Some remarks on the personal name system of Raetic 1

The Raetic language is fragmentarily attested in an epigraphic corpus of a few hundred inscriptions in Iron Age Northern Italy and the Central Alps. The corpus is one of the North Italic corpora, which comprise the evidence for epichoric literacy beyond the river Po before the spread of the Roman Empire, and document a variety of languages: – the Cisalpine Celtic languages Lepontic and Cisalpine Gaulish west of the river Adige, – the probably Italic, certainly Indo-European Venetic in the Veneto and Friuli, – Camunic of unknown affiliation in and around the Oglio valley, – Raetic, a Tyrsenian language, in the Trentino and South and North Tyrol.


Introduction
The Raetic language is fragmentarily attested in an epigraphic corpus of a few hundred inscriptions in Iron Age Northern Italy and the Central Alps.The corpus is one of the North Italic corpora, which comprise the evidence for epichoric literacy beyond the river Po before the spread of the Roman Empire, and document a variety of languages: the Cisalpine Celtic languages Lepontic and Cisalpine Gaulish west of the river Adige, the probably Italic, certainly Indo-European Venetic in the Veneto and Friuli, -Camunic of unknown affiliation in and around the Oglio valley, -Raetic, a Tyrsenian language, in the Trentino and South and North Tyrol.
All these languages are written with very similar alphabets which were first derived from the Etruscan alphabet of Central Italy around 600 BC.The Raetic writing culture is slightly younger than those of Venetic and Cisalpine Celtic, starting in the late 6 th century and coming to an end, like the other North Italic epigraphic traditions, in the late 1 st century BC.Raetic inscriptions come from the area of Verona, from the Alpine foothills up to Trento, the Val di Non, the Bolzano area, the Upper Adige, Eisack, Wipp and Inn valleys and surrounding highlands, including petrographs in the Northern Limestone Alps.Outliers were found in Slovenia on helmets of the Negau type.The corpus comprises almost 400 inscriptions on about 300 objects, but only ca.40% of them are certainly language-encoding.About a third of the documents consists of non-or para-script marks.The language-encoding texts are prevalently dedicational, inscribed on votive or ritual objects made predominantly of bronze and antler; owner's inscriptions may also be represented.
1 The present paper was presented in two parts at the conference Personal Names and Cultural Reconstructions (Helsinki, 21-23 August 2019) and at the GfN-conference Bewegte Namen.Anpassungsprozesse von Eigennamen in räumlichen, zeitlichen und sozialen Spannungsfeldern (Münster, 11-13 September 2019).The research was partly funded by FWF -Austrian Science Fund (project no.P 25495).
The number of funerary inscriptions is low in comparison to the Venetic and Cisalpine Celtic corpora. 2  The Raetic language is related to Etruscan; together with Lemnian in the Aegean, Raetic and Etruscan form the Tyrsenian language family (Rix 1998: 159f.). 2 A complete edition of the Raetic corpus is provided by Thesaurus Inscriptionum Raeticarum (TIR), whose sigla are used in this paper.Sigla for Etruscan inscriptions refer to ET; sigla for Venetic inscriptions refer to Pellegrini/Prosdocimi (1967); sigla for Cisalpine Celtic inscriptions refer to LexLep.The citation of Raetic forms follows the transliteration standard of TIR: transliteration letters represent Raetic characters in a one-to-one correspondence; no phonetic or phonological interpretation is implied.This is particularly pertinent in the area of obstruent spelling, which is not at all transparent (Rix 1998: 50-57;Salomon forthc.).Thus, θ is theta (but is likely to denote a dental stop), z is zeta (but may also denote a dental stop according to Venetic Este orthography), while þ represents the Raetic characters which denote the dental affricate (Schumacher 2004: 304f.).The sound values of phi and chi are unclear.The sign ° indicates that a form is cited without the case or derivational ending with which it is attested, and replaces the hyphen when citing auslauts which are not or may not be grammatical endings.

Personal names in Raetic inscriptions
As is the case with many languages which are only attested epigraphically, the prevalent text types in Raetic inscriptions are ones which contain mainly personal names -dedications name the donor, marks of possession name the owner, funerary inscriptions name the deceased.Raetic personal names are attested either as single individual names (e.g., SZ-8 kaθiave, VN-9 lavise, BZ-2 enikes [gen.],IT-2 χaisurus [gen.]) or as elements of a two-part name formula, which consists of an individual name and a surname (e.g., . Surnames are derived from individual names by suffixation of -nu or, less common, -na, e.g., kaniś-nu ← kanise (vel sim.) + -nu.Personal names are identified internally by their context (appearance as part of a name formula), by their grammatical form (appearance in the genitive or pertinentive case) or by their typically vocalic auslaut -(i)e, -i, -a, -u, externally by comparison with names attested in other corpora. 3In the ca.160 inscriptions which lend themselves to linguistic analysis (many of them too damaged to be of use), about seventy sequences can be identified as personal names with some certainty; another fifty or so may also qualify.Table 1 gives a list of twelve individual names which can be argued to be attested more than once.(See table 4 for a list of names which are attested both as individual names and as bases of surnames.) 4piθamne MA-1 piθamne, AS-18 piθamn[, AS-19.1 piθamn[, *MA-2 piθanme, IT-8 piθan[, BZ-9 piθame5 lavis(i)e VN-9 lavise, CE-1.1 lavise, VN-1 lavisie, WE-1 lavise-s (gen.), *AV-1 lavise-z (gen.)lasta SZ-1.1 lasta, SZ-15.1 lasta, WE-3 lasta-si (pert.) remi(e) SZ-2.1 remi, SZ-2.2 remi, VR-3 remie-s (gen.)φ(i)rima SZ-2.1 φrima, SR-5 φrima, SZ-1.1 φirima φel(i)turie SZ-14 φeliturie-si (pert.),NO-3 φel(i?)turie-si (pert.),PU-1 φelzurie-s (gen.) piθi About 45 names can be analysed or even etymologised to a certain extent, usually owing to comparanda in Etruscan, but mainly Celtic, Venetic and Roman documents.The Raetic inscriptions share a lot of their onomastic material with pre-Roman and Roman inscriptions from the surrounding areas of Northern Italy, especially the distinct onomastic group of the area around Brescia (Untermann 1959: 151-154).The direction of borrowing is often uncertain, as illustrated by the case of ST-2 esimne°,7 but a considerable number of the personal names attested in the Raetic corpus appear to be loans from the other, mostly Indo-European languages of the Southern Alpine area, while compelling connections with Etruscan onomastic material are scarce.Of the personal name bases in Raetic inscriptions, up to ten can be furnished with plausible Indo-European etymologies: from Celtic esumn-, 8 vaþ-, 9 vinuθal-, 10 kaθ-, 11 kar-, 12 χais-, 13 from Venetic valθ-, 14 usθ-, 15 from an unspecified Indo-European language klev-, 16 φrim-17 .
