Nouns and personal names ending in –εύς –ῆϝος are unique to Greek,Footnote ¹ and have often been deemed pre-Hellenic in origin simply on account of the lack of Proto-Indo-European (PIE) correspondences. Our failure to find convincing etymologies for βασιλεύς, ἑρμηνεύς, and βραβεύς has itself contributed to this view.Footnote ² However, we should hesitate, for general reasons, to posit pre-Hellenic origins for these words, since viable explanations both of βασιλεύς and of ἑρμηνεύς (if not of βραβεύς) lie near to hand. Although the explanation of βασιλεύς that will be proposed below still presents difficulties, I believe that it improves on previous attempts.

As Schindler showed in a short but fundamental article, in Mycenaean Greek and in Homer the nouns in -εύς fall into two groups.Footnote ³ The first group is personal names such as Ἀτρεύς and Πηλεύς, which often appear to be of obscure etymology (and are therefore often deemed pre-Hellenic), but in fact are fully explicable as shortened forms, for reasons that will be given below. The second group is that of common nouns in –εύς, which are nearly all (especially in Homer) denominatives from ο-stem nouns denoting an occupation or profession: compare χαλκεύς ‘smith’ with χαλκός, κναϕεύς ‘fuller’ with κνάϕος ‘carding-comb’, ἱερεύς ‘priest’ with ἱερός ‘holy’, and κεραμεύς ‘potter’ with κέραμος ‘pot’, all of which are attested in Linear B and come from familiar Indo-European or Greek roots.Footnote ⁴

Schindler supplied a convincing explanation from within Greek for nouns in –εύς.Footnote ⁵ As he observed, the derivatives of ο-stems are statistically far more common in Homer than are derivatives from other roots. He concluded that these nouns are originally secondary formations made from ο-stems, comparing the Lithuanian suffix –ius, which is functionally as well as morphologically similar: compare banius ‘saddler’ with banas ‘saddle’, puõdžius ‘potter’ with púodas ‘pot’.Footnote ⁶ In addition, Hajnal has suggested that the Phrygian name of which the genitive appears as Proitavos in inscriptions M-01b and M-02 is derived from a name in *–eus –ēwos.Footnote ⁷

Since there is no *–i– in the endings of nouns in –εύς, Schindler deduces that the original thematic vowel –e– was retained throughout the paradigm. He reconstructs the paradigm as nominative *–e–u–s, genitive *–e–e––os, as in χαλκ–ε–ύς *χαλκ–ε–έ–ϝος, with contraction of *ĕ–ĕ to *ē.Footnote ⁸ The thematic vowel *–e– is followed by the forms of the suffix *–e– with its accent on the final or penultimate syllable (the ‘proterokinetic’ declension), so that χαλκε-ύς *χαλκε–έϝος follows the same pattern as ἡδ–ύς *ἡδ–έϝος. Thus, according to Schindler's theory, the nominative singular –εύς is the original form, rather than derived from *–ēus with the shortening of the long diphthong that is usual in Greek (‘Osthoff's Law’).Footnote ⁹ Schindler also suggests that the original accusative must be reconstructed as *χαλκε-έυμ, which would yield *χαλκήν, by the application of Stang's law, as seen in *De–m > *dēm > घῆν.Footnote ¹⁰ He suggests that this is preserved in the Arcadian (and perhaps Cypriot)Footnote ¹¹ forms such as βασιλήν and ἱερήν,Footnote ¹² but that in the other Greek dialects it has been remade into –ῆα by analogy with the other oblique cases; he holds, conversely, that Arcadian and Cypriot have generalized the accusative in –ήν into a declension with a nominative in –ής.Footnote ¹³ However, Stang's law had ceased to operate before the time of Proto-Greek, and it is easier to suppose that the accusative βασιλῆα reflects *g ^wasilēa contracted from *g ^wasilĕ–ĕ––ṃ. Unfortunately the accusative singular of nouns in –εύς is still not attested in Mycenaean Greek.Footnote ¹⁴

