page 17 note 1 Hittite must now be accepted as an early offshoot from the parent stock of both IE and Hittite. A general view of Hittite studies is to be found in J. Friedrich, Hethitisch und ‘Kleinasiatische’ Sprachen (Berlin and Leipzig, 1931; II.5.1 in Geschichte d. idg. Sprw., cf. §3 end).

page 17 note 2 In this branch belong also the ‘North Aryan’ languages, of which records were recently discovered in Central Asia; cf. H. Reichelt IJ 1.20–36. Tocharian, found in the same region, forms a separate branch, in some respects more closely akin to the languages of Western Europe than to those of Asia; cf. A. Meillet IJ 1.1–19, and now E. Sieg, W. Siegling, and W. Schulze, Tocharische Grammatik (Göttingen, 1931).

page 17 note 3 Languages are commonly divided into three chronological stages: old or ancient; middle; modern or new.

page 17 note 4 Ancient Macedonian probably belonged with Thracian and Phrygian, though O. Hoffmann, Die Makedonen (Göttingen, 1908), sought to show that it was a Greek dialect of an aberrant type.

page 18 note 5 The ancient Venetic of the lower Po valley (§7. VII), though often classed with Illyrian and Messapian, was probably a distinct branch.

page 18 note 1 Another theory places their original home in Central Asia.

page 18 note 2 Also to North and South America, South Africa, Australia, etc., in recent centuries.

page 25 note 1 For a discussion of other sounds assumed by scholars to have been present in pIE, see Hirt, IG 1.246–52.

page 25 note 2 Emission of the breath through the nose during the pronunciation of sounds other than nasals is the cause of what is known as a nasal accent.

page 25 note 3 It is quite possible that the distinction of palatal stops from velar stops in pIE developed by the influence of the following vowels and semivowels (as in English), though it later became a fundamental feature of pIE phonetics. Thus palatal may have come from k before e i y, velar q from k before a o u w, labiovelar q ^w from ku before a vowel. Cf. E. H. Sturtevant, Lang. 6.213–28.

page 25 note 4 While ī and ū are certain, the long liquid and nasal vowels are of less certain status. It seems that ī and ū developed from eyә and ewә, and it is reasonable to suppose that etc. came from the similar groups enә erә etc. Yet the developments in the various languages indicate that these contractions of the liquid and nasal combinations were very rare. Cf. Hirt, IG 2.109–10.

page 25 note 5 A strong spirant y has been assumed for pIE, distinct from the weak spirantic semivowel y, to explain certain peculiarities of development in Greek; cf. Brugmann, Gdr. 1².§922–§923. But E. W. Fay, CQ 9.104–14, has shown that these differences arise from root-contamination in Greek; cf. also Hirt, IG 1.247–8.

page 25 note 6 The reduced a and e are the ‘shwa secundum’ of Hirt, IG 2.79–91, which he writes with the Russian character ь. The reduced o is his ‘shwa tertium’, written ъ; cf. IG 2.98–100. The assumption of a reduced a is perhaps unnecessary.

page 25 note 7 Recent X-ray moving pictures of the vocal organs show that there are important differences in the size and shape of the air-chamber in the larynx, which may be of primary importance in the production of the sounds; but the speaker has no voluntary control over this part of the speech-apparatus, except that by shifting the front part of the tongue he produces changes therein by reciprocal replacement and displacement. The descriptions in the text are therefore left in terms of the front part of the mouth, that the student may in practice be able to form the sounds by following the description.

page 25 note 8 The assumption of another o vowel, written å, was made by Brugmann, Gdr. 1².153–6, to explain the correspondence of Skt. ā in open syllables with in other languages; but the length of the vowels of Skt. has been explained by Buck AJP 17.445–72, as a product of analogy and of analogical extension. Cf. also Hirt, IG 2.19; and E. H. Sturtevant, Lang. 7.115–24.

page 25 note 9 The hook under the vowel is used also as the sign of nasality or of former nasality, as in Lith.

page 25 note 10 Despite Brugmann, Gdr. 1² 721, 724, the development of sh and zh after aspirate stops (by the principle given in Remark 16) is unlikely, being contradicted by such writings as Attic Greek χσένον = ξένον; and an aspirate ought to be as pronounceable before s as before a liquid, cf. Gk. χρ, ὄχλος.

Also despite Brugmann, Gdr. 1². 790–3, the assumption of another row of spirants (þ þh s sh) in pIE, as the reason for the development of dental stops in Greek words where other languages have sibilants, is probably unnecessary, and the differences are to be solved on the basis of root contaminations, differences in suffixes, analogical influences, and new etymologies. Cf. E. W. Fay, Bull. Univ. Texas No. 263 (Jan. 15, 1913), 35–7; Hirt, IG 1.249–52.

page 26 note 11 Brugmann's t^st(h), d^sd(h), with a but partly developed sibilant, is a needless differentiation; cf. Hirt, IG 1.255–6.

page 27 note 1 The term law in 'phonetic law is not to be understood in its usual meaning. A law is normally a prescription that a certain act shall or shall not he done, but a phonetic law is merely a formulation of what has actually happened. It might better be called a formula of phonetic change. Phonetic law, collectively, is the sum total of such formulations, i.e., the principle of regular phonetic change.

page 27 note 2 There are indeed so many factors which interfere with the operation of phonetic law, that some scholars now refuse to accept the law as a valid formulation; but their view is faulty, because they regard the interfering factors as fundamental and think phonetic law a mere interfering tendency, reversing the true relation. Cf. L. Bloomfield's review in Lang., Sept. 1932, of E. Hermann, Lautgesetz und Analogie (Leipzig, 1931).

page 28 note 1 These are possibly to be included rather in the Latin-Faliscan dialects; cf. J. Schrijnen, Neophilologus 7.223–39, and BSLP 32.1.64–64.

page 29 note 1 W. Schulze, Zur Geschichte der lateinischen Eigennamen (Berlin, 1904); A. Ernout, Les Éléments étrusques du Vocabulaire latin, BSLP 30.1.82–124 (Paris, 1930).

page 29 note 2 A. Ernout, Les Éléments dialectaux du Vocabulaire latin² (Paris, 1929).

page 29 note 3 F. O. Wiese, Die griechischen Wörter im Latein (Leipzig, 1882); G. A. E. A. Saalfeld, Tensaurus Italograecus (Wien, 1884).

page 29 note 4 W. Muss-Arnolt, Semitic Words in Greek and Latin, in TAPA 23.35–156.

page 30 note 1 Not the same as the ancient Sicilian of §6. I.D.

page 30 note 2 Not the same as the ancient Venetic of §7.VII.

page 30 note 3 Keeping the k-sound before e and i, cf. §47.I.

page 33 note 1 The old view that the syllabary of the Semites came from the Egyptian hieroglyphic writing (theory of de Rougé, published in 1874) was for a time in disfavor, but seems now, as a result of recent finds of old Semitic inscriptions, to be somewhat reestablished.

page 33 note 2 Some scholars, however, have suggested that the Phoenician syllabary and the Greek alphabet were derived, perhaps independently of each other, from the early Cretan characters.

page 33 note 3 Recently, the Turkish government officially adopted the Latin alphabet and made its use compulsory for Turkish in the Turkish Republic.

page 34 note 1 The theory of A. Kirchhoff, Studien zur Geschichte des griechischen Alphabets (Gutersloh, 1876; ⁴1887).

page 34 note 2 A complete alphabet of 26 letters is inscribed on the edge of a writing tablet of the 7th century B.C. found about 1920 in a tomb at Marsiliana in western Etruria. This alphabet, seemingly brought across the Adriatic from Central Greece, contains two characters (Nos. 15 and 18 in the Table) not actually found in the inscriptions of Cumae, and is probably the source of the alphabets of Italy; cf. A. Grenier, L'Alphabet de Marsiliana et les Origines de l'Écriture à Rome (École Française de Rome, Mélanges d'Archéologie et d'Histoire 41.3–41 [1924]).

page 34 note 1 See §12 n. 2.

page 34 note 2 Roberts, Introd. 16–21 (§11).

page 25 note 1 Mart. Cap. 3.261; Dig. 1.2.2.36.

page 36 note 2 i, v, and x were old Italian numerical signs, the single stroke standing for 1, the outline of the outspread hand (= one thumb and four fingers) for 5, and two such hand-outlines, joined at the wrist, for 10.

page 37 note 1 CIL 1².4.

page 37 note 2 Not occurring, but implied by qvei, CIL 1².7.

page 37 note 3 CIL 1².364.

page 37 note 4 CIL 1².583.

page 37 note 5 CIL 1².1202.

page 37 note 6 Since k was one of the letters dropped from use by the Etruscans.

page 37 note 7 Quint. 1.7.10.

page 37 note 8 Cf. CIL 1².586.

page 37 note 9 Cf. Eph. Ep. 7. p. 51.

page 37 note 10 Cf. CIL 10.960.

page 37 note 11 CIL 9.1175.

page 37 note 12 Quint. 1.7.28–29.

page 37 note 13 Fest. 23.10 M.

page 37 note 14 CIL 1².364.

page 37 note 15 CIL 1².1204.

page 37 note 16 CIL 1².474, 479.

page 37 note 17 G. Hempl, TAPA 30.24–41.

page 37 note 18 Plutarch, Quaestiones Romanae 54.

page 37 note 1 A. Ernout, Le Parler de Préneste, 17–23.

page 37 note 2 CIL 1².378.

page 37 note 3 CIL 1².560; cf. Kent; Lang. 5.21.

page 37 note 4 CIL 1².388.

page 37 note 5 CIL 1².1176.

page 37 note 6 Ter. Scaur. 7.15.1–7 K.

page 37 note 1 CIL 1².3.

page 37 note 2 Unless this character was brought by the Etruscans from Asia Minor.

page 38 note 1 CIL 5.1655.

page 38 note 2 CIL 1².581.

page 38 note 3 CIL 1².11.

page 38 note 4 CIL 1².582.

page 38 note 5 CIL 8.5843.

page 38 note 6 CIL 6.14614.

page 38 note 7 CIL 8.67.

page 38 note 1 Quint. 9.4.39–40.

page 38 note 2 Vel. Long. 7.80.18 K.

page 38 note 1 Varro ap. Prisc. 2.30.14 K.

page 38 note 1 CIL 1².9.

page 38 note 2 CIL 1².581.

page 38 note 3 Fest. 293.32 M; Quint. 1.7.14.

page 38 note 4 CIL 1².608.

page 38 note 5 CIL 1².614.

page 38 note 6 Quint. 1.7.20–21.

page 38 note 7 Mar. Viot. 6.8.3 K.

page 38 note 8 CIL 10.3743.

page 39 note 1 Quint. 1.7.14, 1.4.10.

page 39 note 2 CIL 1².638.

page 39 note 3 CIL 1².365; probably earlier than Accius' time.

page 39 note 4 Kent, TAPA 43.42–52.

page 39 note 5 Ib. 43.52–5.

page 39 note 6 But the rules of Lucilius on writing ei and i preserve the historic relations correctly; cf. Kent, AJP 32.272–93, 34.315–21, Glotta 4.299–302.

page 39 note 7 CIL 8.587.