Marchesini's analysis of the sequence as klan-θur° is formally acceptable (but see n. 48), but remains doubtful in light of the problematic reading (see Salomon 2018: 67).
of Brescia may be Tyrsenian features, Raetic appears to have been on the receiving end of onomastic loans.
The dissociation of the Raetic name material from that of Etruscan stands in contrast to formal similarities which concern two aspects: the standard auslauts of names, and the suffixes used to derive surnames.
Etruscan does, however, also have common nouns in °e to provide models for cognomina and praenomina in °e (Rix 1963: 230f.).While there are many loans from Italic languages, the most common Etruscan praenomina seem to be vernacular, and a number of them end in °e (e.g., larice, śeθre); two can be shown to be formed from Etruscan lexemes with a suffix -e: aule < avile from avil 'year', uśile from uśil 'sun' (the latter word a loan from Sabellic ;Rix 1995a: 723;Wallace 2008: 92;De Simone 1970: 141). 51In light of these native forms, De Simone 1970: 142 considers the loan names from Italic to have been transferred into a vernacular stem class (IE o-stem → Etr.e-stem), the similarity of the resulting forms with Indo-European vocatives being coincidental.The existence of a vernacular e-class can be reconciled with the interpretation of names in °e as Indo-European vocative forms in two ways (Adams 2003: 514).An inherited Tyrsenian e-class and the Indo-European vocative endings may 49 This also applies to words for objects which are borrowed in the accusative; cf.Steinbauer (1993: 287f.).50 Eichner (p.c.) considers the suffix -i to be internally developed from -ie by analogy via the syncopated genitive (-ies > -is).51 Different Eichner (2012a: 33;2012b: 28 [n. 83]), who interprets both names as loans from Italic: avele (avile from weakening of posttonic vowels) ← *aelos 'grandfather [diminutive]', uśile/uśele ← Sabell.*oselos < *asēlos (cf.aurelius) 'manifestation of the sun' (← *asōs 'dawn').Cf.Steinbauer (1999: 402, 493).
have had a reciprocal effect, the existence of a suffix -e supporting the choice of the familiar-sounding Indo-European vocative forms, the resulting dominance of names in °e in turn making the vernacular suffix more productive in onomastics (Rix 1963: 231;De Simone 1970: 142;Stifter 2013b: 50).Alternatively, the Etruscan e-suffix may itself be borrowed from Indo-European languages; this would presuppose that Etruscan had been under considerable Indo-European (Italic?) influence for some time, so that the imported suffix was productive at least at the time of the first literary attestation.Such an extension of borrowed suffixes to vernacular bases is indeed evidenced by -ie from the Italic patronymic suffix *-io-, which became productive in Etruscan as a suffix for nomina in the 7 th century (Wallace 2008: 93f.).
As pointed out by Stifter (2013b: 52), the situation in Raetic is even harder to judge, as the absolute number of names which can be connected with Indo-European ones is much lower, and even in the majority of those cases it is not entirely clear to which language the name originally belonged and exactly in what form it was borrowed.Also, the Raetic names concerned are not nomina or cognomina, but only praenomina (individual names).Leaving aside names whose auslaut is uncertain because they are only attested in suffixed form, 52 the best candidates for Raetic names in °(i)e borrowed from Indo-European names in -()os are enike, esumne and maybe esimne, klevie, kaθiave, lavise and lavisie, lasθe, piθamne, pitale, piθiave, piθ/tie and reiθe.As in Etruscan, these names in °(i)e with plausible Indo-European derivations or at least connections stand beside such names without an established Indo-European connection (e.g., lumene, knuse, tnake), and the same considerations apply: Raetic onomastic -e may be a vernacular name formans which happened to coincide 52 There are five clear cases in which names in °ie appear with case endings (gen.BZ-3 terunie-s, remie-s, φelzurie-s, pert.kastrie-si, φel[i]turie-si).This shows that the case endings at least of inflectional class I (gen.-s, pert.-si) do not mask more complex underlying auslauts.The only exception would be pitie ~ gen.pitis; it is therfore preferable to consider pitis the genitive of a name variant *piti (cf.possible piθi in TR-3).Unlike the case endings, the suffixes -nu and -na can be shown to be attached to simplified auslauts.The evidence of piθamne ~ piθamnu, kastrie ~ kaszrinu and maybe aruse ~ aruśna (see table 4) -implicitly also lavise ~ kaniś°, reiθuś° (if formed with the same suffix) -shows that auslauting °e may be or is regularly dropped in these forms; there is no attestation of a surname in °(i)enu/a (with a possible exception in the irregular MA-14 esiumninu ~ esi/umne).Rix (1998: 30 [n. 41]) mentions the possible relevance of the prehistoric Etruscan uσelna-rule, i.e. a vowel in third syllable being syncopated between continuants.It must therefore be considered that any number of those names which are attested with °i before -nu/-na (e.g., meti-nu, φutiχi-nu) may be names in °ie, not °i.Under this consideration, we may have around thirty-five names in °ie, which agrees with the high frequency of names in -o-in the area of Brescia (Untermann 1959: 153).
