Use the following persistent identifier: http://nrs.harvard.edu/urn-3:hul.ebook:CHS_Benveniste.Indo-European_Language_and_Society.1973.
Chapter 2. The Four Divisions of Society
The tripartition studied in Chapter One is of a functional character and it is by no means identical with the hierarchy of the groups to which a man belongs. These are political divisions that concern societies when studied over their whole extension. Here ancient Iranian has preserved four terms, designating respectively the “family,” the “clan,” the “tribe” and the “country.” But the comparatist often has great difficulty in determining precisely the ancient Indo-European value of these terms.
A close study of the root *dem-, which furnishes the name of a small unit (Iranian dam-), of the “house” as a social entity (Lat. domus, Homeric Greek dô̄), shows that it must be distinguished from the roots *dem(ǝ) ‘to construct’ and *dom(ǝ) ‘to tame’, with which the dictionaries usually associate it. As for the change of sense, observed in several languages, from “house as family” to “house as construction,” this reflects a social change: the breakup of the “Grossfamilie,” which led to the gradual substitution of a society structured according to genealogy by a society subdivided geographically.
We must therefore separate Gr. dómos ‘building, house’ and Lat. domus, which designates not an edifice, but the “home” as a social entity, whose incarnation is the dominus. Consequently, domus entered into contrasting pairs, the second term of which designates what is outside the circle of the home: domi militiaeque, domi : peregre, domesticus : rusticus; the couple domi : foris ‘home-outside’ shows that the word *dhwer– ‘door’ designated the frontier, seen from inside, between the inside and the outside world. As contrasted with Iranian terms, the Homeric words for “family,” “clan,” “tribe”—génos, phrḗtrē, phûlon—attest both lexical innovations and political conservatism.
Finally, if to the Iranian word for “country”—dahyu—there corresponds the Skt. word for “foreign slave,” dasyu, this is because the Iranians naturally called their people, seen from inside, by a derivative of daha ‘man’, whereas for the Indians the same dahyu, seen from the outside, appeared necessarily as a “slave-stranger”: here we find once again another illustration of the importance of the opposition inside : outside.
The tripartite organization which has just been described establishes functional classes within society. This division is not of a political nature, except for the fact that the priestly class, being the first, determined the hierarchy of powers. The social organization proper rests on a quite different classification: society is considered not in the light of the nature and hierarchy of its classes, but as what may be called its national extension, a man being regarded as belonging to circles of increasing magnitude. This structure is clearest in ancient Iran. It comprises four concentric circles, four social and territorial divisions which, proceeding from the smallest unit, increase in size until they comprise the whole of the community. The terms which designate them are:
- dam-, dǝmāna-, nmāna– (equivalent forms which are distributed according to the date of the texts), “family” and “house.” The second form, dǝmāna-, is derived from the first, dam-, by suffixation, and dǝmāna– evolved by sound change to nmāna-.
- Above this, vīs ‘clan’, a group of several families.
- Above this, zantu ‘tribe’, properly “the whole of those of the same birth.”
- Finally, dahyu, which may be rendered as “country.”
Alongside each of these Iranian terms we can put the Sanskrit correspondent: dam ‘house’ (Av. dam-); viś– ‘community, people’ (Av. vīs-); jantu– ‘creature’ (Av. zantu-). To the fourth term, Avestan dahyu– ‘country’, corresponds Vedic dasyu which, in circumstances which we shall try to determine, has taken on the sense of “barbarian enemy population.” But in India we do not find an organic connection between these four terms. They no longer form a whole. The ancient schema is already altered. Iranian society has been more conservative.
The same observation is true of the classical languages. We find words that are the congeners of the first three terms: Gr. démos (δέμος), Lat. domus; Gr. woîkos (wοῖĸoς); Lat. metis; and Gr. génos (γένος) (a neuter in –s), Lat. gens (a feminine in –ti, hence Lat. *genti– as compared with *gentu-, the prototype of the Iranian term). But in the classical world they do not constitute a series any more than they do in India. The correspondence is simply etymological. In Greek and Latin, these inherited words are not arranged as they are in Iranian. There is not even parallelism between Latin and Greek. Far from constituting two distinct social units, Gr. dómos and (w)oîkos signify practically the same thing, “house.” Date, dialect and style govern the choice of one or the other. Nor does Latin present the Iranian structure: vīcus is not the superior grade to domus; it differs from vīs in Iranian and also from (w)oîkos in Greek.
Furthermore, in Greece and Rome, new words unknown to Indo-Iranian joined this ancient series; e.g. Gr. phulḗ (φυλή) and Lat. tribus.
We may nevertheless take it as certain that the Iranian terminology for the social divisions goes back to the Indo-European period. The four terms cited from ancient Iranian reappear in the compound words designating the “chief” (pati) of each division: dmāna–paiti, vis–paiti, zantu–paiti, dahyu–paiti. This hierarchy (because it is one, with deep roots) persisted in the same order into Middle Iranian, in spite of the evolution of the vocabulary and of the language: mānbed, visbed, zandbed, dehbed. The fact is that this structure goes back far in time. We discover it, for two of the terms, in a state previous to Iranian and in the same composite form: Av. dǝmāna–pati has parallels, (1) in Vedic dam–pati– ‘master of the house’, and (2) in Greek despótēs (δεσπότης), while Av. vīs–pati ‘chief of the clan’ has correspondents in (1) Vedic viś–pati– and (2) in Lithuanian vë̃š–pats ‘chief of the clan’, which developed the sense “lord.”
The grouping of these terms shows how they were organized. We must now consider them successively and define each of them individually.
