sponds generally to the short o of the Greeks; and nouns of the o-declension always exhibit this u in Latin: comp. Xúkos, lupus; TOS, equus; &c. It is probably a remnant of the Etruscan U. In the older Latin inscriptions we have seen o used for this value of u. Thus we have consol for consul, Luciom for Lucium, &c.

U-This letter, like 1,, must be considered as a point of contact between I and u. Indeed, it may be doubtful in some cases whether U, has not been written for 1. The passage of this us into an approximate I is of the following nature:-First, a short o is changed into u2. The genitive of the Greek imparisyllabic declension ends in -os: for this the oldest Latin substitutes -us, as in Castorus, nominus, &c. compared with Senatuos, &c. Some of these old genitives remained to the end of the language, as alius, ejus, hujus, illius, &c. Again, the 1st pers. plur. of the Greek verb ended in -ouev = -oμes: for this the old Romans wrote -umus, a form still preserved in sumus and volumus. Again, in old Latin the vowel of the crude form is preserved in the inflexions, as in arcu-bus, optu-mus, pontu fex, &c. But in all three cases the later Latin exhibits an 1: thus we have Castoris, nominis, &c.; dicimus, scribimus, &c.; arcibus, optimus, pontifex, &c. In these cases we observe that U = O passes into a simple 1. But there are other instances in which the transition seems to go still farther. As the reduplication-syllable is generally shorter than the root-syllable in the preterite of verbs, we should expect that the u or o in the first syllable of cu-curri, mo-mordi, pu-pugi, tu-tudi, would be an approximation to us. Then, again, in cultus, culmen, &c. from colo, columen, &c., the u is clearly less significant than o, though the u here may have been partly occasioned by that affinity between u and of which the French furnishes so many examples, and which we also see in the transition from the Greek 'AσKλýπIOS, 'Hракλns to the Latin Esculapius, Hercules. But there are some cases in which we conclude that the u, which is written, has less weight even than 1. This might be inferred from con-culco, the secondary form of calco, which, according to

1 The older writers wrote memordi, peposci, pepugi, spepondi, according to Gellius, N. A. VII. 9, who, however, says of the common spelling, "ita nunc omnes ferme doctiores hujusmodi verbis utuntur."

the above table, should be either con-cilco or con-celco; and also from difficultas, sepultus, derived from difficilis and sepelio. The fact seems to be, that what would be 1 before R, becomes U before L; so that U, I, are both ultimate forms of their respective vowels, and as such are in a state of convergence.

Accordingly, if we should seek to arrange the Latin vowels in regard to their comparative weight, we should, as the result of this inquiry, have the following order:

Ā (as in musâ, &c.); Ū1, Î1; A; O, U2, I2; E; U3, I3.

§ 6. The Greek Letters used by the Romans.

The Greek letters subsequently employed by the Romans were z, K, and Y. The period at which the first of these was introduced is doubtful; for while, on the one hand, we are told that z is found in the Salian songs (Velius Longus, p. 2217: "Mihi videtur nec aliena sermoni fuisse z littera, cum inveniatur in carmine Saliari"), on the other hand, we find that, even in words borrowed from the Greek, this letter is represented by di, as in Sabadius for Zéßalos (Apulei Met. VIII. 170), judaidiare for judaizare (Commodian, Instruct. adv. Gent. c. XXXVII. 634), trapedia for trapeza (Auctor. Rei. Agrar. p. 248), schidia for schiza, oridia for oriza, &c. (vide Schneid. Elementarl. I. p. 386; and Lobeck, Aglaoph. p. 296, note l.) The fact seems to be, that the Romans had two different characters to express the two different values of the Greek z, which was a dental, either assibilated (as od), or softened (as dy). Now, in its latter use it becomes equivalent to the softened guttural; for the dental and guttural, when combined with y, which is the ultimate vocalisation of the gutturals, converge in the sound of our j or sh (New Crat. §§ 112, 216). When, therefore, the Greek z more nearly approximates to the sound ad, either this is preserved in the Latin transcriptions, as in Mesdentius, Sdepherus, for Mezentius, Zephyrus (Max. Victor. p. 1945); or the is assimilated to the σ, as in Messentius, massa, Atticisso, comissor, badisso, malacisso, &c., by the side of Mezentius, μája, 'ATTIκίζω, κωμάζω, βαδίζω, μαλακίζω, &c. ; or else one or other of the two component parts is omitted, as in Saguntus for Zakynthus, or Medentius for Mezentius. In this case, too, we may consider that the letter a occasionally steps in, as in rixa by the side of epi[8]s. When, however, the Greek z is a softened