with the Indo-European vocative ending and supported the choice of vocative forms in Indo-European loan names -in which case it represents an inherited connection with Etruscan -, or it may be an imported auslaut which became a productive suffix.Raetic names in °i, however, can hardly, like their putative Etruscan equivalents, be explained as vocative forms borrowed from Latin.°i alternates with °ie in remi ~ remie° and, arguably, piθ/ti ~ piθ/tie.It may be noted that those names which are attested in °i without suffixes are all short (piri, SZ-10.1 χeli, χari/kari, remi, piθ/ti; the identification of BZ-6 φanaχi as a name is doubtful), so that -i might be a suffix for forming hypocoristic names.Alternatively, it may be a suffix for feminine names (see section 3).
In Etruscan, feminine names could be derived from vernacular onomastic stems with vernacular suffixes (e.g., -θa, -θei) or from masculine names with the suffixes -i, -ia and -a (Wallace 2008: 92), all of Indo-European origin.Indo-European -a or -ā 53 is also the standard suffix for feminine names in Northern Italy.It is likely that any number of the Raetic names in °a are feminine, though the only possible name pairs in Raetic to parallel, e.g., Etr.masc.śeθre ~ fem.śeθra are lasθe ~ lasta and maybe lavise ~ VR-14 lav(i)śa (Salomon 2018: 48f.).If we discount Rix' uncertain direct equation of φrima ~ Ven.frema (see n. 17) on account of the fact that Pellegrini/Prosdocimi (1967 I: 95f.) posit a nominal base Ven.*frema which is homophonous with the hypocoristic feminine name, we do not have any full equations of Raetic names in °a with demonstrably feminine names in other corpora.We do, however, have such equations for masculine names in °a: Untermann (1959: 143, 147) remarks upon the frequency of masculine names in °a in the area of Brescia, citing, among others, CIL V 4376 vassa (see n. 9) and CIL V 5070 tula (cf.very uncertain tula-nu° in ST-6).He notes that none of the cited names contains o, which is generally rare in the onomastic material of the area of Brescia, and tentatively suggests a sound change /o/ > /a/.It might be considered whether this scarcity of o is in fact not due to a sound change within an Indo-European language, but the reflection of Raetic sound substitution. 54In any case, some of the Raetic names in °a may belong in this group.If ‹a› should reflect the Raetic rendition of Indo-European [o], these names could be loans from Indo-European ōn-stems with nominatives and vocatives in °o; however, some Raetic names in °u make a bid for the same position.
According to Rix (1963: 180-192), Etruscan cognomina in °u are common nouns formed with the lexical suffix -u, while, in nomina, the suffix is introduced analogically from the cognomina or imported with Italic praenomina in °o (the two most frequent Etruscan nomina in °u being petru ~ Ital.petro and pumpu ~ Ital.pompo, both formed from Indo-European numerals with the suffix *-ōn-). 55Again, both options are viable for Raetic individual names in °u.Untermann (1959: 153) considers the possibility that Raetic °u reflects the ōnstems' long °ō.The only name in °u with a suitable parallel from another corpus is φausu ~ bauso.The name has no Indo-European etymology; bauso may only be the Latinised version of a name borrowed into Raetic from another source, so that it is not certain that °u in the Raetic form reflects Indo-European °ō (Schumacher p.c.), but it is certainly likely.On the other hand, Rix (1998: 20 [n. 23]) suggests that the Raetic suffix -nu may contain the lexical Tyrsenian suffix -u; we cannot at this point exclude that -u was productive in Raetic onomastics.
Etruscan has numerous praenomina with consonantal auslauts in the casus rectus (e.g.aranθ, venel, velθur, laris) -they are the consequence of prehistoric apocope, as can be gathered from the fact that the original vocalic auslauts are preserved in suffixed forms (e.g., gen.venelu-s).In Raetic, only one individual name attested without suffixation ends in a consonant: laθur appears as suffixed laθuru-si (pert.),providing a parallel with Etruscan.Though laθur (and likely the other two names in -uru; see n. 48) can hardly be the only Raetic name with a consonantal auslaut, no other individual name which is attested unsuffixed ends in a consonant. 56While it cannot be excluded that such names do exist but are not recognised as names by circular reasoning, there must still be a decided statistical preponderance of auslauting vowels.
55 Cf. also Steinbauer (1993: 294-297).56 As said above, auslauting °e may be dropped before -nu/-na, so that, again, suffixed names which appear with a consonant before the suffix may have an underlying vocalic auslaut.Drawing by Gudrun Bajc ©TIR.

The distribution of -nu and -na
According to Schumacher (1998: 101;also 2004: 296;Rix 1998: 19), the opposition between the Raetic suffixes -nu and -na, which derive surnames from individual names in Raetic, is likely to reflect the bearer's gender, with -nu for masculine, -na for feminine names.This seems intuitively plausible, because auslauting °a is rarer, and by far the most common feminine marker in Northern Italy with its three Indo-European languages.Neither fact, however, is compelling.Firstly, women are prominent among donors in Northern Italy; there are entire sanctuaries which, judging by the names in votive inscriptions as well as typical donations, were predominantly frequented by women -e.g., the stylus votives of the Reitia sanctuary in Venetic Este, and the burnt-offerings site on the Demlfeld in North Tyrol (cf.Blecha 2013).Secondly, as discussed in section 2, °a is, if anything, better established as an auslaut for masculine than for feminine individual names in Raetic.Also from a Tyrsenian perspective, a distribution of -na for feminine and -nu for masculine names is not easy to explain.The suffixes -nu and -na are generally considered to be connected with the Etruscan derivational suffix -na, which forms adjectives of appurtenance 'belonging to, associated with'.While it is also known from lexical contexts in Etruscan, -na was the most widely used suffix to form the prehistoric adjectival patronyms, which were eventually turned into family names (Wallace 2008: 93); it is the vernacular Etruscan equivalent of the Italic *-io-suffix (Rix 1963: 295).These Etruscan patronyms are mostly derived from individual names (that of the father or the "titular head of the family" [Wallace 2008: 79]), and are still found as the standard surnames in archaic Etruscan in the 8 th and 7 th centuries.In the course of the 8 th century, they were turned into family names (nomina) as part of the Central Italian Sprachbund shift towards inherited family names.
