Stylobate and Roof in the Olympian at Akragas

as Tu N Stylobate and Roof in the Olympieion at Akragas In: American Journal of Archaeology, 84 - 1980 DIS De 359 - 372 DELL, "N Dass Les [GatcodoLig ber= to Pythagorica. — BELL M., Stytonale and roof in the Olympieian at Akragas jinfluence pwihagoricienne]: cf. N° 6669. ASA By 1990le Amer can Tournel B FAP. | . ‘ > V/37T Ben. M. Stylobate and roof in the Olympieion at Akragas : AJA LXXXIV 1980 359- 372. | The numbers and proportions used in tie design of the temple indicate Pythagorean influence and suggest that the temple should be associated with Theron, ruler of Akragas in 488- 472 B.C. Construction was pro[6569 bably begun prior to the battle of Himera in 180 B.C. 1642. BELL M., Stylobate and roof in the Olympieion at Akragas: AJA 84 1980 359-372. Les nombres et les proportions utilisés dans l'architecture de l'Olympieion d’Agrigente témoignent d'une influence pythagoricienne. Ces éléments impliquent en outre que le nom de Théron doit être associé à l'édification de ce temple. On peut donc en fixer la construction entre 488 et 472, et probablement avant la bataille d’Himère en 480.

Source: American Journal of Archaeology, Vol. 84, No. 3 (Jul., 1980), pp. 359-372 Published by: Archaeological Institute of America Stable URL: http://www.jstor.org/stable/504712 Accessed: 11/11/2009 08:24 Your use of the JSTOR archive indicates your acceptance of JSTOR's Terms and Conditions of Use, available at http://www.jstor.org/page/info/about/policies/terms.jsp. JSTOR's Terms and Conditions of Use provides, in part, that unless you have obtained prior permission, you may not download an entire issue of a journal or multiple copies of articles, and you may use content in the JSTOR archive only for your personal, non-commercial use. Please contact the publisher regarding any further use of this work. Publisher contact information may be obtained at http://www.jstor.org/action/showPublisher?publisherCode=aia. Each copy of any part of a JSTOR transmission must contain the same copyright notice that appears on the screen or printed page of such transmission. JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship. For more information about JSTOR, please contact support@jstor.org. Archaeological Institute of America is collaborating with JSTOR to digitize, preserve and extend access to American Journal of Archaeology. http://www.jstor.org

Stylobate Roof in the Olympieion at Akragas MALCOLM BELL (Pl. 47) Tutto in quel tempio e degno di studio; perche tutto ha il suo perche,e saldissimoperche. N. Palmeri, I832. Abstract The Temple of Zeus at Akragas is poorly preserved and imperfectly known. This article offers an analysis of its dimensions and of the principles which might have determined the anomalous arrangement of the columns. Suggestions are made for the interior plan and for the roofing of the structure, which probably included an hypaethral section in the center. Pythagorean beliefs and number theories may have influenced the shaping of this unusual temple. Among Doric temples the Olympieion of Akragas stands apart.1 Anomalous in scale and architectural form, the building seems intended to set forth a radical solution to the problem of Doric design. In size it was the largest temple in Sicily and the largest ever attempted anywhere in the Doric order. The strange design (ill. i) transforms the anticipated peristyle into a perimeter wall, creating interior spaces of unprecedented vastness. The order is treated with great freedom, with regard both to its structure and to the inclusion of non-Doric elements, some Ionic in origin and others unique. The temple bore the only major pedimental sculptures in Sicily, now almost entirely lost, and it incorporated into its design a series of colossal male 1 I am indebted to the National Endowment for the Humanities for a fellowship which enabled me to study the Olympieion in I976, and to Prof. Ernesto De Miro, Soprintendente alle Antichita di Agrigento, for his kind assistanceduring my visits to Agrigento. I also wish to thank Frank E. Brown, John F. Kenfield, Daniel T. Devereux, and Pamille Berg, who read and criticized a draft of this paper; and Erik Thorkildsen, who drew the plans and diagrams which appearin ills. I, 2, 4 and 7. 2 J.. Winckelmann, Anmerkungen iiber die Baukunst der alten Tempel zu Girgenti in Sicilien (1759), in Saimtliche Werke, vol. II (repr. I825 ed., Osnabriick I965) 312-26. 3 This pamphlet tradition combines varying portions of erudition, observation,and eloquence. It has also made several significant contributions to our understanding of the temple. The most prolific of the early writers was Raffaele Politi; others include L. Klenze, G. Lo Presti, N. Palmeri (a quotation from figures, or Atlantes, which for more than a century have provoked debate as to their function and meaning. Modern interest in the Olympieion begins in the eighteenth century, when Diodoros' description of the temple challenged the imagination of Winckelmann.2 His observationsremain perceptive, despite the fact that he had no first-handknowledge of the building. Haphazard excavation was commenced shortly after i800, bringing to light the first of the Atlantes and stimulating the appearanceof numerous pamphlets by local antiquarians and enthusiasts of the antique.3 Although the temple also attracted English, German, and Italian scholars and architects, it was not until the arrival of Koldewey and Puchstein in I896 that study of the Olympieion was based on close observation of the building and extensive knowledge of the traditions of Greek architecture.4The plans and drawings of Koldewey and Puchstein still offer after eighty years the most accurate account of the temple. More recently new evidence has been produced by the excavations of Marconi, Cultrera and Ricci, although only those of Marconi were ever published.5 Several specialized studies have also elucidated aspects of the building, but as yet there has been no attempt to whom begins this paper) and the Marchese Haus. The engineer D. Bentivegna first suggested that the Atlantes support the architrave in every intercolumniation (in Memoria sulla precisa collocazione dei giganti nel tempio di Giove Olimpico [Girgenti I875]); he did so some twenty years before Koldewey and Puchstein arrived independently at the same conclusion. A recent addition to this literature is A. Prado, II tempio di Zeus in Agrigento (Agrigento I973). Prado publishes plans of the 1940 excavations of G. Ricci. 4 R. Koldewey and O. Puchstein, Die griechischen Tempel in Unteritalien und Sicilien (Berlin I899) I53-66, and pl. 23. 5 P. Marconi, "I telamoni dell'Olimpieion agrigentino," BdA (1926-27) 33-45; "Studi agrigentini, l'Olimpieion," RivlstArch I (1929) I85-231; Agrigento, topografia ed arte (Florence 1929) 57-66; "Novitd nell'Olimpieion di Agrigento," Dedalo x2 (1932) I65-73. For the excavationsof G. Ricci, see infra n. 24.

