^:^^m:^^^^-''- o KM o;^/: iPififl \r\rsr\i '(^RmHr^^ P/4 THE ZARNCKE LIBRARY COI,I,BCTEB BY FRIBDRICH ZARNCKE THE GIFT OF 1893 y ^ J.6m^^ ^f/t± Cornell University Library PA 2424.W23 3 1924 021 622 950 Cornell University Library The original of tiiis book is in tine Cornell University Library. There are no known copyright restrictions in the United States on the use of the text. http://www.archive.org/details/cu31924021622950 One of the most certain and frequent results of the un- conscious attempt to lighten difficulties of pronunciation, which plays so important a part in the history of the Indo-European languages, is rhotacism, or the change of s to r. The condi- tions under which this change takes place are different in the diiferent languages, and consequently the extent to which they are affected by it, differs as well. I shall endeavor in the following dissertation to examine the character and extent of this change, so far as concerns the old Italian languages, to consi- der the conditions under which it occurs, and to discuss the exceptions to its operation. A. Latin. In no one of the Indo-European languages has rhotacism been so wide -reaching in its effects as in this one of the Italian group. Next to the change of diphthongs to simple vowels, no one influence was so powerful in giving to the Latin language the shape in which Cicero and Caesar used it, and the abun- dant remains of the literature furnish us with a comparatively copious material, from which to draw our conclusions respecting the law of its occurrence. Ehotacism takes place in Latin, (1) regularly, when a single s is brought by inflection or word -formation to stand between 1* — 4 - two vowels; (2) in the nominative singular of most masculine substantives of two syllables with stems in s and of some neu- ters; (3) before nasals and voiced consonants in a few cases; (4) by assimilation in roots where the combination rs occurs. I. Between two vowels. This is the best-known and most frequent in its occurrence of all the cases in which rhotacism appears. Following is a list of words where it is certain or highly probable that it has taken place. The list could without doubt be increased, if examples of the earliest Latin were more numerous. Cf. Corssen, Aussprache, I^ 229sqq. a- s belongs to the root. ara cf. asa. Ter. Scaurus, 2252 P. Osk. aasa, Umb. asa, Volsc. asif. Sab. asum. arena asena. Varro. L. L. 7. 27. fasena. Velius Longus. 2230 P. Awelius Auselius. Festus p. 23. aurora Sk. ushdsd. aurum Aurunci auris ausculto. vid. Curtius. Gr, Etym. n. 619. ero, eris etc. eso. VaiTO 1. 1. Sk. asmi. eram, eras etc. Gr. ItfTi. Eso, "quoted by Varro from the Carmen Saliare, is un- doubtedly genuine,- and probably other forms could also have been brought forward from the older monuments of the language; esum, on the other hand, also found in Varro, L. L. IX, 100, is an inference only, and in this case an incorrect one. The e must have been dropped before the u was developed into a full vowel; otherwise we should have instead of sum, *erum of which there is absolutely no trace. erim is-te, Osk. ei0o, Umb. 'eru. Festus p. 162. feriae fesiae. Paul. Epit. p. 86. festus. 'Furius Fusius. Quint. I. 4. 13. Pom- 'ponius, Digest 1. 2. 2. § 36. gero gestus. haereo haesus. haurire haustus. Lares lasibus. Paul. Epit. p. 264. Varro L. L, 6. 1. maereo maestus. muris mus. nares Sk. nasa. 0-H-G. nasa. nefarius nefas. oris OS. quaero quaeso. queror questus. ruris rus. sero satus. The present is redupHcated with rhotacism from the ro^ soror Sk. svasar. speres spes. sperare uro , ustus. virus Sk. visKam. Zd. visha Corss. Auss. I''. 533. Curt Gr. Etym.'n. 591. b. 5 bell ongs to th^ stem, aceris cf. acus »x 6 — aeris aes. Sk. ajas. Umb. ahes- nes. a/rhoris arhos and arbosem. Paul. Epit. p. 15. angoris angustus. auguris augustus. Cereris Ceres. cineris cinis. clamoris clamos. Quint. I, 4, 13. coloris colos. Plaut. Mil. 1179. corporis corpus. crepera crepusculum. cruris crus. cucumeris cucumis. curiones cusianes. Varro L. L. 7. 26. Corss. Auss. 1^. 353. (V). decoris decus. Etrvria Etruscus. facinoris /acinus. Falerii Falisci. favoris faustus. femoris. femus. App. Met. 8,31. Appuleius is indeed a late writer, but he was familiar with the earliest Latin , and fond of tising archaic forms , and there is no reason to doubt the genuineness of this form, which is vouched for by trustworthy manuscripts, vid. Neue Formen- lehre. I. 579. feno'ris fenus. floris flos. Osk. Fluusai. foederis foedus. frigoris frigus. funeris funus. generis genus. gliris glis, glisis, restored by an almost certain conjecture in Naevius. Com. 65. ed. Ribbeck. glomeris glomus. gloria Sk. gravasya. vid. Corss. Auss. I^. 360. gloris glos. lieri hesternus. honoris honos. injuria jus. janitoris janitos. Varro L. L. 7. 27. (?j juris jus. laioris lalos. lateris lotus. leporis lepos. leporis lepus. Moris litus. major ibus maiosibus. Paul. Ep. p. 264. maris mas. melioribus meliosibus. Paul. Ep. 1. 1. morihus mos. muneris munus. nemoris nemus. numerus Osk. Niumsius. Numerius odoris odos. Plant. Capt. 815. oleris olus, helusa, Paul. Ep. p. 100. oneris onus. operis opus. Osk. upsannam. Papirius Papisius. Cic. ad Fam. 9,2 1.2 pectoris pectus. 8 — pecons pecus. penoris penus. pignoris pignus, pignosa. Fest. p. 213. plurimus plusima. Varro L. L. 7, 26. pluris plus. ponderis pondus. puberes pubes. pulveris pulvis. picris pus. rauderis raudus. rohoris robustus, robosem, Paul. Ep. p. 15. roris ros. ruder is rudus. rumoris rumusculus. sceleris scelus. spware Spurius stercoris ■ telluris temporis tergoris ulceris umerus Valerius velleris Veneris veteris pusula. vid. Curt. Gr. Etyin. n. 652. Snovdiov. Dion. H. III. 34. stercus. tellus. tempus. tergus. ulcus. Sk. asa-s, Goth. amsa. Gr. ajtti(ra3..Curt. Gr. Etym. n. 487. Valesius. Paul. Ep. p. 23. vapos. Lucr. VI, 953. vellus. Venus, venustas. vetus. — 9 — Veturius Vetusius. Liv. III. 8, vires vis. visceris viscus. vomeris vomis. Verg. Georg. vulneris vulnus. c. In the inflections of nouns. The ending of the genitive plural of a and o stems, rum, stands for original sam. Compare the genitive plural of the Sanskrit demonstrative pronoun tesJidm with Lat. is-torum Gothic 4M0O. So pennarum for *pennasam, honorum for *bonosom etc. d In'the inflections of voice. The formation of the passive is most simply explained by the addition of the reflexive pronoun se to the active. Rhota- cism took place at a very early period, after which the final e was dropped, and simple r remained in the popular conscious- ness as the passive sign. In spite of the hesitation occasioned by the presence of the same passive sign in Oscan and Keltic, where rhotacism has not as yet been proven in other cases, this theory deserves the preference before any other as yet brought forward, because of its simplicity and the completeness of the explanation which it furnishes. The s of the Oscan genitive' ending shows that this dialect, shares in the tendency to alter the sound of 5 between two vowels which was the chief cause of rhotacism in Latin, and the manifold changes of s in Keltic under similar circumstances show the same thing. It is quite conceivable that a tendency which at a very early period affected several different lan'guages alike , may have been deve- loped differently in two of them, and in a third have been deve- loped scarcely at all. In this way is the fact satisfactorily ex- plained that the only trace of rhotacism in Keltic is the passive sign in question, and on the other hand that the disappearance of — 10 — s between two vowels, very frequent in Keltic, is also not without example in Latin. Vid. Corss. Beitr. p. 465. So amor for * amose. regor *regose, darier *dasjese. Vid. Corrs. A uss. II. 478. agier * agiese. e. In the formation of verbal substantives. (I). The infinitive active was originally the dative of a substantive, cf. Sk. vakshe, maishe etc. So dare for * dase, amare for *amase etc. (2). The infinitive passive was formed in the same way with the addition of a nominal suffix, which is still to be found in the I; the passive sign was in the later language regularly dropped. So dari for dasi, Paul. Ep. p. 68, amari for *amasi etc. Vid. Corss. 1. 