SOURCE: An introduction to A Greek Critic: Demetrius On Style, University of Toronto Press, 1961, pp. 3-56.

[In the following essay, Grube offers background on the philosophical and rhetorical traditions of Greek literary criticism, examines the content, nature, and structure of On Style, and considers the problems of determining the authorship and date of composition of the work.]

The Background

Greek criticism of literature was derived from two distinct and independent sources, the philosophical and the rhetorical. The philosophers were first in the field. As early as 500 B.C. we find Xenophanes and Heraclitus vigorously censuring Homer for his immoral and untrue stories about the gods.¹ Thus started what Plato was to call the ancient quarrel between poetry and philosophy, in which the philosophers stressed the social responsibility of the poet, and the importance they attached to this reflects the vital place of poetry in the life of Classical Greece. Formal education consisted, as is well known, mainly of physical training, music, and poetry, especially Homer. The Olympian gods cared little for the conduct of their worshippers; except for a very few traditional requirements such as the sanctity of an oath, respect for parents, and the laws of hospitality, they insisted only on the performance of due ritual. There was no preaching in the temples, and men turned to the poets for guidance in the art of living. Hence the deep-rooted feeling that the poets were the teachers of men; we find this point of view first clearly formulated in Aristophanes, but it is, as a feeling if not a theory, very much older. This moral responsibility might well have surprised Homer, but Hesiod would have accepted it, so would most of the lyric poets, and by the time of the great tragedians it was well established. In any case, where poetry is a vital force in society, it cannot live in an ivory tower. Art for art's sake is a theory which does not arise until poetry has retired to the study and music to the studio. It was therefore very natural that criticism of literature—and this up to the third quarter of the fifth century meant criticism of poetry exclusively—should, particularly in Greece, have begun as moral criticism, as criticism of content rather than of form, and this philosophical approach was rarely absent in the better critics of antiquity.

From the middle of the fifth century, however, a quite different approach to literature was being developed by the teachers of rhetoric. First in Sicily, then in Athens, with the growth of democracy, the art of swaying assemblies and juries was the road to political power, and this was clearly recognized by the ambitious. Teachers of rhetoric were in great demand, and the Sophistic movement arose to fill this demand, for, in spite of their individual differences, the Sophists all had an interest in language in common.

When Gorgias of Leontini came to Athens in 427 B.C. and brought with him from Sicily all the tricks of his rhetorical trade, he is said to have taken Athens by storm. Certainly, the Athenians were predisposed to appreciate the new art of speech: their education had endowed them with a sensitive appreciation of poetry; they eagerly discussed the works of the great dramatists; they had applauded Pericles and other orators in the assembly. Indeed it may well be argued that they were already thoroughly familiar, in practice, with Gorgias' antitheses, homoioteleuta, balanced clauses, and so forth, and it is probable that the direct influence of Gorgias on the style of the great writers of the time, Euripides and Thucydides for example, has been exaggerated by both ancient and modern critics. Nevertheless, as the first theorists of the art of language, the Sophists did have a very great influence on the development of Greek style.

When Gorgias praised the power of Logos, the spoken word, he was claiming for prose a place by the side of poetry as a sister art and he clearly wakened the Athenians to a new awareness of the art of speech. Indeed, when Aristotle speaks of the old simple style of writing,² he means writers before the time of Gorgias, including Herodotus. As a theoretical innovator, however, Gorgias went too far, and it was easy for Aristotle and later critics to ridicule his poetic diction, his farfetched metaphors, his too neatly-balanced clauses, his word-jingles, rhymes, and the rest. Indeed his fiagments fully deserve their censures.

Gorgias and the other teachers of rhetoric were interested only in the art of persuasion, in rhetorical tricks to arouse the emotions of an audience, and Plato was no doubt right when he maintained that they felt no moral responsibility whatever. "The word" said Gorgias "is a mighty power; … it can end fear, remove pain, bring joy, and increase pity." He went on to extol the power of words to play upon human emotions, and to show how a speech (logos) can delight and persuade a great crowd "not because it is spoken with truth but because it is skilfully composed." Gorgias may be regarded as the first conscious technician of the art of speech in continental Greece; he brought with him an already well-developed technical vocabulary. From him ultimately derives that tendency to analyze figures of speech and thought which, in the rhetorical works of later criticism, often looks upon literature, from Homer down, as a mere treasure-house of rhetorical devices.

This emphasis on the means of persuasion and on rousing the emotions, fear and pity in particular, without regard for morality or truth, naturally went hand in hand with scepticism, with a questioning of all accepted values. The new teachers undoubtedly helped to undermine the traditional education of Athens and the traditional social morality. Because of this they aroused the anger and resentment of conservative Athenians, whose spokesman was Aristophanes, the great comic poet (ca. 450-385 B.C.).

The comedies of Aristophanes bear convincing witness to the important place which poetry—tragedy in particular but not tragedy only—held in Athenian life. They are full of literary allusions, parodies, and criticisms which he expected his audience, the people of Athens, to appreciate and enjoy. His hatred of the new education and the new scepticism is especially clear in the Clouds (423 B.C.), where Socrates is his chief butt. For twenty years he attacked Euripides as the exponent of the modern spirit. But Aristophanes' greatest contribution to literary criticism is the famous contest in the Frogs between Euripides and Aeschylus for the Chair of Tragedy in Hades. The comedy was produced in 405 B.C., soon after Euripides' death. Aeschylus had been in his grave for fifty years, and he stands here as the defender of tradition. Euripides is attacked for his immoral subjects and evil influence, but there is also much criticism that is purely aesthetic, where the younger poet is censured for his realism, his innovations in metre and music, his use of a narrative prologue, his excessive use of lyric monodies, his prosaic everyday language. The whole debate puts before us two different views of drama, and of literature generally, which are perennial and irreconcilable. The differences can in part be explained historically: the tempo of Aeschylean tragedy was already archaic in 405 B.C. and so was a good deal of its language; the more "sophistic" techniques of Euripides are also partly due to his date. Essentially, however, the conflict goes very much deeper, for it is the conflict between the romantic and the realist, the former believing that many true things are better ignored, the latter that truth, the whole truth, will make men free. The grand manner of Aeschylus requires the grand style and impressive language; the realism of Euripides inevitably requires a simpler diction. Aristophanes' sympathies were all with Aeschylus, but the criticisms of him which he puts in the mouth of Euripides also neatly hit the mark, and in the end he refuses to judge between them as dramatists or as poets. The contest is one of the most vivid pieces of literary criticism in ancient literature, as well as the most amusing. For all his dislike of the new techniques, Aristophanes was clearly thoroughly familiar with them.

So was Plato (who was growing to manhood at this time) in spite of all the bitter things he said against the Sophists and rhetoricians in his dialogues. In practice he was a most careful stylist, and one likes to remember the story told by Dionysius of Halicarnassus that, when Plato died, tablets found among his belongings showed how he had tried many different word-orders for that simple, easy-flowing sentence with which he begins his Republic.³ His style takes its place in the history of fourth-century Greek prose as the superb culmination of that process of development which Gorgias and the rhetoricians had started about the time of Plato's birth. But Plato was a philosopher, the disciple of Socrates. He knew, better than Aristophanes, that Socrates had laboured for a much deeper, more philosophic education than that of the Sophists, an education which aimed at philosophic inquiry into the nature of reality by way of painful self-knowledge and self-criticism. When, a teacher himself and the spiritual heir of Socrates, Plato opened his Academy in the eighties of the fourth century, it was natural that he should examine critically the claims of those other teachers old and new, the poets and the rhetoricians. It is thus that he approaches rhetoric in the Gorgias, and poetry in the Republic. What are their claims to knowledge and what is it they can claim to teach? The world of literature has never forgiven Plato for banishing the poets, or at any rate most of them, from his ideal republic, but that banishment is essentially a challenge to the poets to recognize their social responsibility, a challenge which has never been completely answered. Every civilized state except ancient Athens has adopted Plato's theory of censorship—the subordination of the artist to the legislator, in some form or other. Nor have they waited, before doing so, for the establishment of the ideal republic or the rule of the philosopher-king! Plato was deeply convinced that poetry, music, and the arts had a tremendous influence upon the formation of character, and he was terribly afraid of an uncritical emotional response to that influence, especially in drama where impersonation makes the response more immediate and more complete, both for the actor and for the audience. Hence the vigour of his attack, and his forbidding the impersonation of any evil at all upon the stage.

His theory of art as imitation of life does not mean, of course, that the best painting is a coloured photograph or the best drama a mere record of actual conversations, though it must be confessed that his ironically emphatic language almost seems to say so at times. It does mean, however, that the artist, and especially the dramatist, must draw his material from life and be true to life. This is, to him, an accepted truism⁴ rather than an original theory, and, after it had been more calmly stated and more fully worked out by Aristotle, it was never challenged in antiquity. Moreover, when he insisted that the poet could not directly imitate or represent the eternal verities that are the Platonic Forms, but could do so only indirectly as they are mirrored in actual life, was he not right at least to the extent that drama must be represented through individuals, and that a drama of pure ideas is not drama, or indeed poetry, at all?

The Phaedrus is a corrective of the too intellectual and social approach of the Republic with its apparent attack upon passion and emotion. The myth of the Phaedrus is a superb vindication of passion and inspiration, but the inspiration must come from the gods. This, translated into philosophical terms, means that the passion of the poet must be directed towards beauty and truth, and directed by reason.

The second part of the Phaedrus is written in a calmer mood. It sets out to discover how to write well, whether in prose or verse, and it contains a statement of basic critical principles. Plato states clearly, for the first time, the difference between criticism of form and of content.⁵ He insists that the writer must know his subject and adds ironically that he will find this useful even if his aim is only to deceive. He must define his subject. Every logos, every work of prose or poetry, should have a definite structure, with a place and function for each part, like a living organism, with a beginning, middle, and end, and with every part in its proper place, in its proper relation to the whole. The technique of writing or speaking are a preliminary requirement, but technique is not art. Plato dwells on this last point at some length: the man who knows the notes of the musical scale but cannot relate them to each other is no musician; the man who knows the effects of drugs but not when to use them is no doctor; the man who can make speeches, long or short, to arouse pity or fear, but knows not when to make them, is no tragic poet. Sophocles and Euripides would laugh at his pretensions "knowing well that the art of tragedy is no other than the interrelating of these elements in a manner fitting to each other and to the whole work." And Plato goes on to pour ridicule upon the rhetoricians' boast that they can arouse and calm emotions at will, as also upon their ever more complex technical vocabulary and their neglect of the fundamentals of their own craft.⁶

Plato discussed poetry and music once more, in the second book of the Laws, the work of his old age. Mousikê, which includes both, is a gift of the gods with two functions: the training of the emotions in youth and the recreation of emotional stability at all ages. Poetry and music have their roots in primary human needs and instincts, in the natural need for motion and utterance. As the random movements of the infant, gradually brought to orderly control by the human sense of harmony and rhythm, culminate in the dance, so the same sense of rhythm and meaning brings under control the infant's random cries until this process culminates in reasoned speech and ultimately in poetry. We all speak and move; we are all to some extent poets and musicians.

There are, in the Laws, three criteria by which art must be judged: one of these is still the moral criterion; the second is pleasure, even though Plato insists that it must be the pleasure of the educated and that art must not be judged, as we might put it, by box-office receipts. The third criterion is artistic or aesthetic, even though the formulation of it is rather rudimentary, that is, the "correctness" of the imitation. Both pleasure and artistic perfection are thus recognized as criteria. We are further told that the poet need not be the judge of the moral values of his work but he must, in that case, accept the judgment of the legislator who in turn must, in order to give an adequate judgment, understand the aim of the artist.⁷

The Poetics of Aristotle continues the philosophic approach to literature and at many points tries to answer Plato. It is important to realize that Aristotle accepts, in the main, the moral approach of Plato and his philosophic predecessors. This is quite clear from the Politics, the only place where he discusses the function of mousikê in society. Even in the Poetics tragedy is "the imitation of a morally good action."⁸ He accepts the principle of censorship, and the place of poetry and music in both education and recreation. He accepts, too, Plato's theory of art as "imitation," mimêsis, though he adds that the poet may imitate or represent things as they are, as they were, as they were thought to be, or as they ought to be, thus making it clear that the Greek mimêsis does not mean copying. He modifies the moral criterion to the extent that any evil act or speech in a play must be judged not in itself but in relation to the effect of the play as a whole, and in relation to the character concerned. His theory of catharsis: "tragedy … by means of pity and fear achieves the purgation (catharsis) of such emotions" is now generally accepted as a medical metaphor. The effect, as he explains in the Politics, is the same as that of orgiastic music which through exciting emotions to a crisis has an ultimately calming effect. And he answers Plato by suggesting that "the more vulgar parts of an audience, mechanics and general labourers … whose souls are perverted from their natural state" and in a state of over-excitement, need this catharsis to recreate emotional stability and that this kind of dramatic performance should therefore be allowed. The passage makes it abundantly clear that the cathartic effect of drama is mainly restricted to these weaker types. The Aristotelian philosopher, who has perfect emotional control, presumably remains unaffected. Even Aristotle's theory of the tragic hero as neither villain nor saint, but a man with some flaw in his character and therefore more "like ourselves," recognizes the need at least for weaker men, if not for himself, of that emotional identification which Plato so greatly feared. Moreover, where he insists upon the unity of plot (the only kind of unity he does insist on) and explains how one incident should follow the other as inevitable or at least probable, with the end emerging from the plot itself, he is elaborating upon the Platonic conception of the unity of a work of art as an organism. He adds important suggestions of his own: the preference for an unhappy ending, the importance of peripeteia or the sudden change of direction toward misfortune, and of recognition and discovery. The Poetics is a triumph of unemotional analysis; it is extraordinarily suggestive in detail; some of its limitations are due, however, precisely to the fact that the author's emotions are nowhere engaged.

What we have called the philosophic, as against the rhetorical, approach to literature thus began with moral criticism and continued to insist upon the writer's responsibility to society, but in Plato and Aristotle it also developed a considerable body of literary theory. Of this approach in its purer form the Poetics is our last extant example.

During the century between the famous visit of Gorgias to Athens and the publication of Aristotle's Rhetoric, textbooks on rhetoric multiplied, and it is these which both Plato and Aristotle regard with contempt. Most of them, undoubtedly, were purely technical and amoral. Of this considerable literature very little remains. We know that Thrasymachus of Chalcedon, a contemporary of Gorgias, was an adept at arousing and again calming the passions of his audiences, that he paid some attention to rhythm, wrote in short rhythmical clauses, and affected the use of paeonic feet. He seems to have been able to arrange his material clearly, to express his thoughts with succinct compactness, and to have developed a kind of prose diction that was neither too poetical, like that of Gorgias, nor lacking in distinction.⁹ We have a short essay of another rhetorician, Alcidamas, on the necessity for an orator to speak extempore.