E m s I -6 0 Q - 0-Z R~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ h,L "-Qmmp ILL. I. Olympieion at Akragas, plan

build on the evidence produced by the excavations of this century.6 Although several recent works on Greek architecture have expressed interest in the design of the Olympieion, there have been few attempts to account for the remarkable extent of its divergences from Doric norms.7 One effort at this sort of exegesis was made forty years ago by H. Drerup, who saw in the temple a reflection of the architectural traditions of the ancient Near East, consciously adopted by the Greeks of Akragas after their victory over the Carthaginians in 480.8The Punic architecture of Carthage was then to be regarded as an intermediary. Although Drerup attempted to account for some of the unusual features of the temple, which was also placed in an historical context, his views have proved difficult to accept because of the absence of similar buildings in Punic sites of North Africa or even of buildings with similar details.9 The disastrous condition of the Olympieion, which is owed as much to human depredations as to the workings of nature, has always been a major obstacle to study. Because of the temple's ruinous state there are still many gaps in our understanding of its original appearance. We are, moreover, still very much in the dark about the more general ideas and intentions of the architect and builder, which find such remarkable expression in the building. Consideration of two aspects of the temple's design-the dimensions of the stylobate and the nature of the interior spacesmay help to explain at least some of its unusual features and also to suggest ways of approaching the larger problems of its originality. F. Krischen, "Das Olympieion von Akragas," AA 1942, P. Griffo, "Sulla collocazione dei telamoni nel tempio di Giove Olimpico in Agrigento," Quaderni di Archeologia, = "Recenti ipotesi ricostruttive del Arte, e Storia 2 (1952) tempio di Giove Olimpico in Agrigento," Atti VII. Congresso Nazionale per la Storia dell'Architettura(I956) 229-38; E. De Miro, "Nuovo frammento di Telamone dal tempio di Zeus in Agrigento e nuova ipotesi ricostruttiva,"CronCatania8 (I969) 6 I-I9; 47-52. 7 W.B. Dinsmoor, "The Basis of Greek Temple Design: Asia Minor, Greece, Italy," Acts of the Seventh International Congress of Classical Archaeology I (Rome 1961) 355-68; V. Scully, The Earth, the Temple, and the Gods (New Haven 1962) 142-44; F.E. Winter, "Tradition and Innovation in Doric Design I: Western Greek Temples," AJA 80 (1976) 143-45; J.J. Coulton, Ancient Greek Architects at Work (Ithaca 1977) 82-85. 8 H. Drerup, "Der Tempel des Zeus Olympios in Akragas," Bericht iiber den VI. InternationalenKongress fu'r Archiologie, 361 THE STYLOBATE The first task is to determine the length of the foot used by the architect.The Olympieion belongs to the small group of Greek buildings whose dimensions were recorded by an ancient writer. In his brief but valuable description Diodoros tells us that the temple measured 340 feet in length, i60 in width, and 120 in height.10If Diodoros' figures are compared with the metric stylobate dimensions recorded by Koldewey and Puchstein (W: 52.85 m., L: IIo.oo m.), we find that two slightly different feet result for the width and length (W: 0.3303 m., L: 0.3235m.), indicating that Diodoros' figures are not exact and may have been rounded off." Yet these two "feet" are close enough to one another to suggest that the actual foot length was about 0.32-0.33m. Indeed the use of such a foot would not be surprising, for the normal Doric foot in the fifth century B.C. has a length of ca. 0.326 m.12 Using this foot, the width of 52.85 m. would become I62.II6 F, and the length of iio.oo m. would equal 337.423 F. These dimensions depend of course on the accuracy of the modern measurements. They can be checked by working from the length of the interaxial spaces (the distance between the axes of the engaged columns), which is almost the same throughout the temple (ill. i). The width and length of the stylobate equal the sum of the interaxials on ends and flanks, with the addition of two half-columns representing the extension of the stylobate beyond the axes of the corner columns (W - 6 I + i col.; L = 13 I + I col.).13 The intended interaxial was evidently 25 F (or 8.i5 m.), and the column diameter was half Berlin 1939 (Berlin 1940) 379-87. 9 It seems unlikely that a Greek architect of the early 5th century would have used barbarianmodels, even under the circumstances suggested by Drerup. This apparently did happen in the 4th century, at Selinus-where there was a strong Punic element in the population; see D. White, AJA 71 (I967) 341-52. 10 Diod. 13.82.1-5. Diodoros gives the width as sixty feet; the emendation to I60 is owed to Winckelmann (supra n. 2). 11 I have used the measurements recorded by R. Koldewey and 0. Puchstein (hereafterK-P). Dinsmoor's figures, except for width and length, are derived from averages,in W.B. Dinsmoor, The Architectureof Ancient Greece (London 1950) 338. 12 W.B. Dinsmoor, (supra n. 7) 359-60. Henceforth the abbreviation F will be used for measurementsin Greek feet. 13 The abbreviationI will be using hereafterto refer to interaxial spaces. Although the columns of the Olympieion are in fact engaged half-columns, they are centered on the stylobate and are treated by the architectas if they were normal columns.

