1Much information has come to light about dedications of armor and weapons in the past quarter century. Christopher Simon, and more recently, Maria Gabaldón Martίnez have compiled surveys of votive arms and armor throughout the Greek world.1 To my knowledge, however, there has been no attempt to evaluate this data quantitatively. Therefore, using both archaeological and epigraphic evidence, I compiled a database of 146 sanctuary sites and corresponding dedications of weapons and armor (both full-size and miniature) from the eighth through the fifth centuries BCE.2 My goal was to find out whether conventions of arms dedication resulted in a pattern that favored certain deities, and if so whether such patterns were correlated to the gender of the deity. If certain categories or combinations of arms were typical gifts for specific gods and goddesses, a quantitative analysis might yield this information. I also hoped to discover whether the local character of a given deity outweighs any other factors in shaping the choice of votive gifts of arms, or whether consistent practices can be identified across the Greek world that correspond to the Panhellenic persona of individual deities.

2Current methods of analysis of votive patterns tend to rely primarily on the subjective impressions of individual scholars, however knowledgeable and skilled, about the relative frequency of votive finds in specific contexts. The method proposed here allows us an opportunity to test objectively the assumptions we have formed through regular study of these materials, but it suffers from its own characteristic limitations.3 First, my information is necessarily incomplete because of gaps in the material record and in the reporting of that record. Second, although this approach is quantitative, many subjective and even arbitrary judgments are involved in the organization of the data. To give just one example, I have identified the Gortyn acropolis as a sanctuary of Athena, despite some uncertainty about the identity of the acropolis goddess in the Archaic period. The method used here also demands the identification of a “primary deity” for each sanctuary, in spite of the fact that many sanctuaries housed a pair or a number of deities. Finally, future finds could change the conclusions I draw here, but it is a useful exercise to ask what the existing data have to say about these questions.

Fig. 1. Relative frequency of finds: offensive arms

3Because of potential gaps in the data due to soil conditions, looting, and other factors, relying on absolute numbers of objects uncovered is not a sound method. Instead, I have focused primarily on whether a given site yields examples in a specific category of object: spears, shields, helmets, and so on. I assumed that the presence of a broader range of arms, both offensive and defensive, correlates with a stronger military focus in a given cult, or with special functions of certain sanctuaries that relate to the display and manufacture of arms. Other types of votives, such as terracotta warriors, can also signal a deity’s military interests, but weapons and armor differ from figurines in their direct connection to the work of the battlefield, in their intrinsic value, and apparently in their relative rarity as votive offerings.

4In 1974, Kendrick Pritchett stated that bronze arms have been found in “almost every excavation in Greece.” Christopher Simon and others echo this assessment, treating arms and armor as routine finds.4 As it happens, this is an exaggeration. Many, if not the majority of excavated sanctuaries have not yielded arms at all, and large concentrations of surviving arms, as are found at Panhellenic sanctuaries, seem to be quite rare in the archaeological record.5 Offensive arms, especially spears and arrows, appear far more often than armor, but even spearheads, the most commonly reported arms dedication, are present at only slightly more than half the sites (Fig. 1). When spearheads do appear, they typically are not preserved in large numbers, even at sites where other types of bronze dedications are plentiful. Of the forty-nine sites where numbers of spearheads are recorded, thirty-four (69 %) had ten or fewer examples, and only five (10 %) of the sites had more than one hundred examples. Finds of basic hoplite armor are relatively unusual (Fig. 2). Shields and helmets are found at 33 % and 22 % of the total sites respectively, while specialized armor such as greaves and corselets is rare (at 8 % or less of the total sites). About a third of the sites in the database had only a small scattering of finds, and nearly a third had only one type of object, usually spearheads or arrowheads.

Fig. 2. Relative frequency of finds: defensive arms.

5Additionally, at sites where arms are present, they tend to be a tiny proportion of the total number of preserved votive offerings. In these cases, should the offerings be dismissed as anomalous, or, as more often happens, should they be given nearly as much weight in evaluating the cult as superabundant types of offerings, such as pins or terracottas? How many different types of arms, and how many individual objects should be present before we can speak of a votive practice that is characteristic of a cult? These are difficult questions with which we have barely begun to come to terms.

6In the absence of epigraphic clues, archaeologists who excavate the sanctuary of an unknown deity must draw what conclusions they can from the votive offerings. Weapons dedications are sometimes described as expected, routine finds when the primary deity of a sanctuary is male, but draw more comment when the primary deity is female. If, as is often said, early Archaic votive offerings reflected the concerns of the donor rather than the character of the deity, we should expect the pattern of arms deposition to be gender-neutral with respect to the deity. But even if the category of object to be given was arbitrary, the donor still had the ability to determine the recipient. Did the highly gendered status of weaponry in daily life affect the selection of divine recipients, in the same way that Greek deities tend to receive victims and have priests of their own gender? Can we assume that the presence of many votive arms means that the resident deity in an unidentified sanctuary is male?