page 39 note 8 Verg. Aen. 12.316 ms.

page 39 note 9 J. Christiansen, De apicibus et i longis, Husum, 1889.

page 39 note 10 CIL 3.2. p. 774.

page 39 note 11 CIL 10.2311.

page 39 note 12 CIL 6.10006.

page 39 note 13 CIL 11.1940.

page 39 note 14 There is no inscription of any length and no manuscript, in which all the long vowels are indicated; some of the long vowels are always left unmarked, apparently through the indifference of the writer. But cf. J. C. Rolfe, Proc. Amer. Philosophical Soc. 61.80–98.

page 40 note 1 CIL 1².626.

page 40 note 2 CIL 4.680.

page 40 note 3 CIL 1².626.

page 40 note 4 Enn. Ann. 178 V².

page 40 note 5 Cic. Or. 160; Cassiod. 7.154.1 K.

page 41 note 1 CGL 5.238.

page 41 note 2 CIG 3435.

page 41 note 3 Verg. Aen. 1.393 ms.

page 41 note 4 Aus. Ecl. 17.5.

page 41 note 5 Prisc. 2.36.17 K.

page 41 note 6 Prisc. 2.24.6, 2.36.21 K.

page 41 note 1 CIL 7.8.

page 41 note 2 CIL 6.2495.

page 41 note 3 Cassiodorus 7.148.8 K states that the Greek digamma character was occasionally used in Old Latin to represent initial v; but there are no examples in inscriptions.

page 41 note 4 Quint. 1.4.11; Vel. Long. 7.54.16 K.

page 41 note 5 Prisc. 2.303.4, 2.14.10 K; cf. Kent, TAPA 43.35–38.

page 41 note 6 CIL 2.1065.

page 41 note 7 CIL 2.4157.

page 41 note 8 CIL 2.1953.

page 41 note 9 Pl. Most. 981 ms.

page 41 note 1 Tac. Ann. 11.14; Quint. 1.4.7–9, 1.7.26; Prisc. 2.15.5 K; Vel. Long. 7.75.17 K.

page 41 note 2 CIL 6.2041.

page 41 note 3 CIL 1².1.p.247.

page 42 note 1 Cic. Nat. Deor. 2.93.

page 42 note 2 Quint. 1.4.9.

page 44 note 1 The written or printed letter is only the graphic representation, to the reader's eye, of the sound produced by the speaker's breath and perceived by the hearer's ear. We should therefore not say ‘the sound of a letter’, but ‘the sound represented by a letter’.

page 45 note 1 Sturtevant, Pr. §115–21.

page 45 note 2 Mar. Vict. 6.33.1 K; Serv. 4.421.16 K; Pomp. 5.102.4 K.

page 45 note 1 Sturtevant, Pr. §122–5. In one period of. Old Latin, curiously, ō was more open than ; for while the consonantal u disappeared before the closer back vowels, it remained in the combination qvō, but was lost in qv (§153.III, §181. VI), which argues that at that time was closer than ō.

page 45 note 2 Ter. Maur. 6.329.121 K; Mar. Vict. 6.33.3 K; Serv. 4.421.17 K.

page 46 note 1 Sturtevant, Pr. §111–4.

page 46 note 2 Lucil. 9.352–5 Marx; Ter. Maur. 6.328.111 K.

page 46 note 1 Sturtevant, Pr. §115–21.

page 46 note 2 Cons. 5.394.19 K; Ter. Maur. 6.329.119 K; Mar. Vict. 6.33.2 K.

page 46 note 3 Antevocalic i was sometimes written with i longa in inscriptions, as in dIes (CIL 6.7527), prIvsqvam (CIL 6.10239); and it developed in Romance like ī, as in oFr. di, It. di, Sp. dia. It must therefore have become close in pronunciation in Late Latin.

page 47 note 1 Sturtevant, Pr. §122–5.

page 47 note 2 Ter. Maur. 6.329.138 K; Mar. Vict. 6.33.8 K.

page 47 note 3 Pl. Men. 651–4.

page 47 note 1 Sturtevant, Pr. §126–9.

page 47 note 2 Pl. Bac. 129.

page 47 note 3 Pl. Bac. 362.

page 47 note 4 Cic. Or. 160; Ter. Scaur. 7.26.13 K; Caper 7.105.17 K.

page 47 note 5 parisvma, CIL 1².7; facilvmed CIL 1².581; etc.

page 47 note 6 As in recipit and other compound forms lf capio, by the influence of reficit and other compound forms of facto; and in the -imus of the first person plural of verbs, by the influence of the endings -is -it -itis. ⁷ infimo and infvmvm, CIL 1².584, in 117 B.C.; maxvmam and maximvm, CIL 1².593, in 45 B.C.

page 47 note 8 Quint. 1.7.21.

page 47 note 9 CIL 9.2608.

page 47 note 10 Quint. 1.4.8; Vel. Long. 7.49.16, 7.67.3, 7.68.3, 7.75.12 K; Mar. Vict. 6.19.22 K.

page 47 note 1 Sturtevant, Pr. §145: Nigid. ap. Gell. 19.14.6.

page 47 note 2 oLt. ou was probably an orthographic imitation of Greek ov, the sound being already ū: iovsit (CIL 1².614), LOVCANAM (CIL 1².7).

page 47 note 3 Pl. Am. 344.

page 47 note 4 Ter. And. 534.

page 47 note 5 Lucr. 3.918; Ter. And. 458.

page 47 note 6 Verg. Aen. 1.131.

page 47 note 7 Verg. Ecl. 3.58.

page 47 note 8 Hor. Sat. 1.3.101.

page 47 note 9 Lucr. 3.1035.

page 47 note 10 Ter. And. 408.

page 47 note 11 Hor. Sat. 2.6.67.

page 47 note 12 De-hinc Verg. Aen. 3.464.

page 48 note 1 Greek ∊ɩ went through a similar development to ē and ī; early borrowings show the ē stage, as in balneum from βαλαν∊iον, where the ē was shortened in Latin before the vowel. Pēlīdēs from Πηλ∊ίδης shows the later value.

page 48 note 2 CIL 1².581.

page 48 note 3 Sturtevant, Pr. §145.

page 48 note 4 Verg. Ecl. 3.96.

page 48 note 5 Verg. Aen. 1.726.

page 48 note 1 oinvorsei, CLI 1².581.

page 48 note 2 Sturtevant, §138, §145. Greek oɩ and ῳ appear in Latin as oi, later oe; but as Greek ῳ presently became a monophthong, late borrowings represent ῳ by a mere ō: thus cōmoedia was borrowed early from κωμῳδίᾱ, while melōdia was a later borrowing of μ∊λῳδίᾱ.

page 49 note 1 CIL 1².581. At the time when ai had recently been replaced by ae, we find occasionally a transitional spelling aei, as in caeicilivs (CIL 1² 633). The writing ai recurs also on some archaizing inscriptions of the reign of Claudius, as in caisare (CIL 6.353).

page 49 note 2 Sturtevant, Pr. §130–4.

page 49 note 3 The transliteration of Greek ᾳ shows the same phenomena as that of ῳ: earlier Thraex and later Thrᾱx both come from Greek Θρξ.

page 49 note 4 Implied also by Ter. Scaur. 7.16.6 K.

page 49 note 5 Varro LL 7.96, 5.97; Lucilius 1130 Marx; Diom. 1.452.17 K.

page 49 note 6 In some words, special influences produced early confusion in spelling: caelum often became coelum by the influence of Greek κοῖλον ‘hollow’; cēna was often written coena through the influence of Greek κοινόν ‘common’, as well as caena, the erroneous late spelling. Latin scaena from σκηνή was an over-correction of a supposed rustic ē for urban .

page 49 note 7 Sturtevant, Pr. §145.

page 50 note 1 Sturtevant, Pr. §142–4.

page 50 note 2 Quint. 1.7.27; Ter. Scaur. 7.28.1 K.

page 50 note 3 Sen. Tro. 852; Mart. 1.104.22; Juv. 3.49.

page 50 note 4 Prudentius, Cathem. 12.67.

page 50 note 5 Verg. Ecl. 1.20, 4.56.

page 50 note 6 6.345 ff. K, especially verses 671–84, 760–77.

page 50 note 7 E. H. Sturtevant, TAPA 43.57–66.

page 50 note 8 Greek νɩ, which is always antevocalic, is transliterated by yi, probably y, pronounced üi-, as it develops in Romance like y (§34.III): Greek, ἄρπυια > Latin harpyja > Ital. arpia (accented on the penult).

page 50 note 1 Neuter was trisyllabic, = ně-b-ter. Neutiquam had real elision of the e, = n'utiquam.

page 50 note 2 Sturtevant, Pr. §139–41; Agroecius 7.122.13 K.

page 50 note 3 Mar. Vict. 6.66.28 K.

page 50 note 4 CIL 14.4101.

page 51 note 1 Sturtevant, Pr. §135–7.

page 51 note 2 Prisc. 2.109.22 K.

page 51 note 3 CIG 191; CIL 9.6229; CIG 6665.

page 51 note 4 Ter. Maur. 6.339, 467 K.

page 51 note 5 Cf. CIL 9.1365.

page 51 note 6 CIL 6.667.

page 51 note 7 Fest. 182.13–8 M; Suet. Vesp. 22; Prisc. 2.39.8 K.

page 51 note 1 That is, with prolongation of the sound; they were not pronounced double, as is often incorrectly stated.

page 51 note 2 Mar. Vict. 6.9.22 K; Cassiod. 7.162.10 K; Vel. Long. 7.61.10, 7.79.19 K; Consentius 5.394.25 K; Prisc. 2.46.6 K.

page 51 note 1 Sturtevant, §198a-200; Ter. Maur. 6.331.186 K.

page 51 note 2 English p is the voiceless bilabial stop, but is aspirated unless preceded by s or followed by t.

page 51 note 3 Cf. Suet. Aug. 88; Mart. 1.109.

page 51 note 4 But unassimilated pt and ps remain in Roumanian.

page 51 note 1 Sturtevant, §198a-200; Ter. Maur. 6.331.186 K.

page 51 note 2 Quint. 1.7.7; Varro LL 10.56–8; Vel. Long. 7.61.5 K.

page 51 note 3 CIL 2.34.

page 51 note 4 CIL 9.2845.

page 51 note 5 In popular speech, perhaps as early as the first century after Christ, cf. libertavs CIL 1².1330 = lībertāvus, for -ābus; Sommer, Hdb.²198.

page 52 note 1 Sturtevant, Pron. §195–7; Mar. Vict. 6.33.24 K; Ter. Maur. 6.331.202 K. Note that t d n were not alveolar as in English; that is, were not made with the tip of the tongue touching the alveolar ridge (§4.3). Also, English t is aspirated unless preceded by s.

page 52 note 2 Serv. 4.445.8 K; Papirian ap. Cassiod. 7.216.8 K; Pomp. 5.286.11 K.

page 52 note 3 CIL 8.9927. Crescentsian us in an inscription of 140 A.D. (CIL 14.246) is a very doubtful reading.

page 52 note 4 CIL 12.2086.

page 52 note 5 CIL 1².5.

