1 Introduction
The question of the origin of the classical temple is still a burning one. In recent years, three monographs have been devoted exclusively to this subject (
Barletta, 2001;
Wilson Jones, 2014;
Pierattini, 2022), and this is because, after centuries of controversy, it remains unresolved. The general question about the origin of the temple actually stems from the question about the origin and
raison d’être of some of its formal features that, despite persistent attempts to explain them, continue to defy logic and/or the universal principle of design that Louis Sullivan enunciated as “form ever follows function” (
Sullivan, 1896, p. 408). Among all these formal features, we will focus on the following: (1) The so-called “ornaments” and architectural sculptures are many meters high, where they look small, deformed by a low angle perspective and, in the case of the frieze of the
cella, in very little light. (2) The entablature does not seem to serve any function. Vitruvius’ famous explanation of the origin of entablature―that the Doric frieze and mutules and the Ionic dentils derived from structural parts of the roof (4.2.2)―has been accepted by modern scholars for lack of a better explanation. However, the current academic consensus has rejected it due to many well-founded objections that can be read in
Bowen (1950, pp. 113—114),
Barletta (2001, pp. 130—137) and
Wilson Jones (2014, pp. 63—87) among others. (3) Similarly, the peristasis does not seem to have any particular function, either utilitarian or symbolic. (4) And in the more sophisticated temples, such as the Parthenon, there are very few straight lines, which has been put down to certain optical corrections.
In this paper, we will test a completely original hypothesis accounting for the origin of some surprising formal features of the classical temple, which will also reveal the origin of the temple itself. It has been suggested by what at first glance appears to be a coincidental similarity between two prominent words in the Greek lexicon: the words
naus “ship” and
naos (Attic dialect
neos)
“temple” are so similar that even the genitive case of the former (
naos, meaning “of the ship”) is indistinguishable from the nominative case of the latter.
† So far, linguists maintain that there is no etymological relationship between the two words, and that
naos “temple” derives from a proto—Greek form *
naswo-only hypothetical, unattested, which has no cognates in any other language, Indo—European or not (
Lejeune, 1987, p. 136;
Frisk, 2021, s.v. naos). However, from a formal point of view, it is much easier to derive
naos “temple” from
naus “ship”. The nominative case
naos may derive from the genitive case of
naus as used in the expression
ho naos oikos “the house of the ship (i.e., the arsenal)”, first becoming a masculine adjective (
ho naos oikos “the naval house”) and finally becoming a noun (
ho naos “the naval”, i.e., “the temple”).
† Probably, the relationship has not been formulated until now because etymologists have not seen any connection between the objects to which they refer, the ship and the temple. However, we are confident that this relationship will be revealed in the following pages.
Our hypothesis at first sight will seem outlandish: that the Greek temple is the representation in stone of a war or merchant galley flipped over and stored upon support walls.† The ancient Greek seafarers who spent several months overseas, and camped on the shores of strange lands, needed to keep the hulls of their ships safe from Teredo navalis shipworms, for which they routinely dragged them onto the beach. But they also needed to keep their interiors dry, preventing rainwater from pooling inside them, and to shelter themselves. They would have solved both problems by turning the hulls upside down, placing them on stanchions and sheltering under them. Such temporary shelters could be complemented with a stone fireplace at the entrance for cooking (Fig. 1). Years later, in the Greek homeland and once their service life had ended, these ships could be placed on purpose-built walls with apsidal plan, and adapted in some way to serve as rulers’ dwellings or as gathering and banqueting halls, and eventually as temples† (Fig. 2).
This hypothesis is compatible with Vitruvius’ claim (4.2.1—5) that the oldest preserved temples were the imitation in stone of older buildings constructed in wood, a hypothesis that has its detractors (a summary in
Zuchtriegel, 2023, pp. 1—40) but is being confirmed, at least partially, by field archaeology (
Hellner, 2020). The petrification process is usually considered to be an imitation of Egyptian architecture, but it may also have been inspired by a passage in the
Odyssey in which Zeus encourages his brother Poseidon, the god of the sea, to turn a wooden structure into stone. This structure, though, is not a temple, but a ship. Zeus says: “turn [the ship] into stone near the land,/looking like a swift ship, so that all may marvel” (
Homer,
Odyssey 13.156—158), which Poseidon finally does.
2 Methodology
Our multidisciplinary research approach combines architecture, nautical archaeology and ethnology, and primarily uses methods from the semiotics of visual arts (the comparison between classical temples and Greek upturned galleys), linguistics (the proposal of new etymologies for some architectural terms) and art history (analysis of social, economic, and cultural context). The formal comparison between works of art, potentially endowed with meaning, and objects in the real world to which they may refer is routinely used as a heuristic tool in the semiotic study of iconography (
D’Alleva, 2005;
Huys and Vernant, 2014).
Specialized nautical terms, as they are not commonly known, are defined in parentheses the first time they are used.
3 Ethnographic parallels
Researchers of the origin of vernacular architectures (
Vellinga et al., 2007;
Jarzombek, 2013) have not considered overturned ships as a noteworthy germ of housing, but they have sometimes been used as temporary or permanent roofs, either by camping under them, recycling them or imitating them.
3.1 Camping
In 1916 the members of Edward Shackleton’s failed Antarctic expedition arrived at the frozen Elephant Island in two boats. To shelter, two sailors suggested upturning the boats, setting them on small walls five spans high and using tarps on the sides (Fig. 3); they camped this way for four months (
Alexander, 1998, p. 170). Until recent times, eel fishermen on the east coast of Skåne, Sweden, used to bivouac under their upturned boats (
Svensson, 2020, p. 86). The Inuit and Yupik Eskimo peoples of the Arctic regions of America protected themselves from the summer winds by camping in the safety of their
umiak, the spacious “women’s boats” used for transporting cargo. They upturned them and rested one side on the ground and the other on stanchions or on oars used for that purpose (
Lee and Reinhardt, 2003, pp. 104, 150—151, Figs. 113, 114, 165) (Fig. 4). Other Native American peoples camped in the same way as shown in the oil painting “Voyageurs at Dawn” (1871) by Frances Anne Hopkins.
