MA S TER NEGA TIVE NO. 92-80599-25 MICROFILMED 1992 COLUMBIA UNIVERSITY LIBRARIES/NEW YORK as part of the "Foundations of Western Civilization Preservation Project" Funded by the NATIONAL ENDOWMENT FOR THE HUMANITIES Reproductions may not be made without permission from Columbia University Library COPYRIGHT STATEMENT The copyright law of the United States - Title 17, United States Code -- concerns the making of photocopies or other reproductions of copyrighted material... Columbia University Library reserves the right to refuse to accept a copy order if, in its judgement, fulfillment of the order would involve violation of the copyright law. A UTHOR : FRANK, TENNEY TITLE: ATTRACTION OF MOOD IN EARLY LATIN PLACE: LANCASTER, PA. DA TE : 1904 t» COLUMBIA UNIVERSITY LIBRARIES PRESERVATION DEPARTMENT Master Negative U BiDuocR APHic MrrRnmRM tarhft mmmmmmm^^utitm Original Material as Filmed - Existing Bibliographic Record BK3/PR0D Books FUL/BIB IMYCG92--B324a9 Record i of — JRecord BddBdi today Acquisitions NYCG-PT DCF:? INT:? RTYP:a. CRC: ? Rpr » 7 REP: ? DM: MOD; ' BIO:? CPIz? RR FRMs SN.R; F I C 5 ? FBI:? COL: ns s ATC;; COW:??? J L C EHL s « T» *■> '? '■> 040 inn t 245 10 II:? GEM : Restrictions on Use: ID:MYCG?2-D32409 CC:9iS68 BLTsaw CP:aau L:eng PD: 1904/ OR: POL: NWOIcNNC Frank |, Tenney , « I dt a.76_t f 39 • Attraction of mood in early iBtin- i hT.mi cy of orm'.] , t .ancaster, Pa» , s I bPre^s of the Mew era printing company,-" lcl9QA 1 p-5 59 p.=jc24 cm. aC fttt f9|.<*l'. AD: 05 -08-92 IJD: 05-08-92 TECHNICAL MICROFORM DATA -»-> FILM SIZE:__i_^ IMAGE PLACEMENT: lA REDUCTION RATIO: // -< IIA; IB IID >^7 /> C. DATE FILMED:_i.r_tfX_ I HLMEDBY: RESEARCH PUBLICATIONS. INC WOODDRIDGE. cf NITIALS BoE c Association for Information and Image IManagement 1100 Wayne Avenue, Suite 1100 Silver Spring. Maryland 20910 301/587-8202 Centimeter 12 3 4 5 Mll|llllLl|imlll|illi|ilMMlmilim llli nil I T \ 8 I I W llllMI TTJ m¥fV Inches 9 „1,1 Mill 10 11 12 13 14 "'l""l""l '"l'"lllMll ITT T 1.0 |50 m II" lUku. 1.4 2.5 2.2 20 1.8 1.6 I.I 1.25 15 mm MRNUFRCTURED TO fillM STfiNDfiRDS BY RPPLIED IMRGE, INC. ^^ ^ «b^ ATTRACTION OF MOOD IN EARLY LATIN TENNEY FKANK Zbc mnlversit? of Cbicago 1904 Ube Tamvcrsitg of Cbicago FOCHDKB BT JOHM D. EOCKlTEIiEB ATTRACTIOF OF MOOD IN EARLY LATIN A DISSERTATION Submitted to the Faculty of the Ghaduate School op Akt^ AND Literature in Candidacy for the Degree of Doctor of Philosophy (DEPARTMENT OF LATIN) BY TENNEY FRANK niEM 6r "nm Niw MA raiNTiNa aomun 1904 ATTRACTION OP MOOD IN EARLY LATIN. i INTEODUCTIOK Notwithstanding the great activity of scholars in the field of Latin syntax during the last half century, the subject of attrac- tion of mood in the Latin language has passed almost unnoticed. The reason is not that the subject is already clearly understood. On the contrary, most of the statements in regard to it in the handbooks are, at least in part, erroneous, and confessedly are based upon very meager data. Neither is it a matter of smaU importance; for at few points does fine feeling for the shades of meaning of moods more affect the interpretation of a Latin sentence than in the usages of this construction. Nevertheless It has been slighted. I find but one treatment of the subject that pretends in any way to be thoroughgoing. Even this touches a very limited field, namely a part of Cicero; does not get com- plete data from this ; makes little attempt at digesting the mate- rial gathered ; and uses a method of treatment which is somewhat antiquated. I refer to the program of Franz Peters,» published in 1861. The substance of his treatise may be found in Draeger,* whose lists are as full and valuable as any yet pub- lished on the subject. Draeger's examples are taken mainly from Cicero. The same is true of Kiihner's examples.» Ziemer* should also be mentioned for some suggestive remarks on the subject Several dissertations upon special constructions have of neces- sity touched upon the subject in passing, but have hardly affected the discussion of the problem, since they have approached it only ? ' De attractione quadam temporum ac modorum linimae latinae Deutsch-Crone, 18«1. « HiBtorische Syntax der Lat. Sprache, I, sec. 151, 2-5. »Lat. Grammatik, II, 2. •Junggrammatische StreifzUge; Colberg, 1883. 1 Progr., ATTEACTION OF MOOD IN EARLY LATIN. from tie peiiit of view of their special topic. Thus Liibbert^ places certain cum-clauses ill Plautns Tinder the category of at- traction. Elste,^ Kichardson' and Boettger* explain certain examples of the subjunctive with dum as due to the same influ- ence. Schubert,* Lange® and Rodenbuscl^ should also be men- tioned as having touched the subject incidentally, the latter two at least with some insight. The lists of Holtze,® which are sup- posed to cover the field of Early Latin, are not half complete, con- stantly betray a lack of understanding of Early Latin usage, and make no analysis of the material. Nevertheless they have been of value as presenting the largest collection of such niaterial from that important period. To all of the above I have given due credit where they have been found serviceable. My attention was first called to this subject by Professor Hale, who in several of his works has thrown new light upon the obscure problem. Keferences to these will be found in the body of this treatise, in which it will become apparent how great is my indebt- edness to him, not only for specific suggestions, but also for my point of view, method, and even grammatical nomenclature and classification. In saying this, however, I do not wish to be under- stood as implying that Professor Hale agrees with me in the Inter- pretation of every example, nor in all of my reasoning. As regards my way of approaching the subject, it may be said that, after working for a considerable time over the whole range of Latin literature, I discovered that I was dealing with a chang- ing construction, and that, therefore, a historical treatment was the only logical one. Furthermore, it became apparent that the origins of tiie use of the subjunctive by attraction are not to be looked for in the language of artistic prose, with its various con- scious artifices; for the construction appears before the time of such writing, and in simpler form. On the other hand, since the iDie Syntax von Quom, 1870, cf. p. 125 if. * De Dum Particulae Usu Plautino, 1882. »De Dum Particulae apud Prise. Scr. Lat., 1886. ♦ De Dum Particulae Uau apud Terentium et in Rel. Trag. et Com., 1887. >Zum Gebrauch der Temporalconjunktionen bei Plautus, 1881. •De Sententiarum Temporalium apud Prise. Ser. Lat. Syntaxi, 1878. 'De Temporum Usu Plautino, 1888. •Syntaxis Prise. Scr. Lat., 1861; and Syntaxis Frag. Scaen. Poet Rom., 1882. INTBODUCTION. 3 construction, in the very nature of the case, belongs to hypotaxis and a fairly well developed complex sentence, we need not become involved in the mists of parataxis. Early Latin, as we have it, may therefore confidently be called on to explain the beginnings. I have accordingly tried to make a thorough survey of Early Latin down to Lucretius. From the point of view thus gained I propose in a second paper to sketch the later usage of the con- struction. The so-called construction of " attraction by the infini- tive " has some things in common with the one here treated, but since the data offered by early Latin are too meager to ensure safe generalization, I propose to postpone its discussion to a later paper in which I intend to use statistics gathered from classical authors as well. The field covered by this paper includes Plautus,^ Terence, Cato, the early dramatic fragments, which are found in Eibbeck's^ collection, and the remaining fragments of the early Roman poets, which are found in the sixth volume of Baehrens'* collection. I have, with few exceptions, adopted the text as given in the editions named. The lists of examples being large, the errors are likely to balance one another. On the other hand, the discussion of all examples for which the manuscript readings vary would swell the work to impossible limits. My treatment falls under two heads: (1) the sources of the con- struction; (2) the uses* of the construction. > I have used the following editions: for Plautus, that of Gotz and SchSll, Leipzig, 1893-1898; for Terence, that of Dziatzko, Leipzig, 1884; for Cato'g Agri Cultura, that of Keil, Leipzig, 1894; for the fragments of Cato, that of Jordan, Leipzig, 1860. « Scaenicae Romanorum Poesis Fragmenta, Leipzig, 1896-7. » Poetae Latini Minores, VI, Leipzig, 1885. *I regret to say that I have not been able to avail myself of a study upon the same subject which was announced after my paper had gone to print. I refer to the paper of F. Antoine ( L' Attraction modale en Latin, Melanges Boissier, Paris, Fontemoing). Judging by the accurate and sane work of that scholar, I feel that the loss to my study must be great indeed; and I can only make it good by referring the reader to his work which may sup- plement and correct the views expressed by mine. CHAPTER I. The Soubces of the Consteuction. A number of years ago Professor Hale pointed out the way in which the origins of the construction now under examination must be studied. He says in his * Sequence of Tenses ' (American Journal of Philology^ VIII [1888], 1, p. 54; and American Jour- nal of Philology, IX [1889], 2, pp. 175-6): "In complex sen- tences made up of a main sentence with subjunctive verb and one or more subordinate sentences, the modal feeling in the speaker's mind which expresses itself in the main sentence is, in the nature of things, very likely to continue in the speaker's mind in the subordinated sentence or sentences, either quite unchanged or but slightly shaded. If, for example, I say in Latin, ' Let him send whom he will,' mittat quern velit, the mood in vellt is not a case of ' attraction ' or ' assimilation * at all. Velit is as much a jussive as mittat is. The meaning is, * Let him choose his man, and send that man ' ; or, in older English, * Choose he his man and send him.' In sei ques esent quel sibei deicerent necesus ese Bacanal habere (C. L L., I, 196), the deicerent is as much a future condition (=sei ques deicerent) as esent is." Again, " the frequent recurrence of such examples gives rise to the occa- sional use of a dependent subjunctive with only a formal likeness to the main subjunctive, and no true modal feeling." Later he reiterates the statement in a paper on the Anticipatory Sub- junctive* (p. 63, footnote). This wording implies, of course, that the construction in ques- tion is a thing brought about and developed. As for Professor Hale's interpretation of the mood of velit and deicerent, there can be no dispute, and it is my belief that, in the other main divi- sions of the subjunctive also, a great number of the cases usually treated as examples of attraction are to be interpreted in the same » University of Chicago Press, 1894. Reprinted in Studies in Classical Philology of the University of Chicago, Vol. I, 1895. THE SOUECES OF THE CONSTEUCTION. 6 way ; and that these must be distinguished from those which are in reality due to the habit of attraction pure and simple. The distinction might well be expressed in the terms of Professor Hale, who speaks of the mood of " like feeling " and that of " formal likeness."^ We are to seek, then, for combinations in which the modal feel- ing expressed by the main verb may naturally cast its shadow over the subjunctive clause. Obviously, there would be a lim- ited number of conditions under which this is possible. The conditions are favorable when both the verbs (main and subordinate) are in the same grammatical tense, and are placed in the same actual time-sphere. Such a state of affairs may be illus- trated by the examples above, or by Aul. 491, quo lubeant, nubant, '• let them marry where they please," i. e., " let them choose their place, and marry there." These clauses occur freely in dependence upon subjunctives of '^ volition," " wish," " permission " and the like. I add further illustrations : With a subjunctive of permission: Bacch. 656, furetur quod queat, " let him steal what he can " ("what he shall be able to steal "). With a subjunctive of wish : Hec. 197, di uortant bene quod agas, " May the gods further what you do " (" shall do "). With a substantive volitive clause : Bacch. 