this figure (I2.5 F or 4.075 m). The width of the stylobate is therefore 25 X 6 + I2.5 = 162.5F; the length is 25 X 13 + 12.5 337.5 F. These dimensions correspond closely to the figures in feet derived from division of Koldewey and Puchstein's metric width and length by a foot of 0.326 m. (162.116 F and 337.423F). If the dimensions of 162.5 by 337.5 F are correct, and the temple was accurately laid out in the manner suggested above, then its total metric dimensions ought to be 52.975 m. (I62.5 X 0.326 m.; as against 52.85 Koldewey and Puchstein, 52.74 Dinsmoor) by 110.025 m. (337.5 X 0.326; as against II0.00 m. Koldewey and Puchstein, 110.095 Dinsmoor). The slight differences can be explained as the result of errors either in the modern metric measurements or in the laying out of the temple. It should be noted that accurate stylobate dimensions are today very difficult to obtain because of the ruinous state of the temple; for example, the width of the stylobate must be calculated on the east end, where neither corner is preserved.Projection of the flanks is therefore necessary, and there is strong probability that the metric dimensions are not entirely accurate.14 A similar procedure in planning the relationship of interaxials to stylobate had been followed a generation earlier, in the laying out of the temple of Athena at Poseidonia.15 There the desired total flank measurement of Ioo F consists of 12 interaxials of 8 F, with the addition of 4 F corresponding to two column halves. But the similarity of the temple of Athena to the Olympieion lies not only in the way the interaxials of constant length were placed on the stylobate, but also in the fact that stylobate dimensions were themselves predetermined and fixed. In the Olympieion the primacy of stylobate dimensions over other factors is demonstrated by the disparity between the ideal interaxial of 25 F (8.I5 m.) and the slightly different interaxials on ends and flanks, as recorded by Koldewey and Puchstein (average on east: 8.071; on south: 8.I2I). The difference in end and flank interaxials is explained by 14 As is in fact suggested by the differences in Dinsmoor's and K-P's figures for length and width; the largest difference (o.II m.) is in the width. 15 For the dimensions of the temple of Athena at Poseidonia, F. Krauss, Die Tempel von Paestum I, Der Athenatempel (Berlin 1959) 13-19. 16I have used a base diameter of I3V1 F (4.40I m.); Dins- [AJA 84 the necessity of accounting for the bases of the corner columns, which could not overlap the stylobate and consequently required that the axes of their columns be set back.16Moreover, the bases were themselves set back slightly from the edge of the stylobate (ill. 2). The total amount of this setting back was then divided between the interaxials of ends and flanks, producing in consequence slightly wider spaces on the flanks, where there were more interaxials to absorb the adjustment, and smaller ones on the ends. 11. 2 depicts an ideal interaxial. Assuming first that the column bases had a diameter of I3.5 F (or. 4.401 m.), which allows for an extension of 2 F around the column, and then that the bases were set back another /4F, a total of i 2 F had to be subtracted from the interaxials of ends and flanks. This amounts to 0.0815 m (or precisely /4 F) per interaxial on the ends, and 0.0376 (not quite ?8 F, or on the flanks, or interaxials of 8.o685 m. 0.04075) on the ends and 8.1124 m. on the flanks. These adjusted interaxials are in general agreement with the average interaxials recorded by Koldewey and Puchstein (8.071 m. and 8.I2I m.).17 It is necessary again to emphasize the difficulty of obtaining accurate measurements. It is possible that there were further adjustments in the corner flank interaxials, as is suggested by Koldewey and Puchstein's slightly smaller measurements on the south (first interaxial: 7.99 m., second: 7.98 m.). If these figures are accurate,there may have been slight angle contraction, with the corner interaxials reduced by about /2 F to 24/2 F (from 8.I5 to 7.987m.). Yet there is reason to question the significance of the short interaxials recorded by Koldewey and Puchstein. The first two interaxials are still covered by tons of debris and their measurements could only have been made from the centers of the interior pilasters. The amount of contraction is also quite insufficient,and there is, moreover, no contraction on the east end. The presence of angle contraction on the south flank therefore seems doubtful. It is interesting to note that a similar situation moor assumes a base of ca. 4.30 m. On this and related questions, see the appended note on ills. i and 2. 17 The average interaxialsare here calculated by dividing the total interaxial measurements(the distances between the axes of the corner columns) by the number of interaxials on the ends and flanks, or 6 and I3.

a '?jI X?~E LL u E X~~C |~ <-~ 1 : _-_ r oU1- LL. I I -^11 ^J \ J 0I $4 ?r(~~~~~

was faced by the architect of the temple of Athena at Poseidonia, who also had to make small adjustments in the interaxials because of a fixed stylobate dimension (there the length of oo00 F). The baseless Doric columns were set back ca. 0.04 m. from the edge of the stylobate, and consequently ca. o.o8 m. had to be subtractedfrom the total axial measurement of the flank. The flank interaxials therefore have a length of 7 15/16 F, rather than the planned 8 F. On the ends, where a fixed dimension was not a consideration, an amount equivalent to the margin (14 F) was then added to the width of the stylobate, and so-on the east end, at any rate-the interaxials approximate 8 F.18 Why could not an amount corresponding to the thickness of the column bases and their setback have been added to the length and width of the stylobate, thereby avoiding the necessity of making small adjustments to the interaxials? And, if we accept the presence of angle contraction on the flanks, why then could not the stylobate have been reduced by the proper amount, again avoiding adjustments? Such procedures were not followed in the Olympieion simply because the architect desired above all to preserve the dimensions of the stylobate, as expressed in a width of 162.5 F and a length of 337.5 F. In his comments on the Olympieion, W.B. Dinsmoor noted that, according to the dimensions recorded by Diodoros (i60 by 340), the periphery of the stylobate measured in toto oo000F.19 But the actual periphery was, according to Dinsmoor, only 994/2 F, a figure which resulted from his proposed foot length of 0.32712 m.20As a consequence of using the slightly smaller-and better documented-foot of 0.326 m., the periphery of the stylobate returns in its total length to the resounding figure of iooo F. (337.5 + I62.5 + 337.5 + 162.5 - I000). This agreement with the total dimensions of Diodoros seems both to confirm the foot of 0.326 m. and to suggest that Diodoros' figures, while rounded off, still produce accuratelythe 18 For the dimensions of the Temple of Athena, see Krauss (supra n. I5). For a different explanation of the stylobate of this temple, J.J. Coulton, BSA 69 (I974) 8i (using Dinsmoor's figures). Despite the obvious differences, the Temple of Athena has many similarities with the Olympieion; interest in number is combined with great freedom in the handling of the order. See also infra n. 51. 19 Dinsmoor (supra n. 7) 36I. 20 Dinsmoor derived this foot from a hypothetical normal [AJA 84 larger number which was a major factor in the planning of the temple. It may be that in the early planning stages the architect began with the dimensions recorded by Diodoros. With a width of i60 F and a length of 340 F the temple would have the desired periphery of iooo F, and its ends and flanks would be related in the ratio of 8:I7. Yet it was still necessary to place columns on the stylobate. The architect apparently had clear ideas about these as well, for the number of columns on ends and flanks (7 by 14) was to be related in the ratio of i :2. Small changes had to be made if these were to be fitted on the hypothetical stylobate proportioned I60 by 340. Working with interaxials of 25 F and 7 by 14 columns, the total interaxial measurements came to 50o by 325 F; to these figures had to be added the width of one column, the diameter of which in relation to the interaxial was determined by the same ratio that had governed the number of columns on ends and flanks. The columns therefore had a diameter of 21/2 F, which, added to the total interaxial figures, gave a width 2? F greater than i60, and a length 212 F smaller than 340-and, still, a total peripheral dimension of oo000F. The ratio of width to length was now I3:27. The problem of fitting the columns with their broad bases on the stylobate required small adjustments, as apparently did the angle contraction on the flanks, but these were made in the interaxials and not in the overall dimensions of the stylobate, which were not altered. If this account of the stylobate dimensions of the Olympieion is correct, then there is no reason to assume, as has been done, that the building's length was derived from temple GT at Selinus, and the width calculated therefrom.21The fact that the two buildings have almost the same length is coincidental, for in both the length was determined by internal considerations. The dimensions of the Olympieion begin with the decision to build a great temple to Zeus with a peripheral measureflank interaxial of 8.I80 m. = 25 F. The dimensions of the stylobate would then be simply the sum of a sequence of interaxials of slightly varying size. This hypothesis does not account for the difference between the "normal"flank interaxial and the smaller interaxials on the ends. In the present study it is assumed that stylobate proportions determined the length of the interaxials, rather than vice versa. 21Coulton (supra n. i8) 71.