But if rhetoric finally established itself as the higher education from the fourth century B.C. onward—and to this place of rhetoric in education the rhetorical nature of our later critical texts is largely due—the credit or discredit for this should be given mainly to Isocrates (436-338 B.C.). Several years older than Plato, he died at the time of the battle of Chaeronea which established Philip's supremacy over Greece. He was the teacher of most of the great Athenians of his day. He called himself a philosopher, though neither Plato nor Aristotle would have conceded him the title. He had little respect for "useless" knowledge; he was the apostle of general education, which for him consisted in being able to speak well on great subjects—and this also meant to be able to write. He rejected, however, the amorality of the rhetorical technicians; he insisted that you cannot speak well on noble subjects without practical knowledge of them, and, furthermore, since an orator (or writer) must make a good impression on his audience, he will desire to be a good man: "To speak well is the greatest sign of intelligence; a truthful, lawful, just speech is the outward image of a good and loyal soul."¹⁰ The later theory of Cicero and Quintilian, that the good orator is a good man vir bonus dicendi peritus, as old Cato put it¹¹—ultimately derives from Isocrates' theories of education, superficial as these obviously were.

It may be added that Isocrates, though included in the later canon of the Ten Attic Orators, was prevented from public speaking by a physical handicap; he published his speeches as pamphlets, and logos obviously means, to him, both the written and the spoken word. Of poetry he says very little. He was a pupil of Gorgias and a very careful stylist: antitheses and balanced clauses follow one another in carefully constructed periods, and he avoids any hiatus like the plague. His patriotism was sincere, but it never caused him to write an inelegant sentence, and the total effect—we have a considerable number of his "speeches"—is one of deadly monotony. His contemporary influence however was very great (greater than that of the Academy or the Lyceum); his posthumous influence was no less—we can trace it clearly in the texts we possess, in Cicero, Dionysius, and Quintilian.

Although Isocrates was a teacher of the art of speech, he was hardly a rhetorician in the strict sense; he did try to communicate to his pupils a general philosophic outlook. For an example of the more strictly rhetorical approach at this time we have to go to a treatise preserved for us among the works of Aristotle, the Rhetorica ad Alexandrum. The dedication, a letter from Aristotle to Alexander the Great, is an obvious forgery. The work itself, however, is definitely dated in the fourth century B.C. and often thought to be the work of a contemporary of Aristotle, the rhetorician and historian Anaximenes. It is, at any rate, our sole remaining example of the more sophistical treatises of the period. It displays the completely cynical, amoral attitude which was so repugnant to Plato; it is concerned exclusively with the question: what kind of arguments will, in particular kinds of cases, be convincing? These are listed, named, and described at considerable length and with great precision. Even if this particular work was not written before Aristotle's Rhetoric, there can be no doubt, that many works of the type were in existence, and it is against this kind of background that the Rhetoric of Aristotle must be judged.

Unlike the Poetics, the Rhetoric had an immediate and lasting influence in antiquity. It is very different kind of book, for here Aristotle meets the rhetoricians on their own ground. He writes the kind of book which should be written about their craft. Indeed, it might well be said that he establishes the art, and even Isocrates' theory of education, on a much more solid philosophic foundation. Aristotle himself, in the Poetics, refers us to the Rhetoric for all that concerns the expression of thought in words, be it in poetry or prose. For we should never forget that rhêtorikê was the art of expression as a whole, even if oratory was the art of expression par excellence. And in part Aristotle is still, as so often, answering Plato: he sets forth in the first two books the kind of knowledge of politics and psychology which an fator should have, and which can suffice, the kinds of arguments and proofs based on probability, which he should employ. The first twelve chapters of the third book then deal with style, and concern us closely; we shall see that the author of our treatise On Style is thoroughly familiar with them, and much indebted to them. Many of the later critical and rhetorical formulae of the schools appear here for the first time, at least for us.

Among these are the statement that the diction of poetry is necessarily different from that of prose; the division of rhetoric into three kinds: the forensic, the deliberative, and the epideictic; the division of style into diction (or the choice of words) and the arrangement of the words thus chosen; the further division of word-arrangement into the running or strung-along style of the "ancient" writers and the periodic structure. Here too are the divisions of a speech into proem or introduction, narrative, proofs, and epilogue. Aristotle will allow only these four; he would prefer two only, statement and proof, and he mentions with Platonic contempt the over-subtle subdivisions affected by contemporary rhetoricians. Prose must have rhythm but not metre, a statement repeated for centuries by Greek and Roman critics.

Aristotle recognizes only one "virtue" of style, lucidity. This is most easily attained by the use of current, everyday language, but men like the strange and the new, so we must introduce a certain number of unusual words and a degree of ornamentation, in order to strike a happy mean, always making our style suitable and fitting… to the occasion, the audience, and the speaker. Aristotle deals at length with metaphor, the chief ornament he allows in prose; it is also the one thing that cannot be learned from others for it involves a capacity to see similarities in things. Like every Greek, Aristotle is well aware of the importance of semantics, of whether, as he puts it, you call Orestes his mother's murderer or the avenger of his father. The discussion of style ends with an attempt to analyze the reasons for successful sayings: their success is related to man's delight in learning something new—a delight that can be derived from a good metaphor, a particularly apt word, a clever antithesis or argument. All these gain from brevity, for example, Pericles' famous remark after many young Athenians had been lost in battle that "the year had lost its spring." Vividness, riddles or half-riddles, similes, proverbs, hyperboles, all these can contribute to success. Moreover, there is not one perfect style suited to all occasions; each kind of rhetoric has its own appropriate style. The style of written work is not that of debate; it is more precise, less histrionic. Nor will Aristotle accept brevity as necessarily a good thing: one should be neither too concise nor too verbose, but seek the right mean.

The rest of the book is more definitely concerned with rhetoric in the more restricted sense: it deals in turn with the aim and purpose of each of the four parts of a discourse and how the speaker should deal with each part (13-17). This section is also much drier and more technical in style. If Aristotle shared Plato's contempt for the technicalities of the Thetoricians, he had a good deal more patience in dealing with them.¹³

From the time of Aristotle to the first century B.C. we have no extent critical texts, unless indeed it be our treatise On Style. We know that Theophrastus, the disciple and successor of Aristotle, wrote a book on style.… The references to this work in later writers, who usually mention him along with Aristotle, are scrappy and tantalizing. Scholars have attempted to reconstruct his critical theories, but the evidence is insufficient. On the whole it would seem that he did not depart very far from his master's theories which he expanded and explained.¹⁴

During the third and second centuries B.C., Alexandria developed a school of literary scholarship rather than of rhetoric or literary criticism. The scholars of the Museum edited, with commentary, all the great classical writers. The catalogues of the great library formed the basis of the first histories of literature, while the commentaries did evolve some critical principles, notably those of Aristarchus (ca. 217-143 B.C.). He stated that Homer must be interpreted in the light of the social customs of his day and not those of a later age; that any statement must be judged by reference to the character who makes it or, as he put it, "all that is said in Homer is not said by Homer"; that poets must be given some licence in dealing with historical facts: they need not tell us every detail of what happens but may leave certain things to the reader's imagination. Aristarchus championed the unity of Homer and he may have defended the artistic unity of the Iliad and the Odyssey.¹⁵

We know that the so-called Asiatic style developed rapidly as the Greek language spread over Asia Minor and that Hegesias of Magnesia (fl. 250 B.C.) was criticized by Dionysius and Cicero as the great exponent of that florid, over-rhythmical, artificial manner. In the next century Hermagoras revived rhetorical studies and seems to have written a great work in which he classified all the possible kinds of court cases, the different kinds of issues and how each should be dealt with. It must have been during this period too that the various formulae which we find in the Roman rhetorical writers of the first century B.C. were developed: that of the three main styles, for example, or that of a specific number of rhetorical virtues, but we cannot trace the origin of either for lack of evidence. Nor did the philosophers, as far as we can make out, make any further substantial contribution. While the Peripatetics probably continued in the tradition of the Rhetoric to analyze style and to use both prose and poetry to illustrate the points they made, the Stoics seem to have concentrated on allegorizing Homer so as to make a Stoic of him, and on pure linguistics. Their main contribution seems to have been to add "brevity" to whatever list of rhetorical "virtues" then came into fashion, and though brevity is often to be commended it is not in itself, as Aristotle knew, necessarily a virtue.¹⁶ But the Stoics seem to have despised all conscious stylistic effort, and the Epicureans, by and large, seem to have taken little interest in literature of any kind.

When we come to the first century B.C., however, we again have a large number of texts, partly Latin and partly Greek. As we are here primarily concerned with a Greek-treatise, it will not be necessary to follow, in any detail, the development of Roman rhetoric or criticism, for the Greek rhetoricians hardly ever mention a Latin writer or a Latin theorist. This is not primarily due to arrogance, or to discretion, but to the simple fact that they had no need to mention them. They were concerned with Greek literature and with Greek style; they took their illustrations from the Greek authors of the classical age and, to a lesser extent, from those of Alexandria. Nor could they be required to mention Roman critics or rhetoricians, for the Romans took their theories from the Greeks. There were, to be sure, specifically Latin or Roman problems such as, for example, the quarrel between the champions of the ancient Latin writers and those "moderns" who took the Greek classics as their models, but these were of no concern to a Greek writer.

Suffice it to say, therefore, that in the Roman tradition there is nothing to correspond to what we have called the purely philosophical approach. The ancient quarrel between poetry and philosophy was long forgotten, and in any case Rome produced no original philosophers. The approach was therefore rhetorical, but of two kinds: the strictly rhetorical on the one hand, and on the other what may well be called the Isocratean, that more general approach which, though still rooted in rhetorical training, nevertheless stood for a more cultivated outlook and a general interest in human affairs. Our first Latin text, the Rhetorica ad Herennium, is of the first, more professionally rhetorical kind. It is preserved among the works of Cicero; it was most probably not written by him, but was in any case written at the time when he was a very young man, probably between 86 and 82 B.C.¹⁷

The Ad Herennium is of considerable interest as the first extant attempt to Latinize the Greek rhetorical vocabulary, and, although the author is impatient of the over-elaborate subtleties of the Greek rhetoricians, all the formulae are carefully set down: the three kinds of rhetoric which we saw in Aristotle; the main formula, also used by Cicero, of the capacities the orator must possess: he must be able to think of what he should say (inventio…), to order his material (dispositio…), he must have style (elocutio…), memory, and a good delivery (actio…). Each part of a speech, of which the author recognizes six,¹⁸ is discussed; we then proceed to inventio in relation to the types of argument to be used in each type of case, and so with the other parts of the main formula. The fourth and last book discusses the three styles, the plain, the grand, and the intermediate, all of which the good orator should be able to use at the right moment, and various kinds of qualities, ornaments, and figures (sixty-four of these altogether).

The more technical of Cicero's rhetorical works are of much the same type—the De Inventione, the Topica, the Partitiones Oratoriae. He shows the same impatience with Greek subtleties and the same inability to shake them off, a greater virtuosity in the translation of technical terms, and an occasional purple patch, usually the introduction. His more general works, however, take a broader, more Isocratean view; in them the technical formulae take second place. The Brutus is largely a history of Roman oratory introduced by a brief sketch of Greek oratory. It also, however, contains many passages of more general interest. Cicero wrote very quickly and he is extremely careless in the use of technical terms, but his main concern, as in the three books of the De Oratore and in the Orator, is to rescue rhetoric from the study of mere techniques, to insist that oratory is an art which must be solidly rooted in a general education in philosophy, history, and jurisprudence, an education in the liberal arts which will ensure a moral education as well. He goes into a good deal of detail but he never forgets his main purpose: the general education of the orator. In spite of his own predilection for the grand manner, he recognizes that the orator must instruct (docere) and entertain (delectare) as well as rouse the emotions (movere) and he must therefore be able to use each of the three styles at the right time (apte). His lists of "virtues" of style vary from one work to the other but purity of language (Latinitas), lucidity (dilucide), appropriateness (decorum) and stylistic ornament (ornatum) are the most frequent.

When we turn from Cicero to Dionysius of Halicarnassus, who settled in Rome about twelve years after Cicero's death, we find ourselves in a world still dominated by the rhetorical education but definitely more literary than oratorical. Dionysius never mentions Cicero, nor Horace whose contemporary he was; yet, as we know from his history and from the introduction to his work on the orators, he was extremely well-disposed to the Romans. It was to the cultured Romans of his day that he gave the credit for stamping out, as he thought, the plague of Asianism in style. We have a considerable bulk of his critical writings. One of them …, On Composition or Word-Arrangement, is a masterpiece. It deals with the collocation of words from the point of view of sound, the music of language which results from the sound of letters in combination, from rhythm and pitch and stress. The Greeks were extremely sensitive to this music of language even in prose; Dionysius pursues its aesthetic appeal to its very elements and discusses the sound of each letter and its contribution to the total effect, incidentally giving us a good deal of information about the correct pronunciation of Greek. When dealing with rhythm he tries to prove too much, for he reduces his examples to metrical feet whereas, in prose, it is the total effect that matters, as Theophrastus seems to have realised.¹⁹ Dionysius insists that there must be variety, and that the language-music must be appropriate to the matter, the occasion, and the emotions which the speaker or writer wishes to arouse.

Dionysius recognizes three main styles of word-arrangement, the dry or austere …, the flowery …, and the intermediate. The first is the severe style of Thucydides: the words stand apart and cannot be run together, harsh collocations are deliberately used, with an abundance of broad syllables; the rhythms are impressive, the clauses not balanced equally. Sense and sentence-structure do not correspond, smoothness of every kind is avoided and even grammatical sequence at times disregarded. The other extreme is the flowery, elegant word-arrangement of Isocrates, and there the words are run together in flowing continuity, and only the periods come to a definite and distinct end. The clauses are carefully balanced, the rhythms neither too heavy nor too light, all harsh collocations are avoided, as is any hiatus between words. The general effect is like that of a painting where light and shade everywhere merge into one another. The intermediate type, which uses the effects of both extremes at the appropriate time, is the composition or word-arrangement which we find in Homer, Sophocles, Plato, and Demosthenes.²⁰

Dionysius' other extant works include fragments of a work on mimêsis or emulation, separate studies on the style of Lysias, Isocrates, Demosthenes, and Thucydides, and three short treatises known as the Three Literary Letters. The work on Emulation consisted of brief critical valuations of classical authors much in the manner of Quintilian's historical sketch of Greek and Roman writers in the tenth book of his Institutio Oratoria; indeed Dionysius may have been one of Quintilian's sources, but of course only for the Greeks. Dionysius' enthusiasm for Isocrates is for a practitioner of the true practical philosophy, an educator who made his pupils "not only clever speakers but men of good moral character, men who served their house, their city, and the whole of Greece."²¹ This is akin to the liberal education of the orator as understood by Cicero and Quintilian, but with more emphasis on rhetoric and less on general education.