The word for “house,” which comes first, is one of the best known elements of the Indo-European vocabulary. Moreover, it is connected with a verbal root in a manner which seems immediately comprehensible and satisfying. The Iranian form dam– can be linked with the word family of the Latin domus. If in Latin domus (fem.) is a stem in –u-, we know from indirect evidence of Latin itself that it coexisted with a masculine stem in –o-, for *domo– is presupposed by the derivative dominus. The Greek form dómos confirms this. In Greek, side by side with dómos ‘house’, we have the feminine domḗ (δομή) ‘building’ and the agent noun *domós (*δομός) with the accent on the suffix, which enters into the compound oiko–dómos (οἰκο-δόμος) ‘he who builds the house’. The thematic form is also known in the Ved. dama– ‘house’. The stem in –u-, attested by the Latin domu– and the Old Slav. domŭ, is also seen in the derived Vedic adjective damū–naḥ ‘domestic (protector) of the house’, and further in the Armenian compound tanu-(tēr) ‘(master of) the house’.
Both *domo– and *domu– come from an ancient root noun which may have the forms *dem-, *dom-, *dm, *dm̥-. It appears both as a free form or in compounds; e.g. the Skt. expression patir dan and dam–patiḥ, Av. dəng paitiš (where dəng represents *dams) “master of the house,” the Greek correspondent of which is despótēs or déspoina (δέσποινα). These two Greek compounds were no longer analyzable in historical times, but the elements are easily recognizable, and their combinations also occur elsewhere: –pótēs (-πότης) and –poina (-ποινα) represent respectively the ancient masculine form *poti ‘master’ and the ancient, archaic feminine *potnya ‘mistress’; the compound des–poina has as its Vedic counterpart dam–patnī ‘mistress of the house’.
There is further evidence for the root-noun *dem in Greek, e.g. the Homeric expression hēméteron dô̄ (ἡμέτερον δῶ) ‘our house’, originally *dōm (like Armenian tun ‘house’), which was later on extended to dō̂ma. It is generally accepted that dámar (δάμαρ) ‘legitimate wife’ belongs to the same word-family and is analyzed into dam– ‘house’ and –ar from the root of ἀραρίσκω ‘to order, to arrange’; the meaning is thus “she who administers the house.” The zero grade of *dem-, that is, *dm-, appears in the Homeric mesó–dmē (μεσόδμη), in Attic mesómnē (μεσόμνη), which designates the central beam that joins together two uprights, two pillars in the interior of the house. Apart from this we have *dm–ōu– in dmṓs (δμώς) ‘servant’, genitive dmōós (δμωός), feminine dmōḗ (δμωή) ‘female servant’, the term meaning “he (or she) who belongs to the house.”
This whole group of noun forms is traditionally attached to a verbal root *dem– ‘to construct’. The forms of *dem– testify to what is called a disyllabic root: *dem–ə– and *dmā, Gr. démō (δέμω), perfect passive dédmētai (δέδμηται) cf. neódmātos (νεόδματος) ‘recently constructed’, démas (δέμας) ‘form, physical appearance’, properly “structure.”
From different stems of this root a number of noun forms are made. Particularly noteworthy are those Indo-Iranian derivatives with the suffix –ana-, Avestan dəmāna-, Old Persian *māna-, Vedic māna– (from *dmānā-), and those with the suffix –ro-, the Germanic *dem–ro-, Old and Modern English timber ‘wood for construction’, German Zimmer ‘wood-work’, then “room,” as well as the ancient Gothic denominative verb timrjan ‘to carpenter.’
Finally, scholars consider that this root *dem– ‘construct’ has yielded, apart from the word for “house,” a derived verb from this noun, signifying “to tame,” a verb represented in Latin by domare, in Greek by damáō, etc. The basic sense is posited as “to attach (an animal) to the house, to domesticate.”
The whole of this etymological group is listed in recent dictionaries under the same heading *dem-, and in their arrangements the entries start from the basic notion “construct.” However, Meillet expressed some reservations about the morphological connection between *dem– ‘construct’ and *dem– ‘house’.
At first sight, this great etymological reconstruction, comprising a large number of forms culled from all the languages of the family, raises no major difficulty. The proposed connections between the notions are at least plausible. It seems quite natural that words designating “house” and common to almost all languages should be derived from a verbal root no less ancient signifying “construct.” It would follow that the first social unit, the “house” or “family,” owed its name to the technique of carpentry.
But a demonstration cannot be regarded as certain simply because it is not improbable. Each of these lexical groups thus brought into relationship reveals, on closer examination, peculiarities of form and sense which seem original and irreducible; these must be brought out before we can collate them. Only this preliminary analysis will enable us to pass judgment on the genetic relationship of the forms. The comparative method is here put to the test over the whole extent of our investigation. We must start with the data basic to this comparison and attempt their description with all proper precision.
If we examine the word for “house,” we shall soon notice that domus in Latin and dómos in Greek, although they appear, apart from the morphological difference of stem (Lat. –u-, Gr. –o-), to tally completely, differ in many respects in their lexical usages. In Homer, dómos is accompanied by descriptive epithets; the house is “great, high, well constructed, wide,” etc. That is to say, it has the characteristics of a construction; the dómos includes a vestibule, which is called pródomos, ‘the front part of the dómos’. In Latin we find nothing comparable: domus always signifies “house” in the sense of “family,” which is quite foreign to dómos. Moreover, certain case forms of Lat. domus are fixed in an adverbial function: domi, domum, domo. In Greek, these adverbial uses are impossible with dómos and dô̄ma; they certainly exist, but the word concerned is oîkos. We have, corresponding to Lat. domi, domum, domo, Gr. oíkoi (oἴĸoı), oíkade or oîkonde (oἴκαδε, οἶκονδε), oíkothen (οἴκοθεν).
