{The following is a verbatim copy of chapter 3 from The Latin Language – a historical outline of its sounds, inflections, and syntax by Charles E. Bennett, Allyn and Bacon, Boston, 1907. (Also published in earlier editions under the name Appendix to Bennett’s Latin grammar.) Any deviations or additions to Bennett’s text have been set within curly brackets.

A very thorough review of The Latin Language, written by Truman Michelson, appeared in The American Journal of Philology, Vol. 29, No. 1. (1908), pp. 84–93. His corrections and comments are cited or lightly paraphrased in the relevant places.

The issue of hidden quantity was also discussed independently of Bennett’s work in a series of articles in The Classical Review some years later. Of these, the most convincing, in my opinion, is the one by Carl D. Buck, Hidden Quantities Again, The Classical Review, Vol. 27, No. 4 (Jun., 1913), pp. 122–126. On most points, his views are in accord with Bennett’s, except for the quantity in usquam and related words.

This work, even with Michelson’s ammendments, is naturally somewhat dated. To at least partly remedy that, I have tried to also note where W. Sidney Allen (Vox Latina 2nd ed. 1978) disagrees with Bennett (or Michelson); mainly, these notes concern length due to the “revised Lachmann’s Law” (as stated by Maniet, in Hommages à Max Niedermann (1956)), which, with some exceptions, primarily postulates long vowels in the past participle of the verb when the present stem ends with g and the vowel is not i.

This site employs quite a lot of Unicode characters, which might not show up correctly on all systems. Amongst the more problematic we have y with macron: ȳ; vowels with both macron and breve: ī̆; small uppercase v with apex (acute accent): ; eng: ŋ; and polytonic Greek characters: άὰᾶᾳἀἁἄἅἂἃἆἇᾰᾱ, etc. If you can’t see them, at least you know what is supposed to be there.

p. 36 follows below.}

Chapter III. Hidden Quantity.


A hidden quantity is the quantity of a vowel before two consonants. Such a quantity is called hidden, as distinguished from the quantity of a vowel before a single consonant, where the metrical employment of the word at once indicates whether the vowel is long or short. The quantity of a vowel before a mute with l or r is hidden unless the syllable containing it appear in verse used as short.

The methods of determining hidden quantity are the following:1

1. Express testimony of ancient Roman writers, e.g. Cicero, Orator, 48. 159, where the principle for the length of vowels before nf, ns is laid down (see § 37); Aulus Gellius, Noctes Atticae, ii. 17; iv. 17; ix. 6; xii. 3. Nearly every Roman grammarian furnishes some little testimony of this kind, and though some of them belong to a comparatively late period, their evidence often preserves the tradition of earlier usage, and hence is entitled to weight.

2. The versification of the earlier Roman dramatists, especially Plautus and Terence, with whom a mute before a liquid never lengthens a syllable whose vowel is short. Hence, before a mute followed by a liquid, the quantity of the vowel always appears in these writers, being the same as the quantity of the syllable, just as in case of a vowel followed by a single consonant.

Furthermore, Plautus and Terence not infrequently employ as short many syllables which in classical poetry would be invariably {p. 37} long by position. Examples are the following: juvĕntūs, Plautus, Mostellaria 30; Curculio 38; volŭntās, Trinummus 1166; Pseudolus 537; Stichus 59; volŭptās, Mostellaria 249, 294; Amphitruo 939, and elsewhere. These cases are to be explained by the fact that the vowel was short and the following consonants failed to ‘make position.’ {Michelson: The cited examples are generally explained as cases of shortening by the jambic law, in as much as the syllable containing the e of juventus, etc. is also used as a long by these authors; e. g., at Captivi 69, 470. The point is, that vowels long by nature “in position” are never, or very rarely, subject to the law of iambic shortening. In this way the e of juventus is shown to be short by nature.}

In some instances, it must be confessed, even long vowels are used as short, e.g. bonĭs mīs, Plautus; Trinummus 822, forĭs pultābō, 868. But these cases are of a peculiar sort and may be explained on metrical grounds, or by the iambic nature of the words, as in the examples cited. Cf. § 87. 3. {Michelson: read § 88. 3.}

3. Inscriptions.—Since the middle of the first century b.c. the apex (or point) appears added to the vowels a, e, o, u to indicate their length. Long i was designated originally by I (rising above the other letters and hence called i longa) and by ei; later, ī took the apex. Examples are tráxi, CIL. x. 2311; príscvs, CIL. xi. 1940; ólla, CIL. vi. 10006; quInqve, CIL. vi. 3539; mIllia, Monumentum Ancyranum, i. 17; fecei, CIL. i. 551.

Before the employment of the apex the length of the vowel in case of a, e, u was indicated by doubling the vowel, e.g. paastores, CIL. i. 551; peqvlatvv, CIL. i. 202; o is never doubled in this manner. This peculiarity belongs to the period from 130 to 70 b.c.

A thoroughly consistent use of these methods of designating the vowel quantities is found, it must be admitted, in but few inscriptions. Of the vowels contained in syllables long by position only a portion are marked, as a rule, in any single inscription. Certain official inscriptions of the late republican and early imperial period form an exception to this, and exhibit very full and reliable markings, e.g. the speech of the Emperor Claudius (Boissieu, Inscriptions de Lyon, p. 136) and the Monumentum Ancyranum, containing the Rēs Gestae Dīvī Augustī. This latter, among a great number of correct markings, contains also some false ones, e.g. clv́pei, svmmá. Such errors also occur occasionally elsewhere.

{p. 38}

4. Greek transcriptions of Latin words.—This method is most fruitfully applied in case of the vowels e and o. The employment of Greek ε or η, ο or ω makes the quantity of the Latin vowel certain, wherever faith may be reposed in the accuracy of the transcription. Thus we may write Ēsquiliae in view of Ἠσκυλῖνος, Strabo, v. 234, 237; Vĕrgilius, after Οὐεργίλιος; Vesŏntiō, after Οὐεσοντίων, Dio Cassius, lxviii. 24.

The quantity of i may also often be determined by Greek transliterations. Thus ει before two consonants regularly points to Latin ī, e.g. Βείψανιος, CIG. 5709, = Vīpsānius; Greek ι points to Latin ĭ, e.g. Ἴστρος = Ister.

Inscriptions are naturally of much greater weight in such matters than are our texts of the Greek writers. Cf. § 3. c).

5. The vocalism of the Romance languages.—These languages, particularly the Spanish and Italian, treated e, i, o, u with great regularity according to the natural length of the vowel. It will be remembered that Latin ē and ō were close; Latin ĕ and ŏ open. Now the Romance languages have not preserved the original quantity of Latin vowels; for both the long and the short vowels of the Latin have become half-long in Romance; but they have very faithfully preserved their quality. Thus Latin ē appears as a close e in Italian and Spanish; Latin ĕ as an open e or as ie. Latin ō appears as a close o in Italian and Spanish; Latin ŏ as an open o or as uo (ue). Similarly Latin ī remained i, but ĭ became a close e; Latin ū remained u, but ŭ became close o. Examples:

mēnsis.mese (with close e).
honĕstus.onesto (with open e).
mōnstrāre.mostrare (with close o).
dŏctus.dotto (with open o).
dĭctus.detto (with close e).
dŭctus.-dotto (with close o).

{p. 39}

The Romance languages, however, authorize conclusions only with reference to the popular language as opposed to that of the better educated classes. In the popular speech the tendency was rather toward the shortening of long vowels than toward the lengthening of short ones. Hence where the Romance languages point to a long vowel in the popular language, it is safe to assume that the vowel was long in the literary language. When, on the other hand, the Romance languages point to a short vowel, this testimony is not necessarily conclusive, particularly if other facts point clearly in the opposite direction.

Again, the Romance languages authorize conclusions only in case of words inherited from the Latin. Many Romance words represent mediaeval borrowing by the learned class, as Italian rigido, cibo, metro, tenebre, pustula, lubrico. All such words retain the Latin vocalism. In some cases it is difficult to decide whether a word has descended by the popular or the learned channel, e.g. luxus, urna.

With all the assistance furnished by the methods above enumerated, there nevertheless remain some words whose vowel quantity cannot be determined. It is customary to regard all such vowels as short, until they are proved to be long.

The following are the most important works of reference on this subject:

Marx, Hülfsbüchlein für die Aussprache Lateinischer Vokale in Positionslangen Silben. 3d ed. Berlin, 1901. A work valuable for its collection of evidence, but frequently untrustworthy in its conclusions. {The 2nd edition of this book (which actually in the title has der lateinischen Vokale rather than Lateinischer Vokale) is available on Google books.}

Seelmann, Die Aussprache des Latein. Heilbronn, 1885. p. 69 ff.

Gröber, Vulgärlateinische Substrata Romanischer Wörter, a series of articles in Wölfflin’s Archiv für Lateinische Lexikographie, vols. i–vi.

Körting, Lateinisch-Romanisches Wörterbuch. 2d ed. Paderborn, 1901.

Lindsay, The Latin Language. Oxford 1894. p. 133 ff.

d’Ovidio, in Gröber’s Grundriss der Romanischen Philologie. Strassburg, 1888, i. p. 497 ff.

Meyer-Lübke, Grammatik der Romanischen Sprachen. Leipzig, 1890.

{p. 40}

Christiansen, De Apicibus et I Longis. Husum, 1889.

Eckinger, Orthographie Lateinischer Wörter in Griechischen Inschriften. Munich, 1891.

Heraeus, Beiträge zur Bestimmung der Quantität in Positionslangen Silben in Wölfflin’s Archiv für Lateinische Lexikographie, Vol. xiv. pp. 393 ff., 449 ff.

Further literature up to 1901 is cited by Marx, p. xiv f.

General principles for the determination of hidden quantity.

37. Vowels before ns, nf.

A vowel is always long before ns and nf, e.g. cōnsul, īnfēlīx. This principle rests upon the following evidence:

a) Cicero, Orator, 159, expressly states that in compounds of con and in, the vowel was pronounced long when followed by f or s.

b) Before ns the vowel is often marked in inscriptions with an apex, as CIL. xii. 3102 Cénsor; CIL. vi. 1527 d. 64 cónsto; CIL. xi. 1118 ménsvm; the apex occurs less frequently before nf, e.g. CIL. xi. 1118 cónficivnt. But i longa occurs repeatedly before both ns and nf, e.g. CIL. iii. 67 Inspexi; vi. 647 Instrvxervnt; CIL. ii. 4510 Inferioris; CIL. xiv. 1738 Infanti; CIL. x. 4294 Inferri.

c) Greek transliterations of Latin words often indicate a long vowel before ns, as Κρήσκηνς (= Crēscēns); Προύδηνς (= Prūdēns).

38. Vowels before gn, gm.

Until recently the doctrine was current that all vowels are long in Latin before gn. In the Appendix to my Latin Grammar, I showed that this general principle was altogether too sweeping and that at most we could go no farther than to recognize with Priscian the length of the vowel before the suffixes -gnus, -gna, -gnum and in such other individual words as may be sup-{p. 41}ported by specific evidence. Admitting the validity of Priscian’s testimony for the length of the vowel before -gnus, -gna, -gnum, I showed that there was certainly no evidence to support the doctrine of Marx (see his Hülfsbüchlein, p. 1) that the vowel is long before gn in gignō, agnōscō, agnātus, cognōscō, cognātus, ignārus, ignāvus, ignōrō, ignōscō, etc. Marx holds that the vowel in these latter forms was long as the result of compensatory lengthening, ignārus being for *in-gnārus, cognōscō for *con-gnōscō. But no such theory of compensatory lengthening is tenable. Marx’s appeal (p. 1) to the fact that Plautus always uses the syllable before gn as long, is of no weight, since we should naturally except gn to ‘make position’ in Latin just as γν regularly does in Greek.

