Translated into English by Iulia Waniek
Motto: O, multum ante alias infelix littera theta (Isidore of Seville) "Thou letter theta, much more ill-fated than all others! "
The TDM Palatal
Kretschmer (Einl. 231) seems to be the first to have noticed the existence of a thracian consonant rendered in Greek alternately by [θ] and other letters and groups of letters, sometimes unexpected and seemingly incompatible. Kretschmer called this consonant, which he believed to be a spirant, 'the irrational spirant'. The term was taken up by Mihailov, who, in his La Langue (p. 68 et passim) in the paragraph called La 'spirante irrationelle' thrace et l'alternance de θ ~ τ – asserts: Dans quelques noms propres thraces apparaît une alternance de θ, θι, τι(t), σ et ζ. He illustrates this alternance with some examples, pertinent but few (see below), without examining the phenomenon.
Here are Mihailov's examples:
-an attribute of Heros, probably of toponomastic origin, in the sanctuary of Lozen (former Diniklij, Harmanli), with the variants: Γεικεθιηνος, Γεικεσηνος, Γικεντιηνος, Γινκατιηνος and Γινκισηνος, Γεικαι[.... Mihailov also adds here Γεσιηνος ( [Ἥρω]ι Γεσιηνῳ IGB III 1, 1497, from Borec (former Salalij, Plovdiv), a names that is difficult to read). The alternating terms are θι/σ/τι/
-an attribute of Zeus, encountered in several places in Thracia, with the variants: Ζβελθιουρδος, Ζβελθουρδος, Ζβελσουρδος, Ζβερθουρδος, [Z]beltiu[rd]us, Zp[ert]urd[us] or Zḅ[el][t]iurd[us], *Svelsurdus (= Iovis Velsuri in Cicero (Pis. 35,85) emendated to Iovis *<S>velsur<d>i). The alternating terms are: θι/σ/ti
-the names of the Danthaletes tribe, of its members and their strategy, with the variants: Dentheleti, Δενθελῆται, Δενθηλήτοι, Δανθηλῆται, Δανθαλῆται, Δανθηλητικὴ but Denseleti, Denselatae, Densela, Dansala. The alternating terms are: θ/s
-the attestations of Βρου-θενις and Βρ]ου-ζενις, for which, however, there is no clue that they were variants of the same name. The alternating terms would be: θ/ζ.
The Bulgarian scholar's conclusion is (p. 68): Il est évident que dans tous ces cas le θ a une certaine prononciation spirante. Cette hésitation dans l'ortographe indique que le son thrace, la 'spirante irationelle' comme l'appelle Kretschmer Einl.201 ne pouvait être transcrit avec précision par aucune lettre grecque. Despite this assertion I do not believe the spirant pronounciation is obvious at all. Though it is possible (and in that case the consonant concerned would have been pronounced [þ] as in English), there are clues that plead to the existence of (at least in some attestations) of either an occlusive (close maybe to/č/) or an affricate (of the [ţ]=/ts/ type in Romanian). The linguistic facts uncovered by the two linguists and their explanations are essentially justified but they need many additions and explanations up-to-date.
I would make a special remark upon the correctness of Mihailov's procedure in choosing for illustration variants of the same name, copiously attested. Future thracological studies should comply to this rule which ensures that 1.the variants belong to the same names and 2. they belong to the same space and time, very important requirings for the analysis.
What sound(s), precisely, could be meant in these examples ? Mihailov (after Schwyzer, Gr. p. 204 sq.) believes them to be spirants. He shows that the old aspirates - which had the values p+h, t+h, k+h (thus ph, th, kh)  in classical Greek - became spirants, dans des localités diverses à diverses époques (not too far apart though). We could sum up this process by the notation /ph/, /kh/, /th/ > /f/, /h/ (both as in Romanian or English), /þ/ (as in English [th] in thin, thick, thorn). This is the essence of the process of aspirates turning into spirants, occurring both in Greek and the Germanic languages. On page 65 Mihailov gives as a reliable example of this process the variants of the Greek name Βασυλλος / Βαθυλλος, where indeed the alternance of θ/σ has every chance to record a /þ/. No one should have any reason to doubt the Greek spirantization, as long as modern Greek no longer has aspirates but only spirants (the letters φ,χ, and θ are not read as aspirates [p+h, k+h and t+h] but as spirants [/f/, /h/ and /þ/]. However, here we are not interested in the situation of these sounds in Greek but in Thracian, and in Thracian there seems to have existed a quite different situation.