In the same order of ideas, we observe that domi, domum, domo, signify only “the home,” with or without movement, as the point of arrival or departure. These adverbs oppose the “home” to that which is outside it (foras, foris), or to foreign parts (peregre); or they contrast everyday occupations, the works of peace, domi, to war, militiae. Such ideas could hardly be reconcilable with the word for “house” if we had to take it in a constructional sense. It is clear that these adverbial uses imply a moral rather than a material connotation for domus. Let us consider for the present the accepted connection between domus and a root *dem– ‘construct’. If the “house” was simply the “construction,” we would expect to find a verb *dem– in Latin. But the verb corresponding to Greek démo ‘construct’ is absent from Latin vocabulary, which removes domus still further from Gr. dómos. The divergence between the two languages and the distance between the two notions is clearly brought out if we examine the expressions for “to construct (a house).” Greek has a verb oikodomeîn, the denominative of the composite expression oiko–dómos ‘house builder’, where we note that the agent noun from *dem– has as its object oîkos and not dómos. What is the Latin equivalent of oikodomeîn? It is a compound verb: aedificare. Thus to Gr. –domeîn corresponds Latin facio and not a verb from the same root; to oiko– corresponds not domus, but aedes. The formation of aedificare is thus a clear proof that the true value of domus has nothing in common with that of aedes, and consequently that domus cannot have been an architectural term. If further confirmation were needed, it will be found outside Greek or Latin in a third compound verb of the same sense: in Oscan “construct” is tríbarakavúm (infinitive in –um). This verb is formed from trííb– (= treb-) ‘house’ and ark– ‘to enclose, to entrench’ (cf. Lat. arceo). This may be an Oscan calque on the Greek oikodomeîn, the result, like a number of other Oscan borrowings, of the influence of Greek civilization. But in Oscan also, the material “house” has a special word, *trēb.
We therefore have in each of these three languages a verb indicating the material construction which is a compound including the name for “house”; now, this noun is never made from *dem– ‘construct’. This is a new indication which marks the difference between the sense which had been reconstructed for domus and the sense actually found.
This clarifies the problem in Latin. Two nouns, aedes and domus, can equally be translated by “house”; but they are not equivalent, and they differ greatly in their derivation. Aedes, meaning “house,” “temple,” viewed as a construction, gave rise to a derivative aedilis, the magistrate in charge of the construction of houses and more especially temples. From domus we have no comparable derivative: *domilis does not exist. Conversely, two derivatives are peculiar to domus: (1) domicilium, the second term of which is itself derived from ancient –cola which figures in agricola; now domicilium ‘seat of the domus’ indicates the house as a residence and not as a construction; (2) dominus, a social term. For us, domus and dominus are different words, but the Romans felt them as closely linked. For instance, in one of those etymological verbal games favored by the early Latin poets, we find: “O domus antiqua, heu quam dispari dominare domino”; in Cicero: “domus erat non domino magis ornamento quam civitati”; finally in St. Jerome : “in navi unus gubernator, in domo unus dominus.” So the dominus is in no way responsible for the construction of the house.
Finally, the usages of domus in Latin exclude all allusion to construction: frequently used with possessive pronouns, domus mea, apud me domi, it always means “home.” From this comes the turn of phrase aliquid est mihi domi ‘I have something at home’, equivalent to “I possess.” Thus in Plautus, cui argentum domi est ‘he who has money (at home)’. All these features characterize domus as a family, social and moral notion, but never as a material one.
In Cato we read an ancient prayer, addressed to Mars on the occasion of the lustration of the fields. It consists of archaic formulae, transmitted from generation to generation and reproduced literally. The person making the offering, after having performed the rite, calls for the protection of the divinity mihi, domo familiaeque nostrae. Thus domus takes its place between the person of the celebrant and his familia.
When in Virgil Aeneas calls out as he disembarks: Hic domus, hic patria est he joins domus and patria in their common membership of the sphere of social and moral notions.
But the term which it is most important to define, because it itself defines domus, is the derivative dominus. Its peculiar formation by itself arrests our attention. The stem is domo-, not domu-; the formation is peculiar, with –no– as a secondary suffix, that is to say, applied to a noun already existing in the language. This type of derivation is not very common. The suffix occurs in a small series of words, the meaning of which is instructive: first, tribunus, which stands to tribus as dominus (stemming from *domo–no-) does to domus. Apart from this, the formation is found in some proper names, all names of gods. Portunus is the god charged with protection of the ports and the wealth accumulated there; he has in his service a flamen portunalis, and festivals are devoted to him—the Portunalia. From this name it is clear that he is the god of the portus, meaning strictly the mouth, but also the crossing of a river. Neptunus is not analyzable in Latin itself; but by means of the comparative method we can restore a noun *neptu– (stem in –u) which would signify “humidity, aqueous element.”
The formation of Fortuna demands an explanation. In the traditional, but not altogether clear, expression forte fortuna ‘by chance’, we can see that fors and fortuna constitute a single phrase, but it is not immediately evident how the two words are coordinated. Fors is ancient *forti-, going back to *bhr̥–ti-, an action noun from the root of fero; we must remember that the root *bhr̥- does not simply signify “to bear,” but rather “to bring” and also “to take away,” so that fors is “the action of bringing,” “what fate brings.” Fortuna, for its part, is not a simple doublet of Fors; it is an adjective which qualifies Fors and gives it greater precision. The Fors Fortuna is the divinized Fors of *Fortu-; the existence of the form *fortu is confirmed by the adjective fortuitus. As a female personage Fortuna stands to *fortu as Portunus does to portus.
Finally, we have Tiberinus, a figure of ancient Roman mythology. Ancient prayers invoke Pater Tiberinus, the god whose name is derived from Tiberis, the Tiber. This secondary formation in –nus thus comprises a certain number of divine names for divinities who preside over an element or a force, and two terms designating a social function, dominus and tribunus.
This lexical peculiarity is revealed in its full significance when we find that the same suffix is employed outside Latin with the same function. In ancient Germanic, we have a group of words with this suffix which comprise, just as in Latin, on the one hand words for social functions, and on the other proper names: þiudans, the word for the king in Gothic, goes back to an ancient *teuta–nos, which signifies the “chief of the *teutā,” of the tribe, the community; Gothic kindins ‘chief of the clan’, from *genti–nos, chief of the gens. By combining the evidence from other Germanic dialects, we get also *druxti–nos, Old Icel. drottinn ‘chief of the troop’, cf. Old English dryhten; the basic term is drott– ‘troop’ in Old Icelandic. This type of formation reappears in Old Icelandic Herjan, the second name of Odin, which is coined from herr ‘army’. The prototype is *koryo–nos, which recalls Gr. koíranos ‘chief’. Certain of these terms feature in personal names, even outside Germanic, e.g. Gaulish Toutonos, Illyrian Teutana and Gaulish Coriono–totae.