1. f. In the personal endings of Verbs. As in the substantive verb the s of the root has suffered rhotacism, so in the personal endings of verbs where this root is used. So in the pluperfect -eram, -eras etc., future-perfect -ero, -eris etc., perfect subjunctive -erim, -eris etc., imper- fect subjunctive, -arem, erem, Srem, irem etc., perfect indi- cative 3"* plural -runt. Corssen, Aussp. I^, 230, explains the form coemise as for coemerunt; if this is correct, we have one example where the s of the ending still remains ; but whether correct or not, there can scarcely be any. other explanation of this ending runt, except that it is for sunt, earlier esont. Paulus Epit. p. 121, gives the form loebesum as archaic for liberum, and in the same sentence loehertatem for liber- — 11 — tatem. The difficulty in reconciling the s and r was perceived by Mtiller. If loebesum is correct, *loebestatem or at best *loebesitatem would be the only regular form; for that the words belong to two different periods of the language, in one of which rhotacism had already taken place and in the other not, is hardly possible, and rhotacism before t is without example. The Oscan word luvfreis increases the difficulty, for only on the strongest possible evidence can we accept rhotacism there. It seems clear to me therefore that the s is due to a mistake of the transcriber, or possibly even of Paulus himself, who may have confused the adjectiye with the name of the Sabine deity Loebasius or Lebasius. The .corresponding part of Festus is lost. That the word was brought forward to illustrate chiefly the ancient diphthong, may be inferred from the Greek illustra- tions he gives, loifiriv and Isi/Ssiv. Corss. Beitr. p. 201. Miiller has restored loebeso in Varro L. L. VI, 2 for libero of the manuscripts, which must be false. But it is as easy a con- jecture to write loebero, and nearer to the manuscript reading, and Varro's object, to bring examples of words to which custom had given a different shape, would be just as well attained. Lasibus is the next example he gives , in which rhotacism would receive its illustration. II. In the nominative singular of masculine substantives with stems in s, and in neuters. arbor cf. arbos, arbustum. clamor clamos, Quint. I. 4, 13. color colos. honor honos, honestus. janitor. janitos. Varro L. L. 7, 27. Cross. Aussp. P. 213. (?) - 12 — labor laios. odor odos. pavor pavos. Naev. Trag. 45. Ed. Eibbeck. timor timos. Naev. 1. 1. vapor vapos. Quint. 1. 1. The comparative ending ior in the nominative masculine singular has arisen by the change of final s to r. The original form of the suffix was jans. melior of. melios. Varro, L. L, 7, 26. major magis. Further in augtir augustus. angor angustus. fulgur fulgus. Paul. Ep. p. 92. robwr rohus. Paul. Ep. p. 264. femwr femus. / rumor rumusculus. So also in ciner cinis. cucumer cucumis. pulver pulvis. vomer vomis. Vid. Neue Formenlehre. I. 176. veter vetus. Varro. L. L. VI. 2. jecur jecusculum. Since Sk. jaJcrt and Gr. ^naq both show an r where it can hardly be supposed that an s originally stood, it may ap- pear doubtful whether the diminutive alone, which is not at- tested before Cicero and could have been formed after the false analogy of arbusculum, munusculum, would justify us in assuming that an s was once found in the nominative. But the formation of the Sanskrit and Greek words is certainly — 13 — different from that of the Latin, as is shown by the t occur- ring in both the other languages, but of which no trace is to be found in Latin; further by the forms jocimis, jocinoris, (Charis. L p. 47. K.), which show the Indo-European suffix as unmistakeably. Since now the suffix as is so frequent in Latin, and in at least three cases, rolur, femur, fulgur, has taken the form ur in the nominative singular of a neuter noun, we are justified in. assuming that this was the case mth. jecur and that the' diminutive was correctly f6rmed. Simitur. This form is found only once in an inscription of the Augustan period in the same sense as the Plautian word simitu, and the r is explained by Corssen as -the result of rhotacism, (Beitr. p. 400) while simitu was formed by drop- ping the s before rhotacism had taken place. But as Bugge remarks (Curt. Stud. IV. 350), if simitu is to be derived from the adverb simitus, the length of the u is not easily explained. The Plautian word is rather the ablative of a sub- stantive ^simitus, formed from the denominative verb *simire, from the stem simo-, old Latin semo-. The form with r, on the other hand, if correctly written, contains the same suffix as igitur, the origin of which is not clear. It is not necessary to assume that it originally ended in s; except in the doubt- ful case of quirquir, there is no instance in Latin of final s suffering rhotacism, when not induced by analogy. Instead of the Oscan passive forms proving that final s sometimes suffers rho- tacism, they furnish one certain instance, if the ordinary expla- nation of the passive be accepted, of rhott,cism between two vowels in Oscan. The e'must have remained in the.. passive until rhotacism had taken place, as is shown by the compara- tively late appearance of final r for final s in Latin and Um- brian. StiD less do Osk. amfr Umb. ambr, ampr prove rho- — 14 — tacism of final s in Latin. Even if amfr be not a formation tike super (Mommsen. U. D. p. 249) , which seems to me more probable , the r even then may be due to rhotacism between two vowels ; neither the Oscan nor the Umbrian word occurs except in composition, with the root i , "to go", and the most natural explanation, if it is a parallel formation with dficpig, is that after rhotacism the vowel before the r was dropped. Vowels have been dropped in Oscan and Umbrian in other cases be- tween labial s and r. Vid. Corss. Auss. 11.^ 535. Beit. 354. Varro in some old formulas' of the augurs read quirquir for quisquis, L. L. 7. 8. Johannes Schmidt, Zt. XIX, 196, explains the r as arisen from rhotacism of s brought between two vowels by the addition of the enclitic i. Of. Gr. ovzoffi, Goth, thizei; so quir for *quisi. Ehotacism would in this case have taken place as in dirimo for *disemo, after which the final vowel fell away as in amor for *amore from *amose. This ex- planation is in perfect accordance with the Latin phonetic laws and the Umbrian pisi gives a perfect analogy in meaning, and the hesitation occasioned by the absence of any other trace of this enclitic particle in Latin, is more than counterbalanced by the difficulty in beheving that contrary to all analogy in Latin and with no such external reason as the oblique cases of veter furnish, rhotacism of final s should have taken place in priestly formulas which contain several archaic forms, and in which arbos has not yet become cMrior. in. Before nasals and voiced consonants. (1). Before m. carmen cf. Sk. casman Carmentis Casmena. Varro L. L. Carmena. VII. 26. — 15 — (2). Before n hodiernus dies. Biespiter. Vid. Corss. Aussp. P. 232. verna Sk. vas, "to dwell". Vesta. Curt. Gr. Et. n. 206. veternus vetus That rhotacism took place after the addition of the suffies and not before, is clear when we compare other derivatives in which the derivative suffix begins with a sharp consonant, Dies- piter, vetustus, the latter of which words is at least as old as veternus. There are besides no forms from the stem of dies in which rhotacism has taken place to suggest the change in the nominative as is the case with veter for vetus. ' That a monosyllabic stem should suffer rhotacism, that vas should have become *var or *ver, is also contrary to all analogy, cf. mus, ros pus etc. Finally the form Casmena is one incontestible example of rhotacism occurring through the influence of a nasal, which we may therefore accept for the cases given above. (3). Before v. Larva cf. Lases. Minerva Sk. manas "spirit," "mind". Grassmann, Zt. XVI, 113, considers the r here the result of rhotacism between two vowels, that *Lasva, through vocali- zation of the V became *Lasua, then through rhotacism Larua, iinally through hardening of the u to v, Larva; similarly with Minerva. But v usually remains unchanged in Latin after s; there is indeed no clear instance of its vocalization except in suus, which Plautus constantly measures in one syllable and in a few words evidently peculiar to poetry, suadeo, suetus, etc.; in suavis, consuetus, suasum it remains generally a con- sonant. The vocalization of the v therefore which we find in — 16 — Larua, the oldest attested form in Latin (Ritschl, Opusc. II. p. 599), can well have taken place after rhotacism; the near relationship of «> to u renders it not at all improbable that it should sometimes have the same influence on a preceding con- sonant as M would have; carmen proves that rhotacism before a voiced consonant has actually taken place in Latin, and there is no trace of a form like *Menerua to justify us in inferring a form *Menesua. IV. In roots, by assimilation. Vid. Corss. Beit. p. 401. currus Sk. harsh, "to draw". errare Goth, airseis. far for farr from Umb. farsio. fars ferrum Norse u. Eng. brass. horrere Sk. harsh, "to shudder". porro Gv. naqdw porrum nqdaov. It is unlikely that the last two words are borrowed from the Greek, for in the later history of the language rs was assi- milated to ss not to rr. Assimilation to rr has at all events taken place , and if these words are borrowed, it may have been due to the different pronunciation of the Greek r, of which there are elsewhere traces (Curt. Gr. Etym. p. 446), that the usual later course of assimilation was reversed. Gr. terra, torrere Sk. ta/rsh , "to thirst' zigtiofiai. terrere Sk. tras, "to tremble". turris Gr. TVQdig. verves Sk. varsh, "to fill". verrere Sk. karsh "to draw". — 17 — In the superlatives, acerrimus, maturrimus, veterrimus, minerrimus etc., it seems certain that the second r is due to assimilation, what- ever the origin of the suffix. The change of s to »• by assimilation in the case of an s arisen by assibilation , is easier to accept than ordinary rhotacism. The only instance of rhotacism of such an s is in the second singular of the passive, and there, as the Sanskrit, Zend and Church-Slavonic show in comparison with the Greek and Latin , the