The term βαϲιλεύς first appears in the Linear B archives of c. 1350–1200 b.c. as , that is qa–si–re–u, which represents Mycenaean Greek /g ^wasileus/ as accurately as the conventions of that script permit. Scholars almost universally believe that its etymology is insoluble. Words for kings and rulers are often borrowed from one language to another, as in the case of πρύτανις and τύραννος, if not also ϝάναξ;Footnote ¹⁵ hence βαϲιλεύς is deemed to be another such term of non-Indo-European origin. Thus Chantraine remarks ‘il est vain de chercher une étymologie à βαϲιλεύς’, though he continues by noting that the initial labiovelar in Mycenaean qa–si–re–u ‘peut faire penser à une origine indo-européenne’.Footnote ¹⁶ Others ascribe it to the pre-Hellenic substrate population which we know to have been present in Greece before the arrival of the earliest Greeks, as does the author of the latest etymological dictionary of Greek, citing precisely the fact that it contains a labiovelar.Footnote ¹⁷ However, languages do sometimes use inherited words for ‘ruler’, as in the case of English ‘king’ and ‘leader’; in Homeric Greek, κοίρανος is agreed to have come from the Indo-European root *kor(i)o– ‘army’,Footnote ¹⁸ and Mycenaean ko–re–te ‘mayor’ has been linked to this same root.Footnote ¹⁹

In the light of Schindler's work on nouns in –εύς, an old etymology for βασιλεύς deriving it from two very familiar Greek roots at last begins to make more sense. In the 1840s Benfey and Curtius suggested that g ^wasileus is a compound of *g ^wṃti– and *lāwos, that is, ‘he who makes the host march’;Footnote ²⁰ they adduced the analogy of German Herzog ‘leader of the army’, ‘duke’, which is a compound of Heer ‘army’ (from Middle High German heri < PIE *koryos) and ziehen ‘pull’, ‘lead’. Another possible meaning of g ^wasileus would be ‘he who marches with the host’, since the element –λαος in names can refer to an oblique case, as, for instance, Πρωτεσί–λαος or Φιλόλαος.Footnote ²¹ But ever since 1916, when Boisacq mentioned this etymology sceptically, it appears simply to have been forgotten. It probably fell out of favour because the formation of hypocoristic nouns in Greek was not understood until relatively recently.Footnote ²²

The second half of the compound, –λεύς, is a shortened or hypocoristic form that appears to belong to the second class of nouns in –εύς, that is masculine personal names. The suffix –εύς was no longer productive of names in historical times, but many mythical and Mycenaean names are of this kind. Von Kamptz showed long ago that even apparently opaque forms such as Εὐρυσθεύς and Μενεσθεύς are shortened forms of intelligible Greek names such as Εὐρυσθένης and Μενεσθένης.Footnote ²³ Thus even as peculiar a form as Ἀ–τρεύς, which von Kamptz still considered pre-Hellenic,Footnote ²⁴ is in fact a shortened version of ἄ–τρεσ–τος or *Ἀ–τρέσ–της ‘he who does not run away’, from τρέω;Footnote ²⁵ compare Ἀγα–μέμνων and Μενέ–λαος, members of the very same mythical family, whose names likewise evoke standing one's ground in battle.Footnote ²⁶ Likewise Mycenaean /Alekseus/ is short for Ἀλέξανδρος or the like,Footnote ²⁷ while classical Ἀνδρεύς would be an abbreviated form of Ἀνδροκλέης and Νικεύς of Νικόδημος.Footnote ²⁸ However, such forms are often so truncated that the original long form of the second half of the compound is unrecognizable.

In the case of g ^wasileus, the second half of the compound is shortened from a word beginning with lambda.Footnote ²⁹ As such it is open to many possible interpretations, but the most obvious candidate is indeed λα(ϝ)ός ‘host’, ‘army’. It would be based on an unattested full form *g ^wasilāwos, with which we may compare names such as Ἀγησίλαος. The shortened form of λαϝός appears as –λεύς in one very famous name, Νηλεύς/Νειλεύς, ‘he who brings home the host’, a compound of *nese– ‘return’ (the root of νέομαι and νόστος) and λαϝός; the first half of this name recurs in that of Neleus' son Νέσ–τωρ. The full form appears in Mycenaean as Ne–e–ra–wo /Nehe–lāwos/.Footnote ³⁰ Another such name containing this suffix is Ἀχι–λεύς, which sometimes has a hypocoristic doubling of the liquid. As Palmer showed, Achilles' name means something like ‘grief to the host’; the same semantic elements appear in the hero's female altera ego Πενθεσίλεια and in the name Πένθιλος, both of which end with shortened forms of λαϝός.Footnote ³¹ A different kind of shortening of the same element appears in the Homeric pair Σθενέλαος and Σθένελος and the historical names Ὀνησίλαος and Ὀνήσιλος.Footnote ³² Finally, Ὀϊ-λεύς/ϝι-λεύς may mean ‘strength to the host’, from ἴς ‘strength’, the cognate of Latin vīs. βασιλεύς looks very like a personal name of this type, a puzzle to which we will return below.