But there is a far more famous name which belongs to this series; that of the great god of Germanic mythology, Wodan (Wotan, Odin) : Wōda–naz (a derivative in *-nos) made from a term *wōδa-, an ancient form of German Wut ‘fury’. The problem is only how to interpret the name. In these secondary suffixations in –nos, the root term designates generally a group of men, a social division. For an abstract notion like wōδa to find its place in this series, we must transpose the abstract into a collective noun and understand *wōδa as “the people possessed by fury.” This interpretation is not without support, if not in the language, at least in the conceptions of ancient mythology. This notion is that of the Wild Hunt known from the literatures of the Middle Ages; a band of the dead who once a year, led by their chief, return to the land of the living, and after devastating everything in their path vanish into the underworld. Wotan–Odin would then be their chief. This is a plausible hypothesis. We note also that it accords with the surname of Wotan, Old Icel. Herjan, literally “chief of the army,” cf. Gothic harjis ‘army’, German Heer. 
We thus possess, to illustrate the function of the derivative in –no-, a considerable body of facts which give us a good deal of help, but we do not find in them in all cases the notion of “chief.” For some of them, this translation is well suited: Lat. Portunus is effectively the master of the ports, and Gothic þiudans that of the people. It is difficult to interpret the name of Neptu-nus in the same way. The connection between Neptunus and the element of water cannot be transposed as such into the social domain. In fact here we have an incarnation, not the exercise of authority: Neptunus personifies the watery element, he represents it. We can therefore say that þiudans in the same way personifies his people. This must be understood in the light of the nature of the institution itself. We use the term “personifies” here, bearing in mind the manner in which the king was appointed by the Indo-European peoples of Central and Western Europe.
There was no hereditary king, but only a king by election; he was chosen from among the people, says Tacitus. Similarly, in India, the assembly had to choose from among a certain class the one who was to represent it. Seen from this angle of “representation,” “personification” or “incarnation” of the basic notion, the two series of words formed by the suffix –nos can be brought together. The list can in fact be increased. There are secondary derivatives in –no elsewhere; thus the Greek word for “moon,” Selḗnē (Σελήνη < *σελασ-νᾱ), derived from sélas ‘(lunar) radiance’, is a noun formed in the same way as Lat. tribunus or Gothic þiudans. We see in Selḗnē the personification of the particular luminosity of the moon. In this way we can find a single principle in this mode of derivation which later became specialized so that it formed the names for heads of social sections.
This brings us back to the formation of dominus. The personage called dominus has authority over the domus; he represents and incarnates it. This leads once again to the same conclusion regarding the meaning of domus. This word does not signify the material construction. It is within an exclusively social and moral conception of domus as a human group that domus and dominus find their respective explanations, and this illuminates their relations.
This is supported by the meaning of another derivative, domesticus, the formation of which is parallel to that of *rowesticus (> rūsticus), if it is not coined on the basis of the latter. The adjective domesticus qualifies what belongs to the house, as against what is foreign to it; it does not imply any connection with the material aspect of the house.
Do we have to suppose that Latin has transformed into a social notion a word which originally had a material significance, which was inherited, and which was the basis of Latin domus? Such a transformation, if it took place, could not have been a total one; it would have left traces in the Roman world itself. But there appears to be no reason at all to suppose this. We are of the opinion that there was an unbroken continuity between the Indo-European sense of the word and that of Latin domus. We can project back into the Indo-European period the correspondence of Skt. dam patiḥ with Gr. despótēs ‘master of the house’. It is true that the sense of “master of the house” has been effaced, or at least weakened, in Greek, where despótēs signified at an early date “master” in general, and not only of the house, so much so that in the New Testament it was felt necessary to create oiko–despótēs to express “master of the house.”
This was because in despótēs the word for “house” was not felt any more. As early as Attic prose we find phrases like οἴκου, or οἰκίας δεσπότης ‘despόtēs of the oîkos, the oikía’, when he exercises his authority within the house. Now this archaic compound *dem(s)-poti– ‘master of the house’ refers in the first component to “the house qua family” and not to the “house qua construction.” We find this confirmed in a parallel expression to dam–patiḥ (or patir dan), that is, in Vedic śiśur dan ‘son of the house’, with a term of kinship, śiśus, which implies the “house” as a family and social entity.
Now that we have completed this examination, it appears that *dem– (*domo-) ‘house’, in Indo-European as in Latin, had an exclusively social value. Many other indications can confirm that there is no connection between the notion of “house” and that of “building.” Even in a language which has abolished many traces of its Indo-European past, in Armenian, the term tanu–tēr ‘master of the house’ applies to the head of the family. Similarly, the adjective Skt. damū–na qualifies the divinities particularly honored by the family.
We can also bring out this relationship of domus with the family by a comparison with the term immediately superior to domus in the social hierarchy. This is the Avestan expression vīsō puθra, which designates the heir of a noble family, literally “the son of the vīs.” He is, according to this appellation, the son of the *weik-, the social unit corresponding to vīcus in Latin and (w)oîkos in Greek. This word can only be understood if we regard vīs as a social and family group (in the wider sense of Gr. oîkos) and not as a collection of houses. His designation by vīsō puθra– is thus, at the next higher level, parallel to that of Vedic śiśur dan ‘son of the house’. The two expressions provide mutual support.
After having established the agreement of these testimonies, we must now examine the Greek facts, some of which bring striking confirmation of our conclusions. Not only despόtēs but also dámar is no longer analyzable in Greek itself, a word which denotes “she who administers the house”; dmṓs ‘the servant, the slave’, dmoḗ ‘the female servant’, that is as a whole, “those who form part of the household,” like the famuli of Latin. Finally, the Homeric form dō̂ in ἡμέτερον, sometimes ὑμέτερον δῶ ‘in my house, at my home’, ‘at your home’, parallel with Latin domi, domum, conveys the notion of the house as “inside.” Here is a lexical series which has clearly carried on in ancient Greek the sense of *dem-, *dom– which we have already recognized as Indo-European.
