INTRODUCTION

I
Summary of the de Compositione

A general account of the life and literary activities of Dionysius will be found in the volume entitled Dionysius of Halicarnassus: the Three Literary Letters, where the de Compositione Verborum is briefly described in connexion with the other critical essays of its author. Here a fuller summary of the treatise seems necessary before an attempt is made to estimate its value and to follow up some of the highly interesting questions which it raises.

The date of the de Compositione is not known, but may be conjectured to lie between the years 20 and 10 B.C. The book is a birthday offering from Dionysius, as a teacher of rhetoric in Rome, to his pupil Rufus Metilius.

c. 1. This book is a birthday present which deals with the art of speech, and so will be found particularly useful to youths who look forward to a public career. Oratorical excellence depends on skill exercised in two directions—in the sphere of subject matter and in the sphere of expression (πραγματικὸς τόπος and λεκτικὸς τόπος). In the former sphere, maturity of judgment and experience is required: in the latter the young are more at home, but they need careful guidance at the start. The λεκτικὸς τόπος has two subdivisions, ἐκλογὴ ὀνομάτων and σύνθεσις ὀνομάτων. The composition of words is to be treated now: the choice of words is to be treated next year, if Heaven keeps the author “safe and sound.” The chief headings in the present treatise are to be the following:—

(1) The nature of composition, and its effect;

(2) Its aims, and how it attains them;

(3) Its varieties, with their characteristic features and the author’s preferences among them;

(4) The poetical element in prose and the prose element in verse, and the means of cultivating both—of imparting the flavour of poetry to prose and the ease of prose to poetry.

c. 2. “Composition is, as the very name indicates, a certain mutual arrangement of the parts of speech, or elements of diction, as some prefer to call them.” The parts of speech recognized by Theodectes and Aristotle and their contemporaries were three in number, viz. nouns, verbs, and connectives. The number was raised, by the Stoics and others, to four through the separation of the article from the connectives. Later were added the adjective, the pronoun, the adverb, the preposition, the participle, and certain other subdivisions. These principal parts of speech form, when joined and set side by side, the cola (‘members,’ ‘clauses’). The union of cola completes the “periods,” and these make up the entire discourse. The functions of composition are to arrange the words fittingly, to assign the proper structure to the cola, and to divide the discourse carefully into periods.

In its effects, though not in order of time, the composition of words comes before the choice of words.

c. 3. Our thoughts are uttered either in verse or in prose. In both alike, composition can invest the lowliest words with charm and distinction. By way of foretaste, two passages (one of poetry, the other of prose) may be quoted in illustration. The first is from the opening of the 16th Odyssey, where the lines allure not by elaborate language or lofty theme, but by the sheer beauty with which the words are grouped. The prose example is furnished by that passage of Herodotus (i. 8-10) which describes the unworthy behaviour of Candaules towards his wife. Here, too, the charm resides not in the incident nor in the words which describe it, but in the deft arrangement of the language.

c. 4. The powerful effect of composition will be still further realized if some choice passages of verse and prose be taken and the order of the words disturbed. Homer and Herodotus once more provide examples. Certain lines in the twelfth and thirteenth books of the Iliad are chosen, and transformed, with disastrous effects, from hexameters into two varieties of tetrameters. A short passage of Herodotus is turned about in a similar way, one of the two versions being in the style of Thucydides, the other in the odious manner of Hegesias. Composition may in fact be likened to the Homeric Athena, who with a touch of her magic wand could make the same Odysseus resemble either a beggar or a gallant prince. The neglect of composition has lamentable results in writers like Duris, Polybius, Chrysippus, and others. Failing to find the subject satisfactorily treated by previous authors, Dionysius has himself endeavoured to discover some natural principle to form a starting-point (φυσικὴ ἀφορμή). He has not succeeded, but he will describe his attempt.

c. 5. It had occurred to him that, in a natural order, verbs would follow nouns and precede adverbs, while things which happened first in time would come first in narration. But these (and other) rules were seen to be untrustworthy, when tested by the actual practice of the great authors.

c. 6. As far as words (or elements of discourse) are concerned, the art of composition operates in three ways—through (1) the choice of elements likely to combine effectively; (2) the discernment of the particular shapes or constructions (i.e. singular or plural number, nominative or oblique case, active or passive voice, etc.) to be given to each element in order that the structure may be improved; (3) the perception of the modification which these shapes need in view of the materials. Each of the processes can be illustrated from the arts of house-building and ship-building—of civil and marine architecture. This analogy is developed at some length.

c. 7. In the case of the cola, the processes are two. (1) The cola must be rightly arranged. For instance, in a passage of Thucydides (iii. 57) the order in which they come makes all the difference. So, too, in Demosthenes de Corona § 119.

c. 8. (2) The right “turn,” or “shaping,” must be given to the cola, so that they may faithfully reflect the various aims and moods of the speaker or writer. A good example will be found in Demosthenes de Corona § 179.

c. 9. Under (2) it is to be noted that the cola may be lengthened or shortened for the sake of literary effect. Examples are given from Demosthenes, Plato, Sophocles, and again Demosthenes.—The same remarks will apply to periods as to cola. Further, the art of composition must determine when it is fitting to employ periods and when not.

c. 10. Next come the aims and methods of good composition. The two chief aims are charm and beauty or nobility: the ear craves these in composition, just as the eye in a work of pictorial art. The two qualities are, however, not identical. Thucydides, for example, and Antiphon possess beauty but lack charm. Ctesias, on the other hand, and Xenophon are charming (pleasing, agreeable), but deficient in beauty. Herodotus combines the two excellences.

c. 11. The chief sources of charm and beauty (or nobility) are four: music, rhythm, variety, and propriety. Charm and beauty, themselves, have many subdivisions. The instinctive appreciation of music and rhythm on the part of a popular audience may be noticed during a performance in some house of entertainment. Variety, too, and propriety are indispensable. As to the music of speech, it is to be observed that there is a sort of oratorical cadence which differs from music proper in quantity only, not in quality. The speaking voice does not rise in pitch above three tones and a half: it confines itself to the interval of the Fifth. The singing voice, on the other hand, uses a greater number of intervals, not only the Fifth but (beginning with the Octave) the Fifth, the Fourth, the Tone, and the Semitone, and, as some think, still slighter intervals. Other points of difference are that, in singing, the words are subordinate to the air, and the length of the syllables is regulated by the musical time. So the speaking voice can show good melody without being “melodic,” and show good rhythms without being “rhythmic.” There is, in fact, music in speech, but not the whole of music.

c. 12. Various sounds affect the ear in various ways. The cause lies in the nature of the letters; and as their nature cannot be changed, there should be a judicious intermixture of pleasant with unpleasant sounds. Short words, too, must be mingled with long, and long with short. The same variety, too, must be practised in the use of figures, and in other ways. But even variety must not be carried to excess: uniformity is sometimes equally pleasant. Tact is needed, and to impart tact is no easy task. It is to be remembered that not even the commonest words need be shunned by good writers: they can all be dignified by means of composition, as is seen in Homer’s poems.

c. 13. Beauty of composition will be attained by the same means as charm of composition,—by melody, rhythm, variety, propriety. And the nature of the letters themselves will play an equal part in determining the character of the composition.

c. 14. The twenty-four letters of the Greek alphabet are now examined from the phonetic point of view. The object is to trace to some of its ultimate elements the secret of the variety and music found in beautiful language. The nature and the qualities of the letters must be understood by the writer who would know how to vary his style in an ever-changing and musical way. The letters (γράμματα), or elements (στοιχεῖα), may be divided into vowels (φωνήεντα, φωναί) and consonants (ψόφοι), and the consonants into semivowels (ἡμίφωνα) and mutes (ἄφωνα). The vowels can be pronounced by themselves; the semivowels sound best when combined with vowels; the mutes cannot be uttered at all except in combination. There are seven vowels: two short, ε and ο; two long, η and ω; and three common,—α, ι, and υ. The semivowels are eight in number: five single, viz. λ, μ, ν, ρ, ς, and three double, viz. ζ, ξ, ψ. The nine mutes may be classified as: ψιλά (tenues) κ, π, τ; δασέα (aspiratae) χ, φ, θ; and μέσα (mediae) γ, β, δ. Or they may be arranged according to the part chiefly concerned in their production: whether it is the lip,—π, φ, β; the teeth,—τ, θ, δ; or the throat,—κ, χ, γ. That is to say, Dionysius recognizes (though he does not use the technical adjectives) a division into labials, dentals, and gutturals. Among these various letters a regular hierarchy is established by him. Long vowels are held to be more euphonious than short vowels. The order of euphony for the vowels is, from the top downwards, as follows: ᾱ, η, ω, υ, ι, ο, ε; and (for the semivowels) first the double consonants, then λ, μ, ν, ρ, and lastly ς, which is condemned in strong terms. Among the mutes, the rough (the aspirates) are regarded as superior to the middle, and the middle to the smooth. The physiological processes by which the several letters are produced are described with some particularity in the light of the phonetics of the day.

c. 15. Syllables, as well as letters considered singly, contribute to variety of style. Of the syllables (or small groups of letters) there are many different kinds. The principal difference is that some are short and others long. But the difference does not end there, since some are shorter than the short and others longer than the long. The fact is that, from the metrical point of view, the vowels and final consonants alone count in determining the length of a syllable, whereas in actual delivery the initial consonants also have to be considered. For instance, a speaker will find that the initial syllable of στρόφος takes more time to utter than that of τρόπος; and so with τρόπος by the side of Ῥόδος, and with Ῥόδος by the side of ὁδός. In the same way, σπλήν is really longer than the vowel η standing by itself. And further: syllables differ not only in quantity but in sound, some being pleasant and others unpleasant, according to the nature of the letters which compose them. Great poets and prose-writers have an instinctive perception of these facts, and skilfully adapt their very syllables and letters to the emotions which they wish to portray; e.g. Homer in Odyss. ix. 415, 416, and in Il. xvii. 265, xxii. 220, 221, 476, xviii. 225.

c. 16. Poets and prose-writers frame, or borrow from their predecessors in earlier generations, such imitative forms (words whose sound suggests their sense) as ῥοχθεῖ, κλάγξας, βρέμεται, σμαραγεῖ, ῥοῖζος: all of which are found in Homer. Nature is here the great teacher; she prompts us to use, in their right connexion, words so expressive as μύκημα, χρεμετισμός, φριμαγμός, βρόμος, πάταγος, συριγμός, and the like. The first writer to broach the subject of etymology was Plato, particularly in his Cratylus.

With regard to the music of sounds, the general conclusion is that variety and beauty of style depend upon variety and beauty of words, syllables, and letters. To clinch the matter, Dionysius quotes (with appropriate comments) further illustrations from Homer—Odyssey xvii. 36, 37, vi. 162, 163, etc. Theophrastus, in his work on Style, has distinguished two classes of words—those which are beautiful (or noble) and those which are mean and paltry. Our aim should be to intermingle the latter kind, when we are forced to employ them (as sometimes we are), with the better sort, as has been done by Homer (Il. ii. 494-501) in his enumeration of the Boeotian towns.

c. 17. Rhythm, also, is an important element in good composition. For our present purpose, a rhythm and a foot may be regarded as synonymous. Of disyllabic and trisyllabic feet the following descriptive list is given:—


	A. Disyllabic Feet.
Name.	Quantities.	Qualities.
1. ἡγεμών, πυρρίχιος.	ᴗ ᴗ	Wanting in seriousness and dignity.
2. σπονδεῖος.	– –	Full of dignity.
3. ἴαμβος.	ᴗ –	Not lacking in nobility.
4. τροχαῖος.	– ᴗ	Less manly and noble than the iambus.
	B. Trisyllabic Feet
Name.	Quantities.	Qualities.
1. χορεῖος, τρίβραχυς.	ᴗ ᴗ ᴗ	Mean and unimpressive.
2. μολοττός.	– – –	Dignified and far-striding.
3. ἀμφίβραχυς.	ᴗ – ᴗ	Effeminate and unattractive.
4. ἀνάπαιστος.	ᴗ ᴗ –	Stately.
5. δάκτυλος.	– ᴗ ᴗ	Contributes greatly to beauty of style.
6. κρητικός.	– ᴗ –	Not lacking in nobility.
7. βακχεῖος.	– – ᴗ	Virile and grave.
8. ὑποβακχεῖος.	ᴗ – –	Virile and grave.