TDM languages did not have aspirates. In TDM languages all arguments plead for the lack of aspirates, which also makes improbable the existence of spirants too (at least of those coming from aspirates). The first and most convincing argument for such a reality is the insignificant number of TDM names rendered with Greek or Latin aspirate letters: φ/f and χ/ch. These are found in less than 20 roots overall, all of them at the periphery of TDM regions and due to foreign influences most likely: scytho-sarmatian in the north and greek-asiatic in the south.
In Decev's TSR these are: Φάγρης (Pieria; greek), -phara (in Breierophara, placename in Rhodope; copying error) as variant of -para; Φάρνουτις (egyptian?), Pharnacis river in Bithynia (surely iranic), -φιγοι in Πιέφιγοι (Ptol. in Dacia, but probably a copying error, cf. Pieporus), Φίττακος and Φίτταλος (more likely Greek phonetic variants of Πίττακος and Πίτταλος), Φορύννα and Iamphoryna, Τρίφουλον, Φύλλις river in Bithynia; Χάβρις river in Chalkidike (surely an iranic or asianic variant of thr. Kebros), -χαιτης (in Δρομιχαιτης), same as Greek -χαιτης in personal names/anthroponims (as Δρομι- too), Χαρ(ι)- in some compounds (Χαριμορτος is Greek [of the series Χαρι-λέως, Χαρι-μένης, Χαρί-ξενος aso.], Χαρκοτυς, Χαρναβῶν Dacian king with a name of sarmatic resonance), Χερδούσκερα and Χεσδούπαρα non-thracian or hybrid toponyms in Procope, Βαραχτεσ<τες> is more likely a celtic Varactes, Χρῶπες and Κιγχρῶπες [=gr.κέγχρος 'millet', the tribe being located in the area where Xenofon places the melinofagi Thracians (Anab. 7.5.12): ...διὰ τῶν Μελινοφάγων καλουμένων Θρᾳκῶν εἰς τὸν Σαλμυδησσόν through the country of the so-called 'millet eater' thracians, up to Salmydessos]; corrupted plant names (Diosk.) φιθοφθέθελα and χόδελα. The above names are too untipical and too scarce (as compared to the over 3000 TDM names) to sustain the existence of a Thracian aspirate. Most likely they are a proof of the absence of the aspirate series.
To the same conclusion lead, as it was pointed out by phonetic research in these languages, the Thracian adaptations of aspirated Greek names, in which the aspirates are replaced with the coresponding unvoiced or voiced nonaspirates. For example:
-επιπιος[epipios] (inscr. Glava Panega) instead of Gk. ἐφίππιος[ephippios] "on horseback, riding" (Decev TSR 166); Αὐλαρκηνος[aularkēnos] for αὐλαρχηνός [aularkhēnos], Καλλίμορπος [kallimorpos] for Καλλίμορφος [kallimorphos], τυγάτηρ [tygatēr] for θυγάτηρ[thygatēr], Βιλίστα [bilista] (macedonian form) for Φιλίστα[philista], Δρεπτῷ[dreptō] for Θρεπτῷ [threptō, maybe pronounced /dhreptō/], all found in Mihailov La Langue 64-65, or the name Diopanes, considered to be the variant of Diuppaneus [=Diurpuaneus] (v. TSR sub voce), while I think it is the phonetic adaptation of the Greek name Διοφάνης[diophanēs], Pulpu (in Pulpudeva) for Philippos and others. Mihailov tends to consider them vulgar Greek pronounciations, showing that there are, moreover, variants with an aspirate instead of a non-aspirate. These are, however, extremely rare and seem to be rather hypercorrect forms: not pronouncing the aspirates, Thracian scribes must have been unsure whether to use aspirates in some Greek words or not, therefore putting sometimes aspirates randomly. But even if such pronunciations are indeed vulgarisms the Greek vulgar language must have picked them from somewhere, and the most plausible source is in the first place the Thracian substratum itself. If the way Mihailov explains this situation is somewhat questionable, the small percentage of names with aspirates is on the other hand a solid proof of the absence of this consonant series in TDM languages.
