved under a thick backfill. This includes a cesspit backfilled with some finds, including pig bones with butchery marks, spaces with screed floors and a courtyard with stone paving. The features in Area U indicate extensive rebuilding and alterations dating from Late Antiquity and the Early Islamic to the Late Crusader periods (Fig. 3).

Fig. 3: Site plan of Area U (survey and drawing by Slava Pirsky and Annette Zeischka-Kenzler)

Area T is located close to the southern town wall, southeast of Area U. From 2012 to 2016 almost 450 m² were stratigraphically excavated. In contrast to Area U, it revealed numerous walls forming small-scale residential structures, installations and units. Remains of at least three, and possibly four, building complexes dating from the Fātimid to the first phase of the Crusader period (1101–1187) were uncovered in the middle of a densely built district. A typical Levantine appearance characterises the complexes: diverse rooms are arranged around an inner courtyard, among the rooms a small yard with the household area and a refuse pit (Fig. 4).

Fig. 4: Site plan of Area T (survey and drawing by Slava Pirsky and Annette Zeischka-Kenzler)

Geography, climate and vegetation

Arsur is located in the southern Sharon Plain about 15 km north of Tel Aviv on a kurkar (fossilised dune sandstone) cliff, some 28 m asl (Tal 1999). The climate of the Southern Levant is biseasonal, with hot, dry summer months and mild rainy winters. The modern mean annual precipitation is 520 mm, with most of the rainfall occurring from December to February,² therefore allowing rain-fed agriculture. Studies on the reconstruction of the palaeo-climate show an increase in arid environmental conditions. Analysed pollen cores from the Dead Sea all show low olive pollen values from the Early Islamic to Crusader times (Neumann et al. 2010). Accordingly, the level of the Dead Sea fell drastically from -370 m to -400/408 m asl during the Middle Ages (1500 to 400 cal BP) (Migowski et al. 2006). Increased aridity was also observed by an increase of δ¹⁸O isotope values from the Soreq Cave with a sharp decrease of the calculated annual palaeo-rainfall from 1000 BP onward, recovering again over the subsequent six centuries (Bar-Matthews and Ayalon 2004). The palaeo-climate proxies reflect climate trends, but inter-annual variations in precipitation may have had a great impact on the agriculture as well. These variations often resulted in severe drought years, requiring irrigation for the cultivation of crops (Bruins 2012). Thus, the Crusader town itself might have developed several methods to ensure water supply. The Palestine Exploration Fund site map prepared during a 1874 survey illustrates two springs at the foot of the cliff (in the shallow sea water), one below the southwestern corner of the Crusader castle and the other below the southwestern corner of the walled city (Conder and Kitchener 1881: 18, 36). In addition, surveys and excavations within the town area resulted in documenting several Byzantine and Early Islamic cisterns and pools for collecting rainwater, which might have been used in Crusader times as well (Roll 1999).

The coastal strip north of Tel Aviv is composed of sand dunes and kurkar ridges alternating with sandy clay. The sandy shore at Arsur takes up a narrow strip of land, with its sand dunes ascending steeply to the kurkar ridges, comprising sandy clay layers and friable calcareous sandstones. The sandy clays are fertile soils low in water retention (Zohary 1962: 12–13). It seems that the plot-and-berm agriculture was central to the period and region in question (see, e.g., Roskin and Taxel 2021; Taxel and Roskin 2022). Further inland, the Sharon is covered with transported terra rosa and rendzina soils from the Samarian Hills. Due to their physical makeup, these soils were assigned by Zohary to the alluvial soil variety (1962: 10). In modern times, the Sharon Plain (central Coastal Plain) has been known as vast farmland in which citrus fruits and other intensively irrigated crops were cultivated (Richter 1981).

The Excavations

Area U

Fills 6414 and 6422 in Cesspit 6407

The rectangular cesspit (Locus 6407) is 2.20 m long, 1.64 m wide and ca. 1.43 m deep with a capacity of some 5,000 litres and plastered (1 cm thick) inner walls (Fig. 5). At the bottom the plaster has been broken out in order to allow liquid drainage. The genuine ḥamra soil at the bottom is contaminated by organic material.

Fig. 5: Locus 6407, looking north (photo by Annette Zeischka-Kenzler)

In the northeast, the wall shows a clear enlargement with semi-circular breakout. The negative shape could indicate a former pipe. This shows that wastewater was discharged into the structure from the northern direction. Opposite projecting stones make it appear like a broken-out former construction and could indicate an older utilisation phase.

The backfilling layers (Loci 6410, 6411, 6414 and 6422) of the cesspit differ mainly in terms of colour, but pottery fragments from different layers were found to be contemporaneous. The fillings are thus from the same period. It is remarkable that many charcoal pieces (especially in Locus 6414) and other organic material were part of the fills. Samples were taken from the middle (Locus 6414, 27.93–27.59 m asl) and lowest layers (Locus 6422, 27.59–27.46 m asl). The whole construction was originally vaulted, as was typical of such constructions. It may even have been a small cistern or water reservoir in an earlier phase, because the ground seems to have once been plastered. The condition in which it was found shows a secondary use as a refuse pit or cesspit in the late Crusader era. Pottery from the 13th century CE was found in the backfill.