674, occasio . . . fuit . . . ut quantum uelles tantum sumeres, "there was opportunity to take as much as you wished " (" as much as you should wish "). The conditions are less favorable when the subjunctive verb, though in the same grammatical tense, is not in the same time- sphere as that of the main clause. An example may be seen in Cas. 252, iam domuisti animum . . . ut, quod uir uelit fieri, id facias ? '^ have you brought your mind to do (i. e., will you do ?) that which your husband desires ? " Between the more favorable conditions and the less favorable lie the conditions in which it is impossible to tell from the con- text whether the subordinate verb is in the same time-sphere with » A. J. p., Vol. IX, p. 176. e ATTRACTION OF MOOD IN EARLY LATIN. the main verb, or, though grammaticallj in the same tense, is not in the same time-sphere, — or, to state the matter in another way, the conditions in which either relation is reasonable, and there is nothing to determine which is meant. Such a state of things may be seen in Epid. 283, iam igitur amota ei fuerit omnis consultatio nuptiarum, ne grauetur quod uelis, " dismiss at once then the whole question of marriage, lest he shall begrudge you what you wish." Either " wish " or " shall wish " is here rea- sonable, and nothing in the context bars either meaning. Such combinations formed a bridge from the use of the subjunctive of " like feeling " to the use of the subjunctive of " formal like- ness " in the dependent clause, and thus carried the mood over from its original domain in the former field to a new domain in the latter. Distinctly unfavorable are the conditions when there is a shift in tense as well as in time-sphere, as when a verb in the past depends grammatically upon a future verb of willing or wishing, as in Cas. 503, ut quod mandaui curet, " that he may execute the commissions which I have given"; or, to take a subjunctive, in Kud. 1243, ut cum maiore dote abeat quam aduenerit, " that she may depart with a larger dowry than she brought with her." Here aduenerit, a verb of past time, could not share in the future volitive idea expressed by the mood and tense of aheat. In sen- tences of this type, then, is found the extreme of the fully devel- oped habit of attraction. With this preliminary explanation, we pass to an enumeration and discussion of the various kinds of subjunctives after which the verb of the dependent clause is at times found to contain the same modal feeling as that of the main clause (" subjunctive of like feeling"). I find these kinds to be as follows:^ 1. Volitive Subjunctive Jussive and Permissive. " Deliberative " and its extensions (Subjunctive of Surprise or In- dignation). Dependent Volitive. 1 The order in this list is that of the treatment which follows, and is de- termined by practical considerations of exposition. THE SOURCES OF THE CONSTRUCTION. 7 2. Optative Subjunctive — independent and dependent. 3. Subjunctive of Obligation or Propriety. 4. Anticipatory Subjunctive, for the present or the past (pres- ent-future or past-future). 5. Conditional Subjunctive. 6. Subjunctive of Ideal Certainty (as in Subjunctive Conclu- sions). 7. Subjunctive in Indirect Questions. A. Subjunctives depending upon subjunctives with simple volitive or optative force have been sufficiently treated above (pp. 5-6). o. The clauses which depend upon a subjunctive of "surprise or indignation," though of volitive ("deliberative") origin, may well be treated in a separate paragraph, since they show some important peculiarities. Through the very nature of this construction, it matters little whether the main and the depend- ent verbs are in the same time-sphere; for "surprise or indignation is expressed as readily at an act that took place in the past as at one of the present time. Ordinarily, in this group, it is easy to tell from the context whether the secondary verb has naturally the feeling of the main verb. The following sentence affords a good illustration of this class : Phorm. 970-3, ubi quae lubitum fuerit peregre feceris neque huius sis ueritus feminae primariae, . . . uenias nunc precibus lautum peccatum tuomi ^ The mood of uenias expresses the speaker's (feigned) indigna- tion. That same feeling extends,* it seems to me, through feceris, and sis ueritus, since these dwell upon the very acts which caused the anger. It probably does not extend into the explanatory clause quae lubitum fuerit, or, if it does, it extends with far less force. I should say that feceris and sis ueritus are in the sub- junctive for the same reason that uenias is, but that lubitum fuerit 1 In the following examples, in which the main verbs are in the indicative, I should prefer to interpret the dependent subjunctives as due to indignation and surprise rather than to an adversative feeling. Pers. 76, sumne ego stultua qui rem euro publicam uhi aint magistratus quos curare oporteat! 8 ATTRACTION OF MOOD IN EARLY LATIN. is a case of attraction. This statement may become clearer if we compare the following sentence, in which the explanatory clause, which is much like the one just discussed, is left in the indica- tive: Men. 560, Egone hie me patiar in matrimonio Ubi uir compilet clanculum quicquid domist? In the following the subordinate verbs are, if my interpretation is right, in the subjunctive, not by mechanical attraction, but Pseud. 184, Eo uost uostros panticesque adeo madefactatis quom ego aim hie siccus! Here Lange (De Sententiarum Temporalium apud Prise. Scr. Lat. Syntaxi, p. 45) needlessly proposes to emend madefactatis to madefaciatiSf in order to get a raison d'etre for tlie mood of aim. Capt. 892, dubium habebis etiam, sanete quom ego iurem tibi! In the first of these examples the force is not felt with such strength, at the moment when sum and euro are uttered, as to call for the subjunctive ; but the speaker's indignation rises as he proceeds, and finds full expression in sint and oporteat, I would add to this list Ad. 166, noui ego uostra haec " nollem factum, dabitur ius iurandum, indignum te esse iniuria hac," indignis quom egomet aim acceptua modis. The tone of the last clearly is " to think of coming i^ith such excuses, after I have been aituaed in this way ! " Lange ( ibid., p. 43) goes far afield in saying that the mood of the cum-clause is due to its dependence upon the infinitive, and Liibbert (Die Syntax von Quom, p. 140) makes it one of the two clauses in Terence which are in this mood because of a causal quom. If we are right in recognizing this force in the dependent temporal clauses above (with a main verb in the indicative), we shall also recognize it when a subjunctive clause with ubi or cum is found in dependence upon a subjunctive; for the existence of such clauses after an indicative proves that those which we find after a subjunctive are not necessarily in that mood because of attraction, but possibly by their own inherent force. I should like- wise suggest that many of the so-called causal and adversative gui-clauses in Plautus and Terence should rather be explained as due to the presence of this other force. For, if we are right in recognizing the subjunctive of surprise or indignation after a cum (cf. Bacch. 1192, Egon quom haec cum illo accubet inapectem^,) and ubi (Epid. 588, Non patrem ego te nominem ubi tu tuam me appellea filiam?), we should naturally do so after qui in clauses of the same nature, as in Pers. 27, deisne aduorser quasi Titani, cum eis belligerem, quibus sat esse non queaml and in Rud. 1244, Egone ut quod ad me allatum esse alienum aciam celeml Note how closely, in the two examples following, the qui-clause and the cum-clause correspond: M. G. 964, Vah, egone ut ad te ab libertina esse auderem internuntius, qui ingenuis satis responsare nequeaal Most. 896, Tibi obtemperem quom tu mihi nequeasl I do not wish to be understood as giving my assent to the view adopted by Dittmar, Lateinische Modus-Lehre, a view credited to Luchs by Guthmann (Uber eine Art unwilliger Fragen, Niimberg, 1891, p. 1; see also Stolz-Schmalz Gram. [1900] p. 370), that this force was the origin of the sub- junctive in qui-causal clauses. An examination of all the data will disprove the THE SOUECES OF THE CONSTRUCTION. 9 because of their own inherent force of surprise, indignation, or the like. Thus, the first of the following examples is practically equal to: Haec cum illo accubet, et ego inspectem! cf. Cic. Cat. 2, 8, 18, tu rebus omnibus copiosm m, et dubites! " You a man provided with everything,— and you hesitate ! " * With quom: Bacch. 1192, Egon, quom haec cum illo accubet, inspectem? Most 896, Tibi obtemperem, quom tu mihi nequeas? Heaut. 413-15, Verum quom uideam miserum hunc tarn ex- cruciarier eius abitu, celem tam insperatum gaudium quom ilh pericli nil ex indicio siet ? Hec. 341, Non uisam uxorem Pamphili, quom in proximo hie sit aegra ? Andr. 944, Egon huius memoriam patiar meae uoluptati oh- stare, quom ego possim in hac re medicari mihi ! Bacch. 285, Adeon me fuisse fungum ut qui illi crederem, Quom mi ipsum nomen eius Archidemides Clamaret dempturum esse! With ubi temporal: Epid. 588, Non patrem te nominem, ubi tu tuam me appelles filiam ? Men. 560-1, Egone hie me patiar in matrimonio, Ubi uir compilet clanculum quidquid domist Atque ea ad amicam deferat? Ubi local: Bacch. 1190, Egon, ubi filius corrumpatur mens, ibi potem? I should add to this list, for reasons given in the footnote (p. theory. One of the numerous weaknesses in Dittmar's general procedure lies in the fact that his theories are bolstered up by a few cleverly chosen examples which are interpreted to suit the theory in question. As for the interpreta- tion of the cum-clauses given at the beginning of this note, I am well aware of the fact that it is by no means certain. It may be that all of the sub- junctives with cum in Plautus and Terence (outside of cases of attraction or indirect discourse) are due to unconscious corruption on the part of copy- ists who were following the usage of their own times. Cf. Hale, Cum-Construc- tions, p. 211. Ubi aint of Pers. 76 may likewise be adversative. » This is classed in the Hale-Buck Grammar, § 503, as illustrating the extreme development from the volitive question. It is ordinarily classed as a Potential subjunctive. 10 ATTEACTION OF MOOD IN EAELT LATIN. 8), the qui causal-adversative clauses which show the same force. Cf. Asin. 838, An tu me tristem putas? Putem ego quern uideam aeque esse raaestum ut quasi dies si dicta sit ? C. Akin to the subjunctive of indignation in feeling, though of a different origin,^ is that which expresses " obligation or pro- priety/' and I accordingly treat it next. In Hec. 658, nunc, quom eius alienum a me esse animum sentiam, . . . quam ob rem redducam? the main verb asks a question of propriety in a tone of surprise, which latter feeling has full possession of the subordinate verb and makes it subjunctive. The same is true in Eun. 566, quid ego eius tibi nunc faciem praedicem aut laudem, Antipho, quom ipsus me noins quam elegans formarum spectator siem ? Again, I should add to this list many of the so-called qui-causal clauses which may be interpreted as expressing the same feeling, and depend upon verbs in the subjunctive of propriety. So, for instance, I do not see why subjimctives like the following, with qui, should be placed in a different class from those just cited with quom: M. G. 426, Quin ego hoc rogem quod nesciam ! M. G. 556, Vidi: cur negem qu4)d uiderim! Amph. 434, Quid ego ni negem qui egomet siem! The number of these examples in Plautus and Terence is large. There are also certain subordinate clauses depending upon verbs like mereo, which belong to this group. Thus in Men. 1067, non edepol ita promeruisti de me ut pigeat quae uelis, " you have deserved better of me than that you should express a wish and be disappointed," I should say that the mood of uelis is not due to attraction, but conveys the idea of propriety quite as clearly as pigeat does. The feeling of remonstrance is also strong in the cum-clauses of the following, in which ut siet must be classed as a substantive clause of propriety (I should not, with Liibbert, p. 81, take the Subjunctives with cum as due to mechanical attrac- tion) : iCf. Hale, Proceedings Am. Phil. As., Vol. 32 [1901], p. 120. THE SOURCES OF THE CONSTRUCTION. 11 Bacch. 140-2, Non par uidetur neque sit consentaneum, Quom haec intus intus sit et cum amica accuhet, Qjwmque osculetur et conuiuae alii accuhent, Praesentibus illis paedagogus una ut siet. In none of the above, then, is the dependent clause mechanically attracted. It may be said to be in the subjunctive because it is within the penumbra of the subjunctive shadow extended by the main verb. And it seems highly probable that the mechanical habit of assimilation was helped on by the occurrence of a great number of such instances. The part played by the anticipatory subjunctive* is probably the most important of all in the creation of this construction. In the first place, all futures, when thrown into the past, of neces- sity are expressed by the subjunctive.^ Epid. 501, Conducta ueni ut fidibus cantarem seni dum rem diuinam faceret. In which faceret represents a future dum faciet projected into the past. Contrast : Epid. 47, Ipse mandauit . . . ut fidicina quam am/ibai eme- retur sibi, in which amahat represents a present amat thrown into past dis- course. Other examples of the past future are: » The existence of such a type of the Subjunctive is now coming to be gen- erally recognized. See Hale's "The Sequence of Tenses" A. J. P., VIII (1887) p. 48 (act "in view"); Rodenbusch, De Temporum Usu Plautino, 1888; Sonnenschein, CI. Rev., vol. VII, Feb. (1893) ; Hale, CI. Rev. vol. VIII, April (1894), and Anticipatory Subjunctive, 1894; Schmalz, Lat. Gram. (1900) p. 370; Blase, Hist. Lat. Gram. (1903) III, p. 124. « See the Hale-Buck Latin Grammar, § 508-9, and Hale's Anticipatory Sub- junctive, p. 64, footnote. It is customary, of course, to treat the mood in such cases as due to attraction, or, if after verbs " sentiendi et declarandi," as due to indirectness of discourse. So Liibbert treats Amph. 128, and Bacch. 