Various lines are quoted from the poets in order to illustrate the effect of these several feet.

c. 18. As each word has a rhythmical value (great or small) which cannot be changed, all depends on the skill with which we arrange the words at our disposal so as to blend artistically the inferior with the better. To illustrate his meaning, Dionysius quotes, and gives a rhythmical analysis of, passages from Thucydides, Plato, and Demosthenes. The excerpt from Thucydides is a part of the Funeral Oration attributed to Pericles (ii. 35). The rhythms here used are shown to be dignified ones, such as spondees, anapaests, dactyls, etc. Thucydides, we are told, deservedly has a name for elevation and for choice language, since he habitually introduces noble rhythms. From Plato is taken a short passage of the Menexenus (236 D); and this too is shown to owe its dignity and beauty to the beautiful and striking rhythms that compose it. If Plato had only been as clever in the choice of words as he is unrivalled in the art of combining them, he “had even outstript” Demosthenes, as far as beauty of style is concerned, or “had left the issue in doubt.” Demosthenes is the foremost of orators, and may be regarded as a model alike in his choice of words and in the beauty with which he arranges them. The opening of the Crown, with its careful avoidance of all ignoble rhythms, will prove his pre-eminence. Deficiency in this respect can be illustrated just as conspicuously by the writings of Hegesias, who would seem to have shunned good rhythms out of sheer wilfulness. A passage is quoted from Hegesias’ History—a passage which, if well written, would have moved to sympathetic tears rather than to derisive laughter. With it are contrasted some famous lines of the Iliad (xxii. 395-411) which, we are told, owe their nobility largely to the beauty of their rhythms.

c. 19. The third element in good composition is variety (ἡ μεταβολή). In the use of rhythms to impart variety, prose enjoys much greater freedom than poetry. Epic poets must needs employ the hexameter line: the writers of lyric verse must make antistrophe correspond to strophe, however greatly they may strive for liberty in other respects. That prose style is best which exhibits the greatest variety in the way of periods, clauses, rhythms, figures, and the like; and its charm is all the greater if the art that fashions it lies hidden. In point of variety, Herodotus, Plato and Demosthenes hold the foremost place: Isocrates and his followers are distinguished rather by monotony of style.

c. 20. The fourth element is fitness or propriety (τὸ πρέπον). Propriety is described as the harmony which an author establishes between his style, and the actions and persons of which he treats. Common experience proves that ordinary people, in describing an event, will vary the order of their words (and the point here is the arrangement, not the choice of words) in accordance with the emotions which it excites in them. Similarly, artistic writers should follow their own aesthetic instincts in the matter. Homer has done so with surpassing effect. A fine instance is furnished by the lines (Odyssey xi. 593-598) which depict the torment of Sisyphus—the slow upheaval of his rock, and its rapid rolling down the hill once it has reached the top.

c. 21. After these theoretical and technical discussions there arises the question: what are the different kinds of composition or arrangement,—what are the different harmonies? The answer given is that there are three: (1) the austere (αὐστηρά), (2) the smooth (γλαφυρά), (3) the harmoniously blended (εὔκρατος) or intermediate (κοινή).

c. 22. The characteristic features of austere composition are set forth in considerable detail: both generally and in reference to words, clauses, periods. Among its principal representatives are mentioned: Antimachus of Colophon and Empedocles in epic poetry, Pindar in lyric, Aeschylus in tragic; in history, Thucydides; in oratory, Antiphon. The beginning of a Pindaric dithyramb and the opening sentences of the introduction to Thucydides’ History are minutely examined from this point of view. [Any attempt to summarize fully this chapter and those which follow is hardly possible owing to the nature of the subject matter. The chapters are important, and will repay a careful study.]

c. 23. Smooth composition is next characterized in a similar way. Its chief representatives may be taken to be: Hesiod, Sappho, Anacreon, Simonides, Euripides, Ephorus, Theopompus, Isocrates. In illustration are quoted (with sundry comments) Sappho’s Hymn to Aphrodite and the introductory passage from Isocrates’ Areopagiticus.

c. 24. “The third, the mean of the two kinds already mentioned, which I call harmoniously blended (or intermediate) for lack of a proper and better name, has no form peculiar to itself, but is a judicious blend of the other two and a selection from the most effective features of each.” This third is the best variety of composition because it is a kind of golden mean; and its highest representative is Homer, in whom we find a union of the severe and the polished forms of arrangement. On a lower plane are other votaries of the golden mean: among lyric poets Stesichorus and Alcaeus, among tragedians Sophocles, among historians Herodotus, among orators Demosthenes, and among philosophers Democritus, Plato and Aristotle. Illustrative examples are, in this case, unnecessary.

c. 25. These discussions lead up to a final question,—that of the relations between prose and poetry. And first: in what way can prose be made to resemble a beautiful poem or lyric? It is in metre, even more than in the choice of words, that poetry differs from prose. Consequently prose cannot become like metrical and lyrical writing, unless it contains, though not obtrusively, metres and rhythms within it. It must not be manifestly in metre or in rhythm (for in that case it will be a poem or a lyric and will desert its own specific character), but it is enough that it should simply appear rhythmical and metrical. It will thus be poetical, although not a poem; lyrical, although not a lyric. Passages are then taken from the opening of the Aristocrates and the Crown of Demosthenes and are subjected to a minute metrical analysis. The result of the scrutiny is (it is claimed) to show that many metrical lines are latent in good prose, the author having taken care to disguise slightly their metrical character. In an eloquent passage Dionysius then submits that the great end in view warranted all these anxious pains on the part of Demosthenes. Demosthenes was no mere peddler, but a consummate artist who had the judgment of posterity always before his mind. Isocrates, also, and Plato spent no less trouble on their writings, as witness the story about the opening passage of the Republic. It is, further, to be noticed that such careful processes, though deliberate at first, become in the end unconscious and almost instinctive, just as accomplished musicians do not think of every note they strike on their instrument, nor skilled readers of every single letter which meets their eyes in the book that lies open before them.

c. 26. Secondly (and lastly) comes a question which is the counterpart of that asked in c. 25: namely, in what way can a poem or lyric be made to resemble beautiful prose? The two principal means are: (1) so to arrange the clauses that they do not invariably begin and end together with the lines; (2) to vary the clauses and periods in length and form. These things are more difficult to do where the metre is uniform, as in heroic and iambic verse. In lyric poems the task is easier, since the variety of their metres brings them a point nearer to prose. At the same time, while avoiding monotony and while generally causing his verse to resemble beautiful prose, the poet must remember that the so-called “prosaic character” is a defect. We are, however, here thinking not of vulgar prose but of the highest civil oratory. In order to show that, in poetry, clauses can be of different sorts and sizes, and can also be so far independent of the metre as almost to give the effect of an unbroken prose-narrative, Dionysius draws some concluding illustrations from the 14th Odyssey, the Telephus of Euripides, and the Danaë of Simonides.

The following Tabular Analysis may help to make the general structure of the treatise still clearer:—

I. Chapters 1-5. Introductory. The nature of composition, and its effect.—Instances of the fatal neglect of composition.—The secret of composition not to be found in grammatical rules.

II. Chapters 6-20. General Theory and Technique of Composition:—

1. cc. 6-9:

(α) Three processes in the art of composition, c. 6.
(β) Grouping of clauses, c. 7.
(γ) Shaping of clauses, c. 8.
(δ) Lengthening and shortening of clauses and periods, c. 9.

2. cc. 10-20: Charm and beauty of composition, and the four means of attaining these qualities:—

(α) Preliminary remarks, cc. 10-13.
(β) Four means:
(1) μέλος, cc. 14-16.
(2) ῥυθμός, cc. 17, 18.
(3) μεταβολή, c. 19.
(4) τὸ πρέπον, c. 20.

III. Chapters 21-24. Three Modes of Composition:—

(1) σύνθεσις αὐστηρά, c. 22.
(2) σύνθεσις γλαφυρά, c. 23.
(3) σύνθεσις εὔκρατος (or κοινή), c. 24.

IV. Chapters 25, 26. Relation of Prose to Poetry, and of Poetry to Prose.

Note.—The existing division into chapters is not always a happy one. As a help to the reader, a few words of summary have been prefixed to each chapter of the English Translation.

The Greek Epitome is about one-third the length of the original. It is of early but uncertain date (cp. Usener de Dionysii Halicarnassensis Libris Manuscriptis p. viii, n. 7), and is preserved in the following codices: Darmstadiensis, Monacensis, Rehdigeranus, Vaticanus Urbinas. It has survived along with the original; and instead of superseding and extinguishing the unabridged work, as ancient epitomes seem often to have done, it contributes not a little to its elucidation. Had it been preserved at the expense of the original, we should have still possessed the Sappho, but should have lost the Simonides. Towards the end, the Epitome is executed with less care than at the beginning.

II
The Order of Words in Greek

The strong and the weak points of the de Compositione Verborum will appear from the foregoing summary, and still more from the treatise itself and the notes appended to it. Dionysius’ book is unique: no other of its kind has come down to us from classical antiquity. Its immediate subject is the Order of Words in Greek. But its author is happily led to raise fundamental questions such as the relations between Prose and Poetry, together with incidental points of Greek Pronunciation and Accentuation; and generally to take so wide a range that no English title less comprehensive than On Literary Composition seems to fit the contents of the work.[3] The discursive enthusiasm of the writer is obvious. Not less striking, however, is the sound literary taste which converts his quotations into a true anthology and preserves some priceless remains of Sappho and Simonides. It will be necessary to point out certain weaknesses of Dionysius from time to time. But his weaknesses are far more than counterbalanced by his great excellences. Some of his shortcomings are those of his age,—an age which was a stranger to the modern method of comparison as applied to literary investigation. Others, again, are more apparent than real. When, for example, certain omissions are observable in some directions along with ample expatiations in others, it is to be remembered (1) that Dionysius is dealing with the department of expression and not with that of subject matter, (2) that, in the department of expression, he is concerned with the composition (or arrangement) of words and not with their selection, and (3) that, in regard to composition, he is here interested primarily not in lucidity nor in emphasis, but in euphony. Hence we must not expect him to dwell on that great governing principle of literary composition,—logical connexion. To its importance, however, he is fully alive, as is clear from a passage in his essay on Isocrates: “The thought” [in Isocrates, who pays excessive heed to smoothness of style and a pleasant cadence] “is often the slave of rhythmical expression, and truth is sacrificed to elegance.... But the natural course is for the expression to follow the ideas, not the ideas the expression.”[4] And though, in the de Compositione, it is his business to discourse rather upon sound than upon sense, yet the orderly way in which the subject matter of the treatise is presented shows in itself that Dionysius was well aware that the chief essential for a book is a basis of clear thinking and broad logical arrangement, and that, as a consequence, its excellence is to be sought even more in its chapters and its paragraphs than in its flowing periods.[5] It may be well to touch, with a similar regard to sequence and with occasional references to modern parallels or contrasts, upon one or two aspects of his main theme which his own treatment of it suggests as suitable for further discussion and elucidation.

A. Freedom and Elasticity

In his fifth chapter Dionysius shows, with no difficulty and with much vivacity, that it is impossible to lay down universal rules governing the order of words in Greek. He admits that he had been inclined to entertain a priori views on the question of the natural precedence of certain parts of speech and to hold that nouns should precede verbs, verbs adverbs, and so forth.[6]

But he had proceeded, with that sound practical judgment which distinguishes him, to test his theories in the light of Homer’s usage. He had then found them wanting. “Trial invariably wrecked my views and revealed their utter worthlessness.” The examples of variety in word-order which he quotes from the Iliad and the Odyssey are most interesting and instructive. But a modern reader, familiar with languages whose paucity of inflexions often offers freedom only at the price of ambiguity, has more cause than any ancient writer to wonder at the liberty which Greek enjoys in this respect. No doubt the long gap between πολὺν and χρόνον in the Frogs has, and is intended to have, a comic effect. But there is no sort of ambiguity in the sentence, since the poet takes care to use no noun with which the adjective could agree until the right noun at length comes and relieves the listener of his suspense and growing curiosity,—

εἰ δ’ ἐγὼ ὀρθὸς ἰδεῖν βίον ἀνέρος ἢ τρόπον ὅστις ἔτ’ οἰμώξεται,
οὐ πολὺν οὐδ’ ὁ πίθηκος οὗτος ὁ νῦν ἐνοχλῶν,
Κλειγένης ὁ μικρός,
ὁ πονηρότατος βαλανεὺς ὁπόσοι κρατοῦσι κυκησιτέφρου
ψευδολίτρου κονίας
καὶ Κιμωλίας γῆς,
χρόνον ἐνδιατρίψει.

Aristophanes Ranae 706-13.

Here as many as twenty-one words divide an adjective from its noun, though noun and adjective are usually placed close together.[7] But, even in serious poetry, the same thing is to be noticed, though on a less surprising scale. For example:

ἦν δ’ οὐδὲν αὐτοῖς οὔτε χείματος τέκμαρ
οὔτ’ ἀνθεμώδους ἦρος οὔτε καρπίμου
θέρους βέβαιον.

Aeschylus Prometheus Vinctus 454-6.

Here the adjective follows the noun, but (as before) there is no ambiguity, though there is much added emphasis due to the apparent afterthought. Similarly:

ἐν δὲ νομὸν ποίησε περικλυτὸς ἀμφιγυήεις
ἐν καλῇ βήσσῃ μέγαν οἰῶν ἀργεννάων.[8]

Homer Iliad xviii. 587, 588.

And in prose the dependence of a genitive may be quite clear, though the distance between it and the words on which it depends be great: e.g.

