2. Was the Colossus of Rhodes Cast in Courses or in Large Sections?

  • Ursula Vedder, Kommission für Alte Geschichte und Epigraphik des Deutschen Archäologischen Instituts, Munich


The paradoxographer Philo of Byzantium (De septem mundi miraculis 4) claimed that the Colossus of Rhodes was cast in situ in horizontal courses buried gradually by an earthen embankment. The study of large-scale ancient bronzes and foundries, however, has provided evidence only for casting in large sections. Here it is argued that the technology used for the Colossus was no exception.

How do we reconstruct the fabrication process of an exceptionally large and lost statue? First, the general parameters for the working steps must be known. Therefore, the indirect lost-wax process in antiquity is compared with better-known methods used to create two extant colossal statues, the Great Buddha in Nara (cast in courses) and the Bavaria in Munich (cast in large sections). Then the various steps of the working process attested in ancient times are examined. The analysis reveals basic differences between the three casting methods.

Philo’s text contains a certain level of technical knowledge but lacks important details and indeed states an important falsehood. A possible explanation for this discrepancy is that he used a written reconstruction of a working process. This means that we can give the archaeological evidence greater weight than this text.

In the argument over where the monumental statue known as the Colossus of Rhodes was located, its casting has hitherto seldom been considered. The statue—made of cast bronze with a height of 70 cubits (30–35 m, or 98–114 ft.)—must have left at least some remnants of its production. Furthermore, these remnants must be discoverable in the city of Rhodes. However, a requirement for recognizing these remnants is knowledge of the casting technique that was used. For this reason, it is worth undertaking a reconstruction of the manufacturing process of this lost statue.1

Philo of Byzantium (De septem mundi miraculis 4) wrote the longest text describing bronze casting to survive from antiquity, and also left a description of a casting technique said to have been used for the Colossus.2 His text has been given a lot of authority in our scholarly tradition. In this paper, a more analytic approach is taken: it is a technical-archaeological commentary on the text by Philo, verifying whether the details he mentioned are compatible with the ancient situation.

Philo claims that the Colossus was cast in a different way than normal statues. He explains: “For the individual metal sections could not be moved.”3 But given that the monumental architraves of the Hellenistic temples in Asia Minor are heavier than any large bronze piece, this is clearly not true.4 Further, he writes: “after the first part had been cast, the second was modeled upon it, and for the following part again the same method of working was adopted…. After the casting of the new course upon that part of the work already completed…. the artist heaped up a huge mound of earth round each section as soon as it was completed, thus burying the finished work under the accumulated earth, and carrying out the casting of the next part on the level.” Thus, according to Philo, the Colossus of Rhodes was cast in situ in horizontal courses buried gradually by an earthen embankment.

Fortunately, a lot of bronze-workshops in the city of Rhodes have been uncovered and provide good information about the casting process used on the island at the time of the Colossus.5 They prove that, contrary to Philo’s report, colossal statues were cast in large sections.

The largest known casting pit was excavated at the southern slope of the Acropolis in the city of Athens, on the Mylonas property. It was part of a workshop that operated at the beginning of the third century BC, the time when the Colossus was manufactured. It measures at its center 7 meters long and 3.25 meters wide and it is 3.6 meters deep (22 x 11 x 12 ft.). G. Zimmer reconstructed in it a colossal upper part of a male body.6 Thus he proved the existence of tall statues cast in large sections up to 15 meters (49 ft.) high. This is an archaeological proof, as no statue of these dimensions has survived.

Could the Colossus of Rhodes—a statue twice the size of the one Zimmer reconstructed—also have been cast in large pieces? How realistic is the method described by Philo? As no ancient examples remain, two extant colossal bronzes will help to answer these questions. They provide the general parameters of the working process and the keywords to be searched for in the ancient literary sources.

The first and older of the two was brought into the discussion by D. Haynes in 1992.7 The Great Buddha Vairocana in the Todaiji monastery in Nara (Japan) is 14.96 meters (49 ft.) high and was manufactured in AD 743–57.8 The statue was cast in a method similar to that described by Philo and proves that this method is technically possible. The second statue, the Bavaria at the Ruhmeshalle, was completed between 1837 and 1850 in Munich.9 Its large cast pieces were assembled into a figure 18.52 meters (60 ft.) tall. These sculptures, though enormous, are roughly half the size of the Colossus. As a sitting figure, the Buddha at least has similar proportions to the standing Colossus. Contemporary documents relating to the manufacturing process exist for both the Buddha and the Bavaria.

Five working steps are necessary for both life-size statues and for extremely large-scale statues such as the Buddha and the Bavaria.