ment of Iooo F. The architect'schoice of this number is probably significant, as are the stylobate proportions of I3:27. Reasons for the use of such numbers will be suggested in the conclusion. THE CENTRAL SPACE Our evidence for the interior of the Olympieion (ill. i) is even more unsatisfactory than for the exterior. The rows of piers which divided the building into three long spaces were constructed of relatively small blocks, as were the intervening screen walls and the two cross walls, and consequently it was the interior that suffered most when the temple was used as a quarry in the eighteenth century and earlier. Only three piers are preserved above the interior floor level (south pier I, north i and 12), although others can be traced at varying foundation levels. There are no known blocks which can be assigned either to the pilasters which formed the outer sides of the piers, or to the massive piers themselves; and so very little can be said about the parts of these supports which would interest us most, namely their capitals and entablature. Despite the paucity of evidence, it is still possible to come to an understanding of the interior arrangement of the Olympieion which is accurate in its general outlines. The most important evidence was recovered in the excavations of P. Marconi and G. Ricci. In 1926 Marconi found numerous fragments of terracottaroof tiles in the south hall, including several pieces which belong to kalypteres (ill. 3).22 The presence of such tiles, which were intended to run along the summit of a pitched roof, within the south hall indicates that this part of the building had such a roof with two slopes, and these can only have run east-west-that is, along the long axis of the temple. A similar roof over the north hall can be assumed. These huge roofed halls (ill. 4) must then be the spaces which Diodoros calls cr-oatand which in their size and height greatly impressed him.23The inner slopes of 22 . Marconi, "Studi agrigentini II," RivlstArch i (1929) 220-24. 23Diod. 13.82.4. Marconi considered this solution but rejected it. 24G. Ricci, Le Arti 3 (1940-4I) 135-36; H. Fuhrmann, AA 1941, 687-88; P. Griffo, "Ultimi scavi e scoperte in Agrigento," Quaderni di Archeologia, Arte, Storia 3 (1946) I8-I9; A. Prado, II tempio di Zeus in Agrigento (Agrigento I973) 20, fig. 5 ILL. 3. Olympieionat Akragas,terracottakalypter. (Drawn by R. Carta) the roofs then must have been inclined towards the central space of the temple, and so it is hard to avoid the conclusion that this space was from the start intended to be open to the sky. The proof, if it be needed, for an open court was found by Ricci in the important war-time excavations of i940, which unfortunately have never been published.24Concentrated in the eastern sector of the temple (ill. I), the excavations laid bare the foundations of piers I-3 in the north and south rows and pier 4 in the north row, and also revealed for the first time the foundations of the pronaos wall between the second pair of piers.25A second important discovery was made at the time, the significance of which has not been appreciated. Piers S 2 and S 3 (pl. 47, figs. I-2) were revealed to be hollow, with rectangular shafts extending deep into their foundations.26 The shaft in S 2 is at least as deep as the tenth foundation course, while that in S 3 is present in the uppermost surviving foundation course immediately below the ancient floor level, and extends down at least nine courses. The bottoms of both shafts are today covered by debris and so their full depths cannot be ascertained. The foundation courses are 2 F in height (ca. 0.63-0.65 m.), and thus the depth of both shafts below the ancient floor level is more than 20 F, or six meters. Both shafts are rectangular, measuring 1.95 by 2.60 m. (6 by 8 F, thus (the only published plan of the I940 excavation). 25 The piers have been numbered beginning at the east: N 2 indicates the second pier in the northern row. Of the piers excavated in I940, only S I and N I are preserved above floor level; S 3 survives to the uppermost foundation course, while S 2 and N 2 and 3 are reduced to the lowest foundation levels. N 3 and 4 have been covered over again. 26 Griffo (supra n.

, a [AJA 84 . ^ 4 .. . ,?, . - . , ;.~~~~~~~~~~~~~~:11 , R ILL. at elevation. ILL.5' 5. Capitolium atCosa, Cosa,restored restored elevation. Capitoliumn F.E. Brown) (By (By F.E. Brown) I? L ::;::i; s ILL. 6. Capitoliumat Cosa,restoredsection. (By F.E. Brown) 0 10 20 30 40 M ILL.4. Olympieionat Akragas,roofingdiagram proportioned 3:4), and they are so placed within the piers as to leave a thicker wall on the sides which face the south hall (pl. 47, fig. 2). The inner surfaces of both shafts are covered with a layer of hydraulic cement, 0.025 m. in thickness.2 27 Griffo 28 Winter (supra n. 24) I9. (supra n. 7) I44. 29 One of the piers above the interior floor level may have housed a wooden stair, giving access to the roofs; so Ricci, This cement is an unmistakable sign of the function of the two shafts: they were cisterns, and the water that filled them must have come both from the runoff of the inner slope of the roof over the south hall and from the hypaethralcourt. The massive piers extend into the central space much more than into the two halls, and it has recently been argued that they were therefore intended to carry a roof at pediment level.28Yet the shafts in piers S 2 and S 3 must have been carried through to the tops of their respective piers, and it seems unlikely that hollow supports of this sort could have been used to sustain such a high roof.29 (supra n. 24) 136; the other shaft must then have been open to the roof, serving to channel rainwater into the cistern. In this case the two cisterns may have been connected by a passage at their lowest level.