The lack of this series confirms the placement of TDM languages within the satәm group, in their historical period, no matter what their character may have been in a more distant period. In the languages of this group aspirates have dissapeared, being replaced by the corresponding unvoiced or voiced non-aspirates.
As you noticed, the preceeding observations do not include the aspirate [θ] (theta). This letter, unlike the other two, appears in a large number of attestations of Thracian names (almost 200 base-words), of which here are the most important:
Ἀβλουθιης, Αθιουτικη, Ἀθρυΐλατος, Ἄθρυς, *Αθυπαρα(<Αθυπαρηνος), Ἀθύρας, Ἄθυς (Ἄθης), Ἄθως + var, Αἰθαζιης, -αλθης, †Ἀνθιηνός (Greek), †Αὐθιπάρου [=Σαυθιπάρου], Αψίνθιοι, Βασκιδίθιας, Βειθυς / Βιθυς (+ compounds and derivatives), Βιθύαι, Βιθύας, Βιθυνοί (+compounds and derivatives ), Βισάνθη, βόλινθος, Βορκηιθιας, βουδαθλα (Dacian plant), Βουρθειθης, Βρουθένης, Βρωλυθρισβεις?, -βυθος, -γεθης, Γειθυκειλας, Γεικαιθιηνος + variants, -γενθης, Δανθαληται (+var.), Δενθάλιοι, -δενθης, Δενθις, Zaecethures, Ζαρμιζεγέθουσα + var , Ζβελθιουρδος, Δορζενθης, Ζηλικίνθιος, Ζηρινθια, ζιβυθίδες, Ζortha, Ζουθις, Ithazis, Ithiostla, †Ιθουσχις (very uncertain reading), Ἰσάνθης, Caeletharidas, Calathus, καροπίθλα (Dacian plant), - κενθος, Κενθος, Κοθων, Κουθειν-Κουθιας, Κουθιουρας-Κουθιουλας, Κουνθισίης, -κ(ο)υθης, Μαθανός, μαριθάς, Μένθης, Μέσθλης, Μοκκοθας, -nithus, Ξανθεια-Ξανθ(ε)ιοι (grec), Οἰρίνθης, Ὄλυνθος, Uthis, Πέρινθος, Πύθια, ῥαθίβιδα, Ρησκύνθιον - Ρησκυνθίς, †Ροθος, Σαικιθης, Σα[υ]θρης, Σεύθης +var, Σθορυς, Sithalcus, Σιθωνες, Σινθος, Σκίθαι, Σκυθοδοκος, Sothimus, Σουθιος, Σπάθιζος, Sparthon, Σπινθοπυρηνος, -thalcas, -thari, Ταρουθινας, Τηρετιθος, -titha- Τιθα + var, Τιλθαζεις, Τιμαθοχιωμ, φιθοφθέθελα (most certainly corrupted form of a Dacian plant name). Almost all of them can be found in Decev's repertory.
The large number of occurences of [θ] cannot be explained, as Mihailov proposes, by the fact that it might have transcribed the spirantization of /th/ to /þ/ in Greek, because:
-as we have seen, Thracian did not have aspirates at all, so it couldn't turn them into spirants.
-why would there be such an imbalance between the frequence of the spirant /þ/ (that would be transcribed by /θ/, with many attested occurences) and the other two: /f/ and /h/, practically absent ?
The only reasonable explanation of these facts is that, in the names listed above, the letter [θ] was used to transcribe a sound different from an aspirated or spirant dental. This sound is evidenced by the multiple alternances that the above mentioned scholars spoke about. Because the main element of this sound is [θ], it seems appropriate to call it shortly the [θ] alternance (or [θ] alternant, when we refer to any one of its members). It is more extended than Mihailov had presented it to be, its members being θ, θι (± vowel), τι (± vowel), σ, σι, ζ, ζι(± vowel), τσ, τζ. It is obvious that this alternance reveals, in most cases, the effort of the scribes to render an autochtonous consonant for which neither Greek nor Latin had an appropriate letter.
