On Some Ancient and Modern Etymologies Author(s): Minton Warren Source: Transactions and Proceedings of the American Philological

Association, Vol. 32 (1901), pp. 110-120 Published by: The Johns Hopkins University Press Stable URL: http://www.jstor.org/stable/282613 . Accessed: 23/09/2013 03:10
Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at . http://www.jstor.org/page/info/about/policies/terms.jsp

.
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship. For more information about JSTOR, please contact support@jstor.org.

The Johns Hopkins University Press is collaborating with JSTOR to digitize, preserve and extend access to Transactions and Proceedings of the American Philological Association.

http://www.jstor.org

This content downloaded from 2.228.10.5 on Mon, 23 Sep 2013 03:10:31 AM All use subject to JSTOR Terms and Conditions

110

Minton Warren.

[go,

VII. -On

and Modern Etymologies. Sote Anicient

BY
PROF.

MINTON WARREN,

HARVARD UNIVERSITY.

Periero and peiero have long puzzled scholars. Lindsay, Latin Language, p. I99, says: " Pejtro and efero (cf. conierat, coniurat,G.G.L. IV. 322, 33) have not yet been thoroughly explained." See also p. 587. More recentlyFerdinand Sommer, Indogermtanische Forschungen,Bd. XI. (1900), p. 56, istpjiirare (Osthoff, Perfekt, S. I I 5 says: " Sehr zweifelhaft Anm.) das, falls es zu peior gehort, auch die Vokalstufe .P5`-- enthaltenkonnte; das Wort ist und bleibt eine crux; von Stolz (H.G. I70) befriedigen auch die Ausfiihrunigen nicht." BeforeOsthoff, Alcuin had connectedpeiero withpeius, but in a different way, cf. Keil, G.L. VII., p. 307. "Periurus qui ma/e iuirat, peiero vero verbumr non debet habere; est enim quasi peius iutro." This was doubtless a current popular etymology,which did not lose sight of the connection of in I884, explainedpeiero as a the word with iuro. Osthoff, peilus in the originalsense of " entstellen, denominativefrom verhunzen." In I885 Gustav Meyer, Zeits. fiir Oest. Gym. 36, p. 280, gave a similarexplanation. Brugmann,Gdr. II., p. 402, admits the explanation as possible. L. Havet, AMmoiresde la Soci,t/ de Linguistique, VI., p. 22, without mentioningOsthoff,gives essentially his view. "Peierare signifiedonc a l'origine 'rendre pire, alterer,fausser,violer.' lus Peieratum est un droit corrompu (par faux serment). L'etymologie populaire crut voir dans ce derive de peius un qui ne parait pas concilicompose de per et de i'lro, doctrinie lois able avec les phonetiques." Wharton, in his Etyma swear falsely; ' make worse,alter, Latina (I890), has "peryero, beside ma4/or)." Despite this violate,' fr.pe-es-(cf. ma-jestas it seemisto me that there is verylittle to arrayof authority,

This content downloaded from 2.228.10.5 on Mon, 23 Sep 2013 03:10:31 AM All use subject to JSTOR Terms and Conditions

Vol.xxxii.] Some Ancient and Modern Etymnologies.

III

be said for this explanation. It disregards the fact that perierare is better attested than peierare in Plautus, and ignores the fact that such denominatives from comparatives are of late development in the history of the Latin to believe that deiero, language. Moreover, it is difficult which is found in Plautus, although evidentlya compound of iziro, owes its e solelyto the analogy ofpeiero. A common explanation must be found,I think,forperiero, deiero,eiero, and the formconierat(coierat) found only in Glossaries and explained as coniiirat. The difficulty of course is in accounting for the shorte in these compounds,and I do not thinkit is met by assuming,with Stolz, the existence of a primitive verb iuiro with shortu alongside of iuiro. Before proceedingto state my view,I shall speak briefly of the formsfoundin Plautus; There are some sixteen cases of the verb periuro(periero) in Plautus, in three of which the reading is disputed. Only seven of these passages are found in the Ambrosianus,namely,Cist. 500, Merc. 539, Poen. 480, I242, Pseud. 354, I057, Stich. I92. In all except the last of these passages A has the spelling withi. In Stich. I92, ni vereperierit, si cenassit domi, Bugge would read perieraverit, but Leo takes perierit in the sense of interie;it. In Poen. I242, A has periures with the other Mss. where the metre requiresperieres. In two otherhave periurem and perizuret where the metrerequiresthe formS with e'. Perierat is well attested for As. 293. There are nine passages where,so faras the metre is concerned,forms. withe mightbe substitutedforthe betterattestedforms with it. Moreover, in Stich. 229, A alone has perieratiunculas,. the other Mss. havingperizuratiunculas. It is quite possible,. thatperierowas the more commonformin Plautus,. therefore, and thatperiulro has been substitutedforit not simplyin the three passages where the metrerequiresperiero. The forms found are periuras Poen. 480, perieras Ussing, Leo, perierat As. 293. Peierat Men. 8I4 without Ms. authoritySchoell. The Mss. read delurat or delirat, and Leo reads deierat.