terfered fatally with the runoff of the hall roofs. The presence of cisterns within two internal piers thus precludes a roof over the central space of the temple. The Olympieion evidently belongs among those temples of Zeus identified by Vitruvius as being hypaethral (De arch. 1.2.5, 3.2.8). Cisterns are normally filled by the runoff of a roof or of an exposed paved area at ground level, and the presence of cisterns within a building indicates that there must have been an opening in its roof, in the form of a central court or compluvium. Cisterns are not therefore normally found in peristyle temples, and I know of no parallel among Doric temples for the cisterns in the Olympieion. As far as I know, cisterns do not occur in hypaethral temples, where they might be expected. In both the archaic and later temples at Didyma, drains were provided to carry off any water that fell in the central court.30The arrangements in the archaic Artemision at Ephesos, presumed to have been hypaethral, are evidently not known.31 An interesting parallel for the Olympieion is found in the Capitolium of the Roman Republican colony of Cosa.32This podium temple in the Italic tradition was built towards the middle of the second century B.C. Under the porch is a large cistern, I 0 I 10 I 20 1 30 1 40 M ILL.7. Archaictemple of Artemisat Ephesos, roofing diagram.(After A. Bammer) Moreover, the thicker walls on the outer sides of the piers correspond to the pilasters of the south hall, which, as is shown by Marconi's kalypteres, did carry a roof. An even greater objection to a roof over the central space is raised by these lateral roofs. Marconi's tiles show that they had two slopes, and the cisterns reveal what became of the rain that fell on their inner surfaces. Another pitched roof over the central space would have in- 30 T. Wiegand, Didyma I, H. Knackfuss, Die Baubeschreibung (Berlin I94I) 54, pl. 105, I49 (later temple); I2Ib (earlier temple). See also B. Kehr, "Zur Geschichte des Apollonheiligtums von Didyma," MarbWinckProg I97I-72, 56. 31 Not discussedin A. Bammer, Die Architekturdes jiingeren Artemision von Ephesos (Wiesbaden I972). Temple GT at Selinus was hypaethral and cisterns may well have been used there in the generation preceding the planning of the Olympieion. Orsi believed that the Athenaion at Syracuse was hypaethral; although he has not been followed in this, the circumstancewhich led him to such a conclusion remains problematical: a large drain runs away from the north side of the temple, connecting with a N-S channel which emerges from the third foundation course below the euthynteria. If the drain continues into the temple at a right angle to the flank, as it appearsto do, it must come from the area of the cella. Below the pronaos of the Athenaion are two open rooms, formed by N-S foundation walls; these spaces give no evidence of ever having served any specific function, and they were certainly not cisterns. Yet the existence of open spaces beneath the pronaos suggests at least the possibility that there might be others beneath the cella, and the drain raises the possibility that these might have been cisterns. The source of the water which left the temple via this drain is indeed mysterious, and Orsi's proposed hypaethralcella offers a solution to the problem; on the drain, P. Orsi, MonAnt 25 (1918) 380-go, figs. 15-16; on the rooms beneath the pronaos, P. Orsi, NSc 1910, 537-38, pls. A, B. 32 F.E. Brown, in Cosa II (MAAR 26, I960) 59-65, figs. 75. I am indebted to Prof. Brown for the elevation and section which appear in ills. 5-6.

which can only have been filled by rainwater falling from a compluvium in the roof above (ills. 5-6). Frank E. Brown's reconstruction of the wooden roof of the Capitolium gives us a valuable parallel for the roof at Akragas. Despite the differences in size and function of the openings, the smaller pitched roofs which flank the compluvium at Cosa are quite similar to the roofs here proposed for the two halls of the Olympieion. At Cosa the main roof is placed at the level of the exterior gables, as would be expected, and this was probably also the case at Akragas (ill. 4). Central ridge-poles will then have extended from the apex of the pediments as far as the two crosswalls (ill. i) which define the pronaos and adyton (or opisthodomos, as it may be), and there will have been internal "pediments," perhaps of wood, above these walls. A lower roof over pronaos and adyton, at the level of the roofs of the two long halls, seems unlikely, for then the east and west pediments would have had no more function than a Baroque church facade, rising high above the roof behind. A solution which places a high pitched roof over pronaos and adyton, and lower pitched roofs on the flanks, is similar to the roofing system proposed by A. Bammer for the first Artemision (ills. 4, 7).33 The hypaethral court must have served as a great light-well for the illumination of the two flanking halls. The walls which connect the piers are not so thick as their counterparts between the columns and so must not have been as high; rather, these were screen walls which terminated at a certain height in order to let light into the two halls. They are analogous to the exterior screen walls of temple F at Selinus. As for windows in the exterior intercolumnar walls, enough of the fallen sections of this wall remains together at the second column of the south flank to prove that exterior windows never existed.34 The presence of elements of the Ionic order in the Olympieion has long been known.35 The engaged columns of the exterior are set on high 33 Bammer (supra n. 31) 13, fig. 8; see also G. Gruben, "Das archaische Didymaion," Idl 78 (1963) 151, fig. 37 (hipped roof on ends). 34 I intend to deal with the problems of the elevation of the Olympieion in a later and more extended study. 35 F. Krischen, "Das Olympieion von Akragas," AA I942, I-i9; Winter (supra n. 7) I45. The freedom with which Ionic [AJA 84 bases, decorated with narrow mouldings; Ionic mouldings were used in the elevation; and the use of Atlantes as a supporting device is evidently derived from the caryatids of Ionic origin. The great size of the Olympieion and its high, stepped platform are also eastern features, and it seems possible that even the pilaster order of the south and north halls is derived from some source in Asia Minor, perhaps the archaic Didymaion.36Not least among the Ionic elements of the Sicilian temple is the hypaethral court, which must have been suggested to the architect by the colossi at Ephesos and Didyma. AN INTERPRETATION Although the Ionic features of the Olympieion can be interpreted as an appropriate response to the decision to build a very large temple, they explain neither that decision nor the essential strangeness of the building which resulted from it. One approach to these questions is suggested by the numbers and proportions which were used in the temple's design. Although an interest in number would hardly be surprising in an architect of the early fifth century, here it is a question of the particular numbers and proportions which were used. Three areas seem especially significant: the total peripheral dimension of the stylobate, the number of columns chosen for the ends and flanks, and the ratio of width to length in the stylobate. In each case the numbers or ratios appear to have significance in Pythagorean numerology. The size of the Olympieion on the stylobate was evidently determined by the total dimension of Iooo F, and we have seen that the architect preferred to make small adjustments in the interaxials rather than adopt the easier solution of altering the width and length of the stylobate. The number Iooo is the cube of io, which as is well known was the most meaningful of numbers for Pythagoras and his disciples. Ten is the sum of the first four digits, which form the components of the most important intervals of the musical scale: the octave features are combined with the Doric in the Olympieion suggests a deliberate improvement on-even criticism of-Temple GT at Selinus, where the purely Doric solution seems, even today, clumsy and inappropriatein a temple of such large scale -however daring the conception may be. 36G. Gruben (supra n. 33) 172.

