• 1 McEnroe 1982; McEnroe 1990; McEnroe 2010; Shaw 2009; Shaw 2015.
• 2 See for instance Dollfus 1975; Goldberg 1979; Rosen 1986; Chazelles 1997; Sauvage 1998; Cammas 2015 (...)
• 3 For contemporary ethnographic studies on earthen architecture see Fathy 1969; see Basu 1993; Jerome (...)

1The Minoan civilisation that flourished on the island of Crete during the 3rd and 2nd millennium BCE is famous for its remarkable architectural undertakings which incorporate a wide variety of building materials and techniques. Amongst the most impressive are the ashlar walls that form the palaces and other specimens of elite architecture. But most constructions generally incorporate humbler building materials, such as rubble stones and mudbricks1. Earthen architecture has been the subject of numerous studies focusing on collecting and investigating data obtained from mudbricks in archaeological contexts in different parts of the world, especially Europe, Central America, and the Middle East. Macro‑ and microscopic analysis of mudbrick architectural remains have shed light on various social aspects of mudbrick production and construction, thus significantly improving our understanding of the social context of past building practices2. Likewise, ethnographic and architectural studies have been of utmost importance to deepening our understanding of different building techniques still common in Syria, Egypt, Yemen and Central Asia (i.e. rammed earth, mudbricks, and cob), of production and construction processes, and of modalities of earthen skill transfer mechanisms over time and between neighbouring areas3. In this academic landscape, the study of earthen architecture in Greece, and especially in Crete, has so far received limited attention.

• 4 Although adobe is the common engineering term, the extensive use on excavations of Cretan Bronze Ag (...)
• 5 Koulidou 1998; Friesem et al. 2014a.
• 6 Doumas 1983; Palyvou 2005: 114.
• 7 See for example Evans et al. 1964: 136; Evans 1972: 117‑118; Warren 1972: 256‑257; Zois 1976: 45; p (...)
• 8 Love 2012: 153.
• 9 Binici et al. 2005; Fernandes et al. 2010; Zhai and Previtali 2010; Gran-Aymerich and von Ungern-St (...)
• 10 Wright 2005; Wright 2009; Minke 2006: 108, 137.
• 11 Doat et al. 1979: 126; Tsakanika-Theohari 2006; Tsakanika-Theohari 2009.
• 12 Binici et al. 2005; Wright 2005; Galán-Marin et al. 2013.
• 13 Aurenche 1981; Aurenche 1993; Wright 2005.
• 14 Minke 2006; Hadjri et al. 2007: 143; Jaquin 2010: 2; Homsher 2012: 2‑3; Llampas et al. 2014.

2Minoan mudbricks are sun‑dried4, so when a building is abandoned, they progressively dissolve into shapeless earthen debris. Only in some exceptional instances when a violent destruction by fire occurs, are substantial remains of mudbrick walls preserved. Recent investigations on the ways earthen building materials react to destruction by conflagration or abandonment, or endure post-depositional vicissitudes after such events, have had little echo in studies of earthen architecture in Minoan Crete5. In fact, due to their limited preservation, mudbricks are often dismissed from synthetic studies on Minoan architecture and are barely mentioned in excavation reports. In the Bronze Age Aegean, this research trend is enhanced by the negligible use of mudbricks at Akrotiri-Thera, the well‑named ‘Pompeii of the Cyclades’, which deprives Minoan archaeologists of invaluable in situ comparanda6. Exceptions occur7, but compared with other building materials, mudbricks are without doubt the neglected twin8 of Minoan architecture. This dismissal is also sustained by misguided prejudices regarding this building material: it is often considered structurally weak, with limited bearing capacity and low resistance to atmospheric conditions. However, such limitations are often unverified and, when they do exist, they can be easily overcome using specific earthen construction strategies. Over the past 20 years, laboratory tests on the tension and compressive strength of sun‑dried mudbricks have explored the many factors that can enhance the performance of earthen architecture9. The load‑bearing capacity of mudbricks has been verified10, and it can even be improved by the use of a wooden framework set within the wall11, or by the incorporation within the mudbricks of fibrous material that increases their tension and compressive strength12. Their limited resistance to moisture is overcome by a stone socle that prevents erosion by capillarity and protects mudbricks from potential floods. Wall erosion is also prevented by the coating of the walls with plaster and by its regular maintenance13. With such provisions, mudbricks show marked advantages in comparison with other materials: they offer better resistance to lateral shocks, their thermic performance allows the slow transmission of heat or cold, they better regulate humidity, and they can resist fire destructions which might ‘freeze’ the walls instead of provoking their collapse14.

• 15 Based on Bietak 2003, 2014, Cherubini et al. 2014, MacGillivray 2014, Warren 2009, Warren 2010, War (...)

Table 1 — Chronology of Minoan Crete15.

• 16 This is the case for the wall east of room VIIa, initially made of mudbricks set of a low limestone (...)