assibilation of the t is so an- cient that it can well have escaped the popular consciousness, and the s may have been subject to changes which other letters, simi- larly formed later, did not suffer. But even of the assimilation of such an s there is no certain example , and a theory which avoids this would deserve the preference, if no other difficulties stood in the way. Such a theory Roby seems to me to have proposed in his »Latin Grammar", p. LXI, according to which the super- lative ending umo is added directly to the comparative stem, as in plurimus, and by a strong contraction is produced the superlative in use. So *miseriosimus, *miserisimus, *misersimus, miserrimus. We are not able to fix accurately the period when rhota- cism in Latin was most active and when it ceased to work. The written records of the language contain many allusions to it, which are however mostly incidental and show only that the fact was well known to the Roman rhetoricians and grammarians. The instances of the retention of the s where the later usage shows r are extremely rare in the literature proper. With the exception of arbos, colos, honos, Idbos, and almost solitary instances of a few other words of like formation, absolutely no such forms are found in Latin literature. Our knowledge of — 18 - them is derived from citations by the grammarians, from occa- sional allusions in other -writers, and from comparison, with other forms in Latin itself and in kindred dialects and tongues. The testimony of well-authenticated inscriptions would of course be the best possible, but even here it would have to be considered whether particular reasons had contributed to pre- serve the older form, after the ordinary speech had suffered the change. Of the two instances furnished us by inscriptions, one presents no difficulty. Lases is vouched for not, only by this inscription, but by Varro and Festus as well. The song of the Arval Brothers, which is preserved to us in a copy of 218 A. D., is undoubtedly one of the oldest monuments of the lan- guage, and there is no difficulty in ascribing the song to the time before the Papirius or the Claudius to whom traidition as- cribes the introduction of the r. The word viasieis on the other hand, C. I. L. No. 200, is more difficult. The first syllable is conjectural, but tjiere is no reason to doubt its correctness. This inscription contains fragments of an Agrarian law, the date of which is 111 B. C, a time, when, from all other indications the impulse to rhota- cism had long ceased to operate. The orthography of the In- scription is as a whole that of the Gracchi period, inclining in- deed in some respects to the more modern system e. g. in most cases puiUcus, instead oi poplieus. Coelius in a. letter to Ci- cero, Ad Fam. 8, 6, 5, speaks of a lex viaria, an expression which it is natural to connect with the viasieis of the inscrip- tion,, in which therefore rhotacism must have taken place in the sixty years between the inscription and the letter. But it is hard to believe that while miser, asinus and nasus remained unchanged, this one word alone should have be.en rhotacized, obeying a law which was in all other cases pow^rjess. In the same connection is found in Coelius alimentariam, and the suffix — 19 — seems to be the same in both cases. Now Corssen has shown, Beitr. p. 331 sq , that in many cases at least there is no reason for supposing that an s was ever present. Even the only example he gives of the suffix asio passing into , ario, Pmasius, rests upon a conjecture of MuUer, a probable one, indeed, but still a conjecture, and it would still be uncertain whether the suffix was the same as the frequent Latin suffix ario. The suffix ari& is found in Plautus in alearius, lapida- riUs, singulanus and others, in Cato in (Mrtieuilarius, molarius, in Terence in jocuJarms, palmarius, and there seems to be no trace of an s in any of the manuscripts. In inscriptions of about the same time with the Agrarian lajv of ill stand admorsa- rium, arvorsario (No. 198), aerarium, aerario (No. 197), janua- Has (No. 199), vicarium, viearios (No. 202), and in this law itself stipendiarieis. In no later inscription is s retained. Ad- jectives in aris, from which in some cases the corresponding words in arius have been derived (Corss. Beitr. p. 336), are also to be found in Plautus and the older writers. Festus cites many obsolete and rare words in a^ius and aris and it is sur- prising, if the s was still to be found in the older Latin, that Festus, who thought it necessary to explain w&osem uni pig- nosa, made no mention of these still more remarkable rarities. It' is not likely that the Umbrian suffix asiu is the same as the Latin cmo. Rhotacism has gone farther in Umbrian than in Latin and the formation of the few Umbrian words which show this suffix is not entirely clear, asiu may correspond to the Latin asio in amasius, where the s having arisen from assibi- lation, did not suffer i-hotacism. This suffix is not common in ■ Latin, but is at least attested, and can very well be present in Umbrian, though stems corresponding to Latiji stems in nt have not yet been pointed out there. In view of these considerations, it is impossible to resist 2* — 20 — the conclusion that viasieis, if not due to an error of the tran- scriber, which is hardly probable, considering the otherwise excellent orthography, is either, first, an instance of the retention in legal language of a form already obsolete in the popular speech. It is possible that provincial peculiarities may haye contributed to this result also ; for those persons entrusted with the care of the highways would find their chief employ- ment in the country and in the neighborhood of the smaller villages, and it would be natural to designate them by a term already in use in such regions. Or, second, what seems to me a more probable supposition, that viasieis is another instance of a formation like amasius, from the present participle of viare, which was later used in the sense of ire; this must have been however a revival of a word long fallen into disuse (cf. Quint. VIII. 6. 33), and this meaning may have been an infer- ence from viator; the notions to go and to mahe one^s way are moreover often nearly synonymous in English. Viarius on the other hand is formed like molarius, palmarius, from the stem via-. With this agrees the fact that viasieis is sub- stantively, viarius adjectively used. The citations by Varro, Festus etc. are generally unaccom- panied by definite statements as to the age of the examples brought forward, but there can be no doubt they were in most cases taken from authentic sources. Of course it is much more likely that old forms have been modernized than that newer forms have been archaized. So while we may suspect that some modern forms have usurped the place of older, there is less reason for doubting the genuineness of the rarer and more ancient words, especially when they proceed from so trustworthy and learned authorities as Varro and Verrius Flaccus But all must have been taken from sources older than any inscription which we have, many no doubt from the ritual of the priests, — 21 — which was in actual use at a comparatively Jate period, and which preserved archaic forms and words long after they had vanished from the popular, speech. Our oldest inscriptions contain such forms as dedrot (C. L L. 173), curia (177), aire (181), dederi (177), fourio (63), plowume, d/imnoro (32), Jionos but Jionore, quairatis (34), morihus, majorum, laetantur, generis (47) etc. The „Mamuri Veturi" of the Salii, Varro L.L. VI. 45 is perhaps incorrectly transcribed with r for 5; we know at all events from Livy that Vetusius was written at a time probably long after the composition of the college of Salian priests. It can hardly be doubted that as far as concerns rhotacism the language had assumed its classical shape substantially, at the time of the first Punic war. With this agrees what we learn from the other data fur- nished us by the literature. Cicero ad Fam IX. 21. 2, says: "Post hunc (L. Pap. Mugillanus, censor) XIII fuerunt sella curuli ante L. Papirium Crassum, qui primum Papisius est vocari desitus; is dictator cum L. Papirio Cursore, magistro equitum, factus est post Romam conditam CCCCXV". So Pomponius Dig. 1, § 36. says. "Idem Appius Claudius (who built the Appian way, censor CCCCXLI)R Htteram invenit". Of course, if "Papirius" was written some twenty years before Appius* censorship, Appius can hardly have found the letter r, but it may well be that he was the first to bring into general use in writing what had before been recognized only in the spoken language. Vid. Osann. Comm. ad Pomp. 1. 