Benfey derived the first part of g ^wasileús from the familiar root *g ^wem–, from which are derived Greek βαίνω ‘to step’ or ‘march’, Latin ueniō, Sanskrit gácchati, and English come (Gothic qiman).Footnote ³³ However, since Hermann Müller proved in 1879 that Proto-Indo-European had three laryngeals which have subsequently been lost, we can be more specific about the root of βαίνω. Until late in its prehistory Greek retained a series of three sounds called ‘laryngeals’, that is gutturals of different kinds; these survive directly only in Hittite, where they are often written as ḫ, and in its Anatolian relatives and descendants. In certain phonetic environments, laryngeals turned into vowels (mostly a, but i in Sanskrit).Footnote ³⁴ Greek is the only Indo-European language where the different laryngeals became different vowels when they fell between two consonants. Thus the so-called e-coloured laryngeal *h ₁ turned into e, as in θε–τός (*d ^hh ₁–tós) from τίθημι, the a-coloured laryngeal *h ₂ into a, as in στα–τός (*sth ₂–tós) from ἵστᾱμι, and the o-coloured laryngeal *h ₃ into o, as in δο–τός (*dh ₃–tós) from δίδωμι.

In the light of laryngeal theory, we now know that Greek uses two related roots to mean ‘go’. βαίνω derives from *g ^w(e)m–, but the other tenses derive from the root *g ^wh ₂–, with the ‘a-coloured’ laryngeal *h ₂, which in Greek became α in many phonetic environments. This root is seen in its other tenses βήσομαι, ἔβησα, βέβηκα, and ἔβην (the exact cognate of Sanskrit ágām ‘I went’), as well as in βίβημι (βίβᾱμι), with its exact Sanskrit cognate jígāti, and in words such as βῆ–μα (compare Avestan gā–man– ‘step’).Footnote ³⁵ It is difficult to determine whether, in the compounded form g ^wasileús, βασι– goes back to (i) the root *g ^wh ₂–, where the laryngeal *h ₂ has become ᾰ between two stops (compare στατός from *sth ₂–tós and ϕατός from *b ^hh ₂–tós), or to (ii) the zero-grade of the root *g ^wm–, where the syllabic nasal *ṃ has become ᾰ, as in βατός from *g ^wṃ–tós (compare Sanskrit –gatáḥ, Latin –ventus). Footnote ³⁶ The noun βάσις, derived from *g ^wṃ–tis, is formed from the latter root, since it is identical in formation with Sanskrit gátiḥ, Latin (con)uentiō, and Gothic (ga)-qumϸs; the nominative, originally *g ^wém–tis, has been reshaped by analogy with the oblique cases such as the genitive *g ^wṃ–téis.Footnote ³⁷

The analogy of the latter two forms suggests that βασι–λεύς has the same origin, but proof is beyond reach. However, on neither hypothesis did the root βασι– contain the thematic vowel e, since otherwise it would have to have taken the form βησι– (*g ^weh ₂–), as in the Homeric personal name Ἀνα–βησί–νεως ‘he who makes one board the ship’Footnote ³⁸ and the Homeric (Aeolic) noun βητάρμων ‘acrobat’,Footnote ³⁹ which originally meant ‘he who steps onto a wheel’, from *βητι–άρμων, a compound of Mycenaean a–mo /arhmo/ ‘wheel’.Footnote ⁴⁰ The assibilation of the dental τ before i in compounds of this type, such as τερψί–μβροτος and indeed βασιλεύς is standard in East Greek; the failure to assibilate in words such as βωτι–άνειρα, ἀρτι–επής and βητ(ι)–άρμων is typical of Homeric words which survive from the Aeolic phase of the epic tradition. Thus Ionian epic should have retained the Mycenaean preposition po–si /posi/ (compare Arcadian πός), but in fact uses the West Greek form π(ρ)οτί beside πρός; indeed πρός can be proved by statistics to have largely replaced ποτί, which must have been in proto-Aeolic also.Footnote ⁴¹ The same replacement apparently occurred in Lesbian, which borrowed πρός under Ionic influence: the Homeric formula προτὶ Ἴλιον ἱρήν, with Lesbian ἱρός, confirms this.Footnote ⁴² βωτιάνειρα has both –τι– and an Aeolic treatment of metrical lengthening.Footnote ⁴³