But against this group we have to posit a group of forms which in Greek must be recognized as distinct and belonging to another family. First the noun dόmos, which applies to buildings: “house,” “temple” and also “a room,” and sometimes “nest.” Herodotus takes it in the sense of “an arrangement of stones or bricks” serving for the construction of a wall, or of a house. It is exclusively to construction that mesόdmē ‘the large transverse beam’ of a building refers. An essential term of architecture is oikodόmos, with the derived verb oikodomeîn ‘construct’, which was the model for Latin aedificare. We also cite the Homeric expression busso–domeúein (βυσσο-δομεύειν) ‘to build in depth, to intrigue, to plot secretly’. Latin offers a parallel expression, which may be a literal translation, in the shape of endo–struos, Classical Latin industrius, literally “constructing inside, in a secret manner.” The parallelism of the formation reveals the equivalence of Gr. –domeúein and Lat. struere. Finally, there is a primary verb démō ‘construct’, which governs objects such as teîkhos ‘wall’ and oîkos ‘house’, this combination being seen in the compound oikodómos; or, furthermore, (h)amaxitós ‘way’: ἀμαξιτός… δέδμηται in Herodotus (VII, 200). We add here the noun démas ‘physical shape, stature, appearance’ which was used adverbially as “in the manner of, literally “according to the appearance, the form of…”
These forms grouped around the verb démō are not a creation of Greek alone. They also have exact correspondents in Germanic: Got. ga–timan, German geziemen ‘to be in accord, to agree’, literally “to be constructed in the same manner”; there is a derived noun *dem–ro-, Old and Modern English timber ‘wood for construction’. From this noun stem *dem–ro– Gothic formed a verb timrjan ‘to work in wood’ (German zimmern) and an abstract word ga–timrjo ‘construction’. If we compare these terms, we see that they require us to posit a root *dem– which, according to the technique involved, had the sense of “construct in tiers” for masonry, and “construct by joinery” for timber construction.
We must recognize another and quite different group. These are the noun forms or verbal forms of a root signifying “to tame,” Lat. domāre, Gr. damáō, a – dámatos ‘untamable’, etc. The sense has no connection with the idea of “house,” but to quite a different notion by a much more satisfactory link. Hittite presents a present tense damaš– ‘to do violence, to oppress, to subject’. It is from this sense that the meaning “to tame” develops by specialization, and we know that the Gr. verb damáō at first referred to taming of horses as practiced by equestrian people, a technical development of sense at first limited to a dialectal area, which cannot be attributed to the Indo-European period.
To sum up, we must carve up the lexical conglomeration which figures in our etymological dictionaries under *dem– ‘construct; house’ into three distinct and irreducible units.
- *domā– ‘to do violence; to tame’ (Lat. domāre, Gr. damáō, Skt. damayati, Got. gatamjan, etc.);
- *dem(ǝ) ‘construct’ (Gr. démō and its derivatives, Got. timrjan);
- *dem– ‘house, family’.
We dissociate, therefore, in the common Indo-European period, the term *dem– ‘family’ from all verbal connections. There is nothing more than homophony between *dem– ‘family’ and *dem(ǝ)- ‘construct’. But it cannot be denied that contaminations came about between the forms issuing from these two roots, as for instance in Homeric Greek between dô(m) ‘house qua family’ and dómos ‘house qua construction’. This is due to a tendency in all the terms of the series to identity social groups with material habitat. 
The same fact recurs at a higher level of society, in the forms of the nominal stem *weik-, *woiko-, denoting the unit formed from several families. They appear everywhere in the Indo-European area, except in Celtic. The social sense is well established by the concordance between Indo-Iranian viś– ‘clan’ (cf. Vedic viś–pati above) and the Lithuanian vẽ̈š–pats ‘lord’. But it has evolved to the material sense of “group of houses,” “village,” “town,” in Latin vīcus ‘town, quarter of a town’, Old Slavic vĭsĭ ‘village’, Gothic weihs ‘village, domain’. Gr. (w)oîkos occupies an intermediate position: first “(large) house,” in which all the descendants of the head of the family lived, then a word substituted for dómos, as we have seen above, and finally “house, building” in oiko–dómos ‘builder, architect’ with its numerous derivatives and compounds. Thus the word for a social unit has been transferred to the material sphere which delimits that unit. A new relationship then becomes established between those grouped together in the same habitat: this relationship is illustrated in Latin by the connection of the sense of between vicus and the derivative vicinus ‘who belongs to the (same) vicus’, hence “neighbor.” In separate languages, the representation of ancient *weik– enters into a given specific series and so in each language takes on the sense assigned to it by its place in the series. But it is still clearly apparent in historical times that in the period of Indo-European unity this word was one of the terms referring to a division of society.
It is thus clear that the Indo-European terms have been subjected to profound changes of sense. Through these changes we can detect an important fact of civilization, a transformation of the institutions themselves, to which the vocabulary gives indirect witness.
What *dem– and *weik– once signified in the Indo-European organization, namely the divisions at different levels of society, are in languages of the historical period designated by new terms, such as *genti– or *teutā-, in a part of Western Indo-European. In Latin, once vicus had become the term for a “quarter” of a “village,” new designations had to be devised: tribus and civitas.
This change is just as far reaching in Greek, but it takes on a different aspect. The largest unit is that of the génos, which was much more extensive than the Roman gens, and is not to be confused with the phratría, a division which is also purely Hellenic. The phratríai, in their turn, are grouped into phulaí.