page 52 note 1 Sturtevant, Pron. §195–7; Ter. Maur. 6.331.199 K; Mar. Vict. 6.33.24 K; Mart. Cap. 3.261.

page 52 note 2 Like t, §45, where references are given in note 2.

page 52 note 3 CIL 8.8424.

page 52 note 4 CIL 3.2654.

page 54 note 1 Sturtevant, Pr. §190–3d; Ter. Maur. 6.331.204 K; Mar. Vict. 6.33.28 K; Cornutus ap. Cassiod. 7.149.1 K; Mart. Cap. 3.261. English k and c (when not a sibilant) are aspirated unless preceded by s or followed by t.

page 54 note 2 The Plautine pun on Sōsiam and socium (Am. 383–4) proves nothing, for he elsewhere puns on arcem and arcam (Bac. 943).

page 54 note 3 The argument for earlier change to the sibilant sound, made by Oskar Hey, ALL 14.112, and Albert Becker, ALL 15.146, on the basis of apparent alliterations of c with s in Ausonius and in the pseudo-Quintilan Decl., is not conclusive.

page 54 note 4 CIG 3888.

page 54 note 5 Plutarch, Vita Cic. passim.

page 54 note 6 1.7.10.

page 54 note 7 Cf. Eph. Epig. 7.51.

page 54 note 8 CIL 10.7173.

page 54 note 9 CIL 8.1091.

page 54 note 10 Also by s without the accent; a careless writing.

page 54 note 11 Cf. A. J. Carnoy, TAPA 47.145–52. The products of t and of c in these combinations were not quite identical; cf. Sommer, Hdb.² 218.

page 54 note 12 CIL 6.975.

page 54 note 13 Vel. Long. 7.75.10 K; Pomp. 5.104.15 K; Donat. 4.367.16 K.

page 54 note 14 CIG 2637.

page 54 note 15 CIG 1770.

page 55 note 1 Sturtevant, Pron. §190-§194.

page 55 note 2 CIG 279.

page 55 note 3 Prisc. 2.7.21 K.

page 55 note 4 Original b is proved by Greek πέντ∊; but the length of the i came from quīn(c)tus: all vowels became long in Latin before nx and nct.

page 55 note 5 CIL 14.1386.

page 55 note 6 Isidorus, Etym. 20.16.5; cf. Probi App. 4.198.11 K.

page 55 note 1 Cic. Or. 160; Quint. 1.4.15; Ter. Scaur. 7.14.4 K.

page 55 note 2 Enn. Sc. 176 V².

page 55 note 3 Cic. Or. 160; the b may represent Illyrian mediation, cf. Walde-Hofmann LEW³ s.v. ballaena.

page 55 note 4 CIL 5.2884.

page 55 note 5 CIL 1².1204.

page 55 note 6 CIL 4.1623.

page 55 note 7 CIL 1².593.

page 55 note 8 Rarely the reverse, which was graphic only.

page 56 note 1 Sturtevant, Pr. §181a-b; Sommer, Hdb.² 199–202, Krit. Erl. 72–4; Cic. Or. 160; Catullus 84; Quint. 1.5.20; Gell. 2.3.1–4; Vel. Long. 7.69.13 K; Ter. Scaur.7.20.2 K; Mar. Vict. 6.34.3 K; Porph. ad Hor. Serm. 1.8.39; and puns in Plaut. Ps. 736, Bac: 362, Capt. 274. In Old Latin braccium (classical bracchium) from Greek βραχων, Accerūns (clXat. Ăcherōn, with learned restoration of n after the Gk. original) from Ἀχέρων, the lengthened stop was a product of Etruscan mediation; but cf. also J. Vendryes, Intensité Initiale 116.

page 56 note 2 CIL 1¹.380. This aspiration leaves no trace in Romance after t and c, but the combination ph not infrequently appears as f; Ital. golfo < late acc. golfum (ALL 2.442) < classical nom. colpus < κόλπος. F is found in Late Latin for ph of any origin: sulpur, sulphur with spurious h, then sulfur. But Greek θ, when borrowed in post-classical times with the spirantic pronunciation, might develop differently; cf. C. C. Rice, Lang. 7.259.

page 56 note 3 CIL 1². 1211.

page 56 note 4 So Niedermann² 114–5, comparing Cic. Orator 160.

page 56 note 1 Sturtevant, Pr. §188a–c; Quint. 12.10.29; Mar. Vict. 6.34.9 K; Ter. Maur. 6.332.227 K; Mart. Cap. 3.261. The sound f did not occur in Greek, so that Greeks when speaking Latin had difficulty with it; Quint. 1.4.14.

page 56 note 2 CIL 1².1420.

page 56 note 3 CIL 1².584.

page 56 note 4 Mar. Vict. 6.18.18 K; cf. Prisc. 2.29.19 K.

page 56 note 1 Sturtevant, Pr. §180–4. H does not help make syllabic position, despite Pomp. 5.117.14 K, citing Verg. Aen. 9.610; see W. M. Lindsay, CQ 10.97–9. The 350 examples of lack of elision before initial h- in Plautus (Th. Birt, der Hiat bei Plautus und die lateinische Aspiration bis zum X. Jhd. nach Chr. 55–68; Marburg. 1901) do not indicate that h was a strong spirant in his time, but are only in keeping with the large amount of hiatus before initial vowels; for Plautus also made elision freely before h-.

page 56 note 2 Quint. 1.5.19–21; Ter. Maur. 6.331.212 K; Prisc. 2.35.24 K; Albinus 7.303.10 K.

page 56 note 3 Catullus 84; Nigid. ap. Gell. 13.6.3; Probi App. 4.199.20, 200.11, 201.35, 202.35, 203.12 K; Charis. 1.103.21 K; Albinus 7.303.13 K; Augustine, Confessiones 1.18.29.

page 56 note 4 CIL 3.3750.

page 56 note 5 CIL 12.230.

page 56 note 6 ahenam CIL 1².581.

page 56 note 1 Sturtevant, Pr. §185–7; Ter. Maur. 6.332.239 K; Mar. Vict. 6.34.16 K. English s is alveolar.

page 56 note 2 CIL 1².49.

page 56 note 3 An archaic usage of which Lucretius often made use.

page 56 note 4 Cic. Or. 161; Quint. 9.4.38.

page 57 note 1 Sturtevant, Pr. §202–5; Quint. 12.10.28; Vel. Long. 7.50.9 K. English z is alveolar.

page 57 note 2 Cic. Fam. 9.21.2.

page 57 note 3 Varro LL 7.27.

page 57 note 1 Sturtevant, Pr. §201–1b.

page 57 note 2 Mart. Cap. 3.261.

page 57 note 3 CIL 10.4546.

page 57 note 4 Max Ihm, ALL 8.589.

page 57 note 5 Probi App. 4.197.28, 198.28–9 K.

page 58 note 1 Sturtevant, Pr. §173–5; Ter. Maur. 6.332.235 K; Mar. Vict. 6.34.12 K.

page 58 note 2 CIL 1².7.

page 58 note 3 Prisc. 2.29.15 K; Pomp. 5.287.7 K; Vel. Long. 7.54.2 K.

page 59 note 1 As in French; not alveolar as in English.

page 59 note 2 Sturtevant, Pron. §176–9; Ter. Maur. 6.332.236 K; Mar. Vict. 6.34.13 K; Prisc. 2.30.7 K.

page 59 note 3 Sturtevant, Pron. §178a-9; Nigid. ap. Gell. 19.14.7; cf. the attempt to use g for it, §20.

page 59 note 4 Mar. Vict. 6.16.6 K.

page 59 note 5 CIL 1². 593.

page 59 note 6 CIL 2.4587.

page 59 note 7 CIL 8.1000.

page 59 note 8 CIL 6.3375.

page 59 note 9 CIL 4.1945.

page 59 note 10 CIL 3.2377.

page 59 note 11 Vel. Long. 7.79.1 K; Charis. 1.58.17 K; cf. Sturtevant, §177a-b.

page 59 note 12 CIL 1².8.

page 59 note 13 CIL 1².9.

page 59 note 14 CIG 7206.

page 60 note 1 Sturtevant, Pr. §170–2; Lucil. 377–8 Marx; Persius 1.109; Ter. Maur. 6.332.238 K; Mart. Cap. 3.261.

page 60 note 2 Cf. CIL 3.4222.

page 60 note 3 CIL 9.2827; cf. Pomp. 5.283.13 K.

page 60 note 1 Sturtevant, Pr. §165–9; Ter. Maur. 6.332.230 K; Mar. Vict. 6.34.10 K.

page 60 note 2 Prisc. 2.29.8 K; Consent. 5.394.22 K.

page 60 note 1 Sturtevant, Pr. §156–64.

page 60 note 2 Nigid. ap. Gell. 19.14.6; Quint. 1.4.10–1; Mar. Vict. 6.5.17 K; Prisc. 2.13.11 K.

page 60 note 3 CIG 4483.

page 60 note 4 CIG 9361.

page 60 note 5 CIG 3923.

page 60 note 6 CIL 9.413.

page 60 note 7 CIL 10.2466.

page 60 note 8 CIL 10.2170.

page 60 note 9 CIG 6710.

page 61 note 1 Sturtevant, Pr. §148–55.

page 61 note 2 If v were spirant, we should have ∗faftor.

page 61 note 3 Mar. Vict. 6.18.18 K.

page 61 note 4 Quint. 1.4.10–1, 1.7.26, 12.10.29; Mar. Vict. 6.5.17 K; Prisc. 2.13.11 K.

page 61 note 5 Nigid. ap. Gell. 10.4.4.

page 61 note 6 CIG 1732.

page 61 note 7 CIG 2783.

page 61 note 8 Cic. Div. 2.84; Phaedrus App. 21.

page 61 note 9 The attempt of the emperor Claudius to introduce a special sign for v, without proposing one for j (§26), does not prove that v was already a spirant, nor that there was a greater difference between u and v than there was between i and j. Imitation of the F of Greek dialects, which was always pronounced w, was the cause of his proposal.

page 62 note 10 Vel. Long. 7.58.16 K.

page 62 note 11 CIL 10.3305.

page 62 note 12 CIL 6.7582.

page 62 note 13 Probi App. 4.199.22, 200.9,22, 201.4, 9 K; Adamanti sive Martyrii de B vocali et V vocali ap. Cassiod. 7.165–99 K.

page 62 note 14 The use of Gothic and Old English w (which in those languages represented the semivowel sound) in loanwords, to represent Latin v, does not necessarily prove the semivowel nature of Latin v, since those borrowings were made in the fourth and fifth Christian centuries or later, when Latin v was a bilabial spirant. This usage shows merely that w was the nearest approach to the Latin sound that the borrowing languages possessed at the beginning of a word (despite Gt. medial b, written b in haban). Thus Latin vīnum gave Gothic wein, oE wīn; vāllum gave oE weall. But when Germanic words beginning with w were borrowed by Late Latin, which was not before the fifth century, the w was represented by gv, which shows that Latin v no longer closely resembled w; for if it had, there would have been no reason to take refuge to such an awkward substitute for the Teutonic w: Gothic warjan ‘to protect’ gave Late Latin gvarīre, whence Ital. guarire, Fr. guérir ‘to cure’.