the fifth 3:2, and the fourth 3:4. The number 369 columns on ends and flanks in the unique proportion of seven to fourteen, but the proportion of column diameter to interaxial is in the same ratio io was symbolized by the tetraktys, the equilateral triangle formed of ten dots arranged in four rows. The Pythagoreioi called this "the source and stem of i :2. The number of columns seems particularly of ever-flowing nature," and on it they swore curious; the choice of seven columns for the ends oaths.37The tetraktys embodied all numbers, but required that there be a central column, rather than its basic elements were three and ten. The number a void, something carefully avoided in normal Iooo, ten to the third power, may be understood as Doric design. Although a number of contemporary deriving from the tetraktys, and the temple may or slightly earlier temples have peristyles with six then be seen as a physical embodiment of the by twelve columns, none has seven by fourteen. sacred symbol, generated from the number ten. Other measurements based on the major intervals The number Iooo also appears to have had sym- of the scale are found elsewhere in the building.42 bolic value. In Pythagorean doctrine about metemA third possible appearance of Pythagorean psychosis each cycle of incarnation was believed to number theory is found in the ratio of width to last a fixed number of years or seasons, which were length in the stylobate, I6212:337/2 or 13:27. Both often calculated as exponential multiples of ten. 13 and 27 have symbolic meaning in the musical Thus Empedokles, who was Akragas' greatest son, theory of the fifth century Pythagorean Philolaos.43 declared that the soul returns to a new body after In his calculations of the smaller intervals of the Io,ooo years.38Plato later described cycles of Iooo diatonic scale, 27 representedthe whole tone and 13 years.39It is generally assumed that Plato drew on represented the diesis, or smaller semitone.44 It Pythagorean sources but it has not been deter- would therefore seem that the ratio of width to mined whether his thousand year cycle was an ar- length in the Olympieion consists of those numticle of belief in the fifth century.40Although the bers which signified for Philolaos the tone and stylobate dimension of iooo F in the Olympieion semitone, the basic units of the scale-just as the probably was not derived from doctrine about total dimensions appear to have been generated metempsychosis, the number itself evidently was from the number io, which symbolized, among other things, the major musical intervals.45 significant in Pythagorean thought. A second possible manifestation of Pythagorean The occurrence of significant numbers and prooccurs in the in the of the use, numerology design portions in the Olympieion raises the question of of based on the octave. This whether the religious consciousness which shaped temple, proportions was "the most beautiful of the (musical) intervals the temple's remarkable interior may not also have and the best of ratios."41Not only are the engaged had a Pythagorean orientation. The followers of 37 On the tetraktys,A. Delatte, Etudes sur la litteraturepythagoricienne (Paris 1915) 249-68; W. Burkert, Lore and Science in Ancient Pythagoreanism,trans. Edwin L. Minar, Jr. (Cambridge, Mass. 1972) 72, n. 120. 38 H. Diels and W. Kranz, Die Fragmente der Vorsokratiker I (Berlin 1951) Empedokles II5. 39 Plat., Rep. 615a, Phaedrus 249a. In (Iamblichos) Theologoumena Arithmeticae (ed. de Falco) 52.1off., the cycle of incarnation consists instead of 216 years; 216 is the cube of 6, as oo000is the cube of o0. 40 See A.E. Taylor, Plato, the Man and his Work (Cleveland and New York 1956) 308, n. i; R. Hackforth, Plato's Phaedrus 43 Burkert (supra n. 37) 389, 399-400. 44 Twenty-seven is the difference between the interval of the whole tone 243:216 (9.8 = 243:216). Thirteen is the difference between the interval 256:243; this is the remainder when two whole tones are subtractedfrom a fourth: (4:3):(9:8):(9:8) = 256:243. This is the Xei7rov of Plato, Tim. 36a-b, called the diesis by the Pythagoreans.The theory of Philolaos is discussed by Burkert (supra n. 37). According to Plutarch, De an. procr. I4.ioi9a, it is with the numbers 13 and 27 that "the mathematicians measure clearly the melodic intervals, calling one the diesis, the other the tone." 45 It is interesting to note that the significance of the ratio (Cambridge 1952) 83, n. I; W.K.C. Guthrie, A History of 13:27 lies not in the ratio itself but in the symbolic value of its Greek Philosophy II (Cambridge i965) 253. components. In this respect "Pythagoreandesign theory" differs 41 Ptol., Harm. (ed. from usual Greek practice, in which simple ratios of small During) II.21-22 . . . TrV re avcuowe'or KacXXITTr)KaLripv Xoywv o 87rXdaotos whole numbers play a more important role. The viv Xj 8&a7raTTCv proportions apto'ros. of the Olympieion are derived from a symbolic harmony, 42 Even the blocks of the exterior wall measure 2 by whereas those of the Parthenon (W:L= 4:9) are based more 4 feet; other intervals are found in the elevation, as that of the sima on a harmony of perception-even if 22:32 = 4:9. The Olymand geison (3:2). No other temple has a peristyle of 7 by 14 pieion also belongs to the tradition of designing with sacred columns. The proportionsof the Temple of Poseidon in Plato's measurements, as exemplified in the o00 foot temples of the Atlantis are 1:2 (Kritias II6c). archaic period. See also infra n. 51.

