Reference / Paper · 2020
The Antikythera Mechanism, Rhodes, and Epeiros
Festschrift chapter summarising Iversen's case for a Rhodian maker, an Epirote calendar and the 205 BC epoch.
- antikythera
- calendar
- dating
Reference / Paper · 2020
Festschrift chapter summarising Iversen's case for a Rhodian maker, an Epirote calendar and the 205 BC epoch.
The Antikythera Mechanism, Rhodes, and Epeiros Paul Iversen Introduction I am particularly honored to be asked to contribute to this Festschrift in honor of James Evans. For the last nine years I have been engaged in studying the Games Dial and the calendar on the Metonic Spiral of the Antikythera Mechanism,1 and in that time I have come to admire James’s willingness to look at all sides of the evidence, and the way in which he conducts his research in an atmosphere of collaborative and curious inquiry combined with mutual respect. It has long been suggested that the Antikythera Mechanism may have been built on the island of Rhodes,2 one of the few locations attested in ancient literary sources associated with the production of such celestial devices. This paper will strengthen the thesis of a Rhodian origin for the Mechanism by demonstrating that the as-of-2008-undeciphered set of games in Year 4 on the Games Dial were the Halieia of Rhodes, a relatively minor set of games that were, appropriately for the Mechanism, in honor of the sun-god, Helios (spelled Halios by the Doric Greeks). This paper will also summarize an argument that the calendar on the Metonic Spiral cannot be that of Syracuse, and that it is, contrary to the assertions of a prominent scholar in Epirote studies, consistent with the Epirote calendar. This, coupled with the appearance of the extremely minor Naan games on the Games Dial, suggests that the Mechanism also had some connection with Epeiros. The Games Dial and the Halieia of Rhodes The application in the fall of 2005 of micro-focus X-ray computed tomography on the 82 surviving fragments of the Antikythera Mechanism led to the exciting discovery and subsequent publication in 2008 of a dial on the Antikythera Mechanism listing various athletic games now known as the Olympiad Dial (but which I will call the Games or Halieiad Dial—more on that below), as well as a hitherto unknown Greek civil calendar on what is now called the Metonic Spiral.3 I begin with my own composite drawing of the Games Dial (Fig. 1). As one can see from the composite drawing, the Games Dial on the Antikythera Mechanism is divided into four quadrants labeled as in Table 1.4 The new reading of ΑΛΙΕΙΑ in the final position of year 4 is published 1 The core of this paper was delivered at a conference at the Lorentz Center in Leiden in June of 2013. I especially want to thank my collaborator John D. Morgan, whose perceptive comments have been invaluable. I also wish to thank the Department of Classics at Case Western Reserve University, the Dean of the College of Arts and Sciences, Cyrus Taylor, the Freedman Center for Digital Scholarship, and the Baker Nord Center for the Humanities for their moral and financial support. 2 3 Price 1974, pp. 13, 57-62. Freeth, Jones, Steele and Bitsakis 2008. 4 The 5 is the common abbreviation symbol for ἔτος (year), and the letters Α, Β, Γ, Δ stand for the ordinals 1st, 2nd, 3rd, and 4th. Instruments – Observations – Theories: Studies in the History of Astronomy in Honor of James Evans, ed. Alexander Jones and Christián Carman, 2020, DOI: 10.5281/zenodo.3928498, pp. 17–38. Chapter DOI: 10.5281/zenodo.3975715. Open access distribution under a Creative Commons Attribution 4.0 International (CC-BY) license. 18 Iversen Figure 1. The Games/Halieiad Dial. here in full for the first time and rests upon the following evidence from the micro-focus X-ray computed tomography.5 5Α ΄ 5Β΄ Ἴσθμια Ὀλύμπια Νέμεα̣ Νᾶα Year 1 Year 2 Isthmia Olympia Nemea Naa 5Γ ΄ 5Δ΄ Ἴσθ̣μια [Π]ύθ[ι]α Νέμεα Ἁλίεια Year 3 Year 4 Isthmia [P]yth[i]a Nemea Halieia Table 1. Inscriptions of the Games/Halieiad Dial. It should first be noted that each CT-generated photo is a slice of the Mechanism at a particular angle, direction and depth (and these slices can be combined into an accurate 3D X-ray volume and viewed with CT viewing software). On each slice only parts of some letters are visible, some more sharply than others. One must, therefore, stack up several image slices (in Adobe Photoshop) with the same absolute placement and read the images in a sequence. As the angle, direction and depth of each X-ray changes slightly, some letters or letter strokes that were blur5 This article is partly based on data processed, with permission, from the archive of experimental investigations by the Antikythera Mechanism Research Project in collaboration with the National Archaeological Museum of Athens (see T. Freeth, Y. Bitsakis, et al. 2006. In particular, I want to thank Mike Edmunds for releasing some of these data to me, and Tony Freeth and Alexander Jones for responding to my numerous