955 (cf. pp. 86 and 93). Holtze's treatment is similar. See pp. 192-5 passim. It must be understood, however, that the verbs in such cases could not have been in the indicative, and therefore it is impossible to say that they have been attracted. They are, as explained in the Supplement to the Hale- Buck Grammar, not due to the habit of using the subjunctive, but factors contributing to bring about that habit. 12 ATTRACTION OF MOOD IN* EAKLY LATIN. Pseud. 57, Ea causa miles hie reliquit symbolum . . . ut qui hue adferret eius similem symbolum cum eo simul me mitteret. Amph. 128, ut ne qui essem familiares quaererent uorsari crebro hie quom uiderent me domi. Cf. also Epid. 316, faceret; Epid. 356, redisses; Epid. 386, in- spexissent; Amph. 83-4, mandasset, fecisset; Amph. 225, uicti sint; Bacch. 955, scinderetur; Trin. 1144, darem; 314, esset; Adel. 109, eiecisset; Ilec. 545, egissem; Cure. 346, attulisset. It is not a great step from sentences like these to the following, in which one cannot be positive whether the original verb of the dependent clause was a future or a present: Bacch. 550, ille . • . accuratum habuit quod posset mali faceret in me. Cf. 554. Here posset could stand for either potero or possum. From such examples the step is short to the use of the imperfect subjunctive for a dependent clause thrown from the present into the past. Cf. Merc. 152, me rupi causa currendo tua ut quae scirem scire actutum tibi liceret ; Bacch. 788, orabat quod istic esset scriptum ut fieret. To such cases it is correct to apply the word attraction, since the verbs could be in the indicative if emphasis required it. Cf. Eun. 574, (ut) . . . essem una quacum cupiebam. In the second place, our study of early Latin seems to bear out the belief of Professor Hale* that " in such a case as di tibi dent quaequomque optes (* the gods grant you whatever your heart shall wish '), Plant., As. 44, we have the descendant of an orig- inal determinative anticipatory clause." " I have long believed," Professor Hale says,^ " that the anticipatory subjunctive supplies a large factor to the development of this construction (attraction). Especially in Plautus and Terence, a considerable proportion of the subjunctives of * assimilation ' after primary tenses seem easily to be accounted for as simply anticipatory, if we assume that the anticipatory power still remained to the mood at that time. This has been shown to be the case for Terence, in a paper written for the degree of A.B. by Mr. F. O. Bates, a student of mine at Cor- nell, in the year 1891-2." 1 Antic. Subj., p. 63, footnote. *Ibid., p. 64, footnote. THE SOUECES OP THE CONSTETTCTION. 13 An examination of the expressions of this kind in early Latin certainly gives strength to the belief that the anticipatory power still persisted in them, though the Roman grammarians, like the modern, may have lost sight of the origin and referred the mode to " attraction." The case of dum, meaning " while " and " as long as " and pointing to the future, furnishes a good illustration. Whenever its clause is subjunctive in early Latin and depends upon a subjunctive, it is the practice to say that the mood is due to attraction. In this way Boettger (De Dum Particulae Usu, etc, p. 20) explains dum-clauses in the following: habet haec ei quod dum uiuat usque ad aurem obganniat, Phorm. 1030 ; ita uelim me promerentem ames, dum uiuas, mi pater, ut me . . . dolet? Adel. 681 ; and ut dum uiuat meminerit semper mei, Heaut 951. In the same way the dum-clauses in the following are explained by Elste (De Dum Particulae Usu Plautino, p. 26) : quid dotis? egone; ut semper, dum uiuat, me alat. Cure. 664; neu sessum ducat, dum histrio in scaena siet, Poen. 20 ; f aciam ut mei memi- neris, dum uitam uiuas, Pers. 495 ; quaeso dum uiuas uti omnes tui similes hospites habeas tibi, Rud. 499. And Richardson (De Dum Particulae apud Priscos Scriptores Latinos Usu, p. 72) similarly explains the dependent clause in Epid. 501, \diich is a sentence of the same kind, in dependence upon a verb in the past. But the occurrence of subjunctives after dum with the same ' meaning, which do not depend upon subjunctives and which clearly show anticipatory force, proves that the cases just men- tioned should be considered as instances of the anticipatory sub- junctive. The following will illustrate what I mean: True. 716, Ego^ interim hie restiti tricis praesidebo iste dum sic faciat domum ad te exagogam. Cf. TOippa ' . . xT^fiaT' idouTou Off pa xe xecun^ toozov iyji ^oov. ^ 123. ofp di^ fiiv xeu doOpaz" iu dp/iopcjjffeu dp^prj, zinpf abzoh pLtvio), e 361. True. 103, Oenus eorum aliquis^ oculum amicae usque oggerit dum illi agant. The same subjunctive with other particles is illustrated by the following : Amph. 439 : ubi ego Sosia nolim esse, tu esto sane Sosia.* »Thi8 interpretation is given by Lange, De Sententiarum TemporaUum apud Prise. Scr. Lat. Syntaxi, p. 41. ATTBACTION OF MOOD IN EABLY LATIN. Cf. ijrijv d^ fJ^ot diuaricentur quos relinques. Cato, R. R. 103, qui fastidient cibum ut . . . adpetant . • . spar- • gito, Cato, R.R. 142, ut quae opus sunt, parentur. Most. 1095, ne hue confugere possint quaestioni quos dabit Amph. 608, caue quicquam nisi quod rogabo te mihi responderis. Bacch. 989, uolo ut quod iubebo facias. M. G. 254, inducamus uera ut esse credat quae mentabimur. Cur. 432, quaeso qui has tabellas adferet ut ei detur. M. G. 81, qui auscultare nolet, exsurgat foras. cum. Cato, R. R. 23, quom pluet, quala parentur. Cato, R. R. 25, quom uinum coctum erit facito ut ser- uetur. Cato, R. R. 40, quom praecides caueto ne librum conuellas. Cato, R. R. 45, ne liber laboret cum dolabis. Cato, R. R. 45, ne librum scindas cum adiges. Cato, R. R. 54, ne sectentur cum arabunt Pseud. 163, haec quom reuortor facite ut offendam parata. M. G. 1176, quom extemplo hoc erit factum, facito ut uenias. Mer. 146, ne laborem capias quom illo uti uoles. Rud. 1206, adoma ut rem diuinam faciam, quom aduenero. Asin. 372, mox quom Saurean imitabor, caueto ne suscenseas. Amph. 197, meditabor quo modo dicam quom aduenero. Bacch. 826, ut auferam quom ilium rescisces. Hec. 575, uereor ne orata nostra nequeat celare quom sciet. Hec. 769, the sources of the construction. 23 ubi. qaando. quotiens. quo. quam. dum (= while). dum (= until). quom tu eris satura ut puer satur sit facito. M. G. 811, ut cum hie agit, actutum partis defendaa tuas. Cato, R. R. 86, ubi coctum erit lacte addat. Cato, R. R. 156, ubi libido ueniet . . . decumbat Cato, R. R. 156, ubi uersus ibit heminam . . . bibat. Cato, R. R. 95, ubi erit crassum . . . sinito frigescat. Pers. 86, curate . . . ne mihi morae sit ubi intro aduenero. Bacch. 36, ubi me fugiet memoria . . . facito ut subuenias. Capt. 342, ubi erunt indutiae illuc, qui conueniat pa- trem. Epid. 595, ubi noles, ne fueris pater. Pers. 384, uideto, me ubi uoles nuptum dare, ne faciat. Most. 403, neu quisquam responset quando aedia pultabit. Bacch. 224, ueniat quando uolt. Hec. 619, utrum illaec fecerint quando haec aberit. Aul. 78, ut faciam litteram longam, meum laqueo collum quando obstrinxero. Poen. 1409, quando ex nemo emissu's, compingare in carcerem. Cato, R. R. 151, quotiens opus erit, purges. Cato, R. R. 151, quotienscumque opus erit facito ut addas. Cato, R. R. 21, ne foramina maiora fiant quo inden- tur. Men. 1044, dicam ut abeat liber quo uolet Cato, R. R. 66, quam diligentissime poterit, tollat Rud. 779, ego dum abes ut abeat non sinam. Rud. 558, tibi copiast dum lingua uiuet qui rem soluas. Poen. 1421, dum auctionem facio opuat ut maneas. Rud. 880, suadeo ut ad nos abeant potius, dum re- cipis. True. 874, ut . . . sinas . . . dum ali- quo miles circumducitur. Phorm. 513, ut me maneat . . . triduom hoc dum id . . . aufero. Eun. 894, uin interea dum uenit domi opperi- amur ? Rud. 773, oro ut illas serues . . . dum ego erum adduce meum. M. G. 1333, ne inter- ueneris quaeso, dum resipiscit Add. 786, nisi • . . dum haec silescunt, abeam. 24 ATTEACTION OF MOOD IN EAKLY LATIN. priusquam Epid. 304, ne abitas priusquam ego ad te uenero. antequam, Bacch. 381, priusquam malum istoc addis cer- tumst dicam. Gato, R. R. 161, ne ante sarueris quam asparagus natus erit. Asin. 448, nunc adeam optimust priusquam incipit tinnire. Phorm. 719, ut conueniat banc priusquam hine abit Phorm. 898, priusquam dilapidat . . ut auferamus. M. G. 1408, obsecro . . . te . . . ut audias priusquam secat Andr. 558, prius- quam . . . redducunt animum, uxorem demus. Poen. 1399, ut minam . . . reddas priusquam . . . abducere. Bacch. 440, at . . . priusquam septuennis est, si attingas eum. donee. Cato, R. R. 86, lacte addat donee crassus erit f actus. Bacch. 758, ne quoquam exsurgatis donee a me erit signum datum. Before we draw conclusions from these tables, a word of warn- ing must be given. It will be noticed that the usage of Cato diverges noticeably from that of the other early writers, a fact which is easily understood by the reader of the De Re Rustica. When Cato has once fairly entered his subject, his work becomes practically a string of sentences shaped after this mould : quom (ubi, si, etc.) . . . erit, facito, with remarkable monotony. There are over three hundred sentences of this form in the little volume. Naturally, then, the future indicative becomes so fixed in the dependent clause, that one could hardly expect it to be attracted to the subjunctive when, for any reason, a jussive sub- junctive is used instead of an imperative in the main clause. That I am right in saying that the future indicative becomes fixed, is proved by the fact that in the first few chapters, several cases of attraction occur before the use of the regularly recurring future indicative with the imperative becomes noticeable ; after that they are remarkably rare. I would also add that conclusions drawn from my lists in regard to priusquam-, dum-, and si-clauses, would be erroneous if accepted without modification (see pp. 52-5). After dum = until, it is customary to classify the verbs as inher- ently subjunctive ; but the very fact that about sixty per cent, of the verbs after dum = until are in the indicative^ in early Latin, 1 Hale-Buck, Grammar, 571, and footnote, for the explanation of this fact. THE SOtJECES OF THE CONSTRUCTION. 25 and that over half of those which are in the subjunctive are also in dependence upon subjunctives and infinitives, goes to show that many of the latter must be due to attraction. In general the same condition of things obtains with the priusquam-clauses. With si-clauses, the feeling of the less vivid future is so subtile in early Latin that, in the majority of cases where the present subjunctive is found with si in dependence upon another subjunctive, I believe it is impossible to say with certainty whether the writer would have used an indicative, if the clause had not been thus dependent (see p. 54). In cases of this class, accordingly, it must be under- stood that the list of subjunctives is not as complete as that of the indicatives. A fair comparison of the two uses may be made in the temporal clauses with cum, ubi, and quando, as well as in the qui-clauses of Plautus and Terence. The following table will show a marked preponderance of the subjunctive in clauses which refer to the future and depend upon another subjunctive. Subjunctive Indicative This proportion of the subjunctive to the indicative becomes sig- nificant when one finds that, in early Latin, attraction is rather the exception than the rule, and that about thirty-five per cent, of the clauses which, so far as function and position are concerned, are capable of being attracted, are in fact so affected. The tense, therefore, is significant; and the examination has confirmed the theory that the anticipatory subjunctive was still a force in Latin during the time of Plautus. If then it is found to " supply a large factor to the development of attraction," it will readily be seen from the list just given that the factor is exceedingly important, for this class alone makes up about 55 per cent, of all the early Latin verbs that are usually classed as subjunctives by attraction in the grammars. £. The note quoted above (p. 4) from Professor Hale's paper on the Anticipatory Subjunctive gives a good illustration, from conditional sentences, of the kind of verb now under discussion: C. I. L., I, 196: Sei quis esent quei sibei deicerent necesus ese cum. vbi. guando. ««I. Tolai. 