τῶν μὲν οὖν λόγων, οὓς οὗτος ἄνω καὶ κάτω διακυκῶν ἔλεγε περὶ τῶν παραγεγραμμένων νόμων, οὔτε μὰ τοὺς θεοὺς οἶμαι ὑμᾶς μανθάνειν οὔτ’ αὐτὸς ἐδυνάμην συνεῖναι τοὺς πολλούς.

Demosthenes de Corona § 111 (cp. § 57).

In prose, again, the extremely antithetic and artificial arrangement of words possible (without complete loss of clearness) in a highly inflected language may be illustrated from Thucydides:—

καὶ οὐ περὶ τῆς ἐλευθερίας ἄρα οὔτε οὗτοι τῶν Ἑλλήνων οὔθ’ οἱ Ἕλληνες τῆς ἑαυτῶν τῷ Μήδῳ ἀντέστησαν, περὶ δὲ οἱ μὲν σφίσιν ἀλλὰ μὴ ἐκείνῳ καταδουλώσεως, οἱ δ’ ἐπὶ δεσπότου μεταβολῇ οὐκ ἀξυνετωτέρου, κακοξυνετωτέρου δέ.

Thucydides vi. 76.[9]

The following sentence of Demosthenes, with its carefully chosen position for the main subject Φίλιππος and the main verb ἐπηγγείλατο, shows how well suspense and the period can be worked in such a language:—

ὡς δὲ ταλαιπωρούμενοι τῷ μήκει τοῦ πολέμου οἱ τότε μὲν βαρεῖς νῦν δ’ ἀτυχεῖς Θηβαῖοι φανεροὶ πᾶσιν ἦσαν ἀναγκασθησόμενοι καταφεύγειν ἐφ’ ὑμᾶς, Φίλιππος, ἵνα μὴ τοῦτο γένοιτο μηδὲ συνέλθοιεν αἱ πόλεις, ὑμῖν μὲν εἰρήνην ἐκείνοις δὲ βοήθειαν ἐπηγγείλατο.

Demosthenes de Corona § 19.[10]

In an analytical language such as English a separate introductory sentence [11] would be almost necessary in order to bring out the point of a familiar passage in the Cyropaedia:—

παῖς μέγας μικρὸν ἔχων χιτῶνα ἕτερον παῖδα μικρὸν μέγαν ἔχοντα χιτῶνα, ἐκδύσας αὐτόν, τὸν μὲν ἑαυτοῦ ἐκεῖνον ἠμφίεσε, τὸν δὲ ἐκείνου αὐτὸς ἐνέδυ.

Xenophon Cyropaedia i. 3. 17.

And the force and variety gained by juxtaposition, or by chiastic arrangement, is obvious in such examples as:—

(1) τίπτε με, Πηλέος υἱέ, ποσὶν ταχέεσσι διώκεις,
αὐτὸς θνητὸς ἐὼν θεὸν ἄμβροτον;

Homer Iliad xxii. 8, 9.

(2) τί δῆτα, ὦ Μέλητε; τοσοῦτον σὺ ἐμοῦ σοφώτερος εἶ τηλικούτου ὄντος τηλικόσδε ὤν;

Plato Apology 25 D.

(3) οὐ γὰρ ἐπὶ κρίσει μέν τις δικασθεὶς οὐκ ἂν ἐπὶ τῶν δικαίων καὶ καλῶν ἐλεύθερος καὶ ὑγιὴς ἂν κριτὴς γένοιτο· ἀνάγκη γὰρ τῷ δωροδόκῳ τὰ οἰκεῖα μὲν φαίνεσθαι καλὰ καὶ δίκαια.

Longinus de Sublimitate c. xliv.

(4) καὶ τῶν κώλων ... ἀνίσων τε ὄντων καὶ ἀνομοίων ἀλλήλοις ἀνομοίους τε καὶ ἀνίσους ποιούμενοι τὰς διαιρέσεις.

Dionys. Halic. de Comp. Verb. c. xxvi.

The two last examples of elegant variation might, no doubt, be closely reproduced in modern languages. To the more important matter of emphasis, which arises in some of the other instances, a separate section must be devoted later.[12]

B. Normal Order

Though Dionysius does right to deny the existence of a natural or inevitable order in Greek and to emphasize the essential freedom of the language, he might well have recognized more explicitly that there is what may be termed a normal or usual order, and that it is precisely the departure from this normal usage which does much to give a definite character (good or bad, as the case may be) to the style of individual Greek authors. For instance, it is usual in Greek for an adjective to follow its noun, and for a negative to precede the word or words which it qualifies. There are, further, certain customary positions for the article (according as it is attributive or predicative); for the demonstrative pronouns in conjunction with the article; for αὐτός, according to the meaning which it bears; for the particles; for prepositions, conjunctions, and relative pronouns; and so forth. There is, in short, a grammatical order sanctioned by prevailing usage, an order which might be shown to hold good, commonly though not universally, in some of the grammatical constructions indicated by Dionysius in his fifth chapter. Now between this normal order, and lucidity of expression, there exists a close connexion.

C. Lucidity

It might easily be concluded, by a reader who knew the de Compositione alone among Dionysius’ critical essays, that he set little store by that clear writing which, as it presupposes clear thinking, is a rare and cardinal excellence of style. As the noun σαφήνεια occurs but once in the treatise and the adjective σαφής not much oftener, it might be supposed that he underrated a quality to which Aristotle and other writers of antiquity assign so high a place. Aristotle, indeed, regards it as a first essential of good style, which must be “clear without being mean” (λέξεως δὲ ἀρετὴ σαφῆ καὶ μὴ ταπεινὴν εἶναι, Aristot. Poet. xxii. 1: cp. Rhet. iii. 2. 1). Similarly Cicero puts clearness (sermo dilucidus) before ornament, asking how it is possible, “qui non dicat quod intellegamus, hunc posse quod admiremur dicere” (Cic. de Orat. iii. 9. 38). Horace’s approving reference to lucidus ordo has become proverbial.[13] And Quintilian allots the primacy to the same great quality: “nobis prima sit virtus perspicuitas, propria verba, rectus ordo, non in longum dilata conclusio; nihil neque desit neque superfluat” (Inst. Or. viii. 2. 22), and puts a high and not always attainable ideal before the orator in relation to his judicial auditor: “quare non, ut intellegere possit, sed, ne omnino possit non intellegere, curandum” (ibid. viii. 2. 24).

If Dionysius in the present treatise says little about lucidity, the sole reason is that he assumes it as a necessary and indispensable quality of style. In the de Thucydide c. 23 it is classed (together with purity and brevity) as one of the ἀρεταὶ ἀναγκαῖαι (in contradistinction to the ἀρεταὶ ἐπίθετοι, such as ἐνάργεια, ἡ τῶν ἠθῶν τε καὶ παθῶν μίμησις, etc.). The Greek critics recognized, however, that the plainer styles were more likely than the more elaborate ones to excel in lucidity,—that, in this respect, a Herodotus and a Lysias might be expected to surpass a Thucydides and a Demosthenes.[14] Among these authors let us choose Lysias and Thucydides, and see what praise or blame Dionysius awards to them upon this score. In the fourth chapter of the de Lysia, the lucidity of Lysias is contrasted with the obscurity often found in Thucydides and Demosthenes; and it is pointed out that this excellence is, in him, all the more admirable in that it is combined with a studious brevity, an opulent vocabulary, and a mind of great native force. And no finer example of pellucid clearness of narration could well be imagined than that quoted from Lysias in the sixth chapter of the de Isaeo: ἀναγκαῖόν μοι δοκεῖ εἶναι, ὦ ἄνδρες δικασταί, περὶ τῆς φιλίας τῆς ἐμῆς καὶ τῆς Φερενίκου πρῶτον εἰπεῖν πρὸς ὑμᾶς, κτλ. To the obscurities of Thucydides, on the other hand, as seen in his History and particularly in his Speeches, constant and mournful reference is made in the essay which has the historian for its subject. “You can almost count on your fingers,” says Dionysius, “the people who are capable of comprehending the whole of Thucydides; and not even they can do so without occasional recourse to a grammatical commentary.”[15] Dionysius, further, gives it as his opinion that the language of Thucydides was unique even in his own day; and he combats the view that a historian (as distinguished, say, from an advocate) may plead in excuse for an artificial style that he does not write for “people in the market-place, in workshops or in factories, nor for others who have not shared in a liberal education, but for men who have reached rhetoric and philosophy after passing through a full curriculum of approved studies, to whom therefore none of these expressions will appear unfamiliar.”[16] Obscurity and eccentricity, he says in effect, are not virtues except in the eyes of literary coteries; presumably a speaker speaks, and a writer writes, in order to be understood.[17]

D. Emphasis

Dionysius’ inadequate recognition of a normal order is naturally attended by some uncertainty in his attitude towards that kind of emphasis which a departure from the normal order produces. It may, indeed, be thought that the effect of emphasis, and the best means of attaining it, are considered at the opening of the sixth chapter of the treatise, and that it comes under the heading both of σχηματισμός and of ἁρμονία. In the fifth chapter, however, we should have welcomed a clearer recognition of the emphasis which, as it seems to modern readers, falls upon ἄνδρα, μῆνιν, and ἠέλιος, when they come at the beginning of the line and so are the first words to accost the ear. Certainly in his own writing Dionysius shows that he appreciates the emphasis gained by thrusting a word to the front of the sentence: e.g. καιροῦ δὲ οὔτε ῥήτωρ οὐδεὶς οὔτε φιλόσοφος εἰς τόδε χρόνου τέχνην ὥρισεν ([132] 21). Towards the end of chapter 7 he quotes from Demosthenes the words τὸ λαβεῖν οὖν τὰ διδόμενα ὁμολογῶν ἔννομον εἶναι, τὸ χάριν τούτων ἀποδοῦναι παρανόμων γράφῃ. He changes the order to ὁμολογῶν οὖν ἔννομον εἶναι τὸ λαβεῖν τὰ διδόμενα, παρανόμων γράφῃ τὸ τούτων χάριν ἀποδοῦναι, and then asks whether the passage will be ὁμοίως δικανικὴ καὶ στρογγύλη. To us it would seem that the chief loss is the loss of emphasis which is entailed (in Greek) by removing from the beginning of the clauses the important and contrasted phrases τὸ λαβεῖν τὰ διδόμενα and τὸ χάριν τούτων ἀποδοῦναι. Possibly this loss of emphasis is implied (among other things) in the words “δικανικὴ καὶ στρογγύλη.”[18]

Where it occurs in Dionysius, the word ἔμφασις bears the sense of ‘hint,’ ‘suggestion,’ ‘soupçon’ (de Thucyd. c. 16 ῥᾳθύμως ἐπιτετροχασμένα καὶ οὐδὲ τὴν ἐλαχίστην ἔμφασιν ἔχοντα τῆς δεινότητος ἐκείνης): a sense which is akin to its technical use of ‘hidden meaning’ (“significatio maior quam oratio,” Cic. Orat. 40. 139; cp. Quintil. viii. 3. 83, ix. 2. 3, 64).[19] In our sense of emphasis due to position, the word ἔμφασις is perhaps hardly used even in the scholiasts; and it is possible that Greek has no single term to express the idea, though it may doubtless be one of the elements in view when a writer uses such expressions as ἁρμονία, σχηματισμός, and ὑπερβατόν.

A modern student of Greek, having to feel his way with practically no help from ancient authorities, will probably reach the conclusion that the rhetorical emphasis he has in mind is attained by placing a word in one of the less usual positions open to it. The word thus emphasized may come at the beginning, in the middle, or at the end of a sentence, the real point being that the position should be (for that particular word) a little out of the ordinary. In Greek, however, as contrasted with English, the emphasis tends to fall on the earlier rather than the later words.[20] In delivery, it would seem that the Greeks found it more natural to stress the beginning than the conclusion of a sentence. But an emphatic word may be found at the end as well as at the beginning, and may sometimes be placed neither at the end nor at the beginning.[21]

Allusion has already been made to the rhetorical emphasis which falls upon the opening words of the Iliad and the Odyssey. As with “arma virumque cano” in the Aeneid, the words μῆνιν and ἄνδρα seem to strike the keynote of the following Epics. And, in a less degree, a certain emphasis due to initial position (and contributing either to emotional effect or to logical clearness) is to be discerned throughout the poems: e.g. in the sixth book of the Iliad:—

δυστήνων δέ τε παῖδες ἐμῷ μένει ἀντιόωσιν.

Homer Iliad vi. 127.

and

πέπλον δ’, ὅς τίς τοι χαριέστατος ἠδὲ μέγιστος
ἔστιν ἐνὶ μεγάρῳ καί τοι πολὺ φίλτατος αὐτῇ,
τὸν θὲς Ἀθηναίης ἐπὶ γούνασιν ἠϋκόμοιο, κτλ.

Homer Iliad vi. 271.