What sound would this have been? The solutions found by the ancient scribes to transcribe it offer us two or three clues that enable us to put up a working hypothesis:
-1. the alternance with [s] and [z] seems to show that the sound was fricative (or spirant; this means it was made not by a complete closure of the fonic apparatus but by narrowing it, as in the sounds s, f, h) or that it had a fricative component;
-2. the graphic renderings [ts], [tz] seem to indicate a palatal occlusive (or affricate) (that is of the type of /č/ or /tz/ in Romanian).
-3. the graphic renderings by an –[i]- (more often followed by a vowel), especially in the groups [θι] and [τι] before a vowel indicates the existence of a palatal element.
So, although we have no clear elements to allow us a precise identification of this Thracian consonant, graphically rendered by Greeks and Romans as I have shown above, yet clues that significantly restrict the number of possible candidates for identification do exist. They allow us to assume that it was a spirant or an occlusive with palatal character, and consequently I propose to call it Thracian palatal. Let's give it a definition:
Thracian palatal. The Thracian palatal is a consonant with uncertain articulation, spirant or occlusive, most probably of a palatal character, specific to the Thracian languages (possibly only to some of them). Its presence is indicated by the fact that in ancient sources it is written alternatively by θ, θι (± vowel), τι (± vowel), σ, σι, ζ, ζι(± vowel), τσ, τζ. For methodological accuracy it will be transcribed phonetically by [ċ], and its voiced pair by [ġ]. It is not at all unlikely that it had voiceless and voiced articulations (voiceless /þ/ or /ts/, voiced /đ/ or /dz/), possibly in special dialectal variations.
Identification of 'Thracian palatal' in the sources. Doubtlessly, the most important question we ask is how do we identify the Thracian palatal in the source texts? Firstly, as we have seen, the mere presence of [θ] in a name can indicate this consonant, since Thracian did not have the Greek sound transcribed by [θ]. It is also almost surely indicated by the renderings [tz] and [ts], and quite frequently by the group τι + vowel. As for the other letters, σ, σι, ζ, ζι(± vowel), for which there are no means of proving whether they transcribe or not our palatal, the fact that we should permanently keep in mind that they could have this value is a gain by itself.
Other occurences of the Thracian palatal :
-Θρᾴκη/Ζραικη. The name of Thracia (Θρᾴκη- Thracia) and of the Thracians (Θρᾷκες=Thraces, Thraci) is Greek. Its oldest form, with a diphtong on a long degree /āy/ (which became in Ionian dialect, ēy), was Θρᾱικᾱ (Ion.Θρηικη, as we find everywhere in Herodotos). By monophtongation of the diphtong it then became Θρᾴκη, with the old iota of the diphtong remaining as iota subscriptum. But where did the Greeks take the old form from? Among the relics of Thracian languages is the name of the strategy Ζραικη (which is doubtlessly a grecization with the desinence -η, of an autochtonous Zraykā). It appears in the famous inscription of Flavius Dizalas, son of Ezbenis (IGB 4.2338, from Nicopolis ad Nestum, former Gărmen, from the end of 1st c. AD), who was, as he himself states,
στρατηγὸς Ὀλυνθίας καὶ Ῥοιμηλη-
τικῆς καὶ Δρησαπα̣ϊκῆς καὶ Θ̣ουκυσιδαν-
 τικῆς καὶ ....σηλητικῆς καὶ Ζραικῆς καὶ
Αθιουτικῆς καὶ Βιολητικῆς...
(3) strateg of Olynthia and of Roimele-
tica and of Dresapaica and of Thukysidan-
tica and of...seletica and of Zraika and of
Aċiutica and of Bioletica...