Two important transformations have come about:
1) the break-up of the “Grossfamilie” into separate families. The ancient period was characterized by the “Grossfamilie” in which, after marriage, all the sons continued to live together, bringing up their own families, while sometimes even the daughters brought their husbands. At this stage, there was no individual property; the whole family domain was an undivided property. We cannot properly speak of inheritance because the “Grossfamilie” itself remained the proprietor, and its rights over its possessions never lapsed. Then the “Grossfamilie” broke up. For economic reasons, the sons left at an early age. The terms which applied to this “Grossfamilie” were more and more rarely used, for the notion itself no longer corresponded with any real institution; the “Grossfamilie” was divided up into much smaller units when the descendants in their turn went off to found new homes.
2) The second transformation was the establishment of the Achaean warriors in a pólis (πόλις), a common township. This evolution slowly abolished the earlier social framework in favor of new territorial divisions. The old social divisions founded on genealogical line of descent were progressively replaced by groupings determined by a common habitat.
This habitat is no longer the privilege of those with a common origin. In the pólis or the kṓmē (κώμη) it was chance or war which brought together those who lived in it. Aristotle, at the beginning of his Politics, does no more than codify an established situation when he characterizes the elements of the society qua “community” (koinōnía). The ultimate unit he describes as the oîkos (the Romans would say domus); for him it is the smallest division and the first form of society which existed, and he defines it as a community of husband and wife, of master and slaves: this is the notion of the Roman familia. The oîkos is, in fact, constituted by the daily participation in food and worship. After this, Aristotle posited a progress upwards to the village (kṓmē) and the city (pólis).
Today we see things differently; such a reconstruction, which starts from a social cell and proceeds by successive accretions, is false. What existed from the start was the society as a whole and not the family, then the clan, then the city. Society from its origin was divided into units which it comprised. The families are necessarily grouped within a unit, and so on. But Aristotle makes into a universal phenomenon and a philosophic necessity what was represented in his own society: he makes an absolute of a particular social state of affairs.
It is this great process of transformation which is reflected in the vocabulary: like dómos, the term oîkos from then on became a word for a habitat. In Greek prehistory, as we have seen, the “house” was not a building; similarly, the Homeric expression designating the Dorians as trikhaí–wikes ‘those divided into three tribus’ preserves wik-, related to (w)oîkos, in its first sense of a social grouping. But soon oîkos took the place of the ancient *dem– ‘house’ and so came to mean “house” as building. The change which came about in society produced (1) the new reference of the terms to the material sphere; (2) the “hierarchical” transfer of the term to the place of another: the sense of *dem– passed to oîkos in Greek; hence the locative oîkoi, etc., which corresponds to Lat. domi, etc. and signifies “at my home,” “at your home.” Thus oîkos has taken over the whole of the ancient semantic domain of *dem-. In general, we observe the abolition of the Indo-European structure and the advance of new terms. The old genealogical words become emptied of their institutional and social meanings and become a terminology of territorial divisions. Each language proceeds to a new adaptation of its terminology. The very way in which this transformation takes place in different languages is highly instructive, because the languages are not Indo-European in the same way. Latin is Indo-European in its fidelity to ancient usage, to the vocabulary of institutions, even when this vocabulary relates to new realities: Greek, conversely, is Indo-European in the persistence of the primitive model, around which it organized a new series of terms.
The category of meaning in which the word for “house” finds its Indo-European value determines also the same notion in its other aspects. Among the uses of domus we must now consider the adverbial form domi and the opposition which Latin usage has established from the beginning between domi ‘at home’ and foris ‘outside’, or, with reference to movement, between domum and foras.
We have here, on closer examination, an opposition which could not have been foreseen, and which contrasts two terms that are not by nature antithetical, because one is the word for “house” and the other the word for “door” (fores). Here a new notion came into play with lexical consequences, that of “door.”
There are in the Indo-European languages several words for “door”: the distribution is haphazard. The word may even be restricted to a single family of languages. Thus in the Italic dialects, Oscan ueru ‘portam’, Umbrian uerofe ‘in portas’ with a postposition –e. The word goes back to an ancient neuter form *werom ‘closure’, derived from the root *wer– (Skt. vr̥ṇóti, ‘it closes, it encloses’, German Wehr), a localized term which apart from Oscan and Umbrian has a correspondent form only in Slavic and Baltic. In other languages, on the contrary, a multiplicity of terms invites our attention. In Latin, we have four: fores, porta, ianua, ostium. Even if some authors seem to use them indiscriminately, we know that at an ancient date they did not have the same signification.
Of all the words the one represented in Latin by fores has the widest distribution; it is attested in nearly all the other languages. The Indo-European form is *dhwer-, in the reduced grade *dhur, Gr. thúra (θύρα), generally in the plural, because it seems that the door was conceived of as having various elements.
*dhwer– is an unanalyzable term by itself, which cannot be attached to any verbal root, and its etymological signification escapes us; but is it possible that we have here a term for a material object which owes its name to the functions which it fulfills?
What is important to stress is the concordance of the adverbial usages of *dhwer– in Latin and in other languages. Some of them, in fact, present uses exactly comparable to that of Latin fores ‘door’ and foras ‘outside’: Gr. thúra ‘door’ and thúra–ze (θύρα-ζε) ‘outside’; Armenian durkc ‘door’ and durs (acc. plur.) ‘outside’. We also have in Gothic a compound faura–dauri, literally “out-of-doors,” which translates plateía (πλατεία) ‘street’.
We have here an adverbial form which was fixed at a very early date and became independent, so much so that thúraze, having lost in Homeric times its connection with thúra ‘door’ (of the house), it was possible to say ἁλὸς θύραζε ‘out from the sea’ in the Odyssey (5, 510; cf. Il. 16, 408). In the Slavic languages, the connection between the two terms continues; on the one hand dvĭrĭ ‘door’, but also, in all modern Slavic languages: Russ. na dvorě, Serbian nadvor, etc. “outside,” literally “at the door.” Such correlations, the antiquity of which is evident, explain the nature of the idea. The “door,” *dhwer-, is seen from the inside of the house: it is only for the person inside the house that “at the door” can signify “outside.” The whole of the phenomenology of the “door” proceeds from this formal relation. For the person who lives inside, *dhwer– marks the limit of the house conceived as an interior and which protects the inside from the menacing outside. This notion is so deeply and enduringly inscribed in the Indo-European languages that, for us too Fr. mettre quelqu’un à la porte, lit. “to put someone at the door,” is “to put him outside”; “open or close one’s door to somebody” is “to admit or not admit him into one’s home.”