1. Orthography generally lags behind orthoepy, and if Appius first made general the use of written r for written s in certain classes of words, we can safely assume that for some indefinite time previously the orthography had not agreed with the pronunciation., Accor- ding to Livy, III, 8, Vetusius was at least written CCXCI, and nothing prevents us from assuming that it was so spoken as — 22 — well. As long as the restraining influence of a recognized liter- ature did not exist, orthography would remain nearer to the actual pronunciation, even if it did not represent it actually; on the other hand, for the same reason changes would take place much more rapidly, and one hundred years would probably suffice to bring about the change which we are discussing. But if this were the case, that rhotacism did its work in the com- paratively late period between Veturius and Papirius, it would be difficult to account for the fact that no trace remains of the s of the passive and of the genitive plural, where the comparison of Sanskrit and other languages enables us to assert with greater or less confidence that s was once found. It can of course be mere accident that an example of each has not been preserved, just as the s of the infinitive is preserved to us only in the one form dasi, given by Festus, which could easily have been lost. But when we compare the Umbrian and Oscan passives, and the genitives in arum and asimi, it seems more probable that here at least rhotacism was much older, perhaps even older than the separate existence of the Latin language. I should prefer therefore to consider the statements of Cicero and Pomponius as indicating a period at which we may assume that rhotacism had ceased to be so active an ele- ment in shaping the Latin language. The few instances of s in proper names in later inscriptions are due almost without doubt to an archaizing tendency, which in some cases shows itself as well in other ways. In one class of words we are able to watch the completion of the process of rhotacism, viz: in the nominative singular of dissyllabic masculines with stems in os. Some of these words remained in good usage till Cicero's time with nominatives in OS. honos occurs far more frequently in Cicero than honor, mrlos is found often in poetry, labos in Sallust and Catullus, — 23 — colos and odos in Lucretius and Plautus, and other words are cited by grammarians , and lepdr for Upos, though given in the lexicons , seems not to occur at all. cf. Neue I. p. 167. It is probable that many other examples could be given, if the manuscripts in every instance gave exactly the words of the authors, but as the IMer forms ended in eveiry ilnstance in r, the force 'of analogy may haVe brought it about that many nominatives in s as uSed by the authors, yielded in the manu- scripts to those in r. There would be no aj)parent reason why a copier shoMd change labor tO lalos, while the opposite error would bfe a natural one. It is impossible to say why a few words so retained the s, while the great mass so early changed it to r. It may pet-hafs be connected with a popular feeling that drbos, tabos etc. were primitive words, not like amos timos, pavos, clantos, which are attested indeed, but Only once each and not in the best age of the literature, derivatives from verbs in common use. RhotaciSm before voiced consonants and by assimilation to a priett6diag r can not be traced in the. existing monuments of the language, s does not occur before jM, n, v, in Latin words, whUe r is very common, and it is not unhkely that in some cases rhotacism has taken place. But the etymology and formation are often obscure, and nothing definite can be asserted, especially since ill most instances where we can trace s with certainty before sonants it is dropped, cf. qualunij remus. The combination rs ih later Latin was regularly assimilated to ss instead of to rr; cf. prosSus, rusus, advoseAith etc. There is no exftmple of the reverse process in the sources tO which we have access, and the proper name Hwrutus (Osann. Comm. Pomp. 1. 36) is either an entirely different word from the ad- jective Mtsutus (cf. hirrio. Paul. Ep. p. 101), or is due to an error bf the copyist. - 24 — Phonetically, the change of s to r is not difficult to explain, and must have been • substantially the same in all languages. The sharp spirant s demands a greater precision and firmness in the position of the organs of speech than is necessary for the articulation of the movable trilled r. s is pronounced by sending the breath through a very narrow aperture formed be- tween the tip of the tongue and the upper gum; a retraction and elevation of the tongue toward the palate, give the sound of sh; a projection against the teeth gives either t or th, according to the completeness of the contact. It will have been observed that rhotacism occurs in Latin chiefly between vowels and before or after flat or voiced consonants. Now a little negli- gence in sharply distinguishing between voiceless s and the voiced letter which precedes or follows, changes the sharp s to the flat z. This is justly assumed by Corss. Ausspr. l\ 295, as the transition stage to r. If during the articulation of s the tongue is not held firmly but its tip is withdrawn and narrowed, the result is r, either the ordinary English r, which is articulated without trilling, or the r heard in some regions of the continent, where the tip of the tongue itself is set in motion. The r thus produced would in no case be the guttural r, such as is heard in Scotland and most continental countries. That this dental r was the ordinary Italian r can also be inferred from the well-known change ot d to r in Latin and Umbrian. Vid. Corss. 1. 1. That such was the ordinary course of rhotacism can also be inferred from the forms still found in Oscan, such as egmazum, poimd^ censazet etc. in Gothic, such as thizos, aizis^ and in Umbrian, menzaru etc. Though this would sufficiently account for all the cases of rhotacism in Latin, the actual course of things seems, to have been different in one class of words. That final s could become r without the influence of analogy is plain from the Umbrian - — 25 — and Greek; but it is surely no accident which restricts rhotacism of final s in Latin to those substantives whose oblique cases show a rhotacised s. Aside from this there is no trace of a tendency to rhotacism under such conditions, and it must be the analogy of the oblique cases which has brought it about here. This is confirmed by the fact that most neuters remain unaffected ; there was no *temporem in the accusative to induce a *tempor in the nominative. It is also curious that in two of the four neuters where this change seems to have taken place, femur andjecur, there are two or more different stems for the oblique cases. Assimilation need not be discussed particularly, but it may be remarked that the progressive assimilation presumed here, though contrary to the general usage of the Indo-European languages (Sievers, Grundsuge, p. 137), is not without example in Latin and Greek. Compare velle for vel-se and Gr. noQqm for noQtioa. It is universally recognized as a law of the language that s, standing between two vowels, is changed to *■. How is it then, that we can hardly read a page of Latin without finding words where s stands between two vowels unchanged, in many words has, so stood from the earliest time to which we can trace them? If the exceptions are so numerous, is it worth while to lay down the rule? Upon consideration however, we shall find that most of the exceptions can be classified under a few heads, which can be easily explained as not contradicting the- law when properly stated. (1). Foreign words. We have seen above that the language had ceased to feel the impulse to rhotacism by the time of the first Punic war, — 26 — and that probably for a long period before this it was no longer felt as a shaping influence. There is no certain instance of a word borrowed early enough to be subjected to its influence, for ius followed Latin analogies in its declension , just as mina and thesaurus did, and probably such a form as *tusis never existed. The greater number of foreign words was introduced after the fifth century, and as might be expected, they have preserved an s between two vowels unchanged. So from the Greek, Ephesus, AncMses, pMlosophus, nausea cite; from the related Italian diakcts, Aesernia, Ausonia; from the Keltic, Esus, gaesum, omasum etc. In most instances the foreign derivation is obvious; a few words where this is not so plain will be discussed below. (2). Compounds. When s is brought by composition between two vowels the compound nature of the word seemed to be present to the popular consciousness, and the s remained unchanged; desino no more became *derino than de sua *de rua. So with prae- sideo, prosequor - reservo, trisulcus, odorisequus, vesanus, nisi, etc. Gaussin (M6moires de la societe de Ling. I. 2. 