Similarly with the following ten miscellaneous examples of various emphasis, taken chiefly from Dionysius’ favourite speech:—

(1) ἐκεῖνος γὰρ πολλοὺς ἐπιθυμητὰς καὶ ἀστοὺς καὶ ξένους λαβών, οὐδένα πώποτε μισθὸν τῆς συνουσίας ἐπράξατο, ἀλλὰ πᾶσιν ἀφθόνως ἐπήρκει τῶν ἑαυτοῦ.

Xenophon Memorabilia i. 2. 60.

(2) καὶ ταραχώδης ἦν ἡ ναυμαχία, ἐν ᾗ αἱ Ἀττικαὶ νῆες παραγιγνόμεναι τοῖς Κερκυραίοις, εἴ πῃ πιέζοιντο, φόβον μὲν παρεῖχον τοῖς ἐναντίοις, μάχης δὲ οὐκ ἦρχον δεδιότες οἱ στρατηγοὶ τὴν πρόρρησιν τῶν Ἀθηναίων.[22]

Thucydides i. 49.

(3) Ἀναξαγόρου οἴει κατηγορεῖν, ὦ φίλε Μέλητε, κτλ.

Plato Apology 26 D.

(4) οὐ γὰρ τὰ ῥήματα τὰς οἰκειότητας ἔφη βεβαιοῦν, μάλα σεμνῶς ὀνομάζων, ἀλλὰ τὸ ταὐτὰ συμφέρειν.

Demosthenes de Corona § 35.

(5) οἱ μὲν κατάπτυστοι Θετταλοὶ καὶ ἀναίσθητοι Θηβαῖοι φίλον, εὐεργέτην, σωτῆρα τὸν Φίλιππον ἡγοῦντο· πάντ’ ἐκεῖνος ἦν αὐτοῖς· οὐδὲ φωνὴν ἤκουον εἴ τις ἄλλο τι βούλοιτο λέγειν.

id. ib. § 43.

(6) οὓς σὺ ζῶντας μέν, ὦ κίναδος, κολακεύων παρηκολούθεις, τεθνεώτων δ’ οὐκ αἰσθάνει κατηγορῶν.

id. ib. § 162.

(7) καὶ τότ’ εὐθὺς ἐμοῦ διαμαρτυρομένου καὶ βοῶντος ἐν τῇ ἐκκλησίᾳ “πόλεμον εἰς τὴν Ἀττικὴν εἰσάγεις, Αἰσχίνη, πόλεμον Ἀμφικτυονικόν, κτλ.”

id. ib. § 143.

(8) ὃς γὰρ ἐμοῦ φιλιππισμόν, ὦ γῆ καὶ θεοί, κατηγορεῖ, τί οὗτος οὐκ ἂν εἴποι;

id. ib. § 294.

(9) ἀλλ’ οἶμαι οὐ δυνάμεθα· ἐλεεῖσθαι οὖν ἡμᾶς πολὺ μᾶλλον εἰκός ἐστίν που ὑπὸ ὑμῶν τῶν δεινῶν ἢ χαλεπαίνεσθαι.

Plato Republic i. 336 E.

(10) μηδ’ εἵμασι στρώσασ’ ἐπίφθονον πόρον
τίθει· θεούς τοι τοῖσδε τιμαλφεῖν χρεών.

Aeschylus Agamemnon 921.

It will be seen from some of the above examples that words may have emphasis if, though not actually placed at the very beginning of a sentence or a clause, they come as early as they well can. The three following passages will further illustrate this point:—

(1) καὶ ἐς Νικίαν τὸν Νικηράτου στρατηγὸν ὄντα ἀπεσήμαινεν, ἐχθρὸς ὢν καὶ ἐπιτιμῶν, ῥᾴδιον εἶναι παρασκευῇ, εἰ ἄνδρες εἶεν οἱ στρατηγοί, πλεύσαντας λαβεῖν τοὺς ἐν τῇ νήσῳ, καὶ αὐτός γ’ ἄν, εἰ ἦρχε, ποιῆσαι τοῦτο.

Thucydides iv. 27.

(2) ὅ τι μὲν ὑμεῖς, ὦ ἄνδρες Ἀθηναῖοι, πεπόνθατε ὑπὸ τῶν ἐμῶν κατηγόρων, οὐκ οἶδα· ἐγὼ δ’ οὖν καὶ αὐτὸς ὑπ’ αὐτῶν ὀλίγου ἐμαυτοῦ ἐπελαθόμην· οὕτω πιθανῶς ἔλεγον. καίτοι ἀληθές γε, ὡς ἔπος εἰπεῖν, οὐδὲν εἰρήκασιν.

Plato Apology init.

(3) ἀλλὰ μὴν τὸν τότε συμβάντα ἐν τῇ πόλει θόρυβον ἴστε μὲν ἅπαντες, μικρὰ δ’ ἀκούσατε ὅμως, αὐτὰ τἀναγκαιότατα ... οἱ δὲ τοὺς στρατηγοὺς μετεπέμποντο καὶ τὸν σαλπιγκτὴν ἐκάλουν, καὶ θορύβου πλήρης ἦν ἡ πόλις.

Demosthenes de Corona §§ 168, 169.

Sometimes, however, emphatic words will be thrust right to the front through such devices as the postponement of an interrogative particle: e.g.

ἑστάναι, εἶπον, καὶ κινεῖσθαι τὸ αὐτὸ ἅμα κατὰ τὸ αὐτὸ ἆρα δυνατόν;

Plato Republic iv. 436 C.

and

οἷον δίψα ἐστὶ δίψα ἆρά γε θερμοῦ ποτοῦ ἢ ψυχροῦ, ἢ πολλοῦ ἢ ὀλίγου, ἢ καὶ ἑνὶ λόγῳ ποιοῦ τινος πώματος;

id. ib. iv. 437 D.[23]

An uninflected language may well envy the grammatical resources which enable Greek or Latin poets to secure at once clearness and the utmost height of emotion in such lines as:

Ζεῦ πάτερ, ἀλλὰ σὺ ῥῦσαι ὑπ’ ἠέρος υἷας Ἀχαιῶν,
ποίησον δ’ αἴθρην, δὸς δ’ ὀφθαλμοῖσιν ἰδέσθαι·
ἐν δὲ φάει καὶ ὄλεσσον, ἐπεί νύ τοι εὔαδεν οὕτως.

Homer Iliad xvii. 645.

Me, me, adsum qui feci, in me convertite ferrum,
O Rutuli.

Virgil Aeneid ix. 427.[24]

The end as well as the beginning of a clause or sentence may bring emphasis when it is an unusual position for the particular word or phrase which stands there. Illustrations may perhaps be drawn from expressions conveying the idea of “death,” which (according to Dionysus in the Frogs) is the “heaviest of ills,” and which (be that as it may) is as little likely as any to be entertained lightheartedly, or to be mentioned without some degree of feeling and emphasis. At the beginning of a sentence, τεθνᾶσι clearly has emphasis in

τεθνᾶσ’ ἀδελφοὶ καὶ πατὴρ οὑμὸς γέρων.

Euripides Hercules Furens 539.

And in the following passage of Plato, it will be seen that the τὸν θάνατον which comes near the beginning of a clause is more emphatic than the τὸν θάνατον which comes at the end of a clause:—

οἶσθα δ’, ἦ δ’ ὅς, ὅτι τὸν θάνατον ἡγοῦνται πάντες οἱ ἄλλοι τῶν μεγάλων κακῶν;—καὶ μάλ’, ἔφη.—οὐκοῦν φόβῳ μειζόνων κακῶν ὑπομένουσιν αὐτῶν οἱ ἀνδρεῖοι τὸν θάνατον, ὅταν ὑπομένωσιν;—ἔστι ταῦτα.

Plato Phaedo 68 D.

The τὸν θάνατον before ἡγοῦνται is here emphatic on the same principle as the θάνατον before εἰσέθηκε in the passage (already alluded to) of the Frogs:—

θάνατον γὰρ εἰσέθηκε βαρύτατον κακόν.

Aristophanes Ranae 1394.

But a word like θάνατος may also come with emphasis at the end of a sentence, if that order is rendered unusual by the interposition of additional words or by any other means which create a feeling of suspense and even of afterthought. For example:

τί δέ; τὰν Αἵδου ἡγούμενον εἶναί τε καὶ δεινὰ εἶναι οἴει τινὰ θάνατου ἀδεῆ ἔσεσθαι καὶ ἐν ταῖς μάχαις αἱρήσεσθαι πρὸ ἥττης τε καὶ δουλείας θάνατον;

Plato Republic iii. 386 B.

Here the θάνατον seems intended to repeat with emphasis the preceding θανάτου to which, itself, a considerable degree of prominence is assigned. So, perhaps,

ἀλλὰ νόμον δημοσίᾳ τὸν ταῦτα κωλύσοντα τέθεινται τουτονὶ καὶ πολλοὺς ἤδη παραβάντας τὸν νόμον τοῦτον ἐζημιώκασιν θανάτῳ.

Demosthenes Midias § 49.

and

... καὶ φοβερωτέρας ἡγήσεται τὰς ὕβρεις καὶ τὰς ἀτιμίας, ἃς ἐν δουλευούσῃ τῇ πόλει φέρειν ἀνάγκη, τοῦ θανάτου.

Demosthenes de Corona § 205.

Some miscellaneous examples of words coming emphatically at the end of a clause or sentence are:—

(1) αἰτοῦμαι δ’ ὑμᾶς δοῦναι καὶ νῦν παισὶ μὲν καὶ γυναικὶ καὶ φίλοις καὶ πατρίδι εὐδαιμονίαν, ἐμοὶ δὲ οἷόν περ αἰῶνα δεδώκατε τοιαύτην καὶ τελευτὴν δοῦναι.

Xenophon Cyropaedia viii. 7.

(2) ἀλλὰ καὶ τούτους κολυμβηταὶ δυόμενοι ἐξέπριον μισθοῦ.[25]

Thucydides vii. 25.

(3) ὑψοῦ δὲ θάσσων ὑψόθεν χαμαιπετὴς
πίπτει πρὸς οὖδας μυρίοις οἰμώγμασι
Πενθεύς.[26]

Euripides Bacchae 1111.

(4) ἴστε γὰρ δήπου τοῦθ’ ὅτι πάντες οἱ ξεναγοῦντες οὗτοι πόλεις καταλαμβάνοντες Ἑλληνίδας ἄρχειν ζητοῦσιν, καὶ πάντων, ὅσοι περ νόμοις οἰκεῖν βούλονται τὴν αὑτῶν ὄντες ἐλεύθεροι, κοινοὶ περιέρχονται κατὰ πᾶσαν χώραν, εἰ δεῖ τἀληθὲς εἰπεῖν, ἐχθροί.

Demosthenes Aristocrates § 139.

(5) δεῖ δὲ τοὺς ἀγαθοὺς ἄνδρας ἐγχειρεῖν μὲν ἅπασιν ἀεὶ τοῖς καλοῖς, τὴν ἀγαθὴν προβαλλομένους ἐλπίδα, φέρειν δ’ ἃν ὁ θεὸς διδῷ γενναίως.[27]

Demosthenes de Corona § 97.

(6) εἶθ’ οὗτοι τὰ ὅπλα εἶχον ἐν ταῖς χερσὶν ἀεί.

id. ib. § 235.

(7) εἰ γὰρ ταῦτα προεῖτ’ ἀκονιτεί, περὶ ὧν οὐδένα κίνδυνον ὅντιν’ οὐχ ὑπέμειναν οἱ πρόγονοι, τίς οὐχὶ κατέπτυσεν ἂν σοῦ; μὴ γὰρ τῆς πόλεώς γε, μηδ’ ἐμοῦ.

id. ib. § 200.

(8) ... ἡμῖν δὲ τοῖς λοιποῖς τὴν ταχίστην ἀπαλλαγὴν τῶν ἐπηρτημένων φόβων δότε καὶ σωτηρίαν ἀσφαλῆ.[28]

id. ib. § 324.

It may be added that, occasionally, both the earlier and the later positions are emphatic in the same clause or sentence: e.g.

(1) τέκνα γὰρ κατακτενῶ
τἄμ’.[29]

Euripides Medea 792.

(2) ὦτα γὰρ τυγχάνει ἀνθρώποισι ἐόντα ἀπιστότερα ὀφθαλμῶν.[30]

Herodotus i. 8.

(3) νῦν δὲ τὸ μὲν παρὸν ἀεὶ προϊέμενοι, τὰ δὲ μέλλοντ’ αὐτόματ’ οἰόμενοι σχήσειν καλῶς, ηὐξήσαμεν, ὦ ἄνδρες Ἀθηναῖοι, Φίλιππον ἡμεῖς, καὶ κατεστήσαμεν τηλικοῦτον ἡλίκος οὐδείς πω βασιλεὺς γέγονεν Μακεδονίας. [31]

Demosthenes Olynthiacs i. § 9.

(4) πολλάκις δὲ τοῦ κήρυκος ἐρωτῶντος οὐδὲν μᾶλλον ἀνίστατ’ οὐδείς, ἁπάντων μὲν τῶν στρατηγῶν παρόντων, κτλ.