3The present paper focuses on the production and use of mudbricks throughout different phases of construction and occupation of the Bronze Age Palace at Malia (1900-1430 BCE, table 1), a Minoan settlement situated on the north coast of Crete (fig. 1 and 2). In the Malia Palace, mudbrick walls are mainly found in groups that form distinctive architectural units, but also in the form of walls added as partitions altering the master plan of the building. Only the walls that are well-preserved are considered, while those that are heavily restored and possibly reconstructed are excluded16. This study relies on macroscopic observations of the mudbrick walls and on geoarchaeological analyses of selected mudbrick samples from different areas within the building, which properly reflect the variability of earthen architecture over time. Taking this building as a starting point, we explore the transmission of mudbrick recipes throughout the Bronze Age, the strategies adopted by the builders, and the logistics of mudbrick production and construction in a monumental site. This paper also advocates the recording and analysis of mudbrick architecture as a useful tool to answer relevant archaeological questions such as skill transfer and craft specialisation. By doing so, it aims at providing data that will help and encourage the record of macroscopic information on mudbrick architecture discovered in current excavations, as well as exploring the potential of combining macro‑ and microscopic study of earthen remains.

• 17 O. Pelon suggests a construction during the MM IA phase (Pelon 1983: 698‑700, fig. 1, 21; Pelon 198 (...)
• 18 Poursat 1988: 74‑75; Pelon 2005: 186‑190.
• 19 Pelon 1984: 884; Pelon 1986a: 19, fig. 14‑15; Poursat 1988: 75.
• 20 Pelon 2005: 191‑196; Darcque et al. 2014: 181‑182.
• 21 Pelon 1997.
• 22 The use of mudbricks in the building is hindered by the heaviness of some of the restorations. Cons (...)
• 23 Chapouthier and Charbonneaux 1928: 7, 36‑38; Pelon 1966: 1008‑1011; Pelon 1969: 1051‑1056; Pelon 19 (...)
• 24 This project, directed by M. Devolder in collaboration with I. Caloi and placed under the aegis of (...)

4The ‘First’ or Protopalatial Palace at Malia is built in MM IA (ca. 2050-1900 BCE) or MM IB (ca. 1900-1800 BCE) on top of the remains of a Prepalatial occupation (EM II and III, ca. 2650‑2050 BCE), some of which are incorporated into the Protopalatial edifice17. It is destroyed in the general catastrophe that strikes the site at the end of the MM IIB phase (ca. 1700 BCE)18, and rebuilt and largely expanded during the Neopalatial period, at the transition between the MM IIIB and LM IA phases (ca. 1600 BCE)19. It is occupied until the end of the LM IB phase (ca. 1430 BCE), despite at least one intermediary destructive event perhaps connected to the eruption of the Santorini volcano some time during the LM IA phase (ca. 1525 BCE) (table 1)20 and evidence for localized LM II reoccupation in the area of the North Court (ca. 1430-1390 BCE)21. Mudbricks are used in different areas of the Malia Palace (fig. 2)22. Early and ongoing study of the architectural and occupational sequence of the Palace makes it possible to date mudbrick construction to the EM IIB phase (ca. 2450-2200 BCE; mudbrick walls in room I 1), and to the Protopalatial  (ca. 1900-1700 BCE; mudbrick walls in Area XI or ‘Eastern Magazines’) and Neopalatial (ca. 1700-1430 BCE; mudbrick walls in Areas II, IV, V, VI and XIII, in rooms C 1-C 2, IX 1‑2, XXI 1 and in the North Portico) periods (table 2)23. Current study of the stratigraphic soundings led under the latest floors of the Palace will help refine the sequence, especially regarding successive phases within each period24.

• 25 Dimou et al. 2000: 438‑439.
• 26 Plastic earthen building material (PEM) is a building material similar to mud mortar but to which v (...)
• 27 Devolder 2016: 144‑148, fig. 4, 5a‑c.

5Mudbrick courses are generally laid on top of a stone socle made of rubble and/or boulders of local ironstone (a hard, grey‑bluish limestone) or, more rarely, conglomerate, a stone which is also procured locally (table 2)25. Only in rare instances is the stone socle absent, in which case the mudbricks are set directly on a low foundation course (fig. 15a and 24d). Both exterior and interior walls show substantial stone socles, from 0.36 m to 0.58 m in height as visible in the interior and exterior walls of Areas II, VI, VII, IX, XI and XIII, although the stone socles tend to be lower in the interior walls of Area XIII (from 0.15 to 0.20 m‑high) (fig. 3, 4, 5, 6, 16 and 21). In some interior walls, however, mudbricks are incorporated quite low in the wall base, and are usually associated with rubble to form the socle (north walls of rooms V 3 and VI 9 and west wall of room IX 1, fig. 7, 8 and 9). Whatever its height, the socle is generally closely associated with a 4 to 12 cm‑thick levelling layer of plastic earthen building material (PEM)26, the top of which is carefully flattened in order to support the upper courses of mudbricks (north wall of room VI 10, fig. 4). In several walls multiple layers of PEM (increasing in thickness from 7 to 9 cm) function as an additional buffer, separating the top of the stone socle from the mudbricks (north wall of room XI 1, fig. 9). This levelling is somewhat reminiscent of the intermediate flattening of superimposed layers of rubble stones incorporating large quantities of soil in the Protopalatial walls of the Western Magazines of the Palace27, and it is especially noticeable in the mudbrick walls of rooms V 3, VI 10, VII 3, IX 1‑2 and XI 1‑7 (fig. 3, 4, 5 and 7a). Stones of small dimensions (5‑10 cm) are stuck into the PEM layer and form a roughly regular line within it, functioning as stone wedges and increasing the stability of the socle. Only in the walls of Area XIII are the mudbrick courses set directly on top of the stone socle, without the use of a PEM and stone wedges levelling layer.