page 64 note 1 A syllable ending in a diphthong seems to be a closed syllable, if one may judge by its phonetic behavior.

page 64 note 2 Condensed practical rule for syllabic division : Of intervocalic consonants, the last, and the last only, went with the following vowel; except that stop + liquid always went with the following vowel unless making position, when they were divided.

page 64 note 3 The weak consonant h, when intervocalic, went with a following vowel, as in ve-hit. Otherwise h had no value at all; ph, th, ch, rh each counted as one consonant, as in A-thē-nae, A-chil-lēs; and inhabilis was divided i-nha-bi-lis. Qv normally had the value of one consonant; see III.D. Stop + liquid normally had the value of one consonant; see III. C.

page 64 note 4 The fact that e remained in the penult before stop + liquid, as in in-teg-ra, shows that at the time of the weakening of to i such penultimate syllables were normally closed and not open; since in-te-gra would have become ∗in-ti-gra; cf. faciō factus and the compound cōnficiō cōnfectus, where the short vowel in the open syllable was weakened to i, but in the closed syllable only to e. Yet before the time of Plautus, the normal syllabic division shifted to the position before the stop, since syllables containing a short vowel, if followed by stop + liquid, are regularly short in the early dramatists, as they are also in Classical Latin prose.

page 64 note 5 Probi App. 4.198.18 K; cf. also W. M. Lindsay, LL 87.

page 64 note 6 A consonant group that may begin a word, is said by the ancient grammarians to go with the following vowel (references are collected in W. Dennison, CP 1.47–68): ma-gnus u-xor pro-pter a-bdōmen pote-stās a-mnis. But this is wrong (Dennison, l.c.; cf. also W. G. Hale, Harvard Studies in Class. Phil. 7.249–71): the fact that these groups (except stop + liquid) ‘make position’, argues for closed syllables, as in om-nis potes-tās; and four fifths of the inscriptions that mark syllabic division (either by the way in which they divide words at the ends of lines, or by actual interpunctuation) favor division of the groups: pris-cilla consump-tas alvm-nae sanc-tae adiec-to modes-ta cons-tantino campes-tri caeles-tes ivs-tina op-timo.

page 64 note 7 Verg. Aen. 6.33.

page 64 note 8 Verg. Aen. 6.412.

page 64 note 9 CIL 9.1866.

page 64 note 10 CIL 4.1278

page 64 note 11 CIL 10.5974.

page 64 note 12 CIL 1².583.

page 64 note 13 CIL 6.541.

page 64 note 14 CIL 12.1971.

page 64 note 1 Hirt, Indogermanische Grammatik, vol. 5 (Heidelberg, 1929).

page 64 note 2 I firmly believe that variations of accent were the chief factor in producing the ablaut grades (§69-§75).

page 65 note 1 Cf. §122-§128.

page 68 note 1 Sturtevant, Pr. §206–14.

page 68 note 2 F. F. Abbott, CP 2.444–60; Kent, TAPA 51.19–29, 53.63–72, Rev. des Études Latines 3.204–14, Revue Internat. de l'Enseignement 45.321–35.

page 68 note 3 Prisc. 2.130 K.

page 68 note 4 Prisc. 2.130.2 K.

page 68 note 5 Prisc. 2.128.23, 130.6, 302.16 K; Arpīnā's for Arpīnātis and other similar forms are probably dialectal, cf. J. Vendryes, Intensité 174.

page 68 note 6 Prisc. 3.67.10 K.

page 68 note 7 Fest. 19.7 M.

page 68 note 8 Vel. Long. 7.53.2 K; Prisc. 3.27.8 K.

page 68 note 9 Nigid. ap. Gell. 13.26.1–3; cf. Prisc. 2.302.6 K.

page 68 note 10 A. S. Pease, PAPA 29.xxv–xxvi; Sommer, Hdb.² 340.

page 68 note 11 Consent. 5.392.4 K.

page 68 note 12 Quint. 1.5.58–64; Serg. 4.526–7 K; Serv. ad Verg. G. 1.59.

page 68 note 13 Consent. 5.392.2, 11 K.

page 69 note 1 The Roman grammarians (Diom. 1.433.19 K) apply the rules of Greek directly to Latin, and hence authorize armá-que generá-que and the like; but there are dissenting opinions (Prob. 4.145.22 K), and the evidence of the accent of poetry favors the view taken above, cf. Hor. Carm. 1.2.16, Verg. Aen. 1.2, 7.186, Ovid Met. 10.308; cf. C. Wagener, Neue philol. Rundschau 1904, 505–11, = Beiträge zur lat. Gram. 1.1–7. In the type līmina-que, ending in three short syllables, the accentuation was probably līmínaque, though Wagener argues for lī'minaque, with a secondary accent on the final syllable.

page 69 note 1 But prepositions were orthoone when used as adverbs: īnfrā, suprā; and when following their nouns: Siciliam versus; unless a limiting genitive follows, when they were enclitic: virtūtem propter imperātōris. Cum when following certain pronouns was enclitic : sḗcum, vōbscum (Prisc. 3.27.6, 3.33.1 K).

page 69 note 2 CIL 1².593.8

page 69 note 3 CIL 1².593.24.

page 69 note 4 CIL 6.35767.

page 72 note 1 H. Hirt, IG vol. 2 (Heidelberg, 1921); K. Brugmann, Gdr. 1².482–505; F. Sommer, Hdb.² 47–55; Stolz-Leumann 68–75.

page 72 note 2 Thus πατρός has -tr- before a vowel, but πατράσι (< ∗pәtṛsú) had -tṛ- before a consonant.

page 72 note 3 The accents which produced the ablaut grades were those of pIE, not the Greek accents seen in the examples.

page 72 note 4 The Gk. examples in B and C illustrate the phenomena, but are not themselves examples of ablaut unless the processes took place in pIE times.

page 73 note 1 Fairem either by syncope from ∗paterem, < ∗pәterṃ, which is the original form, or possibly by analogy to patris patrī, which had the zero grade of the suffix, namely -tr- without a vowel.

page 74 note 1 Or by a consonant; but a consonant does not increase the number of syllables.

page 74 note 2 Or ‘dissyllabic root’; there is no clear logical differentiation between the terms ‘root’ and ‘base’, except that ‘root’ is commonly reserved to indicate the ultimate element common to the largest possible group of related words.

page 76 note 1 All pIE words are of course hypothetical and should be starred; so also many later words and forms are hypothetical reconstructions, to be marked in the same way, like Primitive Latin ∗estes, which gave Latin estis ‘you are’.

page 76 note 2 This is not treated in the present volume. Cf. M. Bréal, Essai de Sémantique (Paris, 1897); J. B. Greenough and G. L. Kittredge, Words and their Ways in English Speech (New York, 1901); E. H. Sturtevant, Linguistic Change (Chicago, 1917); J. Vendryes, Le Langage 205–271 (Paris, 1921); G. H. McKnight, English Words and their Background (New York, 1923); A. Carnoy, La Science du Mot (Louvain, 1927), with history of the subject and bibliography 379–400.

page 77 note 1 These values were much modified in Avestan, by later changes.

page 77 note 2 For Oscan and Umbrian, Roman letters show representation in the local alphabets, and italic letters show that in the Latin alphabet. But if italic only is given, the representation is the same in both alphabets.

page 77 note 3 The values in oIc. and oE differ from those of oHG only as follows: oIc. a = oHG a, e; oE ā = oHG ei; oIc. and oE ō = oHG uo.

page 77 note 4 But η in Attic Greek except after ∊, ɩ, ρ; and η in Ionic everywhere.

page 77 note 5 Umb. often e.

page 77 note 6 Umb. often o in Latin alphabet.

page 77 note 7 Umb. often e.

page 77 note 8 The writings ú and í are Oscan only, in the local alphabet.

page 77 note 9 Oscan often ií.

page 77 note 1 In all dialects, ṃ ṇ ṛ ḷ in some words apparently become um un ur ul; but here the pIE was probably om on or ol, with reduced o.

page 77 note 2 Except y, w.

page 77 note 3 The three values of probably rest on pIE values , mā emә, as ablaut variations of ∗∗emā. A similar interpretation is to be made of long ṇ ṛ ḷ. But the long and , though theoretically possible, probably did not occur anywhere. Cf. §4.n4.

page 78 note 1 The other falling diphthongs and all rising diphthongs (§4.10) may be dealt with as two separate sounds, which should be looked up in their respective tables.

page 78 note 2 The short diphthongs appear in oIc. as ī, ei, ei, jō or jū, au, au; in oE as ī, ā, ā, īo or ēo, ēa, ēa.

page 78 note 3 The Oscan and Umbrian values in Roman letters are those written in the local alphabets; those in italics are the writings in the Latin alphabet.

page 78 note 4 The long diphthongs were not very common, and the table is here filled out only as far as is useful for the present purpose; for the principles of development of the long diphthongs, see §120.

page 78 note 5 This became ŋ in Attic Greek except after ∊, ɩ, ρ; and ŋ in Ionic everywhere.

page 80 note 1 The groups resulting from two dental stops (§4.17) in pIE times are here included for convenience.

page 80 note 2 The sign = before a letter means that a sound had there been lost with lengthening of the preceding vowel.

page 80 note 3 The oIc. and oE differed from Gt. only as follows: pIE p, ph > oIc. f, w, oE f; pIE b > oIc. b, f; pIE t, th > oIc. þ, s, oE s, d; pIE d > oIc. s, d; pIE dzdh > oIc. dd, oE rd; pIE q ^w, q ^wh > oIc. hw, oE hw; pIE ^w > oIc. kw, k, oE hw; pIE q ^wh > oIc. gw, w, oE ʒ, w; and wherever Gt. had k and g for a palatal or a pure velar, oE had respectively c and ʒ.

page 80 note 4 In Umb., written either p or b in local alphabet.

page 80 note 5 In Umb., written t in local alphabet; but pIE d > U ř, rs (written also s) intervocalic.

page 80 note 6 In both languages, written k in local alphabets; but earlier k before palatal vowels > Umb. ç, σ (written also s).

page 80 note 7 In Umb., written k in local alphabet; but g > Umb. i, i (consonantal y) before palatal vowels.

page 80 note 8 Only two Av. values are given, because all voiced aspirates lost the aspiration in pIr., and the representation of the voiceless aspirate is uncertain; note also that Av. γ ž δ w were developments from earlier Av. g n d b, and that Av. x, which represents the sound of Gm. ch, was commonly a development from an earlier k.

page 80 note 1 I do not believe in the existence of sh and zh in pIE, nor in the existence of the so-called þ-sounds (§4.n10).

page 80 note 2 oIc. and oE had the same values as oHG.

page 80 note 3 Lost intervocalic.

page 80 note 4 Always s in the local alphabet.

page 81 note 1 The oIc. agrees with the Gt. in these sounds, and the oE agrees with the oHG.

page 81 note 2 All nasals in Skt. and in Av. in some conditions became a mere nasalization of the preceding vowel.

page 81 note 3 But v at the end of a word.

page 81 note 4 A mere nasalization of the preceding vowel; but where the dash is printed in the lines of pIE y and w, it indicates complete loss of the sound in some conditions.