40 15 8 35 08 10 5 5 10 30 ATTRACTION OF MOOD IN EARLY LATIN. Bacanal habere. Cf. Trin. 472, siquid tibi placeat qiiod illi con- gestum siet, edisne? in which the conditional feeling is nearly as strong in the quod-clause as in the «i-clause. In the following there is more or less of the conditional feeling. It will be noticed that the dependent clause in the second example is, strictly speak- ing, in grammatical dependence upon the apodosis of the sentence, that is, upon a subjunctive of Ideal Certainty. Quod uideas is equivalent to si quid uideas. Amph. 871, Nam mea sit culpa quod egomet contraxerim Si id Alcumenae innooenti expetat. Men. Ill, Ni mala, ni stulta sis . . . Qtiod uiro esse odio uideas tute tibi odio habeas. There are numerous examples like the following: Capt 205, At pigeat postea nostrum erum si uos eximat uin- culis aut solutos sinat quos argento emerit, in which there is a break in time-sphere, and the dependent clause, if emphasis were given to its actual time-feeling, would not be in the subjunctive ; but the speaker for the moment conceives of the act from the ideal rather than the actual point of view, and ex- presses this feeling by using the same mood as in the si-clause. Cf. Bacch. 778, Ni facta cupiam quae is iielit (strictly, " what he now wishes"; ideally, "whatever he may wish") tua iam uirgis latera lacerentur probe. See also Cure., 269; Bacch., 564; Hec., 555. F. In the same way a clause depending upon a subjunctive of Ideal Certainty may, strictly speaking, state a general truth or an objective fact, and at the same time express that fact ideally, that is, with the same mental attitude with which the rest of the clause is uttered. The dependent verb is thus thrown into the same mood and tense with the verb on which it depends. A case in point in the present tense may be seen in Cis. 497, quodcumque optes tibi uelim eontingere, " if you should desire a thing, I should like you to have it" Contrast Cure. 82, Nam istunc qui fert affictum uelim. The following is an example of a clause depending upon a verb contrary to fact: Bacch. 488, si opperiri uellem . . . , plus uiderem quam deceret, " more than would be fitting," instead THE SOURCES OF THE CONSTRUCTION. 27 of " more than is fitting." Contrast Adel. 108, si homo esses, si- neres nunc f acere dum per aetatem licet. Deceret has to my mind the same tone as uiderefn. The same relation holds true in the dependent clause of the following in Rud. 1261, dum praedam habere se cemeret, interim praeda ipsus esset In the following, on the other hand, we seem to have a clear case of attraction: Poen. 681, uidere equidem uos uellem quom huic aumm darem, " I should like to have you see when I give him the money." The fact seems to be that, after conditional subjunc- tives and subjunctives of ideal certainty, Latin freely conceives of the subordinate act from the ideal point of view, unless the fact of Its objective reality is to be emphasized, as in Aul. 482, inuidia nos minore utamur quam utimur,OT unless its time-sphere is clearly different from that of the governing clause, as in Asin. 860, nun- quam f aceret ea quae nunc facit. Complete lists of examples may be found on pp. 40-1. G. The indirect question was evidently passing through an im- portant change in the time of Plautus. We need not here discuss the process of the change; suffice it to say that the subjunctive was coming to be the mood for these expressions, and that a feel- ing was arising which demanded the subjunctive for all indirectly quoted questions ; and when the quotation included a dependent clause, this too naturally came to be put into the same mood if It was felt to be an integral part of the original question. Thus, m the foMowing sentence, Mera 623, quin percontatu's, hominis quae f acies foret qui illam emisset ? the quoted question includes a relative clause, and the whole question " hominis quae f acies erat qui illam amiserat " is thrown into the subjunctive. For the sake of comparison, I quote the following, in which the relative clause IS not conceived of as a part of the quotation : Trin. 7, quae iUaec siet hue quae abiit intro dicam. I would suggest that the origins of attraction in clauses of this sort go even deeper, and that the soKjalled relative clause in many instances was still, as in its origin, actually felt as an interroga- tive in dependence upon the main verb of the sentence.^ I mean >Cf. Sommer, Handbuch der Lateinischen Laut- und Fonnlehre, 293- Hervorgegangen ist die relativ FunkUon hauptsacUich aus den indirekte^ 28 ATTRACTION OF MOOD TS EARLY LATIN. II i that in a sentence like the following: Aul. 29, Is scit quae »it quam compresserit, the speaker may have felt practically the two questions, quis est? quam compressiif Compare Poen. 1027, Narra, quid est, quid ait! ibid. 711, Quid est? quid uoltis, tes- tes ? It must be noted that in the example quoted, the interroga- tive feminine becomes quae in the indirect quotation, though in the direct question it would be quis as a substantive in Plautus. This fact would facilitate confusion between the relative and in- terrogative in the case of the feminine pronouns. In the case of the masculine, since qui and quis^ both occurred frequently as in- terrogatives, there must have been frequent confusion. Of course, in the majority of indirect questions, the possibility of interpret- ing the dependent relative clause as an interrogative is precluded by the unmistakable nature of the clause. It is entirely out of the question in sentences like the following, as a glance will suffice to show: Pseud. 451, quanto satius est . . . exquaerere sint ilia necne sint quae tibi renuntiant; Aul. 17, Coepi obseruari ecqui maiorem filius mihi honorem haberet qiuim eius habuisset pater. Sentences like the following, however, would allow the possibility : Amph. 1016, pergam exquirere quis fuerit quern propter corpus 8Uom stupri compleuerit; M. G. 261, hominem inuestigando operam huic . . . dabo qui fuerit conseruous qui hodie sit sectatus eimiam. Fragesfttzen, die ja oft der Bedeutung von Relativsatzen sehr nahe kommen. Delbrtick, Vergleichende Syntax, III, p. 403, illustrates well the ambiguity of some relatives. He says: Das Relativum entsteht in abhangigen Fragen. £s giebt viele abbftngige FragesUtze, in welchen man das Pronomen ebenso gut interrogativisch wie relativisch aufTassen kann z. B. concrepuit ostium, videamus, qui bine egreditur Men. 348, * seben wir nach, wer kommt bier heraus?* oder * seben wir nacb dem, der bierauskommt.' Tbe literature on the question of the relation of tbe relative pronoun to the interrogative and indefinite pronouns can be found by means of tbe list of references in Delbrtick, ibid., p. 400. » Cf . True. 708. Nunc speculabor quid ibi agatur, quia cat intro, qui foras tieniat. Cf. also Sommer, Handbuch der Lateiniscben Laut- und Formlebre, 293: FUr die Flexion machte urspHinglicb der Bedeutungsunterschied nicbts aus, das Paradigma war fiir alle drei Verwendungen dasselbe. In R. R. 145, homines eos dato, qui placebunt aut custodi aut quia eam oleam emerit, Cato uses quis as a relative. What has been said in this paragraph must, however, apply to written rather than spoken sentences, for I am aware of the fact that the stress of an interrogative pronoun is often not that of a relative. Cf. Seyffert, Berl. Phil. Woch., 1891, p. 108. THE SOURCES OP THE CONSTRUCTION. 29 We have also noticed (p. 21) that the anticipatory indirect questions must be reckoned with in this account, if we are right m supposing that their force will help to account for the mood of verbs like adimat in the following: Phorm. 161, expecto quam mox ueniat qui adimat banc mihi consuetudinem. Summary. We have now examined the more objectiTO forces which were at work m developing the construction in question, and with our eye upon the facts so far found, are ready to make a partial state- ment in regard to what attraction is and what it is not. It has become clear, I think, that attraction is more than a « Streben nach Concinnitat des Ausdrucks » (Draeger, I, p. 316) in each indi- vidual instance,— it is not even a habit which springs purely and simply from such a streben (" hervorgegangen," ibid.).' The phrase " Streben nach Concinnitat " implies a feeling for stvle which comes with a highly developed prose, and I fear definitions like the one just quoted come from an examination of attraction as used by Cicero in his best prose,— where it actually merges into such an " efifort at concinnity,»— and not from an historical study of the construction. This is rather a habit which took shape under the influence of the several similar constructions that we have dis- cussed in the preceding, and it worked outward from those definite starting-points along the lines governed by the laws of analogy. True It IS that in the examples of the original type, the subjunc- tive of the dependent clause was due to the fact that this clause happened to contain the same modal feeling as the governing clause. But in the growing construction of attraction proper, Identity of mental attitude is not a sine qua non of its occurrence. Attraction then appears to be in the main a phenomenon of nat- ural linguistic evolution which falls under the category of analogy working mechanically, rather than of a conscious (or unconscious) mental attitude» connected with stylistic considerations. As we have seen, its origins are found in the language, not of artistic n have been careful to say that this is only a " partial statement," for I fully understand the danger of neglecting the element of i,vxu^ S^tau: in th« discussion of a construction as subjective as this. There are psychological considerations that must be taken into account before the final definition is given. I shall have something to say on this point in a later paper. 30 ATTBACTION OF MOOD IN EARLY LATIN. prose, but of such prose as we find in Early Latin. They neces- sarily belong, of course, to the post-paratactic period, but to that part of the period while style was still quite simple and unadorned. The very fact that these origins are to be placed in a fairly well- developed stage of the language, and yet before the feeling for the intricate and finely wrought periods of Cicero's day arises, is a matter of great significance, and again points to the same conclu- sion, that the construction is not to be traced to considerations of style. What the real genetic forces were we have pointed out in the preceding pages, and have illustrated them by sentences in which the dependent clause directly expressed the same feeling as the independent clause, or a feeling kindred to it. These sentences we have found in connection with the following subjunctives: Jussive, Permissive, Deliberative (and its extensions), Dependent Volitive, Optative, Subjunctive of Obligation or Propriety, An- ticipatory (present-future and past-future), Conditional Subjunc- tive, Subjunctive of Ideal Certainty, and Subjunctive in Indirect Questions. In other words, the origins of this construction are found to exist in connection with almost every kind of subjunctive of any importance in Early Latin, and the actual instances of the original kind are there found to be more than half of the verbs usually considered as cases of attraction. m CHAPTER II. The Uses op the Consteuction, Tables. Having discussed the genesis of the construction, we must now see how It behaved, how far it extended, and what forces opposed its workings. ^ I shall first give a list of subjunctives attracted by volitives in the primary clause, placing in a parallel column unattracted verbs of the same general nature, for convenience of comparison. I shall group these with reference to the relative closeness of time between the dependent and independent verbs, for we have already seen that a shift of time-sphere is a most important consideration in the matter of attraction. The remaining groups will then be treated m the same way in their proper order. It is also necessary to note that identity of grammatical tense does not by any means imply identity of actual time. A univer- sal present may, for instance, depend upon a present volitive sub- junctive; both may be in the present subjunctive, while, in respect to actual time, the volitive is future, and its subordinate verb is present. Of course, for the moment the speaker may be in the future attitude of mind with regard to both verbs, but he is not necessarily so. This general consideration must be applied to verbs m the other tenses as well. The following tables do not contain the verbs which refer to the future, as a full list of these has been given. On the other hand 1 have included the examples of the « subjunctive of like feeling " even when they have already been listed in the preceding, for It IS of course quite impossible to state in any given case that the force of mechanical attraction was not at work. For past-futures see p. 12. Except in the case of indirect questions, where I found It necessary to give complete statistics, I have excluded the verba which were clearly in indirect discourse-explicit or " implied " Such verbs usually depend upon clauses introduced by impero, postulo, etc., or by causal quod. ' 31 32 fW. i ATTBACTION OF MOOD IN EAELY LATIN. Group I. In which the main verb is in the subjunctive of will or wish. 1. Both the governing and the dependent verbs are in the present tense. ^iubmIo. Dependent verb not attracted (Indicative). (a) The governing Rud. 1229, habeas quod di dant. Asin. 644, facias quod suades, Bacch. 