• 28 Tsakanika-Theohari 2006; Tsakanika-Theohari 2009; Tsakanika-Theohari 2017; Shaw 2009: 91‑109; Devol (...)
• 29 Chapouthier and Joly 1936: 9‑10; Chapouthier and Demargne 1942: 9; Pelon 1980: 205.
• 30 Poursat 1992: 28.
• 31 Schmid and Treuil 2017: fig. 232 (0.45 m high), 235 (0.89 m high), 238 (0.97 m high), 239 (0.90 m h (...)

8The incorporation of wood as a structural element in Minoan walls has been stressed in several studies, but surprisingly little indication of its use appears in the mudbrick walls of the Palace at Malia28. Three gaps in the North wall of rooms XIII 1 and 2 – 0.40, 0.16 and 0.45 m‑wide, respectively – may have accommodated vertical supports for a wooden structure, perhaps framing the mudbrick wall in the south-eastern entrance of the Palace (fig. 6)29. Impressions of vertical wooden supports are also visible in the mudbrick walls of rooms IV 9 and V 3, but the examples are quite rare when compared to the large number of preserved mudbrick walls in the building (fig. 13). The scarcity of indications of the mudbrick-wood association in the Palace may be due to the fact that a wooden framework was set higher up on the wall, while mudbrick walls in the edifice are rarely preserved above 0.80 m‑high. In this case, a parallel should be drawn with Quartier Mu, where the wooden framework is set high (at mid‑level30) in the mudbrick walls31. However, it is perhaps worth stressing that even the 1.93 m‑high standing Prepalatial mudbrick walls of room I 1 in the Palace show no indication of the use of wooden elements (fig. 16).

• 32 For the size of the inclusions, see Krumbein 1938.

Table 2 — Description of mudbrick walls components and features in the Palace at Malia32.

• 33 Pelon 1966: 1008‑1011; Shaw 2015: 54, 57, 58, fig. 2.24; Devolder 2016.
• 34 Pelon 1966: 1008, fig. 1; Pelon 1980: plan 16.
• 35 Chapouthier and Charbonneaux 1928: 7; Pelon 1966: 1008‑1011, fig. 2‑6; Pelon 1980: 19, n. 3, pl. 19 (...)

10The walls of room I 1 or ‘casemate’ were formed during a succession of construction phases during the Pre‑ and Protopalatial periods and incorporated different building techniques33. Largely hidden within Protopalatial layered-rubble masonry, the Prepalatial mudbrick walls form a rectangular-shaped structure with a broadly east‑west orientation slightly different from that of the other walls of the Palace (2.5° more to the north‑west) (fig. 14 and 15)34. Three of the four walls of the structure are bonded (the westernmost one is badly preserved). A foundation deposit including a Vasiliki teapot and several juglets discovered in a narrow chasm between rubble and clay walls sandwiched in the south‑eastern part of the mudbrick structure indicates an EM IIB date35. The Prepalatial mudbrick walls, perhaps already hardened by some fire destruction, were later incorporated in the Proto‑ and Neopalatial Palace. They constitute the earliest standing architectural remains and the highest preserved mudbrick walls (1.93 m) in the building.

• 36 Chapouthier and Charbonneaux 1928: 36‑38; Chapouthier and Demargne 1942: 1‑5; Pelon 1980: 198‑203.
• 37 Pelon 1980: 6, 198‑200. This stone shed was erected in 1931 and taken away when a new, non‑invasive (...)

11The rooms XI 1‑7 or ‘Eastern Magazines’ belong to the ‘First’, or Protopalatial, Palace at Malia (fig. 17 and 18). They show a series of low sandstone platforms covered with plaster upon which storage vases were placed, so that the liquid that leaked out of them could be collected into vases set within the floor36. The mudbrick walls that delimit the rooms provided with such platforms were strongly remodelled after the MM IIB destruction of the Palace. Indeed, during the Neopalatial period they were largely rebuilt using limestone rubble stones and boulders, a reconstruction further blurred by the building of a modern shed on top of the rooms (fig. 18)37.

• 38 Structural bonding is visible in several places, namely the northwestern corner of room VI 10, the (...)

12The walls of Areas II, IV and VI in the north-western part of the West Wing incorporate large amounts of mudbricks (fig. 19, 20 and 21). Their study indicates the strong architectural cohesion of rooms II 1b‑c, VI 6‑7 and 9‑10 which were obviously erected during a single building event38. North of these rooms other mudbrick walls are recorded in the small lobby IV 10 and are characterised by their belonging to one distinct building event. The features of the two mudbrick walls north and south of room VII 3 suggest these belong to the same construction project as Areas II and VI further north (fig. 22).

• 39 Part of a mudbrick wall is still visible to the south‑east of room IX 2, but it is badly preserved (...)