Demosthenes de Corona § 117.

(5) καὶ μὴν καὶ Φερὰς πρώην ὡς φίλος καὶ σύμμαχος εἰς Θετταλίαν ἐλθὼν ἔχει καταλαβών, καὶ τὰ τελευταῖα τοῖς ταλαιπώροις Ὠρείταις τουτοισὶ ἐπισκεψομένους ἔφη τοὺς στρατιώτας πεπομφέναι κατ’ εὔνοιαν· πυνθάνεσθαι γὰρ αὐτοὺς ὡς νοσοῦσι καὶ στασιάζουσιν, συμμάχων δ’ εἶναι καὶ φίλων ἀληθινῶν ἐν τοῖς τοιούτοις καιροῖς παρεῖναι.

Demosthenes Philippics iii. § 12.

(6) οὐ λίθοις ἐτείχισα τὴν πόλιν οὐδὲ πλίνθοις ἐγώ, οὐδ’ ἐπὶ τούτοις μέγιστον τῶν ἐμαυτοῦ φρονῶ.

Demosthenes de Corona § 299.

(7) ὑπὲρ τῶν ἐχθρῶν πεπολίτευσαι πάντα, ἐγὼ δ’ ὑπὲρ τῆς πατρίδος.

id. ib. § 265.

In connexion with the imperfect appreciation which the de Compositione Verborum shows of a normal order and of an emphasis produced by departure from it, attention may be drawn to the fact that the treatise contains no reference to the ‘figure’ hyperbaton; and this although the figure had been recognized long before Dionysius’ time, and continued to be recognized long afterwards. It is first mentioned by Plato, who probably took over the notion from the Sophists: ἀλλ’ ὑπερβατὸν δεῖ θεῖναι ἐν τῷ ᾄσματι τὸ “ἀλαθέως” (Plato Protag. 343 E, where the reference is to a poem of Simonides). The author of the Rhetorica ad Alexandrum (c. 30) indicates it in the following terms: ἐὰν μὴ ὑπερβατῶς αὐτὰ [sc. τὰ ὀνόματα] τιθῶμεν, ἀλλ’ ἀεὶ τὰ ἐχόμενα ἑξῆς τάττωμεν. Quintilian treats of it in the passage beginning “Hyperbaton quoque, id est verbi transgressionem, quoniam frequenter ratio comparationis et decor poscit, non immerito inter virtutes habemus” (Inst. Or. viii. 6. 62).[32] The author of the Treatise on the Sublime describes and defines it thus: ἔστι δὲ λέξεων ἢ νοήσεων ἐκ τοῦ κατ’ ἀκολουθίαν κεκινημένη τάξις καὶ οἱονεὶ χαρακτὴρ ἐναγωνίου πάθους ἀληθέστατος (Longinus de Sublim. c. 22).[33] And, later still, Hermogenes and other writers on rhetoric are well acquainted with the figure. Dionysius, however, mentions it but seldom in any of his writings, and even then (e.g. τὰς ὑπερβατοὺς καὶ πολυπλόκους καὶ ἐξ ἀποκοπῆς πολλὰ σημαίνειν πράγματα βουλομένας καὶ διὰ μακροῦ τὰς ἀποδόσεις λαμβανούσας νοήσεις, de Thucyd. c. 52; cp. c. 31 ibid.) is clearly thinking not of desirable but of highly undesirable “inversions.” He may have thought that its proper place was in poetry rather than in prose.

E. Euphony

A modern writer on style would probably lay more stress on clearness and emphasis than on euphony. The ancient critics, on the other hand, seem to have taken the two former elements more or less for granted. Because they were easily attainable in languages so fully inflected as Greek and Latin, their attainment was regarded as an important matter indeed, but one which called for no special recognition of any kind. As Quintilian says, in reference to clearness, “nam emendate quidem ac lucide dicentium tenue praemium est, magisque ut vitiis carere quam ut aliquam magnam virtutem adeptus esse videaris” (Inst. Or. viii. 3. 1).[34] Dionysius, too, in the de Compositione Verborum, passes more readily over the two qualities of clearness and emphasis because he is not concerned with the πραγματικὸς τόπος.[35] He keeps rigorously to his real subject; and that is not the relation of words to the ideas of which they are the symbols. It is, rather, their relation to their own constituent elements (letters and syllables of diverse qualities and quantities) and to the pleasant impression which the apt collocation of many various words can make upon the ear. His task is to investigate the emotional power of the sound-elements of language when alone and when in combination—their euphonic and their symphonic effects. Hence the constant recurrence, throughout the treatise, of words like εὐφωνία, εὐρυθμία, εὐστομία, λειότης, ἁρμονία, σύνθεσις. The illustrative excerpts which he gives are so numerous and so happily chosen that no others need be added here.[36] A careful study of his examples, in the context in which they occur, will suggest many reflexions upon the freedom and adaptability of Greek order. But no absolute test of euphony

can be based upon them. Dionysius himself formulates no invariable rules upon the subject. In the last resort, the court of appeal must, as he sees, be the instinctive judgment of the ear (τὸ ἄλογον τῆς ἀκοῆς πάθος).[37] The part played by the ear has been well described by Quintilian: “ergo quem in poëmate locum habet versificatio, eum in oratione compositio. optime autem de illa iudicant aures, quae plena sentiunt et parum expleta desiderant et fragosis offenduntur et levibus mulcentur et contortis excitantur et stabilia probant, clauda deprehendunt, redundantia ac nimia fastidiunt” (Inst. Or. ix. 4. 116). Naturally the ear in question must be the individual ear (“aurem tuam interroga, quo quid loco conveniat dicere,” Aulus Gellius Noctes Att. xiii. 21); the criterion is subjective, not absolute.[38] But it is assumed that the ear in question has been trained and attuned by constant converse with the great masters, and that (like Flaubert in modern times) an author never writes without repeating the words aloud to himself. Thus trained, the ear will work in harmony with the mind: “aures enim vel animus aurium nuntio naturalem quandam in se continet vocum omnium mensionem” (Cic. Orat. 53. 177). Both Cicero and Dionysius are well aware that style is personal and individual,—that it is no uniform and mechanical thing. Dionysius’ own position has been misunderstood by those who have judged the de Compositione as if it were a complete treatise on the entire subject of style. In the eyes of Dionysius, words are not what dead stone and timber are in the eyes of the ordinary workman. They are, rather, the living elements which, in the secret places of his mind, the master-builder views as potential parts of some great temple.[39] They are what an individual makes them. Hence, just as Cicero writes “qua re sine, quaeso, sibi quemque scribere,

Suam quoíque sponsam, míhi meam; suum quoíque amorem, míhi meum”:

so Dionysius long ago anticipated the saying that the style is the man.[40]

Among the minor debts we owe to him is the fact that his minute analysis of rhythms, or feet, in passages of Thucydides, Pindar and others, helps to disclose the inner workings of the beautiful Greek language and to impress us with the importance attached by the ancients to what we moderns find it so hard fully to appreciate,—the effect on a Greek ear of syllabic quantity in prose as well as verse. And he insists no less upon the charm of variety,—the paramount necessity of avoiding monotony. He saw, for example, that the Greek inflexions (notwithstanding the many advantages which they brought with them) had at least one drawback: they are apt to lead to a certain sameness in case-endings. Accordingly he would, for instance, have approved (though he does not mention this particular passage) of the separation of the words σωτηρίαν ἀσφαλῆ from the other accusatives at the end of the de Corona: ἡμῖν δὲ τοῖς λοιποῖς τὴν ταχίστην ἀπαλλαγὴν τῶν ἐπηρτημένων φόβων δότε καὶ σωτηρίαν ἀσφαλῆ.[41] Further reference to these minutiae of style may fitly be made later, when the topics of “rhythm” and “music” are considered.[42]

F. Greek and Latin compared with Modern Languages, in regard to Word-Order

Something has already been said, incidentally, about certain differences in word-order between the ancient and the modern European languages. In such a comparison Greek and Latin may be placed upon the same footing, as their points of contact are vastly more numerous than their points of divergence, considerable though these are.[43]

The points of contact become manifest when an attempt is made to translate into Latin, and into English, the sentence from Herodotus which Dionysius quotes, and twice recasts, in his fourth chapter:—

(1) Κροῖσος ἦν Λυδὸς μὲν γένος, παῖς δ’ Ἀλυάττου, τύραννος δ’ ἐθνῶν τῶν ἐντὸς Ἅλυος ποταμοῦ· ὃς ῥέων ἀπὸ μεσημβρίας μεταξὺ Σύρων τε καὶ Παφλαγόνων ἐξίησι πρὸς βορέαν ἄνεμον εἰς τὸν Εὔξεινον καλούμενον πόντον.

Herodotus i. 6.

Croesus genere quidem fuit Lydus, patre autem Alyatte; earum vero nationum tyrannus, quae intra Halym amnem sunt: qui, a meridie Syros ac Paphlagones interfluens, contra ventum Aquilonem in mare, quid vocant Euxinum, evolvitur.

(2) Κροῖσος ἦν υἱὸς μὲν Ἀλυάττου, γένος δὲ Λυδός, τύραννος δὲ τῶν ἐντὸς Ἅλυος ποταμοῦ ἐθνῶν· ὃς ἀπὸ μεσημβρίας ῥέων μεταξὺ Σύρων καὶ Παφλαγόνων εἰς τὸν Εὔξεινον καλούμενον πόντον ἐκδίδωσι πρὸς βορέαν ἄνεμον.

Croesus erat filius quidem Alyattis, genere autem Lydus, tyrannusque earum, quae intra sunt Halym amnem nationes; qui, a meridie interfluens Syros ac Paphlagones, in mare, quod vocant Euxinum, evolvitur contra ventum Aquilonem.

(3) Ἀλυάττου μὲν υἱὸς ἦν Κροῖσος, γένος δὲ Λυδός, τῶν δ’ ἐντὸς Ἅλυος ποταμοῦ τύραννος ἐθνῶν· ὃς ἀπὸ μεσημβρίας ῥέων Σύρων τε καὶ Παφλαγόνων μεταξὺ πρὸς βορέαν ἐξίησιν ἄνεμον ἐς τὸν καλούμενον πόντον Εὔξεινον.

Alyattis quidem filius erat Croesus, genere autem Lydus, earum, quae intra sunt Halym amnem, tyrannus nationum; qui, a meridie fluens Syros inter ac Paphlagones, contra Boream erumpit ventum in mare, quod vocant Euxinum.

In these sentences the Latin follows the Greek order closely, and might be made to follow it still more faithfully were it not that it seems better to diverge occasionally for special reasons: e.g. it is desirable, in rendering the original passage of Herodotus, to secure (as far as possible) a good rhythm. In English, on the other hand, the choice lies between a wide deviation and a rendering which is ambiguous and possibly grotesque. In fact (to recur once more to the main point) the freedom with which the order of words can be varied in a Greek or Latin sentence is without parallel in any modern analytical language, and the attendant gain in variety, rhythm, and nicety of emphasis is incalculable.[44]

Still, the modern languages have great powers, in this as in other ways: powers which will be incidentally illustrated later. M. Jules Lemaître has written, with reference to Ernest Renan: “Je trahis peut-être sa pensée en la traduisant; tant pis! Pourquoi a-t-il des finesses qui ne tiennent qu’à l’arrangement des mots?”[45] These finesses are perhaps, as is here implied, hardly communicable, even though an earlier French writer has commended Malherbe as an author who

D’un mot mis en sa place enseigna le pouvoir.[46]

It may well be that these matters, if not altogether the “mysteries” which Dionysius terms them, are eternally elusive because they depend upon the infinite variety of the human mind. Yet some studies in English literary theory, such as might be suggested by Dionysius’ treatise, could not fail to be of interest, and might be instructive also. Something of the kind has been already done, without reference to Dionysius or other Greek critics, by Robert Louis Stevenson in his essay on Some Technical Elements of Style in Literature.[47] Each language has, in truth, a rhetoric of its own. But the various languages, ancient and modern, can help one another in the way of comparison and contrast.

These methods of comparison and contrast have—as regards word-order—been excellently applied to the ancient and the modern languages by Henri Weil and T. D. Goodell. Weil’s chief service is to have pointed out so clearly the principle that the order of syntax must be separated in thought from the order of ideas, and was by both Greeks and Romans freely so separated in practice, whereas in the modern languages (owing to the lack of inflexions) this practical separation is less frequent. Goodell, starting from the postulate that the order of words in a language represents the order in which the speaker or writer chooses, for various reasons, to bring his ideas before the mind of another, discusses (with constant reference to modern languages) the order of words in Greek, from the standpoint of syntax, rhetoric, and euphony. In the course of a carefully reasoned exposition, he corrects and supplements many of Weil’s observations.