We can understand why in Latin foris is the opposite of domi: the “outside” begins at the door, and is called foris for the one who is “at home,” domi. This door, according to whether it is open or shut, becomes the symbol for separation from, or communication between, one world and the other. It is through the door that the secure and enclosed space which delimits the power of the dominus opens on an extraneous and often hostile world; cf. the opposition domi/militiae. The rites of passage through the door, the mythology of the door, give a religious symbolism to this idea.
It is significant that the adjective made from the word for door does not designate what concerns the door itself, but what is outside, the extraneous world. This is also the meaning of the adjective thuraîos (θυραῖος) ‘extraneous, from abroad’, from thúra ‘door’, in Greek. Similarly, late Latin has coined from foris, foras the derivatives foranus, foresticus, forestis, all of them referring to the outside, the extraneous world. This sense remained alive; it continued to be productive even after the ancient name of the “door” was replaced by new terms, e.g. in the Romance languages, where it has produced adjectival derivatives such as Italian forestiere ‘foreign’; in Old French, specially Norman French, horsain means ‘stranger’, ‘he who is outside, who does not inhabit the locality’—and also in modern French, forain ‘who arrives from outside’ (Lat. foranus). Even the French adverb hors necessarily implies a subject who is inside; to put somebody “hors la loi” implies that the subject is inside the law. Thus, although the notion of the “door” is no longer expressed in the Romance languages by forms of ancient fores, it continues to act like an invisible boundary separating the interior space from what is outside. On the other hand, the material sense of *dhwer– is reflected in certain ancient derivatives connected with architecture, like Gr. pró–thuron (πρό-θυρον) ‘vestibule’ (literally what is in front of the door) or Old Slavic dvorŭ ‘courtyard (of the house)’.
The opposition domi/foris has a variant where foris is replaced by a quite different adjective. The opposite term to domi is here taken from ager ‘field’ (< *agros), in the shape of the adverb peregri, peregre, from which comes the derivative peregrinus ‘stranger’. Here again we have two notions which seem difficult to reconcile with the historical meaning of the terms.
Now, this feature of Latin is not isolated. Other Indo-European languages associate the word for “field,” in an adverbial form, with the idea of “outside.” Whereas in Greek agrô̄i usually means “in the country” in contrast to “in town,” elsewhere we find “in the field” means no more than “outside.” Thus Armenian artakcs ‘outside’ is derived from art ‘field’. In the Baltic languages, Lith. laũkas ‘field’ (Latin lūcus) has an adverbial for lauke ‘outside’. Irish has immach ‘outside’ from *in mag ‘in the fields’.
These different but parallel terms conjure up the image of an ancient relationship: the uncultivated ground, the waste land, as opposed to the inhabited area. Outside this physical community, which constitutes the family or tribal habitat, stretches the waste land. This is where the extraneous world begins, and what is strange is necessarily hostile. The Greek adjective derived from agrós ‘field’ is ágrios, which means “wild,” “savage,” and so gives us more or less the counterpart of what is called in Latin domesticus, which brings us back to domus. Whether we start from an opposition like domi/foris or from the wider one in which it is opposed to the “field” (domi/peregre), we always come to the same conclusion, namely that domus denotes the “house” in its social and moral aspects and not as a construction.
In the light of domus and the related forms we can assess the richness and specificity of a terminology which must be counted among the most ancient of the Indo-European world. The other terms relating to the political structure of society are less well attested, according as they apply to larger entities. We might say that the dialectal extension of terms is inversely proportional to the generality of concepts.
We started, we may remind ourselves, with the Avestan series dam-, vīs-, zantu, (dahyu). Now the data are more abundant for the first than for the second division. Both have in common a tendency to assume the meaning of physical habitat.
The third, zantu, belongs to the same etymological family as Lat. gens and Gr. génos, but it differs from these two in its formation. It differs from Lat. gens in that it contains a suffix –tu against Lat. –ti. The study of the two suffixes and their relationship would involve a long discussion, which we have presented elsewhere.  Both have the capacity of forming abstract nouns; –ti has developed more especially in compounds, –tu in simple words. Nevertheless there are simple words in –ti, and gens is one of them.
From a morphological point of view, Latin gens has a correspondent in the Avestan derivative fra–zanti– ‘line of descent’, as well as in Gothic kindins (from *gentinos) ‘hēgemṓn,governor’, a word which has been analyzed above. But the Avestan word zantu– is limited to Indo-Iranian; moreover, Ved. jantu– ‘living creature; a collection of human beings, race’, which corresponds to it, does not have the institutional meaning which attaches to Avestan zantu-. We can see here that the situation of zantu is different from that of gens, and in spite of the resemblance of the terms, there is nothing to prove that they are of the same date.
The important fact is that, as compared with the neuter génos in Greek, we have here words of what is called the “animate” gender, masculine-feminine. The sense of these terms remains close to that of the root *gen-, which does not indicate only physical birth, but birth as a social fact. A number of nominal derivatives make this clear. In a social organization defined by its classes, the birth is the condition of personal status. Terms are required which by the names themselves characterize the birth as legitimate, because of the rights conferred on those whose legitimacy is acknowledged. Besides, such legitimacy is valid first for the men; it is to the men that the collective nouns derived from the root *gen– are applied, and they designate the group which recognizes a common ancestor in the male line. The following conditions express the essential feature of the notion: free, legitimate birth and male descent. They help us to define better these parallel terms from the same linguistic stock, Av. zantu, Lat. gens, Gr. génos.