13The ensemble formed by rooms IX 1 and 2 of the Neopalatial Pillar Hall is delimited to the west and south with mudbrick walls carefully built with a stone socle incorporating large – 0.74 by 0.34 and 0.74 by 0.26 m – limestone and conglomerate stones and boulders (south wall of room IX 2) (fig. 23a)39. Despite the burden of modern conservation, red mudbricks incorporating large quantities of seagrasses are visible on top of the stone socle, separated with intermediary mud mortar and PEM levelling layers. In the easternmost part of the south wall of room IX 2, however, a niche, or window, is filled with mudbricks of a markedly distinct composition and layout from the rest of the wall (fig. 23b). The specimens are set in headers, stretchers and upright in the wall, and are produced in a yellowish red fabric incorporating a variety of inclusions, such as fine gravel (1-15 mm), sherds (up to 6 cm), straw, chaff, and plaster fragments, but no seagrasses. The difference observable in mudbrick colour and macroscopic composition suggest that this ‘fill’ corresponds to an alteration of the wall’s initial state.

• 40 These were interpreted as high‑standard storerooms by the excavators, after the discovery of golden (...)
• 41 Devolder in preparation.

14Area XIII to the south‑east of the Central Court is made up of rooms XIII 1 to 5 and staircase XIII a‑b (fig. 24)40. The structural connections between Area XIII and the Neopalatial sandstone façade to the south‑east of the Central Court indicate that the mudbrick walls were erected during the Neopalatial period, when the Palace extended significantly towards the south41. The standardisation of the mudbricks’ fabric and dimensions, as well as the characteristics of the masonry, are evidence that these walls were part of a single building programme.

• 42 Guest-Papamanoli 1978: 6; Devolder 2005-2006: 69; Lorenzon 2017. Despite recurrent references to se (...)
• 43 Devolder 2005-2006: 69.
• 44 On this subject, see Lorenzon 2017 and Veropoulidou 2019.

17Macroscopic observation of the mudbricks used in the walls of the Palace at Malia has shown that their macro-fabrics (i.e. macroscopic fabric composition) range between the two extremes of a broad spectrum, from fine to coarse. Mudbricks with a fine macro-fabric, usually reddish in colour (2.5YR 4/6 and 4/8), are produced with a screened sediment (gravel max. 15 mm) mixed with large quantities of marine seagrasses, containing minimal other inclusions (fig. 26a). Mudbricks with a coarse macro-fabric are more varied in colour (generally 2.5YR 5/6 and 5/8) and are manufactured with sediments which have not been screened so that they present numerous large inclusions, such as obsidian, bones, shells, plaster or ceramic fragments (fig. 26b‑e). The macro-fabric of these mudbricks presents fewer seagrasses than the fine macro-fabric ones. Seagrasses have been identified in the bricks on several Minoan sites. They belong to different families of marine weeds, the Posidoniaceae and the Zosteraceae, incorporated in the matrix as vegetal temper42. At Malia only Posidonia oceanica has been identified, but on other sites Zostera marina is present as well43. It might seem surprising that the Minoans used a vegetal temper that grew in a salty environment. Numerous salt efflorescences recorded in the geoarchaeological samples on top of the seagrass impression indicate that the seagrasses were not washed in fresh water before they were incorporated in the matrix. This is also confirmed by the presence of numerous small marine gastropods in the bricks, which were attached to the seagrasses during mudbrick preparation44. Rather, the presence of salt must have been desired, as it helps prevent weeds from growing inside the mudbricks. Mudbricks with a fine micro-fabric are usually better preserved than those with a coarse macro-fabric. Clearly, they erode less easily and in some cases minute details related to their fabrication are still visible on their surface, such as the fine vertical grooves left by the wooden mould when it is removed from the mudbrick. Chaff is also attested in the mudbricks, but rarely, and when employed it is in small proportions combined with seagrasses.

• 45 Wright 2005.

18The mud mortar and PEM interspersed between the courses of mudbricks and in levelling layers under and between them is of a crucial importance to the stability of the wall and it needs to be stronger than the mudbricks45. The matrix used in the Palace at Malia is generally slightly lighter in colour than the mudbricks, ranging from red to reddish yellow (2.5 YR 5/6 and 5/8 and 5YR 5/8) (fig. 27). It is composed of sediments of both fine and coarse grain size (i.e. very fine and fine gravel [max. 80 mm]), but small sherds are often incorporated as well. Seagrasses are used as vegetal temper in the PEM that creates a levelled surface on top of the stone socle, although always in smaller proportion than in the mudbricks. When mud mortar is used as a binder between mudbricks, there is little to no evidence of vegetal temper in the form of seagrasses.

• 46 Wright 2005; Kemp 2000.
• 47 Wright 2005: 93‑95.
• 48 Fodde et al. 2014.
• 49 Devolder 2005‑2006.
• 50 Devolder 2005-2006: 68‑69. The incorporation of seagrasses was observed in EM IIB, Protopalatial an (...)
• 51 Shaw 2006: 199‑229.
• 52 Aurenche 1981: 70; Kemp 2000: 92; Wright 2005: 93‑94.