128) points out a peculiarity in the pronunciation of French corre- sponding to this retention of s in compounds. Though s \)e- tween two vowels has regularly the soft sound, in some com- pounds it is pronounced like initial s. Thus present with soft s, but preseance with sharp. Similarly in English presume with soft s, but presuppose with sharp. That this consciousness of the compound nature was the chief cause for the retention of the 5, is indicated further by the fact that prepositions, though in pronunciation connected with the following word, are not joined with them in writing at all before the time of the Gracchi, and until late fu the imperial times are in by far the great majority of cases writteq — 27 — as separate words. Corss. Aussp. IP, 862. So also the no infrequent cases where the preposition of a compound is written separately, and the cases of tmesis in the poets, indicate the same thing. Initial s never suffers rhotacism; we should expect then, if rhotacism takes place at all in compounds , to find it in such compounds as have final s in the first part. It is not difficult to understand how a final s, brought by composition before a vowel, should become r, for all the conditions are present under which rhotacism regularly takes place. It would be harder to understand how an initial s could be affected by a final vowel of a preceding word. For however words may have been divided in writing, in pronunciation it is hardly possible but that often the consonant was divided between the two' syllables. That the Roman carefully divided vi-ri or pu-e-ri-lis is incredible. The fact that it is especially between two vowels that s suffers rhotacism , points to the same thing. Now it is not difficult to suppose that a final consonant is joined in pronunciation to the initial vowel following, forming practically a part of the initial syllable, a thing which occur* regularly in French and often in other languages, but that an initial consonant be joined to the final vowel preceding is contrary to all experience. These conclusions we find agreeing with the facts. The only cases of rhotacism in compounds are dirimo and diribeo, where r for s belongs to the first part of the compound. These two words, with cisalpinus, a comparatively late formation, and the dna^ Isyofisvov, dishiasco, are the only instances of this com- bination. The s was preserved in the last case by the A, which was probably distinctly sounded; moreover the word was too seldom used to be subject to many changes; the h of haieo has disappeared in other compounds, cf. praeieo, probeat. Spite of Pott's etymology, which makes disertus a compound of dis — 28 — and the root of ars, (E F. I. 66. n. 92) the word does not belong here; for aside from the fact that we should expect an r according to analogy, (a desire to avoid two rs so near each other could have preserved the s), a much more satisfactory meaning is furnished by the ordinary derivation from dis and sertus. Compare the constant use of dissero, "to discourse", and bisellium for the short vowel and single s, and for the differ- ence in treatment between disertus and disseptus, compare comedo and coeo, where com is treated similarly. The intensive force of dis moreover is at best rare, and is easily derived from its strict meaning, whereas in disertus from dis and ars, this is not easily done. (3). In many words where a single s occurs between two vowels, it did riot originally stand alone. This is shown in many cases by the older spelling, and' in many others by in- disputable etymologies. Here the consciousness of the omitted letter, which showed itself probably in a sharper pronunciation, preserved the s until the period when the danger from rhotacism was past. The perfects and supines of verbs with stems in d or t, often in the earlier Latin spelled with double s, have in the later peiiod lost one of them. So rasi, trusi, plausi, misi, ussus, casus, divisus, suasus etc. So also causa was written by Cicero with double s. Vid. Quint. I. 7, 20. The suffix osus, as in formosus, odiosus etc. has lost the lettern; compare /ormonsMs, grammonsis. Corss. I.^ 254. Simi- larly the suffix esimus in the ordinal numbers is for ensimus. Corss. I.^ 645. So in the suffix ensis the w often disappears, as Alhesia, Megalesia etc. Lachmann restores vensicula for vesicula in Lucr. VI, 130, where the manuscript gives vessicula. Vid. Lachman ad Ic- So quasi for *quamsi. These three classes of words embrace by fair the greater number of exceptions to the general rule, and in most cases — 29 - the apparent anomaly can be explained easily by bringing it under one of these heads. There remain a few words however> which present greater difficulties, and these must be examined separately. Some words which belong only to postclassical Latin, as amasio, camisia, salisatio etc., can be disregarded, as being either borrowed or formed after a false analogy. Only those words concern us, which, either belonging to the original native stock, or having become thoroughly naturalized, show a single s at a period when rhotacism had accomplished its work. Words like adasia and pesestas therefore, attested by Festus, apparently belonging to the ritual of the augurs, do not need to be explained; if remaining in the popular speech at all, they may have suffered rhotacism, as indeed is certain in the case of adasia, if the form adariarius in Forcellini is correct. casa. From the root skad, "to cover"; a participial form with the ordinaiy change of < to s *skadta, *skassa, *hassa, casa. Corss. Beitr. p. 448. Curt. Gr. Et. n. 112. fusus. From the root spad, "to twitch", "to move quickly"; with the ordinary assibilation and with aspiration of p , fusus for *spudtus. Curt. Gr. Etym. pp. 247, 494. caseus. From the root skad, "to cut into bits" , comes the participle form 'casus for skadtus, lengthened by the suffix eo, caseus, that which is divided, separated. Compare tf^t^w yii^cc, "to curdle milk", in Dioscorides. c/tCw comes from another form of the same root. quasillum. Evidently from the participle *quasus. cf. qua- lum for *quaslum. From the root kart, "to split", with the usual assibilation and with labialization, *quarsus, *quasus, "the split or cloven thing". From *quasus with the suffix lo, qua- lum, with the diminutive suffix, quasillum. By metonymy the — 30 - material gives the name to the thing made. Cf. aena, "a copper vessel". vasa. From the same root, with dropping of the guttural. , Compare currus and verrere, both from the root 1c(wsh. The older form vasum points to a dropping of the ending urn, as in nihilum, noenum, after which the word passed into the Z^"^ de- clension in the singular. Corss. Aussp. 11.^ 594. The original meaning is that of a vessel for containing, which remained its chief meaning. ' suasum. From the root svard, a lengthened form of svar, as seen in sordes, swart, suasum is therefore for *svarsum Curt. Stud. V. 243 ff. rosa. This word was horrowed from the Greek ^oSsa, qoMa, with the same consonant changes as in Claudius, Clausus. Curt. Gr. Etym. n. 515. Hehn p. 516. nasus. The ordinary view of this word is that it is connec- ted with naris, Sk.ndsa, O-R-G. nasa. The formation is the same, the meaning the same, and the identification seems unavoidable. We have then in nasus a word of genuine Latin stamp and . found in the earliest literature, retaining the s through all periods of the language, and , existing side by side with it, and from the same root, and found at least as early as Lucilius, nares which has nearly the same meaning, but has suffered rhotacism. Such an irregularity is in the highest degree surprising. We are justified in looking with distrust at a view which involves the disregard of a well-established law of the language, and in seeking for some other explanation in harmony with it. It is possible that a partial reason for the retention of the s in this case is, that the distinction between nostrils and nose was thus preserved. The distinction is however hardly so im- portant that of itself it had been sufficient to keep the s, when all others, similarly situated, were undergoing the change to r, — Bl- and indeed nares is often hardly to be distinguished from nasus in its use. The chief cause must rather be sought elsewhere. Does the mere fact of similarity in structure and meaning compel us to regao-d two words as etymologically connected, ' when the one disregards a. fixed law of the language? The answer is plain when we consider udus and unda, fe-mina and fe-cun- dus. In these words we have, it is true, other than phonetic reasons for denying the relationship; but if iyqag and uJc- shami had been lost, who would have doubted the common ori- gin of unda and udus? It is also true that it is very difficult tO; find a satisfactory etymology for nasus, but we have at least one indication that a consonant has disappeared before the s in the word nctssUerna, used by Plautus, and explained by Festus. The manuscripts give the double s, which Scaliger would change to the single, to favor his etymology, nasus, ter- nus, "a watering pot with three noses", an etymology highly improbable. The connection with nasus however is most prob- able ("caiicem nasorum quattuor" in Juvenal, and the correspon- ding use im modern languages), but the correct inference from that would be rather that nasus was fomerly spelled with, ss, than that nassiterna was spelled with s. We have, it is true, no other indications of the double s in nasus, but in other words, where it was. certainly found, e. g. casa, caesus, there are also no traces in the manuscripts. Corssen I.