Dionysius is unique among our extant critics in that he uses his critical-rhetorical formulae as a means of evaluating the style of an author, and not his authors merely to illustrate the formulae. He quotes freely from their writings and discusses certain passages at considerable length; he compares them with passages from others. This method of comparative criticism seems to have been characteristic of Greek criticism, at least in Rome, in his day, for we know that his contemporary and friend Caecilius of Calacte wrote critical essays on Lysias, whom he preferred to Plato, and attempted comparative valuations of Cicero and Demosthenes, of Demosthenes and Aeschines.²²

Dionysius has certain weaknesses: he is somewhat naïve and lacking in imagination, as in his strictures on Plato's diction; his moral earnestness betrays him into an unbounded admiration for Isocrates and condemnation of Thucydides as unpatriotic, or as inferior to Herodotus in his choice of subject: "one war, which was neither noble nor fortunate, which had much better not have happened, and, when it had, should have been left to silence and oblivion, and ignored by posterity."²³ It is only fair to add that this kind of nonsense is quietly dropped in his later essay on the historian. The moral earnestness itself, however, shows him to be more a man of letters than a rhetorician in the stricter sense, even though Demosthenes, for him as for Cicero, is the greatest of all writers.

Horace's letter to the Pisos on the Art of Poetry is, as already mentioned, contemporary with Dionysius. There is little in this informal, delightfully phrased advice that is original. We may note, however, that Horace is, in a sense, trying to do for poetry what Cicero attempted to do for prose: he insists that his poet must understand life and man, and that his work must be true to that knowledge. He emphasizes the need for unity and appropriateness, the need for talent and training, the importance of the choice of subject, of the choice of words, of structure, and so on; behind the deliberate informality of the epistle we can easily trace a thorough knowledge of the critical-rhetorical formulae of the day. But Horace is also concerned with specifically Roman problems, as for example, the controversy between the "ancients" and the "moderns."

After Horace we find ourselves in the world of the rhetorical schools, where the practice of declamatio was becoming increasingly fashionable and almost superseding every other method of teaching. Declamations were of two kinds: the suasoria in which the pupil had to imagine himself in some historical dilemma—should Alexander, for example, cross the Indus?—and devise a speech appropriate to the occasion; and the controversia, where he had to speak in an imaginary, often highly artificial and improbable, law-suit. The masters of rhetoric themselves gave display declamations of both kinds before an admiring public. The elder Seneca published a curious collection of such suasoriae and controversiae for the benefit of his sons, in which he records the adroit and clever things said by the rhetoricians of the Principate. His book is extant, and from it we gain a better understanding of how this practice, which put all the emphasis on ingenuity and clever epigram—for the same subjects were dealt with again and again—affected silver Latin style. The dangers of declamations and of their increasing artificiality are vigorously denounced throughout the first century by the elder Seneca himself, by Petronius, by Persius, by Tacitus, and at the close of the century by the great Quintilian, but they flourished in spite of them all.²⁴ It was in the last years of the century that Quintilian published his massive work in twelve books on the education of the orator, the Institutio Oratoria. Quintilian was Professor of Rhetoric by imperial appointment, and his book, complete, authoritative, lucid, sensible, not too original and at times more than a little dull, is an almost perfect pattern of what a professorial work should be. We find in it, in proper historical perspective, all the best thought of Rome on education, literature, criticism, and rhetoric.

We have no critical or rhetorical texts in Greek for this period, though it is fashionable among scholars today to assign to the first century A.D. both our Demetrius' treatise On Style, about the date of which more will be said later, and the famous short treatise On the Sublime which tradition attributed to Longinus in the third century A.D. In the second century A.D., however, we have the considerable works of Hermogenes on various qualities of style and forms of argument. They are purely rhetorical text books, dry and over-subtle in distinctions and classifications. We note that the theory of styles is now quite abandoned, and that Demosthenes is solidly established as the one supreme model for the young. The works of Hermogenes continued to be edited, commented on, and studied for the next ten centuries, but no really original mind appears in the vast collection of rhetorical writings which we still possess from that millennium.

Something more needs to be said, however, about the short and fragmentary treatise known as "Longinus on the Sublime."²⁵ Whatever its date it is a work of original genius. The author knows and uses many of the usual rhetorical formulae, but he remains their master, they never master him. It is not a work on the grand style or any other particular style, indeed the theory of particular styles is completely absent. Rather the author seeks to find the secret of the kind of great writing which suddenly sweeps one off one's feet. He traces it to five sources: vigour of mental conception, strong and inspired emotion, the skilful use of figures, noble diction or the proper choice of words, and dignified and spirited word-arrangement. The first, vigour of mental conception, implies nobility of mind, the power of grasping great ideas, and undoubtedly a certain grandeur. This, however, does not mean grand words; indeed it sometimes requires no words at all, as when Homer makes Ajax stride away in silence when Odysseus addresses him in the underworld, or the simplest of words, as in Ajax' famous prayer to Zeus to clear away the mist from the battlefield "that we may die in daylight." It does imply, however, the capacity to select the significant details and to weld them into a meaningful picture, and of this quality Longinus gives as an example the famous ode of Sappho, preserved here only, which begins "… [phaimetai moi keinos isos theoisi]."

The treatment of passion is lost. To illustrate the proper use of figures—a favourite subject in rhetorical critics who usually discuss it with great care and dullness—"Longinus" first gives a full and brilliant analysis of one sentence of Demosthenes, the famous Marathon oath in his speech On the Crown, and then gives examples of other figures to show how "as dimmer lights are lost in the surrounding sunshine, so pervading greatness all around hides the presence of rhetorical devices." Of the treatment of diction much is lost; what remains shows Longinus to be fully aware of the importance of the choice of words which "endows the subject matter with a speaking soul." He very properly allows vulgarisms in the right places, and insists that there is no limit to the number of metaphors that may be used in a passage provided that successfully conveyed passion can make them convincing.

As for composition, that is, word-arrangement and sentence structure, he glorifies the music of language in poetry and prose, a music allied with meaning and thus making a powerful appeal to the soul and mind of man. Like all the critics, he rejects both metrical, as against rhythmical, prose, and also broken, hurried, or monotonous rhythms. He insists that the anatomy of the sentence must be such that each part fits into the whole like the parts of a living organism.

"Longinus" sees the dangers of attempting greatness, and he names four vices into which the attempt may fall: turgidity, puerility ("the thinking of the schools which ends in frigidity through over-elaboration"), misplaced or artificial passion which leaves the reader unaffected, and frigidity which is often due to strange conceits. Faults and virtues are both illustrated from the greatest writers, and not even Homer is spared criticism. Indeed one of the most striking passages is where he argues that greatness is always accompanied by faults, for genius is careless, but it is always to be preferred to flawless mediocrity. The whole work is full of quotable passages such as the famous comparison of Cicero and Demosthenes: "The greatness of Demosthenes is for the most part abrupt, that of Cicero is like a flood. Our man is violent, swift, strong, intense; he may be compared to a lightning-bolt which burns and ravages. Cicero is like a spreading conflagration which rolls and ranges far and wide.…"

With "Longinus"—whether we place him before or after Hermogenes in time—the living stream of Greek criticism reaches its end. We saw that the philosophers of the fifth and fourth centuries posed the problem of literature's social function and responsibility, and built up the beginnings of literary theory. The contemporary development of rhetoric led to a study of style and stylistic devices as such, and it is this approach which triumphs after Aristotle. A continuous output of rhetorical textbooks was published throughout the centuries, but the more responsibly-minded men of letters, while they remain in the stylistic tradition of the rhetoricians, yet have absorbed from Isocrates and the philosophers a sense of the social function of literature, and it is upon literature as a whole, of which oratory is only a part, that they direct their attention. That the Roman rhetorical and literary theories remain derivative, while the Greeks still retained some originality, is proved by Dionysius and, gloriously so, by "Longinus" at the very end. Where, in this general line of development should we place the work known as "Demetrius on Style"?

The Treatise on Style

The author of our treatise obviously belongs to the rhetorical, not the philosophic tradition. Moreover, the moral concern for the character and education of the orator—the good man skilled in speech—which we have traced from Isocrates through Cicero to Quintilian, is totally absent. "Demetrius" is concerned with style exclusively. There is in his work no comparative criticism such as we have noted in Dionysius. His approach is quite objective: the theoretical framework is stated and illustrated from great and less great writers. On the other hand, he seems to be a man of letters rather than a professional rhetorician: he says nothing about types of cases, arguments, or issues, about ways of convincing a jury, or about methods of handling the different parts of a speech. His interests are obviously literary rather than rhetorical in the strict sense: the orators are frequently quoted, but only as practitioners of one kind of literature among many. We have here an example of literary criticism from a cultured man with a very good knowledge of Classical and early Alexandrian Greek literature, a man rhetorically trained, but not a mere rhetorician. The work is in many ways unique, the more so if it belongs to Hellenistic times, as it was traditionally thought to do, for we have no other extant critical text from this period.

The date of the work is, however, uncertain, and modern scholars have argued for various dates from the late third century B.C. to the middle of the second century A.D., but they have all but unanimously rejected the manuscript tradition which gives the author as Demetrius of Phalerum.²⁶ This Demetrius was the pupil and friend of Theophrastus, a distinguished man of letters and a voluminous writer. He ruled Athens from 317 to 307 B.C. on behalf of Cassander, the king of Macedon. He then fled from Athens to Thebes and later, after the death of Cassander, to Alexandria, where he enjoyed the patronage and friendship of Ptolemy Soter. Nothing is known of his later life or the time of his death except what we are told by Diogenes Laertius (5.78):

Hermippus says that, after the death of Cassander, in fear of Antigonus, he made his way to Ptolemy Soter. There he stayed a considerable time, and advised Ptolemy, among other things, to hand on the kingly power to his children by Eurydice. The king did not take his advice but passed on the diadem to his son by Berenice (Ptolemy Philadelphus) who, after the death of his father, decided to have Demetrius kept under guard in the country, until he decided what to do with him. There Demetrius lived in discouragement. He was somehow bitten in the hand by an asp while he was asleep, and died.…

Diogenes (or Hermippus?) does not directly connect the death of Demetrius with the king's displeasure, though he may seem to imply it, but Cicero, in his speech against Rabirius Postumus (9.23) quotes the case of Demetrius as one of those who owed their death to the enmity of a despot, and clearly suggests that he was murdered (aspide ad corpus admota). Ptolemy Soter died in 283/2 B.C.

Before we consider the evidence, internal and external, for both date and author, however, let us first consider the content and nature of our treatise, and, to avoid awkward circumlocutions, we may as well call the author Demetrius without assuming him to be Demetrius of Phalerum.

Nature and Structure of the Work

It falls into five sections. The first, which is also the shortest (1-35), is introductory and deals with different kinds of sentence structure, while the other four discuss four different "styles" or manners of writing, the grand, the plain, the elegant, and the forceful.

It is natural enough that the treatment of sentence structure should form a general introduction, since, with some exceptions to be mentioned later, different kinds of sentences can be used in all four "styles," and Demetrius begins with the general advice that the structure of the sentence must correspond to the structure of the thought. This correspondence must apply also to the clauses (the kôla, the basic units in prose as the verses are in poetry) and even to the shorter phrases or kommata. After he has repeated, but in his own terms. the Aristotelian distinction between the periodic style and the looser, non-periodic sentence structure, he differentiates three kinds of periods (a formula not found elsewhere).

At one extreme is the involved, Demosthenic period which he calls rhetorical; at the other extreme the looser, simpler, apparently effortless period of dialogue or conversation, which approaches the loose, non-periodic style; intermediate between these is the period he calls historical. There are some further remarks on different kinds of clauses, on the relation of sentence to sense, and a correction of Aristotle's definition of a clause (34).

After these general remarks, Demetrius proceeds to a description of what he considers the four main manners or styles of writing. This is his notorious theory of four "styles." But the word "style" is misleading, first because we think of style as something peculiar to the individual writer whereas the ancients thought of it more objectively, but mainly because, in ancient criticism, we associate the word [kharacter] mostly with the formula of the "three styles" which is found in later and mostly Latin writers.²⁷ There we find three separate styles rigidly differentiated, and while an orator is supposed to be able to use each of the three as the occasion demands, only one style can be used at a particular time.²⁸ But the Greek term [kharacter] is very general in meaning, and Demetrius' use of it is much less rigid. His "styles" can, with the exception of the plain and the grand, be mixed, that is, used at the same time. One can be elegant and forcible, plain and elegant, forcible and plain. The same word [kharacter], which is applied to these four "styles," is also applied to the faults to which each of them is prone; and Demetrius speaks of the "frigid style," "the affected style," and so on. He also speaks of the "epistolary style." The point is of some importance if we are to understand his intention: he is not drawing up a list of four styles by which you may judge this author or that, or different writings. He believes that there are four main elements of style, four qualities or manners of writing or speaking, and he examines how these are to be practised.

Each "style" is analyzed under three aspects: diction or the choice of words, composition or the arrangement of words, and subject-matter. On this last he has least to say.²⁹ And, somewhat irregularly under one heading or the other, he also brings in the figures of speech or thought that are most appropriate to each style.

The distinction between content and style is, as we have seen, found in Plato, and may be older. The subdivision of style into diction and word-arrangement is, if not explicitly formulated, certainly implicit in Aristotle, for chapters 2-7 of the third book of his Rhetoric deal with diction, while 8-9 deal with word-arrangement. The formula may have been clarified by Theophrastus.

It is important to realize that these subdivisions do not quite correspond to modern categories. Diction, which was later more precisely termed the choice of words, includes not only the choice of current or unusual terms, of rare or newly-coined words, but also words which express passion and character (an angry man uses different words from those he uses when sorrowful), different forms of the same roots, and so on. Further, diction also includes the use of loaded words, of metaphors and similes.³⁰ It includes all this in Aristotle as well as in later critics. We have already seen that the Greeks were shrewdly aware of the importance of semantics, illustrated in Aristotle by the story of Simonides who refused to write an ode on the victors in a mule race, "half-asses" as he contemptuously called them, but when the fee was increased he wrote the poem beginning: "Hail, ye daughters of storm-footed steeds, …" but, the philosopher comments, they were still the daughters of asses.³¹ Diction then includes in part the expression of emotion as well as the writing in character, in so far as these follow from the use of certain words.

Synthesis, or the arrangement of words once chosen, ("composition" though etymologically correct, is a misleading translation) has three things in view: the structure of the sentence, the sound of the words in collocation, and rhythm. The first of these Demetrius has already discussed in the introductory section. The second (much neglected today) is part of that music of language to which we have seen that the Greeks were extraordinarily sensitive. It too has a part to play in the expression of emotion, for, though we are less aware of this, an angry man uses harsh, guttural sounds where a lover or a suppliant quite unconsciously will use mutes and labials. It is still true that words that can be run into one another, without pauses between, are softer than those that willy-nilly make us stop. No one, however hard he tries, can (meaning apart) put the same emotional tone into two phrases such as "You accursed crooked cur" on the one hand and "My lovely angel sweetheart" on the other. The third point, closely connected with the other two, is rhythm, and this, in Greek prose as in Greek poetry (except when it was set to music) has nothing to do with stress or pitch, but only with the length or shortness of syllables. Here Demetrius, like all ancient critics, repeats the Aristotelian dictum that prose must be rhythmical but not metrical, that is, its rhythms must be more varied and never repeated regularly. It is then with these three things in view—word-arrangement, diction, and content—that Demetrius discusses each "style."