in A, Bacch. I030 and I042, thebestMss. passag es notfound

This content downloaded from 2.228.10.5 on Mon, 23 Sep 2013 03:10:31 AM All use subject to JSTOR Terms and Conditions

I12

Hinton Warren.

[1901

Perierant True. 30 Schoell. PerieratdumnLeo and GS., periurabo As. 322, periuravisti Pseud. 354, periuravit Mere. ? or perieritStieh. 192. Perierem Baccb. 539. Per-ieraverit I030, periuremBCDFZ, per-icresPoen. I242, periurcs A with Periuret Mss., periuraris other Mss., perieretBaech. 1042. Cure. 268, periurareCist. 500 As. 562 and 570, periuiraverint

and Pseud. I057.

r,peiero,the evidenee without For the existeneeof the form is very slight for Plautus, although it is occasionally found in FZ and has been introduced by editors (cf. the critical apparatus to Asin. 293, Bacch. I042, Poen. 480 and I242, and the noun peinruis Pseud. 1057, Truc. 30). The acijective peiuriumrare somewhat better attested. See Studemund, Rihein.Afihs.21, 588. The explanation,which I have to propose, starts fromthe IOVESTOD of the Stele recently discovered.in the fornm Forum, which, with Hiulsen, Skutsch, Thurneysen, and iisto. scholars, I regard as the earlier formof inistod, otlher In I.. XI., p. 342, v. Grienbergerexplains iovesat at the as equal to iurat, transbeginningof the Duenos inscription lating " es schwortbei den Gottern,der mich sendet." I do not think that this makes good sense, and v. Grienberger himselfadmits his inabilityto establish the connectionwith what follows.' Yet I thinkevery one must admit that the earlierformof iurat would have been iovesat, which by syncopation gave iousat and then by rhotacismioulrat. The diphthongou is abundantlyattested in inscriptions. was someIn the compoundverb the phoneticdevelopment what different. In the early period the accent would of course rest upon the preposition. Now, just as de~novo tu' t4upover induo (cf. Umb., anovilzimu), gives de?uo, erndovo puer (cf. Arclziv, XII., p. 28I), So p/rioverowould give pdriovesowould p/riuero,or, in the stage before rhotacism, have given piriueso. One further change took place.
1 I have attempted, thisas to explain ioveset, Harvard Studies, XI. I64, reafling on the which also restsindirectly an explaniation iussit, of iouset, the older form assumptionithat iovestsd equals iustod.

This content downloaded from 2.228.10.5 on Mon, 23 Sep 2013 03:10:31 AM All use subject to JSTOR Terms and Conditions

Vol. xxxii.]

Some Ancieiztand Modern Etymologies.

I 13

Minerva is now explained (cf. Brugmann,Gdr. I.,2 pp. 232, 3I9, 324) as coming from * menes-oua through menesua, MIineruta, Minerz'a, larva as fromlasua, * lasoua. So I think in periziero, there was a tendencyto resultingfromperiovero, pronounce the vowel u as v (English w), but the difficulty of pronouncingv after consonantali caused the it to drop out. Thus we get instead ofperizzero1 and similarly in eiero periero, and deiero. By this theorythe shorte is fullyexplained,and there is ino necessityto dissociatepeiero fromiuro on the one hand or fromeieroand deicroon the other. The dropping of v (u) aftera consonantis seen inaperio,operio for apverio, opzverio, cf. Brugmann,I.F: I., p. I75, in battere forbattuere, battalia forbattualia,cf.Archiv,I., p. 249 and X., p. 42 I, and Schuchardt,Vocalismus, II., p. 470. The Appendix Probi,Arcyiv,XI., p. 329 hasfebruariusnonfebrarius.See the note of Heraeus. Febrari?us is abundantly attested in inscriptions and has its descendant in the Italian Febbraio. It is foundalso in a papyrusof the first century, firstpublishedin igoo.1 Schuchardtgives (II., p. 467 f.) many examples of u dropped in vulgar Latin. So lanarizus for Ianuarius, Confor Confiuentes, which gives us the modern Coblenz. _iendis An interesting English parallel is seen in swear = iuro, whereas in the compound answer the w, althoughwritten, is not pronounced. As to the form peiero,which seems not to be clearlyestablished forPlautus, but whichis frequentenough in later Latin (see Georges, Lexikon der Lat. Wor/formnen, s.v.), I should explain the droppingof the r as due to the same causes which have led to its disappearance in praestig-iae, crebescere, and otherwords. A formlike perierare or perieraris with three successive r's was difficult to pronounce, but neither the second nor the thirdr could be droppedwithout makingthe formunintelligibleand consequently the firstwas dropped. Cf. Grammont,La Dissimulation consonantique, p. 28. The formsdeierareand. eierare may also have contributedto the disappearance of the r, which was furtherhelped on by a popular etymology connectingthe word withpeius.
1 Archives Militaires du 1er Siecle,JulesNicole et CharlesMorel,Geneve,I900.