Ζραικη [Thracian Zraykā] is thus the name of an administrative region (strategy) that could easily be the autochtonous (Thracian) variant of the Greek name Θραικη [Thraikē]. In the light of what we said before, the θ/ζ alternance, which appears at the initial letter of these two variants of a supposedly single name, should not surprise us at all. The [Z] in Ζραικη shows us that we deal here with a voiced sound and therefore, applying the conventions established previously, we shall transcribe this name by /ġrayk(o)/. At Indo-european level this is obviously an adjectival form, created with a suffix+thematic vowel –iko- (feminine –ika). There is a strong possibility to be equally the etymon of the Greek form Θραικη > Θρᾴκη (region name) and Θραικες > Θρᾶκες (tribe members' name), and of the Thracian form transcribed as Zραικη in the Flavius Dizalas inscription. Even more interesting is the fact that, if we admit, as we proposed above, that the TDM pair [ċ] şi [ġ] come from (originate from) the Indo-european *k' şi *g', then the root /ġrayk(o)/ is the semi-satәm form (if we could say so) of a radical *g'rayk(o). The centum form of the same name is Graykoi with the suffix -k (hence the Latin Graeci) and Grayoi without a suffix (Lat. Grai, Gk. Γραῖοι). This is another proof of the remote kinship of Greeks and Thracians.
-Αθιουτικη/Ασουτικη. Two of the surviving attestations of strategy names seem to be also variants of one and the same name. The first variant, Αθιουτικη, appears in the same Flavius Dizalas inscription (see above). The form Ασουτικη (in IGB III,1, n.1116, found in the Batkun Asclepieum, from Trajan's time) refers almost certainly to the same strategy:
...ηνο[ς] στρ̣ατηγὸς Σ......
...Dizalas of Kotys
...strategos of S(eleticei?)...
and of Asuticei...
The strategos mentioned here, also called Dizalas, is, however a different person. The first one is the son of Ezbenis, the second, of Kotys. They are both strategoi and both living in Thrace around the same time. We ignore the number of strategies existing then in Thracia (Ptolemy enumerates about 15-16, but Plinius, half a century earlier, stated there were 50), but, whatever their number, it is unlikely there were two with such similar names and practically in the same place. It is more natural to assume that both were variants of the same name, which we could transcribe as /aċutika/, a strategy name derived with the same suffix –ik(o) from an */aċut(a)/. See also Mihailov IGB III 1,p. 120: Strategia Ασουτικη eadem atque Αθιουτικη in tit. Nicopoli ad Nestum reperto V. Beševliev SpBAN 70, 1945, 201—211, praesertim 204, vide vol. 4; de θ ~ σ in lingua Thracica vide G. Μihailοv La langue 67 sq., cf. item D. Dečev Charakteristik der thr. Sprache, Sofia 1952, 15 (75). Assuming the same evolution i-e.*k' > Thrac. *ċ, we can propose as etymology for this word the i-e. root *ak'- "sharp", a very suited name for a mountainous county such as that where the two inscriptions were found. Moreover, this etymology fits equally well the oronym Ἄθως (+ var). Documentary sources do not preserve any alternating variant for this name (such as could have been*Ασος, *Αζος, *Ατιος, *Ατζος a. s. o.) which makes me believe that - just as for the name of Thrace itself - Athos is the generalized Greek version of the name.
-θειθης. Mihailov (see above) quotes as occurrence of the 'irrational spirant' the name Βουρθειθης (TSR 81, IGB II, 744 at Razgrad(Abrittus)), although here only the second component, -θειθης is interesting. This one makes an alternance with *-tsitsis, reconstructed from the derivated ending -tsitsinis of the name Burtsitsinis (TSR 83, at Histria (ISM I, 332). Relying on these variants we can reconstruct quite accurately a formant -ċīċis (with a long first vowel, indicated by the alternance Gk./ει/=Lat./i/). Other final formants could also be alternating forms of this one, as for example ζεζες, -θιθ(θ)α, Tzitzis, Tzittas(?) a.o. Despite this closeness in form, I don't know if we can identify this /ċīċis/ with the formant and the name -titha, -τιθος, Τιθα, Θιθθα, which should be transcribed /tiċis/, meaning "breast, tit" (meaning proved by the attribute of Diana Germetitha which doubtlessly meant "(the one) with warm bosom").