"Neighbouring" each successful style or manner is a particular vice or faulty style into which an unsuccessful attempt will fall. An unsuccessful attempt at grandeur or impressiveness is apt to fall into frigidity, attempted elegance into affectation, simplicity into dryness or aridity, and forcefulness will become bad taste. This theory of the "neighbouring vice" is found in other critics too—we know it best in Horace and Longinus³²—but whether Demetrius was the first to put it forward we cannot tell. As in so many things, our opinion will depend upon our view of his date. He does not claim to be original at this point.

Structure of the Work

This general plan is quite simple and clear. It is also carried out more systematically than is often alleged, and what appears at first sight to be a digression or repetition seems to be reasonably well placed in the work as a whole. Demetrius does, however, seek variety, and he deliberately begins one section with subject-matter, another with diction, and so on. Certainly there are repetitions: a figure of speech may in one place be said to contribute to elegance, for example, while later we find it contributing to forcefulness. There are many such cases, and this seems at first sight confusing, but is Demetrius not essentially right? Is it not true that the same way of putting things has, in different contexts, a different effect? We will take one example which is found in the discussions of three of our "styles," namely anaphora or the repetition of words at the beginning of consecutive clauses.

In 61 we have the passage on Nireus in the Homeric catalogue of ships. Demetrius says that Homer makes this one mention of Nireus, who led only three ships to Troy and is never heard of again, much more impressive by the use of anaphora. This is the passage (Iliad 2.671-674):

Nireus from Syma brought three curved ships,
Nireus, son of Aglaïa and of Charopus,
Nireus, most beautiful of all the Greeks
Who came to Ilium, save Achilles only.

Here, surely, Demetrius is right. In 141 he is discussing the elegant charm of a poem of Sappho, of which he quotes:

O Evening Star, all things you bring;
You bring the sheep, you bring the goat,
You bring the child to its mother.

Dimly through this inadequate translation, and much more clearly in the Greek of Sappho (see note on 141), we can see that the same figure of speech has a quite different effect from that in the Nireus passage, that the effect here is one of charm, not of grandeur or impressiveness. Then we find anaphora once again in 268, from the great speech of Aeschines against Demosthenes: "You call him as a witness, a witness against yourself, a witness against the laws, a witness against the democracy."³³

Here again Demetrius is right. The effect of the repetition in this passage is like a series of blows, it is forceful, certainly not charming! The fact that these effects are so different may raise doubts as to the soundness of Demetrius' basic categories but, given these, the discussion of the same figure in different sections is justified, and Demetrius is the better critic for having seen that figures of speech depend for their effect on other factors. Not all his repetitions can be justified in this way, and we shall note them as we read the text, but some of them can be, and the point should be kept in mind.

Demetrius has also been criticized for digressions. Rhys Roberts (pp. 28-31) listed no less than fifteen "subsidiary topics" of this kind, in fact every passage where Demetrius ventures upon some general description or explanation which is perhaps not immediately required by his present purpose. Of the nine so listed in the discussion of the grand style, only one is of any length and worth mentioning here; all the others seem perfectly natural where they occur. It is the discussion of metaphors and similes (78-90), a subject which Aristotle had discussed at some length and regarded as the main ornament allowed in prose. Now metaphor is appropriate in a discussion of impressiveness, but not only there, and it is a general discussion of the subject which we find at this point. It is an extremely interesting passage, in which Demetrius discusses, where it first comes up in a natural way (there are several references to impressiveness in the passage), the whole subject of metaphor. And he does not discuss it again, although, obviously, a metaphor can be elegant or forcible!

The digression found under the plain style are clearly due to a change of plan in this section. Demetrius deals mainly with the qualities which the plain style should have, namely lucidity, persuasiveness, and vividness, and he tells us how to achieve these by means which may concern both diction and word-arrangement. Hence, after a very brief statement (190-191) on simple subject-matter and diction, he says that lucidity is here the chief aim, and proceeds to tell us how this lucidity is to be attained (192-203). He then goes on to word-arrangement (200-208); we are told that vividness and persuasiveness are most acceptable to the plain style and we proceed to deal with these two qualities (209-220 and 221-222). After this comes one of his best passages, on the style of letter-writing: this too should be in the plain style (222-235).

Thus the treatment of the plain style follows a different plan. Perhaps Demetrius felt that all that could be said about plain diction and composition was that they should be plain, and that it would be both more profitable and more interesting to discuss the main qualities which this style strove to attain. He may also have welcomed the opportunity for a variation in his approach. It is true that these qualities are not the exclusive possession of the plain style; and it is true that in the discussion of these three qualities, lucidity, vividness, and persuasiveness (especially the last two), he goes beyond the bounds of the plain style. He does not deal with them again.

The discussion of the proper style for writing letters (223-235) can hardly be called a digression since it is a particular application of the simple style and is discussed under that heading. Demetrius may at first sight seem to contradict himself when, after saying that letters require the simple style, he concludes the discussion by saying that it needs to be a mixture of the simple and the elegant. This, however, is not really the case. He begins by disagreeing with Artemon who said that a letter should be regarded as one side of a dialogue. Dialogue as a literary genre tries to reproduce the style of actual conversation, but Demetrius feels that a letter is something more than that. While it is true that the subject and the style should both be simple, a letter is also an expression of friendship, a "gift" from the writer to the recipient; hence it deserves more care (though not too obvious care) on the part of the writer, whose character it inevitably reflects; it must not, on the other hand, become a treatise. Such treatises, it is true, are often addressed to a correspondent, but they are essentially quite another genre. The proper epistolary style, then, is basically simple, but there should be a certain admixture of elegance, and the letter may be said to require a mixture of the simple and the elegant styles.

This is the first discussion of letter-writing in ancient texts and as such of considerable historical importance. The principles it expresses are found reflected in all later theorists on epistolography.³⁴

Finally, in the section on the forceful style, there are two alleged digressions, the treatment of "figured language" (287-298) and that of the hiatus (299-300). The second of these is very brief and, since hiatus is said to contribute to forcefulness, it is no digression at all but a further reference to a subject already more fully discussed under impressiveness (68-74).

The first is more important and more interesting for the light it throws upon our author's method. After treating of the peculiar force of Demades, he is led to mention a figure, much misused (presumably to attain forcefulness) by contemporary orators (287), which he calls … not so much "figured language" as allusiveness or innuendo. This can properly be used for reasons of good taste or discretion. In the first case it is forceful, as is shown by an example from Plato (288-289); then the cases of discretion are explained: the figure is used when addressing a despot, an all-powerful populace, and the like (292-295). The subject of this figure arises naturally in a discussion of forcefulness, to which it often contributes, and, as it is of some interest, its other uses are then added, even though they are not strictly relevant. At this point comes a true digression (296-298) on the different forms in which a thought may be expressed—statement, advice, question, and so on. This arises very naturally from the different effect of allusiveness and bluntness, but it is irrelevant to the subject under discussion, to which we then quickly return.³⁵

From this brief survey of supposed digressions we may conclude that Demetrius does here and there allow himself to pursue for a moment a particular point which has come up naturally, and to do so at greater length than strict relevance would demand. He also varies his method, particularly in the section on plain style where he deals more with the qualities at which it aims than with the means of attaining it, but he does not seriously transgress the scheme he has set himself. Whether we blame or praise him for these near-digressions will depend upon the rigidity of our own minds.

Elegance or Wit?

A much more serious flaw in our treatise, one of ideas rather than of structure, seems to have escaped attention. It is a basic confusion in Demetrius' account of … the elegant or polished manner (128-189). This "style" does not, like the others, express one clear basic idea, and the confusion is reflected in the terminology. The word … by which it is described in contrast to the others, is almost immediately replaced by … charm or grace, and is thereafter little used in the discussion.

At the very beginning (128) we are told that elegant language may be described as a gay playfulness of expression. The charm is then said to be of two kinds: on the one hand the graceful poetic charm of such passages as Homer's description of Nausicaa playing among her handmaidens; and, on the other hand, witticisms. The difficulty is that too much is included under the second: the witticisms quoted at the very beginning (e.g. the old woman's teeth are sooner counted than her fingers) have no trace of charm, or indeed of elegance or grace. Moreover, this is true of many other jests quoted; for example, the grim humour of the Cyclops (130) and Xenophon's jest (134) about the hard-headed Persian from whom it would be easier to strike fire than laughter.³⁸ When Demetrius calls this the most effective charm … (135) he is straining the meaning of that word beyond all bearing. He seems to have included all witticisms, whether gracious or crude, under his second kind of [kharis], and to believe that, because they could all be called [kharientismos], a word that could be applied to any kind of witticism, he could use the root word [kharis] to apply to them also, although in all its usages elsewhere the word implies what we would call charm or grace or graciousness. Even if it could be so used—and no doubt it is presumptuous to challenge the Greek usages of a Greek—the ideas he deals with have nothing in common with each other except a certain cleverness in the handling of words, which is not enough to class them together under one and the same manner of writing. Homer's description of Nausicaa has grace and charm in plenty, but no wit; Sophron's remark about the rascal who made as many drachmas as he deserved strokes of the lash is very witty but by no stretch of imagination can it be called charming or gracious. It is a [kharientismos] but it has no [kharis]. It seems more than a matter of taste, which notoriously varies in different places and ages. Demetrius may well have considered the cruder witticisms he quotes as in good taste—and here lies another verbal confusion for he later uses the word [akhari], bad taste, for the vice neighbouring on forcefulness (302). He might therefore have considered these witticisms as [enkhari] (he uses that term as nearly equivalent to [kharieis] in 156), but witticisms can be in good taste without having grace or charm. The confusion of words implies a confusion of thought.

Demetrius seems to have realized something of this when he, at a later stage in this discussion (163), draws a definite distinction between the charming … and the ridiculous …, but only for a moment, for we find that this becomes only a repetition of his former distinction between poetic grace and jest. The section is confused (163-169) both in thought and language since up to this point he had called both divisions … charming.

This confusion is a serious flaw in the discussion of the elegant style. True, elegant playfulness of language can be divided into charming grace on the one hand and gracious wit on the other. The mistake is to include all witticisms, even the grimmest and the merely witty, under the second head, and not to have excluded here, as he should have done, those grim and satiric jests which he himself later includes under the peculiar forcefulness of Demades (282-286). He does not there draw attention to their wit; yet a comparison with 130-131 will show that they are essentially of the same kind, and that he was aware that a relationship existed between wit and forcefulness.

However, where he is dealing with the elegant and charming properly so-called, he has a great deal to say that is worthy of our attention.

Demetrius and Aristotle

Another feature of our treatise is its peculiar, indeed in many ways unique, relationship to Aristotle. It is obvious, and universally admitted, that much of what Demetrius says can be traced back to the Rhetoric of Aristotle. It is also clear that this dependence is far more pronounced in certain sections of the book than in others. But there is more than that: our author seems to use Aristotle in a manner unparalleled in other critical texts. Sometimes he corrects what Aristotle says and improves on him; at other times he changes the Aristotelian terms in what seems a deliberate manner, now and again making them simpler and clearer. In a word, he uses Aristotle for his own ends. The question is whether such changes as he makes should be attributed to unknown intermediate sources or whether they argue personal knowledge of the Rhetoric and, further, whether his attitude to Aristotle is more likely at an early date such as the early third century. Certainly no parallel can be adduced from Roman times.

The introductory section makes use of Aristotle at every turn; that on the grand manner makes use of him on particular subjects such as rhythm (38-43), metaphor (78-90), and frigidity (116), and there are frequent Aristotelian echoes, particularly at 58, 77, and 93-115. The discussion of the elegant style has very little that is Aristotelian, though his writings are three times used in illustration (144, 154, 157). On the simple style there are a few Aristotelian echoes (190-194) and references to his letters in the section on letter-writing. The section on forcefulness has practically nothing that recalls him. Demetrius' scheme of four styles is of course quite un-Aristotelian, for to Aristotle there was only one right way of saying a particular thing at a particular time; moreover he recognized only one virtue or arete of speech, namely lucidity.

As the greatest use of Aristotle is made in the discussion of sentence structure the peculiar nature of the relationship will emerge most clearly if we deal with it first (1-35). We may note at the outset that the very title … already seems to betray a careful use of words.…

Any reader who will glance at the first five sections of the ninth chapter of the third book of Aristotle's Rhetoric will see at once how much Demetrius' discussion of sentence structure derives from it. All the main ideas are there, but they are mostly applied with a difference. Aristotle says the period limits the sense, should end with it, and that this is the main difference between the periodic and the loose or unperiodic structure. A period should also be capable of being uttered in one breath. All this is in Demetrius, but he goes further: he applies the correspondence with the thought also to clauses (1-3) which to him are the basic unit; they too must be of reasonable length (4-7, cp: Rhet. 3.9.6). He associates brief clauses—and here he brings in the even smaller division, the phrase or komma—with maxims and laconisms (cp. Rhet. 2.21.13). He then quotes Aristotle's definition of the period, but he quotes it only in part, and expands this with the image of runners who have their goal in view (11, cp. Rhet. 3.9.2-3), using his own words and adding the etymology of the word [periodos] which Aristotle does not mention.

There are interesting changes in terminology. Aristotle's term for the periodic style is [leksis katestrammene], Demetrius keeps the adjective but changes the noun to [ermeneia]; Aristotle's term for the looser, non-periodic style is [legis eiromene], a rather difficult phrase—the "strung-along" style. This Demetrius changes to the simpler … "the disconnected style"—an adjective which Aristotle used for a period of which the clauses are not in antithesis (3.9.7). Both writers note the loose style of the older writers (Rhet. 3.9.1 …) but Demetrius quotes Hecataeus instead of Herodotus. He ignores at this point Aristotle's division of periods into those with one and those with more clauses, but later (19) replaces it by a tripartite division of his own into rhetorical, historical, and conversational periods.… He then uses as an example of this single-clause period the sentence of Herodotus which Aristotle used as an example of the non-periodic, and which Demetrius there replaced by an example from Hecataeus.

In 15 Demetrius says, with emphasis on the first person singular, "I consider" that a discourse should not consist of a series of periods. There is no such statement in Aristotle, but it is a commonplace in Dionysius of Halicamassus and in Cicero. The advice to avoid monotony is Aristotelian in other contexts (e.g. on rhythm, 3.8.1). Nor is there anything in Aristotle to correspond to the requirement (18) that the final clause of the period should be the longest.