^ 434, explains the ss, as in caussa, but the formation of caussa is not entirely clear and it is hard tq see why nasus must receive another suffix before being fitted to represent the spout of a watering pot. The word is evidently of comparatively late formation, when nasus must have ab-eady reached its definite meaning nose whatever its original mea,ning. I have thought with some hesitation of deriving nasus from an Indj^iEuropean root nard, "to produce a nOise" "to sound". — 32 — preserved in the Sk. nard, "to cry", "to scream", nasus formed from this root like suasum from svard, woidd thus signify that organ through which the murmur *of the breath is heard , or even if we call it "the snorer", the idea is not more laughable than that which the ordinary derivation calls up, according to which the nose is "the flowing organ". There is no reason to doubt that the primitive Indo-European snored, like his descendants of the present day, and that is a peculiarity at least as striking as the discharge which accompanies an influenza. That the hardy shepherd people from which sprung such a vigorous ci- vilization, should have suffered from a perpetual cold in the head, so that the most prominent feature should have received its name from it, is hardly likely; that after their days of ac- tivity, they should have slept soundly and snored, is very prob- able. If this be granted, the relationship between nares and Sk. ndsa will of course remain undisturbed, and we may still in^ dulge, if we wish, in the pleasing task of picturing to our ima- gination our flowing-nosed ancestors. But I -cannot help thinking that it is perhaps not quite so sure that we are justified by the facts before us in painting them so. That this peculiarity can give a name to the organ is proved by snout, German Schnauze. The Greek fivxTrjQ, compared with Sk. mmJc'dmi and Lat. mucus and Ch-Sl. moh-na-ti, seems another -instance of the same thing, as also ^wd-wv, which however is used almost exclusively of the nostrils in the plural, where this designation is not so surprising. But the more examples are accumulated, the more ground there is for suspecting that some one of them is false. That the Indo-Europeans should have been so im- pressed by this one property of the nose, as to employ no other characteristic in giving it its name , is highly improibable. A ►simpler derivation would be from the root aw, "to breathe", by — 33 — -metathesis na, in accordance with which the nose would be designated as »the breather*. So as ne-far-ius from the root bha, would nares be derived from the root na, from which with equal ease the Sanskrit and German words could be derived. For the nature and process of the metathesis vid. Johannes Schmidt. Zt. XXIH. miser. Here again the same difficulty presents itself as in nasus. In maereo we have an apparently related word, where however as maestus shows, rhotacism has taken place; but the cognate languages do not give us any similar formations to help or hinder a decision, and the agreement in meaning is not so precise as in nasus and nares, just as in general words with an abstract sense are harder to define precisely. It would perhaps be a sufficient reason for the retention of the s here, that by this means the accumulation of rs would be avoided, so unpleasant to a Roman ear. *mtrer would be disagreeable, and a form like *mirerorum would be intolerable. It seems to me however that an additional reason can be found in the etymology, miser and maereo are indeed related, but are not to be derived immediately from the same form. The root mith, lengthened from mi, has in Sanskrit the meaning among others "to do an injury"; this sense is given only in the cata- logues of the grammarians, it is true, but as* the simple root sometimes means "to injure", it is not at all improbable that the lengthened root should have the same sense sometimes. From mith, with the usual changes and shortening of the vowel as in casa, *misa , with the suffix ro, *miserus, miser, that is, "one who has been injured". From another lengthened form of the same root, which appears in Sanskrit mish with the meaning "to fight against", but which appears in the cognate languages to have more nearly preserved the sense of the single root, comes maereo, with vowel intensification and rho- 11 — 34 — tacism. Both roots mitJi and mish are found in Sanskrit, both are lengthened forms of mi, from them can be derived with strict observance of the phonetic laws and with a natural development of meaning, the Latin words, and I can see no reason therefore why miser should be regarded as any more singular in the re- tention of the s than casa. The meanings of miser and maes- tus, "wretched" and ''sad", are dissimilar enough to make a difference in etymology not at all improbable, and similar enough to account for the instances where they are practically syno- nymous. caesius. This word is rare and used only of the eyes, and is explained by the ancients as equivalent to yXavKu-mg. The earliest instance of its use is in Terence, Hec. 440, where Donatus explains it as "glaucis Oculis, quasi felis oculos habens et glaucos". Both here and in Heaut. 1062, the caesius oculus is evidently regarded as a blemish. So in Lucr. IV. 1161, "Caesia Palladium, nervosa et lignea dorcas". Catullus 45,7, uses caesius of the lion, Vitruvius of the northern tribes. Its- use in all these cases is widely different from that of yi,av- Kconig, which is used almost exclusively of Athena, and in the sense of "gleaming eyed". In Cicero. N. D. I. 30. 83, the eyes of Minerva are said to be caesii, as those of Neptune caerulei, but Cicero has , evidently in mind the representations of the gods by the Greek artists and poets, and uses caesii as a matter of course as the translation of yXavxwnig. So Gellius explains the Latin by the Greek word and derives it from cae- lum, "quasi caelia" Gell. II. 26,19. Its actual usage is how- ever much more nearly that of the Greek yXavwg, which also in the later Greek had the meaning "blue-gray", but which originally meant rather "gleaming", Curt. Gr. Etym. n. 133 b. That caesius too can hardly have signified color in its briginal sense, can be inferred from the superlative caesissimus, Varro — 35 — L. L. VIII, 39, 122; for though of some definite color like white (cf. alhissimum , Varro 1. 1.) or green, comparison can with a certain justice be used, it is difficult to conceive of a superlative of so vague a hue as "bluish-gray". Gellius was so far correct in his etymology, that caesius is related to caelum, if not derived from it. From the root Jcu "to be hollow", was formed the verb *cavere; from the present participle was formed a noun stem *caventio-, "having the characteristics of being hollow"; with disappearance of the V and assibilation , caesius. It is therefore the adjective to caelum, but derived not from caelum itself, but together with it from a common root. The specific meaning "bluish gray" can have arisen either through specialization of the general sense from an occasional color of the sky, or, which seems to me more probable, it can have followed the same course as 'ylavxos, having first acquired the meaning "glowing" from the brilliancy of the skies in the southern atmosphere. indusium. From induo. The present participle stem, length- ened by the suffix io gives *iMduentio, with assibilation and disappearance of n Hnduesio, with assimilation of e to t of the following syllable, Hnduisio, (Corss: Aussp. 11.^. 365), with contraction of ui to u indusium. The contraction of ui to u is certain in anu, aspectu etc., Gell. IV. 16, and is very prob- able in the middle of a word in indutiae for Hnduitiae. agaso. This word must be connected with agere, but cannot be derived directly from the Latin verb. It is rather to be derived from an obsolete denominative verb ^agare from "agus. Cf. otqav-'Tiyog ^ysid^ai,. From the present participle is deri- ved a^rasJMS, as amasius from amant-, with the suffix on, and disappearance of the i, agaso. So formosus for *formontius, *for- monsius, formonsus. equiso. From stem equit-, the with suffix io. through assib- 3* — 36 — ilation, *equisso, equiso. For the double s, cf. odiossa, quoied by Corssien Beitr. p. 480 from Marius Victorinus. asilus. This is the Latin word for thte gadfly and is deri- ved from ad and sed, weakened from sad. The d of the pro- position is ^.ssimilated and disappears as in fusus, tasu'i, dstare, aspergo, etc; the a remains short, like the o in positus. ad appears in composition with the samfe root in assiduus, ^here the s does not disappear; so s in possider^ Also refti&ins. asilus is therefore "the sitting animal", which by alighting oii the cattle, drives them to madness, cf. Verg. Georg. III. 148. posui. The s of the simple verb , \Vhi'ch, ill dOnsequence of the dropping of the i in the present, must also be dtWppfed, pono for *posno, *posino, reappfears is sttoh as th^e vocializiatioh of the V permits it, and as the first lettfer of the seCon^d part of a compound does not suffet rhotacism. Moreover the final r of the preposition was assimilated to initial s. So that oriigi- nally double s was pronounced. This protected the ^, aftei* the consciousness of the compound nature of the word had -Vanished. Corss. Aussp. I.^ 420. pusus. Very rare, perhaps found onfy in the epigram 6f Papinius , Varro L. L. VII. 28, but its derivatives pusillus and pusio are found in all periods of the language. The root is undoubtedly pu »to beget*, and the word is either folrHfed ftXitCi the simple root, like cavssa from slcti, for whifeh puMi , V6rg. Catal. 9. 