3.2 Recycling
On other occasions vessels became buildings only at the end of their life, as a form of reuse, with minor or major amputations when necessary. This was the case with the three herring boats that the architect Edwin Luytens turned into sheds on the island of Lindisfarne (England) in 1908, creating a local tradition that persists to this day (Fig. 5). In the first half of the twentieth century, there was a fishing boat converted into a dwelling on the island of Lambay, in the Irish Sea (Fig. 6). And at the turn of the nineteenth and twentieth centuries in Équihen-Plage, on the French shore of the Strait of Dover, there was a whole district called Quartier des quilles en l’air (“Keels to Sun District”) inhabited by fishing families who had built their houses out of discarded boat hulls (Fig. 7).
3.3 Imitating
In some vernacular styles of architecture there are buildings that, despite having roofs in the form of a ship, obviously never sailed. These structures have a naval origin, either because they were inspired by real ships that at one time served as dwellings, or because the forms and procedures developed in naval construction were transferred to architecture on land.
During the Bronze Age there arose, in the eastern Balearic Islands, a culture known as the Talaiotic. One of the most characteristic buildings of this culture between the twelfth and tenth centuries BCE was the
naveta, Catalan for “small boat”, built partially or entirely of stone, whose shape recalled upturned boats and left the imprint of a basilica floor plan on the ground (Fig. 8). In the opinion of archaeologists (
Guerrero et al., 2006, pp. 107—113),
navetes were used often as a dwelling place, and other times as places of collective burial.
According to the testimony of the Roman writer
Sallust, some immigrants who arrived by sea to the coast of present-day Algeria “used as huts the upturned hulls of their ships”, and so in his time, the first century BCE, “still the cottages of the Numidian peasants, which they call
mapalia, of oblong shape and covered by curved gable roofs, are like the hulls of ships” (
The War with Yugurta 18.4—8).
In the North Sea, from the eighth to the eleventh century, the Vikings developed an architecture of boat-shaped long houses of which enough evidence has been preserved to conclude with certainty that they had roofs shaped like upturned hulls (
Komber, 2001) (Fig. 9).
Further afar and centuries later, among the natives of Easter Island, the most common type of house before the arrival of the Spanish was the
hare paenga “house on basalt blocks”, a house for an extended family shaped like a great canoe, lengthened and turned over (Fig. 10), as well as its smaller version aptly named
hare vaka “boat house” (
Budd and Vargas, 1992, pp. 14—17).
For their part, the Torajans of the island of Sulawesi (Indonesia) still build their traditional houses, the
tongkonan, with an oversized roof in the shape of a boat. According to legend, they reached this island in ships which, when they decided to settle permanently, they used as their new homes (
Julistiono and Arifin, 2005, p. 2667).
† The ancient Balearic, the Numidians, the Easter Islanders, the Vikings, the modern Eskimos and European Atlantic seamen shared with the ancient Greeks their dedication to seafaring and a close familiarity with boats. The Balearic and the ancestors of the Numidians, moreover, may have come into contact with the Greek seafarers of the Early Iron Age (c. 1100—700 BCE), either directly or through the Phoenicians. Thanks to their example we have converted into likelihood the idea, initially outlandish, that the ancient Greeks were able to turn over their ships and take refuge under them. Next, we will investigate whether they really did so.
4 Doric entablature and its naval counterpart
Maritime commerce was one of the most lucrative economic activities of the Greek Mycenaean and Iron ages, and that which gave rise to Greece’s resurgence from the eighth century BCE onwards. Thanks to iconographic sources, we know well the form of the penteconter, the most usual galley of the time, that was propelled by fifty oarsmen, even though nautical archaeologists have not yet excavated a single wreck of this type of vessel.
In the broken piece of pottery in Fig. 11, from the Late Helladic IIIC (circa 1200—1090 BCE), we see firstly that the oarsmen of the represented galley are arranged on a single line. Secondly, that above them is a screen that runs from bow to stern, a bulwark. And finally that between the hull and the bulwark is a gallery of windows, one per rower, through which the oars protrude and the rowers can be seen. The same configuration can bee seen in the freeboard (the portion of the side of a hull that is above the water including the bulwark at the top) of the ships of the Sea Peoples on the relief of the naval battle of Temple in Medinet Habu, Egypt, from 1186 BCE (
Wachsmann, 1981, Fig. 12; 1995, pp. 23—33).
At a later date, the bas-relief in the palace of Senacherib in Nineveh (Iraq) that depicts the evacuation of King Luli of Tyre in 701 BCE (the bas-relief being dated shortly after) shows Phoenician ships of two different types, both with two lines of rowers on either side of the ship. One of the two types clearly shows the same structure as the previously mentioned one (Fig. 12). In addition, from the upper bulwark hangs a row of circular shields.
Lucien Basch described years ago (
1968;
1987, pp. 428, 432—435) a clay model of a galley taken out of the water west of Cape Maleas in the nets of a fisherman from the village of Gytheion, in the Peloponnese (Greece) (Figs. 13 and 14). In the absence of an archaeological context, the piece must be dated according to typological criteria. Basch considered it to be from the Roman period, partially because, in making his dating, he focused on iconographic parallels showing a row of shields hanging from a continuous bulwark, and found them in depictions of Roman galleys from the first century BCE and the first century CE. However, he did not mention the clear Phoenician parallel in the bas-relief of Sennacherib’s palace from the beginning of the seventh century BCE, although he discussed it in detail in the same work (1987, pp. 311—318). He also underestimated other clues. He did not take into account that many Greek ships of the Archaic period had rows of circular shields, although in iconography they seem to hang from bulwarks that are not massive (1987, Figs. 404, 460, 504, 506, 510). On the other hand, he himself pointed out that the large curtain hanging from the stern structure of the Gytheion model has very close parallels in Phoenician and Greek ships of the Archaic period (1987, p. 434), and is absent from Roman ships (1968, p. 157), which clearly points to the earliest dating. More recently Olaf
Höckmann (1995, p. 211 and n. 23), on the contrary, convincingly maintained that the model represents a vessel from the Archaic period, i.e., the seventh and sixth centuries BCE. If so, it is therefore contemporary with the mentioned Phoenician ship type to which it so closely resembles and with the first Doric temples.