But there has been a growing tendency in recent years to reject even Priscian’s testimony in favor of the length of the vowel before the suffixes -gnus, -gna, -gnum. The passage is found in Keil, Vol. ii. p. 82: “Gnus” quoque vel “gna” vel “gnum” terminantia longam habent vocalem paenultimam, ut “rēgnum,” “stāgnum,” “benīgnus,” “malīgnus,” “abiēgnus,” “prīvīgnus,” “Pelīgnus.” Some scholars, as Havet, regard this statement as an interpolation. Others, while admitting the genuineness of the passage, impugn its correctness. Buck (Classical Review, Vol. xv. p. 311, ff.) has discussed the question here at issue with great thoroughness and candor, and urges (p. 312) against the long vowel before -gnus, -gna, -gnum: “(1) the fact that, except in words with an original long vowel, the Romance languages point to a short vowel before gn; (2) the fact that the Celtic and Germanic words borrowed from Latin signum also point to a short vowel; (3) the total absence on inscriptions of the apex or I longa in the case of the great majority of words with gn, some of them, like magnus, of so frequent occurrence that this absence can hardly be accidental; (4) the citation of dignitas as an anapaest by Diomedes (Keil, Vol. i. p. 470), who {[…]} has in mind {[…]} only vowel {p. 42} quantity, not syllabic quantity.” Nevertheless certain words of this class seem occasionally (in special localities, perhaps, or in special social strata) to have had a long vowel before gn, as seen in sIgnum, CIL. vi. 10234; seignvm, xiv. 4270; sIgnificabo, vi. 16664; dIgni, x. 5676; privIgno, vi. 3541; Ignis, xi. 826. But these Buck regards as abnormal and exceptional pronunciations. Buck’s argument is a very strong one, and his conclusions deserve at least provisional acceptance. It should be noted, however, that three words, rēgnum, stāgnum, abiēgnus, being derived from stems with a long vowel, were legitimately entitled to their long quantity and always retained it.


Before gm the vowel is long in pīgmentum (see CIL. viii. 1344, pIgmen[t) and in sēgmentum (cf. Greek σηγμέντα), but there is no evidence warranting the formulation of a broad rule embracing all vowels before gm, as is done by Marx (p. 1). Marx appeals to the analogy of gn in support of his attitude; but apart from the dangers of this kind of reasoning, we have already seen that the case for vowel length before gn is of the weakest possible kind, so that, even if we admit the validity of the analogy, there is nothing to indicate regular vowel length before gm.

40. Vowels before nt, nd, ss.

1. All vowels are regularly short before nt and nd, e.g. amandus, montis, amant, monent.

2. Exceptions:

  1. Before nt the vowel is long in
    1. quīntus.
    2. the following contracted words: cōntiō (for coventiō), jēntāculum (for *jējūntāculūm), jēntātiō (for *jējūntātiō), nūntius (for *noventius).
    3. Greek proper names in -ūs, Gen. -ūntis, e.g. Selīnūs, Selīnūntis (Greek, Σελινοῦντος). {p. 43}
    4. Greek proper names in -ōn, Gen. -ōntis, e.g. Xenophōn, Xenophōntis (Greek, Ξενοφῶντος).
  2. Before nd the vowel is long in
    1. the following contracts and compounds: prēndō (for prehendō), nōndum (nōn + dum), vēndō (vēnumdō), nūndinus (novem diēs), quīndecim (qīnque), ūndecim (ūnus).
    2. some Greek names, e.g. Charōndās, Epamīnōndās (-ώνδας).

3. The evidence for the short vowel before nt lies in the fact that, while in the Nominatives of such words as clēmēns, crēscēns, cliēns, fōns, gēns, parēns, pōns, praesēns, the long quantity of the vowel is assured either by the presence of the apex, or by a long vowel in Greek transcriptions, in the oblique cases the apex is lacking, and in Greek transcriptions the vowel is short, e.g. Κλήμης (i.e. Κλήμηνς), CIA. iii. 1094, but Κλήμεντος, CIG. 3757; Κλήμεντι, CIG. Addenda, 1829 c.; créscéns, CIL. xii. 4030, but créscenti, CIL. vi. 9059; Κρήσκηνς, CIG. 6012, c.; but Κρήσκεντι, CIG. Addenda, 1994, f.; Πραίσης (i.e. Πραίσηνς), CIA. iii. 1147, but Πραίσεντι, Πραίσεντα, CIG. 3175, 3991.

Even where a vowel is naturally long, it sometimes becomes shortened before nt, e.g. in linteum from līnum; cf. Greek λέντιον, CIG. 8695.

For the vowel before nd the evidence is not so full. We find the Greek transcriptions Καλένδαις, Lydus, de Mens. iv. 53, 57; Φονδάνιος (i.e. Fundānius), Bulletin de Correspondance Hellénique, ix. p. 439.

4. Vowels are also regularly short before ss, according to the express testimony of Quintilian, i. 7. 20. But see § 47. 1.

41. Pontem, fontem, montem, frontem, frondem.

A slight uncertainty exists as to the quantity of the vowel before nt in the oblique cases of fōns, mōns, pōns, frōns {p. 44} (frontis); and before nd in frōns (frondis). Three sets of facts are to be considered:

a) The analogy of other words in -ns (Gen. -ntis). Such words, so far as they are genuine Latin words, have, without exception, a short vowel before nt in the oblique cases. See § 40.

b) The testimony of the Romance languages. THis is as follows for the different words under discussion:

fōns. The romance languages seem to point to an antecedent fōntis, fōnti, etc. Thus the Italian fonte has a close o; so the Provençal fon. Spanish alone with its fuente points to fŏntem (Gröber, Archiv, ii. p. 426; Körting, Lat.-Romanisches Wörterbuch).

frōns (-ndis). The Romance languages all agree in pointing to frōndem (Gröber, Archiv, ii. p. 426; Körting, Wörterbuch).

frōns (-ntis). Provençal fron and Italian fronte, with close o, point to frōntem. So the other Romance languages, except Spanish, which has fruente, pointing to frŏntem. (Gröber, Archiv, ii. p. 426; Körting, Wörterbuch).

mōns. The Romance languages point unanimously to mōntem (Gröber, Archiv, ii. p. 426; Körting, Wörterbuch).

pōns. Provençal pon and Italian ponte with close o point to pōntem; so the other Romance languages, except Spanish, which has puente, pointing to pŏntem.

If mere numerical preponderance were decisive, we might at once conclude that all these words went back to Latin forms with ō in the oblique cases, and might explain Spanish fruente, fuente, puente (which should be fronte, fonte, ponte, to represent Latin ō) as exceptions to the prevailing law of development. A glance at certain facts, however, in Italian and Provençal, suggests another conclusion. We find it to be a regular law in these languages that an original open Latin o (i.e. short o, see § 36. 5), when followed by m, n, or l, + another consonant, becomes close. Thus Latin tŏndet with open o, becomes Italian {p. 45} tonde, with close o. Similarly respŏndet becomes risponde; rhŏmbus becomes rombo; pŏl(y)pus becomes polpo, all with close o. Just what has brought about this change is not certain. D’Ovidio in Gröber’s Grundriss der Romanischen Philologie, i. p. 522, thinks it was the analogy of words in on + consonant, om + consonant, and ol + consonant, in which close o had developed regularly from an earlier ŭ (see § 36. 5), e.g. rompe (= rumpit); onda (= unda); dolce (= dulcis). In accordance with this principle, whose operation is certain, Latin fŏntem, frŏndem, frŏntem, mŏntem, pŏntem, would (assuming these to be the original forms) regularly become in Italian: fonte, fronde, fronte, monte, ponte, with close o, exactly as we find them. The admission of a long o in the oblique cases of these Latin words is, therefore, not necessary in order to account for Italian and Provençal close o in their Romance descendants. In fact, when we consider Spanish fuente, fruente, puente, all of which point to Latin ŏ, it seems more reasonable to regard Spanish monte and fronde (which point to ō) as the exceptions. Gröber, who (Archiv, vi. p. 389) expresses himself in favor of assuming an original fŏntem, etc., in these words, suggests that Spanish monte, fronde, are loan-words, while fuente, fruente, puente represent an original inheritance.

Briefly, then, a fair interpretation of the evidence of the Romance languages seems to warrant the belief that the oblique cases of the words under discussion came into the Romance languages from the Latin with a (short) open o; that in Italian and Provençal this open o subsequently became close in accordance with a regular law of wide operation. Spanish regularly developed the open o to ue in those words which it inherited from Latin (viz. in fuente, fruente, puente); while Spanish monte and fronde are probably loan-words from Italian.

c) The third bit of evidence comes from Greek transliterations of Latin words as found in Greek inscriptions and Greek authors. {p. 46} Thus we find Φοντήιος (= Fŏnteius) in Plutarch and Appian; also in an inscription, CIG. iii. 5837, b (59 a.d.); Φρόντων, CIA. iii. 1113, 21, 26 (before 161 a.d.), and in texts; all of which point to Latin Frŏntō, and Frŏntīnus, and indirectly to frŏnt-em. Latin Montānus appears as Μοντανός, CIG. Addenda, 4805 b; and we find τριμόντιον, Ptol. iii. 11, 12, et passim; πόντεμ (= Latin pŏntem) is the text in Plutarch, Numa, 9; ποντίφιξ (= pŏntifex), in Dionysius, Dio Cassius, and Zosimus; ποντίφεξ, in Lydus, de Mens. iii. 21; ποντίφικες, in Plutarch, Numa, 9; and ποντίφικα, in an inscription in Kaibel’s Sylloge Epigrammatum, Addenda, 888 a. The Greek never shows an ω in any of these words, either in inscriptions or in Mss. The evidence furnished by that language therefore is unanimous in favor of ŏ for the Latin. Nor can recognition be refused the inscriptions above cited on the ground that they are late. As the annexed dates show, they all belong to the good period of the language.

We thus have the strongest possible grounds for writing fŏntis, frŏndis, etc. The analogy of other words in -ns (Gen. -ntis) favors this view; the Romance languages favor it, and the testimony of Latin words in Greek dress, as exhibited both in texts and inscriptions, favors it. In fact, the evidence is complete.

The isolates apex in Frónt (for fróntem, as the context shows), CIL. v. 2915, is certainly a mere blunder of the stone-cutter, as is often the case in other words, even in carefully cut inscriptions (see § 36. 3. Christiansen, De Apicibus et I Longis, p. 57, cites thirteen such instances for vowels before nt.