When used to form patronyms in prehistoric Etruscan, and later nomina, the suffix -na appears in two forms: -na in masculine names, -nai (> -nei) in feminine names (Wallace 2008: 88f.).The Etruscan suffix' feminine variant is transparently derived from -na with a feminine marker -i; this is an imitation of Italic conditions and dates to the time when the family name system emerged (Rix 1995b: 728). 57The suffix -i itself is probably of Indo-European origin (Agostiniani 1992: 54;Rix 1998: 20;2004: 951); originally, Etruscan had no grammatical gender (Rix 1994: 951).This is likely to be true for Raetic as well.It would have to be assumed that the Raetic suffix was also reformed under Indo-European influence, with -na shifting to the feminine under pressure of Indo-European gendered endings.For -nu, Rix (1998: 20 [n. 23]) suggests that the Tyrsenian agentive suffix -u, which is attested in Raetic in lexical context (deverbal uti-k-u, elu-k-u), may have caused a recasting of the suffix in masculine names.
57 In Etruscan, this innovation is connected with a functional distribution of case allomorphs (genitive/ pertinentive I vs. II), where masculine nomina take class-I endings (-s, -si), feminine nomina class-II endings (-a[l], -ale) (Wallace 2008: 88f.).In Raetic, all surnames are inflected in class II (-[a]le), so that this parameter cannot be used to determine gender.
A gender-based distribution of -nu and -na, though theoretically feasible, is also hard to demonstrate convincingly based on the evidence within the corpus.Table 2 gives all potential name formulae with surnames in -nu/-na attested in Raetic, without inflectional endings.There is no overlap between individual names combined with surnames in -nu or -na, which supports the theory that the suffix variants are gender-specific, but many of these examples are contingent, based on difficult readings, uncertain segmentation and unclear text structure.Of the six individual names going with -na, four end in °a, though φelna vinuθalina is the only really unproblematic example. 59In kunina° θauχrilina, only the individual name appears to be inflected (see section 4).The analysis of the unsegmented sequence karataśna as a name formula is, in my opinion, very likely, 60 but siara kuhilina on the opaque inscription on the wooden stave from the Ritten is quite uncertain.Despite the caveats, the predominance of names in °a combined with surnames in -na can hardly be coincidental, even if the claim that °a is a feminine marker remains tentative.At least one individual name in °a, lasta, appears also accompanied by surnames in -nu.In SZ-1.1, the segmentation of the sequence χikaśiχanu in line 2 is not certain, but lasta and φirima appear in line 1, and at least one of them is likely to go with the surname in -nu; it may be that all three share the same surname.See n. 70 on the relation of the names in SZ-2.1. 61Still, with fifteen of the twenty-four individual names which go with surnames in -nu, certainly masculine names in °(i)e dominate clearly.One, maybe two names in °u (vaþanu, esθu°) are also best counted among the masculine ones.Individual names in -i appear in both columns.azi° may really end in °ie, and the relevance of another (piti°) is questionable due to the uncertain analysis of the text (see n. 69), but piri and remi as well as χeli are unambiguous.
With regard to the overall passable statistics and what was said in section 2 about auslauts of individual names as gender markers, the best conclusion at this point is that both masculine and feminine names could end in °a and °i in Raetic, and that the combination of either of these auslauts with patronyms both in -nu and -na does not constitute evidence against the theory that the suffix variants mark gender.The positive evidence is by no means compelling, but the only outright contradictory case is the name formula aruse θarna, which combines an individual name in °e with a surname in -na.The reading and segmentation of the text on the heavily damaged antler piece are not obvious, but supported by SR-3 aruśnas, which can be analysed as a formally 59 As argued by Schumacher (2004: 337), the damaged St. Andrew's cross after final alpha is best interpreted as a delimiter (cf.NO-7).See section 4 on individual names in °na.60 The bronze is inscribed with two lines of text SZ-14 φelituriesielukusletile | karataśna.Schumacher (1998: 109f.)analyses the second line as one word and the patronym to go with sletile in line 1 (see section 4), while I prefer to interpret the latter not as an individual name, but as a surname sleti-le in the pertinentive to go with φeliturie-si (see table 3), and to segment the sequence in line 2 into a separate name formula kara taśna in the nominative.61 φrima also appears as part of a name formula in SR-5 φrima piθamn[, but here the last letter of the inscription is damaged: only the rightmost tip of the last letter is left, so that it may be alpha or upsilon.
unobjectionable surname from aruse in the genitive aruś-na-s (on s vs. ś see n. 46).See, however, further on this problematic group of names in section 4.
No surname-forming suffixes other than -nu/-na have so far been identified in Raetic, though the wealth of such elements in neighbouring traditions and in Etruscan strongly suggests that they must be there. 62Likely name formulae with surnames which do not end in -nu/-na are given in table 3. CE-1.1 and SL-2.1 in the casus rectus feature one certain (lavise), one possible (siraku) individual name followed by forms with auslauting -i.The other three potential formulae are inflected in the pertinentive case with the surnames ending in °ile, which might be analysed as an auslaut °i or °ie plus the pertinentive II allomorph -le, as usual in surnames.At this point, we can do no more than point to the forms' similarity; whether they are surnames formed with a suffix -i or individual names with the common auslaut -i/-ie serving as surnames (or something else entirely) must for now remain open.