This content downloaded from 2.228.10.5 on Mon, 23 Sep 2013 03:10:31 AM All use subject to JSTOR Terms and Conditions

114

MintonWarren.

[I901

As to the form periurarelittle need be said. It is simply to bringout more clearlythe connection due to recomposition, with iuro, at a period when the formperiero had already become a little mysterious. So in Curc. 268 periuraverint followsin iure iurando of the line before,and in Pseud. 354 periuravistitakes up iuravistinof 352. and adizuro, In the compoundsabiiuro onlythe forms with u are attested for Plautus. See Lodge, Lexicon Plautinum,l although conierat is Fasc. I. So too in the case of conizuro attested by glossaries. Editors read deieravit,Cas. 670 and deiera, Rud. I336, although in the latter passage the Mss. have deilura. Eiero is not attested forPlautus,but occurs in of the Amph. later writers. Exiuiro is found in a fragment in writerslater of eiero For examples periero,peiero,deiero, see Georges,Lex. der Lat. than Plautus and in Inscriptions, Wortforrnen.
SOROR AND FRATER.

From Gellius, XIII. io we learn that the jurist Antistius and that Nigidius Figulus Labeo derivedsororfromseorsum, derivedfrater fromfere alter, "frater est dictus qzuasifer-e alter." Labeo and Nigidius werepurists. They bothbelieved the Latin. Their etymologies in explainingLatin words from are indefensible,but fromanother point of view they may have a certain interest and value. In deriving soror from had in view the shorterform Labeo evidently seorsuim sorsum, which is foundin Plautus and Lucretius. The derivationof frater from fere alter implies a possible offrater as feraterwith the insertion of a parapronunciation sitic e. Now it is quite possible that some persons should insertan e betweenf and r in pronunciation, whentheywould not do so in writing,just as Edon, Ecritureet Prononciation du Latin savant et dii Latin populaire, p. 2 3, tells us that a Persian will pronouncefranfais feranKais. So we find in inscriptionsTerebonioand Terebunifor the usual Trebonius, cf. C.I.L. I. I9o, and Ephem. Epig. I., p. 29, n. ii6. Other
1 It is quitepossiblethattheseare latercompounds wheniuro was the formed in use of the simpleverb. onlyform

This content downloaded from 2.228.10.5 on Mon, 23 Sep 2013 03:10:31 AM All use subject to JSTOR Terms and Conditions

Vol.xxxii.] Some Azncient and Modern Etymologies.