Figure 2.1. Colossus of Rhodes, master model. Drawing based on picture by L. von Schwanthaler (base and workers), statue of Aias (body), and Rhodian coin (head) Image: © Ursula Vedder

The first working step is the construction and modeling of the master model. Colossal statues need a model with the same dimensions as the bronze statue to be cast. And the Colossus of Rhodes was no exception, as illustrated in the author’s schematic drawing (fig. 2.1). The master model provides certainty that the statics of the statue are well done and that after assembly the large sections will fit together.10

The Bavaria and the Buddha both had a master model consisting of a wooden scaffold covered with materials to provide a surface for modeling in clay. A wood sheathing for this purpose is recorded in Munich. In Nara, the walling of the model still consists of wood, linen, and clay. In both cases, completion of the master model took several years of the decade or more it took to complete the statue. A Japanese text records 426 days just to make the clay surface of the Buddha’s master model.11 It remained inside the statue, forming its core. Of relevance here is the fact that it was built around a central mast of the same height as the statue.

The existence of master models in antiquity is proven by the monumental chryselephantine statues of the fifth century BC. Consider, for example, the Athena Parthenos, 26 cubits (ca. 13 m) high, which was in effect a master model covered with precious materials. Traces of her base are preserved in the Parthenon. In its naos, a hole still is visible where the central mast of her scaffold was affixed. Thus its construction was similar to that of the Buddha in Nara.12

Pheidias, the master of the Parthenos, also made the Athena Promachos, which stood on the Acropolis near the Parthenon. The statue was made of cast bronze, and its height is believed to have been between 9 and 15 meters (29–49 ft.).13 It is notable that the ground plan of the statue’s base again has a hole in the center (fig. 2.2). Following the arguments of Palagia, the statue was cast in the same period as the Parthenos was manufactured. The Promachos was dedicated to the main deity of the polis of Athens and stood in her sanctuary. Thus the Promachos is a real predecessor of the Colossus of Rhodes. Pheidias built a scaffold in the location of the bronze statue in order to make the master model for the Promachos. Afterward the statue was cast in pieces on the south side of the Acropolis. The finished parts were brought up to the Acropolis and assembled on the base in place of the master model. Thus for colossal statues with a height up to 15 meters, we have not only archaeological remains of casting pits but also references concerning the master model.

Figure 2.2. Ground plan of the platform for the pedestal of the Athena Promachos in Athens (drawing by T. Tanoulas, Palagia 2013, fig. 11) Image: © Tasos Tanoulas, Olga Palagia

Philo does not mention the master model of the Colossus. But the master model of the Colossus commissioned by Nero for his Domus Aurea in Rome is probably the subject of a half sentence in Pliny (Naturalis historia 34.45–46). The statue was indeed constructed and cast imitating the Rhodian prototype.14

Finally, it can be proved that the Rhodians not only had the technological skills to cast colossal statues but also were able to build a sophisticated timber structure. Such a timber structure was known to be present in Rhodes when the Colossus was planned: the war machine Helepolis was built by order of Demetrios Poliorketes during the unsuccessful siege of Rhodes in 304/303 BC and was left behind when he quit the island. Ancient sources record its height as between 86 and 100 cubits (44–50 m, or 144–164 ft.). According to Pliny (Naturalis historia 34.41), it was sold to make the Colossus, the dedication to Helios after the victory. Perhaps the Rhodians used even its timbers to construct the master model.

The second working step was to make the casting molds. For all casting methods, it is common for workers to prepare the mold and the core so as to fashion a cavity for the thickness of the bronze wall, which would then be filled with the molten metal. In antiquity, beeswax was used for this purpose, giving the name to the “lost wax method.” Details of this process need not be repeated here.15

In Japan they used neither wax nor clay to create the cavity for the Buddha. Over three thousand bronze pins with heads about 3 centimeters thick and 12 centimeters square were fabricated and hammered into the surface of the master model. The thickness of the pinheads determined how much of the clay surface needed to be reduced in order to provide the cavity for the bronze. The pins remained in place, becoming part of the statue’s walling, and are visible today (fig. 2.3). The reduced surface of the master model was burned out by fire, thus transforming the master model into the core of a casting form. Around this core, the walling of the statue was cast in six layers, as seen in the schematic drawing in figure 2.4.