S, and therefore equivalent to a softened guttural, we find that it is represented either by the full combination di, as in the cases quoted above, or else by the vocalised guttural (j) only. Of this latter substitution there are numberless instances: such as, Jupiter, Zevs Taτnρ; jugum, Levyos; &c. Of these the most important are the cases connected with the first-quoted example, Ju-piter = Dies-pater; and I must take this opportunity of returning to one etymology belonging to this class, which has always appeared to me to open the way to a chain of the most interesting associations.

έωυτοῦ,

It has been shown elsewhere (N. Crat. § 116) how the Greek H, originally the mark of aspiration, came to be used as a sign for the long e. Out of that investigation it appeared— (1) that a short vowel aspirated may be equivalent to an unaspirated long vowel; (2) that the vocalised consonants i and u may change their place; (3) that these vocalised consonants may be absorbed into or represented by the long vowel only. To the instances given there, I will now add the iota subscriptum of the Greek dative, and the Ionic Greek absorption of u after w, as in Ovua, ewUTOû, &c. These principles explain the connexion between rap, jecur (Sanser. yakrit); pov, diáμetos, dimidius; and between nuépa = diáμepos, and dies2 (comp. diuturnus, juturna; Diana, Janus, &c.). Now, besides nuépa, we have an adjective μepos, "civilised," "cultivated," &c., the regular antithesis of aryptos; and it has been suggested (ibid. § 150), that this word was originally applied to a country through which there was a road or passage, a country divided by a road (diάuepos); just as άypios was properly applied to a rude, open country, with nothing but aypo3. This is sufficiently

1 In many editions of Herodotus we have these words written wüμa, ewüroû, &c.; but the accentuation of Ovua sufficiently proves that it is a dissyllable; and even if we had not this evidence, it would be contrary to all analogy to infer a resolution of a diphthong in a crasis, the sole object of which is to shorten the word. Why should rautó be written, if it were a word of as many syllables as τὸ αὐτό ?

2 In the name of the city 'Ipépa (another form of pépa, see Böckh's note on Pindar, O. XII. 13-21, p. 210), the preposition diá is represented by the aspirated. In the words anti-quus, posti-cus, from antea, postea, we have i eα =

= eai.

8 Hence x@pos with its old synonym xópos (New Crat. § 280), may be considered as an adjective agreeing with the suppressed word äypos, just

proved by Asch. Eumen. 13, 14: κελευθόποιοι παῖδες Ηφαίστου, χθόνα ἀνήμερον τιθέντες ἡμερωμένην. Pind. Isthm. III. 76 (IV. 97): ναυτιλίαισί τε πορθμὸν ἀμερώσατο. Herod. I. 126 : ἐνθαῦτα ὁ Κῦρος (ἦν γὰρ ὁ χῶρος—ἀκανθώδης) τοῦτόν σφι τὸν χῶρον προεῖπε ἐξημερώσαι ἐν ἡμέρᾳ. IV. 118: τοὺς αἰεὶ ἐμποδὼν γινομένους ἡμεροῦται πάντας. In all of these passages the verb nepów implies making a clear passage or road; and in Plato (Legg. p. 761 a.) the adjective uepos is used as a predicate of ὁδός: ὁδῶν τε ἐπιμελουμένους, ὅπως ὡς ἡμερώταται čkaotaι yiyvwvTal'. That the Greeks connected road-making with civilisation in general, and with the peaceful commerce of man with man, appears from many passages (Aristotle, πepi Oavμaoíwv ákovoμátwv, c. 85, p. 837, Bekk.; Thucydides, I. 2, compared with I. 13, &c.); and this is generally implied in all the legends relating to Hercules and Theseus. But it has not been sufficiently remarked that this road-making was also intimately connected with the cultivation of land. It may, however, be shown, that as the Greek άypos becomes nuepos when divided by a road, by a similar process the Latin ager becomes jugerum di-ager-um.