IIS

similar examples of the same tendency given by Edon are Geracilis for Alexandiri, Petiro, patiri, matiribus, mnateri, to which Gracilis, pateres, magisteres,arbiterlo,teribunatu, others might be added. The same possibilityseems to be indicated by Varro's derivation of Gracchus fromgero (cf. recentderiCharisius Keil, G.L. I., p. 87), although Wolfflin's vation fromgracus, the primitiveof graculus, is more plausible. Servius and Donatus do nothesitate to derivears from apErT4, suggesting that the nom. pl. may have sounded like aretes. Paulus, p. io, gives as one derivationofatrium " qzod a terra oriatur quasi aterrium." Lindsay, Lat. Lang., p. 93, following Meyer-Lubke, says that the development in Romance of a word like patrem suggests an almost trisyllabic pronunciation like paterem,and the Venerable Bede tried to do away with spondaic hexameters ending in words like respergebat, argentiby assuming a pronunciation rsperigebat, arz^Genti, which remindsus at once of the anaptycticvowel in Oscan in words like aragetud. A tendency of this sort in Latin can hardlybe denied. For the droppingof I before t there is little that we can compare in Latin, but in Oscan in this veryword / is dropped foraltrud,although in the Tabula Baitina, which has atrzud elsewhere in Oscan the I is preserved in the few instances where the word occurs. Similarly the Umbrian drops I in to Lat. multa,fine, cf.von Planta, I., whichcorresponds motar, p. 299. In French of course alter becomes autre,but in some of the Romance dialects the I seems to have been totallylost. Thus Sardinian has atter, atteru, Calabrian atru and atu, Genoese aotroand atro,cf. Moli, Introduction a la Chronologie du Latin Vulgaire, p. 278. It seems to me, then, that in giving the derivationof frater fromfrrealter,Nigidius may well have had in mind a dialectic pronunciation of alter in which the I was not distinctly heard.1
1 We are not informed of Nigidius Figulus,but it is peras to the birthplace haps interesting to note in thisconnection thatthe gentilenameNigidius is found accordingto Conway especially in Campanianand Sabine areas, i.e. exactlyin those regionswhere I beforet was not distinctly heard. Cf. Conway,Italic Dialects,II., p. 576.

This content downloaded from 2.228.10.5 on Mon, 23 Sep 2013 03:10:31 AM All use subject to JSTOR Terms and Conditions

ii6

Minton Warren.

[1901

fere al/ermay fraterfrom The fact that Nigidius did derive lend supportto the view that in the expressiondies ater,ater stands foran originalalter. Mohl, p. 277, says: " Que *a/tru alternmait circuledans l'Italie du Sud ou *atru pour *al/rum des une epoque tres ancienne, c'est ce que montrele latin dies a/er,Varron,Ling. Lat., VI. 4, 29: Dies postridieKalenappella/iatri,quodper eos dies novi inciperent. das Nonas Iduts L'expression dies ater signifiedonc 'un jour d'une nouvelle serie, d'une autre division.' Le mot nous reportesans doute a l'epoque oiu les Campaniens etaient encore les educateurs Numa y introduisait de Rome, peut-etrememe au temps oCu le calendriersabin." Before Mohl, Deecke had explainedater it to Etruscan influence. in the same way, but had attributed In Die Falisker(i888), p. go, he says: "Es ist namlichater die etruskischeForm fur lat. alter, mit Wegfall des / vor t, neben haltu,faltu lat. Fallo." Mohl's wie im Beinamen hat/u view seems to me more probable. Wissowa in the article on dies ater in the new Pauly admits a similar meaningfor ater. I quote his words: " Der Name dies ater hat mit ater, 'schwarz' nichts zu thun, sondern hangt, wie 0. Gruppe (Herm. XV. 624) richtig gesehen hat, zusammen mit den d.h. post diem Sep/ima/ruts Bildungen Tria/rus,Q 7inyuatrus, Fest. VI. de 1. 1. 14. (Varro, septimnum quin/tum, tertiumn, deuten etc. irren dass sie Quinquatrus p. 254, die nur darin der Ausgangspost diemquintumetc. Idits, wahrendnaturlich punktder Zahlung jeder beliebige sein kann), bedeutet also nichts anderes als eben dies postriduanus." Gruppe in the to by Wissowa makes no attemptto connect article referred ater withalter. He says: " Ueber den Ursprungdes Wortes vermag ich eine Vermuthung nicht aufzustellen; was den dass es so scheint mir am wahrscheinlichsten, Sinn betrifft, etwa ' nach,' ' nachher' bedeutete." Deecke, however, op. wohl nach der zweiten urspriinglich cit.,p. 9I, says: "trita/ruts, eigentlich der 'drei-andere' Deklination, fur *tri-alter(us), d.i. 'drittnachste' Tag u.s.w." We have the distinct testimony of Festus to. the fact that Triatrus, Sexatrns, and werein use amongthe Tusculans and Decimatrns Septematrns were dialectic among the Faliscans. These words,therefore,

This content downloaded from 2.228.10.5 on Mon, 23 Sep 2013 03:10:31 AM All use subject to JSTOR Terms and Conditions