Area T

Fill 5308 above Courtyard 5319

To the east of this area, the terrain rises and the preservation of the features deteriorates significantly. Towards the west, the boundary of an eastern plot was documented. This includes a small rectangular courtyard (Locus 5319) with a stone pavement (Fig. 6). Its boundaries are marked by remains of walls; thus, the dimensions of about 3.20 × 2.80 m— around 9.0 m² in area—could be well recorded. The irregular pavement consists of flat large slabs of kurkar. Usually, stone floors were used for open spaces such as courtyards, but also for commercial areas. A large number of olive pits was found in a red sand layer (Locus 5308, 34.11–33.82 m asl) covering the stone floor. For this reason, samples were taken at this location. The plot with its structures was seriously damaged during the Ayyūbid conquest of the town under Saladin in 1187.

Fig. 6: Part of Area T: aerial view (photo by Boaz Gross)

Fill 5342 from Cistern 5457

The feature of a sunken installation (Locus 5457), which unfortunately is very disturbed (Fig. 6), raises questions. The first planum already indicated the obviously demolished cistern by means of a sharply delineated, large rectangular soil discolouration, oriented north–south. In the more than 2.0 m deep backfill of a trial trench nearby, hardly any finds were revealed, which leads to the assumption that material from elsewhere was backfilled at this location on a large scale. Obviously connected with the cistern are the stratigraphically later remains of an installation with two chutes (Locus 5456). The locus is 1.85 m wide and was flanked by plastered walls of kurkar ashlars. Both chutes slope steeply to the east. They consist of medium-sized kurkar ashlars covered with a thick, hard, whitish mortar plaster. In any case, the uniformly plastered surface indicates the introduction of a liquid. The southern chute, 1.40 m long and 0.30 m wide, slopes down from the northwest to the southeast. The northern one, 0.85 m long and 0.43 m wide, is less well preserved. Collapsed stones from an upper construction destroyed the chute at its eastern end. The samples were taken from the lower fill (Locus 5342, 33.13–32.93 m asl) under the collapsed stones. The entire installation was damaged and covered over in the course of the reconquest by Saladin in 1187, too.

Destruction layers 5363 and 5395 from Rooms 5405 and 5410

The best preserved plot structures appeared to the west, so Area T was extended further downslope. In the east of this central plot there are three rooms. They were overlaid by a thick destruction layer with remains of burned wooden beams and a great deal of finds. Samples were examined from one of these stratigraphically dateable layers, which are related to the reconquest in 1187. The dark brown, thin, very sandy, coal-like layer (Locus 5363, 33.96–33.95 m asl) extended over the northern (Locus 5405) and middle (Locus 5410) room (Fig. 6).

The middle room with the screed floor (Locus 5410) is ca. 4.85 m long and 3.10 m wide, corresponding to an area of about 15 m². The room is enclosed by quite uneven walls, mostly made of reused mid-sized ashlars but also rubble set in hard white mortar. The room was originally about 3.80 m wide and thus over 3.0 m² larger. In the northern half of the west wall lay the entrance, of which the threshold and part of the door frame are also preserved. The entrance, however, was modified several times. Initially, the level was lower and there was a wide double door, which was later narrowed. In the debris (Locus 5398) on the floor, a trapezoidal keystone was found that may have belonged to this door. In the last phase, the entrance level was raised by stone blocks, and behind them individual ashlars were placed in a semi-circle, forming a kind of staircase into the lower room. The screed floor (Locus 5410), ca. 0.01 m thick, is flat, with pebbles appearing in places. The charcoal and ashy destruction layer (Locus 5395, 33.79/33.57 m asl) above the screed contained the most archaeobotanical remains.

In the southwestern corner of the room, a rectangular platform (Locus 5411), 2.70 × 1.60 × 0.22 m, was set with large kurkar ashlars. Spaces in between were filled with smaller stones. Mortar remains were still visible in places, suggesting that this was originally plastered. Interestingly, there is a stone block with an elongated hole in the south wall. This could have been an anchoring point for a wooden beam to operate a lever-and- screw oil press. A similar finding was documented in Area E (Locus 1397), where the anchoring point is located in the town wall and the majority of the finds date to the early to mid-12th century CE (Ayalon, Tal and Yehuda 2013: 279–281, Fig. 15). The room is located between a roughly circular and conical silo-like installation (Locus 5241), where a conspicuous number of olive pits and a high presence of grains were detected, and a kitchen area was found with a tabun, small pools and a cesspit. Perhaps the processing of food or oil pressing took place here. Although olive presses tend to be located on the periphery because of waste, they are not uncommon in urban areas, where they produced for a smaller demand.³ The entire plot was covered by a massive debris layer, in which a coin of Saladin (1193–1195) was also found. The destruction must have occurred during the Ayyūbid conquest in 1187, and new buildings were then erected on this debris layer in the 13th century CE.