The Gytheion clay model is similar in many respects to the two models mentioned above. Although it is not visible at first glance, on the bulwark there is a row of small holes in which, in all probability, were attached circular shields (
Basch, 1987, p. 428) like the ones depicted in the Phoenician galleys from the palace of Senacherib. But the clay model also provides us with a greater detail, and adds a structure that is not present in the previous ones: it includes an outrigger (a spar that extends beyond on the side to move away the fulcrum of the oars) on each side of the hull (
Casson, 1971, p. 84), which is partially broken and missing in the model. A minor clarification, perhaps unnecessary: in both the bas-relief and the clay model, the craftsmen reduced the number of rowing windows per side (to five and six respectively) to simplify the design. A penteconter with two lines of oars per side had around thirteen windows on each side.
Now if we turn the three mentioned ship models upside down, their freeboard turns out to be identical to the entablature of a Doric temple in the following nine features, and possibly in one more (see numbers in Fig. 14 and a Doric entablature in Fig. 16). (1) The upper elongated screen, the bulwark, turns into the architrave, (2) the windows of each rower’s post become the metopes, inside which men are depicted acting, and (3) the stanchions that separate the rowers’ windows and hold up the deck and the bulwark become the triglyphs. Moreover, (4) the molding which, in the Gytheion model, is the protrusion on the freeboard of the deck, turns to be identical to the taenia which separates the frieze and the architrave.
† (5) The remains of the angled projections which support the outrigger, partially broken in the Gytheion model, seem to be at the origin of the mutules. (6) The reconstructed circular shields fixed to the bulwark with pegs have exact parallels in the architrave of some temples, such as the Parthenon. (7) The slight curvature of the gunwale (the upper edge of the side of a ship) is almost identical to the very slight horizontal curvature of the architrave of the most sophisticated temples. (8) The rope belt that surrounded the hull of narrower galleys may be the origin of the decoration of guilloches and meanders in the cornice in some entablatures. (9) The side edges of the flat bottom of the hull, which pushing the water outwards create a small wave on each side, have a counterpart in the cornice, where there is a line of antefixes whose shape resembles splashes of water. And finally, although with less certainty, (10) the holes through which the oars of the lower line came out may be the origin of the elusive
viae mentioned by
Vitruvius (4.3.6). We will take a closer look at some of these features.
4.1 Metopes and triglyphs
As for the rowers’ windows, which provided ventilation and visibility, they presumably were closed with sliding or removable wooden boards (
pinakes in Greek) in order to protect the oarsmen in combat and reinforce the structure. These
pinakes must be the origin of the metopes. Both in the Phoenician ships and in the Gytheion clay model a
dikrotos (“with two lines of oarsmen”) penteconter is represented. The oars of the lower line protrude from the hull through small openings which in Greek were called, in the singular,
ope “opening, hole”. The compound term
met-ope, which etymologically means “between the openings” (
Frisk, 2021, s.v. metope), can refer to the oarports (holes in the side of a boat’s hull through which the oars pass), indicating that each window of the upper oarsmen is interspersed in the space between two oars of the lower line.
† The surface of the wooden
pinakes used for closing the rowers’ windows could be painted to depict the oarsmen behind them, or idealized versions of them: heroes from mythology. Subsequently, when the ship would be upturned to convert it into the roof of a permanent building, the
pinakes could be flipped over so that the image was not inverted.
In
Euripides’ tragedy
Iphigenia among the Taurians, two characters, Orestes and Pylades, want to sneak inside a temple to steal the statue of the goddess, and Pylades says: “Look for a gap between the triglyphs through which/a body passes” (vv. 113—114). Undoubtedly the expression “between the triglyphs” refers to metopes and, since they give access to the interior of an enclosed space, it does not refer to metopes of the entablature of an open peristasis, but of a closed
cella. Moreover, either the space between the triglyphs is open (
Guadet, 1894, p. 343), or it has a fragile sealing material that can be removed, such as a wooden board. Two art historians of the late nineteenth and early twentieth centuries, Julien
Guadet (1894, pp. 341—347) and Leicester
Holland (1917), saw the origin of the Doric frieze in rows of windows as well: at the top of an enclosing wall the first, and at the top of a
chemin de ronde the second. Although their hypotheses were more reasonable than the one by Vitruvius, they only managed to explain the openings of a frieze located at the top of a closed
cella, but failed to explain why an architect would place a row of windows on top of an open peripteral colonnade, and their proposals were overlooked.
No model is found in ancient naval iconography that reveals the origin of the Ionic entablature as well as the Gytheion model does with the Doric entablature. However, it is very easy to derive it from the freeboard of the same Archaic penteconter that gave rise to the Doric entablature. The three fasciae or bands of the architrave in the Ionic order may correspond to a bulwark of a tapering thickness, as represented in Fig. 15(a) and 15(b), or to a clinker-built bulwark (method of building in which the edges of hull planks overlap).
With regard to the two variants of Ionic entablatures, which lack a frieze or have a continuous frieze, the sequence of events that gave rise to them may have been as follows. The vertical stanchions connecting the hull to the deck and the bulwark could break once the ship was turned over, and this risk would be avoided in two ways. The first would be to eliminate the gap between the two structures by cutting off the stanchions and resting the hull on the bulwark (Fig. 15(c)). The stanchions may have been cut into small pieces and used as transition pieces between the bulwark and the hull, giving rise to the dentils. Two passages from Vitruvius confirm this possibility: he informs that the space between dentils “is called metope in Greek” (3.5.11), exactly the same as the space between triglyphs in the Doric frieze (4.2.4). This first configuration was habitual in the Ionic temples of Asia Minor, such as the Temple of Zeus at Priene.