19The presence of a mud plaster is essential to the preservation of mudbrick walls, since it protects from erosion, humidity, and parasites46. Where it is preserved, the mud plaster is composed of a sediment matrix ranging from red to reddish yellow (2.5YR 4/8, 7.5YR 7/6) with large quantities of straw (up to 60%), in stalks 3 to 10 cm‑long (fig. 28a). The presence of the fibrous material reinforces the structure of the mud plaster. By helping the drainage of excess water outside the revetment, it also reduces the risk of it shrinking and cracking47. Evaporation of groundwater and rainwater also affects earthen architectural structures, as walls can be easily deteriorated by salt crystallisation, which causes weathering and consequent loss of material48. Quite relevantly, only in some rare instances were seagrasses identified in the mud plaster, where their presence seems accidental49. The marked difference in composition between mudbrick and mud plaster is especially useful when collapsed fragments are found, since the presence of seagrasses or straw may help to distinguish the two categories of earthen building material50. Multiple layers of mud plaster are recorded in situ, indicating the presence of successive coatings. They show the gradual refining of the mud plaster composition, with a thick layer (3 to 10 cm) of coarse mud plaster used to even the surface of the wall, on top of which one or several fine layers of mud or lime plaster are applied (fig. 28)51. These finer layers were to be maintained regularly to protect the wall efficiently, likely once every two or three years according to ethnoarchaeological sources52.

• 53 Emery and Morgenstein 2007; Goodale et al. 2011; Goren et al. 2011.
• 54 Jolliffe 2002.
• 55 Hein et al. 2004; Tung 2005; Rumsey and Drohan 2011.
• 56 Bullock et al. 1985; Quinn 2013; Friesem et al. 2014a.

21Geochemical analyses included pXRF and XRD. The pXRF analysis was conducted using a Thermo Scientific Niton XL3t portable X‑Ray Fluorescence Analyser in order to evaluate trace elements in the mudbrick recipe53. The samples were analysed for 120s in soil mode and multiple measures were recorded for each sample. The results of the quantitative analysis have been investigated statistically with both Principal Component Analysis (PCA) and bivariate scatterplot using software R (3.2.2.). All data had been normalised before statistical analysis54. XRD analysis was conducted at the Demokritos Institute in Athens with a Siemens D 500 diffractometer using Cu Kα radiation in the interval 3/60° 2θ at a rate of 3s per step in order to show the mineral composition of the mudbrick samples55. In parallel with geochemical analyses, thin section petrography was conducted in order to evaluate manufacturing techniques and consistency of fabric types. Petrographic and micromorphological observations were made with a LaborLux 12PolS polarizing light microscope (magnification 5x, 10x and 20x). Petrographic thin sections were investigated in order to identify the presence of vegetal, mineral (e.g. quartz), organic (e.g. shells) and human-induced tempering (e.g. lime) during mudbrick manufacturing as a diagnostic characteristic of different mudbricks’ recipes. Microphotographic analysis focused on determining issues related to manufacturing practices and their changes over time56.

• 57 Goodale et al. 2011; Hunt and Speakman 2015.

23of their reliability in pXRF analysis57, with the two main principal components (PC1 and PC2) explaining 73.2% of the total variability (table 4). The resulting PCA graph shows a clear overlapping in raw material sources between samples of the Pre‑, Proto‑ and Neopalatial periods (fig. 29). Indeed, although the number of samples is limited, the quantitative analysis implemented with the confidence ellipses (95% confidence) indicates a general pattern of consistency of raw material sources over time. However, the Neopalatial mudbrick sample MA8 is an outlier in regard to the main sample cluster (fig. 30). This may reflect distinct production phases of mudbricks of slightly different composition during separate building events within the Neopalatial period. PCA analysis is thus relevant in highlighting firstly, the continuity of soil procurement source throughout the use of the Minoan Palace at Malia; and secondly, the existence of discrete building events within the Neopalatial programme of reconstruction of the edifice.

• 58 Aghia Varvara deposits are calcareous formations typical of the eastern Mediterranean geological la (...)

24Bivariate analysis was performed by isolating the values of Fe, Mn and Sr. Sr is associated with carbonated rocks, while Fe and Mn refer to the non‑carbonate content, such as ferromagnesian silicates (i.e. olivine) (fig. 32 and 33). This analysis highlights the small degree of variability between the samples as far as raw source procurement is concerned and indicates the exploitation of the local sediments around Malia (fig. 31). Indeed, the iron‑rich alluvium deposits around the settlement are the main constituent of the mudbricks, with a smaller percentage of marine alluvial sediments originating from the area west of the site, i.e. carbonate rocks of the Aghia Varvara formation58 and talus cones deposits.

• 59 For the methodology, see Bullock et al. 1985; Matthews 2010.