42. Hidden Quantity in Declensions.

1. It is maintained by some scholars (e.g. Marx, Hülfsbüchlein, p. 2; Lane, Harvard Studies, i. p. 89) that the ending -um in the Genitive Plural of nouns of the First and Second {p. 47} Declensions has ū in such forms as Aeneadum, deum, nummum; also in nostrum and vestrum. The facts in evidence are the following:

a) On early Latin coins prior to the First Punic War, we find the final m of many Genitives Plural omitted, e.g. Romano, Corano. Coins of the same date regularly retain final m of the Nominative or Accusative Singular, e.g. Volcanom, propom (= probum). This has led Mommsen (CIL. i. p. 9) to infer that there was a difference in the quantity of the o in the two instances. As the o of the Nominative and Accusative Singular was short, Mommsen thought that in the Genitive Plural it must be long. But the material with which Mommsen deals is extremely scanty. Genitive Plural forms occur in some number; but only a few Nominative and Accusative forms are found, viz. Volcanom, propom. Again, Romanom (CIL. i. 1) and Aeserninom (i. 20) show that Genitives sometimes retained the m. Mommsen attempts to solve this difficulty by taking Romanom and Aeserninom as the Nominative Singular Neuter of the Adjective; but that is awkward. The natural inference must be that there was no system in the omission of final m on these coins. The coins represent no dialect; in fact they represent widely separated localities; hence it is no wonder if the final m (always weak) was sometimes written, sometimes omitted. In the Scipio inscriptions, the oldest of which may date within a quarter of a century of these coins, we find final m freely omitted in the Accusative and Nominative Singular just as elsewhere. It is, therefore, extremely unlikely that Mommsen’s hypothesis concerning the coins is correct.

b) An inscription of Nuceria (CIL. x. 1081) has dvv́mviratvs, which Schmitz (Rheinisches Museum, x. 110) and Lane (Harvard Studies, i. p. 89) regard as evidence that the u of duum (Gen. Pl. of duo) was long. But even conceding the correctness of the apex in this isolated instance, it remains to be shown that the {p. 48} duum- of duumvir and duumvirātus is in origin a Genitive. Such an etymology would involve the assumption that the duum- of the Genitive Plural, duumvirum, became transferred to the other cases, replacing duo in earlier duovirī, etc. Such an assumption is extremely improbable. It is much more likely that duumvir and triumvir are formed after the analogy of centumvir. In the singular especially such forms as duovir, trēsvir would have been extremely awkward, and it seems probable that the singular duumvir, triumvir were for that reason historically anterior to duumvirī, triumvirī. The apex in the Nucerian inscription, if this etyomology be correct, would then be simply a blunder of the engraver, as is altogether probable. The evidence in favor of -ūm in these Genitives must, therefore, be regarded as of no weight, especially in view of the regular shortening of vowels before final -m in Latin. Certainly if -ūm did by any possibility exist in the days of Augustus, the u had become shortened by 90 a.d. For Quintilian (i. 6. 18), as noted by Lane (p. 90), shows that to his ear nummum, Genitive Plural, was nowise different from nummum, Accusative Singular.

2. Words in -er of the Second Declension, and words of the Third Declension in -er and -x, have in oblique cases the same quantity of the vowel as in the Nominative, e.g. ăger, ăgrī; frāter, frātris; ācer, ācris; pāx, pācis; tenāx, tenācis; făx, făcis; rēx, rēgis; nĭx, nĭvis; cornīx, cornīcis, calĭx, calĭcis; fel, fellis, ŏs, ŏssis; plēbs, plēbis. Thus sometimes the Nominative gives the clue to the hidden quantity in the oblique cases (as ăger, ăgrī); sometimes the oblique cases give the clue to the hidden quantity of the Nominative (as cornīcis, cornīx).

3. Words of the Third Declension ending in -ns (Gen. -ntis) uniformly have a short vowel in the oblique cases, as already explained in § 40. 3. Greek words in -ās (Gen. -antis), e.g. Aiās, Aiantis; gigās, gigantis, have the same quantity as in the original (Αἴᾱς, Αἴᾰντος; γιγᾱς, γίγᾰντος). So, also, contracted {p. 49} Greek names of cities in -οῦς, -οῦντος, e.g. Selīnūs, Selīnūntis; and proper names in -ῶν, -ῶντος, e.g. Xenophōn, Xenophōntis. Acheron (not a contract form) has Acherŭntis.

4. In all words of the Third Declension ending in two or more consonants (excepting -ns and -x preceded by a vowel), the hidden vowel before the ending is short, e.g. ŭrbs, sŏrs, ărx. Exceptions to this principle are plēbs and compounds of ūncia ending in -ūnx, e.g. deūnx, deūncis; quīncūnx, quīncūncis. Before -x the vowel is sometimes long, sometimes short, as already explained in 2, above.

43. Comparision of Adjectives.

In the terminations -issimus, -errimus, -illimus, the hidden vowel is short, e.g. carĭssimus, acĕrrimus, facĭllimus. Apparent traces of a long i in the termination -issimus are found in inscriptional forms with i longa. The word of most frequent occurrence is piIssimus; besides this we find a few other words, e.g. carIssimo, CIL. vi. 5325; dvlcIssimo, vi. 16926; fortIssimo, vi. 1132. But many of these inscriptions belong to the last centuries of the Empire, when the use of i longa had become an extremely untrustworthy guide, as may be seen by palpable errors. As regards the frequent occurrence of piIssimae, piIssimo, these may perhaps be explained on the theory that i longa was here used to indicate not merely i, but also the j which developed in pronunciation between the two i’s, i.e. pijissimo. Cf. the similar use of i longa in words like PompeIivs, CIL. ix. 3748. At all events, in the absence of the apex in these superlatives, and in view of the absolute silence of the grammarians, it seems unwise to attach great weight to the occurrence of the i longa alone. Against ī, Lindsay (Latin Language, p. 405) urges the occurrence of late spellings like merentessemo, karessemo, CIL. ii. 2997. Cf. § 6. 1.

{p. 50}

44. Numerals.

As separate words are to be noted:

a) quăttuor, but quārtus (see § 53 under arca).

b) quīnque and its derivatives, all of which have ī, as quīndecim, quīntus, quīngenti, quīnquāgintā.

c) the derivatives of ūnus: ūndecim, ūndēvīginī, etc.

d) mīlle, mīllie, and mīllēsimus.

45. Pronouns.

1. Nōs, vōs; but nŏster, vĕster; nostrī, vestrī, etc.

2. Hunc and hanc have a short vowel.

3. Ille, ipse, iste have ĭ.

4. The suffix -cunque has ŭ.

5. Compounds retain the quantity of the elements of which they are compounded, as quĭsquis, cūjŭsque {= cŭjjŭsque}.


46. Root Forms.

1. Presents formed by means of the infix n have a short vowel, e.g. fŭndō (root fud-); frăngō (root frag-); jŭngō (root jug-). Before a labial, n becomes m, e.g. rumpō (root rup-); lambō (root lab-). Care should be taken not to confuse derivative and contract Presents like vēndō, prēndō, with genuine nasal formations.

2. In most Presents the hidden vowel is short, e.g. nectō, serpō, vertō. But the following exceptions are to be noted:

a) First conjugation: jūrgō (for jūrigō), nārrō, ōrnō, pūrgō, trāctō.

b) Second Conjugation: ārdeō.

c) Third Conjugation: all verbs in -scō(r), except compescō, discō, poscō, vescor.

d) Fourth Conjugation: nūtriō, ōrdior. {p. 51}

3. The quantity of the vowel in the Present regularly remains unchanged (when it becomes hidden) throughout the entire conjugation of the verbs, e.g.:


Thus inscriptions give fIxa, scrIptvm, conscreiptvm, vIxit, veixit.

But the following exceptions to this general principle are to be noted:



The short vowel of the Perfect Participles dĭctus and dŭctus is assured by the statement of Aulus Gellius (Noctes Atticae, ix. 6) and by the testimony of the Romance languages. (See § 52. s.vv.)

b) The short vowel of the Present is lengthened in the Perfect Indicative and Perfect Participle, if hidden, in the following verbs:

dēlinquōdēlinqueredēlīquīdēlīctus{Cf. relinquo.}
fingōfingerefīnxīfīctus{Allen: fĭctus.}
fungorfungīfūnctus sum
pingōpingerepīnxīpīctus{Allen: pĭctus.} {p. 52}
relinquōrelinquererelīquīrelīctus{Allen: relĭctus.}
trahōtraheretrāxītrāctus{Allen: trăctus.}

So also in compounds and derivatives of these verbs.

4. The evidence for the long vowel in the Perfect Participles of the foregoing list is found:

a) In the statements of Gellius, who testifies (Noctes Atticae, ix. 6) to the quantity of the vowels of āctus, lēctus, ūnctus, and in xii. 3. 4 to that of strūctus.

b) In the testimony of inscriptions, which show the following: Áctis CIL. vi. 1377; redácta vi. 701; fxáctvs {sic in Bennet} Boissieu, Inscriptions de Lyon, p. 136; cInctvs CIL. x. 4104; défv́nctis CIL. v. 1326; dIléctvs vi. 6319; léctvs xi. 1826; exstInctos vi. 25617; infráctá ix. 60; iv́ncta x. 1888; seiv́nctvm vi. 1527 e. 38; récte xii. 2494; téctor vi. 5205; coémto Monumentum Ancyranum iii. 11; trá[cta] (not certain) CIL. vi. 1527 e. 14; sáncta v. 2681; Oscan saa(n)htom (= sānctom). {Michelson: The Oscan word should be written saahtúm.}

c) In the retention of a in compounds of actus, tactus, fractus, pactus, tractus (e.g. coactus, attactus, refractus, etc.), which shows that the a was long; short a would have become e in this situation, as for example in cōnfectus for an original *cōnfăctus; acceptus for an original *accăptus; ēreptus for *ērăptus.

d) For cīnctus, delīctus, distīnctus, exstīnctus, fīctus, pīctus, pūnctus, relīctus, tīnctus, the long vowel is assured by the evidence of the Romance, e.g. Italian cinto, delitto, fitto, relitto, tinto.

5. The evidence for the quantity of the vowel in the Perfects of the foregoing list is found: {p. 53}

a) In inscriptional markings, as coniv́nxit (Wilmanns, Inscript. Latinae 104); téxit (CIL. x. 1793); réxit (CIL. v. 875); tráxi (CIL. x. 2311, 18).

b) In Priscian’s statement (Keil, ii. 466) that rexi and texi have ē.

c) In the testimony of the Romance languages, which point to cīnxī, distīnxī, exstīnxī, fīnxī, pīnxī, strūxi, tīnxī, ūnxī.

d) The long ā in sānxī rests upon no specific evidence, but may perhaps be safely inferred after the analogy of sānctus.

Until recently the principle was maintained (e.g. by Marx in his first edition) that all monosyllabic stems ending in b, d, or g had the hidden vowel long in the Perfect Indicative and Perfect Participle wherever euphonic changes occured. According to this theory we should have e.g. scindō, scindere, scidī, scīssus; mĕrgō, mĕrgere, mērsi, mērsus. This principle was first laid down by Lachmann (on Lucretius, i. 805) for Perfect Participles alone, and was subsequently assumed by other scholars to apply to the Perfect Indicative as well; but this position is now entirely abandoned. Each long vowel must be supported by specific evidence.

In the 3d edition of his Hülfsbüchlein (p. 1), Marx lays down the principle that all vowels are long in Latin before nx and nct. These combinations occur almost exclusively in the verbs given on pp. 51, 52. Whether the general principle is sound, may be questioned. For example, we have no definite evidence in favor of the long vowel before nx in anxius, lanx, or phalanx. {Michelson: That a short vowel is lengthened before nct is practically universally accepted. Allen does not question it.}

47. Verbal Endings.

1. The hidden vowel is short before ss (§ 40. 4) and st in the terminations of the verb, e.g. fuĭssem, amāvĭsse; fŭistī, fuĭstis. This is shown not only by the historical origin of these formations, but by such metrical usage as Plautus, Amphitruo, 761, dedĭsse; Menaechmi, 687, dedĭstī, where iss and ist are treated as short {p. 54} syllables by neglect of ‘position’ (see § 36. 2). Contracted forms are, of course, an exception to the above principle, as amāsse, commōssem, redīsse, audīsset, amāstī, nōstis.

2. Formations of the type: dīxtī (for dīxistī), accĕstis (for accessistis), jŭstī (for jussistī), trāxe, surrēxe, exstīnxem, etc., have the same quantity as the regular forms.

48. Compounds.

Marx (p. 8) holds that the vowel of a monosyllabic preposition, if hidden, is long in composition when the preposition loses a final consonant. Thus he maintains a long vowel for the intial syllable of ascendō (for *ad-scandō); di-stinguō (*dis-stinguō); suspiciō (for *sub-spiciō). But this principle rests upon an untenable theory of compensatory lengthening; see § 89.