The second way would follow the opposite path, and would consist of filling the gap with a continuous screen. In the triremes of the Classical period, oarsmen in the upper rows were very exposed to the elements, as it is seen in the famous Lenormant Relief. In combat, protective curtains of hide or sailcloth, called
pararrhymata, were deployed on both sides (
Casson, 1971, ch. 5 n. 57, p. 249). The Ionic frieze would represent the deployed curtains covering the rowing windows, and could be left either smooth, sculpted or painted with scenes representing the warriors who rowed behind them (Fig. 15(d)), similarly as had been done in the metopes of the Doric frieze. This second is, for example, the configuration of the frieze of the North Portico of the Erechtheion.
4.2 Mutules
There is no known etymology of the Latin word
mutulus (
Ernout and Meillet, 2001, s.v.). In our opinion this architectural term can be a variant of the adjective
mutilus “mutilated”, belonging to the lexical family of
mutilo “to cut off, to mutilate”.
† Although non-attested, etymologists reconstruct in Vulgar Latin a word *
mutilio which, on the one hand, meets all the requirements to be a derivative of
mutilo (in this case its original meaning would be “mutilation”) and, on the other hand, is widely considered to be the origin of the Italian
modiglione (
Meyer-Lübke, 1935, s.v. *mutilio and mutulus), an architectural “ornament” very similar to the
mutulus.The adjective mutilus “mutilated” had, more specifically, the meaning of “dehorned”. In the case of the architectural element, it makes sense for mutulus to designate the remaining part of a horn-like projection that held the outrigger and had been sawed off during the conversion of the ship into a building. The guttae of the mutuli would represent the remains of the pegs or nails that would fix to the hull this piece of union. As chance would have it, in the clay model from Gytheion, a part of the outrigger broke off and split projections remained attached to the hull (Fig. 14, no. 5), to which the adjective “dehorned” can be perfectly applied. This explanation, however, poses a difficulty: the pieces connecting the outrigger to the hull are not very wide, while the mutuli on the cornice are wider than the space between them.
4.3 Hanging shields
Alexander the Great dedicated in the architrave of the Parthenon shields of gold from the spoils taken from the Persians after the Battle of Granicus (
Arrian,
The Anabasis of Alexander 1.16.7). The marks left on the stone by the shields are still visible. We have restored them in place in Fig. 16. Shields from other battles were hung on the entablature of the Temple of Apollo in Delphi, as attested by the Greek authors
Aeschines (3.116) and
Pausanias (10.19.4). Although it was habitual for citizens to dedicate objects of varying value to the gods in the temples (the temple itself is an
anathema, a “dedication” to the gods), these objects were kept inside the temples. The placement of the shields, in both Athens and Delphi, is striking, and coincides exactly with the placement of the rowers’ shields―hanging outside the bulwark―in the Phoenician galleys represented in the palace of Senacherib (Fig. 12), and with similar holes in the bulwark of the clay model from Gytheion which, according to
Basch (1987, p. 428), were intended to hang miniature shields (Fig. 14, no. 6). At this point, it is appropriate to refer to an illuminating etymological question. We call a
gallery a row of windows similar to the metopes of the Doric frieze. At the beginning of the seventeenth century the Spanish lexicographer Sebastián de
Covarrubias wrote, about the etymology of the word
galleries: “They could be called like this because of their length and the similarity of the fenestration
to the pavisade† of the galley” (1611, s.v. galerías; our italics), which fits perfectly with our hypothesis. Current etymological dictionaries, though, attribute other origins to the word.
In the peristyle, both Doric and Ionic, of some Roman
domus of the first century CE and the first half of the second stone features known as
oscilla “swings” were hung from center of the intercolumniations. The origin and function of such objects has been interpreted in various ways, but as early as 1881 Maurice Albert identified two types of architectural
oscilla as shields of two types (
Taylor, 2005). The
tondo type would be a derivative of Greek circular shields. Specimens of this type were found
in situ in the peristyle of the House of the Gilded Cupids at Pompeii, and depicted attached to an Ionic entablature on a fresco in the Villa of P. Fannius Synistor in Boscoreale. The
pelta type, to which the specimen shown in Fig. 17 belongs, unmistakably represented the light shields which the Romans attributed to the Amazons.
† Hanging a stone ornament over a passageway is very risky, and architects would have avoided it unless there was a good reason to do so. Although it has been assumed by some scholars (
Taylor, 2005, p. 93) that the function of these shields was votive, Elena
Roscini (2013, pp. 249—255) has clarified that the term
oscilla is used in archaeological literature to refer to very different objects, only some of them of a religious nature. As for the function of the
oscilla of the intercolumniations, she thinks it was merely ornamental. In our opinion, however, it was rather figurative: on penteconters the shields would be attached to the bulwark by their straps and, once the ship was turned up, they would hang down. So, in Roman peristyles shields hung by their straps from an entablature that had once been the bulwark of a ship.
There was another relevant type of architectural oscillum called in Greek pinakes, which were square or rectangular boards painted with figurative scenes. This third type may well derive from the pinakes that would close the rowing windows of the penteconters to protect the oarsmen and reinforce the structure, which, as we have assumed above, were presumably removable. They were thus a kind of hanging metopes.
4.4 Longitudinal curvature
In both ancient and modern wooden ships there are very few straight lines beyond the longitudinal centerline of the keel, which forms the ship’s axis of bilateral symmetry. The high number of curved lines is the result of design decisions forced by the physics of buoyancy and navigation. The slight curvature known in ship design as “sheer” affects the shape of the gunwale and, therefore, the bulwark and the deck. This curvature is clearly visible in the clay model of the galley from Gytheion (Figs. 13 and 14, no. 7).
A few Greek temples, the most refined ones, have curvatures where a modern architect would use straight lines. One of the most well-known is the curvature of the entablature, very slight: in the temple of Aphaia in Aegina, for example, it is up to 3 cm in 24 m, a proportion of 1/800 (
Bankel, 1993, Tables 56—58).