2 gives a starting point, or comes from the lengthened root pus, Sk. push »to nourish* like haesus from Mes. The formation of caussa is not ctear, and I should therefore pi-efer the latter supposition. For the difference in quantity betw^n pusus and pusillus, compare moles and m&lestus. CorsS. Beit. p. 248. pausa and pusula stand in the same I'elation to the roots pau and spu as pusus to pu. quaeso. The s in this word has been preserved to distin- — 37 — gwph it fro^ qff^erPi wjiich does not have the sense »beg«, i^^e^pecli* , {he only Q}^i^§if^ pi quae$o in classical Latin. So f^P^V'A *n^ fiftfi^ ^r^B distinguished from each other by different jQl^a,niijgs, ^p ^ueJfl^t^ is foujid in erdinary prose only in the compound iper.du§lUo. The perfect and supine of this qlder form were transferred tq t|ij? younger, after the tendency tp rhotacism had ce?.sed to operate, when quae^o was scarcely used except in yie present ten^e an,d paj-entjietically. hasium. Not foupd ea,rlier than Catullus. Compare Old- Eng. hqpse, Germ. Jms, Prov. Germ, bussed, diminutive of h^s, bu§^m, s-to kiss«, Sv- puss, »kisg«, jpussa, »to kiss«, Ppl. bffs^af »|;o kiss«, Ijr. and G^tel. bus, pus, »lip, mouth*, bu- sQg, fkig^f?, W. bus »lip«. It is h^j.'dly likely tiat tly^ so widely gpre9,d .w/?Fd ya^ borrowed from the L9.t;in. It Y^as rather borrowed from the Celts by the Romans, and pos§i|)ly by the ,Ce!tg ^911^ t^e Germans. The German words sepm of genuine najtive ?jtock. pisy,jm, From th.e jQpspk Tiiaov. Npt found .earlier than .Qpli^ro^la, ifv^ile thp Gr^^ word js found in Eupolis. On this ^Qupjt, ^j^d J)ecause the RiP^mans in g,!! probability received the pjifint from ;the Greeks, it seems to me more probable that it is bjOrEoyed jt^an |that it is formed from pinso- The proper naigp f^go .on the ^hey h3.nd, found yery early) is to be deri- y^ frpja;i the v/erb pifiso, piso, formed Jjke /SQi^edo, comUbo. gjif,t. ^r. Etym. p |^6 c. flehn. p. 38^, ,^?^fi«^f f ^s i| gej^e^^Uy regarded as pf Semitic origin, cf. pe|l?. f^hon , fthe she-^sp*, "With assibilation pf the t and S. %ee]^ terflQinatipu *6avog The name ]ffith tlie animal wan- dere4 tp Jtajy, pro))ably wjtli t^e early colonists of Cumae. Ther? tjti^g^ jthg ,dijGficulty j^jj^cji the Rpmans felt ii^ prpnoiin- ing t^ coii(ibip3,ti,9p §n, tj)i§ word assumed its classical shape ill J^&tu^, Sipiil^rly ^ ,i V.as inserteql before r in ^ina, tecma — 38 — etc. Corss. Aussp. II.^ 262. From Italy to the Celts, Ger- mans and Slaves. Benfey, Gr. Wurzellexicon, I. 123. Hehn. Culturpf. pp. 114. 502. This certainly corresponds with the probable history of the ass, and there are no special difficulties in the way of accepting it. There would be ample time for the ass to become familiar to the Itahans in the time between 400 B. C, when, as we have seen, rhotacism may be presumed to have spent its force, and the time of the elder Cato, when the ass was in general use on the Roman estates. That ' the ass came into the peninsula with the first entrance of the Italian race, as Hehn conjectures, seems to me improbable, for the ass in its further progress at least followed the cultivation of wine and oil, and Hehn has shown that wine and oil were introdu- ced into Italy subsequently to the original settlement by the Italian race. Caesar, caesaries. That these words belonged to the older language, is shown both by the early appearance of the proper name in history, and by the declension of caesaries, which pre- supposes a *caesarias. Corss. Aussp. H.^ 348 f They are evidently related to Sk. Jcesara, Jcegara, of which the etymology is not clear. Both Sanskrit forms are well attested, but the Latin word makes it probable that the form with s is the ori- ginal. Initial g in Sanskrit sometimes represents original s, Curt. Gr. Etym. n. 20. Sk. gvaguras, Lat. socer, and with this agrees Bopp's explanation of Jcesara as a com- pound; but it is impossible to derive the Latin sa^ from the root M, from which he derives gara. If a cpmpound at all, it is perhaps rather to be connected with the root svar, »to shine*, in accordance with which the hair would be the glory of the head«. The Latin usage of caesaries agrees with this etymol- ogy, for in Latin the word has generally the meaning of some- thing ornamental, beautiful. But however this may be, the 5 — 39 — could not have been retained on account of the compound na- ture of the word. Even if it was originally a compound, the consciousness of this fact must early have faded away, and if there were no other hindrances, the s would have been rhotacised. A variety of causes contributed to preserve the s into the classi- cal period. Rhotacism would have caused the accumulation of »"S so unpleasant to a Roman ear, and the popular etymology connected it with caesus and with Kaeso , which I connect with caesus. An indication of this is preserved in Nonius p. 556. M. »Caesaries dicti qui caesa matre. nascuntur* , and in Phny 7. 9. 7, »a caeso matris utero*. The snifix ar however renders this derivation improbable, and indeed Festus, p. 44, reverses the process, and derives Caesar from caesaries. Nonius'' and Pliny's derivation is only another instance of the so frequent in- vention of a story to explain a word. There are besides a number of words, as asio, pausea, Drusus, immusulus, resina, cisium etc., which show a single s. The origin of most of them is obscure, but there is nothing to prevent us from supposing that they could be brought under one of the classes of exceptions mentioned above. The dr of Brusus points to a foreign origin; the double s of immusulus in some manuscripts, points to a lost consonant ; ^tiviv^ was prob- ably the origin of r/'sina, and so with others. B. Falerian. The remains of this dialect are too meager to enable us to discover its relations to the question of rhotacism. In Abelese and Plenese the s seems to be the same as in Megalesia, Alh- eses in Latin, where its retention is due to the n which has disappeared; in Cesula, the s may be the same as in caesus cf. CaeselUa, but these can be only conjectures. The near — 40 - relationship to Latin would lead us to suppose that here also rhotacism would take place', but we have no certain evidence of it. That the of senatuo, Zertanea represents the weak sound of s which we have seen was the first step towards rhotacism, is highly improbable; rhotacism of initial s is un- known in the Indo-European languages. C. Oscan. The Oscan, in agreement with the conservative character of the people, is in most points more archaic than the Latin, so also in regard to rhotacism, s occurs constantly between two vowels; so aasas, fluusal, the future endings sid,, set, the suffix asius etc. The only example is the r of the passive, and this must stand or fall with the ordinary theory of the origin of the passive. See above p. 9. Examples are saharater = sacratur, compm-ascuster = comparata erit. The start towards rhota- cism seems to have been made in the weakening of s to e. So aeum the ending of the genitive plural, cf. Sk. sam, Lat. arum; -azet, d. fusld, Gr. -ddsi; eisac cf. eisah etc. It is worthy of note that the older inscriptions in the Oscan alphabet show the s,. the younger in the Latin alphabet the s. It must not however be thence inferred that the change of s to ,^ is due to Roman influence. At the end of the second Punic war, which is about the period of the treaty between Nola and Abella, in which g is not found, rhotacism had practically finished its work in the Latin language, and intercourse with the Romans could at most have hastened the development of a tendency already present. We could perhaps trace its full devel- opment if the language had survived. — 41 - Other possible instances are veiy few and at the best uncertain, hafiert seems to be incorrectly written for *hapiest, Kirchhoff, Stadtrecht etc. p. 43. The r can not be due to Roman influence at least; none of the Indo-European languages shows rhotacism in this form of the substantive verb, and no other instance occurs in Oscan of r in this ending, while numer- ous forms in st are found. So also other manifest errors occur in the Oscan side of this table, docud for dolud, fepacid for fefacid. Mommsen U. B. p. 152. One inscription contains the name niumeriis, probably another form of niumsis , niumsieis. The r can only be due to Roman influence, which is in this instance not improbable. The name Numerius in the Fabian gens, where the s was early rhotacized in Rome , must have been known to the Sam- nites, and it is therefore not necessary on this account alone to assume that the inscription is very young; it could have been engraved indeed by a Roman artist , to whom perhaps the statue of Pallas can also be attributed. Momm, U. D. p. 282. For amfr. see afove p. 14. Aurunk is probably dialectic and local. Bugge. Stud. IV. 351. ' D, Siibellian. The only example of rhotacism in this dialfi^ is the passive sign r. ferenter = feruntur. Corss. Zt. IV. 1%: s between two vowels occurs constantly, where in many cases the corre- sponding Latin word shows r. So fasena = arena, asum ?