49. Inchoatives.

Inchoatives in -scō, -scor have a long vowel before -sc, e.g. labāscō, flōrēscō, nitēscō, tremīscō, adipīscor. Gellius (Noctes Atticae, vi. 15) mentions a number of words of this class as having a long vowel, and implies that this was generally true of all. The Romance languages show that -scō and -iscō (-iscor) had ē and ī. But the hidden vowel is short in compescō, discō, poscō, vescor. {Allen lists mĭsceō as well, (but omits vescor.)}

50. Irregular Verbs.

1. The root vowel of esse is short under all circumstances, e.g. ĕst, ĕstis, ĕstō, ĕssem.

2. Edō, ‘eat,’ has a long e in the forms ēs, ēst, ēstis, ēssem, ēsse, ēstur, ēssētur. Cf. Donatus on Terence, Andria, 81; Servius on Virgil, Aeneid, v. 785.

3. Marx (p. 9) lays down the principle that in compounds of eō, forms containing ii have the second i long before st, as e.g. in {p. 55} interiistī. This theory rests solely upon the occurrence of interieisti in CIL. i. 1202. But ei occurs elsewhere in inscriptions, incorrectly written for ĭ, e.g. parenteis (= parentĭs), CIL. i. 1009; faceivndae (= facĭundae). It is altogether probable that interieisti is another instance of the same sort.

51. Word Formation.

1. Substantives in -abrum, -acrum, -atrum, derived from verbs, have ā, e.g. flābrum, lavācrum, arātrum.

2. The derivative endings -ellus (a, um), -illus (a, um), regularly have ĕ and ĭ, but the following have a long vowel, viz.: catēlla, ‘little chain,’ anguīlla, Bovīllae, hīllae, ovīllus, stīlla, suīllus, vīlla.

3. The vowel is short in -ernus (-ernius, -ernīnus), -urnus (-urnius, -urnīnus), e.g. hībĕrnus, tabĕrna, Sātŭrnus. In vērnus (from vēr) the r is not part of the suffix.

4. The vowel is short in the endings -estus (-ester, -estris, -esticus, -estās), -ister (-istrum), -ustus, e.g. caelĕstis, domĕsticus, tempĕstās, capĭstrum, venŭstus. In sēmēstris, jūstus, the long vowel belongs to the stem.

5. The vowel is short in the endings -unculus, -unciō, -erculus, -usculis, e.g. ratiuncula, paterculus, maiusculus, homunciō; plūsculus (from plūs) naturally has ū.

6. In compounds, the connecting vowel i is short, e.g. nāvĭfragus, lectĭsternium.

{p. 56}

52. List of the Most Important Words containing a Long Vowel before Two Consonants.2


abiēgnus: see § 38, end.

acatalēctus: Gr. ἀκατάληκτος.

āctiō, āctitō, āctor: see agō.

āctūtum: like āctus.

adēmptiō: see adimō.

adimō, adēmptus: like emō.

afflīctō: like flīctus.

Āfrica, Āfrī: from Āfer.

agō, agere, ēgī, āctus: see § 46. 3. b).

Alcēstis: Gr. Ἄλκηστις.

Ālēctō: Gr. Ἀληκτώ.

aliōrsum: for *alio-vorsum.

alīptēs: Gr. ἀλείπτης.

Amāzōn: Gr. Ἀμᾱ́ζων.

ambūstus: see ūrō.

Amsānctus: see sānctus.

anguīlla: ī acc. to the Romance.

Aquīllius: AqvIllivs: CIL. vi. 12264.

arātrum: see § 51. 1.

ārdeō, -ēre, ārsī, ārsūrus: like āridus, ārdus. {Michelson: in as much as Lindsay, Sommer, and Brugmann (Grundriss I2, §§ 240, 2), 930; K. Verg. Gr., § 310 but curiously ārdus in § 346) consider the initial a as short, it would have been well to have placed the word in § 53—the list of words in which the quantity of a vowel before two consonant is in dispute—and there to have discussed the arguments for and against ă-.}

Arginūssae: Gr. Ἀργινοῦσσαι.

āspernor: from ā and spernor.

{āstutus, Allen.}

āthla: Gr. ἆθλον.

āthlētēs: Gr. ᾱ̓θλητής.

ātrāmentum: like āter.

ātrium: from āter; also átrivm, CIL. vi. 10025.

āxilla: Priscian, iii. 36.


bārdus,: ‘stupid’: from bārō.

Bēdriacum: Βητριακόν, Plutarch, Otho, 8, 11.

bēllua: for bēs-lua

bēstia, Bēstia: Βητρίας; Plutarch, Marius, 9; Cicero, 3; the Romance would point to ĕ.

bifōrmis: see fōrma.

bilībris: like lībra.

bimēstris: from mēnsis.

bovīllus: from bovīnus.

būbrēstis: Gr. βούβρηστις.

būrrus: ū acc. to the Romance.

būstum: ū acc. to the Romance; cf. combūstus and ūstus.

Būthrōtum: Βούθρωτον.


catalēctus: Gr. κατάληκτος.

catēlla: from catēna; catella, ‘bitch,’ has ĕ.

catīllus: from catīnus.

cētra: better orthography is caetra; see § 61.

chīrūrgus: Gr. χειρουργός.

cicātrīx: ā in Plautus, Amphitruo, 446; see § 36. 2.

cīccus, -um: ī acc. to the Romance.

{p. 57}

Cīncius: cIncia, CIL. vi. 14817 et passim.

cingō, cingere, cīnxī, cīnctus: ī in the Perfect and Perfect Participle acc. to the Romance: see Körting (Wörterbuch): d’Ovidio (Gröber’s Grundriss, i. p. 501 f.); cInctvs, CIL. x. 4104; see § 53 s. v.

clandēstīnus: from clam and dēs(i)tus from dēsinō (cf. positus from pōnō, i.e. po-sinō); hence ‘secretly put aside.’

clātra, clātrī: Gr. κλῇθρα.

Clytēmēstra: Gr. Κλυταιμήστρα.

Cnōssus: Gr. Κνωσσός.

cōgō, cōgere, coēgī, coāctus: see agō.

combūrō, combūrere, combussī, combūstus: see ūrō and būstum. Marx also markse the o long, regarding cōmbūrō as for co-ambūrō, and comparing cōgitō (for co-agitō).

comedō, comēstus: cf. edō; see § 50. 2.

cōmō, cōmere, cōmpsī, cōmptus: ō acc. to the Romance.

compingō, compingere, compēgī, compāctus: see § 46. 3. b).

conjūnx: coniv́nx, CIL. vi. 6592 et passim; but conjux has ŭ.

contingō, -ere, contigī, contāctus: like tangō.

cōntiō: for co-ventiō: § 40. 2. a).

corōlla: from corōna.

crābrō: ā in Plautus, Amphitruo, 707; see § 36. 2.

crāstinus: from crās.

crēscō: créscéns, CIL. xii. 4030 et passim; Gr. Κρήσκηνς; also acc. to the Romance.

Crēssa: Κρῆσσα.

crībrum: ī in Plautus, Mostellaria, 55; see § 36. 2.

crīspus: Creispinvs, CIL. x. 3514. Κρεισπεῖνον, CIG. Addenda, 4342, d. 4. The Romance would point to ĭ; but see § 36. 5 fin.

crūsta, crūstum: in CIL. i. 1199; the Romance points both to crūstum and also to a collateral form with ŭ. Gröber (Archiv, vi. 384); Körting (Wörterbuch).

Ctēsiphōn, -ōntis: Gr. -ῶν, -ῶντος.

cucūllus, ‘hood’: the Romance points to two forms,—one with ū, another with ŭ; see Gröber (Archiv, i. 555; vi. 384); Körting (Wörterbuch); cucullus, ‘cucckoo,’ has ŭ.

cūnctus: cv́ncti, CIL. ix. 60.

cūstōs: Κουστώδης, Lydus, de Magistratibus, i. 46; ū acc. to the Romance.

Cyclōps: Gr. Κύκλωψ.


dēligō, -ere, dēlēgī, dēlēctus: like legō.

dēlinquō, -ere, dēlīquī, dēlīctus: ī acc. to the Romance. {Cf. relinquo.}

dēlūbrum: ū in Plautus, Poenulus, 1175; see § 36. 2.

dēmō, dēmere, dēmpsī, dēmptus: like emō.

deūnx: from and ūncia.

dēxtāns: from + sextāns.

dīcō, dīcere, dīxī, dictus: see § 46. 3. a). Certain of the Romance languages (Fr. dit; Old Ital. ditto, etc.) point to a collateral dīctus, which Osthoff (Morphologische Untersuchungen, iv. 74) thinks belonged to the colloquial language. But possible those Romance languages which point to ī have simply adapted the Participle to the vowel of the Present and the Perfect. See Gröber (Archiv, vi. 385).

{p. 58}

dīctērium: Gr. δεικτήριον.

Diēspiter: for diēs and pater.

dīgladior: for dis + gladior by compensatory lengthening; see § 89.

dīgredior: for dis + gradior by compensatory lengthening; see § 89.

dilēmma: Gr. διλῆμμα.

dīligō, -ere, dīlēxī, dīlēxtus: like legō.

dīrigō, -ere, dīrēxī, dīrēctus: like regō.

dirimō, -ere, dirēmī, dirēmptus: like emō.

distinguō, -ere, distīnxī, distīnctus: ī acc. to the Romance; see d’Ovidio (Gröber’s Grundriss,, i. p. 502); Körting (Wörterbuch); cf. exstinguō; see § 46. 3. b).

dolābra: cf. § 51.1.

dūcō, dūcere, dūxī, dŭctus: see § 46. 3. a); perdv́xit, CIL. xii. 2346 et passim.


ēbrius: ē regularly in Platusu, e.g. Trinummus, 812; see § 36. 2.

eclīpsis: Gr. ἔκλειψις.

edō, ‘eat’: ēst, ēstis, ēsse, etc. See § 50. 2.

effringō, -ere, effrēgī, effrāctus: like frangō.

emō, emere, ēmī, ēmptus: see § 46. 3. b).

ēmungō, -ere, —, ēmūnctus: ū acc. to the Romance; see d’Ovidio (Gröber’s Grundriss, i. p. 515).

ērigō, -ere, ērēxī, ērēctus: like regō.

ēsca: ē acc. to the Romance.

Ēsquiliae, Ēsquilīnus: Gr. Ἠσκυλῖνος, in Strabo, v. 234, 237.

Etrūscus: cf. Etrūria; Gr. Ἐτροῦσκος.

exīstimō: from ex and aestimō; exIstimavervnt, CIL. v. 5050.

exōrdium: from ōrdior.

exstinguō, -ere, exstīnxī, exstīnctus: extInctos, CIL. vi. 25617; cf. distinguō; see § 46. 3. b).

extraōrdinārius: from ōrdō.


fāstus, a, um; cf. fās.

favīlla: favIlla, CIL. v. 3143. The Romance also seems to point to ī.

fēllō: from same root as fēmina; Gr. θῆλυς.

fēstīvus: from fēstus.

fēstus: from the same root as fēriae (= *fēs-iae), ‘holiday’; féstvs in CIL. i., Fasti Praenestini for April 25th. So also in the proper name:

Fēstus: Féstvs, CIL. xii. 3179; Fésti, v. 2627; Féstae, iii. 5353; Gr. Φῆστος, CIA. iii. 635 and frequently. The Romance points to ĕ, indicating that ē of the classical period ultimately became reduced; see § 36. 5.

fīgō, fīgere, fīxī, fīxus: fIxa, Monumentum Ancyranum, vi. 18; ī acc. to the Romance.

fingō, fingere, fīnxī, fīctus: ī acc. to the Romance; see § 53 s. v. {Allen: fĭctus.}

fīrmus: fIrmvm, CIL. iv. 175 et passim; the Romance points to ĭ, showing that ī of the classical period had become reduced; see § 36. 5. {Michelson: The truth is, fĭrmus and fīrmus are on a par with dĭgnus and dīgnus; that is, in certain circles or strata of society short vowels were lengthened before r followed by a consonant. Cf. Sommer, 1. c., p. 135; Brugmann, K. Verg. Gr., p. 219; and Mr. Bennett in § 53 under arca.}

flābrum: see § 51. 1.

flīgō, -ere, flīxī, flīctus: afleicta, CIL. i. 1175; the Romance also points to ī.

flōsculus: from flōs.

flūctus: ū acc. to the Romance.

fluō, -ere, flūxī; ū is probably long in flūxī in view of flūxus.

flūxus: ū acc. to the Romance.