† This curvature is replicated in the stylobate and in the roof (Fig. 18, with exaggerated proportions). Vitruvius (3.4.5) and modern scholars consider it a refinement for correcting a supposed optical distortion which would result if straight lines were used. But the truth is that present-day architects, when they design similar rows of columns, do not feel the need to correct any optical distortion,
† which may well be imaginary. In fact, according to
Rankin (1986, pp. 38—39), only the thickening of the angular column of the peristyle is due to an optical correction.
The curvature of the entablature coincides exactly with the naval sheer. The curvature of the central part of the ship is minimal, and more pronounced in the bow and stern, but these two parts are absent in the upturned penteconter of split ends. We can deduce, then, that in the temple, the primary curvature is that which affects the entablature and the roof equally, and that the curvature of the stylobate is merely a consequence of the first. And also that the curvature of the entablature is an architectural sophistication, but not an optical or aesthetic one, but rather of the mimesis “imitation”, an accurate representation of its naval model.
On the penteconter there were obviously only rowers’ windows port and starboard, not in the prow or stern, therefore there should only have been windows in the long sides of the temple, as in Fig. 1. But in temples with peristasis the entablature goes round the entire building, including the shorter sides. This forces us to reconstruct a ship longer than the temple, of which the bow and the stern would have been sawn off while almost the entire length of the freeboard would have been preserved, as shown in Fig. 19. The parts of the freeboard that would protrude forward and aft would later be folded inwards in order to enclose the two short sides of the building, thus creating the base of the pediment. As a result, the entablature of the short sides would have the same entablature with the same longitudinal curvature as that of the long sides.
4.5 Guilloches and meanders
The longest and narrowest galleys had a belt of twisted thick ropes (
hypozomata “under-girdle”) that dampened the bending of the hull caused by the passage of the waves (
Casson, 1971, pp. 91—92). The under-girdle surrounded the hull all around its perimeter at the freeboard level, either above the lines of rowers, as can be seen in Fig. 20, or below them, as in another contemporary frieze (1971, fig. 125). The variant of under-girdle that surrounded vertically the stern or the bow and then ran amidships is also well attested (1971, p. 435), as in the model of Gytheion (Fig. 14, no. 8).
Therefore, the perimetral version of this naval gear may be the origin of the line of single and double guilloches and meanders that is often represented on the outer face of the geison, as the one in Fig. 21; this particular guilloche can represent two intertwined ropes, each dyed a different color to differentiate them, and the central circle the head of a bronze nail that fastened them to the hull. Meanders can also represent ropes twisted around nails fixed to the hull and intertwined with each other, but in a rectilinear style characteristic of the Geometric period. Anyway, at this point we must be cautious, because guilloches and meanders have had a long tradition as decorative elements in other arts, as on pottery.
4.6 Antefixes
On any boat the hull pushes the water aside as it moves forward, creating a small side wave (Fig. 15(a) and Fig. 14, no. 9). In the temple, the cornice is the edge where rainwater is evacuated from the roof. Along the cornice are lines of antefixes, ornaments of volutes and palmettes whose shape resembles small breaking waves (Fig. 22). Although the comparison may seem forced, the Greek lexicon confirms its validity: the word
kyma meant “wave”, “fetus” and “young sprouts of plants” (
Liddell and Scott, 1940, s.v.).
† The three meanings are represented in the antefixes: from bottom to top, the wave in the two volutes, the fetus on the male or female young face, and the plant shoot on the palmette, which makes the antefix an emphatic significant for the meaning “wave”, as if to say, in the style of Gertrude Stein’s famous poem, “a wave is a wave, is a wave”.
†Another fact confirms the identification. The largest wave caused by a ship moving forward is at the bow, in front of the ram (in the case of a warship equipped with this feature) or the cutwater. On the roof of the temple, the bow and stern have hypothetically been sawn off, and the closest point to the missing ram or cutwater is the apex of the pediment, where the acroteria are. Actually, although the Greek word
akroterion had a generic meaning of “extremity”, the dictionary informs that it referred specifically both to the “ornament of ship’s stern- or stem-post” and to the “ornaments placed on the angles of a pediment” (
Liddell and Scott, 1940, s.v.). In that place we find, among other figurative motifs, large volutes and palmettes which, like the antefixes, resemble the shape of a splash of water but on a larger scale (Fig. 23). In many Laconian temples, such as the temple of Hera at Olympia, the acroteria were large discs with concentric circles and an outer rim of projections that may be a representation of a bow wave, which often forms a thin, almost complete circle of water (Fig. 24).
Marilyn
Goldberg (1982, p. 196) quantified the motifs represented in the archaic acroteria known up to 1982. According to our hypothesis, two-thirds of them (161 out of 237) are related to the prow of galleys: the volutes/palmettes and the discs for the reasons already explained, the horses and riders because of the strong rocking motion experienced at the end of a ship’s prow, reminiscent of galloping, and the Victories with clothes tightened by the wind as auspicious bow figures, as evidenced by the example of the Winged Victory of Samothrace. One-third of the motifs (76 out of 237) remains unrelated to the bow for now: the sphinxes, other groups of figures, and the gorgons.
As for the two angular acroteria, they have to correspond to the cat-heads (a beam at each side of the bow) where the two main bow anchors were raised and secured, and which were also used as an offensive weapon to break the oars of enemy ships (
Casson, 1971, p. 85, fig. 107).
Classical temples had acroteria at the apex of the two pediments, on the east and west sides. This does not pose a big problem for our interpretation, since all rowing boats can move forwards and backwards, simply by the rowers turning around in their seats. Although there have been ships with a bow at each end in various nautical cultures (
Wachsmann, 1998, pp. 166—167), it does not appear that they were built in Greece during the Archaic and Classical periods.
4.7 Viae
In a very brief passage of disputed meaning, Vitruvius (4.3.6) refers to some elusive
viarum directiones “alignments of the paths”, located under the cornice and aligned with the center of the metopes, which scholars usually identify with the space between the mutules (
Morgan, 1910, pp. 2—4;
Rowland and Howe, 1999, p. 221). In Latin via had among others the meaning of “channel, duct” (
Glare, 2012, s.v. via 3), and so it could be used to refer to the openings in the hull of the lower line of rowers (Fig. 14, no. 10) which were called
ope “hole” in Greek. However, Vitruvius’ reference is unique among ancient sources, and there is no way to verify the accuracy of this hypothesis.