But the size of the group which each of these terms designates may vary from society to society; they do not occupy the same place in the vocabulary of social and territorial divisions. If in the Iranian series, zantu is the third largest social group, génos on the contrary is the starting point of the Greek series. We come back to the great transformation which in Greece culminated in a new organization of the ancient structure. In Athens, in the ancient form of society, above the génos was the phratría, and above the phratría the phulḗ. According to the Athenian constitution, thirty génē (plural of génos) were necessary to form a phratría, and three phratríai constituted a phulḗ. Here, then, we have specific words which were applied to new entities. But the words themselves are old Indo-European formations, and the notions which they convey formed part of those which informed the ancient Indo-European societies.
This transformation of the ancient structure which finally resulted in the kṓ̄mē and the pólis cannot be connected with any external event, except perhaps the establishment of the Greeks in their historical home and the new circumstances of this habitat. We cannot discern any foreign influence. Everything seems to proceed from native Greek sources—the structure, as well as the vocabulary, of these institutions.
As we proceed from génos to phratría, we pass from a group founded on common descent to a group formed by the totality of “brothers.” These are not blood brothers, but brothers only insofar as they recognize themselves as descended from a common ancestor. This mythical relationship is a profoundly Indo-European notion; and Greek has preserved, better than any other language, the original sense of phrátēr. This is also the case for the correlative term patḗr in Greek (and partially in Latin, too). 
This conservatism is still apparent in many social usages described in the epic. The Heroic Age of which it tells was an historical age. We have in certain respects, in the Homeric usage of certain words, in the connection between the different human groups, the image of what common Indo-European society must have been—in civil life and in war. The manner in which the family, the clan assemble, what their chiefs speak about and how they act, must reflect quite closely the behavior of the warrior class in the Indo-European world. We quote only two passages:
ἀφρήτωρ ἀθέμιστος ἀνέστιός ἐστιν ἐκεῖνος
ὃς πολέμου ἔραται ἐπιδημίου…
ὃς πολέμου ἔραται ἐπιδημίου…
Iliad 9, 63f.
This condemns the man who wages a “civil” war, πόλεμος ἐπιδήμιος, against his fellow citizens within the same dêmos.  Such a man is without phratry, without thémis, without a hearth (hestía). The notions of phratry and hestía are correlative, and between the two, thémis is the customary law which holds good in the family.  The nature of these notions, and especially their interconnection, reproduce those which we have studied from a different angle above.
We thus have, in inverse order, the series: hestía, the “hearth,” that is to say domus; then thémis, the customs which constitute the law, and lastly the phratry. Only the two first divisions of society are mentioned or implied here, because what is concerned is a personal crime. But in war, it is the large social units which are involved, and this is what tests their solidarity. When battle is joined, it is this solidarity which must be maintained among the members of the same clan and the same tribe. This condition governs the disposition of the troops and the plan of battle.
Nestor says as much to Agamemnon:
κρῖν’ ἄνδρας κατὰ φῦλα κατὰ φρήτρας, Ἀγάμεμνον,
ὡς φρήτρη φρήτρηφιν ἀρήγηι, φῦλα δὲ φύλοις.
ὡς φρήτρη φρήτρηφιν ἀρήγηι, φῦλα δὲ φύλοις.
Iliad 2, 362-363.
“Position the men by phûlon and by phrḗtrē so that phrḗtrē may aid phrḗtrē and phûla phûla.” To be victorious in the great trial of strength which battle represents is everybody’s affair; the organization of the army must conform to the structure of society. In this way it will have the greatest effectiveness.
We find in the ancient texts of India and Iran similar recommendations. “Friend” fights with “friend”: each social group must maintain or reconstitute its unity in all circumstances in which the whole of the society is engaged. This principle is not always stated in so explicit a manner as in Homer, but it is no less inherent in the functioning of the institutions of each class.
It remains to consider the last term of the series. This is, in contrast to the two others, limited to Iranian. The Avestan word for “country,” dahyu (ancient dasyu) has as its Sanskrit correspondent dasyu. In spite of the complete identity of form, some scholars doubt the connection because of the difference in sense. In Avestan and Old Persian, dahyu signifies “country”; in Vedic, dasyu is a foreign slave. But the difference can be explained in the light of the older stage of these notions.
In Indic, dasyu may be taken as an ethnic. The dasyu are a foreign people which the Aryans had to fight; they are barbarians, slaves.
But in Iranian, dahyu is part of the traditional and official vocabulary. Darius proclaims himself as “King of Countries (dahyu).” This is a reference to each of the “countries” Persia, Media, Armenia, Egypt, etc., the union of which constituted the Achaemenid Empire. This term must have had a long history in Iranian. It even originated in Iranian society. Today we have some possibility of analyzing its formation. An eastern Iranian dialect, Khotanese, possesses the word daha ‘man’. We know from other sources that in the Iranian world there were a people, the Dahae, as they are called by Latin authors. This people, like many others, simply called themselves “the men.” Thanks to this connection, the sense of dahyu becomes clear: it is a derivative based on the root *das-, of which we have little evidence, signifying a group of men, the most extensive in the tribal order, and hence also the territory they occupy.
We can now understand the strange sense of Skt. dasyu. If the word referred at first to Iranian society, the name by which this enemy people called themselves collectively took on a hostile connotation and became for the Aryas of India the term for an inferior and barbarous people. Thus the connection between the senses of dahyu/dasyu reflects conflicts between the Indian and Iranian peoples. 
[ back ] 1. Cf. Book One, Chapter Eight.
[ back ] 2. On the homophonous roots for “tame,” “construct,” “house-family,” see our article “Homophonies radicales en indo-européen,” Bull. de la Soc. de Linguistique de Paris, vol. 51 (1955), pp. 14-41.
[ back ] 3. Noms d’agent et noms d’action en indo-européen, Paris, 1948, 2nd part.
[ back ] 4. Cf. Book Two, Chapter One.
[ back ] 5. On dē̂mos see Book Four, Chapter Nine.
[ back ] 6. On thémis cf. Book Five, Chapter One.
[ back ] 7. Cf. Book Three, Chapter Five.