queries. The mention of the possible reading ΑΛΙΕΙΑ by Zafeiropoulou (2012, p. 247) was based on a communication of my reading to her by other members of the AMRP. The Antikythera Mechanism, Rhodes, and Epeiros 19 Figure 2. Area of damage above the initial ΑΛ of ΑΛΙΕΙΑ. ry or invisible in one image come into view or even into focus in another, some more sharp than others, while others become blurrier or disappear altogether. Thus one must make a composite drawing on a graphics tablet by tracing what one sees as one moves through the images, each image adding more strokes to complete the picture. There are sometimes stray shadows or marks of damage (some of this a result of the image slice being at a slightly different angle or level) and often the letters themselves change appearance from looking like deep black grooves to letters in relief, so the technique of reading these images is not foolproof and is subject to interpretation. Nevertheless, the following analysis will demonstrate that the correct reading is ΑΛΙΕΙΑ. The final two letters, ΙΑ, are clear in several photos, so I believe all would agree these are not in dispute (see Figs. 3, 4 and 5). The interpretation of the first 4 letters, however, is more difficult. Their decipherment begins about ¼ millimeter above them on the surface of the Mechanism, where there are accretions and damage. Here one sees that an area of damage sits above the first two letters, which I have circled in red (see fig. 2).6 As will become clear, some of this area of damage runs down into the level of letters to interfere with their reading, and it is important to distinguish this damage from the letters themselves. Going to a slice a little below, one can see that the outlines of 6 letters appear, along with some of this area of damage circled in red (Fig. 3). Here it is especially important to note that to Figure 3. Damage at the level of the letters. 6 Special thanks to the late John Seiradakis and to Magdalini Anastasiou, who sat down with me in Thessaloniki and using the Studio Max VG software snapped these pictures. I also wish to thank Magdalini for sharing the beautiful Antikythera Mechanism font that she created. 20 Iversen Figure 4. The six letters of the games in year 4. the left of the red oval, some new strokes are visible (an apex along with a diagonal) that were not visible in Fig. 2 in this same area. This suggests that these are, in fact, letter strokes. In any case, as is clear from all relevant image slices, there are six letters, no more and no less. Stepping back a bit, all six letters are visible (Fig. 4). That these are the only six letters is made even more likely by the fact that the inscriber was careful to center names of the games around the Games Dial (with the exception of the Olympia in year one because he ran out of space next to the Metonic Spiral). Adding another letter to the left, the only place where an extra letter could have possibly gone and escaped the notice of the micro-focus X-rays, would have spoiled the appearance. Fig. 5 is a close-up of what one sees when one removes the layer of noise above the level of the letters. First it is important to note that with the area of noise above removed, all that is visible in this photo is on the same level of the letters. Thus any visible marks that were not visible in fig. 2 are likely to be letter strokes; conversely, any strokes in Fig. 2 that are still visible in Fig. 5 are likely to be damage. Hence the neat and crisp joining-up of the diagonal and cross-bar of the first letter on the same level as the rest of the letters and not visible in Fig. 2 strongly implies these are purposeful letter strokes, not damage. In addition, in some slices an apex and a left diagonal of this first letter are visible (see Figs. 3, 4 and 7, for example). Finally, these strokes are on an orientation that is slightly rotated, which is exactly what they should be to follow the arc of the Figure 5. Close-up of 6 letters, with about ¼ millimeter noise above removed. The Antikythera Mechanism, Rhodes, and Epeiros 21 Figure 6. Close up of Ε. circle around the Games Dial as the letters of the other games do. Thus the reading of Α for the initial letter is extremely likely, if not assured. To the right of the initial Α there is a thick, dark groove. This groove is on an orientation that is the same with respect to the damage above seen in Fig. 2. This is, therefore, very likely to be a part of the area of damage that starts above and continues down into the level of the letters. To the right of this thick, dark groove, one can see the second letter with a right diagonal as well as an apex. Enough of this second letter is preserved in several different slices (see Figs. 3, 4 and 5) to see it cannot be Α, Δ, Μ or Ν, but only Λ. It too is in the correct orientation for a letter on the arc around the Games Dial. The third letter consists of one faint vertical hasta that fits between two wider letters, and while a deep