4.8 Other features
Other features of the Doric entablature do not challenge our hypothesis. The triglyphs’ two vertical grooves appear not to have had any function other than to reduce the weight of a wooden piece without compromising its solidity. As for the guttae, it has long been said that they represent wooden pegs or nail shanks whose heads have been removed, a view still held by current scholars (
Ruggieri, 2022, p. 59). Finally, regulae may be pieces of reinforcement. Grooves, nails, and reinforcements are routine in wood carpentry.
Other plausible connection between temple entablatures and oared ships is the standardized way in which the temples’ proportions were calculated from minimum units that Vitruvius called modules. One of these modules, closely connected with the frieze of triglyphs and metopes, was intercolumniation (Vitruvius 3.2.5 ff.). This practice may have come from naval architecture where it is justified on strictly functional grounds. In the design of oared ships the basic unit for calculating the length of the whole ship is the
interscalmium, the space between the thole-pins (pins set upright in the gunwales or equivalent surfaces to serve as a fulcrum in rowing) of two oars, that defines the size of the “room” or space occupied by each oarsman. The
interscalmium of the ancient oared warships measured two cubits (Vitruvius 1.2.4), i.e., 0.98 m, the Doric cubit being 0.49 m (
Morrison, 1991). This gives rise to an estimate for the single line penteconter of about 25 m for the central rowing section, plus a few meters more at the bow and stern, giving a total of about 36.5 m (
Casson, 1971, pp. 54—55). If one of its ends were cut off, it would reach a length of 30.75 m or 100.88 feet, virtually the same length as the ubiquitous Archaic temple type known as
hekatompedos naos or “one hundred feet temple” (see Fig. 28).
The Gytheion model carries, hanging from the stern, a large piece of cloth like a curtain. Whatever this curtain is,
† it is strongly reminiscent of a component of the Panathenaic procession in which Athenians carried a large
peplos “mantle” to the Acropolis for the goddess in the Parthenon. One aspect of the ritual may have a strong bearing on our hypothesis: the
peplos was carried to the Acropolis in the Panathenaic Ship, hung like a sail from its yardarm (Strattis,
Makedones fr. 31
Storey;
Mansfield, 1985, pp. 6—7, 55).
5 An objection: double entablature and peristasis
If the explanation of the origin of the classical entablature given here is true, the presence of an inner entablature and its frieze at the top of the cella in temples with peristasis (Fig. 25) supposes a difficulty: if the entablature was made with the freeboard of an upturned ship, then would a building with two entablatures, an inner one over the cella and an outer one over the peristyle, have been done with the freeboards of two ships?
At the present time, due to the lack of rigor in the primary excavations and the construction of many temples on the remains of others, it is difficult to know what the first temple to have a peristasis was. In the opinion of the most skeptical scholars there is no certainty that any temple from the seventh century BCE had it (
Wilson Jones, 2014, pp. 48—53). However, the origin of the peristasis can be explained independently of the date and placewhere it arose, attending to its shape in the preserved temples of the sixth and fifth centuries BCE.
In architecture, the usual procedure is to first construct the walls and then the roof, its dimensions in agreement with the space it is intended to cover. However, in the construction of walls that are to support an upturned ship, the order is inverted: the ship was built earlier, and the walls later, to fit the ship that already existed. This way of proceeding inevitably led to a long-term problem. The passing of time and the action of the elements would end up ruining the ship’s hull (Fig. 26(a) to 26(b)). One solution would be to cover the open space with a standard gabled roof (Fig. 26(c)). Another one, to cover it with the upturned hull of a new ship preserving, whether for reasons of economy or respect, the supporting walls and the remaining parts of the first ship. But for this purpose it would not be at all easy to obtain a ship of the same dimensions as the previous one, while a smaller ship would be completely useless. If the open space was eventually covered with a larger ship (Fig. 26(d)), the measurements of the new “roof” beyond the dimensions of the existing building would force the creation of a perimetral colonnade in order to support the freeboard of the cantilevered ship; thus creating, at the same time, an empty perimetral space with no practical function in itself. The preservation of the freeboard of the first ship would explain the presence of the entablature above the cella of the previous SHIP.† The passing of time and the action of the elements would eventually ruin the hull of the second ship, as had happened with the first one, and the architects would end up settling, this time, for erecting a standard gabled roof (Fig. 26(e)).
5.1 Objections to the hypothesis of the enlargement
The explanation just given for the creation of the peristasis runs into a difficulty. The
cella of the temple of Poseidon at Isthmia (mid-seventh century BCE), possibly one of the first to have peristasis (
Wilson Jones, 2014, p. 50;
Fehr, 2015, p. 68), had a width of 5.75m, which coincides with the presumed width of a merchant penteconter (
Casson, 1971, p. 158). But the combined width, including the peristasis, measures up to 14 m, and in that period in Greecegalleys of such beam were not built.
A finding made in the sanctuary of Hera at Samos provides the solution to this difficulty. Archaeologists there unearthed a statue of an Archaic
kouros, from the sixth century BCE and therefore contemporary with the first buildings of the sanctuary that had peristasis. The Kouros of Samos, 5.25 m tall, would, with the missing feet, be three times the height of a person. Its excavator wrote that “the gigantic size of such statues may be linked to the conception widely expressed in Greek literature ever since Homer, that the heroes of earlier times were much larger than the ‘people alive today’” (
Kyrieleis, 2005, p. 120). From the late eighth century BCE numerous fossil remains of megafauna, much larger than human bones, were considered relics of mythological heroes and displayed in sanctuaries (
Mayor 2011, pp. 104—129, table 3.1). It is reasonable to deduce that, in making the enlargement proposed in Fig. 26(d) and 26(e), the architects did not use real ships. Since they presumably wanted to represent the bigger ships of the Heroic Age, they would enlarge the pre-existent buildings as in Fig. 26(d) with complete false wooden hulls; or perhaps directly as in Fig. 26(e), with only their freeboard.