Pythagoras did not ignore the traditional gods; the philosopher himself was identified with Apollo, and to the question, "What is most just?," his disciple replied, "to sacrifice."46In Pythagorean cosmological thought Zeus was associated with the fire which burned at the center of the universe. This central place was referred to allusively oras died after a battle between Syracusans and Akragantines, when he refused to retreat across a field of beans.53Pythagoras' death in Metapontion in southern Italy is better attested and Hermippos' account is rightly identified as a piece of antiPythagorean ridicule.54 There is no evidence to suggest that the philosopher ever visited Akragas, as the vkaK '/ Ato',47 Zavos 7rvpyo0S, povos Atos48 or even Sicily. In Iamblichos' list of famous Pyand oLKoAto0s:49Guardhouse, Tower, Throne, and thagoreans the only Akragantine is Empedokles, House of Zeus. Could it be that one or more of who was born in the 470s and came to maturity these images, which are concrete and strikingly ar- after the building of the Olympieion.55 Yet Emchitectural, lies behind the strange form of the pedokles was strongly influenced by Pythagoras, Olympieion, that the temple should be seen as a whose ideas are likely to have been current in symbolic representation of Zeus' seat at the center Akragas at least as early as the youth of the of the universe? The light-filled court at the center younger philosopher. of the temple, an appropriate focus for the cult of The best evidence for knowledge of PythagoZeus as a celestial god, could also have been reanism in Akragas in the first quarter of the cenplanned as a place for a central fire or hearth.50 tury is found in a literary work written by a poet Unfortunately we have no literary or archaeologi- from the Greek homeland. The second Olympian cal evidence to explain the function of the interior Ode of Pindar was composed for Theron, ruler of spaces, so such hypotheses cannot be tested. Yet the Akragas, to commemorate his chariot victory in the interior of the temple is unique in the sacred archi- games of 476. To celebrate the victory Pindar came tecture of the classical period. Given the possibility to Akragas. In the unusually personal second of a Pythagorean contribution to the planning of Olympian, Pindar describes the dynast's expectathe building, speculation about the function and tion of an exalted existence after death, promised meaning of the interior spaces, and about the form by his achievements, high position and virtue. Pinof the building as a whole, seems justified.51 dar alludes to metempsychosis, a judgment in the To what extent are Pythagorean ideas likely to underworld, and the attainment of an heroic exhave circulated in Akragas in the early fifth cen- istence after death. Pindar hardly needs to invoke tury? The sources which associate Pythagoras him- the name of Pythagoras to remind us that such self with the city are all late and of little value. beliefs form the core of Pythagorean doctrine about Although Iamblichos includes Akragas among the the survival of the soul. cities which the philosopher is supposed to have set free, such an event is otherwise unattested and diffi- Truly wealth patterned with prowess cult to reconcile with historical tradition.52 The Brings the moment for this or for that, Hellenistic writer Hermippos claimed that Pythag- If it rouses deep ambition to range afar, 46 On the identification of Pythagoras with Apollo, see Burkert (supra n. 37) 141-46; Iamblichos, De Vita Pyth. I44 (the commandment "first honor the gods .. ."). 47 Aristotle, De caelo 293b.I-II; also Aristotle, frag. 204 (Simplicius, in De caelo 5II.25). 48 Aristotle, frag. 204. 49 Diels and Kranz (supra n. 38) 44 A I6, where other terms including "altar," "hearth" and "meeting point" are used. In Plato, Kritias I2Ic, Zeus assembles the gods in the center of the universe. 50 See Burkert (supra n. 37) 350, n. 64. I take Empedokles, frag. I42, to refer to the oikos Dios literally. With its plan derived from significant numbers, the Olympieion might also be called a IAETrpov fvaews (Diels and Kranz, A 44 I6). The Atlantes of the Olympieion may also have a symbolic meaning; in Theologoumena Arith., 80.5, 82.5, the number ten was identified "by the Pythagorikoi" with both Atlas and Ananke. In Hesiod, Theog. 517-19, Zeus punishes Atlas with his Ananke, requiring the rebellious Titan to uphold the sky forever. The Atlantes of the temple should be viewed in such terms, not as symbols of the defeated Carthaginiansafter Himera. 51 Pythagorean number theory has been proposed for the Temple of Athena at Poseidonia by R.R. Holloway, "Architettura sacra e matematica pitagorica a Paestum," ParPass 21 (1966) 60-64. 52 Iamblichos, De vita Pyth. 20.I. 5 Diog. Laert., Pythagoras 40, 45; also schol. Plato, Rep. 6oob. The battle fought in 476 between Syracusansand Akragantines (Diod. II.53.4-5), won by the former, might be considered a possible candidate for this event, but the participation of the aged philosopher seems most improbable. 54 Aristotle, frag. I9I; Aristoxenos, frag. x8. 55 Iamblichos, De vita Pyth.

A transcendentstar,the truestlight for a man. -If any man has it, and knows what shallcome,That of thosewho die here The lawlesssoulsat once pay penalty, And sins done in this kingdomof Zeus Are judgedby one below earth With harshinexorabledoom. nor does he refer specificallyto the Olympieion.58 W.B. Dinsmoorarguedthat the dateof the temple should be pushed back to ca. 5Io B.C., on the groundsof the absenceof appreciableangle contractionand his own assumptionthat the Olympieion was an immediateresponseto the building of Temple GT at Selinus, which is dated about of the Olym525.59 In light of the interpretation But in nightslike oursfor ever, which has been offered neither of pieion here, With a like sun in theirdays, these seems Morevalid.60 arguments particularly The good win a life withoutlabor, in B.C. had been in existence over, for 51o Akragas Nor bracetheirhandsto troubleearth and could have sustained only seventy years hardly Or the sea'swaterfor an emptylivelihood. the constructionof such a large and costlyedifice. Besidethe most honoredof Gods Yet Dinsmoor'sassertionthat the Olympieion Those who delightedto keep theirword was begunbeforethe battleof Himeraseemsto me Passunweepingdays;the others to be correct.Becauseof the late archaicsculptural Supporta burdennot to be lookedupon. styleof the Atlantesand the profilesof the capitals, All who have enduredthreetimes which point togetherto the late 470s,thereis every In a sojournin eitherworld, reasonto believe that the Olympieionwas begun To keep their soulsutterlycleanof wrong, well beforeTheron'sdeathin 472.61His rule over Go by God'sroadto the Tower of Kronos, Akragas is usually assumed to have been estabWherethe Airs, daughtersof Ocean, lished in 488, but he is alreadymentionedwith a Blow roundthe Islandof the Blest, certainawe in the sixth Pythianof Pindar,written And the flowersare of gold.... for his brotherXenokratesin 490.62Although a OL. II, 52-74, trans. Bowra date for the beginningof work after 480 remains These beliefsare not so much the poet'sown as possible,an earlierstartduring the decade490-480 his patron's.Pindar "turnsinto high poetryThe- seems to me to be more probable.Sufficienttime ron'sown views, and he is able to do this because would then be allowedfor the planningphasesand he must to some extent have absorbedthem into for constructionup to the level of the capitalsat his imaginativeconsciousness.Theron must have aboutthe time of Theron'sdeath.In this view the instructed Pindar on his own esoteric beliefs."56 Carthaginianwork force will have contributedto Theron was given heroic honors after his death; the completionof a building project already in his tomb was of extraordinary size, as would befit progress. The apparentPythagoreanelementsof the Olyma man who was believed to have become a god.57 Theron ruled Akragas from 488 until his death pieion suggestthat the templeshouldbe associated in 472. The building of the Olympieion is generally with Theron himself,ratherthan with the victory thought to have commenced only after the great over the Carthaginians.The dynast must have victory over the Carthaginians at Himera in 480, known of the great buildingsbegun by such auin which Theron played a decisive role, but there tocraticrulersas Polykratesof Samosand the Peiis no conclusive evidence to support this view. A sistratidsin Athens, and so on his accessionto passage in Diodoros indicates that the city's largest power a project of comparablescale may have buildings were constructed by Carthaginian pris- seemedfitting to him. The storyof Theron'sseizoners captured at Himera, but Diodoros says noth- ing of poweris in fact associatedwith the building ing about the inception of work on such buildings, contractwhich he held for a templeto Athena,and 56 M. Bowra, Pindar (Oxford 1964) I2I. See also U. Wilamowitz von Moellendorff, Pindaros (Berlin 1922) 243. 57 Diod. 11.53.2, 13.86.2. 58 Diod. 11.25; see also Winter (supra n. 7) 143. 59 Dinsmoor (supra n. i ) Io . Angle contraction is (largely) ignored because of the primacy of stylobate dimensions; the similarity of the two build60 ings is limited to their length and, as has been suggested above, this is coincidental. 61 Dinsmoor (supra n. i ) IoI, n. I. 62 Pindar, Pyth. 6.46. The beginning of Theron's reign is assigned to 488, as Diodoros states that he ruled sixteen years and died