At 22-24 we come to antithesis, also discussed by Aristotle (3.9.7) who does not distinguish between antithesis of thought and of words (a doubtful distinction). He does, however, also condemn the fake verbal antithesis (3.9.10) and Demetrius uses his example of it, adding a comment of his own.

At 25 there is another change of terminology. Where Aristotle (3.9.9) restricts paromoiôsis to similarities of words or sounds at the beginning or the end of consecutive clauses, and uses parisôsis for balanced clauses of equal length, Demetrius uses the adjective XXX to cover both—a more natural use since the word simply means similar—but he uses XXX, a much more expressive word, to refer to balanced clauses. Of his three examples, the first comes directly from Aristotle in this passage, the second is used by the philosopher in another connection (3.9.7); the third is not found in the Rhetoric. Neither is Demetrius' comment that the use of these studied devices is inadvisable in forceful or emotional passages, but he illustrates his point by examples from a lost work of Aristotle.

The definition of the enthymeme (30-33) is Aristotelian (Rhet. 1.2.8) and Demetrius' insistence on the difference between it and the period is in accord with that meaning.⁴⁰

The last references in this section are the most interesting. Demetrius quotes (in his own words) the Aristotelian definition of a clause as "one of the two parts of a period" and naturally complains that this contradicts the existence of the one-clause period. He then quotes and accepts the correction of one Archedemus into: "a clause is either a simple period or a part of a composite period" and he ends his discussion of sentence structure by another statement (35) about Archedemus: that he seems to imply that a period may consist not only of one or two, but of three or more clauses. Demetrius had already stated (16) that he considered four clauses the advisable limit. Let us be quite clear that Archedemus (whoever he was) is quoted only for his definition, with a comment that this definition appears to imply that Archedemus established no limit to the number of clauses.⁴¹ I am at a loss to understand the position taken by Roberts (and others), when he says (p. 218) that this reference to Archedemus "is of such a nature as to suggest that the author of the [Peri] may have drawn a good deal of his doctrine from him."

In the other sections the dependence on Aristotle is less constant. When Demetrius begins his discussion of the impressive style with a discussion of rhythm, he is obviously making use of Rhet. 3.8.4-6. But the way he uses it should warn us to be cautious in the use we make of such alleged quotations where we do not have the original text before us. Demetrius' words are: "A paeonic arrangement is, as Aristotle says…,⁴² impressive" (38). Occurring as this does in a discussion of the grand manner, it might be thought to imply that Aristotle said that the paeonic rhythm was appropriate to the impressive style, and one might well deduce from this that the theory of several styles goes back to Aristotle. Fortunately we have Aristotle's text, and he says nothing of the kind. His discussion of rhythm has nothing whatever to do with any theory of styles, nor does he use the adjective [megaloprepes], grand or impressive. What he does say in his discussion of rhythm is that one should aim at dignity and take the hearer out of himself. … and he then goes on to advise the use of the paeonic metre at the beginning and end of clauses, presumably as having this effect. Since both dignity and passion belong, for Demetrius, to the impressive style, he is quite justified in saying that, as Aristotle says, the paeonic is impressive, for what Aristotle says does fit in with the description of the grand manner. There is no intention to deceive, only to use Aristotle for his own purposes. He then goes on (41) to use Theophrastus to improve on Aristotle. Even if we cannot have paeons exactly at the beginning and the end of a clause, we can still have a general paeonic effect, "and this is what Aristotle, it would seem, recommends," and he gives an example of his general paeonic rhythm from Theophrastus. But Aristotle did not say so, and probably did not mean this.

Sometimes our author obviously has a passage of Aristotle in mind, but does not mention him. Aristotle said that connectives… should correspond exactly (Rhet. 3.5.2) but Demetrius says they should not, at least in the grand style (53), and gives an example from Antiphon.

The discussion of metaphors and similes (78-90) is full of Aristotelian echoes. A number of the same examples are used. There are also several things which Aristotle does not say: that a metaphor is sometimes more lucid and precise than the specific term (82), and that the transference should, at least in the grand style, be from the greater to the less, not vice versa (83), that in metaphors from one species to another, the transference does not always work both ways, so that you can say the foot of a mountain but not the slope of a man (79). These are good points, not in Aristotle.

The treatment of similes in the two authors is perhaps the most interesting of all. Demetrius says that when the metaphor is too bold we should convert it into a simile, and similes are safer in prose (80); the examples he gives conform to Aristotle's classic distinction between the two (Rhet. 3.4.1), namely that "Achilles leapt like a lion" is a simile, while "A lion, he leapt" is a metaphor. But when Aristotle goes on to say that "the simile is useful in prose, but rarely, for it is poetical" while he allows metaphor as the ornament in prose (3.2.8-15), we are confused, for it is simply not true that "Achilles, a lion, leapt" is less poetical than "Achilles leapt like a lion." This strange advice is almost certainly due to Aristotle's having the extended Homeric simile in mind, but the core fusion remains. Demetrius clarifies all this: to him the simile, as stated above, is safer, but he adds that the simile must be brief (89); if it is extended it becomes a "poetic comparison" (… i.e. a Homeric simile) and this can only be used in prose with the greatest caution (90). There can be no doubt that Demetrius is improving on Aristotle.

Enough has been said to show that our author's relation to Aristotle is unique in extant critical texts. It might be described as respect falling short of veneration, and he does not hesitate to improve upon the master. He is quite ready to give credit to others, as when he quotes Theophrastus always very aptly, but he does seem to be himself quite familiar with Aristotle. He may be quoting from memory, and this may account for some inaccuracies, but it would not seem to account for his changes in terminology which impress one as purposeful. The use he makes of Aristotle is everywhere subordinated to his own categories.⁴³ This way of using Aristotle cannot be paralleled in any other critic whether Greek or Roman, in the first century B.C. or later, and it seems unlikely that it derives from one or more intermediate sources.⁴⁴ I am again at a loss to understand Rhys Roberts' view (p. 249) that "the relation … of the [peri hermeneias] to Aristotle suggests a follower far removed in time." As far as it goes, it suggests to me the exact opposite.

We cannot prove, of course, that this or that point was not derived from an intermediate source, but we must not automatically assume that this applied to everything Demetrius says where no sign of such sources is found in our text and we have no evidence elsewhere that the opinions found here were held by anyone else. Our author's attitude as a whole seems to be far more easily explained as that of a man who had read Aristotle himself, and used him independently. Such a man is. much more likely to have lived in the late fourth or early third century than at a much later date.

Theophrastus

Next in importance to Aristotle himself comes Theophrastus. He is quoted four times in all, each time on a specific topic, and each time he was well worth quoting. Curiously enough, the particular theories attributed to him here are not found elsewhere. At 41, Theophrastus' example of a paeonic clause which does not begin or end with a paeon leads to the improvement already noted on Aristotle's conception of prose-rhythm, and for this improvement Theophrastus was probably responsible, though the improvement may have been implicit rather than explicitly stated in Theophrastus. At 114 a general definition of frigidity from Theophrastus precedes quotation of Aristotle's analysis of frigidity of diction into four kinds (116). At 173-174, Theophrastus' analysis of the causes of beauty in words is quoted in preference to Aristotle's remarks on the subject (Rhet. 3.2.13, in the discussion of metaphors, of which Demetrius makes use elsewhere), probably because his terminology was clearer and because he dealt with the subject specifically. Finally, at 222 we have Theophrastus' suggestion that some things should be left to the imagination of the reader; to this there is no parallel in Aristotle.

This seems to imply that Demetrius knew his Theophrastus as well as his Aristotle. If he were quoting from an (unknown) intermediate source, it is unlikely that he would have restricted himself to these four cases or that, in each case, the reference would have been so apt; it seems very much simpler to conclude that Demetrius made his own selection. Nor do I see any supporting evidence for the suggestion that "Theophrastus is probably followed in many other places." It may be so, but it is at least as likely that Demetrius took Aristotle as his primary guide in certain sections, and quoted Theophrastus when he found something particularly good in his works.

The Question of Date

References to Aristotle and Theophrastus do not, in themselves, affect the question of date, though the manner of them may, if we believe it to indicate considerable familiarity and a freer use of them than is found in writings of Graeco-Roman times. References to later writers, however, may well affect the date of our treatise. We will now examine further references which have been considered incompatible not only with the authorship of the Phalerean Demetrius but also with an early Hellenistic date.

1. References to persons.

One unusual feature of our treatise is the large number of references to, and quotations from, persons who are known to have lived in the late fourth and early third centuries. These are interesting in themselves; moreover, they would seem at least as compatible with an earlier as with a later date.

There is, first of all, Demetrius of Phalerum himself. Words spoken by him about Craterus are quoted as an example of innuendo at 289. As long ago as 1594 Victorius took this as a proof of the authorship of that Demetrius, which it obviously is not. On the other hand, Rhys Roberts expresses the view now generally accepted when he says: "No literary reference throughout the De Elocutione is so damaging to the traditional view as this," presumably on the assumption that such a self-revelation is in impossibly bad taste. But that is not true either. Plato and Xenophon, for example, do mention themselves in their works. All we can be sure of is that if our author did tell a story about himself, he would certainly have told it, as it is told, in the third person. The story itself is no evidence one way or the other. One interesting point about it, however, is that it is a late fourth-century anecdote told by no one else.

At 193 where Demetrius has been speaking of disconnected sentences as more appropriate for oral delivery, he continues: "That is why Menander, whose style lacks connectives, is mostly acted, while Philemon is read." Menander's date is 342-291, and Philemon's 361-262. This way of referring to them, we are told, "seems to be the judgment of posterity" (Rhys Roberts, p. 53). But is it? There is a parallel in Aristotle where he too is discussing the difference between oral and written style, a passage which our author most probably had in mind.⁴⁵ Aristotle says: "Poets who write to be read are in everybody's hands, like Chaeremon … or Licymnius.…" Licymnius lived a generation or two before Aristotle, but Chaeremon, the tragic poet, was certainly his contemporary. Aristotle never hesitates to mention contemporaries where appropriate. Our author need not avoid doing so. His reference to Menander and Philemon does not prove them contemporaries, of course, but neither is there any reason to consider this "the judgment of posterity." In fact the passage proves nothing at all, except that the treatise cannot have been written before Menander and Philemon were well known to their contemporaries, but then they both were well known before the end of the fourth century!

The lengthy reference to Ctesias (212-216), whom others criticized for prolixity while Demetrius defends him as being vivid, offers no difficulty of dating, since he was writing at the end of the fifth century, and it is difficult to see the point of Roberts' remark that he was "not yet a classic" in the time of Demetrius of Phalerum. He had been dead over a hundred years by the early third century and could, whether a "classic" or not, be the subject of literary controversy.

A sentence is quoted at 182 from Dicaearchus, a contemporary of Theophrastus, and Nicias, the painter, is quoted at 76: (f7. 340 to after 306). We cannot build much on the tense of these various references, though "Nicias used to say that the choice of subject was no smaller part of the art of the painter" is at least as compatible with personal knowledge or oral tradition as with quotation from another intermediate source.⁴⁶

The three references latest in time seem to be Clitarchus who is criticized for bad taste in 304; Praxiphanes, who is quoted in 57 as a literary critic, and Sotades referred to in 189. Clitarchus' exact date is uncertain. Roberts puts his floruit at 300 B.C., the Oxford Classical Dictionary speaks of him as writing some time after 280 B.C. Praxiphanes cannot be dated exactly; he is said to have been a pupil of Theophrastus, and Callimachus (305-240 B.C.) wrote a book against him. Sotades was a contemporary of Ptolemy Philadelphus and seems to have been old enough to object to that king's marriage in 289.⁴⁷ All three seem to have belonged to the first half of the third century, and indeed to have been active in the first quarter. Sotades was notorious later for his broken and soft verses, known as Sotadeans. Roberts says: "the use of the term 'Sotadean' for feeble and affected rhythms is probably of still later date." We might reply that the term need not have been in general use when it was used for the first time, but it is more profitable to reflect that the Alexandria of, say, the second quarter of the third century was a highly sophisticated literary society where reputations were quickly made (and unmade). We know that there was a great deal of literary controversy among contemporaries. The literary war between Callimachus and Apollonius of Rhodes is famous enough; the book of Callimachus Against Praxiphanes has already been mentioned. In that kind of atmosphere references to contemporaries (in a work like ours) would be very natural, and even the term "Sotadeans" might well be current during the lifetime of Sotades.

This does not prove that our author was writing in Alexandria at that time, but not one of these references is incompatible with this assumption, nor are they evidence of a "later" date. And if we ask at what time these unusually frequent references to writers of the later fourth and early third centuries are most likely, we may well answer that they would be much more likely in the first half of the third century than at the "later date," such as the first century B.C. or A.D., to which our treatise is now commonly assigned. Indeed it would be hard to find any author in those centuries as familiar with the late fourth century and early third as our Demetrius seems to be.

The references to the orator Demades (fl. 350-319 B.C.) as an exponent of forcefulness where he gets a brief section all to himself (282-286) is also significant. The works of Demades were, according to Quintilian (2.17.13), not committed to writing, so that they were not extant in Roman times. We may, if we wish, suppose that the examples given here were taken from a "collection" of his sayings, which would then have to have been made shortly after his death and survived. We have no evidence of such a collection, nor is he often quoted by later writers. It is at least as likely that these very striking sayings were remembered in oral tradition at the time of our treatise; they would certainly live on for some time. And we are reminded that Theophrastus valued Demades more highly than he did Demosthenes.⁴⁸

I have not so far mentioned the unknown three, Archedemus (34-5), Artemon (233) and the mysterious Gadereus (237), because we know nothing about them. Archedemus corrected Aristotle's definition of a kôlon or clause: there is no reason whatsoever to connect him with a Stoic philosopher of that name who may have lived as late as 130 B.C., except that they have the same name. Nor do we know anything about Artemon except what we are told here; true, we know of several other men of that name, one of whom also may have been living in 130 B.C., but there is no reason to identify him with our Artemon.⁴⁹ However, the real display of ingenuity is …: "as Gadêreus wrote (in trivial fashion) about the battle of Salamis." This improbable name is likely to be a corruption, and editors have amended it to … "the man from Gadara." This Gadarene is then identified with Theodorus of Gadara, the famous rhetorician and tutor of Tiberius, but if he wrote on the battle of Salamis, trivially or otherwise, we have no record of it. The identification is then used as evidence for a later date, and so is the manner of referring to him by the name of his city, a later practice. When scholars thus argue on the basis of their own conjectures as evidence, it might be salutary to remember that we teach students to respect Socrates because he knew he did not know what he did not know. The truth is that we have no means of knowing who the mysterious "Gadereus" is intended to be, just as we are quite unable to identify either Archedemus or Artemon. These three unknown names provide no evidence at all.