31Quartz grains visible in the red matrix are present in substantial quantities, which is characteristic of raw‑material procurement in alluvial deposit. The presence of human-induced tempering (i.e. lime fragments), often added together with vegetal temper, was employed to prevent shrinking. In MA14 the presence of grog is particularly evident, and it is very likely that it indicates the incorporation of reprocessed archaeological material, i.e. pottery or mudbrick fragments (fig. 35a). The minerals are generally medium in size (0.5 mm < size < 0.1 mm), poorly sorted and with only a limited presence of bigger grains. The coarse fraction comprises 40% of the fabric and the fine fraction never exceeds 60%. The numerous voids indicate the average percentage of vegetal temper in the recipe, namely 30‑40%, an estimation based on point counting during the analysis of the thin sections59. The voids are mainly elongated (approx. 40%) with a few vesicular voids and primarily pertain to the loss of vegetal temper. MA5 is an exception to the main fabric, since it presents a calcarenite fabric, with a brownish colour and the presence of micritic calcite grains (fig. 36).

• 60 Friesem et al. 2014a; Friesem et al. 2014b; Hamard et al. 2017.
• 61 Cammas 2015; Cammas 2018.
• 62 Friesem et al. 2014b; Cammas 2018.

32The study of the petrographic fabrics also highlights several other aspects of mudbrick manufacturing60. First, the presence of different types of vegetal tempers is indicated by the voids and impressions left by chaff and seagrasses. Voids left by seagrasses are wider and longer than those left by chaff, which are thin and elongated. Chaff voids are mainly visible in the Pre‑ and Protopalatial samples, whereas seagrasses are clearly visible in all the Neopalatial samples (fig. 37a‑b). Second, micro‑fissures visible in the microphotographs were caused by the progressive evapotranspiration of water‑lain sediments, as the seagrasses inside the bricks caused long‑term water-retention61. Third, rotating features were caused by kneading during the manufacture (fig. 37c‑d) 62. Finally, fine fabrics tend to show a more homogeneous matrix. Coarse fabrics, on the contrary, show fewer traces of pugging and kneading, and present a clearer separation between fine and coarse fraction, both features perhaps linked to a hastier manufacture.

• 63 Lorenzon 2017; Lorenzon forthcoming.
• 64 Friesem et al. 2014b; Cammas 2015: 60‑63; Lorenzon 2017.
• 65 Lorenzon 2017.

33In earthen architecture, voids and sediment patterns are significant elements that make it possible to investigate the use of vegetal temper or manufacturing practices such as kneading and pugging. For instance, vesicles and irregular planar voids visible in the microphotographs are created by the interaction of water with the soil material, including the process of evapotranspiration, while the elongated regular planar voids are caused by fibrous vegetal temper. Comparing chaff and seagrass voids, the latter are often more than 5 cm long and wider, which is consistent with the macroscopic observation conducted on the mudbricks and reflects the incorporation of Posidonia oceanica in the mudbrick mixture63. Seagrasses, which have also been identified macroscopically in the samples analysed (fig. 26a), have the capacity to retain water for a long period of time and release it slowly, thus accounting for salt crystallisation occurring during evapotranspiration of the mudbricks, as well as for the presence of water‑lain features visible in the matrix64. The Pre‑ and Protopalatial samples (MA1, MA11 and MA12) present a similar percentage of vegetal temper (30‑40%) to the Neopalatial ones (40%). It is not consistent, however, with the data available in the settlement of Malia, where an extended investigation of vegetal temper highlights that the use of seagrasses increased progressively in the Neopalatial period at the expense of straw65.

• 66 Oliver 2008: 98.

34Geoarchaeological analysis of a selection of mudbrick samples in the Palace at Malia demonstrates that raw materials were collected locally throughout the Pre‑, Proto‑ and Neopalatial periods. Sediments with illite-muscovite clay‑beds were favoured, which are common in the terra rossa alluvium of the Malia plain. Marine sediments and other local geological formations, such as deposits of talus cones and limestone formations were also exploited, although in a more limited way. Raw materials can be linked to deposits inside a 2‑km radius from the core of the settlement (fig. 31). Preferred use of alluvium for mudbrick manufacturing is easily explained as it contains a favourable ratio of clay, silt and sand suitable for mudbrick production with a fine fraction accounting for 40% of the total matrix66. However, one cannot entirely rule out the possibility that tiny processed fragments of lime plaster were deliberately added to the mixture in order to harden the bricks and make them more resistant to erosion (MA6 and MA11) (fig. 35). Similarly, processed fragmented mudbrick and/or pottery was identified in the shape of grog used as mineral temper in several samples (MA4, MA5, MA9 and MA14) (table 5 and fig. 35).

• 67 Lorenzon forthcoming.
• 68 Pacheco-Torgal and Jalali 2012; Lorenzon 2017: 275.

36Vegetal temper was procured locally by collecting seagrasses, which are naturally deposited by tides along the nearby shore. An increase in the quantities of seagrasses incorporated in the mudbricks from the Pre‑ to the Neopalatial period is noted, although the limited number of pre-Neopalatial samples must be acknowledged. The use of seagrasses as vegetal temper reflects the increasing exploitation of a readily available raw material that significantly enhanced the performance of mudbricks. Posidonia oceanica released the water content in the mudbricks slowly, thus preventing them from shrinking too abruptly while drying. Also, the fibrous nature of the seagrass increased the tension and compressive strength of the mudbricks, and therefore their load-bearing capacity67. Salt efflorescences and small marine gastropods observable microscopically in the mudbricks suggest the seagrasses were incorporated in the matrix unwashed, with the presence of salt increasing the water-retention and thus slowing down the drying process. Additionally, the presence of salt water and the natural properties of Posidonia oceanica as a fungicide may also have prevented vegetation and microorganisms from growing in the unfired mudbricks68. It is noticeable that seagrasses were found in lesser proportions in the mudbricks produced in a coarser fabric, a feature perhaps in part responsible for their poorer state of preservation.

