SOME REMARKS ON THE LATIN CURSE TABLETS FROM PANNONIA

Summary: This paper gives a short review of the research from recent years on texts of Latin curse tab- lets from Pannonia. In the last decade, four new lead tablets of quite long and well-readable texts came to light in well documented archeaological context in Pannonia. On one hand, these findings have not only doubled the small corpus, but they presented new data from both the field of magic and linguistics. On the other, in connection with the examination of the new pieces, the reconsideration of earlier ones could not be delayed any longer.


INTRODUCTION
For more than one hundred years, curse tablets have been representing one of the main direct sources of Vulgar Latin. Besides being essential for scholars in field of ancient magic and religion, these small sheets of metal usually containing a unique and 'hand-made' text may contribute to the results what linguists can obtain from the much more formulaic and even once checked and proved stone inscriptions.
In the past few years new pieces were found in Pannonia, 2 the grown number of these objects can now deservedly be considered as a source for linguistical research - and conversely, with the help of linguistic research former readings can be reconsidered. 3 The newly found curse tablets (coming from properly examined archaeological circumstances) not only have added new evidences to the Latin of Pannonia, but they present characteristics which were previously not known or were documented only from later texts in any part of the Latin speaking territories.
This study aims to answer three questions: (1) To what extent do curse tablets contribute to getting to know the spoken language of Pannonia? (2) Do they alter our view of the general language conditions in Pannonia reconstructed on the basis of other sources, mainly inscriptions? (3) Is the statement still tenable that the language of magical communication was Greek in Pannonia? 4 Curse tablets represent obviously a small unit of sources of Vulgar Latin. However, their number is growing day by day, since new pieces are found constantly in archaeological excavations, or even in repositories of museums. were found in the Eastern Greek language provinces of the Roman Empire -beside the thousand more pieces in Greek.
In each Latin language province, the proportion of Latin curse tablets in Latin inscriptions is about 0.1%. Britain has ten times more curse than the average. But excluding Britain, the number of Latin curses in Pannonia compared to other inscriptions could be also a mean value. In six civil towns of Pannonia, 10 Latin and 2 Greek curse tablets were found, 6 of which all are quite long and well readable texts. In Aquincum, three pieces were found recently 7 (Appendix Nr. 1-3). In Savaria, after the first Greek tablet of an 1975 excavation campaign 8 , two more Latin curses came to light in the last few years 9 (Nr. 4-6). One of the curses of Siscia is Latin, the other one is Greek (Nr. 7-8). From Carnuntum, Emona and Poetovio, one curse was published, alike . Finally, the Hungarian National Museum preserves one piece coming from an unknown Pannonian, now a Hungarian finding spot (Nr. 12).
Most of the curse tablets found in Pannonia has a surprisingly good command of language and style. In reality, however, we cannot exclude the so-called professionals (of a certain level of education) who were well aware of the formalities of this genre and might have used magical recipes. These persons and the circumstances of production are non-identifiable, nonetheless, the corpus of curses obviously attests that the texts were composed from pattern books on hand, or at least with full knowledge of magical formulae.
In respect to the Pannonian curse texts, they are in conformity with the conventions and the requirements of magic. 10 The tablets themselves represent expertise, they were made of lead as usual, hidden and found in customary places. The target person(s) were named definitely, sometimes the cursing people were mentioned too, usually deities were addressed, and special magical formulae were used, rarely in a certain creative sense. This last characteristic, i.e. the linguistic features indicate the importance of this rather small collection.