2. Other "late" references.

As far as references to persons in our treatise are concerned, the unusual familiarity with many who lived in the late fourth and early third century would seem to favour an early date, in fact one almost contemporary or at least within the reach of oral tradition; on the other hand the references to Clitarchus, Praxiphanes, and Sotades would seem to make the first quarter of the century improbable but offer no difficulty if we assume a date about 270 B.C. or later. However, the absence of references to later authors is no argument for an early date, for it was the practice of even later critics to draw their illustrations from the writers of Classical Greece and Alexandria. Nor would it be safe to draw any conclusion from the absence of references to later critics, though we may well feel that in a later work we might expect some reference to Hegesias (250 B.C.) and to the Asiatic style in general, for both were universally condemned by critics from the time of Cicero on, and these references would have been especially apt in connection with the affected style. Indeed it has been suggested that some examples of it, such as "a Centaur riding himself and the pun on Olympias (187) are typical of Asianism and may well be derived from Asiatic writers of the third century.⁵⁰ But if we remember that Hegesias' floruit was the middle of that century, a taste for affectation obviously developed before his day, and our author may well be quoting from writers with "Asiatic" tendencies earlier in the century, before the name of Asianism or the reputation of Hegesias had developed. Nor need we regard historical inaccuracies (at 238 Aristides is reproached for not being present at Salamis) or excessive hyperboles as betraying clear traces of the artificiality of the rhetorical exercises of Roman times. Even Isocrates was careless of historical truth and some pretty wild hyperboles are quoted by Aristotle.⁵¹

Our author mentions certain characteristics of rhetoric in his day. He tells us that his contemporaries are apt to confuse grandeur with eloquence itself (38); that forcefulness is in fashion (245); that the term [kakozelos] was commonly applied to affectation or preciosity (186) and this was apparently a recent coinage (239), which, in word-arrangement, was applied to broken rhythms (189) like those of Sotades; finally, that such affectation when combined with aridity of thought was called [kserokakozelia] (189 and 239). We are told also that the orators of the day did not know how to use innuendo without making it ridiculous (287). These are all faults of taste; they are quite insufficient to give any clear indication of date, but they seem quite appropriate to the early Hellenistic period when Cicero tells us the fatty or corpulent Asiatic style came into fashion.⁵² Certainly these faults are not particularly characteristic of Greek style from 100 B.C. to 100 A.D., and in so far as they were current then, they were certainly not new.

Certain specific phrases are also supposed to betray a late date. For example, Demetrius' use of the expression [oi arkhaion], the old writers or the ancients, to designate the Classical writers. This, we are told, is a late usage. The word is used three times: in 12, the adjective is applied to the loose, unperiodic sentence-structure; we have the same contrast in 244 and we are told again that "ancient" writing is unperiodic because "the ancients were simple men." Obviously, this does not mean the Classical writers but those who wrote before periodic structure came into fashion. Aristotle uses the word in exactly the same sense (Rhet. 3.9.1) and in the same context, and gives, as an example of the "ancient" manner, a sentence from Herodotus. Indeed Aristotle goes further, for in the previous sentence he compares the period to "the antistrophes of the ancient poets," … which must include at least Pindar, and probably the three fifth-century tragedians as well. The truth is that [arkhaios] is a very elastic term, like our own word "old," and, if Aristotle could apply it to fifth-century poets, we should not be surprised if a couple of generations later it could include Demosthenes in the one place (67) where Demetrius uses it of the Classical writers as against those of his own day: "the old writers use many figures but they use them so skilfully that they seem less artificial than those who use none." The contrast between old and contemporary writers is there, but such expressions may mean no more than half a century. Modern critics of the twenties and thirties certainly used such expressions to contrast their contemporaries with the writers of even the late nineteenth century.

In 204, Demetrius says that "the new comedy" uses shorter verses. We are told that "the triple division of comedy is Alexandrian and this suggests a later date than Demetrius of Phalerum." How much later? Does the expression "the new comedy" really imply a tripartite division? Aristotle already used a very similar expression when he said the difference between crude and polished wit is seen in "comedies old and new."⁵³

In 181, prose of a general rhythmical nature is said to be practised by Plato, Xenophon, and "the Peripatetics." The objection is that only "later" writers would thus have referred collectively to the school of Aristotle and to writers of that school. Perhaps, but we do not know when the Aristotelians became Peripatetics (LSF traces it back to about 200 B.C.). Is it not the kind of nickname that should be almost contemporary with the first two heads of the school?

The expression [hoi Attikoi] to designate the inhabitants of Attica, where Attic dialect is contrasted with Doric at 175-177, is even less significant. The adjective is, of course, quite Classical and Aristotle uses the phrase [hoi Attikoi hretores] for the Attic orators (Rhet. 3.11.16). There is no reason to consider this a gloss, but in any case Plato speaks of "the Attic speech" and "the old Attic speech."⁵⁴

There is one other phrase the meaning of which presents a pretty problem. In 108, Demetrius says "altogether, ornamental addition … may be compared to those external displays of riches, (ornamented) cornices and triglyphs and broad purples…" The meaning of this last phrase is very obscure, but the context requires some ornament used to display wealth, almost certainly in connection with buildings, probably awnings or drapes of some kind. The fact that Lucian in the second century A.D.⁵⁵ uses a somewhat similar expression to refer to the laticlavus, that is, the broad purple band which Roman senators wore on their toga, is hardly enough to see a reference to it here. In any case the senatorial purple was not a sign of wealth but of caste and prestige, nor did mere wealth entitle one to it, certainly not in the first century B.C. or A.D.

Finally, the citing of 172 as a proof of a later date by Roberts, Radermacher, and others seems a good example of faulty logic. Demetrius tells us that certain jibes may be allowed, and he clearly implies that they are allowed because their frequent use … has blunted their sting. The two examples he gives are that a tall dark man is called "an Egyptian clematis" and that a fool on the water (whether rowing or swimming is not clear) is called "a sheep at sea," the word "sheep" … being common parlance for a stupid fellow also in other contexts. Now we are told by Seneca (De Constantia 17) that Chrysippus (280-207 B.C.) mentioned someone who was indignant because he was called a sheep at sea, vervex marinus; and from Diogenes Laertius (7.1.2) we learn that in a work on proverbs or saws Chrysippus said that Zeno, who was tall and dark, was called an Egyptian clematis. As both expressions are traced to Chrysippus, this proves, we are told, that Demetrius is extracting them from Chrysippus' collection of proverbs. It surely proves nothing of the kind: it is in no way surprising that two jibes quoted by Demetrius as in frequent use should reappear in a collection of such proverbial sayings by Chrysippus one or two generations later. Proverbs hardly find their way into a collection until they have been in use for a considerable time.⁵⁶

Language and Vocabulary

Both Radermacher and Roberts, and indeed all those who have argued that On Style must have been written much later than the traditional date, have put a good deal of emphasis upon the language of the treatise. It is one of their main arguments, at times their chief argument. The fullest study of the language of Demetrius is that of Dahl, and most of his arguments will be found in Rhys Roberts' edition. In this question of language we are admittedly on very slippery ground, for we have no third-century prose works to give us a standard of comparison. We know that linguistic habits were changing by the second half of the fourth century. The number of word-forms, apart from technical terms, which appear for the first time in Aristotle and Theophrastus is very considerable. The Rhetorica ad Alexandrum is of special interest in this respect, for it is a fourth-century text and, being a rhetorical work, it is roughly of the same kind as On Style. Even a superficial study of its language and vocabulary makes quite evident that it contains a substantial number of words, expressions, and constructions which would certainly have been condemned as "late" if the date of the treatise had not been so firmly established.⁵⁷

At the end of the fourth century, Greek spread rapidly all over the Near East until it became the common language, the Koine of the Eastern Mediterranean world. All our later authorities agree that this was the time when the florid, over-rhythmic Asiatic style was born; all of them associate this style with the name of Hegesias whom they unanimously condemn, and the floruit of Hegesias was as early as 250 B.C. Clearly the degeneration was rapid. Obviously, many changes took place before then.

What kind of Greek, then, would be written in Alexandria in the first half of the third century, and written by the scholars of the Museum who had been attracted there by the patronage of the Ptolemies from all over the Greek world? They studied and edited all the great Classics, they were students of all dialects. They wrote Greek, but it is pretty well established that they did not write Attic Greek, nor any particular dialect, unless indeed a particular type of Greek was required by convention for a particular genre of poetry. One would expect to find Attic and Ionian, poetic and prose forms appearing cheek by jowl with colloquialisms we know only in Aristophanes; it would be a mixture of the old and the new, and a more heterogeneous mixture than either the later Koine of the New Testament, or the deliberate Atticism of Roman times. We should not be surprised, therefore, to find in a scholarly text of that period both forms and constructions that are specifically Attic, … and even an occasional use of the dual, which we know had almost disappeared. We should also expect a relaxation of that grammatical precision which is so characteristic of the best Classical authors (the beginnings of this process too we can trace in Aristotle), for example, a less exact use of prepositions, a slackness in the use of [an] with the optative, in short an erratic kind of inconsistency. And this is precisely what we do find.

As for the words themselves, we should always remember that our evidence is incomplete for any period. Unless the method of word-formation is itself late, when we call a word "late" it only means that it does not appear in our extant texts before a later period. Even where the manner of derivation is late, we cannot be sure, for the Greeks always formed words very easily, especially compounds. There is surprisingly a large number of words that occur only once in extant literature. Our treatise has sixteen such words.⁵⁸

Where the form is regularly derived from a word in common use in Classical times—when, for example, the verb and the adjective are used in Classical writers but the noun is not, or vice versa,—there seems very little point in considering such forms as "late."⁵⁹ It is true that later Greek shows a tendency to use compounds, and double compounds (verbs formed with two or even more prepositions) carelessly, where the simple word would do just as well, while the Classical writers are, with occasional lapses, more careful. This tendency too began in the fourth century. Our Demetrius, however, is here, at least on a number of occasions, on the side of precision. He uses double compounds with each preposition having its full meaning so that the double compound is fully justified.⁶⁰

In fact, Demetrius has a tendency to be precise also in his technical terms. We have already seen that he on occasion changes the technical terms of Aristotle, even where he follows him closely, from an apparent desire to simplify. We are very ignorant of the technical rhetorical vocabulary which, by the end of the fourth century, had been developing for well over a century. Plato quotes a few such terms in the Phaedrus in order to ridicule them; Aristotle gives us more, but his attitude to rhetorical subtleties is not very different.⁶¹ There are some more rhetorical terms in the Rhetorica ad Alexandrum; of Theophrastus we have only a few uncertain fragments on matters of style. Yet, clearly, the vocabulary of rhetoric and criticism had, by this time, reached considerable proportions. Most of these terms are, for us, inevitably "late." Moreover, even in the first generation after the foundation of the Museum, the learned men of Alexandria were no doubt adding technical terms of their own. We must therefore expect some "new" and some "late" technical words in Demetrius in any case. There are altogether about twenty. It would be very surprising if there were fewer.⁶² These terms do not appear again till the first century, but then we have no extant critical or rhetorical works in which they could appear.

It is perhaps more surprising that some words that are very common in later criticism do not appear in Demetrius at all. The most obvious of these are [tropos] in the sense of "trope," [apheles], later applied to the simple style, a word also used by Aristotle for the "simple" period, which Demetrius avoids, as he seems to avoid or not to know the phrase [eklage onumaton], the choice of words, meaning diction. Moreover, Demetrius uses a number of terms in a sense closer to the Classical than to the later meaning⁶³

If we carefully review all the words and expressions suspected of being "late," (as is done in Appendix I), we shall, I believe, conclude that the evidence for a date much later than the traditional one is, as far as language goes at any rate, not convincing. It is true that the effect of such suspicious words is cumulative; if enough words are listed as suspect, the reader is tempted to believe that there must be something in the argument, but we should not forget the truism that poor arguments, however numerous, do not lead to sound conclusions. There are a half-dozen Classical forms used in a "non-Classical" sense. Of the fifty or so words said to be "late" in form, most could pass anywhere and the exceptions are less than a dozen. There are some deviations from the Classical usage in the use of adverbs, prepositions, and syntactical constructions. All this is, I believe, quite consistent with an early third-century date.

Is it, however, consistent with the traditional author? It is true that Cicero regards Demetrius of Phalerum as the best Greek exponent of the middle style, yet sharply marks him off from the Classical writers and regards him as an exponent of a less vigorous, softer style. Quintilian says his style is more suitable for display than for the struggle of the courts. We know from other sources that he was very affected and easy-living as a young man. Cicero actually links him with the later "Asiatics," at a time when he was more kindly disposed to the latter than he was in his later years.⁶⁴ We may conclude that his affectations did lead him to affectations him to new ways of style. How would such a man write after some years in Alexandria? How far does an expatriate scholar adopt the ways of speech of his adopted country and of the community in which he lives? The question is hard to answer, especially in the case of a man like Demetrius. We may be quite sure that some non-Attic forms would slip in; yet he was an Athenian, a contemporary of Theophrastus and Menander, and we may well feel it hard to believe that he would write in this somewhat medley mixture of Attic and non-Attic. While other aspects of the treatise, its relationship to Aristotle and Theophrastus for example, would suit Demetrius very well, the language of the treatise may well be felt to be a strong argument against his authorship.

General Critical Views

To infer from the occurrence of the four "styles" here that "the writer lived at a time, considerably later than Aristotle, when the doctrine of the types of style had undergone many developments"⁶⁵ is to confuse our author's scheme of four main manners of writing which can all be used together (except that the plain and the grand do not mix) with the far more rigid formula of the separate and distinct three styles which we first meet in the Ad Herennium, and then in Cicero, but in no earlier Greek extant texts. The "styles" of Demetrius are much more like qualities of style than [kharakteres] in this more rigid sense,⁶⁶ and they seem rather to belong to a time when the three-style formula had not become current.

As for the general standpoint of our author, it does not seem to be "that of the Graeco-Roman period, earlier than Hermogenes and (possibly) later than Dionysius."⁶⁷ Demetrius does not seem to share either the point of view or the concerns of Cicero, Dionysius, Horace, or their successors. They are deeply concerned to restore rhetoric to its Isocratean place as the true philosophy based on a general education, and they insist that the writer, poet, or orator must know life and imitate it. Cicero and Dionysius are vehement in their condemnation of Asianism. There is not a word about all this in our treatise. There is no such comparative criticism as was the fashion in the circle of Dionysius and Caecilius. Demosthenes is not for Demetrius, as he was for Dionysius and Hermogenes, the supreme model; even [deinotes] is not his above all others, for he shares it with others as well as Demades. Demetrius is aware of the basic formula which differentiates matter from style and subdivides the latter into diction and word-arrangement, but the far more elaborate critical formulae of Dionysius do not appear. There is no trace of the rhetorical mimesis or emulation so important to Cicero, Dionysius, and Quintilian. The whole apparatus of criticism seems very much simpler.