groove is not visible in any photo, the outline of an iota, including serifs, is visible in a few slices (see Figs. 3, 4 and 5). Based on spacing (i.e., the need to be a narrow letter) and the outline of a single vertical stroke in some photo slices, I believe the reading of an Ι is assured. As for the fourth letter, in a few photo slices, what looks to be the shadowy outline of a Ν appears. These were the first photo slices I was shown, so this was my first impression (and before I was given the data I spent several weeks of fruitless searching for games that matched the other traces and ended in ΝΙΑ), but when one looks at all the photo slices, it is clear that this fourth letter is an Ε, with parts of all strokes visible in some photos, especially the bottom horizontal and lower left corner (see Figs. 6 and 7), and that the seeming diagonal of a Ν in some photos is Figure 7. Ε from another angle. 22 Iversen actually a result of mistaking pieces of the upper left corner, the middle horizontal, and right side of the lower horizontal of the Ε for the diagonal of a Ν. This fourth letter was the key to reading the games, and if I have correctly identified the letter strokes and the damage, epigraphically speaking the reading ΑΛΙΕΙΑ may be given with no dotted letters (see composite drawing, Fig. 1).7 In addition to the epigraphical considerations, this reading is strengthened further by the fact that no other games of six letters consistent with the visible traces and ending in ΕΙΑ are attested in Greek or Roman sources.8 Finally, the AMRP team sensibly argued that these games were listed in chronological order,9 since otherwise it would make no sense that the Isthmian games were listed before the Olympia and Pythia.10 This hypothesis finds strong corroboration by the identification of the Rhodian Halieia, which the Scholiasts to Pindar say ended six days after the Nemean games had finished.11 Apart from the Scholiasts’ testimony that the Halieia were in the same year of the Nemea and shortly after them, from inscriptions we know that the Great Halieia were held every four years (the Greeks referred to this as pentaeteric, or every five years, 7 Based upon my composite drawing of letter traces, I am obligated not to dot any letters (the art of the epigrapher is to distinguish between real letter strokes and damage, make a drawing of the actual letter strokes, and then give a text according to the drawing). There is, of course, always the possibility that all the photo slices deceive in their totality, but this seems very unlikely. 8 The only other two attested possibilities that come even close are the Ἰλίεια of Ilion (see Hesychius: Ἰ λ ί ε ι α · ἑορτὴ ἐν Ἰλίῳ Ἀθηνᾶς Ἰλιάδος καὶ πομπὴ καὶ ἀγών). These are attested epigraphically one time (IG II2 3138, end of 4th/beg. of 3rd bc, but see Frisch 1975, p. 130, n. 52 for other references). These, however, appear to have been called the Panathanaia after 306 bc. There are also the Δίεια in honor of Zeus at Tralleis in Karia, which are also attested epigraphically only one time (SEG 22.350, l. 28). These, however, have only 5 letters, plus a delta in a position on the Mechanism for which there is no evidence of a lower horizontal where there should be, if this were the correct reading. 9 Freeth, Jones, Steele and Bitsakis 2008, p. 20. 10 The only other likely system for the arrangement is that of having the trieteric games on the inner circle, and the pentaeteric games on the outer circle, but given the Scholiast’s information (see note 10), it appears the maker had the games arranged trieterically/pentaeterically and chronologically. 11 The Scholiast to Pindar Olympian 7 147c, who was probably relying upon Istros’ lost work περὶ τῶν Ἡλίου ἀγώνων (On the Games of Helios), says of the Halieia τελεῖται δὲ μηνὸς Γορπιαίου εἰκοστῇ τετάρτῃ ἡμέρᾳ. ἀπέχει δὲ τῶν Νεμέων ἡμέρας ἕξ (“they finished on the 24th day of the month of Gorpiaios, six days from/after the Nemea.”). The Hypothesis to the Nemea d,e also says of the Nemea: καὶ ἔστι τριετὴς, τελούμενος μηνὶ Πανέμῳ ιη ΄ (“they are trieteric, and they finish on the 18th of the month of Panemos”). Most scholars have accepted the authority of these two scholia that the the Halieia finished on the 24th six days after the Nemea had finished on the 18th, but Perlman (1989, pp. 57-60) argued that the word ἀπέχει means “before” rather than its most natural meaning “after”, so that the Nemean games finished after the Halieia around the new moon (last day) of Argive Panamos rather than on the 18th of the month, six days before the end of the Halieia. She also ignored another Pindaric scholion that equated Argive Panamos with Julian Iulius (Mommsen 1867 Nem. Hyp. schol. 