page 81 note 5 A writing for the palatal or velar nasal sound.

page 81 note 6 Various other developments when following a consonant.

page 81 note 7 Initial h regularly; the ζ appearing in some words was due to root-contamination (§4.n5).

page 81 note 8 Remaining in some dialects, but always lost in Attic.

page 81 note 1 By analogy pIE e > pIt. a: ∗epros, as in oHG ebur, > Lat. aper, Umb. abrof ‘apros’, with a by the influence of caper.

page 81 note 2 sovom gen. pl. CIL 1².727.

page 83 note 1 For changes in unaccented syllables, see §123-§127, §129.

page 83 note 2 Enn. ap Non. 504 M.

page 83 note 3 CIL 1².4.

page 83 note 4 Fest. 100.5 M.

page 83 note 5 CIL 1².531.

page 83 note 6 CIL 1².587.

page 83 note 7 CIL 1².698.

page 83 note 8 CIL 1².15.

page 83 note 9 These words were pronounced ī or e-ī, īs or e-īs, etc., never -ī, -īs, etc.; cf. Sommer Hdb.² 419–20, 347–8; Prisc. 2.298.9 K.

page 83 note 10 CIL 1².34.

page 83 note 11 Minīscor and Minerva may equally well be placed under IV.B.

page 83 note 12 CIL 1².553.

page 84 note 1 For changes in unaccented syllables, see §123-§127, §129.

page 84 note 2 The changes of o to u, when occurring after v, were not represented in the writing in republican times, for vo was retained to avoid the writing vv: volgus, avonculus; but in the early empire the writing vv = vu became regular: vulgus, avunculus.

page 84 note 3 The sound v was regularly lost after u, §181. IX; the writing ov is a means also of avoiding vv, cf. the preceding note.

page 84 note 4 CIL 1².1537.

page 84 note 5 Mar. Vict. 6.25.9 K.

page 84 note 6 Cons. 5.392.24 K.

page 84 note 7 CIL 1².082.

page 84 note 8 CIL 1².1024.

page 85 note 1 Pl. Cas. 596.

page 85 note 2 CIL 1².593.

page 85 note 3 Cf. CIL 6.1527.

page 86 note 1 pLat. ri in the initial syllable before a consonant, > ṛ > Lat. er only by analogy of the unaccented position (§124):

page 86 note 1 CIL 9.2450; but norvs may have o by analogy of socrus.

page 86 note 2 Libet perhaps received the i by frequent association with licet, as in Cic. ad Att. 14.19.4.

page 86 note 3 Cf. CIL 4.815.

page 86 note 4 Cf. Varro LL 5.71 ms.

page 86 note 5 Cf. CIL 1².1624.

page 86 note 6 Cf. §193.V; and J. Wackernagel, ALL 15.218–21.

page 87 note 1 But subtīlis < ∗sub-teg-slis, mantīle < ∗man-terg-slom, etc., had ī by analogy of adjectives in -īlis; for fidēlis keeps ē.

page 87 note 2 The examples are too scanty to give confidence in this formula.

page 87 note 3 But fīlius may rather be from ∗dhēilyos or ∗dheilyos, from which -ī- is the regular development; cf. §120.I-II.

page 88 note 1 genoarias CIL 12.934.

page 88 note 1 Cf. Fest. 100.5 M: but perhaps hemō is from ∗ĝhemō, with strong grade of the root, despite Gt. guma with the reduced grade, since the ablaut grades certainly varied in pIE in the paradigm of this word.

page 90 note 1 The Italic examples under I may be explained by III, with syncope.

page 90 note 2 Some of the Italic examples under III, with syncope, may be explained under I, with shortening.

page 90 note 1 deicerent CIL 1².581.

page 90 note 2 Cf. CIL 1².4.

page 90 note 1 bh > π by Grassmann's Law (§82.IV).

page 90 note 2 ∗deiwos ∗deiwom > ∗dos ∗dom by II, then w was lost before (§181.VI), and antevocalic was shortened (§183); from deus deum a whole paradigm was made. But ∗deiwei ∗deiwōi, etc. > dīvī dīvō, etc., whence dīvos dīvom were made.

page 90 note 3 Or ∗meiĝh-yō; cf. Walde LEW² s.v.

page 91 note 1 CIL 1².581.

page 91 note 1 CIL 1².9.

page 91 note 2 Or perhaps with ī after relīqvit, where ī developed regularly, §127.II.

page 91 note 3 Umb. vinu seems by its i to be borrowed from Latin.

page 91 note 4 CIL 1².593.

page 91 note 5 CIL 1².583.

page 91 note 6 Lucr. 3.199 noenu potest has the masc. form ∗noenus with metrical disregard of the final -s (Lindsay LL 615–6); cf. the frequent use of adj. nūllus for adv. nōn (Pl. Trin. 606; Cic. ad Att. 11.24.4).

page 92 note 1 CIL 1².9.

page 92 note 2 CIL 1².8.

page 92 note 1 CIL 1².7.

page 92 note 2 CIL 1². 634.

page 92 note 3 Fest. 115.11 M; Non. 557 M.

page 92 note 1 New eu in Latin, see §40 and §110.III.

page 92 note 2 CIL 1².549.

page 92 note 3 CIL 1².548.

page 93 note 1 CIL 1².549.

page 93 note 2 CIL 1².1.

page 93 note 3 Fest. 121.1 M.

page 93 note 4 Cf. CIL 1².586.

page 94 note 1 Or perhaps from ∗dhei-; see also §98.III.

page 94 note 2 And in primitive Greek also.

page 94 note 3 Unless by syncope for nāvifragus, which also occurs.

page 94 note 4 Ion. νης, after gen. νηός etc.

page 94 note 5 CIL 1².3.

page 94 note 6 CIL 1².379 (from Pisaurum).

page 94 note 7 CIL 1².1445.

page 95 note 1 Pl. Mil. 103; Verg. Aen. 3.354 aulāī, 6.747 aurāī.

page 95 note 2 Kent, Lang. 1.103–6.

page 95 note 3 CIL 1².443.

page 95 note 4 CIL 1².397.

page 95 note 5 CIL 1².561.

page 95 note 6 CIL 1².581.

page 96 note 1 The objections of certain scholars, notably Meillet and Hirt, to the assumption of a pIt. accent, do not seem valid to me; cf. Kent, Lang. 7.179–89 and 6.317–8.

page 96 note 2 The usual view is that syncope was the stage after the weakening of vowels to their utmost limit; but the phenomena of Oscan and Umbrian, with much syncope and no weakening, favor my interpretation. Words borrowed later from Greek, like balineum (§123.I), may show weakening and then syncope.

page 98 note 1 In Plautus the meter nowhere requires the syncopated form of this adverb.

page 98 note 2 soldum CIL 1².593.

page 98 note 3 Cf. Pl. Asin. 357.

page 98 note 4 CIL 1².582.

page 98 note 5 Cf. pūtēfacit, Pl. Most. 112.

page 98 note 6 calfacere CIL 2.5181.

page 98 note 7 Most of these vowels would be lost before an initial vowel also, by elision, as in sub-it, amb-it, ab-i , cf. Gk. ὑπό, ἀμφί, ἀπό. But syncope was a real factor, for neqve took the form nec before consonants, whereas it was doubtless pronounced neqv' before vowels, with elision, but with retention of the consonantal u.

page 98 note 8 Pl. Capt. 359.

page 98 note 9 Some scholars regard the loss in trochaic b-stems (mors, urbs, arx, etc.) as merely analogical, after pure consonantal stems of one syllable.

page 98 note 1 A Sanskrit term for the assumption of vocalic value by a consonant.

page 98 note 2 Or possibly in very early Latin; but this syncope will be spoken of here as belonging to pIt., and the resulting sounds ṇ ṛ ḷ will be labeled pIt. They must be carefully distinguished from pIE ṇ ṛ ḷ, which had already developed into vowel + consonant in pIt., before the new set of vocalic nasals and liquids developed.

page 98 note 3 There happen to be no instances where ṃ resulted.

page 98 note 4 Niedermann, Phon.² 51–2, maintains that they became in er il respectively; but forms with -ill- where -ell- would be expected more probably had i by analogy of other words or forms in which there was an original i.

page 98 note 5 Or perhaps merely with loss of w before b, §181.VI.

page 98 note 6 Fest. 87.1 M.

page 102 note 1 Or paribss < -ets, cf. Stolz-Leumann 264.

page 102 note 2 Pacuvius ap. Fest. 356.27–8 M.

page 102 note 3 Cf. §191.III.D, and Kent, Lang. 6.313–4.

page 102 note 4 Fal. peparai; cf. G. Herbig, IF 32.86.

page 102 note 5 mihei CIL 1².1206.

page 102 note 6 This change is independent of the accent; §88.IV.A.

page 102 note 7 Pl. Trin. 83.

page 102 note 8 Pl. Capt. 760.

page 102 note 9 b before ł + consonant > u independently of the accent, §91.IV.A.

page 102 note 10 CIL 1².439.

page 102 note 11 The inclusion of av and iv at this point is for convenience of presentation.

page 102 note 12 CIL 1².584.

page 102 note 13 Cf. CIL 1².727.

page 103 note 1 That is, of those short vowels that had escaped loss by syncope (§123, §124) or by elision (§199.III), or came to stand in final syllables as the result of such losses.

page 103 note 2 CIL 1².450.

page 103 note 3 By loss of the final vowel of ∗aĝesi, the original penult became the final syllable in Latin; cf. the end of note 1.

page 103 note 4 CIL 1².34.

page 103 note 5 CIL 1².25.

page 103 note 6 I do not believe that final ă remained unchanged (Niedermann, Phon.² 54): the neuter plurals dōna, genera have -a < -ā, as is shown by OU forms (Osc. comono, Umb. uatuo, trio), where -ō is from -ā. The -a of ita was originally -ā, cf. the unweakened -a- in itaque; while itidem followed analogically the -idem in tantidem, indidem, etc.

page 103 note 7 The Gk. has the reduced vowel in the radical syllable; cf. §69.IV.

page 103 note 8 Cf. Quint. 1.4.8, 1.7.22–4. So also when -s was weak and disregarded in writing: militare CIL 1².49 = mīlitāris.

page 105 note 1 I no longer hold that pIE ei in non-initial syllables became ī earlier than oi and ai: H. Ehrlich, Untersuch. ü. d. Natur d. gr. Betonung 66–79 (Berlin, 1912), and Kent, Lang. 6.310–2.

page 105 note 2 CIL 1².581.

page 105 note 3 CIL 1².614.

page 105 note 4 A new form in pLat., carried over from pronouns; the same new formation was made in Greek.

page 105 note 5 CIL 1².581.

page 105 note 6 CIL 1².9; this e denotes a long vowel sound between the normal ē and ī, further attested by the writing of ei for ē in decreivit CIL 1².614, which shows that the two sounds were at the time very similar.

page 105 note 7 CIL 1².581.

page 105 note 8 G. Herbig, IF 32.84–7; but the origin of the Latin ending is uncertain, since Praen. dedi, cofeci, coepi (CIL 1².60, 560) have -ī (< -?), and not -ei (< -ai), cf. Kent, Lang. 6.307–8.