We do not know what the Greek critics of the first century A.D. were concerned with; for the schools of Theodorus of Gadara and Apollonius seem to have concentrated on matters purely rhetorical.⁶⁸ We do know, however, that the Romans were extremely concerned with the evil results of declamations and the decay of eloquence. Indeed the treatise On the Sublime has been dated in the first century largely on the strength of its concluding chapter on this latter subject. There is nothing of all this in Demetrius, nor of the more specifically Roman controversies between the "moderns" and the "ancients," any more than about Asianism vs. Atticism, Analogism vs. Anomalism, or any other topical question of that period. It is hard to see in what respect our Demetrius can be said to share the standpoint of Graeco-Roman times.

As for the place where our treatise was written, the concern with literature in general rather than with rhetoric would suit Alexandria very well, and the apparent familiarity with Egypt would tend to confirm an Alexandrian origin.⁶⁹

External Evidence

There are a few somewhat uncertain references to our treatise in much later authors. There are also one or two references in Philodemus which deserve more attention. than has been paid to them. Also, the references to, and fragments of, Demetrius of Phalerum himself, which have recently been collected by Wehrli,⁷⁰ should be examined anew to see what light they throw on his possible authorship.

These last are somewhat disappointing. The [peri hermeneias] does not occur in the list of his works in Diogenes Laertius (early third century A.D.).⁷¹ The list is not very long, as compared for example with the works of Theophrastus, though Diogenes tells us that he wrote more than any contemporary Peripatetic. He also says that Demetrius wrote rhetorical works, but the only clearly rhetorical title is a Rhetoric in two books. There are, however, some other titles that may be rhetorical: the [peri pisteos], [peri kharitos], and [peri kairon] may well have been On Proof, On Elegance, and On Good Taste⁷² in the rhetorical sense, rather than On Faith, On Favour, and On Opportunity in general.⁷³ The word [kharis] is used by our Demetrius as one of the terms to describe the elegant style, and Radermacher even suggests (p. 95) that our author may have owed a good deal to this work of the Phalerean. It is, at any rate, a point of contact between the two, if a slight one.

Both Plutarch and Stobaeus⁷⁴ tell us that Demetrius of Phalerum advised Ptolemy to read books on kingship as he would find written there the advice his friends did not dare to give him, which may remind us of On Style (292-293) on how despots have to be addressed with care. More directly reminiscent of that whole section is a passage in Philodemus,⁷⁵ who says that Demetrius, together with the Sophistic kind of speeches, added to the deliberative and forensic kinds of rhetoric … that is, the speech which knows how to address all audiences and obtain favour with all. Philodemus seems to mean that Demetrius included this under "sophistic" (which in Philodemus means epideictic) rather than that he made it a fourth kind of rhetoric, but, in any case, he obviously attached great importance to methods of addressing, in particular, popular assemblies and ruling princes. This subject may also have been treated in the … (how to speak as an ambassador?)⁷⁶ which is listed among his works by Diogenes.… Now this … is highly reminiscent of what our Demetrius says about innuendo, which he too specifically links with princes and assemblies (289-293). Here again we have a distinct point of contact between the two Demetrii.

There is another interesting sentence in Philodemus, where he says that "long periods are bad for delivery, as we read also in Demetrius about those of Isocrates… "⁷⁸ We should note that the name "Demetrius" is uncertain, though the restoration has been generally accepted, and some commentators have admitted this as a reference to our treatise. The sentence occurs in a discussion of delivery in which just before "the Phalerean" (that name is quite clear) was said to have disliked the delivery of Demosthenes.⁷⁹ If the name "Demetrius" is correct in the sentence quoted, there is therefore no doubt that it refers to Demetrius of Phalerum, and if the reference is to our treatise, then Philodemus, in the first century B.C., already believed the author of our treatise to be Demetrius of Phalerum.

What Philodemus says of the long periods of Isocrates does accord very well with what we find in our section on forcefulness. From the very beginning of that section we are told that the forceful style requires short phrases rather than clauses (241-242), brevity and lack of smoothness (241-244), vehemence (246), the avoidance of the antitheses and balanced clauses found in Theopompus, the pupil of Isocrates (247). We may have periods but they should be short (252); forcefulness favours harsh sounds and abrupt endings (255-256), but not smoothness or regularly connected clauses (258, 269). Moreover, dramatic delivery is clearly related very closely to forcefulness, indeed at one point all but identified with it (271, and compare 193, where the disjointed word-order is said to be histrionic …). Returning to forcible word-arrangement at 299, we are there told that "smooth word-order, as practised most by the school of Isocrates" is not suited to forcefulness; it is better to do without connectives (301) and a little later (303) long, continuous periods are said to be unpleasant and tiresome in this style. We also remember that our author said (12) that he disliked a style which is entirely periodic like that of Isocrates.

In view of the specific mention of Isocrates at 299, of Theopompus at 247, of the emphasis throughout this section on characteristics of Isocrates' style which are undesirable in the forceful style, and the all but identification of the forceful style with that of delivery and debate, it would seem that Philodemus is indeed referring to our treatise when he says that it is "stated in Demetrius" that the long periods of Isocrates are not suited to delivery, even though where Isocrates is specifically mentioned Demetrius refers to the smoothness of Isocrates' periods rather than to their length. In Isocrates certainly the two went together. It seems, therefore, more than probable that Philodemus believed Demetrius of Phalerum to be the author of our treatise; in any case, we have here a very close similarity of viewpoint between the two Demetrii.

There is one other matter of interest in connection with Demetrius of Phalerum. We are told that, in common with Theophrastus and other cultured Athenians, he was very critical of Demosthenes' delivery; they considered it to lack dignity and simplicity, and they disliked his boldness and violence.⁸⁰ Now the attitude to Demosthenes in our treatise is unusual. He is quoted fourteen times in the section on forcefulness, usually with approval though he is once (250) condemned. No other style is illustrated by a quotation from him. At 80 a metaphor of his is quoted to illustrate the difference between a metaphor and a simile. The other three references are illustrations of periods in the first section. In other words Demetrius regards forcefulness as characteristic of the orator, and that style only. And even that section he shares not only with Demades but with seventeen illustrations from other writers.

How very different from the picture of Demosthenes which we find in Cicero, Dionysius, and Hermogenes! The [deinotes] attributed to him by all the writers of these later centuries is no longer the one quality of forcefulness but an over-all excellence in every kind of style, at times including force but not restricted to it.⁸¹ In our treatise Demosthenes has not reached that supremacy. Not only is he used to illustrate one style only, but other writers are both more frequently quoted and illustrate various styles. As against the eighteen references to Demosthenes, there are thirty-seven references to Homer, twenty to Xenophon, nineteen to Plato, eighteen also to Aristotle, and twelve to Thucydides. Altogether, Demetrius' attitude to Demosthenes is almost as unlike that of later critical texts⁸² as his attitude to Aristotle.

There are no other references to our treatise in writers of the Classical period. Ammonius, the commentator on Aristotle (ca. 500 A.D.) does refer to the [peri hermeneias] of "Demetrius" and explains that Aristotle's work of the same name dealt with a different subject.⁸³ True, he does not add "of Phalerum" but Demetrius of Phalerum is by far the most natural reference in such a Peripatetic context. One feels that if he had not meant the friend and pupil of Theophrastus he would certainly have said so. Specific references to the author as Demetrius of Phalerum do occur in the eleventh century and later, but these are of little value except in so far as they support the tradition of the manuscripts.⁸⁴

Conclusion

None of the arguments in favour of a date much later than the traditional one seem to me at all compelling. On the other hand, our author's familiarity with Aristotle and the way he uses him, his attitude to Demosthenes, his unusual familiarity with people living in the late fourth and early third centuries about whom he gives us a good deal of information not found elsewhere, his general critical outlook, his remarks on contemporary oratory and his silence about later controversies—all these things seem to point to a quite early date. The use of some technical terms in a Classical or near-Classical sense while often ignoring later meanings seems to point in the same direction. There are some references to Alexandrine personalities, on the other hand, which make a date before 270 unlikely. Certainty in such matters is of course impossible to attain, but I incline to the view that our treatise was written about 270 B.C., or not very much later.

As for the author, we have seen that, as far as critical ideas are concerned, there are several points of contact with Demetrius of Phalerum. It must also, I think, be admitted that Ammonius (ca. 500 A.D.) regarded him as the author. Far more important, it seems highly probable that Philodemus, in the first century B.C., also attributed our treatise to him. On the other hand, the un-Attic nature of the language does not seem to suit the Phalerean. Moreover, if we accept the evidence of Cicero and Diogenes Laertius quoted above⁸⁵—and in the absence of any evidence to the contrary it seems that we must accept it—Demetrius of Phalerum most probably died shortly after the death of Ptolemy Soter in 283/2 B.C. He cannot then have been the author of our treatise.

This leaves us with a treatise On Style probably written in Alexandria not much later than 270 B.C. by an unknown author. This author must, obviously, have had strong Peripatetic connections, and to some extent have shared the critical ideas of Demetrius of Phalerum, whom he had probably known, and with whose writings, as with those of Aristotle and Theophrastus on rhetoric and style, he was no doubt familiar. If his name was also Demetrius—for it was a common enough name—this might help to account for confusion between the two even at an early date.

Notes

¹ H. Diels, Die Fragmente der Vorsokratiker⁶ (Berlin 1951). Xenophanes fragments 1 and 11, also 14, 15, 32, and 34. Heraclitus fragments 40 and 42.

²Rhetoric 3.9.1-2. For Gorgias' praise of Logos see Diels, 2. 290.

³On Composition, ch. 25. Diogenes Laertius (3.37) tells the same story and attributes it to Euphorion and Panaetius.

⁴Laws 2. 668b-c.

⁵ At 236a.…

⁶Phaedrus 255-267. Plato selects a few technical terms for mention … This casual selection of technical terms, some of which do not recur elsewhere, obviously from a far larger number, is a salutary reminder of how little we know of the rhetorical vocabulary of the fourth century. Some of these elaborate compounds are of the kind we often consider as "later" because we do not come across them again till Roman times.

⁷Laws 2.653-673. For full references and a discussion of Plato's views on poetry and rhetoric see my Plato's Thought (London 1935 and Boston 1958) 179-215.

⁸ … See my Aristotle on Poetry and Style (New York 1958) xxi-xxii and xiv-xvii for mousik6 in the last chapters of the Politics. For another view see S. H. Butcher, Aristotle's Theory of Poetry and Fine Art (republ. Dover Publications 1951) 198-214.

⁹ For Thrasymachus' capacity to rouse the emotions see Plato, Phaedrus 267c and Aristotle, Rhetoric 3.1.7. Also Rhet. 3.8.4 for his use of paeonic rhythms. His short rhythmic clauses are mentioned by Cicero, who contrasts them with the rounded periods of Isocrates (Orator 39-40). Dionysius of Halicarnassus, in Lysias ch. 6, credits him and Lysias with a virtue of style.… This I believe to mean the capacity to express thoughts succinctly and compactly.… For a full discussion of these interpretations see my "Thrasymachus, Theophrastus and Dionysius of Halicamassus" AJP 73 (1952) 251-267. See also G. A. Kennedy's "Theophrastus and Stylistic Distinctions" in Harvard Studies in Classical Philology 72 (1957). He agrees that "the stylistic distinctions of Dionysius of Halicamassus (i.e., in the first seven sections of the Demosthenes) are made on the basis of diction" (p. 101).

¹⁰Nicocles 7. See also Antidosis 184, 207, 271-272, 344.

¹¹ Seneca Rhetor, Controversiae 1. Praef. 9; cp. Quintilian 1. Praef. 9.…

¹³ Aristotle (Rhet. 3.13) would, as a philosopher, prefer to recognize only two necessary parts of a speech, the statement … and the proofs.… However, he compromises by accepting four: the introduction …, the statement or narrative …, the proofs, and the epilogue or peroration.… He notes and discards as ridiculous further subdivisions such as (contrast), (recapitulation). From Theodorus of Byzantium, he quotes (supplementary narrative), (preliminary narrative), (refutation), (supplementary refutation), and, from Licymnius (improving one's case, lit. wafting along), (digressions, lit. wanderings) and (ramifications).

¹⁴ In particular, scholars have attempted to trace back to Theophrastus the formula of three styles and that of four virtues … of style. See H. Rabe, De Theophrasti libris … (Bonn 1890); A. Mayer, Theophrasti … Fragmenta (Leipzig 1910); and J. Stroux, De Theophrasti Virtutibus Dicendi (Leipzig 1912). See also G. L. Hendrickson, "The Peripatetic Mean of Style and The Three Stylistic Characters," in AFP 25 (1904) 125-146, and my "Theophrastus as a Literary Critic," in TAPA 83 (1952) 172-183. The controversy is in any case somewhat academic, as most of the ideas involved, apart from the actual formulae, were doubtless discussed by Theophrastus, as they were by Aristotle.

¹⁵ See A. Roemer, Aristarchs Athetezen in der Homerkritik (Leipzig 1912) and Die Homerexegese Aristarchs (Paderborn 1924); A. Severyns, Le Cycle épique dans l'école d' Aristarque (Liège 1928).

¹⁶Rhet. 3.16.4. For the Stoics see Karl Barwick, Probleme der stoischen Sprachlehre und Rhetorik (Berlin 1957).

¹⁷ See H. Caplan's edition in the Loeb Library (1954), especially the introduction.

¹⁸ The six parts of a speech are (1.3 (4)) exordium, narratio, divisio, confirmatio, confutatio, and conclusio.

¹⁹ See Demetrius 41, below.

²⁰ The three styles of word-arrangement are described by Dionysius in his Camposition, chs. 22-24.

²¹Isocrates 4-7 (543).

²² See J. W. H. Atkins, Literary Criticism in Antiquity (London 1952) 2. 106-108 for references.

²³Letter to Pompey, ch. 3. For a full account of Dionysius' criticism of Thucydides see my "Dionysius of Halicarnassus on Thucydides," in Phoenix 4 (1950) 95-110.

²⁴ See S. F. Bonner, Roman Declamation in the Late Republic and Early Empire (Liverpool 1949) and H. Bornecque, Les Déclamations et les déclamateurs d'après Senèque le Père (Lille 1902).

²⁵ See my translation Longinus on Great Writing (New York 1957) and Rhys Roberts' Longinus On The Subline (Cambridge 1907).

²⁶ The two standard editions of our treatise were published almost simultaneously: L. Radermacher's Demetrii Phalerei qui dicitur de elocutione libellus (Leipzig 1901) and W. Rhys Roberts' Demetrius on Style (Cambridge 1902). Roberts' text and translation were reissued, with a brief introduction in the Loeb Library in 1927, reprinted in 1953.