990c, iubeo; Asin. 180, lubet; Cist. 768, est; Bacch. 993, iubeo; Most. 594, est; Trin. 351, habes; Trin. 979, lubet; Andr. 393, facts; Heaut. 177, excruciat; Eun. 78, hahet; Hec. 810, refert; Adel. 622, placet; Men. 349, egreditur; Capi 609, uolt; Aul. 542, ha- bent; Men. 353, est; M. G. 1054, oro; Andr. 697, uolunt; Heaut. 464, lubet; Eun. 529, uolt; True. 233, habent; Gate, R. R. 4, mu- to»^; Aul. 776, uolt; Cure. 180, es<; Merc. 991, uolt; Most. 222, iioltin*; Most. 306, gaudent; Pers. 373, uolt; Afranius, 407, lubet; Cato, Frag. p. 79, opus est, necesse est. Phorm. 519 es; Phonn. 165, amo; Heaut. 589, extrudis; Hec. 579, exopto; Hec. 469, nuntiant; Men. 308, habi- tant; Asin. 841, uolunt; Eun. 665, uolunt; Andr. 931, spero; Pseud. 108, dicis; Rud. 158, quaero; Rud. 992, ^a5e<; Stich. 320, refert. True. 163, liJi moriuost quies- Cftt. Trin. 671, copiast. Dependent verb attracted (Sub- junctive). verb is independent. Capt. 548, ne tu quod istic fabuletur auris immittas tuas. Most. 1100, quod agas id agas. Pseud. 570, det locum illi qui queat. Bacch. 656, furetur quod queat. Lucil. 572, concedat homini id quod uelit. M. G. 1038, di tibi dent quaequomque optes. t. THE USES OF THE CONSTEUCTION, TABLES. 33 qUOM' quantofn. dum. ut. qtMin, quantum. quo. qui. Most. 871, quom impluit ne impluat mi. Bacch. 536, adue- nis; Capt. 355, honestas; M. G. 1419, odes; Poen. 668, datis; Poen. 687, uis; Adel. 918, uideo; Trin. 671, est. True. 127, aduenis. Merc. 553, dum potes ames; True. 163, uiuit. Asin. 731, nunc ut est elo- quamur; Adel. 399, est; Tur- pilius 191, meres. Stich. 44, nos faciant quam aequomst. Aul. 785, di quantumst per- duint; Heaut. 870, est; Pseud. 37, est. Cato, R. R. 1, ambulant. Amph. 960, en ut sint, ipse item sit. Bacch. 661, utcumque res sit ita animimi habeat. Hec. 634, turbent quam uelint. Pseud. 936, tantum tibi boni di duint quantum tu tibi exoptes. Aul. 491, quo lubeant nubant (6) The governing Aul. 576, quod habeo ut com- mutet; Aul. 546, sospitent quod nunc habes; Amph. 870, accusat; Amph. 970, opust; Capt. 329, sentio; Capt. 908, pendent; Cas. 311, postulas; Cas. 512, para- ^ttww^; Cist. 632, oportet; Cist. 13, arbitrar; Cure. 159, agimus; Epid. 268, «0?^; Epid. 456, aJttc- mo; Men. 427, uolo; Men. 558, dan<; Men. 955, opus est; Mer. 937, expetis; M. G. 728-9, pro- bast, improbast; M. G. 945, agendumst; Most. 903, mulcet; Pseud. 12, nesao; Pseud. 168, ibist; Stich. 26, metuis; True. 722, MoZo; Andr. 339, e^^; Trin. 654, habes; Heaut. 867, cupis; Hec. 325, cs^; Adel. 706, sunt; Lucil. 36, impendet; Lucil. 440, ver6 w dependent. Cas. 252, domuisti animum ut quod uir wc7t< fieri id facias? Epid. 283, iam amota ei fuerit omnis consultatio nuptianim ne grauetur quod uelis. Pers. 601, ut tibi percontari liceat quae uelis. Trin. 221, pauci sint faxim qui sciant. Amph. 630, ut quae imperes compareant. Men. 994, Caue quisquam quod illic minitetur flocci fecerit. M. G. 41, ut praeolat mihi quA)d tu uelis. Pseud. 207, faciant aduersum eos quod nolint. Phorm. 125, ut orbae qui sint genere proximi eis nubant. 34 ATTRACTION OF MOOD IN EAKLT LATIN. M. dwn. «t intellego; Cato, E. E. 14, sunt; Amph. 879, grauidast; Aul. 251, uis; Bacch. 863, publicat; Cas. 107, rftpem; Cas. 206, Ivbet; Cas. 239, decent; Asin. 256, «o- Z«n<; Cure. 34, palamst; M. G. 227, «un^- M. G. 1050, cwptf; Pers. 74, oppugnant; Pers. 372, lice*; Trin. 341, uis; Trin. 641, protneres; Phorm. 42, hdbent; Phonn. 533, dandumst; Hec. 674, nolo; Hec. 768, opust; Adel. 54, faciunt; Adel. 511, poie5. Pacuu. 282, rogo; Turpil. 146, iioZo; Accius, 509, accolunt; Cato, B. E. 23, sunt; Cur. 428, peto; Trin. 979, tw)?o; Andr. 825, cupis; M. G. 1229, amo; Pers. 293, opto; Eud. 1256, est. Eun. 537, amabo ut illuc tran- seas ubi illast. Hec. 385, cum orata eius rc- miniscor nequeo quin lacrumem; Merc. 178, qttom malum audien* dumst flagitas me ut eloquar. Adel. 711, ne imprudens fa- ciam forte qv^d nolit, sciens cauebo. Trin. 211, non flocci faciunt dum illud qv/)d Ivbeat sciant. Andr. 306, quaeso ... id uelis quod possit. Cure. 29, ne id quod ames . . . tibi sit probro. M. G. 1230, quod cupiom ne grauetup. Eud. 874, ut id quod quaerant inueniant M. G. 595, ne dum absum sor- titae fiat; M. G. 1317, dat; Eud. 123, sudumst; Adel. 312, est; Pseud. 922, dormit; Asin. 914, litigant; Asin. 531, expedamus; Andr. 556-7, datur. Andr. 623, ut sumam suppli- cium ut nolo; Amph. 559, sunt; Amph. 982, intellegis; Cas. 158, est; Men. 861, minatur; Merc. 989, uolt; Phonn. 1020, sunt. M. G. 1086, ne magis sim pul- cher quam sum; Asin. 268, Lu" lentiast; Heaut. 681, uolt; M. G. Andr. 160, ut consumat nunc quom nil ohsint doll. M. G. 1342, nequeo quin fleam qu^OM abs te abeam. Bacch. 907, ut eirai castigem quom haec facta ad hunc faciat modum. Poen. 884, (metus) dum ero insidias paritem ne me perduim. Capt. 343, alium misero — qui tua mandata ita ut uelis perferat. Pers. 237, nunquam extere- brabis, tu ut sis peior quam ego siem. quantum. qno. qui. quantum. cum. qu%. oiiorgum atque. qui THE USES OF THE CONSTEUCTION, TABLES. 35 2, Solent; True. 63, sunt; Men. 192, impetrant; Poen. 694, solent; Capt. 443, sum. Bacch. 348, ut conueniam quan- fum potest; Heaut. 645, est; Pseud. 938, dignu's. Pseud. 470, ut possint quo uolo. 2. Both verbs are in the imperfect tense. Eun. 574^ut essem unaquacum Merc. 152, rupi currendo ut Hec. 729, ne minus hinc im- petrem quam possiem, Andr. 577, is mihi suadet nup- tias quantum queam ut maturem. cupiebam; Epid. 47, mandauit ut fidicina quam amabat emeretur sibi. quae scirem scire tibi liceret. Bacch. 788, orabat quod istic esset scriptum ut fieret. Cist. 85, ut me quern ego amarem sineret cum eo uiuere. Bacch. 674, occasio . . . fuit . . . ut quantum uelles tantum sumeres. Phonn. 733,1 ut facerem egestas me impulit quom scirem infinnaa nuptias hasce esse. 3. Both verbs are in the perfect tense. M. G. 588, quin id adimatur ne M. G. 149, faciemus ut quod Id quod uidit uiderit; Poen. 951, uident ne uiderit. ueni; Pers. 478, credidi. Eun. 82, uereor ne dliorsum atque ego feci acceperit. M. G. 370, numquam deteixe- bor quin uiderim id quod ui- derim. 4. The governing verb in the present tense with the depend- ENT VERB IN A DIFFERENT TENSE. (a) The dependent verb in the perfect tense. 1) live governing verb independent. Aul. 433, utinam auferam quae tuli; Poen. 193, hoc agamus quod cepimus; True. 9, uentumst; Men. 1104, pollicitu's; Men. 451, commentust; M. G. 1010, sum agressa; M. G. 1100, instruxisti; » The mood of scirem may be due to the adversative force of the clause. Cato, R. R. 5, quae dominus imperauerit fiant. Ibid, quod dominus crediderit exigat. 36 ATTILACTION OF MOOD UST EAELY LATIN. «1 eum. ^fM» quanio, quam, vt and com' paruona. Heaut. 31, fecit; Heaut. 745, at* iulerunt; M. G. 1127, instraxisti; Eun. 302, remoratust; Plant. Frag. 21, and Aquilius. 1, rep^ perit; Naeu. 19, protulit Trin. 63, and True. 844, habeas ut nactu's. Poen. 208, ohiulisii. Cato, H. H. 7, salutauit, 2) The governing Amph. 291, imperauit; Amph. 948, noui; Aul. 278 imperauit; Capt 515, oraui&ti; Cas. 503, mandaui; Cas. 512, condiuit; Men. 445, imperatumst; Men. 686, commisi; Men. 991, tm- peraui; Men. 1057, fecisii; Merc. 669, fugit; M. G. 949, conduxi; Pseud. 639, missus sum; Eud. 587, potaui; Trin. 1123, egi; True. 893, perdidi; Amph. 629, imperaui; Asin. 38, locutu's; Aul. 671, /ea<; Cure. 272, por- tentumst; Cure. 433, emi; Baceh. 1020, obiurigauit; Men. 672, dedi; Cure. 464, locaui; Most. 416, turbauimus; Pers. 613, tt«- «f<; Trin. 141, concreditumst; Heaut. 1067, fecit; Hec. 54, com^ misit, Trin. 616, euortit. Merc. 425, dum ne minoris uen- das qvam emi; Men. 1033, fui. Men. 1146, tit iusti; Pers. 616, u< accepi. Naeu. 7, «/ est; Pseud. 1020, ut fuit; Pacuuius 297, ut ego egi, ago, OQcim; M. G. 257, exorsi sumus; Phonn. 31, usi sumus. verb dependent. Phorm. 845, ut haec quae con^ tigerint sciat. Most. 413, quae dissignata sint . . . tranquille cuncta ut proue- niant. Trin. 1105, iubeto Sagarionem qua£ imperauerim curare ut ef- ferantur. Phorm. 272, non causam dico quin quod meritus sit ferat. Heaut. 1040, ut semes quod labore inuenerit. Poen. 7, ut . . . sedeant . . . qui uenerint. Rud. 1242, ut cum maiore dote abeat qu^im aduenerit. iJUl. «1. quam. quu fm. quom. priusquam. quL quu THE USES OF THE CONSTEUCTION, TABLES. 37 (h) The dependent verb in the imperfect or pluperfect iense. Aul. 33, quo facilius ducat qui compresserat; Capt. 939, reli- queram; M. G. 132, fuerat; Men. 426, dederam; Cas. 514, erat. Andr. 543, fuerant. Capt. 247, seruibas. 5. The governing terb in a past tense with the dependent vebb in a different tense. (a) The dependent verb in the present tense. Andr. 583, ne faceres quod uol- gus solet; Andr. 793, ut sciret quae uolumus; Bacch. 689, lo- ', quar; Pers. 433, faciunt; Phorm. 656, debeo. Eun. 93, dolet; Trin. 375, re- percis. Merc. 427, est. : (&) The dependent verb in the perfect tense. Cas. 933, ut quo ego bibi, bi- beret; Merc. 230, ne noceret qwim habui; M. G. 74, const- grmui; M. G. 186, uidit; Pseud. 72, sciui. Rud. 498, adduxisti. Capt. 537, penisti. (c) The dependent verb in the pluperfect tense. Capt. 720, ut melius consu- Cure. 550, tuom qui signum lerem tibi quam iUi quicum ... ad me attulisset nuntium ne exegeram. spernerem. Group II. In which THE main VERB IS IN THE SUBJUNCTIVE OF DELIBEBA- TION OR AN EXTENSION OF IT. (a) The main verb is a subjunctive of deliberation. Hec. 445, quo pacto celem qu^d me orauit; Stich. 675, habito; Eun. 1046, fuit; M. G. 199, uidit. i (in. H. M. U. S8 ATTBACTION OF MOOD IN EABLY LATIN. (h) The main verb is a sv^junctive of *' surprise, remonstrance, or indignation." Eun, 47, non earn quom accen" [cf. p. 9. Most. 896, Heaut. sorf 413-15; Hec. 341; Andr. 944; Bacch. 286; 1192; Trin. 733; Epid. 588; Men. 560-1; Bacch. 1190; Phorm. 970-2.] Bacch. 66, penetrem ubi desu- dasciturf Asin. 94, Ten defrudam quoi esi.^ Asin. 885, habet; Men. 560, domist; Men. 763, expetit; Most. 301, cupiof Heaut. 784, daturas sum; Andr. 271, crcdidit! Adel. 677, ueneramf Phorm. fH), nhi quae Inhitum fuerit peregre feceris . . . uenias! cf. p. 7. Merc. 702, em quoi tua quae habeas commendcs uiro ! Most. 494,* mirum quin uigi- lanti diceret qui . . . occisus forei, Pers. 340,' mirum quin regis Philippi causa . . . potius uen- dam quam mea quae sis mea. Group III. In which the main verb is in the subjunctive of ** OBLIGATION OR propriety." (a) Both verbs are in the present tense. M. G. 55, quid . . . dicam quod [See p. 10 and Hec. 658; Eun. omnes sciuntf Phorm. 345, ea qui praehet, non hunc habeas . . . deum? 566; Amph. 434, etc.] Hec. 760, meritus ... est quod queam ut commodem. Men. 1067, ut pigeat quae uelis. Men. 1100, promeruisti ut mi- quid ores quod uelis. > There arc a great number of verbs in the subjunctive which correspond to this indicative, but since they are usually classed among the qui-causal and adversative subjunctives, I have omitted them. Cf. Amph. 434, quid ego ni negem qui egomet siem, and see footnote p. 8. What is here stated will also apply to the clauses of the same nature in the next group, no. III. In group II, I have not classified as regards tense, for reasons which have been given, p. 7. •Th«9e may be qui-causal subjunctives. THE USES OF THE CONSTRUCTION, TABLES. 39 quam. quom. m. qui. quu Asin. 49, cur filio suscenseam patres ut faciunt ceteri. Eun. 75, ut redimas quam queas minimo ... at quanti queas. Stich. 114, ut quom ambulent ... OS obturent. Bacch. 139-^2, non par uidetur neque sit consentaneum quom haec intus intus sit et cum amica accubet quomque osculetur et conuiuae alii accubent praesen- tibus illis paedagogus una ut siet. (b) Both verbs are in the imperfect tense. Hec. 230, quae hie erant cu- rares; Poen. 391, dicebas. (c) The dependent verh is not in the same tense as the main verb, Cas. 701, cur non impetrem quod coepi; Trin. 1024, periit; Most. 435, uoluisti; Rud. 1397, sum iuratwi; Pers. 637, fuit; Aul. 222, fads; Phorm. 468, domist. qui. UL quam, dum. Group IV. In which the main verb is in the potential subjunctivb. (a) Both verbs are in the present tense. Asin. 234, habeo unde quod Capt. 937, lingua nullast qua poscts dem; M. G. 82, uolt; negem quidquid roges. Heaut. 855, opus sunt. Epid. 445, postulas. M. G. 615, iu'sf M. G. dum ductantj uideas. ' " (b) The dependent verb not in the same tense as the main verb. Merc. 175, quaerebas? Pers. 434, ut faceres quod faciunt. Epid. 115, sumpsi. 40 ATTBACTION OF MOOD IN EAELY LATDT, Group V. In which the main verb is in the subjunctive of '* IDEAL CERTAINTY." (a) A verb in the present tense depending upon a *' less vivid future '* apodosis. Cure. 82, iBtunc qui fert aflBic- Ciet. 497, quodquomque optes turn ueliin; Capt. 237, suadeo; tibi uelim contingere. Cist. 97, si ames . . . quam antes consulas. Asin. 122, mauolet quam non reddat quod promiserit. Pseud. 