page 105 note 9 Or perhaps ∗prō-moiriom; see Kent, TAPA 44.19–24.

page 105 note 10 For variant spellings, cf. Kent, TAPA 43.37 n.

page 105 note 11 CIL 1².583.

page 105 note 12 CIL 1².7.

page 105 note 13 For the -ei, cf. note 1 against Kent, AJP 34.320–1.

page 105 note 14 Pl. Men. 493; only the second syllable was elided before an initial vowel.

page 105 note 15 Pl. Am. 276; cf. Gell. 9.14.21–4.

page 105 note 16 Hor. Sat. 1.3.95.

page 105 note 17 The similar genitive forms are almost parallel in their developments.

page 105 note 18 So C. D. Buck, OU Gram.² 364, rather than ‘capitis’.

page 105 note 19 Sen. Tro. 186 clusit.

page 105 note 20 CIL 1².583; for fraude.

page 107 note 1 Whether the accent plays any part in this shortening is quite uncertain.

page 107 note 2 After the shortening the vowel was weakened, §126.II.A.

page 107 note 3 CIL 1².756, archaic spelling, alongside sit.

page 107 note 4 Enn. Ann. 371 V².

page 107 note 5 The shortening cannot have been finally accomplished until -d after a long vowel had been dropped, for -d after a short vowel remained.

page 107 note 6 Pl. Aul. 457.

page 107 note 7 The only neuter plurals to keep the final long vowel were the multiples of ten from trīgintā to nōnāgintā, which by becoming indeclinable were no longer felt as case forms.

page 107 note 8 Charis. 1.16 K.

page 107 note 9 Hor. Sat. 1.4.93.

page 107 note 10 Hor. Sat. 1.10.42.

page 107 note 11 Ov. Tr. 4.10.3.

page 107 note 12 Pl. Mil. 59; Ov. Fasti 3.852.

page 107 note 13 Pl. Curc. 708.

page 107 note 14 Pl. As. 30.

page 107 note 15 Pl. Am. 353.

page 107 note 16 Pl. Aul. 240.

page 107 note 17 Cat. 6.16.

page 107 note 18 Cat. 2.6.

page 107 note 19 Ov. Am. 3.2.26.

page 107 note 20 Hor. Sat. 1.4.104.

page 107 note 21 Ov. Tr. 1.1.87.

page 107 note 22 Enn. Ann. 463 V².

page 107 note 23 Kent, Revue Internationale de l'Enseignement 45.321–35, esp. 329.

page 107 note 24 Pl. Persa 786.

page 107 note 25 Pl. Poen. 1171.

page 107 note 26 Pl. Aul. 721.

page 107 note 27 Pl. Men. 259.

page 107 note 28 Verg. Aen. 5.261.

page 107 note 29 Hor. Sat. 1.9.38.

page 108 note 1 Pl. Am. 777.

page 108 note 2 Archaic soluit, Cat. 2.13.

page 108 note 3 Verg. Aen. 6.412.

page 108 note 4 Hor. Sat. 1.7.30.

page 108 note 5 Verg. Aen. 5.432.

page 108 note 6 Lucr. 4.66.

page 108 note 7 CIL 6.541.

page 108 note 8 CIL 6.21322.

page 109 note 1 Properly ‘long’ or ‘lengthened’, in sound, §42; but in word formation the term ‘doubled’ is often appropriate, since they were the sum of two sounds, as in Lat. ad-dūcō, at-terō, etc.

page 109 note 2 H. Hirt, IG 1.254–6.

page 110 note 1 I do not accept the theory of C. C. Uhlenbeck, IF 13.213–9, that the pIE voiceless aspirates first became voiceless non-aspirates in pIt.

page 110 note 2 Nor do I accept Walde's theory, IF 19.98–111, that an initial aspirate stop before r or l lost the aspiration in Italic by dissimilation, if the root ended in another aspirate stop.

page 110 note 3 On L bt and bs, see §44.

page 111 note 1 ī after quīnctus §184.II, < i by §88.IV.A.

page 111 note 2 In Osc. and Umb., kw > p regularly; §153.VI.

page 111 note 1 The b by Verner's Law; §82.II.

page 111 note 1 Or perhaps fallō is to be connected with Gk. θολ∊ρός ‘disturbed’, Skt. dhvarati ‘he injures’, oE ge-dwelan ‘to go astray’, root ∗dhwel-.

page 112 note 1 Ter. Scaur. 7.13.9 K.

page 112 note 2 Fest. 102.15 M.

page 112 note 3 Serv. ad Verg. Aen. 7.695.

page 112 note 4 Fest. 102.12 M.

page 112 note 5 The endings of these words vary, but the -bh- is constant.

page 113 note 1 See n.3.

page 113 note 2 mm > m by §185.II.E.

page 113 note 3 If Osc. Safinim came by anaptyxis from ∗safnyom, rather than from ∗safīnyom, then the change of fn to mn was not pIt. in date, but belonged to the early stages of the separate dialects.

page 113 note 4 Cf. CGL 4.308.

page 113 note 5 Cf. CIL 6.14672.

page 113 note 6 Pl. Capt. 481.

page 113 note 7 Verg. Aen. 7.424.

page 113 note 8 Gk. ἀπó, Skt. apa.

page 113 note 9 In ablaut relation to ∗epi, Gk. ἐπί ‘at, on’, Skt. api ‘close by, also’.

page 113 note 10 Gk. ύπό, Skt. upa ‘near’.

page 113 note 11 From ∗ks-upo, a compound preposition, cf. Gk. ἐξὐπ∊ρθ∊ν ‘from above’; ∗ks was in ablaut relation to ∗eks, Lat. ex, and is seen also in super, cf. Skt. upari, nE over.

page 113 note 12 Except where the prepositions were extended by -s in imitation of ex, as in abs-tineō, os-tendō, sus-cipiō; cf. §161.I.

page 113 note 1 CIL 1².4.

page 113 note 2 Not changing to -d because the -i protected it until after that change had ceased to operate.

page 113 note 3 The further fate of pIt. -nd is uncertain.

page 114 note 1 Quint. 1.4.16.

page 114 note 2 Cf. CIL 10.1249.

page 114 note 3 tw- became t- only by loss of the w (§167.II.A): so tōnsillae to Lith. tvinti ‘to swell’, pIE ∗twen-; also tesqua, tinea.

page 114 note 4 tw > t in Osc.

page 114 note 5 -a- from reduced ^e §86.I.

page 114 note 6 Analogical -ā-; Kent, Lang. 3.12–4.

page 114 note 7 Hor. Sat. 2.2.76, metri gratia.

page 114 note 8 postempvs CIL 1².2189.

page 116 note 1 Cf. §127.I and n.1, against Kent, Lang. 6.312–3.

page 116 note 2 oLat. diovem (CIL 1².558), Osc. Diúveí came from ∗diyew-.

page 116 note 3 R. S. Conway, IF 2.157–67.

page 116 note 4 Liv. Andr. apud Fest. 68.10 M.

page 116 note 5 But ciLat. merīdiēs, to medius diēs, does not belong here; it shows dissimilation (§193.V) against a following d, perhaps aided by the analogy of Lat. merus.

page 116 note 6 Cato Agr. 135.7.

page 116 note 7 CIL 1².581.

page 116 note 8 For this interpretation, see R. von Planta, Gram. d. OU Dialekte 2.652.

page 116 note 9 Fest. 26.5 M.

page 116 note 10 CIL 1².5; cf. Mar. Vict. 6.9.16 K.

page 116 note 11 Pl. Mil. 1407; Plautus was an Umbrian by birth.

page 116 note 12 Despite A. Walde, LEW² s.v.

page 116 note 13 CIL 1².4.

page 116 note 14 CGL 4.330.

page 116 note 15 CGL 4.331.

page 116 note 16 Varro, LL 5.172.

page 116 note 17 Fest. 66.15 M.

page 116 note 18 Enn. Ann. 559 V².

page 116 note 19 CIL 1².9.

page 116 note 20 Pl. Capt. 405.

page 116 note 21 CIL 1².401.

page 116 note 22 CIL 1².581.

page 116 note 23 CIL 1^2.25.

page 116 note 24 Pl. Poen. 388, where the final long consonant was sounded before an initial vowel.

page 117 note 1 mamphur, Fest. 132.1 M.

page 117 note 2 Aspiration lost by Grassmann's Law, §82.IV.

page 117 note 3 The long vowel by analogy to nūbō nūpsī; cf. Kent, Lang. 4.187 and n.

page 117 note 1 w lost by §181. V.

page 117 note 2 π by Grassmann's Law, §82.IV.

page 117 note 3 The final vowel was analogically changed in pLat.

page 118 note 1 Cf. Kent, Lang. 8. 18–26; the same processes must be assumed for Celtic and Germanic, where -ss- is likewise the usual product of -tst-.

page 118 note 2 With -b- by §143.IV.

page 118 note 3 Afranius ap. Non. 150 M.

page 118 note 4 The denominative verbs mantāre and pultāre testify to old participles ∗mantus and ∗pultus.

page 118 note 1 Cf. §131.IV.

page 119 note 1 In some words, satәm languages show variations between spirants and stops: the stops are perhaps due to borrowing from a centum language.

page 120 note 1 For pIt. k before w, see §155.

page 120 note 1 For the variation in the initial, see §162. III.

page 121 note 1 Or ∗ĝhmo-, to oIc. gpa ‘to yawn open’.

page 121 note 2 With varying finals in the different languages.

page 122 note 1 Ter. Scaur. 7.11.K.

page 122 note 2 Fest. 84.5 M.

page 122 note 3 CIL 1².166.

page 122 note 4 Varro LL 5.97.

page 122 note 5 Varro ap. Vel. Long. 7.69.8 K.

page 122 note 6 On loss of w, see §181.VI.

page 122 note 7 CIL 1².206.

page 122 note 8 Pl. Am. 512.

page 122 note 9 Present stem with nasal infix; the root is in the zero ablaut grade.

page 123 note 1 Enn. ap. Fest. 111.11 M.

page 123 note 2 Cf. Hofmann-Walde, LEW³ s.v. agnus.

page 123 note 3 Its opposite in meaning.

page 123 note 4 Possibly also after u; but lupus (with OU p, VI) < ∗luq ^wos is against such a view. Dūcō had no labio-velar; Gmc. forms (like oHG zoum ‘bridle‘) do not require it.

page 123 note 5 Lucr. 6.526.

page 123 note 6 But these had ∗g ^wo- with weakening (§125.II.C), cf. Gk. πότ∊ρος, Skt. kataras.

page 123 note 7 Of OU f in Latin, < ^wh, there is no good example.

page 123 note 8 Varro ap. Non. 77 M.

page 123 note 9 With Osc. ū < ō.

page 124 note 1 From ∗omben, cf. §153.VI.

page 124 note 1 No good examples in Italic.

page 124 note 1 From pIt. ∗leχwbs, a b-stem remodeled from the pIE fem.

page 125 note 1 Being , it was shortened by §185.II.A.

page 125 note 2 Enn. ap. Fest. 111.11 M.

page 125 note 3 No real basis for the difference of development is to be found.

page 125 note 1 When a stop was lost in a group, it was in many combinations first assimilated to the following sound, and then lost by §185.11.