• 69 [T]he fermentation produces lactic acids that make the brick stronger and less absorbent, Fathy 196 (...)
• 70 Specimens entirely detached from the walls showed an upper, more regular surface, and a lower rough (...)
• 71 Devolder 2005-2006; Nodarou et al. 2008; Lorenzon 2017.

37Once the raw materials had been procured and, for some of them, processed, they were mixed together with water. Studies on modern mudbrick production have suggested that the blend was left to ferment for a week or more, kneaded daily so that it would release lactic acid to enhance the matrix cohesion and thus the mudbrick performance69. Petrographic analysis demonstrated that traces of pugging and kneading are more apparent in specimens with a fine macro-fabric (fig. 37c‑d). The roughly standardised dimensions of the mudbricks produced during each building event indicate that moulds were used in manufacturing, and that a maximum of two recurring mould sizes were used in each wall or architectural unit (table 6). The vertical impressions left on the side of the bricks attest the use of wooden moulds, while the raised upper edges on the top of the vertical surfaces indicate that the bricks were not regularised by any tool after the mould was removed70. Similar characteristics are typical of Minoan production and are indeed well represented on other sites71. The moulded bricks were then left to dry for a few days, depending on the weather conditions. The deformation of the mudbricks indicates that they had not dried completely before they were incorporated in the Palace walls (walls in rooms I 2‑3, V 3 and VI 10, fig. 7 and 16). It is most likely that the manufacturing took place in an area that combined proximity to the raw materials and a source of water, as well as large availability of space for drying the mudbricks. Mudbrick production may very well have occurred in the immediate vicinity of the Palace: raw materials would be procured from the nearby alluvial deposits, beaches and rivers, and the mudbricks left to dry in its West and Central Courts.

• 72 It is worth underlining here that the opposite conclusion was reached regarding the procurement of (...)

38Mudbrick walls served different purposes in the Palace of Malia: coherent architectural units formed by mudbrick walls sharing similar masonry features and functioning together in terms of structure and configuration; isolated mudbrick partition walls erected for blocking or narrowing accesses within a pre-existing plan; and late additions to already existing groups of mudbrick walls for filling gaps or repairing earlier masonry (table 6). During the Neopalatial period, for which the record is notably more abundant, mudbricks illustrate a marked consistency in dimensions within the same unit or partition wall. In contrast, differences in dimensions between the different architectural units and amongst the partition walls or additions are pronounced (tables 2 and 6). This indicates the use of distinct moulds in the different areas of the Neopalatial Palace and suggests that mudbricks were not mass-produced to be used indifferently throughout the edifice but were produced separately by builders, who erected their own mudbrick architectural unit or isolated wall72. This hints at a high degree of segmentation of the building project, i.e. its completion in separate building events, an impression further enhanced by the diversity in mudbrick fabrics and mudbrick walls constructional features in the Palace (table 2). It is worth noting that while load-bearing underperforming partition walls or additions were generally erected with mudbricks in a coarse macro-fabric, fine macro-fabrics were preferred in the case of larger mudbrick architectural units (with the exception of Area II/VI/VII 3, table 6). However, the type of macro-fabric used is unrelated to the features of the walls’ masonry. For instance, walls provided with features that enhanced their structural efficiency (e.g. a stone socle, intermediate levelling layers, or the avoidance of running bond) could be using mudbricks in one or the other macro-fabric (compare for example Area II/VI/VII 3 and Area XIII).

• 73 This is for example illustrated by sample MA2 which is a statistical outlier in PCA and bivariate a (...)

39Consistency in dimensions, fabric and constructional features within each mudbrick architectural unit stands out from the marked distinction between the different areas supposedly erected during the same building episode in the Neopalatial period. Macroscopic observations of the mudbrick walls highlight their marked diversity within the Palace, in terms of bricklaying and mudbrick fabric and dimensions. These suggest that despite the coherence of the Neopalatial Palace master plan, the edifice was built in separate building events most likely fulfilled by distinct work parties (Area II/VI/VII 3; room IV 10; rooms IX 1‑2; and Area XIII) (fig. 38 and table 2). Each building team would have been responsible for the entire construction process of a given architectural unit or isolated wall within the edifice, from the procurement of raw materials to the production of mudbricks and the actual building of the wall(s). Small variations in composition between mudbricks from the same architectural unit also indicate that in some cases the mudbricks could be produced in several batches before they were used in construction73.

• 74 Devolder 2019c.
• 75 On the discovery of standing and collapsed mudbrick walls and partitions in the town of Malia, see (...)