Channel 5516

One of the most interesting discoveries occurred in Area T with the uncovering of a partially preserved covered channel (Locus 5373, Figs. 6 and 7). The elaborately constructed channel extends to the southern half of the trench. A total length of at least 13.0 m is archaeologically verifiable, but only a central section is well preserved. Its width varies between 0.80 m and 1.10 m (inside 0.37 m wide and about 0.55 m high). The capping, partly still in situ, is made up of several larger kurkar boulders, measuring between 0.20 × 0.45 m and 0.40 × 0.55 m. These boulders were set unevenly and overlap one another in several places.⁴ It became apparent that some large boulders had been robbed. The sides of the channel were built of kurkar stones, showing clear traces of water flow. The construction of the channel required some restructuring of the plot development.⁵ However, the deep modern ground encroachment on the southeastern edge of the trench destroyed all stratigraphic relationships. The question how far the channel extended to the east remains open, but the channel seems to have had a connection to the cistern (Locus 5457). There is, however, no doubt that the channel sloped toward the west. Obviously, there was also a connection to the room with the above-mentioned silo-like installation (Locus 5241). A small sewer, constructed of two parallel upright stone slabs with the bottom covered with stones, represents a connection between the silo-like installation and the covered channel. This small sewer installation is about 0.45 m long and has an inner width of 0.15 m.

Fig. 7: Channel course around Locus 5581 (photo by Annette Zeischka-Kenzler)

Ca. 1 m west of the corner of a wall (Locus 5372), the channel slants toward the north at about a 45° angle from its original course (Fig. 7). It was built exactly along the two wall courses. It bypassed a cesspit (Locus 5581) of a southern plot in order to avoid disturbing it. This means that the cesspit was still in use during the construction of the channel. A small area before the diversion of the channel was covered with a lime screed, in which many burnt olive pits were baked. This paving may have served as a footbridge. On the other side, the southern dwelling area of the central building complex was abandoned due to the construction of the channel. Thus, the continuation of the wall (Locus 5372) and the screed floor (Locus 5388) were cut, reducing the room by almost half. The area south of the channel was disturbed, but robber trenches still survived.

In its course further to the west, preservation of the channel increasingly deteriorated. The surrounding area showed massive traces of robbing and destruction and seems to have been washed out due to overflowing water. In this section, the channel is again flanked by masonry, and in its elongation it runs almost to the middle of the eastern wall of a 13th-century CE building complex. There were obviously problems with the subsoil. The eastern wall sank somewhat and had to be reinforced, and a slightly sunken-in area could be observed in the floor (Locus 5573). The channel’s direction seems to be towards a massive rectangular pool (Locus 5675/5681, ca. 3.30 m in width) at the western border of Area T.

In summary, it appears that the covered channel served as a sewer. Liquids were discharged from at least two locations. There was a kind of connection to the cistern, and it seems to have been possible to flush the sewage from there.⁶ Everything speaks to the fact that in the 12th century CE, service water or waste water from productions, but not from feces, were transported away from their source. The channel’s use was discontinued after the 1187 Ayyūbid conquest of the town under Saladin. In 2014, a cover slab was temporarily removed and the greyish-brownish humous sandy backfill (Locus 5516) was sampled.

Materials and methods

The sampling of sediments for the purpose of archaeobotanical analysis was carried out selectively. We sampled loci that we considered to be of importance based on the identification of the features, as well as on eye inspection. The decisive factor was loci that could be clearly classified stratigraphically and had not been falsified by later soil interventions. Therefore, only soil deposits thought to be rich in archaeobotanical remains from the Crusader period were sampled in order to recover flora and microfauna (see online supplement). Normally, three buckets of sediment material (= 30 litres) per sample were taken for flotation. The buckets were then filled with water to let the sediment soak. After a while, the floating material (light fraction) was skimmed off with a 0.5 mm-wide sieve. Finally, the material left in the bucket was also sieved in order to separate the muddy and sandy sediment from the larger material (heavy fraction), which might have also included heavier botanical remains such as olive stones. The light and heavy fractions of each sample were separately dried in pieces of cloth. The dried fractions were then packed in plastic bags and sent to the Archaeobotanical Laboratory of the University of Tübingen for further analyses.

The heavy and the light fractions of each sample were sieved with different mesh sizes (2 mm, 1 mm, 0.5 mm and 0.2 mm) in order to facilitate the sorting process. The archaeobotanical macro remains were sorted from the light and heavy fractions with the help of a binocular with 10× magnification. The light residues of all samples were carefully scanned for archaeobotanical finds, whereas the small fractions of the heavy residue (0.5 and 0.2 mm) were only roughly scanned due to the large volume of sandy sediment. Charcoal fragments were also sorted, but were separated from the seed and fruit remains as the charcoal was not examined. For identification of the plant material, the reference collection of the archaeobotanical laboratory as well as identification literature (Nesbitt 2006; Neef, Cappers and Bekker 2012; Cappers, Bekker and Jans 2012) were consulted.