-Θιντας, attested twice in Sofia (IGB IV, 1963 and 2004), appears about five times in Latin inscriptions of dacianizing form as Tzinta, Tzinto and Tsinta. We have here a typical alternance, θ/tz(ts), therefore the initial consonant surely must be the Thracian palatal [ċ]. The etymology should be an I-e. *k'int/*k'ent
-*Kutsus. A name with a few dozens of attestations, in many variants, is Κουθιας , of whose forms I would mention here only Κοθος, Κοθιης, Κουθειν, Κουσους, Κουτζης, Κουτζίς, Κουτσης, Κοτ(τ)ιων, Cotia, Cutius, -κουζος, - κουθης , -κυθης, Κουζαιος etc. One of its derivatives, Κουθιουλας, has itself a multiplicity of variants, as for example Cutiula, Κουτιλας, Γουδιλας, Κοθηλας, Gudila and others. This is a yet unstudied onomastic family, a very ramified one, whose components and variants (in which merged different names, sometimes foreign, like Gothic) appear in Decev's repertory as separate names. The alternance θ/θι /τι/σ.
-eithio-. In a series of attributes of Heros, all found in Moesia, there appears, either as a first or as a second formant in compounds, a root ithio/eithio with alternating variants. Here you are its certain or very probable attestations:
-ithi(-o,-a): Ithio-(s)-tla (TSR 215, at Karaissen, Svištov), Βορκη-ιθιας (TSR 75, at Gorna Orjahovica), Βασκιδ( ι) -ιθιας (TSR 43, at Paskalevec, Nicop. ad Istrum), Ebist-ithias (TSR 162, at Gagovo, Popovo), Ζουσυρ-ειθα (attribute of Herakles, in IGB I, 24 at Varna); -iti(-u) in Brigan-itius (TSR 87, at Agatapara); it is possible that the second part of the name Geika-ithienos, presented above, belongs here too.
-eithi -/eiti-/eis-: In the east of Moesia, over a relatively small area, around Šumen-Devnja (Marcianopolis) - Provadija - Varna(Odessus) - Pomorje(Anchialus) they are attested god-epithetes and personal names compound with Ειτι-/Ειθι-/Εισ-, which I take as variants (both phonetic and graphic) of ithio-: Ειτιο-σαρος (θεός) by Čauševo (Šumen), Ειθις Ειθιαλου (by Padina, Varna), Εισηνος (Apollon) by Bata (Anchialo), Εισα-τραλις (by Provadija and Devnja); we could add here also the ending –ιζις of the theonym Γεβελέ-ϊζιζ or of the antroponyms Τσυκολ-ειζις, Ζειβ-ιζιης, as well as a lot of other, much less sure. In those above, the alternating elements are θι/ θ / τι / σ.
-Zi(a)marcus and Thamarcus (IMS II 53 p. 34: M(arcus) Aur(elius) Thamarcus Rat(iaria), cf. TSR 537. IGBulg III,2 1787 Augusta Traiana (Stara Zagora ) τῆς γυναικὸς / Ζια̣μαρκης etc.
Echoes of this palatal sound in the Romanian and Albanian substratum. If things are really as pointed out so far, then the variants of this Thracian palatal could coincide with the remnants of Indo-european consonants k' and g' in the substratum words common to Romanian and Albanian. Indeed, in both these languages there are a few such words coming from Indo-european roots that contain a [k'] or [g'] (that is a palatal /k/ or /g/). The correspondences are: to an Albanian [th] (pronounced /þ/) or [dh] (pronounced /đ/) correspond a Romanian [ţ] or [dz] ( this latter one became [z] in the daco-romanian dialect).Examples:
1. for Alb. [th]= Rom. [ţ]: Alb. thumbullë=Rom. ţâmburuş (there is also Alb. sumbullë=Rom. sâmbure) 'kernel'; Alb. thark = Rom. ţarc 'enclosure for animals', Alb. thep(<thap)=Rom. ţep, ţeapă 'stake, spine, thorn'. In quite a few words the correspondence is also between Alb. [th]= Rom. [s], for example: Alb. kurthë = Rom. cursă 'trap', Alb. tharbët=Rom. sarbăd 'with no (or bad) taste', Alb. thumbullë = Rom. sâmbure 'kernel'.