As for the timing, there is debate among archaeologists as to whether some of the more ancient buildings, such as the Megaron B at Thermon (ninth century BCE) or the Hekatompedos I at Samos (eighth century BCE), were enlarged with a peristasis while the interior edifice, the cella, was still standing. Be that as it may, most stone temples with peristasis were not built in two stages, but rather in only one that included an interior cella and a peristasis from the very beginning. Although this apparently rejects the idea of enlargement, the truth is that many temples were built on an emplacement in which there had been similar previous constructions. Under these conditions, it is plausible that the configuration inner cella + peristasis was a way of simulating a two-phase construction history in order to show the antiquity of the building, giving rise to a style that can be called historicist.
The suitability of the idea of enlargement is confirmed by one more piece of evidence: the plans of some temples contain up to three simulated stages of construction. The enormous Temple G at Selinus (520—409 BCE) can be interpreted as a succession of three buildings (Fig. 27): the smaller of the three would be supported mainly on columns (as in Fig. 1 of this article), and would represent the temporary visit to the site of the first pioneers; the next one would be supported on walls (as in Fig. 2), and would represent the settlement of the site by its first stable inhabitants; the largest one, finally, would be an enlargement of the second (as in Fig. 26(e)) and would represent the continuity and success of the community over time. Similarly, there was a naiskos (a “little temple”) in the center of other major temples composed of cella and peristasis, as in the open courtyard of the temple of Apollo at Didyma.
6 Beyond the entablature
No other part of the Greek temple is as revealing of its naval origin as the entablature, but we must review others parts to check if any seriously contradicts our hypothesis; or, in the contrary, reinforces it.
The pediment’s triangular space would represent the ship’s cavity. Many overseas voyages in the Bronze Age were made to obtain metals. In the Archaic period, wealth was often accumulated in the form of bronze objects, such as cauldrons and tripods,† and later in the form of bronze sculptures of anthropomorphic gods. In fact such sumptuous objects were exhibited to the public in temples, in the room called opisthodomos (“back room”), and in the smaller temple-like structures known as thesauroi “treasuries”. The obvious interpretation of the pediment group of sculptures is, so, that it represents the precious cargo carried by the penteconter: bronze and even gilded statues of the gods, which eventually were petrified as was the whole ship.†
In the single-room temples its main cult statue was at the back of the room, the place that in our hypothesis corresponds to the bow of an upturned ship. Ancient ships carried their two bow anchors here, as well as what was known as the
hiera ankyra “sacred anchor” in Greek (
Lucian,
Fugitivi 13). It weighed twice as much as the others and its use meant the difference between life and death in desperate situations. Hypothetically it was anthropomorphized and may have given rise to statues such as the Victory of Samothrace and other similar ones attested by iconography on the bow of many ships (e.g.
Casson, 1971, fig. 107). When the ship was upturned, the sculptured sacred anchor would “fall” into the place corresponding to the bow of the ship, i.e., the bottom of the
cella. There is evidence of the association of anchors to divinity, either as a votive element or as an aniconic representation of it, in the ancient Mediterranean (
Tito, 2018, pp. 9—42).
Regarding both constructions, the bunkhouse formed by the upturned ship resting on stanchions and the stone fireplace for cooking (Fig. 1), the first would hardly leave a trace of its post holes in the soil; the latter, however, is more durable. This is presumably the reason why in various archaeological sites the stone altar precedes the first archaeologically recorded buildings (
Mazarakis Ainian, 2016, p. 20) by decades or even centuries. In the sanctuary of Hera at Samos (Fig. 28) the altar may date from the end of the Mycenaean period, and the first temple from the eighth century BCE (
Kyrieleis, 2005, p. 100). This sanctuary is located about 150 m from the beach, and there is archaeological evidence that a wooden ship was dedicated to Hera at the end of the seventh century BCE (
Wallinga 1993, pp. 49—51, Figs. 4 and 7(b)). The ship was very similar in size to the oldest temple, the
hekatompedos located immediately west of the altar, and to each of the three segments of the South Stoa, all of which were standing in the seventh century BCE (Fig. 28, in orange). And it was oriented, like the oldest temple and the later ones, as well as other buildings of unknown function, in the direction of the sanctuary’s main altar. All this allows us to hypothesize that the place, in addition to serve as a sanctuary since Minoan times (
Niemeier and Maniatis, 2010, pp. 106—108), served as a beaching place for a fleet (in fact, the first entrance to the sanctuary was from the coast-―
Kyrieleis, 2005, pp. 103, 107), and that it was monumentalized in the sixth century BCE, at the same time that the island flourished and became an international mercantile center whose fleet traded throughout the Mediterranean.
†Scholars agree that the stone temples from the seventh century BCE and later are the continuation of previous constructions from the Mycenaean (c. 1750—1100 BCE) and the Early Iron Ages (c. 1100—700 BCE) excavated in various locations in Greece and its vicinity (
Mazarakis Ainian, 1989). Little is known of these other than the floor plan (a few examples in Fig. 29): many of them have a basilica plan, with an apse at one end and the entrance at the other, often
in antis, and in some of them the long walls are not completely straight, but slightly convex. Greek oared warships were entered from the stern, so all these features roughly match the contour of a ship’s hull at gunwale level. This plan, which was popular during the Mycenaean Period, regained its popularity during the Early Iron Age (
Mazarakis Ainian, 1997, p. 112), immediately before the first temples were built. A few architectural clay models (reproduced in
Wilson Jones, 2014, p. 41) give us information about the roof of these buildings, but none unequivocally supports our hypothesis: those of Perachora, very fragmentary, and Nikoleika have a high roof that can be the rough representation of a ship’s hull, but not others. The measurements of the largest plans of these buildings, which range, with the odd exception, from 20 to 35 m long and from 4.5 to 7 m wide (see for penteconters
Casson, 1971, p. 158;
Rankov, 2013, pp. 90, 93—94), are also consistent with the hypothesis that some supported an upturned ship’s hull. However, it would be totally naive to think that all or many of them did so: some are located at a high altitude above sea level (Ano Mazaraki, 1100m), others we know for sure were of excessive size and had a timber frame with reed thatched roof (Heroon at Lefkandi;
Coucouzeli et al., 2024), and so forth.