A whole seed or fruit was counted as one. Fragmented macro remains of most of the taxa were counted as one if preserved at least in halves. Exceptions were made for cereals, olive and fig. Four quarters of cereal grains were added up to one grain. As the olive pits were commonly broken and some samples contained many fragments, weighing was the main method used for quantifying these remains (Samples 08, 09 and 10; see online supplement). Due to the high volume of sediment residue, the small-sieved fractions (0.5 and 0.2 mm) of Sample 02 and 06 were split into smaller sample sizes with the help of a riffle box. Using this method, the 0.2 mm fraction of Sample 02 and the 0.5 mm fraction from Sample 06 were split four times, whereas the 0.5 mm fraction from Sample 02 was split once. Finally, the seeds of the goosefoot plant (Chenopodium sp.) and the small and very small fig nutlets from the split fractions of these samples were summed up to the whole sample size with the factor 16 and 2 respectively. The calculated sum is recorded in Table 1 and the online supplement.

Table 1: The archaeobotanical plant finds from Arsur’s Areas U and T

Results

Ten samples were collected for archaeobotanical investigations resulting in 12,911 identified seeds and fruits. The largest number of plant finds (12,401) was collected from two layers of a cesspit located in Area U (Samples 02 and 06). Only 510 seeds came from the eight samples from Area T taken from various loci, e.g. the filling of the channel (Locus 5516), an ashy destruction layer (Locus 5395), and the filling of the dried cistern (Locus 5342; see online supplement). The archaeobotanical material from both areas consists of 69 taxa, from which 29 taxa have been assigned to cultivated species, whereas three taxa were assigned to different sizes of Ficus carica nutlets (>1 mm, >0.5 mm and >0.2 mm). We were frequently able to differentiate between carbonised and mineralised taxa of the same species. In what follows, the samples from the two areas, U and T, are discussed separately as their assemblages differ considerably (Table 1).

Area U

Most of the macro remains of both samples were mineralised. The assemblage from Area U is dominated by the remains of Ficus carica (fig; 98%; Fig. 8a–c). Mineralised grape pips (Vitis vinifera), mineralised, dried small berries (together 0.2%), one nutlet of possible white mulberry (Morus cf. alba; 0.08%; Fig. 8f) together with carbonised olive pits (Olea europaea; 0.4%) complement the fruit finds from the cesspit layers. The possible find of white mulberry is exceptional as it is the first find of this species documented so far in the Southern Levant. It is assumed that the nutlets of black and white mulberry cannot be distinguished (van Zeist, Bottema and van der Veen 2001: 31). Despite the mineralisation of the nutlet it was possible to exclude it from Morus nigra—the black mulberry. The nutlet is 3.1 mm long, 2.3 mm wide and in ventral view, 1.6 mm wide. The side view shows a rounded and flat depressed nutlet with a slightly pronounced ridge. The hilum is clearly set back and contrasts the rather straight line of the hilum in Morus nigra (see Amichay and Weiss 2020: 659–660 for the description of black mulberry nutlets). Due to the mineralisation, it was hard to definitively identify the remains of some umbellifers at the species level, but it was obvious that two different taxa were present. The taller mericarps (average: 3.6 mm long, 1.7 mm broad, 1.1 mm thick) have been identified as possible fennel (cf. Foeniculum vulgare; Fig. 8e; cf. Carruthers and Smith 2020: 51). The shorter mericarps are 2.8–3.0 mm long, 1.5 mm wide and 1.2 mm thick on average. The mericarps are elliptical in shape with pointy apical ends as the stylopodium is no longer recognisable. The mineralisation left no clear indicative marks on the surface so that the smaller mericarps were named Daucus carota type due to the shape of the mericarps and the weakly pronounced ridges on the dorsal side (Fig. 8d; cf. Carruthers and Smith 2020: 55; Mueller-Bieniek 2010).

Fig. 8: Images of selected taxa from Crusader Arsur; a) SEM of a mineralised fig nutlet (Ficus carica) bigger than 1 mm from the cesspit; b) SEM of mineralised fig nutlet (Ficus carica) smaller than 1 mm from the cesspit; c) SEM of a mineralised fig nutlet (Ficus carica) smaller than 0.5 mm from the cesspit; d) dorsal, ventral and lateral view of the carrot-type (Daucus carota type); e) dorsal, ventral and lateral view of the possible fennel (cf. Foeniculum vulgare); f ) ventral and laterial view of a mineralised nutlet of possible white mulberry (Morus cf. alba); g) seed of the broad bean (Vicia faba cf. var. faba) (photos by Andrea Orendi)

The wild species are composed of various representatives of ruderal and segetal plants (inter alia, Segurigera securidaca, Malva sp., Plantago sp.). The mineralised seeds of the goosefoot plant (Chenopodium sp., 0.5%) were the most numerous finds. Chenopodium species of the Southern Levant typically grow on roadsides near waste places, waste heaps, or irrigated fields (Zohary 1966: 140–143). The other small to medium-sized seeds were low in number and mainly from taxa of the Poaceae and Fabaceae families.