1, l. 38: ἤγετο δὲ μηνὶ Πανέμῳ ι̅η̅, ὅς ἐστιν Ἰούλιος), the month that normally most closely corresponded to Athenian Hekatombaion, and argued instead for late August or even early September (i.e., the season that normally most closely corresponded to the end of Athenian Metageitnion). Perlman’s heterodox view is now virtually excluded by the evidence of the Antikythera Mechanism. As for Perlman’s contention that the Gorpiaios mentioned here is from the Seleucid calendar, it is more likely it refers to Alexandrian Gorpiaios, since Istros is known to have worked with Kallimachos in Alexandria in the mid third century bc. In the time of the Roman Empire, which is the time when the Scholiast was probably working, Alexandrian Gorpiaios ran from June 25 to July 24, and thus corresponded most closely to Athenian Hekatombaion and Julian Iulius (see Samuel 1972, p. 177). My own thorough review of the evidence of the Nemea has concluded that they normally fell in the lunar month coincident with Athenian Hekatombaion (July/August), although there may have been times when they were in the lunar month normally coincident with Athenian Skirophorion (June/July). The Antikythera Mechanism, Rhodes, and Epeiros 23 because lacking the concept of zero they counted inclusively).12 In sum, when all this powerful and interlocking evidence is taken together, I believe the reading of Halieia is assured; thus the Games Dial is complete. Games Dial/Halieiad Dial vs. Olympiad Dial Naturally, a dial on a Greek time-reckoning device listing athletic games including the Olympia in a four-year period makes one immediately think of Olympiads, the four-year period between Olympic festivals that was a standard means to reckon time in Magna Graecia, but year 1 of an Olympiad in ancient Greek literary sources began with the celebration of the Olympia in even years bc divisible by 4 (200, 196, 192 etc. ) and ran until the same time of year in the next year when year 2 of an Olympiad began,13 whereas on this dial year 1 is beginning sometime between the Halieia of Rhodes (in roughly July of odd years one year before the Olympia such as 201, 197, 193..., if the testimony that they finished six days after the Nemea is reliable, which now seems confirmed by the Mechanism) and the Isthmia at Corinth (in the spring of even years bc) and it is ending with the Olympia, not beginning with them. To put this another way, the Isthmia definitely fell in years 2 and 4 of a traditional Olympiad, not in years 1 and 3,14 hence the years on this dial cannot refer to the four individual years of an Olympiad. It therefore appears that this dial does not reference individual Olympiad years in the traditional sense, which is why I prefer to call it the Games Dial, although one could rightly also call it the Halieiad Dial, since it does accurately represent Halieiad years. What other function this four-year period on the Mechanism had is not clear, but it may be that it was somehow used in combination with the 365 holes for the days of a solar year on the front of the Mechanism to achieve a leap-year function. The Halieia of Rhodes and the Naa of Dodona/Ambrakia The presence on the Games Dial of the Isthmia at the Isthmos of Corinth, the Olympia of Elis in the Peloponnese, the Nemea of Argos in the Peloponnese, and the Pythia of Delphi in Phokis is 12 See especially SER, p. 259, no. 5b (= IG XII,1 730), which lists the Halieia as being celebrated under the priests of Apollo Erethimios in years 8, 12, 16, 20, 24 and 28 of the list, thus assuring their pentaeteric cycle (the Halieia that should have appeared under the priest in year 4 was probably cancelled due to war; I would argue this apparently skipped year corresponds to the games of 85 bc at the very end of the First Mithridatic War). It is true that some inscriptions refer to both the Great and Small Halieia: IK RhodPer 555, l. 14: [Ἁ]λίεια τὰ μεγάλα καὶ τὰ μικρὰ δίς (mid. of 2nd cent. bc). Numerous other inscriptions mention only the Great Halieia: Clara Rhodos 2 (1932) p. 190, no. 19, l. 15 (early 1st cent. bc); Clara Rhodos 2 (1932) p. 188, no. 18, l. 16 (1st cent. bc?); Clara Rhodos 2 (1932) p. 210, no. 48, l. 4 (ca. 100-50 bc); NSERC, no. 36 (Roman period); SER 5, face b, l.3 (AD 4/5?); Tit. Cam. 75 (undated). It is hardly surprising to see the Small Halieia were largely ignored. 13 I have thoroughly reviewed the literary sources, and it is clear that the ancient historians who used Olympiad reckoning and whose works survive in quantifiable amounts (such as Polybios, Diodorus Siculus, and Dionysios of Halikarnassos) began a new Olympiad with the celebration of the Olympia, although they were not always precisely sure when it occurred given the variable nature of Greek luni-solar calendars, and they counted the individual years within an Olympiad by assuming the festival was always roughly at the same time of year (the evidence suggests between the middle and end of summer). 14 Based on a thorough examination of the season of the Nemea, Halieia and Olympia, I would also argue that the Nemea fell at the end of years 1 and 3 of a traditional Olympiad, and that the Halieia fell at the end of year 3. Although the Olympia and Pythia are in the correct Olympiad year, they should be at the beginning of years 1 and 3, not the end. I would also argue that the Naa should be at the beginning of Olympiad year 2, not the end of year 2. 