{p. 59}

fōrma: see Donatus on Terence, Phormio, 28; φώρμη in Greek inscriptions; the Romance also shows ō.

fōrmula: from fōrma.

frangō, -ere, frēgī, frāctus: see § 46. 3. b).

frīgeō, -ēre, frīxī: § 46. 3.

frīgō, -ere, frīxī, frīctus: ī acc. to the Romance.

frūctus: ū acc. to the Romance. Old French froit points to a collateral frŭctus; see Osthoff, Geschichte des Perfects, p. 523.

fruor, fruī, frūctus sum: ū acc. to the Romance.

frūstrā: frv́strá, CIL. vi. 20370.

frūstum: ū acc. to the Romance.

fungor, fungī, fūnctus sum: defv́nctis, CIL. v. 1326; fv́ncto, xii. 3176 et passim.

fūrtim: from fūr.

fūrtīvus: from fūr.

fūrtum: from fūr.

fūstis: ū acc. to the Romance.


geōgraphia: Gr. γεωγραφία.

Geōrgius: Gr. Γεώργιος.

geōrgicus: Gr. γεωργικός.

glīscō: § 49.

glōssārium: from Gr. γλῶσσα.

glōssēma: from Gr. γλωσσῆμα.

grȳllus: ȳ acc. to the Romance.

grȳps: like Gen. grȳpis; § 42. 2.


hāctenus: like hāc.

Hellēspontus: Gr. Ἑλλήσποντος.

Hērculānum: Hércvlániae, CIL. xii. 1357; Ἡρκουλάνεον, Dio Cassius, lxvi. 23; Ἡρκανός, CIA. iii. 1197.

hibīscum: ī acc. to the Romance.

hīllae: from hīra.

hīrsūtus: like hīrtus

Hīrtius and hīrtus: ī acc. to the Romance.

hīscō: see § 49.

Hīspellum: cf. Gr. Εἰσπέλλον, Strabo, v. 227.

Hīspo, Hīspulla: like Hīspellum.

hōrnus: from hōra?

hōrsum: for *ho-vorsum.

hydrōps: like Gen. hydrōpis; § 42. 2.

Hymēttus: Gr. Ὑμηττός.

Hypermēstra: Gr. Ὑπερμήστρα.


īgnis: Ignis, CIL. xi. 826.

illōrsum: for *illo-vorsum.

illūstris: from lūx.

Īllyria: Eillvrico, CIL. i2. p. 77.

impingō -ere, impēgī, impāctus: see § 46. 3. b).

īnfēstus: infésti, CIL. v. 2627; cf. manifēstus.

īnstīnctus: see distinguō.

intellegō, intellegere, intellēxī, intellēctus: like legō.

intervāllum: from vāllus.

intrōrsum: for *intro-vorsum.

involūcrum: ū in Plautus, Captivi, 267; § 36. 2.

Iōlcus: Gr. Ἰωλκός.

istōrsum: for *isto-vorsum.


jēntāculum: see § 40. 2. a).

jēntātiō: see § 40. 2. a).

jūglāns: from Jov- and glāns.

jungō, -ere, jūnxī, jūnctus; see § 46. 3. b).

jūrgō: for jūrigō, from jūs.

Jūstiniānus: from jūstus.

{p. 60}

jūstitium: from jūs.

jūstus: from jūs: also iv́sto, CIL. ii. 210; v. 5919.

jūxtā, jūxtim: from jūgis ‘joined with.’


lābor, lāvī, lāpsus sum: see § 46. 3; dIlápsam, CIL. xi. 3123.

lābrum, ‘bowl’: for lavābrum; labrum, ‘lip,’ has ă.

labrūsca: ū acc. to the Romance.

laevōrsum: for *laevo-vorsum.

lāmna: syncopated for lāmina.

lārdum: syncopated for lāridum.

Lārs, Lārtis: Lárt-, CIL. x. 633.

lārva: like lārua, the early Latin form, e.g. Plautus, Amphitruo, 777; Captivi, 598.

lātrīna: for lavātrīna; cf. Plautus, Curculio, 580; § 36. 2.

lātrō, ‘bark’: ā in oblātrātrīcem, Plautus, Miles Gloriosus, 681; § 36. 2.

lavābrum: see § 51. 1.

lavācrum: see § 51. 1.

legō, -ere, lēgī, lēctus: see § 46. 3.

lēmma: Gr. λῆμμα.

lēmniscus: Gr. λημνίσκος.

Lēmnos: Gr. Λῆμνος.

lentīscus: ī acc. to the Romance.

lībra: ī in Plautus, Pseudolus, 816; § 36. 2.

lībrō: like lībra.

līctor: lIctor, CIL. vi. 699 and often; líctor, Ephemeris Epigraphica, v. 51; λείκτωρ, Eckinger (Orthographie Lateinischer Wörter in Griechischen Inschriften, p. 43).

līmpidus: ī acc. to the Romance.

lingō, ere, līnxī, līnctus: ī acc. to the Romance.

lūbricus: ū in Plautus, Miles Gloriosus, 853; § 36. 2.

lūceō, -ēre, lūxī: see § 46. 3.

lūcta: ū acc. to the Romance.

lūctor: like lūcta.

lūctus: from lūgeō: also lv́ctvm, CIL. vi. 1527 e. 66; lv́ctv, CIL. v. 337; x. 4041. 2.

lūgeō, lūgēre, lūxī: see § 46. 3.

lūstrum, ‘expiation’: lv́strvm, Monumentum Ancyranum, ii. 3, 5, 8; ii. 3, 6, 10; lustrum, ‘haunt,’ has ŭ.

lūstrō: like lūstrum.

lūxuria: see lūxus.

lūxus: ū acc. to the Romance.

Lycūrgus: Gr. Λυκοῦργος.


mālle: for *mag(e) (magis) + velle.

manifēstus: [mani]féstvm, CIL. i. p. 319; very uncertain.

Mānlius: from Mānius; Mánlio, Mánlia, CIL. v. 615; Mánliae, ix. 3942.

manūpretium: ū in Plaut. Men. 544.

Mārcellus, Mārcella: from Mārcus; Márcella, CIL. xii. 3188.

Mārcius: from Mārcus; Márcivs, CIL. v. 555 et passim; Μάαρκιον, CIG. 1137.

Mārcus: Maarco, CIL. i. 1006; xiv. 2802; Márci, Boissieu, Inscriptions de Lyon, p. 143; Μάαρκος, CIG. 887 et passim.

Mārs, Mārtis: Mártis, Monumentum Ancyranum, iv. 21; CIL. x. 809 et passim.

Mārsī: like Mārs.

Mārtiālis: like Mārs.

māssa: Gr. μᾶζα.

mātrimōnium: from māter.

{p. 61}

mātrīx: from māter.

mātrōna: from māter; Mátróna, CIL. v. 5249.

māxilla: according to Priscian, iii. 36. (Keil).

māza: Gr. μᾶζα. See Cramer, Anecdota Oxoniensia, iii. 293.

mercēnnārius: for *mercēd-nārius.

Mētrodōrus: Gr. Μητρόδωρος.

mētropolis: Gr. μητρόπολις.

mīlle, mīllia: mIllia, Monumentum Ancyranum, i. 16; mIlliens, iii. 34; ī acc. to the Romance.

mīlvus: as in the early Latin mīluos.

Mōstellāria: from mōnstrum.

mūcrō: ū in Atta, Frag. 13 (ed. Ribbeck); § 36. 2.

mūlleus: ū acc. to the Romance.

mūllus: ū acc. to the Romance.

mūscerda: from mūs.

mūsculus: from mūs.

mūscus: ū acc. to the Romance.

mūstēla: from mūs.

Mycalēssus: Gr. Μυκαλησσός.


nancīscor: see § 49.

Nārnia: Umbrian Nahar- (=ā).

nārrō: nárrem, Boissieu, Inscriptions de Lyon, p. 136.

nāscor: § 49; náscerer, Monumentum Ancyranum, ii. 44; Náscentibvs, CIL. xii. 3702.

nāsturcium: from nāsus.

nefāstus: from nefās.

neglegō, -ere, neglēxī, neglēctus; see legō.

nēquīdquam (nēquīcquam): from Abl. quīd.

nītor, nītī, nīxus sum: see § 46. 3.

nōlle: by contraction from *nŏvelle (for *nĕ-velle; § 73. 3).

nōndum: from nōn and dum; nóndvm, CIL. x. 4041. 6.

nōngentī: for *no(v)engentī.

nōnne: from nōn.

nōnnūllī: from nōn and nūllus.

Nōrba: Gr. Νώρβη.

nōscō: ō acc. to the Romance.

nūbō, -ere, nūpsī, nūpta: see § 46. 3.

nudiūstertius, quārtus, etc.: see § 86.

nūllus: from ne and ūllus; nv́llvm, CIL. x. 4787.

nūndinae, nūndinum: for *no(v)endinae; noundinae in early Latin; Nv́ndinvs, CIL. xii. 3650.

nūntius: for *nove-ntius? (‘newsbringer’).

nūntiō: like nūntius.

nūptiae: like nūpta.

nūsquam: like ūsquam. {Buck: ŭ.}

nūtrio: like nūtrīx.

nūtrīx: ū in Plautus, Curculio, 643; nūtrīcātus, Miles Gloriosus, 656; nūtrīcant, Miles Gloriosus, 715; § 36. 2.


oblīvīscor: see § 49; oblIvIscemvr, CIL. vi. 6250.

Oenōtria: Gr. Οἰνωτρία.

ōlla: for aulula; ólla, CIL. vi. 10006 et passim.

Onchēstus: Gr. Ὄγχηστος.

Opūs, -ūntis: Gr. Ὀποῦντος.

ōrca: ō acc. to the Romance.

orchēstra: Gr. ὀρχήστρα.

ōrdior: like ōrdō.

ōrdō: órdinis, Boissieu, Inscriptions de Lyon, p. 136; CIL. ix. 5177; xii. 3312; ō acc. to the Romance.

ōrnō: órnare, CIL. xii. 4333 et passim.

ōrnāmentum: órnámentis, CIL. xii. 3203 et passim; cf. ōrnō.

{p. 62}

ōscen: from ōs.

ōscillum: from ōsculum.

ōscitō: from ōs.

ōsculor: from ōs.

Ōstia: from ōs; Gr. Ὠστία.

ōstium: from ōs; ὤστια, scholion to Aristophanes, Plutus, 330; óstivm, CIL. vi. 4710, óstió, Monumentum Ancyranum, v. 14.

ovīllus: from ovīnus.

Ōxus: Gr. Ὦξος, in Strabo.


pacīscor, pacīscī, pactus sum: see § 49.

palimpsēstus: Gr. παλίμψηστος.

palūster: from palūs.

pangō, pangere, pepigī, pāctus: the compounds impāctus, compāctus, point to ā; see § 46. 4. c).

paradīgmā: Gr. παράδειγμα.

pāscō, pāscere, pāvī, pāstus: see § 49.

pāstillus: like pāscō.

pāstiō: like pāstus.

pāstor: like pāstus; paastores, CIL. i. 551; pástóris, CIL. x. 827.

pāxillus: acc. to Priscian, iii. 36.

pēgma: Gr. πῆγμα.