page 125 note 2 This and several other examples lost s with compensatory lengthening of the preceding vowel, §164.IV.

page 125 note 3 Fest. 102.12 M; to Lat. fortis, cf. Skt. bṛhant- bṛḍha- ‘strong’, with root ∗bherĝh-.

page 128 note 1 XII Tab. ap. Gell. 20.1.25.

page 128 note 2 Pl. Rud. 142.

page 128 note 3 With st unchanged in Grac., but t > þ by Grimm's Law, §82.I.

page 128 note 1 Latin f- initial, -b- medial before r; for new -sr-, see §164.IV end.

page 131 note 1 With metathesis of the stops.

page 131 note 2 A voiced stop became voiceless before s; while s became voiced z before a voiced stop.

page 131 note 3 With parasitic -k-.

page 131 note 4 Properly parr, like ess; cf. §198.IV.

page 131 note 5 Enn. Ann. 313 V², perhaps as a reminiscence of Oscan famel.

page 131 note 6 For medial -rts- before a consonant, see IV.

page 131 note 7 But the intermediate stage -ll- medial remained in some circumstances; see §177.III.

page 131 note 8 Cf. Fest. 205.15, 209.10 M; this has -rssn- (for -rsn-, see III).

page 131 note 9 But -rstl- > -stl-, with anaptyxis (§188.I): postulō < ∗pr-s-tlā-yō, cf. poscō < ∗pṛ-sō, and precor.

page 131 note 10 For old -sr-, see §163.III.

page 131 note 11 Pl. Most. 346.

page 131 note 12 Pl. Aul. 815.

page 131 note 13 Cato ap. Fest. 379.10 M.

page 131 note 14 Prisc. 2.516.15 K.

page 131 note 15 Fest. 136.1 M; the form is somewhat Latinized.

page 131 note 16 With additional genitival -s.

page 131 note 17 CIL 1².7.

page 131 note 18 Sounds before which non-initial w was regularly lost (§181.V–VI), for swe- became swo- (§88.III.A).

page 131 note 19 Pl. Aul. 664; other examples in Neue-Wagener 2³.870.

page 132 note 1 CIL 1².592.

page 132 note 2 CIL 1².583.

page 132 note 3 CIL 10.2311.

page 132 note 4 CIL 10.5974.

page 132 note 5 Ter. And. 112.

page 132 note 6 Cic. de Or. 2.6.22.

page 132 note 7 Pl. Rud. 142.

page 132 note 8 Pl. Aul. 528.

page 134 note 1 Fest. 86.7 M.

page 134 note 2 Varro LL 7.27.

page 134 note 3 CIL 1².2.

page 134 note 4 Verg. Ge. 2.66.

page 134 note 5 Acc. ap. Prisc. 2.236.5 K.

page 134 note 6 Enn. Ann. 17 V².

page 134 note 7 Verg. Ge. 1.46.

page 134 note 8 Verg. Ge. 1.162.

page 134 note 9 Pl. Aul. 2; vowel quantity not determinable.

page 134 note 10 Charis. 1.548.29 K.

page 134 note 11 Fest. 299.2 M.

page 134 note 12 Without lengthening, as a consonant preceded.

page 135 note 1 Perhaps also after a single consonant preceded by a long vowel or a diphthong.

page 135 note 2 But this ending was replaced in pIt. by a new -āns, from which the acc. pl. of Latin, Oscan, and Umbrian developed.

page 135 note 3 This process sometimes produced an alternation of l and r in derivatives of the same root, although l does not regularly become r, nor vice versa (except in Indo-Iranian, where l regularly became r).

page 135 note 1 There is no evidence in the other dialects, so that this change may really be pLat. rather than pIt.

page 136 note 1 The long vowel by analogy of perf. ēmī.

page 138 note 1 The single -l- in scālae indicates rather that the lengthening took place when the n was lost, and that later the -ss- was lost after the long vowel without causing further change; for -ssl- intervocalic seems to become -ll-, which remained after most long vowels, §177.III.

page 138 note 2 CIL 1².8.

page 138 note 3 Citations in Neue-Wagener³ 3.552–3.

page 138 note 1 Cf. CIL 9.2305.

page 139 note 1 -rg-sl- must first have become -rssl-, cf. §161 n1, since any other development would yield -rl- > -ll-.

page 139 note 2 Fest. 302.12 M, quoting Pl. Truc. 271; probably for svăssum, with the older orthography, §21.

page 139 note 3 The long vowel was a product of contraction; it was not due to loss of the -r-.

page 139 note 4 On the evidence of oIr. druim, pl. drommenn ‘back, ridge’, Welsh trum ‘back’ < pClt. ∗drot-smen. Or perhaps dorsum is actually from ∗dē-worssom ‘turned away’; so Fest. 69.3 M, accepted by Ernout-Meillet s.v.

page 139 note 5 As in Caes. BG 1.53, and often.

page 140 note 1 Plin. ap. Charis. 1.135.13 K.

page 140 note 2 M. Niedermann, Essais d'étymologie et de critique verbale latines 36–45, in Recueil de travaux No. 7, Univ. de Neuchâtel (1918).

page 140 note 1 CIL 3.2 p.774.

page 141 note 1 Cf. CIL 1².581.

page 141 note 2 With ζ- < dy-, by contamination with the root of δέω ‘I bind’; cf. E. W. Fay, CQ 9.104–14.

page 141 note 3 Analogical weakening for -e- after j, §125.I.B; unless it is the product of syncope and samprasāraṇa, §124. Cf. M. W. Mather, Harvard Studies in Classical Philology 6.83–151; and Kent, TAPA 43.38–41.

page 141 note 4 As the meter shows; Pl. Rud. 769.

page 141 note 5 Pl. Ep. 194.

page 143 note 1 Remade to b-stem, and used as masc. and fem.

page 143 note 2 Enn. ap. Non. 504 M.

page 143 note 3 The slight difference in quality between b b b b and the corresponding long vowels (§29, §30, §32, §33) was not sufficient to prevent the loss of -v- and subsequent contraction; perhaps the qualitative difference of the sounds had not yet developed.

page 143 note 4 Ter. Ad. 61.

page 143 note 5 Pl. Am. 427.

page 143 note 6 The older writings avoided vv even in these conditions, disregarding the ambiguity, or writing ov: iventa = juventa CIL 1².1603; flovio and flvio, conflovont and comflvont CIL 1².584; cf. Kent, TAPA 43.41–2.

page 143 note 7 Pl. As. 835; or from ∗mag(i)s-wolō.

page 143 note 8 For another theory on the origin of mōvī, see W. Petersen, Lang. 4.191–9.

page 144 note 1 The process is therefore very important for the development of original long diphthongs (§120).

page 144 note 2 This process may account for the shortening of long vowels before final -m -r -l (§128.I; though not before final -t), since when the next word began with a consonant the conditions for shortening were present.

page 144 note 3 This marks a shift in the syllabic division, §62. IV.

page 144 note 4 For the -ā-, cf. Kent, Lang. 3.12–4.

page 145 note 1 The vowels were in many instances brought into contiguity by the loss of an intervening consonant; such lost consonants were -y- (§178.II), -yy- (§179.IV.B), -w- (§181.V-VIII), -ww- (§181.XI), -sw- (§164.VII.B).

page 145 note 2 Pl. Aul. 583; Enn. ap. Cic. de Senec. 1.

page 145 note 3 Enn. ap. Cic. de Orat. 3.168.

page 146 note 1 Short vowels are said by Prisc. 2.82.7 K to have been long in Latin before gn; but Romanic forms show that this was not always the case: the -e- in Ital. segno from Latin signum shows -b- (§32). Vowels before gn are therefore to be considered long only when originally long, as in rēgnum to rēx; the vowels before gm are to be viewed in the same way. Similarly, the occasional use of the apex (§22. III) over a vowel, or of i longa (§22.II), before r + a consonant and before x, as in fArtunata (CIL 6.7527), fIrmi (CIL 6.1248), máx (= máximō CIL 6.2080.17), is not decisive; for Ital. fermo shows that Latin firmus had -b-.

page 146 note 2 CIL 1².8.

page 146 note 3 Lachmann, note to Lucr. 1.805; cf. Gell. 9.6.

page 146 note 4 But note the exceptions fbssus to findō, scbssus to scindō, sbssus to sedeō, strbdus to siringō, tbssis to tundō.

page 146 note 5 H. Pedersen, Nordisk Tidskrift for Filologi, 3d series, 5.32–8; so also Sommer Hdb.² 122–3.

page 146 note 6 Analogical extensions are often not thoroughgoing, which accounts for the continued presence of b in sbssus, despite ē in perf. sēdī.

page 146 note 7 Replaced by ēgī, with ē after fēcī.

page 146 note 8 Kent, Lang. 4.181–90.

page 147 note 1 Verg. Aen. 2.664 has hocc erat in a very old Ms.

page 147 note 2 Pl. Poen. 388.

page 147 note 1 A. Meillet, Esquisse d'une Histoire de la Langue latine 166–9, who terms it ‘expressive gemination’; but not all his examples are convincing.

page 149 note 1 Pl. Curc. 359.

page 149 note 2 Association with discō is only a popular etymology; there was no suffix -pulus!

page 149 note 3 Pl. Aul. 93; this form occurs only at the end of the verse.

page 149 note 4 The Late Latin anaptyxis in groups containing a nasal or a liquid was only sporadic in late inscriptions, and can be omitted; cf. Sommer, Hdb.² 139.

page 149 note 5 The importance of this change is that it is regularly represented in French, Spanish, and Portuguese, and yields a living variation in Italian.

page 149 note 6 CIL 8.2438.

page 149 note 7 Cf. CIL 9.6408.

page 149 note 8 Retaining the -s- unlike écrite, since esprit is a semi-learned word.

page 149 note 9 Cf. CIL 5.1720.

page 149 note 10 CIL 12.1971.

page 150 note 1 Thus a double influence was exercised in favor of assimilation, which was lacking in such forms as cecidī to cadō, cecīdī to caedō, pepulī to pellō, tetigī to tangō, dedī to dō. But abdidī to abdō got its -b- by weakening, not by assimilation.

page 150 note 2 Citations in Gell. 6.9.

page 152 note 1 Or ei.

page 152 note 2 CIL 1².3.

page 152 note 3 Or eis.

page 152 note 4 Or ed or od.

page 152 note 5 Varro ap. Non. 83 M.

page 152 note 6 Lucr. 1.43.

page 152 note 7 Such contractions as these were the chief basis for the reclassifying of the pIE verbs into the four Italic conjugations; for pIE had two conjugations only, to which the ō and mi verbs of Greek correspond, while each of the four Latin conjugations includes some verbs of each original pIE conjugation.

page 152 note 8 With syncope; Kent, AJP 52.84.

page 152 note 9 Dative, Lucr. 3.918.

page 152 note 10 Regularly with the diphthong in Latin verse.

page 152 note 11 The influence of incēpī, with the same meaning, probably reinforced the tendency to contraction in this word. But co-ēpī occasionally survives, as in Pl. Merc. 533, Lucr. 4.619.

page 152 note 12 Kent, Lang. 6.313–4.