24 Iversen hardly surprising – these were the four most prestigious Panhellenic games of Greco-Roman antiquity. The Halieia of Rhodes and Naa of Dodona in Epeiros, on the other hand, were relatively unimportant games, especially the Naa, so they do require an explanation. As for the Halieia, 35 of the 37 epigraphical attestations of the spelling Ἁλίεια are found on inscriptions from the island of Rhodes or the Rhodian Peraia.15 Of the remaining two, one attestation is found at Miletus on the west coast of Asia Minor,16 and a second copy of this same inscription at nearby Didyma.17 In sum, if other inscriptions are any guide, any attestation of the spelling Ἁλίεια,18 with an alpha and two iotas, has a 95% likelihood of coming from Rhodes itself, or in reality a higher percentage when one considers that the one attestation at Miletus and the one at Didyma are on monuments honoring the same athlete. That the Halieia were largely local and regional games for second-tier (or possibly older) athletes is not surprising, given that the games apparently often started just days after the Nemea had ended so there was not enough time for premier athletes to participate in both games in the same year. The appearance of the Doric form Ἁλίεια coupled with the fact that the Halieia were largely local and regional games, point to some relationship of the Mechanism to Rhodes, and this is especially notable given that the rest of the preserved inscriptions on the Mechanism employ Attic-Ionic forms.19 Apart from the Halieia, the other odd set of games on the Games Dial are the Naa. These games are even more obscure than the Halieia, with the spelling Νᾶα attested only once at Dodona,20 once at Tenos,21 once at Sikyon,22 once at Priene,23 three times at Athens (two of these on a single monument for Menodoros),24 and two times at Delos (on a single monument, also for 15 IG XII,1 57, l. 8 (#1); IG XII,1 72a, l. 2 (#2); IG XII,1 73, a, l. 3 and b, l. 3 (#s 3, 4); IG XII,1, 74, l. 2 (#5); IG XII,1 75, b, l.2 (#6); SER, p. 259, no. 5b (= IG XII,1 730), ll. 9a, 13a, 21a, 26a, and 32a; line 17a has the spelling ΑΛΕΙΑ, and although this spelling is common later, here it is almost certainly an error for Ἁλ⟨ί⟩εια, the spelling elsewhere on this same inscription (#s 7, 8, 9, 10, 11, 12); IG XII,1 935, l. 2 (#13, partially restored); ILindos 2.322, l. 10 (#14); ILindos 2.392, b, l. 8 (#15, partially restored); ILindos 2.707, l. 2 (#16); Tit. Cam. 63, ll. 22-23 (#17); Tit. Cam. 75, l. 6 (#18); Clara Rhodos 2.188,18, l. 16 (#19); Clara Rhodos 2.190,19, l. 15 (#20); Clara Rhodos 2.210,48, l. 4 (#21); NSERC no. 18, l. 6 and no. 19, l. 9 and no. 36, l. 4 (#s 22, 23, 24); SER, no. 4, face b, l. 3, face b, l. 16, and face c, l. 4 has the spelling ΑΛΙΑ, which is probably an error for Ἁλί⟨ει⟩α as is the spelling elsewhere on this same stone, and nos. 18 and 19 (#s 25, 26, 27, 28, 29); NSER p. 125, no. 25, l. 2 (#30); SEG 39.759, ll. 5 and 16 (#s 31, 32); IK RhodPer 555, l. 14 (#33); SEG 43.527, l. 22 (#34); Zimmer and Baïrami, 2008, pp. 149-154, l. 1 (#35). The restoration of [Ἁλίε]ια at IG XII,1 1039, l. 2 from the islet of Saros/Saria is not certain (the drawing in IG suggests the letter where there should be an Ε is more consistent with a Σ, yielding perhaps [Πύθια ἐν Μαγνη]σίᾳ, or [Ἀρτεμ(ε)ί]σια, or [Ἡμερά]σια), etc. 16 17 A. Rehm et al. 1958, no. 201, l. 11. Gerkan and Krischen 1928, no. 369, l. 20. 18 There are a three instances of the Attic-Ionic form Ἡλίεια (one at Athens, one at Samos and one at Ios), and several instances of the spelling Ἅλεια both around the Greek world and at Rhodes (which seems to be a later spelling that begins at the end of the first century bc). These alternative spellings are interesting, but for the purpose of this argument they can be ignored since they are not the spelling on the Mechanism. 19 For a history of the Halieia games, see my forthcoming article in Eulieme: Studies in Classical Archaeology, Epigraphy, Numismatics and Papyrology scheduled to come out in the fall of 2020. 20 L’Épire 586,71. 22 IG IV 428, l. 8. 21 23 24 SEG 37.709, l. 13. Freidrich and Hiller von Gaertringen 1906, no. 234. Agora XVIII, C-196, crowns 25 and 28; SEG 38.179 (here restored but likely = IG II2 3152+3153), crown X. The Antikythera Mechanism, Rhodes, and Epeiros 25 the same Athenian Menodoros),25 and once at Rhodes, where the games are said to be at Ambrakia.26 There is also the spelling Νάϊα attested one time at Tegea and three times at Messene.27 Of these 14 attestations, it is important to note that four, or almost 30% of all attestations, are in honor of the same Athenian athlete, Menodoros, who had monuments in his honor that were erected in both Athens and Delos ca. 135-130 bc that included his two Naan victories. The relatively rare attestation of the Naa in our sources suggests