Patronyms vs. nomina in Raetic inscriptions
The Raetic forms in -nu and -na have so far been carefully referred to as surnames, but their status as patronymic forms has been evident to researchers of 62 A word-final element -þu occurs three times at Magrè (MA-2, MA-5, MA-23).The element is opaque and could also be a grammatical ending (though there are no Etruscan comparanda).Name formulae and single names occur in equal parts at Magrè, and the inscriptions from that find place abound in hapax legomena, so it is not obvious to expect surnames, but it may be observed that all three elements which are suffixed with -þu (θurie°, kuśi°, usθi°, for the latter see n. 15) may find comparanda in the North Italic onomastic pool -provided that the above readings are correct (MA-2 θurieþu is only reconstructed from the lower tips of the letters, MA-5 kuśi° looks more like dubious kunii°).In MA-2 and MA-5, the sequences concerned are preceded by individual names, with short and obscure sequences in between, while usθiþu would be a single name, unless the questionable zezeve represents an individual name.Cf.Markey (2006: 157) and Markey/Mees (2003: 140).
Raetic since the first successes in text segmentation (Vetter 1954: 74).It is not certain, however, whether this status persisted into the later phases of Raetic.
A shift from the archaic patronymic system to the complex family-name system which is typical for Central Italy occurred in Etruscan in the late 8 th century, and it may well have spread to Raetic at some point during the half-millennium of the language's attestation.Apart from the etymological connection of the Raetic suffixes with the Etruscan patronymic suffix, there are four arguments which support the assumption that Raetic had a productive patronymic system at least during some of the time of its documentation: the statistics of individual and surnames, the evidence of the inscription group ST-1-3, the use of a suffix -alu in VN-1, and the arguable existence of name formulae in which only the individual name is inflected.In an onomastic system based on inherited family names (nomina [gentilicia]), the number of individual names (praenomina) is usually restricted; the majority of nomina derived from personal names are based on praenomina which are no longer used (Untermann 1961 I: 39;Rix 1998: 18f.) -thus in Neo-Etruscan, where the number of commonly used masculine praenomina is limited to seven, with about five more rarely used ones (Rix 1995a: 720;Wallace 2008: 82).In contrast, in a patronymic system, the individual names and the patronyms which are derived from them are attested in equal measure and side by side (Untermann 1995: 733).In Raetic, due to the fact that individual names are often attested on their own, we know overall more than twice as many individual names as patronyms.Up to seven names are attested both as individual names and as bases of patronyms in the Raetic corpus, shown in table 4.There is no chronological pattern to these data, i.e. the individual names do not belong to an earlier phase than the patronyms.Most important is the testimony of the rock inscription group ST-1-3 with the names of three related men: ST-1 kastrie° eθunnu°, ST-2 pitau?e° kaszrinu° and ST-3 esimne° kaszrinu°.The person named in ST-1 bears the surname Eθunnu, while the persons named in ST-2 and 3 bear surnames which cannot but be derived from the individual name in ST-1 -we assume that P/Pitau?e and Esimne are Kastrie's sons (Schumacher/Salomon 2019: 169f.).Unfortunately, the three inscriptions, being petrographs, cannot at this point be dated, so they do not give us a terminus post quem for a transition to a family name system in Raetic.As already suggested by Rix (1998: 18 [n. 18]), the surname in VN-1 lavisie lavisealu appears to be formed with the Celtic patronymic suffix -alo-, well attested in Cisalpine Celtic (Untermann 1959: 87; LexLep s.v.-al with literature). 67The Celtic suffix is not an unproblematic element in itself, 68 but it is one of the productive patronymic suffixes which are assumed to be translated with the genitival formula in Roman inscriptions from Northern Italy (see section 5).That it could, for whatever reason, replace Raetic -nu in an otherwise typically Raetic text (name, text type, support) indicates that the two suffixes functioned in the same way, viz.as productive patronymic suffixes.The antler piece cannot be dated directly; the settlement on the Tartscher Bühel in the Upper Adige valley flourished during the early and middle La Tène period (5 th -3 rd century), 67 If this analysis is correct, it must be noted that, here, Indo-European -os is reflected by Raetic -u rather than -e -assimilation to the typical Raetic auslaut of patronyms?68 The suffix cannot be explained from Indo-European.Lejeune 1971: 52 considers it a thematised version of the Tyrsenian genitive II -al, which appears to be widely accepted (Markey/Mees 2003: 138;Stifter 2020: 26), but is not unproblematic -Lejeune seems to think of a loan from Raetic, where this allomorph is not certainly attested.
but must be expected to have been in use until the 2 nd century, when it was essentially replaced by the Ganglegg settlement (Gamper 2006: 290f.).The last argument may be furnished by a phenomenon noted by Schumacher (1998: 110, 112), viz.that patronyms could apparently remain uninflected beside inflected individual names.Schumacher's examples are SZ-9.1 kunina-si tauχrilina and SZ-14 sleti-le karataśna, both on Sanzeno bronzes dated to the 5 th -4 th century (Gleirscher apud Schumacher 2004: 247).In both inscriptions, when segmented like this, the individual name is inflected in the pertinentive case (with the allomorphs -si and -le, respectively), while the patronym has no grammatical ending.SZ-14 is problematic, as it is not quite clear how the three or four name elements on the bronze relate to each other (see n. 60), but the two names in SZ-9.1 can hardly be anything other than a name formula. 69The practice would clearly be marginal, but in kunina-si tauχrilina at least the non-inflection of the second name may indicate that the writer considered it to be grammatically dependent on the individual name ('by Kunina of θauχrili°' rather than 'by Kunina θauχrilina'), which makes more sense for a genitival patronym than for a nomen (Untermann 1959: 81f.).