2. for Alb. [dh]=Rom. [dz]: Alb. barth (fem. bardhë)=Rom. bardzŭ, bardză '(spotted) white' ; Alb. bredh (< v.Alb. bradh) = Rom. brad (singular form adapted from the plural bradzi, from an old sg. *bradzŭ) 'fir tree', Alb. modhë (and its better known derivative modhullë)=Rom. mazăre and madzire 'pea' , Alb. vjedhullë (="thief", from the verb vjeth "to steal") = Rom. viedzure (which means 'badger'; see how much alike is the English word, whose etymology is unclear); Alb. dhallë =Rom. dzară 'sour milk' and others.
Partial alternances. Partial alternances, considered sometimes as independent from the one we are now dealing with (the alternance of [θ]) should also be studied. Here are a few examples of alternances I found:
-Mihailov quotes a series of names that would display the τ/θ alternance (p.68): Ρεσκου- βιτου, Βιταν, Βιτος versus Βειθυς, Βιθυς, Δεντις / - δενθης, - κεντος - - κενθος, Μεντης / Μενθης, Σαμβατ ί ων - Σαμβαθ ί ων. There appear forms with -τ- instead of -θ-, says he then, in other words as well, like Θιθι-σαττα, Βουρ-θειθης, Θεθεις for which he sends us to Mateescu EDR I 89 sq. On page 69 he also mentions two transcriptions with [θ] for the Iranic /þ/: Μιθραδ ά της, Μιθριδ ά της < Miþradāta, Παρθικ ό ς < Parþa
-here belong also the few attestations of the name of Mesembria Pontica on some of the town's oldest medals.. On these, instead of the normal spelling ΜΕΣAΜΒΡΙΑΝΩΝ we find in a few cases the form ΜΕTAΜΒΡΙΑΝΩΝ, with a special sign, like a T with very long lateral apices, usually transcribed as T.
-a ζ / τ alternance is also mentioned by Mihailov (p. 80) and exemplified by him with:
-forms compounded with -τελμις considered as a variant of -ζελμις: Δωρουτ[ελμις and Ζουροτε[λ]μις, both in the same inscription (IGB II, 738, from Discoduraterae, Moes.Inf.):Διμη[ς Ε]πτεθισι ζ ή σας ἔ τη ἑ κατ ὸ ν κα ὶ γυν ὴ μου Δωρουτ[ελμις] κα ὶ υ ἱὸ ς Βιθυς, θυγ ά τηρ Ζουροτελμις
-the ζ / δ alternance discussed by Mihailov at pages 63-64, considered by him to be a late phonetic evolution of δ > ζ. Occurences:
-Δεβαβενζις (IGB IV, 2292 Laskarevo) and Βενζει (Βενζει θυγάτηρ Μοκαπόρεος IosPE II 223, at Pantikapaion cca. 100-150p) < The name of the goddess Βενδῖς
-names that start with Ζι-(Ζισυρας, Ζεισκωρις, Zipyrus, Ζειποιτης, Ζειτραλις) versus their correspondents beginning with Δι-(Δισυρου, Δεισορου, Διπυρος)
-isolated pairs like Αρσιληνος - Αρδιληνος (Αρδις )
A study of the entire Thracian linguistic material could add still more examples to those above, but I think that even these are sufficient to convince us that the Thracian language had a sound, Kretschmer's "irrational spirant", which, as Mihailov says "could not be transcribed exactly by any Greek letter" (nor Latin, for that matter). To write it, the ancients used various graphical conventions which evolved in time. At the beginning they rendered it by the consonants [θ] and [τ], less by [σ]. It was later considered that the groups [θι] / [τι] would be closer to the original sonority, especially when followed by a vowel, so more so that in spoken Latin the groups ci/ti+vowel had really come to be pronounced that way. Finally, during the Late Empire there appeared the graphical renderings [ts] , [tz] (considered until then barbarian).
1. In the more conservative gypsy dialects these aspirates can still be heard.
2. I have chosen this letter because, being more rarely used it does not lead to confusions and vbecause it suggests (as [č] for exam ple) a palatal sound.
[to be continued]