The full title of Weil’s book is De l’ordre des mots dans les langues anciennes comparées aux langues modernes: question de grammaire générale (3rd edition, Paris, 1879). There is an English translation by C. W. Super (Boston, 1887), with notes and additions. Goodell’s paper on “The Order of Words in Greek” is printed in the Transactions of the American Philological Association vol. xxi. Other writings on the subject are: Charles Short’s “Essay on the Order of Words in Attic Greek Prose,”—prefixed to Drisler’s edition of C. D. Yonge’s English-Greek Lexicon,—which is an extensive collection of examples, but is weak in scientific classification and in clear enunciation of principles; H. L. Ebeling’s “Some Statistics on the Order of Words in Greek,” contributed to Studies in Honour of Basil Lanneau Gildersleeve, and including some valuable investigations into the order in which subject, object, and verb usually come in Greek; inquiries into the practice of individual authors, e.g. Spratt on the “Order of Words in Thucydides” (Spratt’s edition of Thucydides, Book VI.), and Riddell on the “Arrangement of Words and Clauses in Plato” (Riddell’s edition of Plato’s Apology), or various dissertations such as Th. Harmsen de verborum collocatione apud Aeschylum, Sophoclem, Euripidem capita selecta, Ph. Both de Antiphontis et Thucydidis genere dicendi, J. J. Braun de collocatione verborum apud Thucydidem observationes, F. Darpe de verborum apud Thucydidem collocatione; and in Latin such elaborate studies as Hilberg’s Die Gesetze der Wortstellung im Pentameter des Ovid. An interesting book which compares Cicero’s Latin translations (prose and verse) with their Greek originals is V. Clavel’s de M. T. Cicerone Graecorum Interprete. In Harvard Studies in Classical Philology vol. vii. pp. 223-233, J. W. H. Walden discusses Weil’s statement that “an emphatic word, if followed by a word which, though syntactically necessary to the sentence, is in itself unemphatic, receives an access of emphasis from the lingering of the attention which results from the juxtaposition of the two.” Reference may also be made to A. Bergaigne’s “Essai sur la construction grammaticale considérée dans son développement historique, en Sanskrit, en Grec, en Latin, dans les langues romanes et dans les langues germaniques,” in the Mémoires de la Société de Linguistique de Paris vol. vii. The subject is, further, glanced at in the Greek Grammars of Kühner and others. But in modern times, as in those of Dionysius, it has on the whole failed to receive the attention which its importance would seem to demand.

G. Prose and Poetry: Rhythm and Metre

Readers of the de Compositione cannot fail to notice that, catholic as he is in his literary tastes, Dionysius reserves his highest admiration for two authors,—Homer in poetry and Demosthenes in prose; and that he seems to regard them as equally valid authorities for the immediate purpose which he has in view. Homer is quoted throughout the treatise, on the first page and on the last; and Demosthenes inspires (in c. 25) its most eloquent passage. That outburst is a triumphant vindication of Demosthenes’ methods as a sedulous artist. Dionysius sees that he is one of those men who spare no pains over the art they love—that Demosthenes, like Homer, φιλοτεχνεῖ ([200] 18; cp. [154] 20).

In seeming thus to draw no very clear line between verse and prose, Dionysius is at one with most of the Greek and Roman critics; and this attitude is readily intelligible in the light of the historical development of Greek literature, in which Homer (who was a master of oratory [48] as well as of poetry) heralds the intellectual life of all Greece, while Demosthenes is the last great voice of free Athens. But the approximations of prose to poetry, and of poetry to prose, which Dionysius describes in his twenty-fifth and twenty-sixth chapters should not create the impression that, in his opinion, the prose-writer was free to borrow any and every weapon from the armoury of the poet. Of one poetical artifice he says, in c. 6, “this principle can be applied freely in poetry, but sparingly in prose”; and elsewhere he calls attention to qualities which he regards as over-poetical in the styles of Thucydides and Plato.[49] Yet he did clearly wish that good prose should borrow as much as possible from poetry, while still remaining good prose. And although he agrees, in general, with Aristotle’s exposition of the formal differences between prose and poetry, he does not adhere quite firmly to the Aristotelian principles.[50]

In the Rhetoric, Aristotle insists that the styles of poetry and prose are distinct. The difference is this: “prose should have rhythm but not metre, or it will be poetry. The rhythm, however, should not be of too marked a character: it should not pass beyond a certain point.”[51] In the same way, Dionysius (C.V. c. 25) declares that prose must not be manifestly metrical or rhythmical, lest it should desert its own specific character. It should simply appear to be the one and the other, so that it may be poetical although not a poem, and lyrical although not a lyric. But, in practice, Dionysius is found to cast longing eyes upon the formal advantages which poetry possesses, and to wish to infuse into public speeches a definite metrical element, which seems alien to the genius of prose, and which would have failed to gain the sanction of Aristotle, though this appears to be claimed for it.[52] It is not here a question of the ordinary methods of imparting force and variety to word-arrangement. In regard to these, Dionysius’ precepts are, in general, sound and helpful enough; and if, now and then, the process is extolled in what may seem extravagant terms, we have only to think of the vast difference which slight variations of word-order will make even in our modern analytical languages. For example:

Cut is the branch that might have grown full straight.

Marlowe Doctor Faustus.

Killed with report that old man eloquent.

Milton Sonnets.

Schön war ich auch, und das war mein Verderben.

Goethe Faust.

The effect of these lines would be sadly marred if we were to read “the branch is cut,” “that eloquent old man,” and “ich war auch schön.”[53] In Greek prose, no less than in Greek poetry, inversions like those just quoted would be quite legitimate. This at least we can affirm, though it would be rash to attempt to lay down any general rules with regard to the differences between Greek order in verse and in prose. It is better to follow Dionysius’ example and to cull illustrations from both alike impartially, with only two qualifications. First, the Greek word-arrangement is even freer in verse than in prose, though the clause-arrangement and the sentence-arrangement of Greek poetry show (as Dionysius implies in c. 26) a general tendency to coincide with the metrical arrangement. Second, an absolutely metrical arrangement is foreign to the best traditions of Greek prose. It is the second point that is of importance here; and notwithstanding the almost furtive character which he attributes to the metrical lines detected by him in the Aristocrates, it is obvious that Dionysius has in mind a very close and deliberate approximation to the canons of verse and is prepared to strain his material in order to attain it.[54] Here, again, some modern illustrations may be of interest. The writers of the Tudor period seem to have had a special fondness for, and an ear attuned to, what may be roughly regarded as hexameter measures. This predilection appears both in their rendering of the Bible and in the Book of Common Prayer:—

The rhythms into which modern prose-writers drop are usually iambic or trochaic. This is so with Ruskin and Carlyle, and it would be easy to quote examples from their writings.[56] But, as in ancient so in modern times, the best criticism looks with favour on rhythmical, with disfavour on metrical prose. Prose, it is held, loses its true character—as the minister primarily of reason rather than of emotion—if it is made to conform to the rigid laws of metre.

If Dionysius fails to prove that metrical lines, thinly disguised, are a marked feature of the style of Demosthenes, no greater fortune has attended some attempts made in our own day to establish such exact rhythmical laws as that of the systematic avoidance, in Greek oratory, of a number of short syllables in close succession. It is clear that Demosthenes’ ear, with that kind of instinct which comes from musical aptitude and long training (cp. C.V. [266] 13 ff., [268] 12), shunned undignified accumulations of short syllables, but not with so pedantic a persistency that he could not on occasion use forms like πεφενάκικεν or διατετέλεκεν or προσαγαγόμενον. If he formulated to himself a principle, instead of trusting to inspiration controlled by long experience, this principle would be that which Cicero attributes to a critic who was almost contemporary with Demosthenes: “namque ego illud adsentior Theophrasto, qui putat orationem, quae quidem sit polita atque facta quodam modo, non astricte, sed remissius numerosam esse oportere” (Cic. de Orat. iii. 48. 184).[57] The necessary limits to be observed in these curious inquiries are well indicated by Quintilian, who utters some sensible warnings against any attempts continually to scent metre in prose or to ban some feet while admitting others: “neque enim loqui possumus nisi syllabis brevibus ac longis, ex quibus pedes fiunt ... miror autem in hac opinione doctissimos homines fuisse, ut alios pedes ita eligerent aliosque damnarent, quasi ullus esset, quem non sit necesse in oratione deprehendi” (Quintil. Inst. Or. ix. 4. 61 and 87).[58]

On the subject of prose and poetry, Coleridge’s Biographia Literaria (ed. Shawcross, Clarendon Press, 1907) is likely long to hold its unique position. Theodore Watts-Dunton’s article on “Poetry” in the Encyclopaedia Britannica contains an appreciative estimate of the good service done to criticism by Dionysius in the de Compositione. The article by Louis Havet on La Prose métrique (in La Grande Encyclopédie, xxvii. 804-806) deals with what we should call “rhythmical prose,” the French terminology differing here from our own. Some account of enjambement (with ancient and modern illustrations) will be found in the Notes, pp. 270 ff. The recent writings on Greek rhythm and metre are almost endless. Some of them will be suggested by the names of: Rossbach, Westphal, Weil, Schmidt, Christ, Gleditsch, Wilamowitz-Moellendorff, Goodell, Masqueray, Blass.

With regard to the relation between metre and rhythm, there is not a little suggestiveness in the saying of the historical Longinus: μέτρου δὲ πατὴρ ῥυθμὸς καὶ θεός (Proleg. in Heph. Ench.; Westphal Script. Metr. Graeci i. 82). There is also, in our day, an increasing recognition of the intimate alliance between Greek poetry and Greek music; it is more and more seen that lyric stanzas are formed out of figures and phrases, rather than from mere mechanical feet. Nor is it to be forgotten that poetic rhythm may probably be traced back to the regular movements of the limbs in dancing. The views of Blass on ancient prose rhythm are given in his Die attische Beredsamkeit, Die Rhythmen der attischen Kunstprosa (Isokrates, Demosthenes, Platon), and Die Rhythmen der asianischen und römischen Kunstprosa (Paulus, Hebräerbrief, Pausanias, Cicero, Seneca, Curtius, Apuleius); and some of them are summarized in an article which he contributed, shortly before his death, to Hermathena (“On Attic Prose Rhythm” Hermathena No. xxxii., 1906). Probably his tendency was to seek after too much uniformity in such matters as the avoidance of hiatus and of successive short syllables, or as the symmetrical correspondences between clauses within the period. The best Attic orators were here guided, more or less consciously, by two principles to which Dionysius constantly refers: (1) μεταβολή, or the love of variety; (2) τὸ πρέπον, or the sense of propriety. This sense of propriety rejected all such obvious and systematic art as should cause a speech to seem, in Aristotle’s words, πεπλασμένος and ἀπίθανος (Rhet. iii. 2. 4; 8. 1). Still, Demosthenes’ greatest speeches were no doubt carefully revised before they were given to the world; and so the blade may have been cold-polished, after leaving the forge of the imagination. It is to be noticed that, in the matter of hiatus, for example, some of the best manuscripts of Demosthenes do seem to observe a strict parsimony; and this careful avoidance of open vowels may be due ultimately rather to Demosthenes himself than to an early scholar-editor. Whatever the final judgment on Blass’s work may be, he will have done good service by directing attention anew to a point so hard for the modern ear to appreciate as the great part played in artistic Greek prose by the subtle use of time,—of long and short syllables arranged in a kind of general equipoise rather than in any regular and definite succession. How singularly important that part was reckoned to be, such passages of Dionysius as the following help to indicate: οὐ γὰρ δὴ φαῦλόν τι πρᾶγμα ῥυθμὸς ἐν λόγοις οὐδὲ προσθήκης τινὸς μοῖραν ἔχον οὐκ ἀναγκαίας, ἀλλ’ εἰ δεῖ τἀληθές, ὡς ἐμὴ δόξα, εἰπεῖν, ἁπάντων κυριώτατον τῶν γοητεύειν δυναμένων καὶ κηλεῖν τὰς ἀκοάς (de Demosth. c. 39).