Our main reason to consider the existence of inherited nomina in Raetic is the issue of single names which end in °nu and °na. 70Ten names ending in °nu and seven names in °na are not readily identifiable as patronyms: SZ-5.1 vaþanu, BZ-3 laśanu°, NO-17 ketanu°, CE-1.2 velχanu, MA-16 valθe?nu, possibly ST-5 ...) tulanu° and ST-6 ...)estanu°; CE-1.3 lupnu, AV-1 tipruχnu and SZ-87 esminu; SZ-31 remina, SZ-18 χevisiana, SZ-9.1 kunina°, CE-1.5 φelna, SR-3.1 arusna°, possibly VR-3 nakina and VR-1 tinesuna.Of these, vaþanu, kunina and φelna appear as individual names in otherwise unproblematic name formulae (see table 2).Three of the names in -nu -lupnu, esminu and tipruχnu -may straightforwardly be patronyms if we accept the possibility that patronyms can precede the corresponding individual names.But eleven names ending in °nu or °na are not accompanied by a second name of any kind.It is not clear whether some or all of these names are individual names which just happen to have that auslaut, individual names which were derived with the same suffixes that are also used to derive patronyms, or actual surnames formed with the patronymic suffixes used as single names.
We have ample evidence for individual names unaccompanied by surnames, and the three names φelna, kunina and vaþanu demonstrate the existence of individual names which end in °nu and °na.However, a coincidental similarity is hard to argue for all the forms, as there is only one name-forming n-suffix in Northern Italy, and it is very rare.The suffix -an-is attested sparsely forming individual names in Venetic (Untermann 1961 I: 135f.), beingmuch as in Raetic -more common for forming surnames (Untermann 1961 I: 83f.); five instances are known from the area of Brescia (Untermann 1959: 135).The suffix may account for those names in °nu/a which must be interpreted as 70 A potential piece of epigraphic evidence is easily dismissed: in SZ-2.1 φrima remi visteχanu, the first two words are well established individual names which appear to share the surname visteχanu.This might support an interpretation of the latter as a nomen, under the assumption that the text records the donation of a couple and Φrima took on her husband Remi's family name.However, there are many caveats to this interpretation.Firstly, the assumption that the individual names φrima and remi are coordinated is not well substantiated; it stems from the obsolete interpretation of a mark in the form of chi embossed on the edge of the bronze, around which the writing is arranged, as enclitic -χ 'and' (Rix 1998: 21;Schumacher 2004: 334 with n. 208;cf. Salomon 2018: 55f.).Secondly, even if φrima and remi should be coordinated, Φrima's father's name may have been left out to save space, a married woman's patronym being considered unimportant when she is named beside her husband.Furthermore, as discussed in section 2, auslauting °a does not mean that φrima is a feminine name; Φrima may as well be Remis brother, so that they are both visteχanu.Many of the name conglomerates in the inscriptions on the Sanzeno bronzes defy the attempt to segment them into discrete name formulae; this example can hardly bear the burden of evidence.
individual names (Untermann 1961 I: 140 mentions vaþanu and also velχanu as Raetic names which might be formed with -an-), but hardly for the whole conspicuously large group -not least because of the variation in (or absence of) the thematic vowel: names in °inu/a and °Cnu/a (e.g., kunina or aruśna) cannot belong here.It must also be asked why names in IE -an-(i)os should come out as Raet.°anu rather that °an(i)e.The forms concerned do certainly look like patronyms.
One might consider the possibility that the patronymic suffixes were still productive in Raetic in their original lexical genitival function and could as such be used to form individual names.The best cadidate for a lexical form in -na is the well-attested terisna, though its pan-Tyrsenian comparanda Etr.t/ zêrśna (AC a4;Rix 2000: 13), *zeris-na 'belonging to all = public' (Rix 1998: 48 [n. 2]), Lemn.*zariz-na 'for all' (Eichner p.c.) and etymology are still under debate (Salomon 2017: 253-255).Schumacher (p.c.) tentatively suggests an analysis of vaþanu from Celtic *ats-'servant' (see n. 9), that is '[he] of a servant', possibly as the name of a freedman.If the suffixes were lexically productive, this would yet again point to productively derived patronyms.
If, however, we concede that at least some of the single names ending in °nu and °na must be surnames used on their own, this would serve as evidence for them being inherited nomina rather than productively formed patronyms.The demotion of the praenomen in favour of the nomen is a phenomenon typical for the Central Italian family name system, and would not be expected in the context of a productive patronymic system (Untermann 1961 I: 41) -a person may identify themselves only by their family name, but would hardly be expected to give their father's instead of their own individual name.Again, there is no chronological pattern to indicate that the names in °nu/°na belong to a late phase of inherited nomina in contrast to earlier real patronyms which could not stand on their own.
A final remark is in order concerning the question of metronyms.According to Untermann (1959: 143 [n. 43]), metronyms are absent in inscriptions from the area of Brescia, though a few examples can be cited from the Milano area (Untermann 1960: 300 [n. 99]).The Raetic inscriptions, as far as can be judged from evidence which is inconclusive even about basic gender-marking, give no indications that metronyms were used.All surnames in -na are derived from individual names which are not otherwise attested; three have i before the suffix, which was above judged to be ambiguous as a gender marker, but any or all of the underlying names may as well end in °ie.aruse is the only individual name which occurs with a patronym in -na θarna and also as base for a form in -na aruśna°; the latter would thus -unless it is an individual name in °na as discussed above -be a potential metronym.aruse, however, is already suspect for being a presumably masculine individual name in -e combined with a presumably feminine surname in -na (see section 3) -the whole group is problematic and best not used to support arguments for now.