there is much other evidence for his interest in embellishing Akragas.63The Olympieion was the most ambitious of these undertakings. It would demonstrate not only the magnificence of his reign, but the harmony of nature perceived in number. It stood as a microcosm, an offering to Zeus who was perceived as the embodiment of the order of the cosmos. For the Olympieion, Theron chose an architect of great imaginative power, who was, like Pindar, in sympathy with his religious ideas. These then evidently became the determining influence on the building's form. [AJA 84 ros and blocks of which still survive with their relief sculptures)then it may have been carriedby a central pier, substitutingfor the thickerwall at the east. I can offer two observationsin support of this hypothesis: the positioningof the wall in relationto the two piers which it joins to north and south producesthe same setback as that of the E-W screen walls (ca. 2.0 m.), so that the projectionof the supposedcentral pier from the crosswall would agreewith that of the other piers; and an analogouscentralpilasterappearsat the short ends of the north and south halls. The west crosswall has never been fully uncovered;excavationwould no doubt produce more evidence than is available at present. Illustration2 illustratesan ideal interaxial,without DEPARTMENT OF ART adjustments.No interaxialor column is sufficiently UNIVERSITY OF VIRGINIA well preservedto make a completemeasureddrawing VIRGINIA possible,and so this one is a composite,basedon measCHARLOTTESVILLE, 22903 urementstaken in many places. Several details require comment: I) The columns A NOTE ON ILLS. I AND 2 (diam: 4.075 m. or I212 F) are centeredon the styloAny plan of the Olympieionmust be in good part bate, which has a width of 4.564 m. or 14 F. A 4 hypothetical.The plan which appears here as ill. i foot margin results,with the column base accounting assumes original stylobate dimensions of 337/2 by for V1 F and the setbackof the base for the remaining can be made only at E 2, 4 F. These measurements I62?1 feet, as described above; and a foot unit of 0.326 m. The interaxialis slightly less than 25 feet. which has the only surviving column block in situ, Doors are placed in the outermost intercolumnar and where the column base is in fair condition.The spaces of the east facade; the presenceof the inter- base setback is preserved at N 7. 2) A vertical framing columnar wall in the four central spaces (preserved channel between column and intercolumnarwall alonly in interaxialsE 2-3 and 3-4) eliminatesthe possi- lows the centralflutes in the column axis to be fully bility of doors there, as has long been known.64Simi- carved,or almost so. The outer surface of the interlar doors may once have existed in the same positions columnar wall does not, therefore,extend up to the at the west end, which is now totally destroyed.Steps edge of the columns; instead, the shadow-producing must once have been cut into the raisedstylobateplat- channels interrupt it, giving the columns a greater form at the doors, both on the exteriorand interior; sense of plasticity, perhaps even giving the impresthe upper surfaceof the stylobateis so much higher sion that they were in the round. This channelis prethan the top step on the exterior (I.IO m.) and the served only at E 3-4, and has not, to my knowledge, paving blocks on the interior (ca. 0.80 m.) that such been previously discussed. 3) The thickness of the steps were necessary.They do not survive at E I-2 intercolumnarwall is 1.630 m. or 5 F, as measuredat E 2-3 and 3-4. Only in the latteris the outer surfaceof and are not indicatedon the plan. The two N-S cross walls which delimit the hy- the wall preserved,and there only in one small place. paethralcentralspaceat eastand west are shown in the The wall extends /4 F beyondthe axis of the columns; plan in somewhathypotheticalform. The locationand the normal base setbackof ? F was here reducedto thicknessof the walls is establishedby the surviving /4 F. The thickeningof the wall beyond the axes of foundations (thickness: east, ca. 1.95 m.; west, ca. the columns is undoubtedlyowed to the difficultyof 1.35 m.). The east wall is much thickerthan the E-W supportingthe architrave;it would have been masked, screenwalls running betweenthe piers,and so, unlike to some extent,by the framingchannels.4) The pilasthem, must have been a bearing wall, supportingthe ters are set back 2 F from the inner edge of the internal"pediment"behind the east facade (cf. ill. 4). stylobateand thereforedo not coincide with the proAny proposalfor doors in this wall must remaincon- jectedcircumferenceof the columns,which would fall jectural,for the foundationsare preservedno higher 3 F in from the stylobateedge. Flanking the pilasters than six coursesbelow interior floor level. The west- at the points where they join the wall were framing ern cross wall has the same thickness as the screen channels,analogousto those on the exteriorbut shalwall between the piers. If there was also an internal lower. Pilastersand basesare relativelywell preserved "pediment"here (there can be no doubt about the at S 13 and E 2-4, and dimensionshere are much more externalwest pediment,which is mentionedby Diodo- easily obtainedthan on the badly weatheredexterior. 63 Polyainos 6.5I; Diod. 11.25.2-5. 64 To facilitate reference to specific surviving details, I have numbered the columns of the temple: on the east, from south to north; and on the flanks, from east to west. E 2-3 thus refers to the second interaxial from the south on the east end.

BELL . FIG. i. Olympieion . PLATE 47 ' "; . *; ! : at Akragas, piers S 3 and 2, from W FIG.I. Olympieionat Akragas,piers S 3 and 2, frtlm W :i-:;-%-? :,? :A-" I I:_- i?:::e -??:"2::i" .r rl FIG. 2. Olympieion at Akragas, pier S 2, from W