page 152 note 13 From -ets, according to Stolz-Leumann 264; but if with original ē, then no contraction would be expected.

page 153 note 1 Examples are more numerous in Late Latin and in Romanic.

page 153 note 2 German ‘Fernassimilation’.

page 153 note 3 Cf. Ter. Ph. 54.

page 154 note 1 As with assimilation, there are numerous Late Latin and Romanic examples which are here disregarded.

page 154 note 2 A. J. Carnoy, TAPA 49.101–13, gives an explanation of this phenomenon.

page 154 note 3 Sommer, Hdb.² 191.

page 154 note 4 Cf. A. Walde, LEW² s.v. But forma is rather a derivative of ferīre ‘strike’ than by dissimilation from ∗morq ^wh-mā, connected with Gk. μορφή.

page 154 note 5 Gell. 6.9.15–6.

page 154 note 6 Pl. Capt. 524, cf. Pl. Aul. 630 praestrīgiātor.

page 154 note 7 Caecil. Com. ap. Cic. ND 3.29.73.

page 154 note 1 Examples are not numerous in Latin, but are more so in Late Latin and Romanic.

page 154 note 2 Assured metrically in Pl. Curc. 406, and the probable regular pronunciation in Plautus's time, though the mss regularly have trap-. On the -ss-, cf. §24.

page 154 note 3 Phaedr. 1.25.4, 6; Mart. 3.93.7.

page 154 note 4 Both forms in Plautus; the metathesis was aided by the influence of accēdo and other similar words; cf. Hoffmann-Walde LEW³ 63, s.v.

page 154 note 5 CIL 6.11915.

page 154 note 6 CIL 3.p.806.24.

page 154 note 7 CGL 3.588.

page 155 note 1 Containing at least a consonant and a vowel, but not necessarily forming a syllable: the syllabic division was vē-nun-dō, but the psychological division for haplology was vēn-un-dō.

page 155 note 2 As in nE mineralogy, from ∗mineralólogy; popular usage retains the phonetic form minerólogy, but learned usage has restored the other vowel to show the derivation from mineral.

page 155 note 3 Venundō or venumdō is often found, as in Liv. 4.29.4.

page 155 note 1 The rhotacism of -s in Umbrian shows that -z persisted through the pIt. stage into Umbrian.

page 158 note 1 E. H. Sturtevant and R. G. Kent, Elision and Hiatus in Latin Prose and Verse, TAPA 46.129–66.

page 158 note 2 Not merely after the participial -tus -ta -tum! Also, this phenomenon can hardly be based on hypothetical enclitic forms ‘st and ‘s.

page 158 note 3 Examples, assured by metrical values: Pl. Men. 226 nullāst = nulla est; 432–3 negōtīst = negōtī est; 36 ēmortuost = ēmortuos est; 282 īnsānust = īnsānus est; 49 redeundumst = redeundum est; 494 mēcumst = mēcum est; 293 molestus = molestus es; 268 mulierums = mulierum es.

page 158 note 4 Oscan -tust represents ∗-tō est, from earlier ∗-tā est.

page 158 note 5 Pl. Most. 288.

page 158 note 6 Verg. Aen. 1.237.

page 161 note 1 See especially §137.III-IV and notes, §161.I, §164.IV, for ab and abs ā, ex ē, ob and ops- os-, sub and sups- sus-, and their changes; §145, for ad and its changes.

page 169 note 1 CIL 1².555.

page 169 note 2 CIL 1².585.

page 169 note 3 CIL 6.1479.

page 169 note 4 CIL 1².1.

page 169 note 5 CIL 1².25.

page 169 note 6 CIL 1².216.

page 169 note 7 CIL 1².582.

page 169 note 8 Cf. CIL 5.1685.

page 169 note 9 CIL 1².10.

page 169 note 10 Accius ap. Prisc. 2.30 K.

page 169 note 11 CIL 1².10.

page 169 note 12 CIL 14.4106.

page 169 note 13 CIL 5.1361.

page 169 note 14 CIL 1².10.

page 169 note 15 CIL 1².1202.

page 169 note 16 CIL 3.p.776.

page 169 note 17 CIL 1².365.

page 169 note 18 CIL 1².1578.

page 169 note 19 CIL 11.4632.

page 169 note 20 CIL 1².1214.

page 169 note 21 CIL 1².797.

page 169 note 22 CIL 1².581.

page 169 note 23 CIL 1².585.

page 169 note 24 Varro ap. Cassiod. 7.154.1–3 K.

page 169 note 25 Pl. Merc. 925 ms.

page 169 note 26 Pl. Men. 12.

page 169 note 27 Cic. ap. Vel. Long. 7.54.16 K.

page 169 note 28 CIL 6.2041.

page 169 note 29 CIL 1².1.p.247.

page 169 note 30 CIL 6.2034.

page 169 note 1 Prisc. 2.30.13 K.

page 169 note 2 CIL 4.1136.

page 169 note 3 CIL 1².12.

page 169 note 4 CIL 1².86.

page 169 note 5 CIL 1².366.

page 169 note 6 CIL 4.1136.

page 169 note 7 CIL 9.5533.

page 169 note 8 CIL 1².581.

page 169 note 9 CIL 1².582.

page 169 note 10 CIL 1².585.

page 169 note 11 CIL 1².15.

page 169 note 12 CIL 1².2191.

page 169 note 13 CIL 9.552.

page 169 note 14 CIL 1².586.

page 169 note 15 CIL 1².1214.

page 169 note 16 CIL 1².581; engraved dqv-.

page 169 note 17 CIL 1².587; engraved -tvm.

page 169 note 18 CIL 1².681.

page 169 note 19 CIL 1².7.

page 169 note 20 Prisc. 2.36.21 K.

page 169 note 21 Cassiod. 7.153.13 K.

page 169 note 22 CIL 1².581.

page 170 note 1 CIL 4.148.

page 170 note 2 CIL 2.34.

page 170 note 3 Pl. St. 384.

page 170 note 4 CIL 6.7582.

page 170 note 5 CIL 5.1667.

page 170 note 6 Enn. ap. Cic. Or. 160.

page 171 note 1 Verg. Aen. 4.116 ms.

page 171 note 2 Gell. 2.3.3.

page 171 note 3 Quint. 1.5.20.

page 171 note 4 Lucil. ap. Cic. Or. 161.

page 171 note 5 Cf. CIL 7.133.

page 171 note 6 Probi App. 4.198.29 K.

page 171 note 7 CIL 12.5869.

page 171 note 8 CIL 1².593.

page 171 note 9 CIL 3.2.p.782.

page 171 note 10 CIL 6.3637.

page 171 note 11 Varro LL 9.75.

page 171 note 12 CIL 11.3159.2a.

page 171 note 13 CIL 12.934.

page 171 note 14 Cf. CIG 5059.

page 171 note 15 Verg. Aen. 1.2.

page 171 note 16 Mar. Vict. 6.18.22 K.

page 171 note 17 CIL 5.1625.

page 171 note 18 CIG 2572.

page 172 note 1 Pl. Mil. 1331.

page 172 note 2 Hor. Sat. 2.6.59.

page 173 note 1 Sample Solution: agō : -āgēs shows a variation a : ā, found in the a : o series (§71); agō has the normal grade; -āgēs has the long grade.

page 173 note 2 Skutsch, Glotta 2.230–46.

page 173 note 3 The eu is seen in Gt. tiuha.

page 173 note 4 Or ∗weidai; cf. Add. to p. 91.

page 174 note 1 Sample Solution: In -que, e corresponds to ∊ of Gk. τ∊ and to a of Skt. ca. By the table in §79, Gk. ∊ represents (normally) only pIE e; Skt. a represents (normally) pIE e o a, sometimes ә, but the c before the a indicates a palatal vowel (§82. VI), which can be only e. The e in -qve is therefore descended from pIE e: pIE e > pIt. e in most positions (§85.I); pIt. e > Lat. e in most positions in initial syllables (§87.II, §88.II). [It will not be necessary to describe the history of the sounds so elaborately after a few have been done in detail.]

page 174 note 2 pIE e > oIr. e.

page 175 note 1 pIE ē > Gall. ī.

page 175 note 2 These are all liquid and nasal vowels, as indicated in the margin.

page 175 note 3 Or its variants, cf. §108.

page 175 note 1 CIL 1².581.

page 175 note 2 CIL 1².672.

page 175 note 3 CIL 1².366.

page 175 note 4 CIL 1².583.

page 175 note 5 Caper 7.108.6 K.

page 176 note 1 Sample Solution: In abdicō, apo- has lost the -o- by syncope because it is a short vowel in an open second syllable.

page 176 note 2 Cf. Quint. 7.9.12.

page 176 note 3 In ∗ĝheim-rinos, -i- was lost by syncope, and r became ṛ, being interconsonantal; it then developed into er. (For the m > b, see §168.II.)

page 177 note 1 In studying weakened vowels, it is of the first importance to note and state whether the syllable in which the vowel under discussion stands is closed or open.

Sample Solution: In qvīnqvennis, a was the original sound, as is shown by the a in annus, of which qvīnqvennis is a compound. Every a in medial unaccented syllables regularly became e (§125.III); e in closed medial syllables remained unchanged (§125.I.A).

page 177 note 1 Cf. CIL 1².973.

page 177 note 2 CIL 1².439.

page 177 note 1 CIL 1².25.

page 178 note 1 Verg. Aen. 2.16.

page 178 note 2 Verg. Aen. 6.33.

page 178 note 3 Hor. Sat. 2.2.76.

page 178 note 4 Cf. Hor. Epod. 13.2.

page 179 note 1 Mar. Vict. 6.26.2 K.

page 179 note 2 CIL 1².581.

page 179 note 3 Fest. 207.1 M.

page 179 note 4 CIL 1².581.

page 179 note 5 CIL 1².7.

page 182 note 1 Cf. Fest. 205.15 M.

page 182 note 2 Both from non-IE Medit. languages.

page 182 note 3 CIL 1².1.

page 182 note 1 CIL 5.4220.

page 182 note 2 Cf. CIL 14.3679.

page 183 note 1 Of course never pronounced in this way; but postulo was made up of these elements.

page 183 note 1 Verg. Ge. 4.461.

page 183 note 2 With analogical length of the vowel.

page 183 note 3 Fest. 9.13 M.

page 183 note 4 pIE dy > Gk. ζ.

page 183 note 5 Acc. ap. Non. Marc. 500 M.

page 184 note 1 Pl. Mo. 596.

page 184 note 2 So in Plautus, though mss have the later spelling.

page 184 note 3 CIL 1².33.

page 184 note 4 CGL 5.536.

page 184 note 5 CIL 8.2833.

page 184 note 6 CIL 11.5996.

page 185 note 1 Fest. 100.5 M.

page 185 note 2 App. Probi 4.198.33 K.

page 185 note 3 Cf. CIL 4.2124.

page 185 note 4 Fest. 124.7 M.

page 185 note 1 CIL 1².1420.

page 185 note 2 CIL 1².7.

page 185 note 3 CIL 1².3.

page 185 note 4 CIL 1².47.

page 185 note 5 CIL 1².359, from Norba.

page 185 note 6 CIL 1².3.