Raetic surnames in Roman inscriptions
On the basis of the evidence from Roman inscriptions, Untermann (1959: 81f., 91;also 1995: 735f.)In pre-Roman Celtic inscriptions from Northern Italy, synthetic surnames with patronymic suffixes like -al(o)-or -kno-/-gno-correspond to the genitival patronyms in Roman inscriptions and represent their vernacular models (Untermann 1959: 87-89;1960: 308;1995: 736f.).The integration of Venetic surnames -here mostly with the vernacular suffix -o--as nomina into the Roman system is also documented in the Latino-Venetic inscriptions of the 1 st century, e.g., Es VII UANTI As seen in these examples, many of the names include the genitival formula, Roman-style, to give the father's name in addition to the nomen, which indicates that the surnames are not Venetic patronyms which were ad-hoc transformed into nomina.Untermann concludes that, at the time of Romanisation, Venetic surnames were inherited nomina which could be smoothly transferred into the Roman system (also 1961 I: 39-41), arguing that this was probably aided by the comparatively close relationship between the languages and particularly the similar -o-suffix, but that it would not have been possible unless Venetic surnames had been inherited nomina already.Celtic and Raetic surnames, on the other hand, he considers to be patronyms, which had to be translated with the analytic genitival formula (1959: 89-91).
A look at Untermann's map in fig.7, however, shows quite clearly that the area of Raetic inscription finds does not at all coincide with the sphere of genitival patronyms in Roman inscriptions -on the contrary, it corresponds to the gap in the attestations of the genitival name formula in the Alpine area, much as the Venetic area does in the Alpine foreland and the plain.The reason for Untermann's association of the Brescia namescape with the Raetic as well as the Camunic vernacular is due to the fact that, at the time of his investigation, the Raetic and Camunic corpora, then both undeciphered, were often suspected to encode related languages.The mere fact that Camunic remains obscure even after the successful decoding of the Raetic inscriptions strongly suggests that this is not the case.Of course, the many connections between the onomastic material in Raetic inscriptions and Roman inscriptions from the Brescia area remain valid and must be considered, but some of them may be due to positive reinforcement, and particularly the shared onomastic bases can be explained as loans and do not necessarily indicate close relations between the vernaculars.Furthermore, any potential Camunic layer in the Brescia area must be expected to be admixed with a Celtic (Cenomanian Gaulish) layer, which may be responsible to a certain extent for the frequency of the patronymic formula (cf. the Celtic name brigovicus in the example from Erbusco above) -in fact, at the current state of documentation, the main cluster of the Brescia area coincides more with the Gaulish inscriptions of the region that the Camunic ones.
Untermann was aware of the issue and, as observed by Schumacher (1998: 99f.), was not entirely certain about the status of the Raetic surnames.Another reason for Untermann's inclination to subsume Raetic and Camunic under the Brescia namescape may be the lopsidedness of the evidence.Even in the 1950s, the structure of the Raetic name formula was much better understood than any aspect of the Camunic inscriptions, which could not even be reliably read and from which little to no onomastic material could be extracted.Roman inscriptions which yield useful material, in contrast, are thick on the ground in the Brescia area, while the Raetic core area, South Tyrol, is deficient in that regard.As indicated by the map, the patronymic formula is good as absent in Roman inscriptions where they are available from areas associated with Raetic by inscription finds (the pagus Arusnatium, Trento, the Val di Non, and the handful of inscriptions from the rest of the Trentino and South Tyrol), while attested nomina fail to show any signs of being vernacular.Raetic onomastic material in Roman inscriptions is restricted to the south-west -there, we find Raetic names both in the patronymic formula and transferred into nomina.
Trento, which should be expected from its situation to be a centre of Raetic culture, and indeed it is called raeticum oppidum by Pliny (III 130), has not so far yielded a single vernacular Raetic inscription.The only recognisable Raetic material we have from Trento occurs in the Roman inscription CIL V 5033 SASSIUS AD), which is also the only Roman inscription from Trento which features the genitival formula for the surname.The inscription contains two name formulae which consist of an individual name and the father's name in the genitive: sassius remi filius and turis barbarutae filia.Sassius' father bears a name well known from Raetic inscriptions: remi(e) → Latinised *remius. 71The analytic genitival patronym appears to translate the synthetic patronym of a vernacular Raetic name formula: remi f. ~ *reminu 'son of Remi'.This example indicates that Raetic surnames at the time of Romanisation were still functionally equivalent to Celtic patronyms, but it is isolated.
In opposition to the Trento inscription, we have evidence for Raetic surna-mes being integrated as nomina in four Roman inscriptions (Untermann 1959: 86, 134f.) from the Val di Non, a basin north of Trento on the western side of the Adige that was an epicentre of Raetic literacy, and its immediate environs.We can identify the vernacular surnames by the non-Latin n-suffix which represents the remains of the Raetic suffix -nu/-na.In two inscriptions, the bases are, again, recognisably Raetic.The nomen CIL V 5068 LUMENNONES (Romeno; 1 st -2 nd c.AD) is a Latinised Raetic patronym *lumen-nu from the individual name lumene attested twice in the Vinschgau.In CIL V 5023 L•LAUISNO | PATERNUS (Roverè della Luna; 1 st -2 nd c.AD), lauisno is *lavis-nu based on the well attested individual name lavise.Two nomina ending in -no in the damaged CIL V 5067 ]OSICCINO, ]ATINO (Cles; 103 AD) are too fragmentary to allow comparison with Raetic material, but ]osiccino may be structurally compared to patronyms from names with a k-suffix like φutiχinu, valθikinu (see n. 45).CIL Pais 715 Q•TENAGINO | MAXIMUS (Cles; 2 nd -3 rd c.AD) contains a similar form tenagino.An individual name tenaχi°/tenaki° is not attested in Raetic inscriptions, but the nomen is also likely to be the Latinised rendition of a Raetic patronym *tenagi-nu (Untermann 1961 I: 92f.assumes another k-suffix).A member of the family may be identified in Tenagino Probus, governor of Egypt in the 3 rd century AD (Schumacher 2004: 313 [n. 195]).The same nomen, or nomina based on the same Raetic patronym, may also be attested in CIL Pais 695 TENAINUS (Arco) and CIL V 3345 TENIGENONIA (Verona) (Untermann 1959: 86 [n. 15]).The latter inscription is the only potential piece of evidence for Raetic names from Verona, whose environs have yielded seventeen vernacular Raetic as well as eleven Celtic inscriptions.The settlement is called "oppidum raetorum et euganeorum" by Pliny (III 130) and associated with the Celtic Cemomani by Livy (V 35, 1), and appears indeed to have been something of a multicultural centre which partakes in various onomastic traditions (Untermann 1960: 309).