Area T

The different loci resulted in a variety of cultigens and wild plants of a typical Mediterranean composition. The number of macro remains is low, compared to those from Area U. Nevertheless, the samples provide reliable data. The ashy concentration contained the most archaeobotanical remains (Sample 01; N=247; see online supplement), which were dominated by medium-sized to small seeds of wild plants. The wild species are mostly ecological markers of waste places or fallow fields. These include goosefoot (Chenopodium sp.; 8.6%), mallow (Malva sp.; 2.0%) and timothy grass (Phleum sp.; 4.1%). Typical weeds such as darnel grass (Lolium sp.; 8.4%) and corn buttercup (Ranunculus arvensis; 1.0%) point to the cultivation and processing of crop plants. Indeed, crop finds include grains of barley (Hordeum vulgare, 0.6%) and free threshing wheat (Triticum durum/aestivum, 1.4%), as well as chaff remains of hulled wheat (Triticum dicoccum/monococcum, 0.2%). The lentil (Lens culinaris; 2.4%) seems to be the most common edible legume at Arsur in Crusader times. However, the plant finds from Area T point to a diverse protein-rich diet of pulses, with finds of broad bean (Vicia faba; Fig. 8g), chickpea (Cicer arietinum), grass pea (cf. Lathyrus sativus), common vetch (Vicia sativa) and garden pea (Pisum sativum). Olive was abundant in each sample of Area T and in large amounts (33.7%). The most numerous olive pits come from a red sandy layer (Locus 5308) together with many fragmented stones collected from the filling (Locus 5516) of the covered channel (Locus 5373) and mineralised fig nutlets (7.7%). Grapes were only found as single pips (1%).

Discussion

The cesspit finds (Area U)

Cesspits have been used as collecting pits for sewage installations and latrines, as well as refuse pits. In Europe latrines were found as early as the Roman period, and there was a sharp increase of latrine finds in Middle Age towns. The fecal remains contain hard- shelled seeds and fruits or cellulose plant finds not digested by the human body—e.g., fig nutlets or grape pips, but kitchen and household waste might have been discarded into these pits as well. Plant finds from cesspits and latrines allow insights into the dietary habits of the town’s inhabitants. Mineralisation of plant finds is a quite common process in chalk and phosphate-rich layers known from European cesspits (Jacomet and Kreuz 1999: 62). In the Southern Levant, cesspits and mineralised plant finds are not among the most frequent archaeological finds, a fact that renders the finds from Arsur even more outstanding. It may be added that cesspits are known in the region as early as the Bronze Age, generally in connection with palatial architecture (cf., e.g., Langgut et al. 2016). The mineralisation of plant finds is typical of contexts with fecal contents (Jacomet and Kreuz 1999: 87–88). The process of mineralisation in latrines is caused by fecal phosphate salts. Phosphate and calcium replace the endosperm of the plant remains. The result is a negative shape of the inner part of the macro remains, but the surface texture gets lost. Sometimes during the mineralisation process the phosphate salts build a protective layer around the plant remains (Jacomet and Kreuz 1999: 62, 88; Green 1979). However, due to a lack of micromorphological studies on mineralised plant finds of the Arsur discoveries (which had been conducted, for example, on finds from the GivꜤati Parking Lot in Jerusalem; Amichay and Weiss 2020), the degree of mineralisation could not be determined with certainty. Nevertheless, the umbellifer mericarps were preserved as imprints of the original seeds, limiting the identification of mineralised plant remains to species level (see above, Materials and Methods). Observations on the fig nutlets seem to indicate that the seed coat of nutlets bigger than 0.5 mm (Fig. 8b) might have been covered by a protective layer, whereas the smaller fig nutlet finds (Fig. 8c) seem to be mineralised imprints of the decomposed achenes.

The cesspit uncovered in Area U was used as a refuse pit and latrine. The fills from the cesspit date to the 13th century CE, incorporating mineralised and several carbonised plant finds, pottery, some fragmented glass and animal bones, including pig. Moreover, the samples from the cesspit contained small dark globules of calcium conglomerations, which are indicative of feces (oral communication by J. Wiethold; see Sample 06 column in online supplement). The mineralisation of plant remains is rare for Southern Levantine plant assemblages (e.g., Ramsay 2008; Amichay and Weiss 2020). In addition to the finds from Crusader Arsur, there is an outstanding assemblage of mineralised plant remains from cesspits and refuse pits of an Abbasid bazaar (mid-8th to 10th century CE), located south of the Temple Mount, Jerusalem. The assemblages show both similarities and dissimilarities. On the one hand, thousands of fig nutlets, as well as achenes of umbellifers, were uncovered at both sites; on the other hand, the Abbasid assemblage contained some carbonised plant remains with finds of cereals, which are completely missing in the spectrum of herbs, vegetables and fruits that were probably traded and consumed in the Abbasid bazaar (Amichay et al. 2019; Amichay and Weiss 2020). In contrast, the finds from Crusader Arsur represent everyday foodstuffs, which found their way into the cesspit as garbage or fecal matter.