As Curtius noted,Footnote ⁴⁴ compounds such as τερψί–μβροτος often have transitive verbs in the first half, for example, Sophocles' ζευξίλεως ‘he who yokes the host’Footnote ⁴⁵ or Στησίχορος ‘he who establishes the chorus’. In the case of this root we may compare Ἀνα–βησί–νεως ‘he who makes one board ship’. Such formations are often related to sigmatic transitive aorists.Footnote ⁴⁶ By this analogy, we might perhaps have expected to find βησι– with the e-grade rather than the zero-grade of the root. Ηowever, we saw that βητάρμων has *βητι– in an intransitive sense, and that nouns in –εύς always have the accent on the suffix; hence the lack of the thematic vowel in the root of the first half of βασι–λεύς is not surprising. The Mycenaean name Ne–e–ra–wo /Nehe–lāwos/ has the e-grade,Footnote ⁴⁷ but the vocalism in the Kurzformen Νηλεύς and Νειλεύς is presumably based on that of the full form.

The meaning of βασιλεύς may shed light on its history. In Linear B qa–si–re–u denotes neither the supreme ruler, which is wa–na–ka /wanaks/, nor his second-in-command, the ra–wa–ke–ta /lāwāgetās/ ‘leader of the host’. The latter word survived into the first millennium as λᾱγέτᾱς in Pindar.Footnote ⁴⁸ Remarkably, it was borrowed into Phrygian along with wanaks, since it appears in the early inscription over the façade of the ‘Tomb of Midas’ at Midas Şehri near Afyon.Footnote ⁴⁹ Indeed, when the stone-cutter incised lawagtaei : wanaktei (), one suspects that he was supposed to put the Greek dative singular λᾱϝαγέτᾱι, and simply miscopied what should have read *lawagetai : wanaktei by anticipating the ending of wanaktei.

Mycenaean qa–si–re–u clearly denoted some sort of minor chief, mayor, or headman, since the term describes not only an ‘overseer’ of bronze-smiths but also various local officials.Footnote ⁵⁰ In Pylos tablet Jo 438 the a–ke–ro qa–si–re–u, perhaps ‘overseer of messengers’,Footnote ⁵¹ is assessed to contribute gold along with various people of the rank of ko–re–te ‘mayor’, po–ro–ko–re–te ‘deputy mayor’, mo–ro–qa, and [du]–ma, and with individuals who can be identified as holding the prestigious posts of e–qe–ta and da–mo–ko–ro. A man who is known from Pylos tablet Jn 431 to have been a g ^wasileus is, according to tablet An 261, in charge of a ke–ro–si–ja /geronsia/, ‘council of elders’.Footnote ⁵² The important Knossos tablet As 1516 records at Phaistos and Setoia two groups called g ^wasilēwiai of 23 and 17 men each, alongside a larger group at Knossos, consisting of 31 men, which is called lāwāge<si>ā. Each group may well have been another ‘council of elders’. As Ventris and Chadwick noted, ‘the use of these terms suggests that the basilēwes of subordinate districts were in some sense analogous to the lāwāgetās at Knossos’,Footnote ⁵³ and, as Shelmerdine observed, ‘the association and relative numbers could indicate that the g ^wasileus stands not far below the lāwāgetās in the administrative hierarchy’.Footnote ⁵⁴

Benfey's etymology of βασιλεύς accords with the view that the g ^wasileus originally held an office of high rank, the authority of which was greatly reduced when administration was concentrated in the palaces. There is a consensus that the power of the βασιλεύς rebounded after the fall of the palatial centres around 1200 b.c., when, with society ‘decapitated’, authority devolved back to the level of the village; this will explain how the word came to mean ‘king’ in Homeric and especially post-Homeric Greek.Footnote ⁵⁵ If ‘these officials really derived their power from local communities’, this ‘would account for both their survival and their increased status after the palatial system collapsed’.Footnote ⁵⁶