TECHNICAL MISTAKES 11
Curse tablets were essential objects of a ritual, of a magical process. The magician was to execute the ritual the most effective, the most precise way. All elements -including the curse tablet -might have been planned in advance. The persons who carved the text on the tablet in fact, generally must have been skilled and trained professionals. The tablets Nr. 1 and Nr. 2 can confirm this statement: the text-field on both tablet was prearranged by a vertical line or a margin. Further on, surface damages were evaded: Siscia tablet (Nr. 7), before writing on it, was folded into two, and where it became unfortunately damaged the scribe left that line blank. Savaria tablet (Nr. 5) proves the same producing steps: it might have been first pierced, and then written, as the text runs around the whole caused by the piercing.
At the same time, accurate planning could sometimes fail: when the scribe ran out of space to write in (Nr. 1), three lines were inserted between lines 3-5, written with smaller letters. In Nr. 2, on the other hand, the tablet was turned counterclockwise -so the text could continue that way above the previously written letters.
Letters on the Pannonian tablets are usually clear cut, they were written consequently, the texts can be read mostly easily. But the scribe did not always pay attention to the content of the text, since there are letters which were copied mistakenly from the draft: in Nr. 3, as a common mistake of inscriptions, E were written in cursive form, or turned in a wrong way. In Nr. 1 Annianus is mentioned twice, maybe mistakenly, as the scribe probably copied the name from the wrong line, therefore Annianus seems to be cursed and cursing at the same time. A similar phenomenon could have happened in Nr. 2 where the expression qui tibi epistulas tradet is repeated in the middle of the text without any reason. 12 Some misspellings show the total carelessness of the scribe who wrote senseless words or expressions. In Nr. 3 DIROV is meant to be diris because of the following word canibus which is taken by tradas. This form can be explained either by the simultaneous misinterpretation of an unusually curved I of the draft and a V-like (or checkmark-like) cursive S. 13 In Nr. 1 an M was miswritten as NV, creating the senseless form of Oceanunu, instead of Oceanum. Likewise, letters could be omitted (AVERSARIVS, EACVRA) without any linguistic reason since correct forms on the same tablet also occur.
Technical mistakes attest the negligence of the scribe whose work was not supervised. However, supervision cannot be expected for magic was prohibited. Thus, curse tablets can be ideal direct source material for the research of Vulgar Latin.

VULGAR LATIN FEATURES
Beside technical mistakes, all curse tablets from Pannonia show vulgar Latin features in the field of phonetics, morphology, syntax and vocabulary.
Not surprisingly, common mistakes are very similar to the ones of stone inscriptions: they can be either consequent recurrent misspellings, just like atversarius on Nr. 2 (actually a kind of hypercorrection or a pseudo-etymological dissimilation) or that kind of mistakes which occur only once in a text beside more correct forms of the same word, attesting the scribe's confusion due to language changes, e.g. the E-I alternation in OCIANVM/OCEANVM (Nr. 1), the loss of word-ending -m: CONTRA AMENE and CONTRA FELICONE versus CONTRA OCEANVM (Nr. 1), or the confused use of Q or C.
There are mistakes usually regarded as very common features, but their proportion is less considerable then expected. The followings may be regarded rather as Vulgar Latin mistakes and not just as simple technical mistakes or different orthographical traditions: the omission of i in Feliconem can be traced back to palatalization; the absence of the letter H is not surprising from the republican era; and the rare form of Cyllenii as Culeni originates in a geminate shortening and in the less common or maybe archaic transcript of the Greek upsilon. Respectes lingua instead of Respectae lingua and linguas illorum aversas as a subject 14 instead of linguae illorum aversae from (Nr. 1) can contribute to the changes or decline of the classical declension system, as well as the name list of the Siscia tablet (Nr. 7) where seemingly nominative and accusative endings are alternating in the same function. 15 12 Iteration is a usual tool in magic (KROPP, A.: Magische Sprachverwendung in vulgärlateinischen Fluchtafeln (defixiones). Tübingen 2008, 168-170), but unlike the other Pannonian texts this phrase was repeated nonfunctionally here. 13 Or even by a Vulgar Latin mistake, i.e. taken as diros canibus, it could represent an example of the confusion between inflections. HERMAN, J.: Vulgar Latin. Pennsylvania 2000, 49-68. 14  The Greek adstratum can be observed not only in Greek letter-forms (mainly sigmas and epsilons), but the Grecised ending of the Latin name Amoena (Amene). In addition, Greek names (Beroe, Chariton, Eunicus, Zosimus) testify these tablets were produced supposedly in multilingual circumstances. The cognomen in Nr. 2 after contra, BEROVENE must be an accusative form with an intrusive V of the standard common form Beroe. This version can be compared to those of Tyche and Nice from stone inscriptions, where the genitive is Tychenis and Nicenis. 16