Ancient and Modern Etymologies. Vol.xxxii.] Somne

II7

and there is nothingto preventour supposing that Qiiinqzathat therefore, trus was originallya dialectic form. I submit, in view of Nigidius' etymologyof frater and of the facts above stated in regard to alter, this etymologyof Deecke's deserves more serious attention than it has received from p. 549, withoutany scholars. Stolz, HistorischeGrammatik, of attemptto explain the intrusion the r, connects the suffix -atus. His words are "Weiter reiht -atrns with the suffix sich hier auch an tr4m-dtu-s (Analogiebildung nachpr7m-d/u-s Varro u.a.) 'Alter von drei Jahren' Col. Plin., das zugleich auch den Schluissel enthalt fur Qu7nqu-dtrfi-s (*qu7nqu-dtu-s Tage') und die von Festus 340 Th. d. P. 'Feier am ftinften (falisaufgefuihrten entsprechendenBildungen Decim-atrFu-s (tusculanisch)." Sex-dtrui-s, Tri-dtrii-s kisch) und Septem-dtru-s, Afterthis explanation,which does not explain,he coolly dismisses Deecke's view with the words " Eine andere sicher unhalthare Vermuthung iiber diese Substantive findetman bei Deecke Die Falisker S. go f." Deecke is doubtlesswrong in what he says about Etruscan influence, but not I thinkin connecting the words with alter,although therestill remains in regard to the declension of these forms. some difficulty The most plausible supposition would be that Quinquatrus was influenced in its declension by Idus, Idus itself being cf. Brugmann,Gdr. II., p. 297. an old adjective with a u-stem, and Delbruck, Vergleichende aktFun Comparealso Sanskrittrgr Syntax, I., p. I63. offrater,it may be In connectionwith Nigidius' etymology to note that Cicero, in Adfam. II. I5, speaks of his interesting brother Quintus as his alter ego. "quem tamensi reliquissem, dicerenzt iniqui non me plane post annum, ut senatitsvoluisset, quoniamalterummereliquissem."Voss, deprovinciadecessisse, in his Etymologicon, this, goes so faras to say that in writing Cicero had in mind the etymologyof Nigidius, but this is unnecessary. That alter and ater are oftenconfusedin Mss. I need not note. A famousexample is Catullus, XCIII. 2, necscireutrum sis albusan aterhomo, where all the Mss. have alter forater.

This content downloaded from 2.228.10.5 on Mon, 23 Sep 2013 03:10:31 AM All use subject to JSTOR Terms and Conditions

ii8

Warren. lI'inton
SALTEM.

[I90I

Gellius, XII. I4, gives two etymologies for saltem, with both of which he declares himselfunsatisfied," censuimus ig,itur amplizisquaerenduvm." The second derivationmakes sal/em stand for salztem, with the extrusion of u. This seemed to Gellius clever but far-fetched. That it was current in antiquityis shown by the fact that Donatus and Servius both give it. Cf. Donatus, ad Adel. II. 2, 4I and Servius,ad Aen. IV. 327. This etymologyneed not be seriously considered. The other I give in the words of Gellius. "A tque erat, quil diceret,legisse se in grammaticis cominentariis P. Nigidii, sal/emex eo dictum, quod esset'si ali/er' idque ipsutm dici solit/um 'si per defectionem, nam plenam esse sententiam ali/er non po/est.' S4d id nos in isdem commen/arlis P. eos non, opinor,incurioselegissemus,nuisquam Nigidii, cutm invenimuis." He goes on to state a very valid objection. verba is/a ' si a/iternon potest' a szgnifca"V ident/ur auztem tionze non ab/orrere. quidem vocutlae hnius,de qua quaerimuis, Set tot verba ta;nen in pazucissimasli//ei-as cludere,inprobae cuiusdam szubtilitatis est." Wharton, in his EtylmaLatina, from withoutdifficulty.' derivessal/em saltus, leaping,' swiftly, himself more cauLindsay, Lat. Lang., p. 556, expresses tiously. "If saltem is Acc. Sg. of *salti-,a leap, lit. ' witha leap,' 'swiftly, easily, assuredly,'it has taken -temby analogy shows that sal/em was the The derivationfromsalu/tem while saltim is late. In fact,with so many recognizedform, adverbs in -tim as par/im,statim, etc., it is hard to comprehend why an original sa//im should have been changed to saltem. Therefor it seems to me that this connection with sa/io must be given up. The derivationfromsi aliter, as thus stated, cannot be correct,but why may not saltem stand for an earlier si al/em, al/em being a syncopated formfor a/item,an adverb fromthe stem ali- formedafter the analogy of item? The early Latin, as we know from Paulus, had an adverb,alin/ta,which does not appear in literature. Why mayit not have had a/item in the sense of 'otherof au/tem, i/em, etc."