that the Mechanism also had some tie to Epeiros. The Calendar on the Metonic Spiral and Epeiros A further tie to Epeiros is found on the Metonic Spiral of the Antikythera Mechanism. Before summarizing the results of a comprehensive study of this calendar here, 28 first I want to address the question of whether the calendar on the Mechanism can be that of Syracuse, the home of Archimedes. The evidence clearly indicates that this is not possible, as the months Apo[llonios] (restored but virtually certain and not on the Mechanism),29 Karneios (not Kraneios as on the Mechanism),30 Artamitios with a tau (not Artemisios with a sigma as on the Mechanism),31 Panamos,32 and Apellaios33 are attested at Syracuse and its nearby military outpost of Akrai. In addition, the month name Damatrios is attested on an inscription that probably came from the area of Syracuse,34 which is a month name also found at the Syracusan foundation of Tauromenion, but not on the Mechanism. In addition, I have now read the month name Ἑλώρειος on an inscription from Tauromenion,35 but this month name probably originated at Syracuse (Syracuse had military fort at its southern border at the mouth of the Heloros river called Heloros, where we are told games called the Helor(e)ia were held,36 probably as a part of the festival after which the month Heloreios was named). It is thus likely that Tauromenion’s calendar, which is fully known (see Table III), came directly from Syracuse, the co-founder of Tauromenion. 25 IDelos 1957, crowns 25 and 28. 26 Zimmer and Baïrami 2008, pp. 149-154, lines 10-11, where the inscription reads reads … Νᾶα ἐν Ἀμ[- - - -]/ [- - - -]ς παγκράτιον…, and the correct supplement is almost certainly Νᾶα ἐν Ἀμ[βρακίαι· -?-] / [. . ca. 10. .]ς παγκράτιον. See Iversen 2017, pp. 147-148. 27 IG V,2 118, l. 21. The three examples from Messene are as yet unpublished, but will be by Andronike Makres (whom I would like to thank for allowing me to mention them). 28 The fuller article may be found at Iversen 2017. 29 IMagnesia 72, line 3. 31 SEG 42.836, line 4 (from Akrai, a dependent of Syracuse). 33 SEG 42.832, line 8 and probably SEG 42.835, line 6 (from Akrai). 30 32 Plutarch, Nikias 28.1-2. SEG 42.833, line 8 (from Akrai). 34 SEG 47.1462. The provenance of this inscription is in doubt; Manganaro (1997, no. III) originally argued for the area of Syracuse, but later changed his mind (Manganaro 2011 = BullÉp 2012.520) and now believes it comes from Halaisa Archonidea. The dating of SEG 47.1462, however, by the eponymous amphipolis (which is only attested as the eponymous office at Syracuse) and tribal/phratry designations in the form of ordinal numbers on this inscription strongly suggest Syracuse. 35 IG XIV 426, col. II, line 5 (I now read Ἑλωρε̣ίου) and Manganaro 1964, pp. 42-52, col. III, l. 30 (read Ἑλ[ωρείου πρ. Διο]γένης). 36 Iversen 2017, pp. 131-134. Cf. Hesychius Ε, 2253: Ἑλώριος ἀγών· τελούμενος ἐπὶ Ἑλώρου ποταμοῦ. 26 Iversen Antikythera Mechanism Phoinikaios Kraneios Lanotropios Machaneus Dodekateus Eukleios Artemisios Psydreus Gameilios Agrianios Panamos Apellaios Epirote (Cabanes) Kraneios Panamos Apellaios Machaneus Deudekateus Eukleios Artemisios Psydreus Gamilios Agrianios Phoinikaios Haliotropios Order/Seasons of Epirote (Cabanes) 7 (Aug./Sep.) 8 (Sep./Oct.) 9 (Oct./Nov.) 10 (Nov./Dec.) 11 (Dec./Jan.) 12 (Jan./Feb.) 1 (Feb./Mar.) 2 (Mar./Apr.) 3 (Apr./May) 4 (May/June) 5 (June/July) 6 (July/Aug.) Table 2. Comparison of the Mechanism’s calendar with Cabanes’s reconstruction of the Epirote calendar. As for other possible homes for the calendar, contrary to the claims of Cabanes,37 a leading scholar on Epeiros, the calendar on the Mechanism is consistent with all the known available evidence for the Epirote calendar. Here I give the calendar on the Mechanism and the reconstruction of the Epirote calendar by Cabanes side by side. As one can see, Cabanes disrupted the order of the five months in bold, started the calendar with Artemisios rather than Phoinikaios, and assigned the month Phoinikaios to June/July (the month normally coincident with Athenian Skirophorion), and the month Apellaios, in which the Naan games fell, to October/November (the month normally coincident with Athenian Pyanopsion). The main reasons Cabanes argued for this different starting point, order, and the particular seasons are because he believes that there is evidence at Korkyra. where we know Artemitios comes after Eukleios as it does on the Mechanism, that Eukleios was the 12th month, hence Artemisios the first.38 He also believes there is another month attested in Epeiros besides the twelve months on the Mechanism (namely the additional month of Haliotropios), which based on its etymology meaning “turning about the sun” should be placed around the time of a solstice.39 Finally, he believes the Naa games, which are known to have been held in the month of Apellaios, 37 Cabanes 2011, pp. 249-260. See also Cabanes, 2003 = CIGIME 2.2, pp. 275-288. 