VULGAR LATIN FEATURES -INNOVATIONS
Lexical-semantical innovations make these tablets outstanding not only in the field of Latinity used in Pannonia, but in a wider aspect.
Epistularius of Nr. 2, antepistula and Tricerberi of Nr. 3 represent their earliest occurence in ancient sources. The archaeological context dates the tablets to 2nd and 3rd centuries or with a little possibility to the beginning of the 4th century. Epistularius used to be attested much later, only from the 5th century at earliest. According to the Thesaurus Linguae Latinae, the office of epistularis was established in 384, whose occasional variant is the word epistularius. 17 The meaning of the word is worth to be examined. Either it refers to Mercury of the proceding lines (Mercurius Epistularius which would be a hapax legomenon) or someone unknown (maybe a nekydaimon or a spirit of the deceased person of this grave) could have been asked to hand over to Mercury in the underworld those who will hand letters (curses) to him. If the epistularius were the spirit of the buried person it would make the tablet resembling the Greek one of Savaria (Nr. 4). Up to now, antepistula was never found in classical Latin sources. Even the Greek noun was only once attested, whereas the verb was widely used. 18 The fragmentary but undoubtedly complemented Tricerberi is a unique form from such an early evidence. Latin and Greek authors mentioned this version from the late antiquity on. In Nr. 3 19 Ito can be read before Pater, without any more letter fragments. Nr. 2 can prove that it is not an accidental mistake or a misspell, because in line 3-4 the dative ITO PATR/I is used. They correspond to Vulgar Latin changes partially: There are two inscriptions offered to 'DITO PATRI'. Furthermore, there are four instances for the vocative DITE of which three are used in curse tablets. We can see that the classical adjective dis (ditis) used to have a variant o-stem form in Vulgar Latin. On the tablet at issue none of these quite common forms (dative or vocative) appears, but a third one does. As regards the O in the middle of the nominative form, it can be a weak form of the supposed *ditus (Ditus Pater  Ditopater) supported by such compound words' analogies just like mulomedicus, and vicomagister -('doctor 16 [---] and QVICVNQVA ATVERSARIVS / SVRGESERIT. The context makes it evident that the two verb supposed to be the same, but written differently. This form is a hapax legomenon (used in both of the new Aquincum curse tablets), but it is presumably either a hidden variant yet unknown or a newly built form of the verb surgo. It cannot be excluded that this form was created by more linguistic analogies simultaneously. For example, the remarkably similar perfectum of gero/suggero might have had an effect on it (gesserit/suggesserit), or maybe another perfectum, the one of the semantically similar suggredior could influence it (suggressus erit  *suggresserit), or perhaps it was anaptyctically created following the perfectum forms of spargo, mergo (surgsi  *surgesi); or it might be conceivable in the archaizing context of curse tablets that this irregular perfectum form goes back to an imperfectum surgesso by analogy of the types peto -petesso, facio -facesso, capio -capesso. At last, a metathesis cannot be excluded either, i.e. surgexi/surgessi could be created from the standard surrexi. 22 In line 7 of Nr. 1 a persuasive analogy begins: 23 Quomodo hoc ego averso graphio scribo, sic linguae illorum aversae ne possint facere contra hos … ego supra posivi. According to paralells where aversus occurs, a part or the whole of text is written abnormally, backwards or upside down. There are a few exceptions where no modification was made and usually aversus is interpreted there as 'hostile'. 24 Thus, the translation could run as follows: 'Just as I write this in hostile style, so may their hostile tongues be unable to act against these … (whom I mentioned above)'. 4 years after the tablet was found, a bag was opened in the museum's storage room containing other archaeological findings from the closest surroundings of this curse tablet, and there was a style, a stilus among them. However, hairpins, stili, nails come to light very often from cemeteries and they can be interpreted in different ways: they could be simple personal belongings, but they could be set there as apotropaic (against harmful spirits) or protective (defending those alive, against unlaid ghosts) objects. The stilus found next to the Aquincum curse tablet was examined by the Restoration Department of Budapest History Museum. 25 The examination proved that it must have been bent before getting underground (and not during the excavation) as corrosion was found in each side of the bent part; the bending was intended, because the 20 I am grateful for Romain Garnier's comment. 21 BARTA: Ito Pater (n. 2) 106. 22 I am grateful for Oswald Panagl's comment. inner side of the bent part was prepared for bending probably by a semilunar chisel in order to make the bending easier. By comparing this stilus with the evidence of the magical papyri, the concordance is evident: -PGM VII 397: ἐπίγραφε χαλκῷ γραφείῳ 'inscribe it with bronze stylus'. The Aquincum stilus was made of bronze, too. -PGM IV 1847: κυπρίῳ γραφείῳ γράψας ψυχρηλάτῳ 'with a cold-forged copper stylus'. The cross-section of the Aquincum stilus is rectangular, probably the stilus was hammer-harden. -PGM VII 442: γράφε δὲ (τὴν πλάκαν) χαλκῇ βελόνῃ ἀκεφάλῳ 'write [the spell] with a headless bronze needle'. The end part of the Aquincum stylus is definitely outlined. Probably it was produced without head by intention. Both the sources and the physical examination confirm that the bent stylus found next to the curse tablet is the aversum graphium mentioned in the text of the curse text. With the graphium in hand, the interpretation changes: 'Just as I write this with a bent, twisted stylus, so, too, may their bent and twisted tongues be unable to act against these … whom I mentioned above'.