{pēlvis, Allen.}

pentāthlum: Gr. ἆθλον.

perēmptālis: from perēmptus (emō).

pergō, pergere, perrēxī, perrēctus: like regō.

perīclitor: like perīculum.

perimō, -ere, perēmī, perēmptus: like emō.

Permēssus: Gr. Περμησσός.

perrēptō: from rēptō (rēpō).

persōlla, for persōn(u)la, from persōna.

Pessīnūs, -ūntis: Gr. Πεσσινοῦντος.

Phoenīssa: like Phoenīx.

pīctor: like pīctus (pingō).

pīctūra: like pīctus.

pīgmentum: pIgment-, CIL. viii. 1344; ī acc. to the Romance.

pingō, pingere, pīnxī, pīctus: see under fingō, which is precisely parallel. {Allen: pĭctus.}

pīstillum, pīstor, pīstus (from pīnsō), pīstrīnum, pīstrīlla: pIstvs, CIL. v. 6998. The Romance evidence is conflicting, but is favorable to ī.

Pīstōria: like pīstor.

plēbīscītum: = plēbī scītum, and better so written.

plēbs: like genitive plēbis; pléps, CIL. v. 6797; xii. 4333.

plēctrum: Gr. πλῆκτρον.

Plīsthenēs: Gr. Πλεισθένης.

plōstellum: from plaustrum.

plūsculum: from plūs.

poētria, -is: Gr. ποιητρία, ποιητρίς.

Pōlla: = Paulla; Pólla, CIL. xii. 3471; cf. Pōlliō.

pollingō, -ere, līnxī, līnctus: like lingō.

pollīnctor: like pollīnctus.

Pōlliō: from Paullus; Póllio, CIL. vi. 22840 et passim; Πωλλίων in Plutarch, Dio Cassius, and elsewhere.

pollūceō, -ere, -ūxī: § 46. 3.

Polymēstor: Gr. Πολυμήστωρ.

porrigō, -ere, porrēxī, porrēctus: like regō.

praelūstris: like lūx.

prāgmaticus: Gr. πρᾱγματικός.

Prāxitelēs: Gr. Πρᾱξιτέλης (πρᾶξις).

prēndō: for pre-hendō.

prīmōrdium: from ōrdior.

prīnceps: from prīmus and capiō.

prīncipālis: from prīnceps.

prīncipātus: from prīnceps.

prīncipium: from prīnceps.

Prīsciānus: from prīscus.

prīscus and Prīscus: Príscvs, CIL. xi. 1940; PrIscvs, CIL. ix. 4354. c; Πρεῖσκος CIG. 4494 et passim.

{p. 63}

prīstinus: like prīscus.

prōcrāstinō: from crās.

Procrūstēs: Gr. Προκρούστης.

profēstus: from fēstus.

prōmō, -ere, prōmpsī, prōmptus: see § 46. 3.

prōrsum, prōrsus: for *pro-vorsum, -sus.

prōsperus: from prō*spēre? (‘according to expectation’).

prōstibulum: from prō and stabulum.

Pūblicius, Pūblicola: from pūblicus. Poplicola is another word, viz. from poplus,, early form of populus, ‘people.’

pūblicus: from pūbes; pv́blicór[vm, CIL. vi. 1377; ū in Plautus, Miles Gloriosus, 102, 103; Captivi, 334 et passim; § 36. 2; ū also acc. to the Romance.

Pūblilius: like Pūblius.

Pūblius: like pūblicus.

pulvīllus: from pulvīnus; pvlvIllvs, CIL. i. Fasti Cap., a. 297.

pungō, -ere, pupugī, pūnctus: ū acc. to the Romance.

pūnctus: see pungō.

pūrgō: for *pūrigō (pūrus): ū also acc. to the Romance.

pūrgāmentum: from pūrgō.

pūrgātiō: from pūrgō.

pūstula: from pūs; ū acc. to the Romance.


quārtus: quártvs, CIL. iii. 4959; Monumentum Ancyranum, iii. 22 et passim.

quārtānus: like quārtus.

quārtārius: like quārtus.

quiēscō: acc. to Gellius, Noctes Atticae, vii. 15, some persons pronounced quiĕscō in his day; but other -scō formations have practically invariably ē before sc: quiēvī and quiētus also point to quiēscō; qviéscere is found CIL. vi. 25531.

quīncūnx: from quīnque and ūncia.

quīndecim: from quīnque and decem; ī acc. to the Romance.

quīngentī, quīngēnī, quīngentiēs: from quīnque.

Quīnquātrūs: from quīnque; ā in Plautus, Miles Gloriosus, 691; § 36.2.

quīnque: qvInqve, CIL. vi. 3539 et passim; ī acc. to the Romance.

quīnquāgintā: from quīnque.

quīnquennium: from quīnque.

quīnquiēs: from quīnque.

quīntāna: from quīntus.

Quīntiliānus: from quīntus.

Quīntīlis: from quīntus.

Quīntilius: from quīntus; QvInctilio, CIL. iii. 384.

quīntus, Quīntus, Quīnctius: qvIntvm, Monumentum Ancyranum, iii. 1; i longa occurs repeatedly; Κόειντος, CIG. 2003; ī acc. to the Romance.

quīppe: for *quīd (Abl.) and -pe.

quōrsum, quōrsus: for *quō-vorsum, *quō-vorsus.


rāllus: for rār(u)lus from rārus.

rāstrum: from rādō.

reāpse: for rē eāpse (Abl. of ipsa).

rēctē, rēctor: like rēctus.

rēctus: see regō.

redigō, -ere, redēgī, redāctus: like agō.

redimō, -ere, redēmī, redēmptus: like emō; Ῥεδῆνπτα, CIG. 9811; redémpta, CIL. vi. 22251.

{p. 64}

redēmptiō, redēmptor: from redimō.

rēgnum: see § 38, end.

rēgnō: like rēgnum.

rēgnātor, rēgnātrīx: from rēgnō.

regō, -ere, rēxī, rēctus: see § 46. 3. b).

relinquō, -ere, relīquī, relīctus: see § 46. 3. b). {Allen: relĭctus.}

reminīscor, -ī: see § 49.

rēpō, rēpere, rēpsī, rēptum: see § 46. 3.

restinguō, -ere, restīnxī, restīnctus: see distinguō.

rīxa: ī acc. to the Romance.

rōscidus: from rōs.

Rōscius: Róscio, CIL. vi. 2060, 5; Ῥώσκιος, Plutarch, Cicero, 3; 5; Pompey, 15.

rōstrum: from rōdō; ῥῶστρον, Hesychius.

Rōstra: from rōstrum.

Rōxāna: Gr. Ῥωξάνη.

rūctō: acc. to the Romance (Gröber, Archiv, v. p. 370), which points also to a form with ŭ.

rūctus: like rūctō.

rūrsum, rūrsus: for *re-vorsum, *re-vorsus.

rūscum: ū acc. to the Romance.

rūsticus: from rūs; Rv́sticvs, CIL. ix. 4012; ū acc. to the Romance.


sanciō, sancīre, sānxī, sānctus: see § 46. 3. b).

sārculum: like sāriō.

Sārsina: Sássinas in an inscription.

scēptrum: Gr. σκῆπτρον.

scīscō: see § 49; d[esc]Iscentem, Monumentum Ancyranum, v. 28.

scrībō, -ere, scrīpsī, scrīptus: see § 46. 3; scrIptvm, CIL. vi. 2011; conscreiptvm, CIL. i. 206. 87; 109; cónscríptis, CIL. x. 3903; ī acc. to the Romance; Umbrian screihtor = scrīptōs (Nom. Plu.).

scrīptiō, scrīptitō, scrīptor, scrīptūra: see scrībō.

sēgmen: like sēgmentum.

sēgmentum: see § 39.

sēgnis: ségnis in a Herculanean papyrus.

sēligō, sēligere, sēlēgī, sēlēctus: like legō.

Selīnūs, -ūntis: Gr. Σελινοῦντος.

sēmēstris: for *ses-mēstris, *sexmēstris; see § 89.

sēmūncia: from sēmi- and ūncia.

septūnx: from ūncia.

sēscūncia: for sēsqui- and ūncia.

sēscuplex, sēscuplus: for sēsqui- and -plex.

Sesōstris: Σέσωστρις.

sēsqui-: = sēmisque-.

sēstertius: for sēmis tertius.

Sēstius: Gr. Σήστιος, in Cic. ad Att, vii. 17. 2 et passim; Σηστία, CIA. iii. 1450.

Sēstos, Sēstiī: Gr. Σηστός, Σήστιοι.

Sīgnia: Seig[nia, CIL. i. 11.

sīgnum and sĭgnum: seignvm, CIL. xiv. 4270; sIgna, Boissieu, Inscriptions de Lyon, p. 606; see § 38.

sī̆gnificō, sī̆gnō: like sī̆gnum.

sīnciput: for sēmi + caput, i.e. sīnciput for *sēnciput, by vowel assimilation; see § 90. {Michelson: Such an assimilation of ē to ī is quite problematic, and at least not conclusively demonstrated: see Brugmann, Grundriss I2, p. 505, footnote 1, p. 836; K. Verg. Gr., § 100 anm.; Buck, A. J. P. 17, 270 as cited by Brugmann. It is true that Sommer (Lat. L. u. Fl., pp. 77, 115) accepts this sequence, but his explanation of sinciput is the same as Brugmann’s; namely, sēmi + caput became *sēŋciput, whence *sĕŋciput by the law of shortening a long vowel before a nasal + consonant; and then the ĕ became ĭ as it was before ŋ + a consonant. See Brugmann, Grundriss I2, §§ 121, 134; K. Verg. Gr., pp. 216, 218; Sommer, Lat. L. u. Fl., pp. 77, 147.}

sinistrōrsus: for *sinistro-vorsus.

sīstrum: Gr. σεῖστρον.

sōbrius: ō in Plautus, Miles Gloriosus, 812; § 36. 2.

Sōcratēs: Gr. Σωκράτης.

sōlstitium: from sōl.

Sōphrōn: Gr. Σώφρων.

{p. 65}

sōspes: Gr. Σῶσπις, CIA. iii. 1161 et passim.

sōspita, sōspitō: like sōspes.

stāgnō: like stāgnum.

stāgnum: see § 38, end.

{stēlla, Allen.}

stīlla: ī acc. to the Romance.

stīllicidium, stīllo: like stīlla.

stringō, -ere, strīnxī, strictus: strīnxī acc to the Romance.

struō, -ere, strūxī, strūctus: see § 46. 3. b); ū also acc. to the Romance.

strūctor: like strūctus; cf. strVctor, CIL. x. 708; ū acc. to Gellius, xii. 3. 4.

strūctūra: like strūctus.

sublūstris: like lūx.

substrūctiō: like strūctus.

suēscō: as in suēvī, suētus.

sūgō, -ere, sūxī, sūctus: see § 46. 3; ū acc. to the Romance.

suīllus: from suīnus.

sūmō, -ere, sūmpsi, sūmptus: see § 46. 3; ū also acc. to the Romance.

sūmptus: from sūmō.

sūrculus: from sūrus.

surgō, -ere, surrēxī, surrēctus: like regō.

sūrsum: for *su-vorsum.

sūtrīna: like sūtor.

Sūtrium: ū in Plautus, Casina, 324; § 36. 2.

syllēpsis: Gr. σύλληψις.


tangō, -ere, tetigī, tāctus: see § 46. 3. b).

tāctiō: like tāctus.

Tartēssus: Gr. Ταρτησσός.

tāxillus: acc. to Priscian, iii. 36.