Figure 2.3. Statue of Buddha Vairocana in the Todaiji Temple, Nara, Japan Image: © Jakub Hałun (Own work) (GFDL), via Wikimedia Commons
Figure 2.4. Buddha of Nara, reconstruction of the casting mold. (The drawing is a copy of Maeda et al. 1997, fig. 55, omitting Japanese characters and adding numbers and English text.) Image: © Todaiji, Nara, Japan © Ursula Vedder

The Buddha of Nara proves that the casting method described by Philo is technically feasible. But in important details, such as the master model and the use of wax, the description is incomplete and unreliable. Philo also neglects to mention the making of a core. Instead he describes an armature in the inside of the statue, growing higher with every cast course. Hence the inside was thought to be open.

It cannot have worked like that. Without a master model, it would have been impossible to form the course of a statue on the finished section covered with earth. And a sculptor’s copy cage to scale up the object, as Maryon proposed, is not feasible.16 Furthermore, this technique, though used by Roman portrait sculptors, was unknown in the early third century BC.

Turning to our other analog in Munich, the Bavaria was cast in large sections, so the master model was cut into numerous pieces (fig. 2.5). The largest of these are marked in figure 2.6. The fabrication of mold and core was done according to the “sand-casting process,” in which the fabricators used a special sand—a mixture of fire-resistant soils fixed by plaster—instead of clay. Instead of wax, clay sheets were used. And whereas in antiquity the system of funnels, gates, and vents was distributed regularly over the body, in Munich the funnel system was manufactured only above the mold. The channels for the vents were fixed at the lower edge of the mold at the bottom of the casting pit. Mold and core were built up from the bottom of the casting pit in such a way that no chaplets were needed. As illustrated in figure 2.7, the sections of the core can still be seen in the statue’s interior.

Figure 2.5. Statue of Bavaria, part of the Hall of Fame, Munich, Germany Image: © Ursula Vedder
Figure 2.6. Statue of Bavaria. The edges of the large cast pieces are marked Image: © Ursula Vedder
Figure 2.7. Statue of Bavaria. Detail of the interior: edges of the core’s sections visible in the area under the armpit of the left arm Image: © Ursula Vedder

The third working step is the melting of the bronze alloy and casting. These processes are not mentioned by Philo. It seems that he did not understand that Greek melting ovens were small.17 The earthen hill for casting would have to have been more than 70 cubits high, and had to have provided a large surface with space for a great many casting ovens and workers.

In eighth-century Japan, the ovens were larger and took up less space. Both the oven and the mud-brick hill are proved by archaeological finds. In nineteenth-century Munich, they were able to build an oven large enough to melt the metal for one large section at a time.

Working step four is the treatment of the bronze surface after casting. Casting a large bronze in large sections means that the treatment of the bronze surface can be done at the same time as the preparation for the casting of other pieces. In this case, working step five, the mounting of the finished pieces, is the last step.

The Japanese workers, by contrast, had to wait until the casting of the head was complete. Only then could they begin with the removal of the mud-brick hill, the molds, and the surface treatment of the bronze. The mud brick was later spread in the terrain, an important detail for our context: there are no reported finds in Rhodes that can be connected with an artificial hill 35 meters (114 ft.) high.

And another detail is not comparable: in Nara, surviving documents indicate that eight thousand workers and two hundred overseers worked on the Buddha project. It is unlikely that there were that many workers in the Greek context at the beginning of the third century BC.


Summing up, the technique described by Philo is not compatible with the situation proved by archaeology in ancient Greece. An explanation for this discrepancy might be that his text is not a factual account but rather a reconstruction of the casting of the Colossus. The text, or an earlier source from which it was compiled, was written because the original information from the third century BC—the time of Chares of Lindos, the sculptor of the Colossus—was lost. Thus we need no longer credit the idea that the Colossus of Rhodes was cast using a different technique from that attested on Rhodes. The statue must have been cast in large sections.

This understanding leads to the following points useful in the search for the location of the statue and its workshop on Rhodes. We are looking for a workshop in which an extremely large figure was cast in large sections. I believe it was situated near the statue’s base, with space for a workshop to be built and used over a long working period (Pliny Naturalis historia 34.41 suggests twelve years).18 The statue base had dimensions comparable to those of the surviving base of the Colossus of Nero (17.60 x 14.75 m, or 58 x 48 ft.). It was used first for the construction and modeling of the master model and then for the finished statue. The casting pits were nearby. They were not necessarily oval and were wider but not deeper than the pits already known.19

In 2015 I published a proposal for the possible location of the Colossus’s base and workshop in the sanctuary of Helios above the stadium and the odeion on the acropolis of Rhodes (fig. 2.8).20 The ruin northeast of the temple may be the remains of the base and the workshop. In the discussion of this interpretation of the ruin, the casting technique of the Colossus plays an important role.