427, gestant; Poen. 971, iubes; Eud. 96, mactat; True. 349, culpant; Heaut. 642, sciunt; Hec. 794, intellego; Cas. 999, dicitis. Aul. 493, mores meliores sibi parent quam nune ferunt; Aul. 482-4, utimur; . . . metuoni . . . sumus; M. G. 493, facit. Capt. 961,* quod ego fatear credin pudeat quom autumesf Eun. 863,* debeam . . . si id fecerim praesertim quom se ser- uom fateaJtur! (&) Verbs in the present tense (if attracted, in the imperfect) de- pending upon verbs " contrary to fact in the present " ( imperfect subjunctive ) . Adel. 108, sineres ilium facere dum licet. Phorm. 11, minus quam nunc laedit laederet; Baccb. 434, quamst, Phorm. 208, qu^m hoc non po«- sum illud minus possem, Asin. 503, crederes . . . quod fers; Asin. 860, facit; Hec. 657, intellego. 1 1 am not at all sure that we are right in placing subjunctives after cum- causal in the list of " attraction." The data are too meager to furnish trust- worthy results; besides, the question is of no practical importance to classical Latin. Cf. p. 61. The habit of using that mood to express cause with cum may have had its beginning before the time of Terence. Besides, the examples here in question contain a strong suggestion of " remonstrance." «Cf. p. 26. Eud. 1261, dum . . . censeret . . . ipsus esset. Bacch. 488,* si uellem plus uiderem quam deceret, Poen. 681, uidere uos uellem quom aurum darem. qui. ^tM. quam, uL quo, quantum, quom. qui. THE USES OP THE CONSTRUCTION, TABLES. 41 (c) Miscellaneous examples in which the two verbs refer to different Ume-spheres. Heaut. 953, non auderet facere quae fecit; Adel, 314, produxit; Cist. 506, dedi; Most, 202, dedi; True. 349, dedit; Capt. 705, dor bam; Eun.^ 606, simulabar. Group VI. In which the main verb is in the subjunctive op assumption. (a) A verb in the present tense depending upon a ^'less vivid future condition,'' Bacch. 778, ni . . . facta cu- piam quae is uelit, Trin. 472, siquid placeat quod congestum siet, Rud. 1150, si tantillum pec- cassis quod postules. Capt. 906, si memorem ea quae conducunt; M. G. 1429, scio; Rud. 978, memoras; Rud. 1021, quoiust; Trin. 470, uocant; Andr. 165, uolo; Phorm. 171,mt- hist; Pacuuius 407, euentura sunt; Pers. 393, damus. True. 324, lauant Merc. 874, properas; Pers. 206, digna's; Stich. 112, censeo. Asin. 507, praecipis. Lucil. 501, petit, Caecil. 174, qm)m aduenis; Cato. Frag. J. p. 58, cum , . . ivbet. (6) A verb in a past tense depending upon a ''less vivid future condition,'' Capt. 417, si memores quae . . . fecisti; Bacch. 698, dixit; Merc. 419, emptast; Andr. 142, si ilium obiurges qui iulit; Cist 187, uiderat. Capt. 205, si . . . solutos sinat quos . . . emerit, Amph. 871, quod contraxerim si . . . expetat. Cure. 269, si . . . nelint qui periurauerint, Hec. 555, si is posset . . . qv/icum consuesset, nn this sentence the verbs are in the same tense to be sure, but e««em refers to the present, which of course simulahar does not 42 m. ATTBACTION OF MOOD IN EAELY LATIN. (c) Miscellaneous eocamples in which the two verbs are in different tenses, Capt. 754, absque hoc esset qui Bacch. 564, nisi cum ilia qvam hoc fedt; Trin. 967, sunt mortui. mandassem occiperes amare. Pseud. 1236, loqu£re, M. G. 1083, illest. Ghoup VIL In which the main verb is in the subjunctive of result or characteristic. (a) Both verbs are in the present tense, Heaut. 76, ut cures ea quae nil Most. 173, id euenit ut deceat attinent; Most. 396, ut facias quidquid habeas, quod iubeo; Eun. 396, facio; Merc. 1006, arbitri ut sint Andr. 390, sunt certa; Asin. 175, uolt; Cure. 66, nvilist. Adel. 823, duo cum idem fad» unt saepe ut possis dicera Adel. 39, quid sit cafils qtlam ipse est sibi. Pseud. 850, ut appareat quo uenio. qui praetereant. Merc. 840, ubi id eripiatur quod placeat. Bacch. 352, ita feci . . . quan- tum Jubeat reddere ut reddat. Phorm. 153-4, ut qui . . . uelit patrem ut extimescam. Amph. 824, absunt testes qui illud quod dicam adsentiant. Aul. 791, tam parui preti quom pudeat quin purget. Asin. 945, tam ingenio duro quin ubi . . . sit faciat bene. Phorm. 154, patrem ut exti- mescam ubi ueniat. Hec. 608, qui vbiquomque opus sit, possit. Eun. 554, qui me sequatur quo- quo earn. (6) Both verbs in the imperfect tense. Men. 20, ut non posset qu^e Amph. 47, ut exprobraret . . . mammam dabat; Pers. 261, qui daret quoius ingenium nouerat. t quod faceret. Aul. 740, ut id quod non tuom esset tangeres. Bacch. 352, feci ut auri quan^ turn u^llet sumeret. ^i. ut. quom. ubi. quando. quocwnque. THE USES OF THE CONSTRUCTION, TABLES. 43 (c) The dependent verb is not in the same tense as the main verb. Poen. 21, qui dormierunt decet . . . stent; Naeuius 58; opus est. Pers. 825, faciebat. Hec. 150, accepi; Eun. 870, studui. Capt. 467, nee uidi quoi minus procedat quicquid facere occe- perit. Capt. 473, qui . . . quom ede- rint reddant domi. Trin. 621, quoi tuam quom rem credideris sine . . . eura dormias. M. G. 742, quin ubi triduom . . . fuerit, iam odiosus siet. Trin. 699, ut vbi adstrinxeris . . . atque dederis . . . nee sit . . . effugias. Pseud. 725, qui quando prin- cipium prehenderit . . . teneat. Hec. 859, ut uoluptati . . . aduentus tuos quoqttomque ad' ueneris . . . sit. quu Ul, Group VIII. In which the main verb is in the subjunctive of indirect question. (a) Both verbs are in the present tense. Asin. 27, quid sit quod scire expeiis eloquere; Men. 972, sunt; M. G. 1012, quaeris; Pseud. 216, loquor; Pseud. 451, renuntiant; Trin. 257, eget; Naeuius, 60, times. Aul. 800, praedico; Bacch. 400, oportet. Cas. 572, adsitne ei . . . qu^m aduocet, Cato, E. E. 2, possitne quae reliqua sint conficere. Trin. 210, falsone an uero cul- pent quem uelint. Poen. 92, quid id sit hominis quoi Lyco nomen siet. Trin. 306, utrum itane esse mauelit ut . . . censeat an . . . ut , , , uelint. 44 ATTBACTION OF MOOD IN EABLT LATIN. (b) Verbs in a past tense depending upon verbs in the same tense. M. G. 97, quo modo denenerim . . . quoi seruiui; M. G. 345-6, utnim egon id quod uidi uiderim; Bud. 1310, penit. Amph. 1016, exquirere quis fuerit quern propter corpus . . . compleuerit. M. G. 261, qui fuerit . . . qui sit sectatus. Mer. 623, quae facies foret qui illam emisset. (c) The dependent verb is not in the same tense as the main verb, Amph. 106, quantus amator Aul. 29, is scit quae sit quam siet quod complacitumst; Cure. 630, elusit; Trin. 7, ahiit; Hec. 732, iussi; Hec. 873, feci; Andr. 525, dixit, Aul. 65, condidi. compresserit. Heaut. 2, sunt Eun. 522, perii. Ai4 111 coepi obseruari ecqui maiorem filius mihi honorem haberet quam eius habuisset pater. CHAPTEE III. The Uses of the Construction, Limitations. Some of the conclusions to be drawn from these comparative lists are obvious. Mechanical attraction is of relatively rare oc- currence in Early Latin. Even when the dependent verb is in the same tense as the governing clause, very little emphasis is needed to prevent assimilation. In fact, only about thirty-five per cent of the verbs closely attached to a verb in the subjunctive are found m the same mood, and more than half of these belong to the class which seemed in the preceding discussion to be influenced to a great extent by the anticipatory subjunctive. It is impossible, however, to give absolutely accurate ratios, since, as will appear later, it is not often possible to fix in either class the verbs con- nected with conjunctions that may take either the indicative or the subjunctive. Such conjunctions are, for instance, pnusqiMtn dum and si in early Latin. By a closer analysis of the lists of examples given above, we may discover to some extent the conditions under which the habit of attraction began to spread, or, the converse, the conditions under which the new construction did not thrive. The following conclu- sions are based not only upon the examples of mechanical attrac- tion, but on the examples discussed in chap. I, as well ; for it is obviously impossible to state in every case whether the subjunctive is entirely or merely in part due to the force of mechanical attraction. 1. The Time and Tense of the Dependent Clause, The most important element is that of time, as has been implied by the classification itself. To be sure, this is not of equal import- ance in aU constructiorns. So, for instance, attention has been called to the fact that a break in time-sphere does not affect the mood of clauses in dependence upon subjunctives of surprise, remon- strance, or indignation. Of course, conditional and independent 45 46 ATTKACTION OF MOOD IN EABLY LATIN. expressions of wish also present abnormal situations as regards tense. Tabulating the rest with fefei«iiot to the time relationship which exists between the dependent and the governing verb, we find the following ratios of verbs attracted to those not attracted: Subjunctive. Indicative. Verbs in the future depending upon other verbs in the future» (pp. 16-24) 124 71 Other verbs depending upon verbs in the same time-ephere 10 10 Verbs in the present depending upon verbs in the future. 49 162 Verbs in the perfect depending upon verbs in the future. . 16 64 Other verbs depending upon verbs not in the same time- sphere 2 27 2. The Position op the Dependent Clause. The position of the dependent clause is also found to be a fac- tor. A sentence may well open with a subordinate clause, even when the speaker's mind has not yet clearly given the exact shade of meaninfto the main verb Jth which It is finally uttered. Thus in shaping a sentence like the following: Bud. 485, Qui homo sese miserum et mgndicMm uolet, Neptuno credat sese, the mind may well have conceived the relative clause in entirety, before it had any definite feeling in regard to the main verb. In like manner the dependent clause may be forming when the main verb has already been uttered and its modal tone is growing indis- tinct, as may be the case in sentences like the following : Lucil. 440, Ut ego effugiam quod te imprimis cupere apisci intellego. Sometimes the dependent verb is attracted, though it precedes the main verb ; very often when it follows. The favorite position, however, seems to be between the subjunctive verb and its intro- ductory word, when this, — like ut, ne, etc., or any verb, like per- s\mdeo, which calls for a subjunctive clause, — signals the approach «f a subjunctive. A case in point is: Andr. 424, I nunciam intro ne in mora quom opus sit sies. ■This term "future" refers, of course, not only to the future and future- perfect tenses, but also to many of the verbs in the present indicative and to very many verbs of the present subjunctive — especially those of the voUtive aubjunctive. THE USES OP THE CONSTEUCTION, LIMITATIONS. 47 In sentences containing independent subjunctives, and in many conditional periods and in indirect questions, there is usuaUy no possibility of a central position for a dependent clause, but the following table will show the relative importance, under normal conditions, of the three positions mentioned : (When the dependent clause lies before the governing subjunc- tive and precedes all sign of its coming, it may be said to be in the first posUwn. The second position is the central one just iUua- trated. The third position is that in which the clause lies after tne governing verb.) Subjunctive Indicative 5 U td Pontion. 88 87 SdPiuUion. 73 138 It appears that the central position claims fifty-five per cent of the attracted verbs, while only a third of those not attracted are m that place. The dependent clause is occasionally found between the intro- ductory conjunction and ijs verb, in conditional clauses like the lollowmg: Pers. 206, si ut digna's f aciant, odio hercle habeant, and m imprecations and expressions of wish, as in: Aul. 785, ut ilium di immortales omnes deaeque quantumst pei^ duint ; but instances of this kind are not very numerous. The data in regard to the position of verbs depending upon independent voli- tive and optative sentences and upon conditional clauses and clauses of ideal certainty are as follows : Subjunctive Indicative IttPoHtion. 23 td Position, 15 34 SdPoHtioH. 24 78 3. Pkecision in Tense and Modal Feeling. When precision in certain respects is required, the dependent verb is seldom attracted. (a) As regards time, the present subjunctive may stand for a present or a future indicative, and therefore is often of itself somewhat ambiguous. In the same way, the imperfect subjunc- tive may be confusing in that it represents an imperfect indicative or a future thrown into the past. It may be for the sake of avoid- 48 ATTBAOtilMf OF MOOD IN EABLY LATIN. ing this possible ambiguity tBat fti I^W» indicative occurs in a few sentences like the following : Bacch. 