7 Discussion
One could object that the hypothesis set out in the preceding pages must be confirmed archaeologically in the field before being accepted, that a building from the Geometric or Archaic period must be excavated with remains of a wooden entablature or roof that have unmistakable signs of belonging to a ship that sailed: its shape, traces of salt, caulking or of shipworm Teredo navalis… But such a critic would be objecting to a statement that we have not made in these pages. In previous section on “Ethnographic parallels” we distinguished three ways in which builders crafted buildings of naval origin: camping temporarily under ships, recycling them as roof of permanent buildings, or imitating their shape in the roofs of buildings constructed from scratch. In these pages, we have demonstrated that entablatures of Greek stone temples from the seventh century BC onwards belong to the third type. Buildings of similar origin constructed by the Numidian, the Vikings, the Easter Islanders and the Indonesian Torajan (some reconstructions shown in Figs. 9 and 10) would not have those unmistakable elements, because what came from the sea was their shape, not any of their physical components. Furthermore, archaeology already plays a central role in validating our hypothesis, since both the pictorial and sculptural representations of galleys and the entablatures of temples are archaeological remains, and their dating and formal comparison are routine procedures within the discipline of archaeology, which is not restricted to fieldwork. The ancient Greeks likely represented overturned stone ships as roofs of temples to indicate that real wooden ships were recycled as roofs in the past. There may even be one or more such “missing links” somewhere beneath Greek soil, waiting to be unearthed by archaeologists conducting fieldwork, but the truth is that, at present, there is no basis for stating this. It is also possible that such ships recycled as roofs only existed in oral tradition.
It could also be argued that such overturned ships did never exist because ancient literary sources, prominently including the treatise by Vitruvius, say nothing about them. But the following remarks discard it. We would be dealing with a case of fallacy
a silentio: silence proves nothing because often what is known to the sender and the receiver is left unsaid for communicative economy. Furthermore, experts agree that only between 1% and 5% of the literary works of classical antiquity have been preserved (
Gerstinger, 1948, among others). Finally, some preserved texts may have been misinterpreted and will only reveal their true meaning when examined in light of the hypothesis presented in these pages.
Lastly it could be argued that “extraordinary claims” have been made here that “require extraordinary evidence”. However, epistemological philosophers repeatedly point out that this famous aphorism was coined, albeit in different words, by David Hume in the seventeenth century to refute belief in miracles, and that it is often misapplied in our days to unexpected but rational claims that fall outside the mainstream assumptions (
Deming, 2016, p. 1320). The quality of “unexpected”, moreover, is not a scientific category, but a psychological reaction to novelty (p. 1322). In scientific controversy it is reasonable to accept a claim, no matter how extraordinary it is, “as long as it provides the best explanation of the evidence available” (
Schick 2002, p. 332), which we believe applies to the research presented in these pages.
8 Conclusions
From what has been discussed in these pages, it can be deduced that the hypothesis put forward in the introduction corresponds to the facts: the entablatures of Greek temples of the Doric and Ionic styles from the seventh century BCE onwards represent in stone the freeboard of upturned merchant or war galleys used as the roof of communal buildings, which is why this type of building was given the name ho naos (oikos) “the naval (house)” or, in other words, “the temple”. Our thesis answers the four apparent inconsistencies listed in the introduction. (1) The supposed “ornaments” of these temples are high up because that is where the ship to which they belong is located. (2) The function of the entablature and each of its parts is not aesthetic (nor originally structural, as Vitruvius claimed with regard to some of them), but representative, because in reality the entablature is, in its entirety, a sculpture, that is, a signifier that refers to a meaning, the ship.† (3) The peristasis originally had no function, but was an unintended result of the enlargement of a simpler original building. However, its preservation when it is no longer necessary also makes it a signifier, in this case of the building’s construction history. (4) Finally, the longitudinal curvature represents the ship’s sheer, and therefore fulfills the same representative function that inspires the entablature as a whole. Additionally, this thesis provides a rational explanation for the strong tendency toward bilateral symmetry in Greek temples, which was functional in the ships from which it originated, and remained a constant feature of Western monumental architecture until the twentieth century CE.
Unveiling that the supposed ornaments do not seek to produce aesthetic pleasure, but are in fact sculptural elements with a representative function, removes the Greek temple from the exceptional space it has mistakenly occupied in the history of architecture, and makes it intelligible. This building also complies with the rule that “form ever follows function”, but along a slightly more tortuous path: nautical requirements determined the shape of the ship and its freeboard, and then the architects’ desire to represent it in stone, upturned at the top of the temple, determined the presence in that place of nautical and sculptural structures that we have called “ornaments” until today.
This conclusion is consistent with and therefore reinforces the validity of the thesis we ourselves presented a few years ago on the Greek and nautical origin of Buddhist architecture, which began in India in the second century BCE and thereafter (
Ciordia, 2020).
Further research should focus on the possible application of the naval hypothesis to the interpretation of other parts or the classical temple, specially the shape of the columns and capitals, their number and distribution.† Archaeologists should review reports from past excavations or initiate new ones in search of any traces of temporary or permanent buildings that may have had an upturned ship as a roof, both in overseas colonies and on Greek soil. Philologists should incorporate this thesis into the range of possibilities available for interpreting passages, some of doubtful meaning and others apparently not, from ancient written sources. The lack of positive results in these investigations, however, would not invalidate the conclusion we have reached in this work, which stands on its own merits due to its internal consistency and its explanatory power of the data it handles.
2095-2635/2026 The Authors. Publishing services by Elsevier B.V. on behalf of KeAi Communications Co. Ltd.
PDF
(23130KB)