Fig. 3. Distribution of sanctuaries with arms by primary deity

7The examples of Olympia (Zeus), Isthmia (Poseidon) and Delphi (Apollo), the three largest repositories of dedicated arms and armor, might lead us to believe so. However, these Panhellenic sanctuaries are exceptional in many ways. The massive dedications of arms there may have had more to do with the inter-state function of the sanctuaries than with the gender of the presiding deities (and of course, these sanctuaries contained the shrines of numerous gods and goddesses). In fact, a quantitative approach to the available data indicates that the majority of excavated sanctuaries where arms were found are those of goddesses, particularly Athena. In the database of 146 sites with arms, goddesses are the primary deities at 53.4 % of the sanctuaries (with Athena accounting for a plurality of those sites), gods are the primary deities at 30.1 % (with Apollo accounting for half of those sites), 6.8 % are heroes and heroines, and 9.6 % are unknown. If we look at the breakdown by deity (Fig. 3), we see that while Athena and Apollo are at the top of the list, Zeus has surprisingly few sites (and four of his ten sites are in Crete, which is unusually rich in votive arms).6 The “other female” deities include Aphrodite, while the “other male” deities include Poseidon, Hermes, and Ares.

8Clearly, the gendered nature of weapons and armor does not inhibit their dedication at goddess sanctuaries. For our hypothetical excavator trying to establish the primary deity of a sanctuary, the presence of weapons or armor indicates, if anything, a greater chance that the deity is female. On the other hand, occurrence of arms dedications could be proportional to the total number of sanctuaries allotted to each deity, and thus simply a function of how popular the deities are and, more crucially, whether they are typically sanctuary gods. If the deposition of arms is gender neutral, but there are more goddess sanctuaries to begin with, this could explain the preponderance of goddess sites with weapons. We can correct for this possibility by looking to see if male or female deities have more than their expected share of sites with specific categories of arms, where the expected share is based on random distribution. For example, goddesses have 53 % of the total sites, so they are expected to have about the same percentage of sites with shields, or sites with offensive arms. When we analyze the data this way, the goddesses’advantage in having a greater number of sanctuaries disappears. Using this method, there are no statistically significant differences by gender in the numbers of sites with individual weapons types, but we do find that male deities have significantly more sites with full-size defensive arms and with pieces that make up a full or partial hoplite panoply (Fig. 4).7

Fig. 4. Distribution of some arms types by gender

9What about individual deities? Generally speaking, individual types of arms tend to occur in proportion to the number of sanctuaries attributed to a given deity. For example, Athena has 21 % of the total sites, 27.3 % of the sites with arrows, and 25.7 % of the sites with swords. Hera has 7.4 % of the total sites, 7 % of the sites with arrows, and 8.5 % of the sites with swords. Furthermore, arms are not restricted to certain deities, but instead they are widely distributed among a large number of different gods. It seems that nearly any deity can receive arms under the right circumstances. Again, could it be that the greater number of sites in the database assigned to Athena and Apollo simply means that they have more sanctuaries than other deities? Considering the great popularity of Artemis as a sanctuary deity compared to her relatively poor showing in the database, I think this is only a partial explanation. The greater number of Athena and Apollo sites with arms in fact represents a military affinity of these gods that is part of their Panhellenic personality and holds true from one geographical region to the next, and over time. The Athena and Apollo sites are distributed all over the Greek world; the Artemis sites are also widely distributed but there are a few gaps (Crete, Northern Greece).8 The Hera and especially the Demeter/Persephone sites are heavily weighted toward Magna Graecia and Sicily, while the Zeus sites are weighted toward Crete. This data supports the conclusion that dedications of arms to Athena and Apollo corresponded to elements of their Panhellenic personalities, while dedications of arms to other deities were always an option, but were based to a greater extent on contextual factors such as the type and geographical location of the sanctuary.

Fig. 5. Distribution of some categories of arms

10If we examine the data for individual deities, some significant patterns of dedication can again be detected. First, let us recall the percentage of the total number of sites attributed to each deity set forth in Fig. 3. We expect to find sites with each type of weapon distributed along similar proportions, and in general, this is true. No deity has a significantly greater number of sites with any individual type of object. If we look at aggregate categories, however, we find significant patterns (Fig. 5). Athena has a significantly larger share of sites with miniature arms. Apollo and (to a lesser extent) Zeus have a significantly larger share of sites with full-size pieces from the complete hoplite panoply.9 This is not the case for any female deity, not even Athena. Overall, any given type of arms appears at more Zeus and Athena sanctuaries than expected. This general finding is illustrated by the case of the Apollo and Athena sanctuaries at Halieis in the Argolid. The Athena sanctuary yielded miniature bronze shields, helmets and greaves, but no full size armor or offensive weapons. The Apollo sanctuary had plenty of swords and spears, but no miniature arms.10 As for Zeus, he is more like Athena with respect to the regular appearance of a wide variety of arms types at his sanctuaries, but more like Apollo with respect to the focus on functional hoplite arms.