Nomina formed with an n-suffix are also attested in inscriptions from west of the Adige, i.e. from within Brescia namescape itself (Untermann 1959: 134f.), but no Raetic comparanda are available for any of the names.The nomen in CIL V 4014 L•SAMMUC[I]|NO•IUSTUS (Peschiera di Garda; 1-50 AD) is judged by Untermann (1959: 135 [n. 18]) to belong with the Milano namescape; *drūtos in CIL V 4204 L•TRUTINO | PROBUS (Brescia) is in fact Celtic. 72t is evident that a shift to family names happened somewhere between the Steinberg petrographs inscribed by Kastrie and his sons, and the more than fifteen Lumennones of Cles, whose nomen (unless they were all brothers) was evidently not derived from the individuals' fathers' names anymore (Schumacher 1998: 100).Yet the evidence is slim, and none of the name formulae in the abovementioned inscriptions features an additional patronym to demonstrate that the nomen was older than the person named.The Lumennones must have inherited their nomen at least from the previous generation, but, generally speaking, the transfer of patronyms to nomina may only have happened with the names' integration into the Roman system.
The fact that the evidence for Raetic patronyms as nomina is focussed in the Val di Non is probably significant.The Tabula Clesiana (CIL V 5050, Cles, 46 AD; Untermann 1959: 86f.) records the Emperor Claudius' decision to grant Roman citizenship and therefore the right to bear Roman names to three tribes, the Anauni, Tulliasses and Sinduni, owing to their close association with the Tridentini of Trento.These tribes were most likely Raetic -the Anauni can be connected with the Val di Non (Lat.anaunia), the other two tribes may be expected to have settled in the vicinity. 73According to the Tabula, the official decree came after the fact, as the tribes' members had long behaved like citizens already, taking office and serving in the Praetorian Guard -the Raetians of the Val di Non appear to have been quite enthusiastic about the Roman lifestyle.The treatment of vernacular names in this area may be due to conscious efforts to create Roman citizen's names, and may not reliably reflect Raetic conditions.The single attestation of the patronymic formula at Trento, as the irregular case, and also possibly the earliest of the discussed inscriptions, probably bears more significance than the four inscriptions from the Val di Non.
In summary, Untermann's conclusion that Raetic surnames in -nu and -na were productive patronyms at the time of Romanisation holds up to a re-investigation on the basis of our current understanding of Raetic onomastics, even if the question concerning the single names in °nu/°na remains open.Whether the productivity of these patronyms is reflected in the prominence of the Roman patronymic formula in the Brescia namescape is doubtful, but it might be assumed that another such "Namengebiet" to fit between Brescia and Styria/Carinthia would have provided evidence for the Raetic name system, if Raetic names had made it into the Roman documents of the area.The situation in the Raetic north is at this point impossible to evaluate based on the few Roman inscriptions from South Tyrol and the completely barren provincial North Tyrol.Why so little, if any Raetic names are attested in Roman inscriptions from the Raetic south outside the Val di Non -particularly Trento, the pagus Arusnatium and

Fig. 2 :
Fig. 2: Inscription IT-2 χaisurus on a fragmentary Fritzens bowl from the Himmelreich (North Tyrol).The individual name in the genitive -s can be interpreted as an owner's or donor's inscription.Drawing by Gudrun Bajc ©TIR.

Fig. 3 :
Fig. 3: Inscription NO-11 pirikaniśnu on a votive bronze from Dercolo (Val di Non).The name formula in the nominative with individual name piri and surname kaniśnu refers to the donor, or maybe to the owner of the hoard in which the object was found.Drawing by Gudrun Bajc ©TIR.
name formulae with surnames in -nu and -na.Uncertain attestations are marked with an asterisk.
Raetic name formulae with surnames in °i.
identifies areas in Central Europe where vernacular second names are rendered by what he calls the genitival formula, i.e. the Roman patronymic formula which consists of the father's name in the genitive plus filius/filia, while elsewhere these names are turned into Roman nomina.The genitival formula is used between and south of Lago Maggiore and Lago di Como (the Milano namescape), between Lago d'Iseo and Lago di Garda (the Brescia namescape), in the Eastern Alps between the valleys of Sava and Mur, and west of Budapest.

Fig. 7 :
Fig. 7: North Italic namescapes in which the analytic patronymic formula predominates in the Roman epigraphic evidence as established by Untermann (map from Untermann 1959: 92, Karte 1) and their association with vernacular pre-Roman corpora as shown in fig. 1.
Examples for the patronymic formula being used to render Celtic synthetic patronyms in the Milano area include CIL V 4924 BITIO•CARIASSI•F (Zenano), CIL V 6092 TERTIAE | CATTONIS•F (Milano), and CIL V 4710 RUFUS | BRIGOVICIS•F (Erbusco); a particularly clear example can be seen in the Todi bilingua (PG•1), where Cisalpine Gaulish trutikni (gen.)'of the son of Drutos' is rendered as DRUT(E)I•F in the Latin part.The transfer of vernacular surnames into Roman nomina, on the other hand, is associated with the Venetic area, where the genitival formula in Roman inscriptions is almost absent to render surnames.Examples for vernacular Venetic surnames, recognisable by the typically Venetic morphology with the ko-suffix, include CIL V 2327 SECUNDA•CAMMICA•SIPIONIS•FILIA (Adria) and CIL V 2035 SEX•PAETICUS•Q•F | TERTIUS (Castellavazzo).
. On the suffix of vaþanu see section 4.