38 IG IX,12 4 798, ll. 50-51: ἐγδανεισάντω / ἐμ μηνὶ Δυωδεκάτωι καὶ Εὐκλείωι τῶι ἐπὶ Ἀριστομένεος (Cabanes reads instead ἐγδανεισάντω / ἐμ μηνὶ δυωδεκάτωι ⟨τῶι⟩ καὶ Εὐκλείωι τῶι ἐπὶ Ἀριστομένεος, turning what is clearly the month Dyodekatos into the ordinal 12th and positing a scribal error by inserting ⟨τῶι⟩). 39 I have inspected the stones or photos of the all the alleged instances of Haliotropios (with two iotas), and the correct reading on these is Lanotropios (as on the Mechanism), or what appears to be a variant (H)alotropios (with only one iota). Thus Lanotropios at IMagnesia 46, lines 2-3 (Epidamnos) and L’Épire, p. 553, no. 32 (Dodona). There is also (H)alotropios at IMagnesia 45, line 2 (Apollonia), as well as at CIGIME 2.2 76, line 3 and CIGIME 2.2 77, line 3 (both from Bouthrotos). I believe (H)alotropios and Lanotropios are different spellings for essentially the same month (based on two different roots that have a similar meaning), just as Karneios/Kraneios and Dodekateus and Dyodekatos/Deudekateus are. The Antikythera Mechanism, Rhodes, and Epeiros 27 fell in October/November.40 Elsewhere I have demonstrated that all these claims are demonstrably wrong, or likely wrong.41 The Antikythera Mechanism and the Corinthian Family of Calendars In Table III I summarize the evidence for months for what I call the Corinthian Family of Calendars. As one can see, apart from mostly minor orthographical differences, there is remarkable consistency in the names of the months attested at Corinth and its colonies in Northwest Greece and the cities of Epeiros, and that these months are consistent with those found on the Antikythera Mechanism. The most economical explanation of this remarkable consistency, with a total of at least 125 individual attestations,42 is that all the colonies of Corinth in Northwest Greece retained the calendar of Corinth with very few changes, which was the same or very similar to that on the Antikythera Mechanism.43 The evidence from Epirote cities is also quite consistent with the calendar on the Mechanism. The simplest explanation for this is to posit that at some point the Epirote Confederacy as a whole required that its member states adopt a fairly uniform calendar,44 and that calendar belonged to a city that was originally a colony of Corinth that had retained the Corinthian calendar. The Likely Candidates for the Calendar on the Metonic Spiral As for which cities could be the home of the calendar on the Metonic Spiral on this list, as we have already seen the calendar of Syracuse and its colonies in Sicily, can now be eliminated. Of those left in Table III, significant deviations from the Mechanism are found in the months of Alotropios at Bouthrotos and Apollonia instead of Lanotropios, and the apparent form Δυωδέκατος/ Δευδέκατος at both Korkyra and Apollonia rather than the third declension form Δωδεκατεύς. All these places have direct ties to Korkyra, and so it seems likely that Korkyra and its colonies can be eliminated as homes for the calendar on the Mechanism. Of the remaining candidates for the calendar of the Mechanism, the most likely are Corinth, Ambrakia, Dodona, and Epidamnos. If the Mechanism was built after the destruction of Corinth by Lucius Mummius in 146 bc, which is still an open question but seems likely to me,45 then 40 The evidence that the Naa games fell in October/November rests upon a demonstrably wrong argument made long ago by Klee 1918, pp. 54-55. 41 See Iversen 2017, pp. 149-159. Daux 1956 also argued that a month of Datyios is attested at Dodona at L’Épire, p. 534, no. 1, line 19. For a good photo, see the editio princeps of Evangelidis 1956, p. 2. I argue instead that this was a man’s name. 42 The data concerning the number of attestations are a bit skewed due to the 74 attestations of months at Bouthrotos alone, but nevertheless the remainder are an impressive amount of evidence. 43 That it was not unusual for colonies to retain the calendar of their mother city is demonstrated by Olbia, which clearly kept the entire calendar of its mother city, Miletos. See SEG 30.977 and 53.788. 44 The inscription published by Tziafalias and Helly 2007, ll. 57-83 (= SEG 57.510) suggests that the (post 167 bc) Koinon of the Molossoi possessed a common calendar, which was probably the same as the earlier Epirote calendar. 