Figure 2.8. Rhodes, Sanctuary of Helios (said to be of Apollon Pythios) and the possible location of the Helios Colossus (drawing added) Map data: © 2017 Google and Basarsoft


  1. This paper summarizes Vedder 2015, 40–56, figs. 23–55, 60–65.
  2. Vedder 2015, 82–84.
  3. Translation by Haynes 1992, 121–22.
  4. Zimmer 1990, 17 n. 646.
  5. Zimmer and Bairamē 2008.
  6. Zimmer and Bairamē 2008, 38–51, figs. 1–10.
  7. Haynes 1992, 121–22.
  8. Maeda et al. 1997; Rosenfield 2011, 107–15; Vedder 2015, 95–96, 111, figs. 40–48. My gratitude to A. Mori and K. Totsu for important help.
  9. Pangkofer 1854, 21–28; Vedder 2015, 96–98, 111–12, figs. 49–53, 60–65.
  10. Agreeing with Lapatin 2001, 72–73.
  11. Rosenfield 2011, 108.
  12. Stevens 1955, 240–76, figs. 2–5; Lapatin 2001, 63–79; Vedder 2015, 86–87, figs. 33–35.
  13. Zimmer 1990, 62–71; Palagia 2013; Vedder 2015, 85–86, figs. 30–32. My gratitude to O. Palagia for the permission to publish the detail of the ground plan in fig. 2.2.
  14. Vedder 2015, 92–95.
  15. Mattusch 2014, 87–88.
  16. Maryon 1956, 81–82.
  17. Zimmer 1990, 143–52.
  18. Compare Giumlia-Mair 2012, 17.
  19. See the rectangular foundry in Zimmer and Bairamē 2008, 73–77.
  20. The sanctuary is wrongly attributed to Apollon Pythios, Vedder 2015, 57–68.


Giumlia-Mair 2012
Giumlia-Mair, A. 2012. “The Technology of Bronze Statuary.” In Myth, Allegory, Emblem: The Many Lives of the Chimaera of Arezzo, ed. G. C. Cianferoni, M. Iozzo, and E. Setari, 9–27. Rome: Aracne.
Haynes 1992
Haynes, D. 1992. The Technique of Greek Bronze Statuary. Mainz: Philipp von Zabern.
Lapatin 2001
Lapatin, K. D. S. 2001. Chryselephantine Statuary in the Ancient Mediterranean World. Oxford: Oxford University Press.
Maeda et al. 1997
Maeda, Y., T. Matsuyama, S. Hirakawa, D. Nishi, and K. Totsu. 1997. Todaiji Daibutsu no kenkyu: Rekishi to chuzog ijutsu. Tokyo: Iwanami Shoten.
Maryon 1956
Maryon, H. 1956. “The Colossus of Rhodes.” JHS 76: 68–86.
Mattusch 2014
Mattusch, C. C. 2014. Enduring Bronze: Ancient Art, Modern Views. Los Angeles: J. Paul Getty Museum.
Palagia 2013
Palagia, O. 2013. “Not from the Spoils of Marathon: Pheidias’ Bronze Athena on the Acropolis.” In Marathon the Day After: Symposium Proceedings, Delphi, July 2–4, 2010, ed. K. Buraselis and E. Koulakiotis, 117–37. Athens: European Cultural Center of Delphi.
Pangkofer 1854
Pangkofer, J. A. 1854. Bavaria, Riesenstandbild aus Erz. Munich: Georg Franz. http://www.mdz-nbn-resolving.de/​urn/​resolver.pl?urn=urn:nbn:de:bvb:12-bsb10384297-4.
Rosenfield 2011
Rosenfield, M. 2011. Portraits of Chōgen: The Transformation of Buddhist Art in Early Medieval Japan. Leiden and Boston: Brill.
Stevens 1955
Stevens, G. P. 1955. “Remarks upon the Chryselephantine Statue of Athena.” Hesperia 24: 240–76.
Vedder 2015
Vedder, U. 2015. Der Koloss von Rhodos: Archäologie, Herstellung und Rezeptionsgeschichte eines antiken Weltwunders. Mainz: Nünnerich-Asmus.
Zimmer 1990
Zimmer, G. 1990. Griechische Bronzegusswerkstätten: Zur Technologieentwicklung eines antiken Kunsthandwerkes. Mainz: Philipp von Zabern.
Zimmer and Bairamē 2008
Zimmer, G., and Bairamē, K. 2008. Ροδιακά εργαστήρια χαλκοπλαστικής. Athens: Ministry of Culture, Archaeological Resources and Receipts Fund.