Tecmēssa: Gr. Τέκμησσα.

tēctum: from tegō.

tegō, -ere, tēxī, tēctus: see § 46. 3. b).

Telmēssus: Gr. Τελμησσός.

Tēmnos: Gr. Τημνός.

Termēssus: Gr. Τερμησσός.

terūncius: from ūncia.

theātrum: Gr. θέᾱτρον.

Thrēssa: Gr. Θρῇσσα.

Tīllius: tIllivs, CIL. vi. 2043.

tingō, -ere, tīnxī, tīnctus: see § 46. 3. b).

trāctim: like trāctus.

trāctō: like trāctus.

trahō, -ere, trāxī, trāctus: see § 46. 3. b). {Allen: trăctus.}

Trapezūs, -ūntis: Gr. Τραπεζοῦς, -οῦντος.

trifōrmis: from fōrma.

trilībris: like lībra.

trīstis: trIstior, CIG. 6268; ī also acc. to the Romance.

trūlla: for truella. The Romance also points to ŭ. {Sic. Obviously ū is intended.}

trūcta: ū acc. to the Romance.

tubilūstrium: like lūstrum.


ūllus: from ūnus; v́lla, CIL. ii. 1473; v́lli, CIL. vi. 10230.

ūlna: Gr. ὠλένη. {Michelson: Except for fūr and cūr, the rule is that long ō remains in Latin. Suppose we start with a prototype *ōlena. This certainly would appear in Latin as ŭlna through the stages *ōlna, *ŏlna. The shortening of a long vowel before l + a consonant, and the change of ŏ to ŭ before l + a consonant (except l) are too well known to require illustration.}

ūlva: like ūlīgō.

ūncia: like ūnus.

ūnctiō: like ūnctus (ungō).

ūndecim, ūndecimus: from ūnus and decem. {Michelson: Certain Romance languages postulate ŭnd-.}

ūndēvīgintī, etc.: like ūnus.

ungō, -ere, ūnxī, ūnctus: see § 46. 3. b).

ūrō, -ere, ussī, ūstus; ū in the Perfect Participle acc. to the Romance; for the ŭ in ussī, see § 53 s. v. {Michelson: Even if the Romance languages postulate ūstus, yet in view of ostili we must allow ŭ- for ustus in Classical Latin; for the initial o- of ostili points distinctly to a short open ŭ. On the Romance forms see Sommer, 1. c., p. 644.}

ūspiam: like ūsque. {Buck: ŭspiam.}

ūsquam: like ūsque. {Buck: ŭsquam.}

ūsque: ū acc. to the Romance. {Buck: Marx also favoured ūsque on the basis of compensatory lengthening, a view that is untenable; on the contrary, the etymology points to a short ŭ. The only Romance evidence is the modern French jusque, which, on account of the u does seem to point to an original ū. However, the Old French josque is most easily understood if we start from ŭsque, and jusque can be explained as a being influenced by the medieaval Latin usque. Allen does not list ūsque in his list of common words with hidden quantity, which proably indicates that he accepts ŭsque.}

{p. 66}

ūstrīna: like ūstus.

ūsūrpō: ūsū rapiō?


vāllum, vāllus: vállári, CIL. ii. 4509; also Vállivs, Vállia, CIL. xix. 4039.

vāllāris: see vāllum.

vāllō: see vāllum.

vāsculum: like vās.

vāstus: the Teutonic languages point to a long root vowel.

Vēctis, ‘Isle of Wight’: Gr. Οὐηκτίς.

vēgrandis: from vē- and grandis.

Vēlābrum: ā in Plautus, Curculio, 483; § 36. 2.

Venāfrum: the suffix is originally the same as -ābrum; see § 51. 1.

vēndō: from vēnum and dō.

vērnus: from vēr.

vēstibulum: + stabulum? Cf. prōstibulum.

vēstīgium: + steigh-?

Vēstīnī: Gr. Οὐηστῖνοι.

vēxillum: véxillo, CIL. xii. 3167; Byzantine Gr. βήξιλλα; CIG. 4483, οὐηξιλλατι(ῶ)σιν; also acc. to Priscian, iii. 36.

vīctus: from vīvō. The Romance also shows ī.

vīlla: vIlla, CIL. vi. 9834; the Romance points to ī.

vīnaēmia: from vīnum and aēmō.

Vīpsānius: vIpsanI, CIL. vi. 12782; vIpsania, CIL. vi. 8877; Βειψάνιος, CIG. 5709.

Vīpstānus: vIpstanvs, CIL. vi. 2039 and frequently; Οὐειψτανοῦ CIG. 5837, b; CIA. iii. 621.

vīscus: vIsceris, CIL. vi. 1975. {Inscription actually says vIscera.}

vīvō, -ere, vīxī, vīctum: § 46. 3; veixit, CIL. xiv. 2485; vIxit, CIL. ii. 3449; vIctv́ro, CIL. vi. 12,562; βεῖξιτ in an inscription cited by Eckinger (Orthographie Lat. Wörter in Griech. Inschriften, p. 43).

Vopīscus: Gr. Θὐοπεῖσκος {sic in Bennett}; vopIsco, CIL. x. 4872.


Xenophōn, -ōntis: Gr. Ξενοφῶν, -ῶντος.


zōstēr: Gr. ζωστήρ.

53. Words whose Hidden Quantities are in Dispute.

agmen: ā Marx; see § 39.

agnātus, agnōtus, etc.: ā Marx; see § 38.

agnus: ā acc. to many; see § 38.

alliciō: som scholars mark the e of the Perfect long in allexī, illexī, pellexī; and likewise in -spexī (aspexī, cōnspexī, etc.), flexī, pexī, vexī. This marking rests upon a statement of Priscian in ix. 28. But Priscian in this passage simply says that Perfects in -xī have a long vowel before the x only when the vowel is e; he does not state that every e is long before -xī. Moreover, little weight is to be attached to this testimony; for in the paragraph immediately preceding (ix. 27) Priscian lends the weight of his authority to such forms as trăxī, mănsī, dŭxī, which certainly had a long vowel in the best period. Osthoff (Geschichte des Perfects, p. 227) and Brugmann (Grundriss der Vergleichenden Grammatik, ii. p. 1182) support ē in Perfects of this type by arguments drawn from comparative grammar; but the evidence does not warrant a positive conclusion in their favor.

allium: ā Marx; see § 88. 1.

amygdalum: ȳ Marx, without citation of evidence, Gröber (Archiv, i. 240) and Körting (Wörterbuch) give y.

anxius: ā Marx, Brugmann, Sommer, and others; see § 46. 5. end.

Appulus, Appulia: Ā Marx. Āpulus, Āpulia are the better spelling.

aprugnus: ū acc. to many; see § 38.

arca: this word occurs with the apex (árcae) in Boissieu, Inscriptions de Lyon, p. 279, but it is doubtful whether this single instance justifies our recognizing the a as long. The root arc-, ‘hold, confine,’ had originally a short vowel, as is shown by coerceō (for *co-arceō); *ārceō would have retained the ā in composition; see § 72. Nevertheless it is undeniable that a tendency existed in certain localities to lengthen the short vowel before r + a consonant. In some words this resulted in permanent lengthening of short vowels in the classical speech, e.g. in fōrma, quārtus (cf. quattuor), ōrca, and probably in ōrdō, ōrdior, ōrnō. īn case of other words we simply meet isolated local manifestations of the tendency, e.g. in árváli, CIL. vi. 913; libértis, CIL. x. 3523; sérvilio, Henzen, 6490; vIrgo, CIL. vi. 2150; vIrtvtis, CIL. vi. 449; córvinvs, vi. 2041; órfito, vi. 353; Córdiae, vi. 22,915; Nárbóne, xii. 3203; Nárbonénsis, xii. 3163; hórt[os, vi. 9493; cohórt[is, vi. 2993; Fórt[is Fórtvnae, vi. 9493; fórtvnata, vi. 7527. Yet these sporadic inscriptional markings hardly justify our assuming ārvum, ārvālis, lībērtus, sērvus, vīrgō, etc., for the classical speech; and the same applied to arca. See Seelmann, Aussprache des Latein, p. 91.

Arrūns: Ā Marx; see § 88. 1.

arvum, arvālis: see arca.

ascendō, ascrībō, etc.: ā Marx; see § 48.

ascia: ā Marx; see § 89.

Asclēpiadēs: Ā Marx.

Asculum: Ā Marx.

aspiciō, -ere, -exīm, ectus: ēxī Marx and Lewis; see above under alliciō.

assus: ā Marx and Lewis, as if for *ārsus, which is improbable. See Osthoff, Geschichte des Perfects, p. 545.

astus, astūtus: ā Marx, as if for *axtus, etc.; see § 89.

axis: ū {sic in Bennett} Marx, without warrant; Charisius (Keil, i. 11. 22) and Diomedes (Keil, i. 428) both testify to a short a.

balbuttiō: ū Marx; see § 88. 1.

barritus: ā Marx; see § 88. 1.

benignus: ī Marx and others; see § 38.

benignitās: ī Marx and others; see § 38.

bēs, bessis: ē in oblique cases Marx; but in view of Quintilian’s statement (i. 7. 20) that ss was not written after a long vowel in the {p. 68} post-Ciceronian period, it is much more probable that the word followed the analogy of as, assis. Osthoff, Geschichte des Perfects, p. 545.

braccae: ā Marx; see § 88. 1.

caballus: ā Marx, as if a diminutive from an assumed *cabānus, for which there is no warrant.

Camillus: ī acc. to Appendix Probi (Keil, iv. p. 197); i acc. to Martianus Capella (p. 66. 4, ed. Eyssenhardt).

capessō: ē acc. to Osthoff (Geschichte des Perfects, p. 221), who regards capessō, facessō, lacessō, as originally aorists of the same type as habēssō, licēssit, etc. Brugmann (Grundriss, ii. p. 1203), taking a different view of the formation, regards the e as short.

carduus: possibly ā, if from the same root as cār-ex, ‘sedge’ (lit. ‘rough plant’?).

carrus, carrūca: ā Marx; see § 88. 1.

Cassandra: Cāss- Marx; see § 88. 1.

cingō, -ere, cīnxī, cīnctus: Lewis (E.L.D.) regards the i as short in cinxī; likewise in -stinxī, -stinctus; tinxī, tinctus, and in pinxī, finxī. The Romance languages seem to point to ī in the Perfect and Perfect Participle of all these words, e.g. Italian cinsi, cinto; stinsi, stinto; finsi, finto, etc. Inscriptions, moreover, give extInctos, cInctvs. See d’Ovidio in Gröber’s Grundriss, i. p. 501 f.; Körting, Wörterbuch, and Fröhde in Bezzenberger’s Beiträge, xvi. p. 193. {Michelson: Proof for the long ī of cīnctus is also Umbrian sihitu.}

classis: ā Marx, on the basis of an assumed etymological connection with clārus. {Allen: there is no evidence whatever for a long vowel in classis.}

cognātus, cognōmen, cognōscō, and other words beginning with cogn-: the o here is regarded as long by many; but the evidence is not sufficient to warrant this view; see § 38.

combūrō: ō Marx, who explains the verb as for *co-amb-ūrō; cf. cōgitō for *co-agitō.

cōnfestim: ē Marx, after the analogy of manifēstus, which latter is somewhat uncertain.

conjungō, conjūnx: ō Marx, on the basis of cónivgi, CIL. v. 1066; vi. 9914, which are too improbable to merit acceptance.

cōnspiciō, -ere, -exī, -ectus: ēxī Marx and Lewis; see above under alliciō.

cunctor: ū Marx, whose treatment of this word is unintelligible. {Michelson agrees with Marx. (Cf. § 46 end.)}

damma: ā Marx; see § 88. 1.

dēspiciō, -ere, -exī, -ectus: -ēxī Marx and Lewis (E.L.D.); see under alliciō.

dignus: ī Marx and others; see § 38.

discidium, discrībō, distō, distinguō, distringō: dīs- Marx and Lewis (E.L.D.); see § 48.

discō: ī Marx, on the theory of compensatory lengthening (discō for *di-dc-scō); see § 89.

distinguō, -ere, -īnxī, -īnctus: see cingō. For dīstinguō, see above under discidium.

duumvir: ū Marx and Lewis (E.L.D.); see § 42. 1.