989, Volo ut quod iubebo facias. and the present indicative in many sentences like this : M. G. 1054, Age, ... fiat quod te oro. In the same way, the imperfect indicative (cupiebam) in the fol- lowing prevents the interpretation of the verb as a past future of a generalizing clause, which would have been possilte if the verb had become imperfect subjunctive by attraction: Eun. 574, Ut essem una quacum cupiebam. It is at least true that no verb in Early Latin is attracted if it is modified by a temporal adverb which refers to a time differing from that of the governing clause. Cf . : Andr. 339, ubi inueniam Pamphilum ut metum in quo nunc est adimam? (h) Further, the indicative must obviously stand when precision in modal feeling is called for. So often does the tone of the gov- erning clause penetrate to the dependent verb that the mind is continually expecting to find it there. In a great number of cases, though it is not definitely called for, it matters little if there is a shading of it ; but often it is very essential to state a plain fact which is wholly devoid of the volitive, optative or conditional atti- tude of the governing verb. The terms "integral part" or *< essential part " are ambiguous for the reason that they ignore considerations of this nature. A clause may be functionally an integral part of its sentence, and essential to its meaning without having its mood. On the other hand, if these terms are used with the understanding that they convey the modal and temporal feel- ing of the main sentence, such use ignores the fact of mechanical attraction. The following sentence will illustrate what I mean in speaking of precision in mood secured by the indicative : Pers. 2t)3, Di deaeque me omnes perdant — 1 1 eueniant nolo tibi quae optas. Quae optas refers definitely to the prayer just uttered. A sub- junctive here would have been ambiguous, as it might have been equivalent to the above if attracted, or could have been understood as partaking of the optative nature of eueniant, or as conditional THE USES OF THE CONSTEFCTION, LIMITATIONS. 49 In fact both of these points are well illustrated by a number of sentences m which the contrast of time and of modal feeling is brought out by contrasting a subjunctive verb with the indicative of the same verb. See the following: Trin. 351, Quod habes ne habeas et illuc quod non habes habeas. Aul. 482, Et inuidia nos minore utamur quam utimur Et illae malam rem metuant quam metuont magis Et nos minore sumptu simus quam sumus. (c) In fact, as any contrast necessarily calls for an explicit statement of the contrasted parts, wherever such a condition pre- vails there is less likelihood of attraction. So in the follo^ng there 18 a contrast between the subjects tu and ego: Trin. 341, Non eo haec dico quin quae tu uis ego uelim. A contrast of this kind may be brought out by some other word in the sentence, as by pariter in the following: Cap. 329, Nunc hoc animum aduorte ut ea quae sentio panter SCI as. or by the tone of the context, as in the following: Trin. 979, Dum ille ne sis quem ego esse nolo. 4. The Function of the Dependent Clause. A comparison of the determinative with the generalizing clauses will prove that the readiness with which a verb is attracted de- pends to a great extent on the function of its clause as a part of the sentence. The determinative clause is necessarily precise in Its statements. It aims at accuracy in the expression of its modal and temporal feeling, and is apt to possess a peculiar emphasis from this fact In dependence upon a volitive or optative verb it IS therefore likely, by remaining in the indicative, to deny that it partakes m any respect of the tone of the governing verb; and in subjunctive conditional sentences, where the Eoman was so prone to cast the shadow of the ideal even over clauses which expressed facts generally known to be true, the determinative clause is much less frequently attracted than is the generalizing clause. Con- trast, for example, the following, which is determinative: Rud. 978, si istuc ius sit quod memoras, piscatores perierint, with this, which is generalizing: 60 JifUlACTION OF MOOD IN EABLY LATIN. Bacch. 7T8, ni meum Gnatum tam amem atque ei facta cupiam qtuie is uelit . . . tua iam uirgis latera lacerentOT. An examination of the attracted verbs which refer to the present, while in dependence upon volitive, optative, potential, or condi- tional clauses in the present subjunctive, will show that even though they are determining rather than generalizing, and are to be interpreted as referring definitely to the present in connection with the events spoken of in the rest of the sentence in which they are found, they will, with but few exceptions, allow of an ideal interpretation which removes them from the sphere of a deter- mined time or circumstance. They are Hf the type illustrated by the following: Amph. 630, memor sum et diligens ut quae imperes compareant, in which, to be sure, quae imperes refers to the commands just being given, but may in a larger sense mean " whatever commands you give " or " shall give." So in M. G. 1230, (oro) quod cupiam ne grauetur, qu>od cupiam seems to refer to the wishes just ex- pressed, but may here be conceived of as a part of a general prayer, " May not Venus begrudge my wishes ! '* Contrast the more defi- nite statement: Hec. 674, cogis ea quae nolo ut praesente hoc loquar. My point then is that such clauses, while actually serving in their proper context as determinative clauses, are also given a form which suggests a general meaning of universal application, much as the universal presents in conditional sentences may be so expressed as to serve as such, while at the same time they adapt themselves to the feeling of the sentence in which they are found. Cf. Tusc. Disp., Ill, 35 : Diceres aliquid et magno quidem philoso- pho dignum, si ea bona esse sentires, quae essent (i. e,, sunt) homine dignissima. 5. The Temporal Versus the Eelative Clauses. The temporal clauses in Early Latin, when attached to a sub- junctive, are found in the same mood more frequently in propor- tion to the number of occurrences, than the relative clauses under the same condition. There are a great number referring to the the uses of the construction, limitations. 61 future, of the nature discussed under the anticipatory, and we saw that these were as a rule subjunctive, if conditions were not unfavorable. This very fact would naturally give a strong im- petus to attraction in the temporal clauses in particular. Before pving the general table, I would point out the more important habits of some of the individual temporal connectives. Cum.— I have attributed the subjunctive with cum, eleven times to the inherent feeling of surprise, indignation, etc. (see pp. «-11), three times to the necessities of its service as a past-future (see p. 12), forty-three times to its future force after a subjunc- tive (see pp. 16-21), ten times to a more mechanical attraction. After a subjunctive, the cum temporal clause has been found in the future indicative twelve times, in the present indicative seven times, and in the perfect indicative twice. We have found the verb with cum-causal attracted five times, and unattracted seven times; I would, however, call attention to the fact that the causal connection is not functionally a close one, as is seen in the case of the numerous loosely attached qui-causal clauses. Adding to this consideration the interesting fact that the quando<;ausal clause, out of fifteen opportunities, is not once attracted in Plau- tus and Terence, I would suggest that the word attraction must not be employed too frequently as a waste-basket, in attempting to remove troublesome subjunctive cum-clauses from Early Latin. The following verbs after cum I have classed with perfect subjunc- tives rather than future perfect indicatives after comparing their nature and position with those te which no doubt is attached : Men. 543, ut te lubenter uideam quom ad nos ueneris. Trin. 621, Capt 434, True. 234, Capt. 473, Eun. 933, Hec. 694. The following, I think, are indicative: Cas. 130, Heaut. 557, Phorm. 185. Quando.— This conjunction does not seem te be used in Terence as purely temporal.^ The statistics which are found below re- garding this conjunction apply, therefore, mainly to Plautus. As verbs after qmindo-causal are not found to be attracted, I have not even placed the non-attracted indicatives of this class in my tables, but for the sake of completeness give a list of them here : Capt. 12, Men. 834, M. G. 1269, Bacch. 445, Poen. 815, Eud. » Cf. p. Scherer, De Particula Quando, Studemund Studien, II, p. 130. 52 ATTRACTION OF MOOD IN EARLY LATIN. 1182, Adel. 877, Adel. 201, Adel. 287. In the following, which belong to the same class, the conjunction is quandoquidem : Mer. 170, Trin. 351, Trin. 991, Eun. 373, Andr. 487, Hec. 490. As vhi and the remaining temporal conjunctions show no marked pe- culiarities, it will be sufficient to refer to the table given below for the facts regarding them. The following is a synopsis which will give as fairly as can be done the comparative frequency of attraction in the temporal and relative clauses. It summarizes the said clauses which are tabu- lated in the collections of pp. 16-44, omitting the clauses in which the causal or adversative force predominates. Subjunctive Indicative eum, 57 17 IS quando, 9 6 Subjunctive Indicative vbicumqtte qtuim Ut, utcumqut. quotient, primum. 3 2 2 2 Stkaive clauses, 104 269 Total. 4S The proportion of subjunctives in these temporal clauses as compared with the subjunctives in relative clauses is thus seen to be about five to one. A few words of explanation are also needed in regard to the behavior of the conjunctions conunonly used with the anticipatory subjunctive. Dum. — It is not within the scope of this paper to decide when dum takes the subjunctive because of its own force without regard to its position, and I have thought it sufficient to call attention to the fact that the mood of the main clause must be reckoned with, for which purpose I have tabulated the uses of dum in Eafly Liliii with the present and future tenses. IndicatiTe. SubjunctiTe. AJUr After After AJter indicativt. sutjuncHm* indicative. subjunctive. Dum = present 124 li 3 while and as » 1 long as future 11 i 2 9 '1 z=^ until future 18 6 9 9 This table does not take into account clauses in indirect discourse or in dependence upon infinitives. Of chief interest is the behavior of dum = until. After seeing how the indicative predominates after this conjunction, and that it THE USES OF THE CONSTRUCTION, LIMITATIONS. 53 keeps six verbs out of attraction, it is impossible to say that all of the subjunctive examples are due to the regular habit of dum = until to take that mood. Since, however, it was impossible to call any of them undoubted instances of attraction, I have not admitted any of this kind into my tables in Chapter II. For the sake of completeness, I shall give a list here of the nine which are subjunctive and depend upon other subjunctives. They are: Cist. 782, M. G. 1249, True. 482, Andr. 980, Pseud. 1234, Cure. 526, Rud. 328, Rud. 1190, Trin. 757. A rule is sometimes found in the handbooks stating that cer- tain conjunctions which are used with both moods avoid attrac- tion in order to prevent confusion, and that dum is one of these. A glance at the table will show that so far was Early Latin from thus distinguishing between dum = until and its other uses, that there are as many instances (14) of dum = while in the subjunc- tive as of the avoidance of attraction by the same particle. We have seen (p. 13) that all the evidence seems rather to point to a survival of an early anticipatory subjunctive use, even with this meaning, when the verb refers to the future. Priusquam presents some of the same difficulties as dum, because even when it stands free, it takes the subjunctive at times. It is, however, safe to say that a just share of the subjunctives which are in dependence upon other verbs of that mood should be attrib- uted to the growing habit of attraction, since priusquam in Early Latin, when standing free, takes the indicative far more frequently than the subjunctive. Priusquam does not seem to take the sub- junctive in Terence.^ I therefore feel safe in attributing the mood of Eun. 751 to the influence of the main verb. The passages in doubt are: Rud. 456, Men. 846, Mer. 1015, Aul. 154, Epid. 277, True. 523, Pseud. 1031, Bacch. 175, Trin. 886. 6. Adverbial versus Adjectival Clauses. A further reason for the preponderance of the subjunctive in temporal clauses is that these are usually adverbial clauses and as such generally depend more closely upon the governing verb than do relative clauses, which are, as a rule, attached to the sub- ject or object of the verb. Similarly, it is obvious that relative iCf. Lange, p. 36. Adel. 583 is an example of the second singular gen- eralizing subjunctive. 54 ATTBACTION OF MOOD IN EABLT LATIN. clauses in dependence upon the object of the verb are more closely connected with that verb than those which are attached to the subject, and are in consequence more likely to be attracted, as is proved by the table hereto appended. This table includes only such examples as depend directly upon the subject, verb or object. Subjunctive Indicative In Dependenet upon the Subject 26 60 Upon the Verb. 114 70 Upon the Object. 