The carbonised plant finds

The mineralised and carbonised plant assemblages of Arsur differ not only in the quantity of the plants finds but also in the composition of the material, which puts us in the fortunate position of comparing the mineralised and carbonised plant finds. The cultivated species among the carbonised assemblage contained various cereals and legumes not present in the cesspit. This is not surprising as cereals are seldom found mineralised, as are legumes (Green 1979). Instead, the carbonised seeds and fruits complement the archaeobotanical assemblage from Arsur. Without the carbonised plant finds, the cereals and legumes would have been lacking for the interpretation of the Crusader diet and economy. The diversity of seeds and fruits is higher for the carbonised plant finds than for the cesspit material, as many finds of wild plant species were extracted from the ashy destruction layer (Locus 5395) above Locus 5410. Although few in number, the variety of legumes attests to the high value of these protein-rich crops in the Crusader diet. The cereals are very few in number as well, showing no preference either for barley or free-threshing wheat. The carbonised remains of wild species are mainly representatives of the ruderal segetal vegetation growing in (fallow) fields and in open habitats (Zohary 1966; 1972; Feinbrun-Dothan 1978; 1986; Riehl 2010: 39–40). Sedges (Scirpus sp.) and lovegrass (Eragrostis sp.) are hydrophilic plants, with Eragrostis sp. also growing on irrigated grounds (Feinbrun-Dothan 1986: 283–289). In addition, some plants growing on sandy soils (Artemisia sp.) and desert steppe habitats (Medicago radiata) were among the wild plants from the Crusader time strata.

Diet at Crusader Arsur

The two preservation conditions of carbonised and mineralised plant remains complement one another in such a way that a wide range of diverse plant material could be used to study the dietary habits of Crusader Arsur inhabitants. Due to the few finds of cereals and legumes, it is hard to draw any clear conclusion about preferences for any specific grain species. According to the carbonised macro remains from the Crusader loci, barley, naked wheat, and possibly hulled wheat, as well as a wide range of legumes, were part of the available foodstuffs.

The sheer number of thousands of mineralised fig remains from the Crusader cesspit strata is impressive, evidencing their consumption by the site’s inhabitants. Figs can be cultivated in orchards, but they also grow spontaneously on wet ground along stream sides or in gorges. Feral fig trees mainly occupy habitats of farmed land, like terrace walls, edges of plantations, or on stony ground of abandoned settlements or collapsed cisterns (Zohary, Hopf and Weiss 2013: 128). Figs were (and still are) consumed fresh and dry. The dried fruits are rich in sugar and were easy to store and transport. Even in Islamic times, fig was known as a laxative, recommended for cleansing the kidneys, curing stomach problems and used for making vinegar (Ibn Sayyār al-Warrāq 2007: 159–160, 639–640, 736).

During the Crusader period, a rising demand for wine for nutritional, liturgical and medical purposes led to a revival of viticulture by the Franks. Various textual sources (e.g., pilgrims’ itineraries, legal and financial documents) mention intense grape cultivation in the Latin Kingdom. One area of grape cultivation was the Mediterranean coast from Ashkelon to Tyre (Bronstein, Yehuda and Stern 2020: 57–60). Although there is no historical evidence for viticulture at Arsur, the grape finds from this site could point to wine growing in the vicinity of the town. As archaeobotanical evidence for viticulture in the Latin Kingdom is still relatively scarce (cf. Khamisy 2021), the evidence from the site provides more knowledge on grape consumption, as it cannot point directly to the production of wine on site. No wine presses were documented (although the large stone craters found in Crusader contexts at the site could have been used for small-scale production), and almost all grape finds were removed from the cesspit, indicating the ingestion of grapes either fresh or dried (Ramsay and Holum 2015).

Olive must have played a major part in Crusader Arsur. Studies on Early Islamic charcoal material point to the cultivation, or at least the growing, of olive trees in the vicinity of the Early Islamic town of Arsuf (Liphschitz 1999; 2007: 42), and they might have been used and cultivated in Crusader times as well. Olive-oil production was a major agrarian industry in the Latin Kingdom (Boas 2003; Ayalon, Tal and Yehuda 2013); therefore, it is not surprising that the olive is found in abundance at Crusader Arsur. Two olive-oil installations (Loci 1367 and 1397) from the 12th century CE have been exposed in Area E (Room 1372 and 1358), located in close proximity to Area T (cf. Fig. 2) (ibid.: 262–281). The mill stones (basalt and beach-rock) were different in size and attest to small- and large-scale production of this valuable agrarian product. The production of olive oil is also attested by the remnants of olive pits themselves. The olive stones are crushed to small pieces, a typical waste pattern from olive-oil production (cf., e.g., Galili et al. 2021). The smaller mill stones (about 50 cm in diameter) were used as rotary hand mills. These small installations could have also been used for the preparation of food, e.g., the grinding of cereals, legumes, or condiments, which is also evidenced by the many finds of umbellifers from the cesspit layers.