Dyrrhachium: ȳ Marx, who cites the modern name Durazzo.

ēnormis: ō Marx and Lewis (E.L.D.); see norma.

Erinnys: ī Marx; cf. § 88. 1.

exstinguō, -ere, -īnxī, -īnctus: see distinguō.

fastīgium: ā Marx, on the theory of compensatory lengthening; see § 89.

fastus, ‘disdain’: ā Marx, on the theory of compensaroty lengthening; § 89.

festīnus, festīnō: ē Lewis and Marx, on the theory of compensatory lengthening, as though for fendt-; see § 89.

festūca, fistūca: ē and ī Marx, on the theory of compensatory lengthening (see § 89), as though for ferst-.

fingō, -ere, fīnxī, fīctus: see cingō. {Allen: fĭctus.}

flectō, -ere, flexī: flēxī Lewis and Marx; see under alliciō.

forsit, forsitan: Marx writes fōrsit and fōrsitan on the basis of the Romance. But Körting (Wörterbuch) interprets the evidence of the Romance as pointing to ŏ.

fortasse, fortassis: ā Marx, who cites nothing valid in support.

fragmen: ā Marx and many others; see § 39.

frendō, -ere, frenduī, frēsus, or fressus: -ēssus Marx; § 98. 2.

futtilis: ū Marx; see § 88. 1.

garriō, garrulus: ā Marx, who connects with Gr. γᾱρύω; see § 88. 1.

Garumna: ū Marx on the basis Gr. Γαρούνας; but the Romance (Fr. Garonne) points to ŭ.

gignō: ī acc. to Marx and many others; see § 38.

gluttiō, gluttus: ū Marx; see § 88. 1.

grallae: ā Marx; see § 88. 1.

hallūcinor: ā Marx; see § 88. 1.

helluō: ē Marx; see § 88. 1.

hīrcus: the quantity of the i is doubtful, as the Romance words upon which judgment is based may be ‘semi-literary’; see § 35. 5 fin. Cf. Gröber (Archiv, iii. 139); Körting (Wörterbuch). Marx compares hīrtus, with which hircus may be related.

hispidus: ī Marx and Lewis. Marx cites the Romance, but the word is probably ‘literary’ in the Romance; see § 35. 5 fin. Körting (Wörterbuch) regards the i as short.

ictus: ī Lewis; but the Romance points to ĭ.

īlignus: ī acc. to Marx and others; see § 38.

immō: īmmō Marx, in view of īmus and Immo, CIL. iii. 774. The Romance points to ĭ.

īnspiciō, -ere, -exī, -ectus: -ēxī Marx and Lewis; see alliciō.

jubeō, -ēre, jussī, jussus: jūssus Lewis. The only authority for ū in jussu is iv́ssvs, CIL. vi. 77. But the apex here is entitled to no weight. The same inscription has at least one other error in the use of the apex, viz. anniv́s. In favor of jūssī we find iv́ss[it, CIL. xii. 1930; iv́ssit, iv. 25531; and iovsit, CIL. i. 547 a, et passim in inscriptions of the ante-classical period. The simplest solution of the difficulties is to recognize an ante-classical jūsī, which is well attested by Quintilian in i. 7. 21, and a classical jŭssī. The shortening occurs in accordance with the principle explained in § 88. 1. In view of Quintilian’s additional statement that jussī was the orthography of his day, and that ss was not written after a long vowel (i. 7. 20) this {p. 70} is almost a necessary conclusion. The apex in CIL. xii. 1930 is then a blunder, a result of the confusion of jūsī and jŭssī. See Osthoff, Geschichte des Perfects, p. 532 ff.; Brugmann, Grundriss, ii. 1182; Frödhe, Bezzenberger’s Beiträge, xvi. p. 184.

Juppiter: ū Marx; see § 88. 1.

lascīvus: ā Marx, on the basis of an assumed etyomology, which connects the word with the root lās (lār-) of lārua.

lībertus: ē Lewis; see arca.

lībertās: ē Lewis; see arca.

lignum: ī acc. to Marx and others; see § 38.

littera: ī Marx; see § 88. 1.

malignus: ī acc. to Marx and others; see § 38.

Matrona: ā Marx, without citation of evidence.

Messalla: ā Marx; see § 88. 1.

mingō, -ere, minxī, mictum: mīnxī acc. to Marx and Lewis; see § 46, end.

misceō, miscēre, miscuī, mixtus: ī in mīxtus acc. to Marx and Lewis. The Romance points to ĭ (Gröber, Archiv, iv. 117; Körting, Wörterbuch).

mittō, mittere, mīsī, missus: the Romance points to i; a few suspicious instances of i longa occur, e.g. dimIssis, CIL. iii. p. 862 (shown by Osthoff, Geschichte des Perfects, p. 526, to be probably a blunder); mIssione, x. 7890; remIssa, xi. 1585.

Narbō, Narbōnēnsis: ā Marx; see under arca.

nesciō, nescius: ē Lewis; but compare nequeō. The Romance points to e.

norma: ō Lewis and Marx, who connects with Gr. γνώριμος.

nūncupō: ū Marx and Lewis, who connect with nōmen. {Michelson: In point of fact, the first u of nuncupō is short. That nuncupō is etymologically connected with nōmen is beyond dispute; but that does not show that the first u is long: on the contrary we must assume *nōmi-c- became *nōmc-, whence *nŏŋc-, and finally nŭnc-. Cf. the remarks on ulna and sinciput, and see Brugmann, Grundriss I2, pp. 142, 143, 149 and Sommer, 1. c., § 60.}

nūsquam: u Lewis; see ūsquam. {Buck: ŭ.}

ostrum: ō Marx, who connects with austrum.

Paelignus: ī acc. to Marx and others; see § 38. Gr. texts accent Παιλῖνοι.

pannus: ā Marx; cf. § 88. 1.

pectō, -ere, pexī, pexus: pēxī Marx, and Lewis; see under alliciō.

pelliciō: see alliciō.

perspiciō: see aspiciō

pestis: ē Marx, in accordance with an untenable theory of compensatory lengthening; see § 89.

pignus: ī acc. to Marx and others; see § 38.

pilleus: ī Marx; see § 88. 1.

pingō: see cingō

plangō, -ere, planxī, planctus: plānxī, plānctus acc. to many; see § 46, end.

planctus: ā acc. to many; see § 46, end.

plector, ‘be punished’: ē Marx, who compares πλήσσω.

posca: ō Marx, who compares pō-cu-lum; but the root had also a reduced form pŏ- (§ 69); cf. Gr. ποτόν.

poscō: ō Marx, on the theory of compensatory lengthening (pōscō for *porscō); see § 89.

postulō: ō Marx, as in the case of poscō.

prīvignus: ī acc. to Marx and other; see § 38.

{p. 71}

prōpugnāculum: ū acc. to Marx and others; see § 38.

pugna pugnax pugnō pugnus: ū acc. to Marx and others; see § 38.

pulmō: ū Lewis. The Romance points to ŭ.

quoūsque: Lewis u; see ūsque. {Buck: ŭ.}

respiciō, -ere, -exī, -ectus: -ēxī Marx and Lewis; see alliciō.

Sallustius: ā Marx.

sagmen: ā Marx and others; see § 39.

salignus: ī Marx and others; see § 38.

Sarmātae, Sarmātia: ā Marx, who compares the form Sauromātae.

sescentī: sēs- Marx and Lewis, on the theory of compensatory lengthening; see § 89. Marx compares Sēstius (for Sextius), but the ē in that word is exceptional. See Fröhde, Bezzenberger’s Beiträge, xvi. 204.

sordēs: ō acc. to Körting (Wörterbuch), on the basis of the Romance, but the only word he cites, is Italian sorde, which is very likely ‘literary’; see § 36, 5, end.

Sphinx: ī Marx.

spinter: ī Marx.

stannum: ā Marx, on the basis of the ‘by-form,’ stāgnum.

stella: stēla acc. to the Romance; probably the form with two l’s had ĕ.

strenna: ē Marx; see § 88. 1.

supparum: ū Marx; cf. § 88. 1.

suspiciō, -ere, -exī, -ectus; suspīrō: ū Marx; see § 48. On suspexī, see alliciō.

taxō: ā Marx.

testa: ē Marx, on the theory of compensatory lengthening (testa for *tersta); see § 89. The Romance points to e.

testis, testor, testāmentum, testimōnium, etc.: ē Marx, on the theory of compensatory lengthening (tēstis for *terstis); see § 89.

testūdō: ē Marx, as in testa.

tignum: ī acc. to Marx and others; see § 38.

tinguō, -ere, tīnxī, tīnctus: see cingō.

torreō, -ere, torruī, tostus: tōstus Marx, on the theory of compensatory lengthening (tōstus for *torstus); see § 89. The Romance points to o. See d’Ovidio in Gröber’s Grundriss, i. p. 520; Körting (Wörterbuch), Gröber (Archiv, vi. 129).

tressis: ē Marx; see bēs, bessis.

Tuscī: ū Marx, on the theory of compensatory lengthening (Tūscī for *Turscī); see § 89. The Romance points to u.

Tusculum: ū Marx: see Tuscī.

ultrā, ulterior, ultimus, etc.: ū Lewis, on the basis of an alleged apex in vltra, Boissieu, Inscriptions de Lyon, p. 136. But the apex does not occur there. See Lindsay, Latin Language, p. 595. The Romance points to u.

urceus: ū Marx, who cites ōrca; but the Romance points to u.

ūrna: ū Marx and Lewis. Marx compares ūrīnātor; but urna is to be referred to the root arc-, weak form urc- (§ 100. 2), whence ur(c)na. The Italion urna, if a genuine Latin inheritance, would point to ū; but it is probably purely literary; § 35. 5, fin.

{p. 72}

ūrō, -ere, ussī, ūstus: ūssī Lewis; but Priscian (Keil, i. 466. 6) gives ŭssī. See under jubeō.

urtīca: ū Marx and Lewis. Marx compares ūrō.

vehō, -ere, vexī, vectus: vēxī, Lewis; see under alliciō.

vescus: ē Marx, on the basis of the questionable etymology + ēsca.

victor, victus, victōria, etc.: ī Lewis, on the basis of repeated inscriptional markings, such as vIctor, CIL. vi. 10056; 10115; 1058; vIctorinvs, vi. 1058; vIctoriam, vi. 2086; invíctae, vi. 353. But with a single exception no one of these inscriptions can be shown to antedate the third century a.d.; and I quite agree with Christiansen (de Apicibus et I longis, p. 49) in the view that in the classical period the i was short; later, apparently, it was lengthened.

vinciō, -īre, vinxī, vinctus: vīnxī, vīnctus, acc. to Marx and others.

vīscum: ī Lewis; but the Romance points to ĭ.


1 The material here presented is based chiefly upon Marx’s Hülfsbüchlein, cited below, p. 39.

2 The following classes of words are omitted from this list:—

  1. Most derivatives and compounds.
  2. All words containing ns or nf.
  3. Inchoatives in -āscō, -ēscō, -īscō.
  4. Some rare Greek loan-words and proper names.
  5. Nouns and adjectives in -x, whose Genitive (acc. to § 42.2) shows the preceding vowel to be long.

[Back to the index.]