=

Whenever a piece of unemployed ground-of ager, so called was to be taken into use, whether for cultivation, or for the site of a city or a camp, the rules of the ancient limitatio were immediately applied. Now this very word limitatio signifies, the dividing of a certain piece of ground into main-roads (via) and cross-roads (limites); and the same primary notion is conveyed by tem-plum, so obviously derived from tem-no, Gr. Táp-vw, comp. тéuevos, &c. For in all limitation the first thing done was to observe the templum, i. e. as we should say, to take the bearing by the compass. Suppose the

as xapa refers to the suppressed word y: and thus x@pos signifies "land not built on❞—either the open space in a town, or fields in the country (Herod. II. 154: didwσi xwpovs évoikîσai),—and xúpa rather signifies "a region," a territory," in the wider sense.

66

1 The word ἤπειρος = ἡ διαπέραν χώρα, furnishes another instance of the substitution of 7 for diá: comp. the epithet dampúσios, Pind. N. IV. 51, where see the note.

2 Most ancient nations seem to have connected the regiones cœli with the regiones viarum. Thus in old English "the milky way" was called "Watling-street," which was the name of one of the four great roads in this country; see Grimm, Deutsche Myth. p. 330, 2d ed.

augur stood with his back to the north, then the line from north to south would be called the cardo, as corresponding to the axis of the globe; and that from east to west, which cut the cardo at right angles, would be called the decumanus, or "tenth line." For both these lines repeated themselves according to the number of separate allotments into which the land was divided, or the number of separate streets in the city or camp1. Now the Roman actus or fundus = [120 feet] was the unit of subdivision; two of these fundi made a jugerum = di-ager-um, and two jugera constituted the heredium of a Roman patrician : consequently, 200 jugera made up the ager limitatus of a century of the old Roman populus (Fest. s. v. Centuriatus, p. 53). If this ager limitatus, then, were arranged as a square, we have, of course, for each side 20 x 120 feet. Supposing, then, a road between each two of the fundi,-which there must have been, as every two fundi made a di-ager-um,-the cardo which passed between the tenth and eleventh fundus would be properly called the decumanus, and it would consequently be the main road, and would be terminated by the main gate (porta decumana). The point at which the decumanus crossed the cardo was called groma or gruma; and here, in a city or camp, the two cross-roads seem to have spread themselves out into a kind of forum. There is as much probability in the supposition that the immortal name of Rome was derived from this ancient word, as there is in any of the numerous etymologies suggested by Festus (p. 266). From this it appears, that among the Romans it was the same thing to speak of a territory as divided by roads, and to call it cultivated, occupied, or built upon; and the jugerum, or divided ager, implied both. To the same principle

1 It would seem that the word sicilicus (from seco) was properly and originally applied to this apportionment of land. In the Bantine Table (1. 25) we have nep him pruhipid mais zicolois x nesimois; which I have translated (above, p. 127): ne in hoc præhibeat (i.e. præbeat) plus sicilicis x contiguis. According to Klenze (Abhandl. p. 50) x nesimois= decimis; but I cannot understand why we should have an ordinal here. The root of ne-simus appears in nahe, near, next, &c.; and I would understand it of so many adjoining allotments. The sicilicus was 600 square feet, i. e. of the jugerum, or of the actus. Consequently, the 30 contiguous of the jugerum, or of the actus;

1

48

sicilici mentioned in 1.

1

24

17 would be

and the ten contiguous sicilici would, therefore be of the former and

24