The mulberry is not native to the Southern Levant, with its original distribution located in east and central China. Although the black mulberry is mentioned in the Mishna and other rabbinic texts, its first archaeobotanical testimony dates to the Islamic period (Amichay and Weiss 2020). The white mulberry might have been introduced into the Southern Levant as early as Byzantine times. Only after the Arab expansion, however, did the white mulberry tree become an economic factor for this region, as the white mulberry was distributed for the silk industry and as a fruit tree. Nowadays it grows spontaneously in disturbed areas in the coastal regions of Israel, in Jerusalem, at the southern end of the Sea of Galilee, and in the Hulah Basin (Amichay and Weiss 2020; Danin 2000: 306). Since the identification of the possible white mulberry nutlet (Morus, cf. alba) is not definite, we would like to refrain from drawing conclusions regarding the cultivation and the economic value of this species by the inhabitants of Crusader Arsur. assemblages of the Southern Levant, a paucity that highlights the archaeobotanical finds from Arsur all the more. Vegetables are an essential component of the human diet, with a high nutrient density as they are rich in vitamins and dietary fibres. Condiments are valued for their refinement of meals and for their medicinal use. Fortunately, the preservation conditions together with the quality of the cesspit finds allowed the preservation of spice and vegetable plants. The tuber of Foeniculum vulgare is eaten raw or cooked. Fennel seeds as well as the tuber are known for their anise-like taste. The essential oil of the fennel seed was used for various purposes in traditional medicine, including as a digestive remedy (Rather et al. 2016). The carrot (Daucus carota) is also employed as raw and cooked food and is well known for its supply of vitamin A (Bahrami et al. 2018). Carruthers and Smith (2020: 55) assume that carrot seeds might have been used as seasoning when found in cesspits, but could have been used for medicinal purposes as well. Daucus carota is a common wild species in the Southern Levant. As a ruderal plant it grows at roadsides and in fields and is part of the batha vegetation (Zohary 1972: 452–453). From written sources it is known that the carrot was cultivated since Classical times at the latest (Zohary, Hopf and Weiss 2013: 160). Whether the carrot seeds from the cesspit come from cultivated carrots or were gathered from the wild is a question that cannot be answered. However, the find situation of whole carrot seeds inside the cesspit—probably as fecal matter—indicates that the carrot might have been for seasoning or for medicinal purposes.

The composition of the Crusader crop material shows a diet of typical Southern Levantine character that was based on the cultivation and use of indigenous crops. The botanical diet of the site’s inhabitants was in fact similar to that known from earlier periods in the region (Orendi, Lichtenberger and Tal 2021), as well as from contemporary sites. For example, the Crusader samples from Caesarea Maritima contained mainly carbonised plant material, so that cereals and legumes were found in larger amounts than at Arsur (Table 2). The composition of the crop species is similar to that of Arsur. The legumes show a high diversity, with lentils being the most commonly found. Cereals comprise hulled/naked wheat and barley, whereas fruit remains come from olive, grape and fig. Due to waterlogged preservation conditions, the plant finds from Caesarea Maritima revealed shell fragments of walnut and pine (Ramsay 2008: Table A, 260–265). At al-Burj al-Ahmar (the Red Tower), the plant finds are scarce, but the data nevertheless permit some conclusions. Again, the legumes are composed of various species, but in contrast to Arsur and Caesarea Maritima, lentils are not as abundant as chickpeas and broad beans. In addition, the remains of naked wheat, barley and olive are apparent (Hubbard and McKay 1986). These comparative data from Crusader-period sites in the Southern Levant supplement our knowledge on the period and the dietary habits of the inhabitants (Tables 2 and 3). When it comes to the site of Arsur, the archaeobotanical evidence from the Crusader-period occupation at the site, as analysed from Areas U and T, complements the site’s animal-bone remains, as analysed from Areas F and R (cf. Pines, Sapir-Hen and Tal 2017). Both types of organic material remains supplement one another and offer an intimate understanding of the dietary habits of the site’s inhabitants.

Table 2: The plant data of Crusader sites from the Southern Levant in comparison: Caesarea Maritima (after Ramsay 2008); al-Burj al-Ahmar (after Hubbard and McKay 1986)

Table 3: The plant finds of cultigens and usable plants from the Crusader sites in comparison: Caesarea Maritima (after Ramsay 2008); al-Burj al-Ahmar (after Hubbard and McKay 1986)