III
Other Matters arising in the de Compositione

A. Greek Music: in Relation to the Greek Language

For the modern student there is perhaps no more valuable chapter of the de Compositione than that (c. 11) which treats of the musical element in Greek speech. It helps to bring home the fact that, among the ancient Greeks, “the science of public oratory was a musical science, differing from vocal and instrumental music in degree, not in kind” (μουσικὴ γάρ τις ἦν καὶ ἡ τῶν πολιτικῶν λόγων ἐπιστήμη τῷ ποσῷ διαλλάττουσα τῆς ἐν ᾠδῇ καὶ ὀργάνοις, οὐχὶ τῷ ποιῷ, [124] 20). The extraordinary sensitiveness of Greek audiences to the music of sounds is described by Dionysius, who also indicates the musical intervals observed in singing and in speaking, and touches on the relation borne by the words to the music in a song. His statements, further, give countenance to the view that “the chief elements of utterance—pitch, time, and stress—were independent in ancient Greek speech, just as they are in music. And the fact that they were independent goes a long way to prove our main contention, viz. that ancient Greek speech had a peculiar quasi-musical character, and consequently that the difficulty which modern scholars feel in understanding the ancient statements on such matters as accent and quantity is simply the difficulty of conceiving a form of utterance of which no examples can now be observed.”[59] Even Aristotle, Greek though he was, seems to have felt imperfectly those harmonies of balanced cadence which come from the poet, or artistic prose-writer, to whom words are as notes to the musician. And if Aristotle, a Greek though not an Athenian, shows himself not fully alive to the music of the most musical of languages, it is hardly matter for wonder that writers of our own rough island prose should be far from feeling that they are musicians playing on an instrument of many strings, and should be ready, as Dionysius might have said in his most serious vein, εἰς γέλωτα λαμβάνειν τὰ σπουδαιότατα δι’ ἀπειρίαν ([252] 16). It is true that, on the other side, we have R. L. Stevenson, who writes: “Each phrase of each sentence, like an air or recitative in music, should be so artfully compounded out of longs and shorts, out of accented and unaccented syllables, as to gratify the sensual ear. And of this the ear is the sole judge.”[60] Dionysius and Stevenson are, admittedly, slight names to set against that of Aristotle. But this is no reason why they should not be allowed to supplement his statements when he is too deeply concerned with matter and substance to say much about manner and the niceties and enchantments of form. And Dionysius is—it must in justice be conceded—no mere word-taster but a man genuinely alive to the great issues that dignify and ennoble style. He can, for example, thus describe the effect, subsequent and immediate, of Demosthenes’ speeches: “When I take up one of his speeches, I am entranced and am carried hither and thither, stirred now by one emotion, now by another. I feel distrust, anxiety, fear, disdain, hatred, pity, good-will, anger, jealousy. I am agitated by every passion in turn that can sway the human heart, and am like those who are being initiated into wild mystic rites.... When we who are centuries removed from that time, and are in no way affected by the matters at issue, are thus swept off our feet and mastered and borne wherever the discourse leads us, what must have been the feelings excited by the speaker in the minds of the Athenians and the Greeks generally, when living interests of their own were at stake, and when the great orator, whose reputation stood so high, spoke from the heart and revealed the promptings of his inmost soul?”[61]

In addition to D. B. Monro’s book on Greek music, reference may be made to such works as Rossbach and Westphal’s Theorie der musischen Künste der Hellenen, H. S. Macran’s edition of Aristoxenus’ Harmonics (from the Introduction to which a quotation of some length will be found in the note on [194] 7), and the edition of Plutarch’s de Musica by H. Weil and Th. Reinach. The articles, by W. H. Frere and H. S. Macran, on Greek Music in the new edition of Grove’s Dictionary of Music and Musicians should also be consulted, as well as the essay, by H. R. Fairclough, on “The Connexion between Music and Poetry in Early Greek Literature” in Studies in Honour of Basil Lanneau Gildersleeve. The close connexion between music and verbal harmony is brought out in Longinus de Sublim. cc. 39-41. In Grenfell and Hunt’s Hibeh Papyri, Part i. (1906), p. 45, there is a short “Discourse on Music” which the editors are inclined to attribute to Hippias of Elis, the contemporary of Socrates.

B. Accent in Ancient Greek

If there were any doubt that the Greek accent was an affair of pitch rather than of stress, the eleventh chapter of this treatise would go far to remove it. It is clear that Dionysius describes the difference between the acute and the grave accent as a variation of pitch, and that he considers this variation to be approximately the same as the musical interval of a fifth, or (as he himself explains) three tones and a semitone. Similarly Aristoxenus (Harm. i. 18) writes λέγεται γὰρ δὴ καὶ λογῶδές τι μέλος, τὸ συγκείμενον ἐκ τῶν προσῳδιῶν τῶν ἐν τοῖς ὀνόμασιν· φυσικὸν γὰρ τὸ ἐπιτείνειν καὶ ἀνιέναι ἐν τῷ διαλέγεσθαι (‘for there is a kind of melody in speech which depends upon the accent of words, as the voice in speaking rises and sinks by a natural law,’ Macran). The expression προσῳδία itself (cp. τάσεις φωνῆς αἱ καλούμεναι προσῳδίαι, [196] 16) implies a melodic character, and the adjectives (ὀξύς and βαρύς) which denote ‘acute’ and ‘grave’ are used regularly in Greek music for what we call ‘high’ and ‘low’ pitch.[62] It would be hard to believe that βαρύς could ever have indicated an absence of stress.

That such a musical pitch—such a rising or falling of tone—can be quite independent of quantity seems to be proved by the analogy of Vedic Sanskrit, inasmuch as, when reciting verses in that language, the native priests are said to succeed in keeping quantity and musical accent altogether distinct. “We cannot now say exactly how Homer’s verse sounded in the ears of the Greeks themselves; and yet we can tell even this more nearly than Matthew Arnold imagined. Sanskrit verse, like Greek, had both quantity and musical accent; and the recitation of the Vedic poems, as handed down by immemorial tradition, and as it may be heard to-day, keeps both these elements clear. It is a sort of intoned recitative, most impressive and agreeable to the sensitive ear.”[63]

A useful handbook on the general subject of Greek Accentuation (including its musical character) is Vendryes’ Traité d’accentuation grecque, which is prefaced by a bibliographical list. The volume is noticed, in the Classical Review xix. 363-367, by J. P. Postgate, who supplements it in some important directions. There is also a discussion of the nature and theory of the Greek accent in Hadley’s Essays pp. 110-127. As Monro (Modes p. 113) remarks, it is our habit of using Latin translations of the terms of Greek grammar that has tended to obscure the fact that those terms belong in almost every case to the ordinary vocabulary of music. The point of the illustration drawn from the Orestes, in the C.V. c. 11, is that the musical setting in question neglected entirely the natural tune, or accent, of the words. It is not to be assumed that Dionysius approved (except within narrow limits) of this practice or of the corresponding neglect of syllabic quantity ([128] 19). He probably regarded such excesses as innovations due to inferior schools of music and rhythm. In the hymns found at Delphi (and also in an inscription discovered by W. M. Ramsay) there is a remarkable correspondence between the musical notes and the accentuation of the words, as was pointed out by Monro (Modes pp. 90, 91, 116, 141; and Classical Review ix. 467-470). It is the hymns to Apollo (belonging probably to the early part of the third century B.C.), in which the acute accents usually coincide with a rise of pitch, that Dionysius would doubtless have regarded as embodying the classical practice. In early times, it must be remembered, words and music were written by the same man; cp. G. S. Farnell Greek Lyric Poetry pp. 41, 42. The chief surviving fragments of Greek music (including the recent discoveries at Delphi) will be found in C. Jan’s Musici Scriptores Graeci (with Supplement), as published by Teubner.

C. Pronunciation of Ancient Greek

The de Compositione is not a treatise on Greek Pronunciation, or even on Greek Phonetics. The sections which touch upon these subjects are strictly subsidiary to the main theme; they are literary rather than philological in aim. There was, in fact, no independent study of phonetics in Greek antiquity; the subject was simply a handmaid in the service of music and rhetoric. Hence the reference early in c. 14 to the authority of Aristoxenus “the musician,” and the constant endeavour to rank the letters according to standards of beautiful sound. Still, though Dionysius’ object in describing the way in which the different letters are produced is not scientific but aesthetic and euphonic, much praise is due to the rigorous thoroughness which led him to undertake such an investigation at all. And it has had important incidental results.

One modern authority claims that, notwithstanding difficulties in the interpretation of the de Compositione due either to vague statements in the text or to defective knowledge on our own part, it is possible to reconstruct, with essential accuracy, the “Dionysian Pronunciation of Greek,” or (in other words) the pronunciation current among cultivated Greeks during the fifty years preceding the birth of Christ; while another authority has given a transliteration of the Lord’s Prayer, according to the original text, in the Hellenistic pronunciation of the first century A.D.[64] It is, further, maintained that, thanks to the general progress of philological research, we can in the main reproduce with certainty the sounds (including even the aspirates) actually heard at Athens in the fourth century B.C.—with such certainty, at all events, as will suffice for the practical purposes of the modern teacher.[65]

Two circumstances render it unsafe to lean unduly on Dionysius’ evidence in determining the pronunciation of the earlier Greek period. Although he studied with enthusiasm the literature produced by Greece in her prime, and would certainly desire to read it to his pupils in the same tones as might have been used by its original authors, it is hardly likely that the pronunciation of the language had changed less in three or four hundred years than that (say) of English has changed since the days of Shakespeare.[66] The other circumstance is the uncertainty which attends some of his statements, quite apart from any question of the period which they may be supposed to cover. This uncertainty is due to the fact that there was no science of phonetics in his day, and that consequently his explanations are sometimes obscure, either in themselves or at all events to their modern interpreters. But in many other cases he is, fortunately, explicit and easily understood. One example only shall be given, but that an important one: the pronunciation of ζ. In [144] 9-12, it is clearly indicated that ζ is a double letter, and that it is composed of σ and δ (in that order): διπλᾶ δὲ τρία τό τε ζ̄ καὶ τὸ ξ̄ καὶ τὸ ψ̄. διπλᾶ δὲ λέγουσιν αὐτὰ ἤτοι διὰ τὸ σύνθετα εἶναι τὸ μὲν ζ̄ διὰ τοῦ σ̄ καὶ δ̄, τὸ δὲ ξ̄ διὰ τοῦ κ̄ καὶ σ̄, τὸ δὲ ψ̄ διὰ τοῦ π̄ καὶ σ̄, κτλ. The manuscript testimony is here in favour of σ̄ καὶ δ̄ (rather than the reverse order), and it may be noticed that the similar reading, ὐπασ̅δ̅εύξαισα, is well supported in Sappho’s Hymn to Aphrodite ([238] 9). The statement is not in any way contradicted by the further statements in [146] 5 and [148] 6; and taken together with other evidence (e.g. such forms as συρίσδειν = συρίζειν, κωμάσδειν = κωμάζειν, Ἀθήναζε = Ἀθήνασδε), it seems to establish this as at least one pronunciation of ζ. The actual pronunciation may well have varied at different times and in different places. Some authorities think that in fifth-century Greece the sound was like that of English zd in the word ‘glazed,’ while in the fourth century it roughly resembled dz in the word ‘adze’ (Arnold and Conway, op. cit. pp. 6, 7).

The book which deals most directly with the de Compositione in relation to Greek pronunciation is A. J. Ellis’ English, Dionysian, and Hellenic Pronunciation of Greek, considered in reference to School and College Use. In applying great phonetic skill to the interpretation of Dionysius’ statements, the author of this pamphlet has done much service; but he abandons too lightly any attempt to recover a still earlier pronunciation, and shows an uncritical spirit in so readily believing (p. 4) that Erasmus could be hoaxed in the matter of Greek pronunciation. A more trustworthy work is F. Blass’ Pronunciation of Ancient Greek (translated by W. J. Purton), in which the scientific aids towards a reconstruction of the old pronunciation are marshalled with much force. Arnold and Conway’s Restored Pronunciation of Greek and Latin, and Giles’ Manual of Comparative Philology (pp. 114-118: especially p. 115 for ζ), contain a succinct statement of probable results. There is also a good article, by W. G. Clark, on Greek Pronunciation and Accentuation in the Journal of Philology i. pp. 98-108; with which should be compared the papers by Wratislaw and Geldart in vol. ii. of the same journal. The entire conflict on the subject of Greek pronunciation, as waged by the early combatants in England and Holland, is reflected in Havercamp’s two volumes entitled Sylloge Scriptorum qui de linguae Graecae vera et recta pronuntiatione commentarios reliquerunt, videlicet Adolphi Mekerchi, Theodori Bezae, Jacobi Ceratini et Henrici Stephani (Leyden, 1736), and his Sylloge Altera Scriptorum qui ... reliquerunt, videlicet Desiderii Erasmi, Stephani Vintoniensis Episcopi, Cantabrigiensis Academiae Cancellarii, Joannis Checi, Thomae Smith, Gregorii Martini, et Erasmi Schmidt (Leyden, 1740). Erasmus’ dialogue de recta Latini Graecique sermonis pronunciatione (Basle, 1528) was, in its way, a true work of science in that it laid stress on the fact that variety of symbols implied variety of sounds, and that diphthongal writing implied a diphthongal pronunciation. Attention has lately been directed to the fact that Erasmus claims no originality for his views on this subject, and that he had been anticipated, in varying degrees, by Jerome Aleander in France, by Aldus Manutius in Italy, and (earlier still) by the Spanish humanist, Antonio of Lebrixa (Bywater The Erasmian Pronunciation of Greek and its Precursors Oxford, 1908). It may be noted, in passing, that when enumerating the errors of his Byzantine contemporaries, Antonio mentions that they pronounced Ζ “as a single letter, whereas it was really composite, and stood for SD” (Bywater, p. 20). Among the immediate successors of Erasmus in this field the most interesting, perhaps, is Sir Thomas Smith (1513-1577), who, like Cheke, was one of the “etists” and so incurred the wrath of Stephen Gardiner and drew out that edict which threatened various penalties (including corporal punishment for boys) against the practice of unlawful innovations in the province of Greek pronunciation. It was Smith who, in his treatise de recta et emendata linguae Graecae pronuntiatione (Havercamp, ii. 542), detected a lacuna in the text of C.V. [140] 16 as current in his time, and secured the right sense by the insertion of δύο δὲ βραχέα τό τε ε̄ καὶ τὸ ο̄ after τὸ ω̄ (in l. 17). Echoes, more or less distinct, of the long dispute as to the pronunciation of the ancient classical languages may be heard in such various quarters as: (1) [Beaumont and] Fletcher’s Elder Brother ii. 1, “Though I can speak no Greek, I love the sound on’t; it goes so thundering as it conjur’d devils”; (2) King James I. (in an address to the University of Edinburgh, delivered at Stirling), “I follow his [George Buchanan’s] pronunciation, both of his Latin and Greek, and am sorry that my people of England do not the like; for certainly their pronunciation utterly fails the grace of these two learned languages”; and (3) Gibbon’s reference to “our most corrupt and barbarous mode of uttering Latin.” In modern times a constant effort is being made to get nearer to the true pronunciation of the two classical languages; and (to speak of Greek alone) some interesting side-lights have been shed on the subject by the discovery of Anglo-Saxon or Oriental transliterations (cp. Hadley Essays pp. 128-140, and Bendall in Journal of Philology xxix. 199-201). The application of well-ascertained results to the teaching of Greek pronunciation could be injurious only if it were allowed to impede the principal object of Greek study—contact with the great minds of the past. But an attempt to recapture some part of the music of the Greek language is hardly likely to have this disastrous effect.