64 122 Fifty-six per cent, of the subjunctires are m clauses which depend directly upon the verb, whereas but twenty-eight per cent, of the indicatives are in such clauses. It is no doubt the failure to make this fundamental distinction which has led grammarians into the erroneous conclusion that the indicative is kept in "circumlocutions equivalent to a substan- tive," a statement which not only fails to give the real reason for the actual phenomena, but even misstates the facts in the case. A very considerable number of clauses of this kind are in fact attracted. See, for example, the following: Men. 994, Caue quisquam quod illic minitetur nostrum flocci fecerit (quod minitetur = minas eius). Eun. 1026, ut . . . faciam quod iubeat (quod iubeat = iussa eius). Add to these Phorm. 125, Trin. 211, Bacdi. 788, Trin. 715, Eun. 790 and many others. It is not true, therefore, flat verbs of this kind are never attracted. However, the majority are not, and the real cause of this fact is that such clauses usually modify the subject or object of the sentence or modifiers of these, and are not, therefore, as closely bound up with the governing verb as the ad- verbial clauses are. In fact, the table just given shows that almost 70 per cent of such clauses after subjunctive verbs remain in the indicative. Some of the adverbial clauses present diflSculties which call for a few words of explanation. Sirclatises. — I have already spoken of the impossibility of mak- ing a satisfactory statement in regard to the conditional clauses, but it can at least be stated with certainty that the ^t-clause feels the influence of the mood of the apodosis. Rotheimer's* collection «Dc cnuntiatis condicionaUbus Plautinis, 1876, pp. 42-7. THE USES OP THE CONSTEUCTION, LIMITATIONS. 55 i of the conditional sentences of Early Latin shows that when there is a question of mixed conditions, the si-clause is quite regularly in the subjunctive if the main clause is an apodosis of the regular type in the subjunctive. There are but thirteen exceptions, of which the following is an illustration : Mer. 351, si dico ut res est . . . quern ad modum existumet me? This fact becomes the more noteworthy when one finds that 129 exceptions occur in the converse type, which means that the prin- cipal clause of the conditional sentence is very often indicative though connected with a subjunctive protasis, as in: M. G. 673, siquid sumas, sumptus est. This contrast certainly points to assimilation of the protasis to the mood of the apodosis. I find it impossible to be more explicit than this, and as it would be useless to attempt to decide in particular cases whether a si- clause is attracted or not, none of this class will be found in the tables. The ut- and quamrclauses of comparison, — The ut-clause of com- parison is sometimes said, in the hand-books, to avoid the mood used with final- and consecuHve-ut, for the sake of clearness. It is true that though this is an adverbial clause, it is not often at- tracted ; but I doubt whether the real cause of the fact is the rea- son given. This clause often precludes assimilation by its very nature, since, as has been pointed out (p. 49), comparisons and contrasts naturally tend to throw emphasis upon the verbs juxta- posed, which fact entails precision in expression of the modal and temporal force. Cf. : Merc. 874, si hue item properes ut istuc properas facias rectius. But this consideration is equally true of quam in comparisons and all other comparative and contrasting expressions. The fol- lowing may represent a large group of this kind: True. 324, si proinde amentur mulieres diu quam lauant. Again, the next paragraph will show that many of these clauses, adverbial though they are, depend upon adjectives at a second re- move from the verb of the governing clause. These two reasons will account for the fact that such clauses are not attracted as frequently as the other adverbial clauses. 56 ATTBACTION OW llOOD IN EARLY LATIN. In Early Latin the ut-clauses are found to be attracted four times and to avoid attraction twenty-five times. With quam in comparisons the ratio is 6 : 17. 7. The Peoximity of the Dependent w» the Governing Clause. A minuter classification must now be made in order to deter- mine how closely the dependent clause must be attached to the main body of the governing clause, in order to be attracted. It has appeared in the above that not only may those verbs be al^ tracted which are found in adverbial clauses and in direct depend- ence upon the subjunctive verb, but also those which are attached to the subject or object of the same. In Early Latin this state- ment is not to be understood as applying to clauses after ante- cedents complete in themselves, which remain in the indicative (cf. Cato 84, uideto ut bene percocas medium, ubi altissimum est), nor to any of the loosely attached coordinate relative clauses, the " forward moving clauses," or parenthetical asides. It refers to the closely attached clause which is necessary to complete the meaning of the antecedent expressed or understood. 1. In Early Latin the relative clause, when attached to the sub- ject or attribute complemerd of the sentence, is found to be at- tracted under the following conditions: (a) When the antecedent of the relative is not expressed, as in Amph. 630, diligens ut quae imperes compareant (20 examples). (6) When the antecedent is expressed and is a noun or sub- stantive adjective, as in Most. 413, uiri doctist opus quae dissig- nata sint . . . tranquille cuncta ut proueniant (2 examples). (c) When the antecedent is expressed and is a demonstrative pronoun,— is, ille, etc., as in Cure. 29, ne id quod ames . . . tibi sit probro (4 examples). ,• / ir 2. Similarly the relative clause when attached to the object ot the verb is found to be attracted: (a) When the antecedent is not expressed, as in Epid. 6, di dent quae uelis (41 examples). (6) When the antecedent is expressed and is a noun, as m Cur. 550, tuom qui signum ad me attulisset nuntium ne spemerem (5 examples). THE USES OF THE CONSTRUCTION, LIMITATIONS. 67 I (c) When the antecedent is expressed and is a demonstrative pronoun. Cf. Lucil. 572, concedat id quod uelit (18 examples). 3. It may also be attached to the indirect object of the verb. Cf. Pseud. 570, det locum illi qui queat (5 examples). 4. Or to the object of a complementary infinitive depending upon the subjunctive verb, or forming its subject. Cf. Aul. 600, quod iubeat properet persequi; and Mer. 152, ut quae scirem scire actutum tibi liceret (5 examples). 5. Or to a substantive in an adverbial clause which modifies the predicate, as in Hec. 491, exopto ut relicuam uitam exigat cum eo uiro me qui sit fortunatior. So also Cist. 85. 6. As regards the adverbial clauses, the larger part of these (a) (114) depend directly upon the governing verb, usually by means of a temporal conjunction, as in Pseud. 307, quando nil sit, simul amare desinat We have already shown (p. 50) that this is the chief reason why so large a proportion of temporal clauses suffer attraction. Furthermore, many adverbial clauses are attracted which are not directly attached to the verb, but (b) depend upon some modifier of the subject, as Pers. 237, Numquam . . . exterebrabis tu ut sis peior quam ego siem. Cf. True 96. (c) Or of the object. Cf. Hec. 729, uidendumst ne minus propter iram hinc impetrem quam pos- siem. Cf. Pers. 340. (d) Or which depend upon a second adverbial modifier of the predicate, as in Rud. 1243, ut cum maiore dote abeat quam aduenerit Cf. Bacch. 488; Hec. 555; Aul. 17. It is worthy of note that the last three classes contain for the most part adverbial clauses of comparison with quam, and that many of these clauses are of necessity far removed grammatically from the governing verb, and therefore remain in the indicative. On the other hand, it is readily seen that in a sentence like the last one cited (Rud. 1243) the dependent verb aduenerit, even though a second remove from abeat grammatically, is, in simi- larity of function and in position, very near to it; considerations which probably weigh as heavily as that of grammatical prox- imity in determining the mood of the verb. 58 ATTRACTION OF MOOD IN EARLY LATIN. SUMMAEY AND CONCLUSION. The first part of this paper makes an attempt (1) to explain the origin and (2) to describe the growth of the construction of modal attraction. (1) It endeavors to show that from several constructions (enumerated on p. 7) in which the dependent clause contained the modal force of the governing clause, the tendency arose which was the source of the construction of modal attraction. (2) It attempts to describe how this beginning developed by the law of analogy into a habit of placing dependent clauses in the subjunctive when attached to clauses of that mood. The second part of the paper, after giving a complete list of the clauses in dependence upon subjunctives in early Latin, at- tempts to define the limits of the field into which the construction in question spread in that period, and to point out the obstacles which 80 limited it. This discussion shows with more definiteness than can be reproduced in a general statement, (1) that the at- tracted clause is preferably in the same time-sphere as the clause on which it depends; (2) that its favorite position is between the introductory conjunction (when such exists) and the verb of the governing clause; (3) that its verb rarely expresses precise modal and temporal force ; (4) that the clause as a whole is rather of the generalizing than of the determinative type; (5) that it is more frequently a temporal than a relative clause; (6) that it is connected with the predicate more frequently than with the sub- ject or object of the sentence, and (7) that, as a rule, it is an essential clause, and grammatically depends very closely upon the main body of the clause to which it is attached. The above, I think, are practically all and the only statements of importance that can be made regarding the limitations of its uses in Early Latin. It is sometimes said^ that under given con- ditions a clause is inevitably attracted. On the contrary, I be- lieve that the comparative tables given above and the discussion of the same have made it evident that attraction is never abso- iSo for example Riemann ct Goelzer (Gram. Comp., p. 724) : Cos oi^ le BuhjoncHf est ohligatoire.—he subjonctif est nScesmire lorsque la proposi- tion otL il doit se trouver exprime une id^ qui complete et achftve Texprea- Bion de la pens^ eontenue dans la proposition infinitive ou subjonctive ft lAqueUe elle se rattache. The sentence that he gives to illustrate this state- ment is an example, not of assimilation, but of Indirect Discourse. THE USES OF THE CONSTBUOTION, LIMITATIONS. 59 lutely necessary. Practically every example of attraction was balanced by one or several non-attracted clauses which, so far as closeness of attachment is concerned, were of the same nature as the attracted clauses which they resembled. These results may seem somewhat unsatisfactory in that no single short statement has been found adequate in defining the construction. However, a short definition may serve the peda- gogical purposes of those who correctly understand the usages of the construction, and know the actual facts about its limitations as I have tried to present them in the preceding. A rule may be formulated somewhat as follows: When a clause depending upon a subjunctive clause forms an essential part of the thought of the governing clause, it may be put in the subjunctive. In this rule, however, the word " essential " must be interpreted in the light of all of the preceding discussion; and it must be re- membered that the probability of the assimilation of a clause varies with its success or non-success in complying with the quali- fications herein enumerated. These are quite tangible, and, as we have seen, lend themselves readily to definite, statistical treat- ment. To one who imderstands these the study may bring satis- factory results. Finally I would protest against the common custom of treating this construction as a kind of Indirect Discourse. This error is probably due to the fact that the two constructions behave in much the same way, and are to a certain extent affected by the same limitations. It is also true that one often finds it difficult, not to say impossible, to decide whether a verb is influenced by one or the other, or by both. We have seen, however, that there is very little in common between them. Their origins are widely separated, and any grammar that purports to be historical should treat them separately. Again, the construction of ai?simiUtioi' should be carefully dis- tinguished from that which occiii^s after an infinitive. The two have in many particulais: the ^ame; beg:innirfg8^ Ih^jy have prac- tically the same habits, but th6 latter is more closely allied to that of Indirect Discourse than. the. fgnner, and i^.of.rg^rer occurrence. m 'III i*ii m p » t • t I • • < « » • « t • . * u p m i ^1 * Bj^f^- 00 WOT COLUMgi^ APR«