With regard to the ethnicity of the site’s inhabitants, historical sources state that Arsuf was abandoned by its indigenous population after the Frankish conquest (Albert of Aix 1923: 55, Chapter 54). The consumption of pigs during the site’s Crusader occupation may point to the inhabitants’ Christian identity (Pines, Sapir-Hen and Tal 2017: 308, 315). Various aspects speak to the fact that they were of Christian faith (cf., e.g., Tal, Taxel and Zeischka-Kenzler 2018). The burial of a fetus in one of the houses of Area T, although not necessarily a Christian custom, cannot be justified with Muslim or Jewish/Samaritan beliefs. Archaeological evidence shows, in contrast, a certain continuity in the appearance and interior of dwellings (erected) between the Fatimid and the Crusader periods. A comparison of the dwellings in Crusader Arsur with those unanimously identified as being constructed by Franks, e.g., in the Frankish street settlements of Emmaus-Qubeibeh (Bagatti 1993), al-Kurum (Boas 1996) and ꜤAtarot (Edrey, Samet and Boas 2018), shows that these dwellings differed quite remarkably (Yehuda 2010: 141–195). While the Frankish buildings show a somewhat European layout and are devoid of private installations for the processing and preparation of food, the dwellings in Arsur display installations related to food preparation in a familial context. Thus, the tabun ovens, cylindrical clay-made devices for cooking and baking, merit special attention. Tabuns are deeply rooted in the material culture of the Southern Levant and are associated with its indigenous populations (Ebeling and Rogel 2015). Thus, in the case of the dwellings excavated in Arsur, it may be suggested that they were constructed by one population and resettled by another. Such reuse can be observed also in the case of cesspit L6407 in Area U, which was supposedly built during the Fatimid era, whereas the filling of the structure clearly dates to the Crusader period.

The new people using those existing structures and dwellings could have been Franks, yet we cannot reject altogether the possibility that locals returned to the cities and towns that had been conquered by the Franks. The assumption that indigenous inhabitants, and not only Franks, lived in those centres receives additional confirmation in the historical sources. An appeal issued by King Baldwin I in 1115/1116 CE promised Christians from Transjordan transportation for them and their flocks and houses free of rent if they would be willing to settle in Jerusalem (William of Tyre 1986: 536). More than 50 years later, King Guy de Lusignan ruled that Syrian nations, among them Nestorians, Samaritans, Jews and Jacobites, should live for taxation reasons near the fonde en amont in Acre (‘Assises de la Baisse Court’ 1839: 282).

Summary

Our selective sampling strategy may hamper the significance of the plant finds when it comes to the economic value of the agrarian resources. Nevertheless, carbonised and mineralised plant remains complement one another, so that the archaeobotanical finds from the 12th and 13th centuries CE portray a diverse Mediterranean diet. Although low in number, the pulses attest to a diverse diet including protein-rich crops. The mineralised cesspit finds are important additions to the generally carbonised plant finds from the Southern Levant, as is shown with the finds of fennel and carrot seeds. The processing of olive oil is attested by olive presses and crushed olive pits. Although wine production was on the rise during the Crusader period, it is as yet unclear whether wine was in fact produced at Arsur, as wine presses are still lacking from the archaeological record. The white mulberry nutlet is the first of its kind known from the Southern Levantine archaeobotanical record to date. Still, the economic value of this species in Crusader Arsur remains unclear.

Disclosure statement

The authors report that there are no competing interests to declare.

Supplementary material

Supplementary data for this article can be accessed online at http://dx.doi.org/10.1080/03344355.2023.2246822

Additional information

Notes on contributors

Andrea Orendi

Andrea Orendi: Department of Archaeological Sciences, Archaeobotany Laboratory, University of Tübingen, and ArchaeoConnect GmbH, August-Bebel-Straße 16, 72072 Tübingen;email: [email protected]

Elisabeth Yehuda

Elisabeth Yehuda: Research Fellow, Department of Israel Studies, University of Haifa; email: [email protected]

Annette Zeischka-Kenzler

Annette Zeischka-Kenzler: Ceramic Museum Westerwald, 56203 Höhr-Grenzhausen, Germany;email: Zeischka-[email protected]

Oren Tal

Oren Tal: The Jakob M. Alkow Department of Archaeology and Ancient Near Eastern Cultures, Tel Aviv University

Notes

1 The archaeobotanical research was conducted within the framework of a joint Deutsche Forschungsgemeinschaft (DFG)-funded project, titled ‘Die kreuzfahrerzeitliche Stadt Apollonia/ Arsur in Israel: Struktur–Kulturadaption–Stadt-Umland-Beziehungen’, co-directed by Barbara Scholkmann (University of Tübingen) and Oren Tal (Tel Aviv University).

2 Data from the Israel Meteorological Service (http://www.ims.gov.il/) for Tel Aviv.

3 Several comparisons can be found in Boas 2017: 65 and 77, e.g., on the topic of oil production in houses, as well as an increased production of olives and oil in the 12th century CE.

4 The covered channels from the Early Islamic period at Ramla bear similarities, but are much narrower (Tal and Taxel 2008: 91–99).

5 Sewers were usually built under streets or pavements and were not so elevated. This could be an indication that the construction was not carried out by local inhabitants.

6 Cisterns associated with canals are known from Early Islamic Ramla, e.g., Locus 590 (Tal and Taxel 2008: 85–86, Fig. 6.3a–b).