40Standing architectural remains in the Palace show that mudbrick building tradition in Malia goes back to the beginning of the settlement’s history from the middle of the 3rd millennium BCE onwards74. The extensive use of this building material on the site is best exemplified in Protopalatial Quartier Mu but it is also illustrated, when the conditions of preservation permitted it, by many other buildings75. The abundance of local building resources most likely encouraged the extensive use of mudbrick in architecture, perhaps also enhanced by the structural advantages of the material highlighted in the introduction to this paper.

• 76 Chapouthier and Joly 1936: 1‑3.
• 77 Chapouthier and Demargne 1942: 9.
• 78 Chapouthier and Charbonneaux 1928: 19, 23, 30‑33.

41Although the masons favoured coarse mudbricks for blockings and partition walls, it seems that the use of finer mudbricks with enhanced performance is unrelated to the structural needs of the building: ground floor walls that supported an upper level were made of both high‑quality mudbricks (Areas IX76 and XIII77) and mudbricks which performed less well (Area III-IV-VI78) (fig. 38). Similarly, we doubt that aesthetic considerations played a role in the choice of a specific mudbrick macro-fabric, since the mudbrick walls’ components were hidden behind a mud or plaster coating (fig. 4, 8, 9, 21 and 28). Rather than structural and aesthetic reasons, we argue that the varying skills of the builders determined the use of varying macro-fabrics.

• 79 On building forms and techniques created over time by the collectivity and giving birth to local ar (...)
• 80 It is however important to beware of an extreme polarisation between skilled and less‑skilled build (...)

42Extensive use of mudbrick architecture on the site must have contributed to the deve­lopment amongst its inhabitants of the practical knowledge necessary to produce structurally efficient mudbrick walls, i.e. embedded skills acquired through observation and participation in building projects79. The detailed macro- and microscopic observation of mudbrick architecture in the Palace at Malia indicates that the workforce involved in the construction of the mudbrick walls was most probably drawn from the site’s inhabitants, whose skills must have varied markedly. This would explain the variety in mudbrick masonry, composition and performance throughout the edifice, including within the same building episode. While some architectural units seem to have been erected by true master builders, others attest the work of less‑skilled builders80.

• 81 On the debate regarding the socio-political organisation of Minoan Crete, see for example Evans 192 (...)
• 82 Devolder 2018: 362; Devolder 2019a.

43The segmentation in the production of mudbrick architecture within the Palace at Malia did not hinder the general cohesion of the building, and the involvement of distinct building teams appears ancillary to the sophisticated configuration of the Neopalatial complex. Although defining the nature of the leadership that transcended the involvement of distinct teams of builders is arduous81, the architectural integrity of the Neopalatial Palace must not be blurred by the partitioning of the building process. This argues for a centrally-managed architectural development and a craftspeople-based production complemented by a less‑ or unskilled workforce. The diversity in the practices and skills – and perhaps also status – of the agents involved in the materialisation of the Palace’s coherent master plan nevertheless shakes our preconceptions about the degree of control and uniformity of the human and material resources invested in such a monumental building project. This diversity is further exacerbated if one takes into consideration other masonries in the Palace. Recent research has put forward the involvement in the Neopalatial reconstruction of the edifice of Knossian builders specialised in cut‑stone and wood-and-cut-stone masonries82. Participation in the same project of local mudbrick builders with varying degrees of competence thus completes an already composite picture of the Neopalatial Palace’s building materials and techniques.

• 83 Hastings and Moseley 1975; Moseley 1975; Cavallaro and Shimada 1988.
• 84 Schmid and Treuil 2017: 137, fig. 222‑223.
• 85 These are bricks with crosses incised and in relief in Houses Ab (room 15) and Ac (room 17) in Gour (...)
• 86 Hazzidakis 1915: 118‑119, fig. 6.
• 87 Olivier 1980: 219, n^o 201.
• 88 Devolder 2018.

44If separate teams of mudbrick builders were involved in the construction of the Neopalatial Palace at Malia, it is worth stressing the absence of visibly intentional markers that would make it possible to distinguish the production of each one of them, such as the marks illustrated in other crafts in Crete and on mudbricks in other civilisations83. This is a general trend on the site, where even the best‑preserved mudbrick remains of Protopalatial Quartier Mu have produced only the impressions of goat and human feet84, although marks on mudbricks were found in other sites on Crete and in the broader Aegean85. There is one striking example though, of an eight-branch star set in relief on a brick (0.50 by 0.35 by 0.10 m) retrieved by J. Hazzidakis during the first excavation campaign of the Palace in 1915 (fig. 39)86. It is said to come from the southern part of the Western Magazines, although the precise location of the discovery is unknown. It was suggested that the sign was created when the mudbrick was pressed, still wet, on a block bearing a deeply incised masons’ mark87, the like of which are carved on the ashlar sandstone blocks in the Neopalatial Palace88. This latter interpretation seems doubtful, since mudbricks were laid in the walls with at least a dried external surface, and although they could bend under stress, they were no longer plastic enough to take on the shape of the surrounding wall components. Unfortunately, the loss of this specimen makes further study and interpretation impossible. The general absence of makers’ marks on the mudbricks of the Palace at Malia may be a further indication that each building team was in charge of the whole construction process, from mudbrick manufacture to actual construction, thus making it unnecessary for mudbrick makers to mark their own production.