As the class of nouns in –εύς seems hardly paralleled outside Greek, it was probably created within Proto-Greek. Indeed, since there is no evidence at all for a West Greek (or Proto-Greek) form †βατιλεύϲ,Footnote ⁵⁷ the word g ^wasileus should have arisen only after East and West Greek had been differentiated, when East Greek began to assibilate τι in most phonetic environments; to judge by the evidence of Mycenaean, which is indubitably a dialect of East Greek ancestral to Arcado-Cypriot, this differentiation must have occurred by the middle of the second millennium b.c. The rise of the office of g ^wasileus well suits the development of nucleated small states with local rulers and minor tholos-tombs such as we see at this time in southern Greece, which is exactly where East Greek evolved.Footnote ⁵⁸

The remaining puzzle about g ^wasileus is that there is no parallel for a shortened or hypocoristic form of λαϝός that is not within a proper name. No words ending in –λεύς are anything other than normally formed agent-nouns in –εύς. Hypocoristic shortening other than in names is very rare. One possible exception is νήπιος, which seems to be an abbreviated form of νηπύτιος; it has been suggested that this is owed to baby-talk, as befits the meaning of the word, in-fans.Footnote ⁵⁹ Could g ^wasileus have been created by children as a hypocoristic form of a title *g ^wasilāwos, perhaps in some juvenile pursuit such as the game of βασιλίνδαFootnote ⁶⁰ or its Roman equivalent rex eris si recte facies? Or was *g ^wasilāwos shortened to g ^wasileus in order to mock this relatively low-ranking office-holder, who was perhaps merely a file-leader like a λοχαγός, just as a ‘sergeant’ was originally a ‘servant’?Footnote ⁶¹ Or was g ^wasileus originally a personal name which became a noun for ‘ruler’, like ‘Caesar’?Footnote ⁶² If it was once a name, it is remarkable that its meaning is no less appropriate to leadership than is that of lāwāgetās. But the Spartan king-lists contain many names suited to royalty, from Agis, Agesilaus, and Archelaus to Pleistarchus, Pleistoanax, and Prytanis. In the absence of historical and literary records from the Aegean Bronze Age, the full story eludes us.

In the light of his success in showing that nouns in –εύς are a Greek innovation on an Indo-European basis, Schindler warned against resorting too readily to non-Greek explanations for them:

It does not weaken the overall strength of Schindler's hypothesis that the noun ἑρμηνεύς ‘interpreter’, originally ἑρμᾱνεύς as in Doric,Footnote ⁶⁴ is a clear case of the latter phenomenon. No recognized etymology for it existedFootnote ⁶⁵ until Yakubovitch proposed that ἑρμᾱνεύς is cognate with Carian armon ‘dragoman’,Footnote ⁶⁶ since armon corresponds to Egyptian wḥm ‘dragoman’ in a Carian–Egyptian bilingual from Memphis (text E.Me. 8 Adiego). This is a bronze Apis-bull bearing the Carian inscription paraeym: armon ƙi | para ^!eym: sb polo, together with the Egyptian inscription Ḥ3py dj ‘nḫ Prjm p3–wḥm, which means ‘may Apis give life to Prjm (= Paraeym) the dragoman’.Footnote ⁶⁷ Yakubovitch adds that the initial vocalism of the Greek word has been influenced by the name of Hermes, god of communication.Footnote ⁶⁸

I would myself add to this convincing derivation the suggestion that the common origin of Greek ἑρμᾱνεύς and Carian armon may be the familiar and ubiquitous Akkadian term targumānu ‘dragoman’, ‘interpreter’, from which Hebrew targum and our own word ‘dragoman’ are derived. The Akkadian word, which is attested from Old Akkadian and Ur III onwards,Footnote ⁶⁹ has suffered unpredictable phonetic changes during its lengthy transmission, which must have taken place via first Syria and then Anatolia. It has lost its initial dental, perhaps by a mis-division of words. For this phenomenon one may compare the Hittite rendering of the Ahhiyawan name Ta–wa–ga–la–wa, which was rightly identified by Forrer as the Achaean (Mycenaean) name /Etewoklewēs/, Eteocles, with an added initial dental.Footnote ⁷⁰ Akkadian targumānu has also lost the medial syllable –gu–, probably via a form such as *targmān or *argmān, because of a dynamic stress accent such as Carian is known to have had.Footnote ⁷¹ In Carian armon, the original ā has been lowered to o, a phonetic change familiar both in Carian and in many other languages.Footnote ⁷² In Carian, targumānu has lost its final –u, whereas in Greek ἑρμᾱνεύς has been adapted to suit the class of nouns in –εύς. But of course the inclusion of this particular foreign formation within this existing type of noun by no means undermines Schindler's conclusion that this class of nouns as a whole is of Hellenic origin.