RECONSIDERATIONS
After certain period of time, reconsideration of the first editions is required. The Siscia curse tablet (Nr. 7) was published many times, each of them showing differences in certain sections. The two names in line 4 have given rise to much controversy, the publications do not agree.  26 A recent autopsy presented a new reading: The missing D of a -nd-cluster of SECVNO was so common that beside the epigraphical instances it can be found in the Appendix Probi, too. After a thorough personal examination an L came to light in the place of the C. So, counting with an opening of e to a before r, we got the well known nomen VALARVS > Valerius (with a missing semiconsonant in -ius ending). The different endings of the name belonging to one man are recurrent characteristics of the text.
Three editors gave three different drawings and reading to the first line of the Siscia tablet. The first editor, Brunsmid and the AIJ gave it sense, but the latest one did not explain at all what these words should mean. Since each of them worked on base of photocopies they do not agree even on the character number. According to the recent autopsy, all letters could be interpreted and explained: MADATA DATA SS / SAVO. Mandata data and other similar words are often used in curses, referring to the names or lists of names which are handed to the addressed deities. For example Nomina data, delata, legata ad inferos ('the names of our enemies are given, sent and entrusted to the infernal deities' CIL XIII 7550); quorum nomina hic scripta et demandata habetis 'whose names you have got here written and entrusted to you' dfx 11.2.1/31). SS is a common abbreviation in inscriptions and in curses, too. It was used mainly in Britain where the curses had a slight judicial character, they often followed the style of agreements (Donatur deo s(upra)s(crip)to decima pars 'The tenth part of the worth of my stolen things will be given to the before mentioned god ' AE 1964, 168). It occurs also in verb form, too (Adligate linguas horum quos suprascripsi ne adversus nos respondere possint. 'Bind the tongues of these (or those) persons whom I mentioned above' dfx 11.01.01/04). On the base of these examples, two new interpretations arise: Ma(n)data, data s(upra)s(cripta) Savo '(All the names) mentioned above (in the inner side of the tablet are here) given and entrusted to Savus, the river god'. Or Ma(n)data, data s(upra)s(cripto) Savo '(all these names on the inner side of the tablet are here) given and entrusted to the above mentioned Savus, the river god'.
The growing number of curses can offer newer readings to the tablet of uncertain origin, now preserved in the Hungarian National Museum, Budapest (Nr. 12). 27 Slight modifications 28 can offer a more coherent interpretation. In line 4, after totum instead of the genitive corporis the accusative corpus would be a more adequate form, as the phrase was never attested as a partitive. The text targeted a woman and her body, by naming some body parts, too. Between cutem 'skin' and ungues 'nails' the editors proposed the word corpus again for the fragmentary C[---]. On the base of curse texts, other words beginning with c-could be more plausible solutions, 29 for some possible traces of an A after the C, e.g c[apillos], referring to her hair which is also a mark of beauty as skin and nails, and can represent the whole body together with other bodyparts. 30

CONCLUSION
After having considered the main characteristics of the small set of Pannonian curse tablets, it may be possible to give answers to the three starting questions. Due to the proportion of curses in other inscriptions our general knowledge of Pannonian Latin can expand only in a qualitative way. Pannonia was a Latin-speaking province with many people of oriental origin. The written communication was mainly Latin, too, even the Greek texts mention Latin names (the target of the Greek curse of Savaria is a certain Adiectus, son of Cupita). But in the field of vocabulary curses can supply new evidences. The creative world of magic and rituals, and the innovative characteristics of Vulgar Latin are well matched, texts born in these circumstances can be the objects of many further studies. . 29 The first editors saw the traces of an O thre, but on the base of own autopsy I suppose an A. Words beginning with c-from curse texts where body parts are targeted: capillus 'hair', color 'healthy look, i.e. skin colour', consilium 'intelligence', collum 'neck', corium 'skin', cor 'heart', cerebrum 'brain', crus '(lower) leg', cunnus 'the female organ', caput 'head'. 30