This content downloaded from 2.228.10.5 on Mon, 23 Sep 2013 03:10:31 AM All use subject to JSTOR Terms and Conditions

Vol. xxxii.] Some Ancient and Modern Etymologies.

II9

wise,' later supplanted by aliter? For the elision of si and the subsequent contraction, we may compare sis, sultis, and took place, the derivationmight sodes. Afterthe contraction as it was in the case of sirermpse, easily have been forgotten, a word whose derivationis still in doubt. Whether altem or alitemwas the originial formof the adverb,I leave undecided. The analogy of aliter would favor alitem, and, according to F. Sommer,IF. XI. 4, the pronoun alter goes back to an original * aliteros. Altem would then be the syncopated of alitem. Syncope was doubtless very comAllegro formn mon in the early period of t-helanguage, and it is a question whetherwe can safelylay down such definitelaws governing its action as has been attempted by von Planta and Sommer. To trace the developmentof meaningby whichsaltemarrived at the forceof ' at least ' is not easy, as it alreadyhas thatforce in the earliest literature, and the ordinary speaker in the time of Plautus had entirely lost sightof its origin, but we maynote that as we oftenfindsin aliter opposed to si, so we oftenfind saltem following a clause with si. Terence, Hec. 635 ff.,has: esse inter Ego, Pamphile, nos,si fieri potest, Adfinitatem hancsaneperpetuam volo; Sin estut aliter tuasietsententia, Accipias puerumn. The latter propositionmight have been abbreviated into sin aliter, accipias puerurn, and we can see how sin aliter as an alternative mightcome to have the meaning of ' at least.' In Trin. 485, tu hoc facito, Semper Lesbonice, cogites, Id optumum esse tuteutisis optumus: Si id nequeas, saltem sis proxumus. ut optumis At an earlier period of the language we can conceive of saltem, 'if otherwise,'taking the place of the clause si id nequeas, but in the time of Plautus it had lost thisforce and had come to mean ' at least.' Compare,Ter. Eun. 639, si i lud non licet, sal/cm hoc licebit,also Amph. 438, Quis ego sum saltem,si nionsum Sosia Cas. 298, Nam si sic nil impetrare saltem sortiiar. potero,

This content downloaded from 2.228.10.5 on Mon, 23 Sep 2013 03:10:31 AM All use subject to JSTOR Terms and Conditions

120

Minton Warren.

[1901

So in a sentence like Cic. ad Att. IX. 6, 5, eripe mihi hUinc doloremazit minuesaltem,we can conceive that at an earlier i.e. ' if you can't period saltem lhadthe forceof ' if otherwise,' do that.' By the association of ideas aftera time this might appreciaeasily assume the meaningof ' at least.' Some dimn tion of this earlier force seems to have been presentto the author of the etymologywhen he interpreted saltem as 'si aliter nonpotest.'
NOTE

ON frequenter.

on Ter. And. I07, una aderat Fairclough,in his commentary freqiens,says: "frequensused insteadof theadverbfrequenter, which belongs to later Latin." This is a natural mistake,as the dictionariesgive no instance earlier than the Auctor ad Herenniurn, who uses it four times, IV. 32, 46, 48, and 56. of Terence, usesfrequenter (de Agr. But Cato, a contemporary once, 67, i, olcumfrequeznter capiant). Cato also uses rarenter de Agr. I03, and we know from Nonius that it was used by Livius Andronicus, Ennius, Caecilius, Pomponius, and Novius. Later it crops up again in Gellius, Apuleius, and must have been formedon Macrobius. Now I thinkrarenter the analogy offrequenter. It was not necessary,as the early were paired, language had raro and rare,but as raro and crebro so rarenterand frequenter. Of course frequenterwould not be cited by the grammarians, not being for them an unusual word. Indeed it is one of the words which later supplants saepe. Compare my article on Latin Glossaries in Vol. XV. of the Transactions, p. I39. Plautus, of course,does not use occurs infrefrequenter. His usual word is saepe, and crebro but does use raro quently. So Plautus does not use rarenter, and rare. Frequentermust have been known in the time of but neitherdoes Lucilius, but does not occur in his fragments, or raro,bothof which are earlierthan Lucilius. In fact crebro saepe is only found fivetimes in Lucilius and saepius twice. I believe that,if we had a larg,er bodyof early Latin prose,we should find that frequenterwas in common use, and that rarenter, as I have already said, was formedon the analogy of frequenter.

This content downloaded from 2.228.10.5 on Mon, 23 Sep 2013 03:10:31 AM All use subject to JSTOR Terms and Conditions