Fig. 6. Arms in Artemis sanctuaries

11With miniatures we cross from the category of so-called “raw” votives, those not originally produced for votive use, into the category of “converted” votives, and this has important consequences for their distribution and the range of donors.11 Not only are miniatures generally more affordable and thus accessible to a wider variety of dedicators, including women, the very fact that they are miniatures can make them less heavily gendered, and more acceptable as gifts from women. A full-size, functional spear is perhaps not likely to be offered by a woman, but a miniature spear may have been. In such cases, the character of the deity tends to determine the gift rather than the situation or gender of the donor; hence a woman may dedicate a miniature shield to Athena because of the goddess’warlike nature and role in defending the city. The inscribed bronze shield from the Athenian Acropolis dedicated by the bread-baker Phrygia is often mentioned, but the inventories from the Erechtheion also reveal a gold miniature shield dedicated by a woman named Phylarche.12 The use of gold and silver for such objects, while reducing their affordability, removes them even further from the gendered arena of the battlefield. The Classical Erechtheion contained a collection of such choice pieces, including silver and gold shields, silver helmets, and silver spears. It is possible that Athena received a greater number of miniature items because she was perceived for various reasons as more accessible to women.

12It has long been suggested that goddesses who receive arms are given these gifts because of their kourotrophic role as protectors of the state’s youths.13 As this argument goes, a deity concerned with youths and their maturation, and especially with rites of passage to adulthood, will inevitably come to share in military concerns. I began my investigation with this hypothesis, and though there is some evidence to support it, it is not satisfying as a global explanation for arms in goddess sanctuaries. First of all, if the goddess is a poliad or federal deity, or simply the major deity of the dedicator’s hometown, this factor alone is sufficient to explain the presence of spoils and other arms in her sanctuary, just as it is for equivalent male divinities. The goddess may be kourotrophic, but need not be, and it does not follow that this element is at the root of the custom of dedicating arms.14 Second, on this hypothesis, we should expect to find significant amounts and ranges of arms in sanctuaries of Artemis, but the data do not bear this out (Fig. 6). The most common arms in Artemis sanctuaries are arrowheads and spearheads, and these may be associated with hunting rather than war. Of thirteen sanctuaries with arms where Artemis was the primary deity, only one, Kalapodi, yielded a significant number or range of objects other than arrowheads. Kalapodi was, of course, a federal sanctuary of the Phokians, located at a spot of great strategic importance, and shared with Apollo.15 The kouro-trophos theory actually works best for Apollo, particularly at several of his Peloponnesian sanctuaries where the pattern of deposition favors offensive weapons.16 This suggests that it is not a question of spoils, since literary sources and inscribed pieces show that helmets and shields were more highly favored as spoils dedications. Instead, gifts of spears, swords and arrows might be related to the documented use of Apolline sanctuaries as mustering places, arsenals, and training grounds for young warriors.17

13The hypothesis that cults of Hera throughout the Greek world had an important military component that focused on rites of passage for young men has also gained wide currency. Such rituals provide an apt explanation for the miniature and other arms that have been discovered in many goddess sanctuaries. Philip Brize suggested that some seventy-five miniature bronze and terracotta shields found at the Samian Heraion were dedicated when young men received their first arms and became adult warriors. This theory has been embraced with respect to the major cults of Hera, and expanded to cover the sanctuaries of Persephone in Lokroi and her colonies; and such sites as Palaikastro and Prinias in Crete.18 As a poliad deity, Hera certainly took an interest in military matters, but current scholarship relies too heavily on the hypothesis of rites of passage for young men at the Argive Heraion, where surprisingly few arms were found.19

14Better evidence for a rite of passage involving weapons exists at those sanctuaries in the western poleis, such as Metapontum and Naxos, where spearheads and other weapons were deposited together with the remains of sacrifices in connection with what appear to be gentilicial cults. Such cults are associated with a number of different deities, including Hera and Apollo.20 In any case, it is likely that most miniature bronze helmets, corselets and greaves, as opposed to shields, were dedicated as substitutes for full size spoils when individuals could not afford to give the real thing. This is the best explanation of their presence at Apollo’s sanctuary at Bassai, where the earliest examples seem to have some connection with the second Messenian war, and the miniature arms from Delphi and Olympia also illustrate this point.21

15Miniature shields of various materials including bronze, terracotta, and precious metals are a far more complex category and must be considered separately. It is certainly possible that miniature shield votives were used as Brize suggests in certain places and times. Yet shields are polyvalent objects and the rite of passage theory cannot account for all — or even most — of the contexts in which they appear. For example, miniature shields are popular grave gifts and are used as votives in so-called “tomb cults” and in hero cults, such as the deposit found at Menidhi in Attica and the deposits in the Potter’s Quarter at Corinth.22 Finally, no statistically significant concentration of miniature shields is associated with any one deity (though Athena comes closest), nor with female deities over male, once we correct for the larger number of goddess sanctuaries.23

16To conclude, I would like to suggest that quantitative analysis, although methodologically challenging, offers us a path forward in our study of votive offerings, which are so plentiful yet so difficult to interpret. Among other benefits, it allows us to test our assumptions and helps us to filter out those that are unwarranted. One such assumption is that arms and armor are unusual finds in sanctuaries of goddesses, and therefore require a special explanation, such as a rite of passage, when they are present.