= ara, esuc, cf erim, flusare = florali. Most of the Sabellian inscriptions are very old, bu"t even the latest of them show no such trace of a tendency to rhota- cism as is furnished by Oscan s. So the very latest, which Corssen sets about at the same period as the decree de BaQ- — 42 — chanalibus, contains the form flusare, preserving the s un- changed between two vowels. E. Umbrian. In accordance with the generally younger character of the Umbrian, whose decay from its primitive condition must have been much more rapid than that of the Latin, rhotacism has in general gone much farther here. The early contact of the Umbrians with the Etruscans may have hastened this result, in so far as much intercourse with those who speak another language tends to destroy those influepces which keep a lan- guage within its old barriers. (1). s between two vowels regularly suffers rhotacism as in Latin. So from the root as comes the infinitive erom. The same root employed in the formation of the different tenses suffers rhotacism under the same circumstances, henust but henurent. The ending of the genitive plural is arum as in Latin, in Oscan asum, cf. Sk. sam. The compound pronominal stem seen in Oscan eisaJc, Sk. esha appears rhotacised in Um- brian erir,'erar etc, as well as in Latin erim. The Umbrian is however not so consistent as Latin. In benuso, covortuso the s remains, but can perhaps be explained, (Biiai. Les Tab. Eug. p. 361); so also in asa, Latin ara. Auf- recht p. 103, holds that s belonging to the root remained un- changed, but erom would in that case be esom. It is not certain that the s belongs to the root in any of the words where it stands between two vowels , except the two words cited above, while in many it can be proved that a consonant has been lost. aseriato ha'fe the related form anseriato , FisiM but also Fissiu, frosetom cf Lat. frausus, eso but also essu. eesona is not clear butCorssen's supposition, P. 378, which connects it with the Sabine — 43 — aisos, is most probable. In that case it may be a borrowed word, as is true of some of the proper names. In ooserdome, which according to Br6al, p. 49, = Latin * aviservaculum the s is the initial letter of the second part of a compound. It is possible that the retention of the s in asa is connected with the length of the original root vowel; at- all events the cause can not lie simply in its being a root consonant. The passive sign r is also found in Umbrian, as in Latin and Oscan. emantiir = Lat. emantur, terhantivr = tergeantur. (2). Final s of certain inflections becomes r. So in the nominative singular of stems in io: in the genitive singular of stems in a, o, i, u; in the nominative, dative and ablative plural of stems in a, o, i This occurs without exception in the later tables with Latin letters, and is found in a few cases in the older tables with Etruscan letters. Examples are Atiersir, ivengar, prinvatur, pacrer ; totar, popler, ocrer, trifor; dequrier, Atiersir- Cf. Br6al. p. 336. This is not in the least analogous to those cases in Latin where final s becomes r. There, as we have seen, rhotacism was due to the analogy of the oblique cases acting upon the final s of the nominative;, here the final s of the oblique cases changes to r itself without any such external cause. It is impossible to say why stems in a and o were more affected than others: we only call attention to the fact that it is just these stems in Latin which differ most in their declension from the original Indo-European forms. In Fesnere we have according to Br6al p. 337 an. example of the rhotacism of final s before the initial vowel of an enclitic, where according to him it regularly occurs. The other examples he gives of this kind of rhotacism however do not substantiate this view. ereJc, pure, erarunt etc. could be regarded and treat- ed as simple words, like Mcce, idem, in Latin, and if this were the case, s would follow the general rule in becoming r — 44 — between two vowels. But it is more probable that we have liere e^aiBples of the rhotacism of final s, whicb bad already begun when the tables in Etruscfin letters were written, cf. plenasier, Jclmiier, gersnatur. If the two form§ in r and s existed side by side, it yrould be more in agfeeijaent with the tendencies of thp language to append the enclitic to the form in r. The forms of the pronoun in r are regularly used in the later tables, and nothing prevents us from assuming that in ereh and erarunt the enclitics were appended to the already rhotacised forms. In the same way fesnere can be the rho- tacised form of the dative plural with thp appended ,e. In this W;^y can be explained the .exception? tp Br]^§,)'s rule, pisi and pisher. Though the pronoun pis appears a? pir ip the later tables, once, yi, 53, it retaips the ?, ^nd that jtfto befofe a word beginning with a yowel- In these cs^^es thgp, the enclitic was appended tp the older form and rjj^tacisni d^d not t^te place. So seso, if rightly explajned by Br^al as cop- t3,ining the enclitic hont, supports this view; therie is po tr^,cii fif rhotacism in the s of the reflexiye pr,Qnoun, an4 rhot§,&i?fli i^ not occasioned by the aj^pjended pnpytic. The forips of the verb whepe s is final are too few to enable us to come to a satisfaptory conclusion, if thosp forms are disregarded where s becomes §nal by the dr;lrimo\ but What appears in Latin only in word's wMch the popular cGBBeiousness grasped as necessarily connected with each oflier, where the ¥ therefore is permanent and a part of the word) occurs in Sanskrit only tvhere the two words are distinct- ly felt as such. When s is brought before a vowel or voiced consonant by word formatJon, that is, where the adtftion of a suffix or an ending gives precisely the same position, the s may nndeiigo changes according to well-defined laws, but it never be- coiates r. — 46 — Rhotacism in Greek occurs only in two or three of the dialects and under circumstances altogether different from those in the Italian languages. It is found chiefly in final s, to which the only Italian analogy is in Umbrian, and in the middle of a word only before consonants, of which in Latin a very few examples occur. In the remarkable -inscription recent- ly found at Olympia, final s becomes r regularly, in the nouns and adjectives and adverbs alike. There is nothing similar to this in any of the Italian languages, and it is probably connec- ted with some local ipeculiarity of pronunciation. Cf. Curt. Gr. Etym. p. 444. In Gothic we find the first step towards rhotacism in the change of s to s. This does not take place with the same reg- ularity as in Latin, but is found chiefly under the same con- ditions. So when s is brought by inflection between two vowels mais but mai^a , farais but faraza. So between vowel and voiced consonant, a kind of assimilation, ans but anza^ slepan but saiglep etc. In Old High German the process has gone further, and r takes the place of original s, and that regularly in nom. sing, masc. of the adjective, in gen. and dat. sing. fem. and in gen. plur. of ihe adjective, and in the comparative, and in some iso- lated words as Jcar, mer, leran, mir, tMr etc. Grimm, Deut. Gramm. I.^ 121. In the plural of the imperfect r took the place of s, wdrun, birun, etc., but singular was. Final as well as medial s was thus changed, and it is impossible to discover a fixed rule, s brought by inflection between two vowels often became r, but often remained unchanged, los, lurumes but las lasun. The nominative sign s, which becomes r in Old-High- German in the pronouns, in Old-Norse is regularly changed. So sonr , vulfr for Goth, sunus, vulfs. But even in Old-Norse where rhotacism is more frequent than in any other German Ian- — 47 — guage. (Grimm, p. 305), s between two vowels is not seld5m found, ausa cf. Jiaurire etc (Grimm, p. 317). The German languages seem thus to show the greatest similarity to the Latin, but have apparently not been so con- sistent in the development of the common tendency. Moreover by the side of the rhotacism between two v6wels, the German languages showed a tendency to change final s, which is to be found Only in Umbrian among the Italian languages. Natus sum Edyaxdus Lotharius Walter Litchfieldi in op- pido Michiganiensi ante diemlVNon. Feb.ajMDCGCXLV patre Edvino matre Sara e gente Walker. Primis litteranim elementis in schola oppidi natalis imbutus, adii scholam Albionensem, nt . stadia ad academiam praeparantia ad finem perducerem. A. h. s. LXII voluntariam militiam tuli belloque Virginico interfui, donee Sacramento solutus sum propter morbum gravem, quo, susceptis laboribus molestis , correptus sum. Valetudine con- iirmata, ascriptus in civitatem. academicam Michiganiensem, a, h. s. iXVin nactus sum primum gradum -honoris aeademici, et tribus post annis alterum gradum. Exinde adjunctus sum extra ordinem praeceptoribus ejusdem academiae ibique per sex annos id munus praestiti. Hoc tempore officio professoris or- dinarii Latinae linguae, in Europa absentis, per duos annos functus sum. Veni a. MDCCCLXXIVin Europam, atque, Anglia, Scotia, -Ndf vegia , Suecia , Daniaque peragrata , Lipsiensem uni- yersitatem petii, et a rectore magnifico Schmidt receptus, ad studinm philologiae comparativae me contuli. Audivi per quattuor semestria viros docentes hosce : Curtius, Ritschl, Kuhn, Leskien, Lange, Mendelssohn, Osthoff. ^B^lra^tljWi ^^fe&'» ^n