Both editors agree in rejecting the manuscript tradition which attributes the treatise to Demetrius of Phalerum, and would date the work between Dionysius of Halicarnassus and Hermogenes, that is, in the first century A.D. or early in the second. This was the generally accepted view at the end of the last century; see in particular the monographs of R. Altschul, De Demetrii rhetoris aetate (Leipzig 1889), F. Beheim-Schwarzbach, Libellus … qui Demetrii nomine inscriptus est (Kiel 1890), and K. Dahl, whose Demetrius … (Zweibriicken 1894) contains the fullest study of the language of the treatise. Some, like Dahl, and Hugo Rabe in De Theophrasti libris … (Bonn 1890) 19, favoured the second century A.D. because they thought that the language was that of the period of strict Atticism, while others favoured the first century. The last defender of the Phalerean authorship was Hugo Liers in his De aetate et scriptore libri qui fertur Demetrii Phalerei… (Bratislava 1881), while some, notably C. Hammer in his Demetrius … (Landshut 1883), continued to favour an earlier date, about 100 B.C.

This last opinion has more recently been supported by F. Boll in Hermes 72 (1917) 25-33; more recently still, F. Kroll in his article on rhetoric in RE, Suppl. 7 (Stuttgart 1940) 1078-9, has argued for an earlier date, namely the end of the third century B.C. It is clear that the arguments, based on both language and content, which were thought to have settled the matter at the beginning of this century are not as definitive as they then appeared.

²⁷ Dionysius of Halicamassus does indeed have three styles of composition in his Composition, and he classes authors in accordance with these; he also has three styles or types of diction, but the two formulae cannot be telescoped. Isocrates, for example, whose composition is flowery … uses the mean or mixed type of diction. See note 9 (above).

²⁸ See, for example, Ad Herennium 4.16 (ch. 11) and Cicero, Orator 69.

²⁹ Radermacher (p. xii) suggests that the poverty of the treatment of subject-matter is due to the fact that this heading was not dealt with in the books … which our author had before him, and in particular in that of Theophrastus. As we have no knowledge of these "sources" the suggestion cannot be disproved. The poorer treatment may also be due, however, to the fact that the category is not a very sound one, even though it might be traced to Aristotle's dictum that different genres of imitation imitate different objects (Poetics ch. 3.5).

³⁰ For the weakness of this approach to metaphor, which is common to all Greek critics, see W. B. Stanford's stimulating little book, Greek Metaphor (Oxford 1936), especially pp. 6-9.

³¹Rhet. 3.2.14.

³²On the Sublime ch. 3; Horace, On the Art of Poetry 25-31; also Ad Herennium 4.15 (ch. 10).

³³ The translation given here is different from that in the text because there are other figures involved in the Greek passage, which must also be rendered. Here, however, it seemed better to translate the anaphora only, as the other figures are not relevant.

³⁴ See on this point H. Koskennieni's "Studien zur Idee und Phraseologie des griechischen Briefes bis 400 n. Chr." in Annales Academiae Scientiarum Fennicae, B 102 (Helsinki 1956) 18-47. On the general subject of ancient letter-writing see J. Sykutris, "Epistolographie" in RE Suppl. 5 (1931) 185-220. Neither the references to Aristotle's letters nor to one of Thucydides in section 264 (which latter is not likely to have been genuine) should be used as evidence for a later date. The writing of letters in the person of famous men became a literary genre which is often said to have begun in the second century B.C., but for this date there is no evidence. Sykutris suggests that the practice began in the fourth century and points out that there may have been many motives for such writing, quite apart from any intention to deceive. One may add that this would be especially true at a time when the introduction of historical and nearly contemporary personages into dialogues was a well-accepted literary convention. See also, on this point, Ernst Howald's Die Briefe Platons (Zurich 1923) 14-15; J. Harward, The Platonic Epistles (Cambridge 1932) 65-78.

³⁵ Demetrius passes, we might say, from the use of the particular figure of innuendo for a particular purpose (forcefulness) to other uses of it which are not strictly relevant in a discussion of forcefulness, and then from the use of a particular form or figure of speech to different forms in which the same thought may be cast.… A mode m writer would avoid the charge of irrelevance by using footnotes.…

³⁸ It is significant that Demetrius himself, perhaps unconsciously, uses the term "forceful" and "forcefulness," in describing the effect of such witticism at 130 and 131.

⁴⁰ See note on section 30 of the translation and Appendix I, A.…

⁴¹ Demetrius seems to have misunderstood Aristotle. The reference is Rhet. 3.9.5. Aristotle has been criticizing a period he attributes to Sophocles as improperly divided (he seems to take the lines as equivalent to kôla): …

(This country, Calydon, of Pelops' land,
The opposite shore, fertile and happy plains.)

Aristotle's criticism is that if we stop at the end of the first line, the division is haphazard, and does not correspond to sense. He tells us that a complex period must not be divided in this manner.… Aristotle commits himself to periods of two [kola] only. On the other hand, if it refers to the sentence quoted in part, that is, "a colon is either part of this sentence" (which is grammatically preferable), we hardly have the general definition which we should expect. Modern editors seem to gloss over this difficulty (see Frese's Loeb translation p. 289, Cope and Sandys, 3. 97-98, Voilquin and Capelle's Bude edition, p. 343) and generally agree with Demetrius who boldly replaces [tantes] by [phesi]. It is not certain whether Archedemus interpreted the passage as Demetrius does. Even if he did, this does not prove that Demetrius did not have the Rhetoric before him. Modern editors certainly do.

⁴² It may be worth noting that [phesi] does not introduce a direct quotation. As we have the controlling text of Aristotle, no harm is done. But we do not have the text of Theophrastus, for example, and this kind of semi-quotation only too often has been built on excessively.

⁴³ There is an excellent article by Alfred Kappelmacher in Wiener Studien 24 (1902) 452-456, on "Die Aristotelzitate im der Schrift der Pseudo-Demetrius" which shows how Demetrius reproduces the sense rather than the words of Aristotle. In 116 on frigidity, as Kappelmacher points out, the parallel with Rhet. 3.3 is so close that we can with certainty fill in the sense of the lacuna; and yet Demetrius gives the four causes of frigidity in a different order with only one example of each (Aristotle gives many) and one of his examples is not in Aristotle. Kappelmacher also deals with Demetrius 11 and 34 (see note 41) and the "misleading" quotation in 38 (see above pp. 36-37). Even from such an incomplete survey the conclusion is convincingly established.

⁴⁴ As is supposed by F. Solmsen in Hermes 66 (1931) 241-267. Solmsen tries to link the four "styles" of Demetrius with the four virtues of style attributed to Theophrastus, but this too is a conjectural reconstruction of modern scholars, and based on doubtful evidence. See TAPA 83 (1952) 179-181. Solmsen's article, however, contains a useful discussion of Demetrius' indebtedness to Aristotle. Solmsen is right when he says that the four styles of Demetrius are more like qualities than "styles."

We know of course that a number of works on style were written by Peripatetics, but the insistence on intermediate "sources" at all costs seems to assume that no Greek or Roman critic can ever have quoted directly from Aristotle, and that known parallels are always due to some unknown intermediate source. These persistent attempts to explain the known by the unknown seem to me in the present case quite unnecessary.

⁴⁵Rhet. 3.12.2. And there are interesting verbal parallels.…

⁴⁶ It is interesting to note a close parallel to this use of the imperfect in the treatise On the Sublime 3.5, where the author discusses the vice of false enthusiasm "which Theodorus used to call parenthyrsos"—… and that scholars have there argued from the imperfect that "Longinus" must have heard Theodorus and was indeed a younger disciple. See for references my "Theodorus of Gadara" in AJP 80 (1959) 360.

⁴⁷ See OCD, s.v. Sotades and Ptolemy Philadelphus.

⁴⁸ Plutarch, Demosthenes 10. See below, p. 54-55.

⁴⁹ See H. Koskenniemi's "Studien zur Idee und Phraseologie des griechischen Briefes" in Annales Academiae Scientiarum Fennicae, B.102 (Helsinki 1956) 25. He points out that in view of the early collections of the letters of other philosophers, that is, Plato and Epicurus, it is very unlikely that the letters of Aristotle remained unpublished until the second century, and suggests that the Artemon mentioned by Demetrius may well have been a contemporary of Theophrastus.

⁵⁰ By Norden, Kunstprosa 1.148 note I (quoted by Roberts). We may remember in this connection that Quintilian tells us that "fictitious themes in imitation of lawsuits and of deliberative speeches" began in the time of Demetrius of Phalerum, even if these were not as artificial as the Roman type of declamation which he seems to regard as of more recent origin (cp. Quint. 2.4.41 and 2.10.1).

⁵¹ At Rhet. 3.11.15-16. The specific passages said to bear clear traces of the later rhetorical schools (Roberts, p. 251 and Radermacher, pp. 91 and 112) are: the rock of the Cyclops at 115, the Amazon's girdle at 138, the Centaur and the pun on Olympias at 187, the reproach to Aristides at 238, and the obscure phrase … at 239. Radermacher also considers the advice on how to address despots at 292-293 to smack of later declamations, though one would have thought it quite as topical in the third century B.C. Alexander was certainly a frequent topic for suasoriae and the rock of the Cyclops (but not this phrase) is found in Seneca, Suas. 1.12. These were perennial subjects.

⁵²Adipata dictio, Orator 25.

⁵³Nic. Ethics 1128a 22.…

⁵⁴Cratylus 398d. We may also remember Aristophanes' phrase "the Attic look" in Clouds 1176.… The verb is found in the comic writers of the fifth and fourth centuries.

⁵⁵ Lucian, Demonax 41: "Seeing one of the grandees taking pride in the width of his purple …

⁵⁶ We may add one more reference. At 158 Demetrius refers to a legend that "cats grow fatter and thinner as the moon waxes and wanes," and quotes a further elaboration by an unknown writer. Now Plutarch (Isis and Osiris 63, 376f) gives a different version, namely that a cat's pupils grow larger at the full moon. As it seems established that this and other Egyptian tales were first told in the West by Apion in the first century A.D. and that this is Plutarch's source, Demetrius, Radermacher asserts, must have got it indirectly from the same source and got it wrong. But no one suggests that Apion invented the legend, and if our author was writing in Alexandria, or indeed had been in Egypt at all, he might have heard it from anyone, correctly or incorrectly, and so could anyone else. Such arguments surely have little force. I understand from my colleague Professor R. J. Williams that the story is not found in native Egyptian sources.

⁵⁷ See Appendix II.

⁵⁸ As given by Rhys Roberts p. 57.

⁵⁹ Example s will be found in Appendi x I.…

⁶⁰ See Appendi x I.… Rhy s Robert s (p. 229) himself says there are comparatively few such double compounds in our treatise.…

⁶¹ See notes 6 and 13 above.

⁶² See Appendix I, A, where 20 are listed, but some might be added, from other sections, which contain at least semi-technical words.

⁶³ See Appendix I.…

⁶⁴ On the style of Demetrius of Phalerum see Cicero De Oratore 2.95, Orator 92, Brutus 37, and Quintilian 10.1.33 and 80.

⁶⁵ Roberts p. 59; this view seems to be generally accepted.

⁶⁶ As is indeed pointed out by Altschul (p. 31, note 21) but he rather surprisingly takes this as evidence of a later date. It should be noted that Solmsen links the four styles with the four virtues which he traces back to Theophrastus (note 44 above). We first meet the formula of the three styles in the Ad Herennium 4.8-11. It is there clearly stated (ch. II, 16) that every orator must use all styles. So in Cicero (e.g. Orator 20-21, 75-99) who links the three styles with the three officia oratoris (ibid. 69: subtitle in probando, modicum in delectando, vehemens in flectendo).

Dionysius has three kinds … of diction and three kinds of word-arrangement, but these do not correspond (see AJP 73 [1952] 261-267). We do, however, have a passage of Aulus Gellius (6.14) which not only speaks of the three styles and their corresponding vices, but further states that Varro said that Pacuvius was an example of ubertas, Lucilius of gracilitas, and Terence of mediocritas, that is, of the grand, the plain, and the intermediate respectively. As Gellius, however, calls them virtutes as well as genera dicendi, this need not imply any classification according to "styles" in Varro; he may have given these authors as examples of particular qualities only.

The earliest Greek classification by three styles seems to be a doubtful passage in Tryphon… He gives Thucydides and Antiphon as examples of the first, Demosthenes, Hyperides, Demarchus, and Lycurgus as examples of the intermediate, and Aeschines, Isocrates, Lysias, Andocides, and Isaeus as examples of the simple style. The passage is quite irrelevant where it stands, the classification contrary to that of critics like Dionysius; Spengel brackets it.

⁶⁷ Roberts, p. 59. See also note 26 above.

⁶⁸ See my "Theodorus of Gadara" in AJP 80 (1959) 337-365.… In any case the main preoccupations of "Longinus" are quite different from those of Demetrius.

⁶⁹ See sections 77, 158, 172. We may add 97, for Boll (Rhein. Mus. 72 [1917] 27-28) gives good reason to think that the word [skafites] is of Egyptian origin.

⁷⁰ See Fritz Wehrli, Demetrius von Phalerum, Die Schule des Aristoteles 4 (Basle 1949).

⁷¹ Diogenes Laertius 5.80-81 = Wehrli 74.

⁷² The meaning of [kairos] and [kairios] in criticism comes very close to that of [prepon], that is, appropriateness to the occasion. Dionysius complains in Composition 12 that no rhetorician or philosopher has written a treatise on the subject since Gorgias first dealt with it. For an interesting, if somewhat fanciful, discussion of [kairos] in Gorgias see W. Vollgraff, L'Oraison funèbre de Gorgias (Leiden 1952) 21-27.

⁷³ As R. D. Hicks translates these titles in his Loeb version of Diogenes.

⁷⁴ Plutarch, Reg. et Imp. Apopth. 189d, Stobaeus 4.7.27 = Wehrli 63.

⁷⁵Rhetorica IV, Col. XIP.… There can be little doubt that Philodemus is referring to Demetrius of Phalerum, as Wehrli and others have taken him to be.

⁷⁶ So Hubbell, "The Rhetorica of Philodemus," Trans. Connecticut Acad. Arts and Sciences 23 (1920) 304, note. "Obtaining favor with all" is Hubbell's translation.…

⁷⁸Rhetorica IV. Col. XVIa.…

⁷⁹ The point seems to have escaped Roberts who says (p. 60): "it is to be noticed that Philodemus speaks vaguely of 'Demetrius' without any addition, and so may, or may not, have Demetrius Phalereus in mind." Hammer (p. 60) admits the reference is to our treatise which was therefore known to Philodemus, but he does not make the connection with Phalereus. A look at Sudhaus' text makes clear that there is no possible doubt that the two passages go together.

⁸⁰ Besides the passage just referred to from Philodemus, see also Plutarch, Demosthenes 9 and 11 and other references in Wehrli 162-169.

⁸¹ See Appendix I, A.…

⁸² It is true that "Longinus," On the Sublime (33-35), recognises that Demosthenes lacks many virtues which Hyperides possesses, but this is meant to show that those he does possess are by far more important, in a passage which insists that faulty genius is to be preferred to flawless mediocrity.

⁸³In Arist. de Interpretatione, Berlin edition, 4. 4-5, 966-997a. See Appendix I, A.…

⁸⁴ Quoted by Rhys Roberts, pp. 60-61.

⁸⁵ p. 23.