45 Carman and Evans 2014, Freeth 2014 and Jones 2020 have all persuasively argued the Saros Eclipse Dial has a start-up date of April 29, 205 bc. It appears the designer deliberately picked this date because the sun and moon were both very close to their apogees at this month’s full moon. As Alexander Jones has pointed out to me, this does not happen very often (the full moon on May 12, 205 bc is a better candidate than that of May 12, 91 bc or May 12 28 Iversen Mainland/Northwestern Greece Corinthian-Epirote Antikythera Φοινικαῖος Κρανεῖος Corinth Φοινικαῖος Ambrakia Φοινικαῖος Charadros Φοινικαῖος Λανοτρόπιος Dodona Gitana Byllis Φοινικαῖος Epidamnos Λανο[τ]ρόπιος 46 Μαχανεύς ΑΒ Δωδεκατεύς Εὔκλειος Ἀρτεμίσιος Ἀρτεμίσιος Ψυδρεύς Γαμείλιος Ἀγριάνιος Πάναμος Ψυδρεύς Ψυδρεύς Γαμίλιος? Ἀγριάνιος?48 Πάναμος Εὔκλειος Ἀρτεμίσιος 47 Πάναμος Ἀπελλαῖος Μαχανεύς Γαμίλιος Ψυδρεύς Γαμίλιος Πάναμος Ἀπελλαῖος Intercalated month Πάναμος Ἀπελλαῖος ἐμβόλιμος49 72 bc; it is not until the full moon of May 11, ad 308 that another good candidate comes along). Hence, I am (still) of the opinion that the Mechanism was designed within a generation of the Antikythera shipwreck (which sank ca. 70-50 bc), almost surely on Rhodes now that I have deciphered the Halieia on the Games Dial. 46 The intercalation on the Mechanism has now been read at Μαχανεύς in year 11 by various members of the AMRP, including me. 47 I agree with N. F. Jones (1980, pp. 165-177 and 1998) that SEG 30.990 (found on Delos) is likely to come from Corinth, or, as Cabanes and Ceka suggest (CIGIME 1.2.2.A, pp. 50-51), from Ambrakia. See also SEG 56.948. 48 [Ἀγ]/ρ̣ια̣ν̣[ίο̄] may possibly appear on Corinth MF-1975-86, a sacred calendar inscribed on lead found on Corinth’s Temple Hill that will soon be published by me in Hesperia (probably first issue of 2021) in an article of unpublished inscriptions from Corinth’s Temple Hill. The only other likely supplements are [- - τ]ρ̣ία̣ .[- -] or [- - τ] ρ̣ια̣κ̣[άδι]. 49 We do not know after which month the intercalary month was inserted at Epidamnos. The Antikythera Mechanism, Rhodes, and Epeiros 29 Mainland/Northwestern Greece Sicily Korkyraian Uncertain Syracusan Korkyra Issa Syracuse Φοινικαῖος Bouthrotos Apollonia Κρανεῖος Κρανῆος Κρα[νεῖος] Δυωδέκατος Ἀπελλαῖος 50 Μα[χανεύς]? Ἀπο[λλώνιος] Δευδέκατος Εὔκλειος Εὔκληος Εὐκλήιος Ἀρτεμίτιος Πάναμος Καρνεῖος Ἀλοτρόπιος Μαχανεύς Ψυδρεύς Akrai Φοινικαῖος Ἀλοτρόπιος Εὔκλειος ←?→ Ψυδρεύς Ἀρτεμίτιος Ψυδρεύς Γαμείλιος Γαμίλιος Ἀγριάνιος Πάναμος Ἀπελλαῖος Ἀρτεμίτιος [Ἑλώρειος] Δαμάτριος Πάναμος Ἀπελλαῖος Πάναμος Ἀπελλαῖος ἐμβόλιμος51 Tauromenion Ἰτώνιος Καρνεῖος Λανοτρό[πιος] Ἀπολλώνιος Δυοδεκατεύς Εὔκλειος Ἀρτε[μίτιος] Διονύσιος Ἑλώρειος Δαμάτριος Πάναμος Ἀπελλαῖος Ἀπελλαῖος β ΄ 50 If the Μα[- - -] preserved on J. Brunšmid 1898, no. 2-14, column I, l. 1. refers to the month Machaneus, it appears that Issa took the form of the Corinthian calendar that was in use by its closer cousins in NW Greece. 51 We do not know after which month the intercalary month was inserted at Bouthrotos. 30 Iversen Corinth could also be excluded. Of these, Ambrakia is the only one with the spelling Ἀρτεμίσιος (attested on an inscription dating to 167 bc52), rather than the expected Doric tau.53 In addition, Ambrakia and Dodona are attractive candidates since the Mechanism mentions the Naan games, which as we saw based on the sparse number of attestations seem to have been a relatively minor set of games. I would, therefore, argue that the Epirote calendar, as represented by Dodona and Ambrakia, is the most likely candidate for the calendar on the Metonic Spiral of the Antikythera Mechanism. The Antikythera Mechanism, Rhodes, and Epeiros In the foregoing discussion I have shown that the missing games in year 4 on the Games Dial were the relatively minor Halieia of Rhodes, and that the calendar on the Metonic Spiral cannot be the calendar of Syracuse, but that it is consistent with the known evidence concerning the calendars of Ambrakia and Dodona of Epeiros, both associated with the very minor Naan Games, which also appear on the Games Dial. It would appear, therefore, that the Mechanism had some connection with both Rhodes and Epeiros. How to explain this? It is clear that the Antikythera Mechanism is the product of an astronomical and philosophical tradition that developed or was wedded to a tradition of manufacturing such devices. Rhodes was one of the few places where similar devices are attested as having been manufactured (and most of the earliest evidence that concerns itself with this kind of astronomical device is connected with Rhodes in the first century bc, right around the time of the shipwreck, which is not likely to be mere coincidence). I would thus agree with those who have argued that the Mechanism was manufactured on Rhodes (and the reading of the Halieia is the first evidence found on the Mechanism itself supporting the association),54 but I would add the qualification that it was probably made for a client or recipient who came from Epeiros. Whether this owner or someone to whom it came into possession was on the boat when it sank, or the device was being transported for an owner in Epeiros as a stop-off point on the probable journey to Rome is impossible to know. I would further suggest that the Ur-Mechanism may have been designed for the Rhodian calendar, and the Metonic Spiral was merely adjusted by having the names of the Epirote months substituted for the Rhodian. The odd rotation of the Games Dial ca. 8° counter clockwise, or roughly one lunar month, may have also been made as an elegant adjustment for the one month diff