D. Greek Grammar

Grammar, like phonetics, was by the ancients often regarded as a part of “music.”[67] It would not, therefore, seem unnatural to his readers that, in a treatise on euphony, Dionysius should continually be referring to the parts of speech (τὰ μόρια τοῦ λόγου). He also uses freely such technical terms of grammar as: πτῶσις, ἔγκλισις, ἀπαρέμφατος, πληθυντικῶς, ὕπτιος, ἀρρενικός, θηλυκός, οὐδέτερος, ἄρθρον, ὄνομα, πρόθεσις, σύνδεσμος, etc. Though himself concerned more immediately with the euphonic relations of words, he is fully alive to the phenomena of their syntactical relations. His remarks on grammatical points show, as might have been expected, many points of contact with the brief treatise of another Dionysius—Dionysius Thrax, who was born a full century earlier than himself. Dionysius Thrax was a pupil of Aristarchus, and produced the earliest formal Greek Grammar. Some interesting hints as to the successive steps in grammatical analysis which had made such a Grammar possible may be found in the second chapter of the de Compositione, where special mention is made of Theodectes, Aristotle, and “the leaders of the Stoic School.” In c. 5, a useful protest is raised against the tyranny of grammar, which so often seeks to control by iron “rules” the infinite variety and living flexibility of language.

The standard edition of Dionysii Thracis Ars Grammatica is that by Uhlig (Leipzig, 1883). The whole question of ancient views on grammar can be studied in Steinthal’s Geschichte der Sprachwissenschaft bei den Griechen und Römern, mit besonderer Rücksicht auf die Logik (2nd ed., Berlin, 1890-91).

E. Sources of the de Compositione

It must strike every reader of the treatise, that Dionysius combines some assertion of originality with many acknowledgments of indebtedness to predecessors. In this there is, of course, no necessary inconsistency. The work covers a wide field, and implies an acquaintance with many special studies. While referring with gratitude and respect to the admitted authorities in these various branches of learning or science, Dionysius claims for himself a certain originality of idea and of treatment. He is among the first to have written a separate treatise on this particular subject, and he is the first to have attempted an adequate treatment of it.[68]

In making these acknowledgments, Dionysius does not specify any Latin writers, nor indeed any recent writers whatsoever. When Quintilian, in the fourth chapter of his Ninth Book, is himself writing a short de Compositione, he mentions “Halicarnasseus Dionysius” and (with special respect) “M. Tullius.”[69] But Dionysius says not a word about Cicero or Horace, although the former was partly and the latter fully contemporary with himself, and although they, like himself, were students of literary composition. As his work on early Roman history shows, Dionysius was not ignorant of Latin; and it is unfortunate that he did not think of comparing Greek writers with Latin. But the comparative method of literary criticism hardly existed in Greek antiquity, notwithstanding the reference to Cicero and Demosthenes in the de Sublimitate, whose author (it may be added here) not only treats of σύνθεσις in two of his chapters, but also tells us that he had already dealt with the subject in two separate treatises.[70]

To his Greek predecessors Dionysius often refers in general terms. For example, they are called οἱ πρὸ ἡμῶν in [140] 7, οἱ πρότερον in [96] 7, and οἱ ἀρχαῖοι in [68] 9. The last term best suggests Dionysius’ habitual attitude, which was that of looking to the past for the finest work in criticism as well as in literature.[71] And so it will be found that, though the de Compositione Verborum contains incidental references to the Stoics and to other leaders of thought, its highest respect seems to be reserved for Aristotle and his disciples Theophrastus and Aristoxenus.[72] But the question of Dionysius’ obligations to his predecessors (and to the Peripatetics particularly) is so large and far-reaching that it must be treated separately elsewhere. Meanwhile, let it be noted how considerably his various writings illustrate, and are illustrated by, the Rhetoric of Aristotle.[73]

As to its originality, the book may well be left to answer for itself. It does not read like a dull compilation. The learning is there, but it is lightly borne, and none can doubt that the writer has long thought over his subject and can give to others the fruits of his reflexions with verve and a contagious enthusiasm. The work has an easy flow of its own, as though it had been rapidly (but not carelessly) written, out of a well-stored mind, while its author was busy with his teaching and with the many literary enterprises to which he so often refers. It must be conceded that a literary critic who deals with so difficult, many-sided, and elusive a subject as that of composition can hardly avoid some errors of detail, since he cannot hope to be a master in all the accessory sciences upon which he has to lean. But we may well be content if he preserves for later ages much invaluable literature and teaching which would otherwise have been lost,—if he himself maintains (amid corrupting influences) high standards in his literary preferences and in his own writing,—and if he sheds a ray of light upon many a hidden beauty of Greek style which would but for him be shrouded in darkness.

Reference may be made to G. Ammon de Dionysii Halicarnassensis Librorum Rhetoricorum Fontibus and to G. Mestwerdt de Dionysii Halicarnassensis in libro de Compositione Verborum Studiis. One section of the subject is also treated in G. L. Hendrickson’s valuable papers on the ‘Peripatetic Mean of Style and the Three Stylistic Characters’ and on the ‘Origin and Meaning of the Ancient Characters of Style’ in the American Journal of Philology vols. xxv. and xxvi.; and in H. P. Breitenbach’s dissertation on The ‘De Compositione’ of Dionysius of Halicarnassus considered with reference to the ‘Rhetoric’ of Aristotle.

F. Quotations and Literary References in the de Compositione

The greatest of all the lyrical passages quoted in the treatise is Sappho’s Hymn to Aphrodite. But great as this is, it does not stand alone. It has companions, if not equals, in the Danaë of Simonides and in the opening of a Pindaric dithyramb. The very preservation of these splendid relics, as of some slighter ones, we owe to Dionysius alone.[74] The total extent of the quotations made in the course of the treatise may be judged from the references given at the foot of the translation: these illustrative extracts form a substantial part of the work they illustrate. The width of Dionysius’ literary outlook may also be inferred from the following roughly-drawn Chronological Table, which (for the sake of completeness) includes some authors who are mentioned but not actually quoted:—

Chronological Table of Authors quoted or mentioned in the De Compositione

B.C.	Epic Poetry.	Elegiac and Iambic.	Lyric.	Tragedy.	Comedy and Satire.
Before 700	Homer Hesiod	...	...	...	...
700-600	...	Archi- lochus	Alcaeus Sappho Stesichorus	...	...
600-500	...	...	Anacreon	...	...
500-400	...	...	Simonides Pindar Bacchylides	Aeschylus Sophocles Euripides	Aristophanes
400-300	Antimachus of Colophon	...	Philoxenus Timotheus Telestes	...	...
300-200	...	[Calli- machus]	...	...	Euphorio Chersonesita
					Sotades
200-100	...	...	...	...	...

B.C.	History.	Oratory and Rhetoric.	Philosophy.	Grammar; Musical and Metrical Science, etc.
Before 700	...	...	...	...
700-600	...	...	...	...
600-500	...	...	...	...
500-400	Herodotus Thucydides	Gorgias Antiphon	Empedocles (verse) Democritus	...
400-300	Ctesias Xenophon Theopompus Ephorus	Isocrates Aeschines Demosthenes Theodectes	Plato Aristotle Theophrastus	Aristoxenus
300-200	...	Hegesias	Epicurus and the Epicureans	Aristophanes of Byzantium
			Chrysippus and the Stoics
200-100	Polybius	...	...	...

To this list might be added the minor historians, of the third and second centuries B.C., who are mentioned together with Polybius in c. 4, and of whom some account will be found in the notes on that chapter: Phylarchus, Duris, Psaon, Demetrius of Callatis, Hieronymus, Antigonus, Heracleides, and Hegesianax. And it will be noticed, further, that the treatise contains a large number of unassigned verse-fragments, which can only be referred, vaguely, to some lyric poet or to the lyric portions of some tragic poet. By such anonymous fragments, as well as by the poems quoted under the names of Sappho and Simonides, we are reminded of the many lost works of Greek literature and of the happy surprises which Egypt or Herculaneum or the Sultan’s Library may still have in store for us. If the quotations as a whole—identified and unidentified, previously known and previously unknown—are passed in review, it will be found that Dionysius has given us a small Anthology of Greek prose and verse. While strictly relevant to the main theme, his illustrations are chosen with so much taste, and from so wide a field of study, that (to adapt his own words) οὐκ ἀηδὴς ὁ λόγος ἐγένετο πολλοῖς ὥσπερ ἄνθεσι διαποικιλλόμενος τοῖς ἐαρινοῖς.[75]

Two prose-writers mentioned by Dionysius seem to invite special comment: Polybius and Hegesias. It is not without a kind of shock that we find the great historian Polybius classed, along with Phylarchus and the rest, among writers whose works no man can bring himself to read from cover to cover.[76] But we have to remember that the judgment is passed solely from the standpoint of style; and from this restricted standpoint, it can hardly be said that subsequent critics have ventured to reverse it and to maintain that Polybius is (to use the modern expression) an eminently “readable” author. Let one modern estimate be quoted, and that from a writer who appreciates fully the greatness of Polybius’ theory of history, and who, on the other hand, is not concerned to vindicate the soundness of Dionysius’ judgment: “Unfortunately, his [Polybius’] style is a serious deterrent to the reader. We long for the ease, the finished grace, the flowing simplicity of Herodotus; or again, for the terse and rapid phrase of Thucydides, the energy, the precision of each single word, the sentence packed with thought. Polybius has lost the Greek artistic feeling for writing, the delicate sense of proportion, the faculty of reserve. The freshness and distinction of the Attic idiom are gone. He writes with an insipid and colourless monotony. In arranging his materials he is equally inartistic. He is always anticipating objections and digressing; he wearies you with dilating on the excellence of his own method; he even assures you that the size and price of his book ought not to keep people from buying it. Admirable as is the substance of his writing, he pays the penalty attaching to neglect of form—he is read by the few.”[77]

Hegesias is not only mentioned, but quoted, in the treatise. A few detached sentences are given from his writings, and one longer passage. In c. 4 Dionysius rewrites a brief extract from Herodotus in utter defiance of the customary rules (or practices) of Greek word-order, and then exclaims, “This form of composition resembles that of Hegesias: it is affected, degenerate, enervated.” He proceeds: “In such trumpery arts the man is a hierophant. He writes, for instance, ‘After a goodly festival another goodly one keep we.’ ‘Of Magnesia am I, the mighty land, a man of Sipylus I.’ ‘No little drop into the Theban waters spewed Dionysus: O yea, sweet is the stream, but madness it engendereth.’”

In c. 18 Dionysius illustrates the beauty of prose-rhythm from Thucydides, Plato, and Demosthenes. He then assigns to Hegesias a bad pre-eminence among writers who have neglected this essential of their art. Quoting a passage of some length from his History, he asks how it compares with Homer’s description of a similar scene; and he holds the vast superiority of the latter to be due ‘chiefly, if not entirely, to the difference in the rhythms.’ In the words just cited there is obviously much exaggeration. But we must allow for Dionysius’ preoccupation in this treatise (cp. τοῦτ’ ἦν σχεδὸν ᾧ μάλιστα διαλλάττει ποιητής τε ποιητοῦ καὶ ῥήτωρ ῥήτορος, τὸ συντιθέναι δεξιῶς τὰ ὀνόματα, [92] 18-20), and must, at any rate, try to discover wherein the main defect of Hegesias’ rhythms is supposed to lie. It is probable that no single thing in the passage offends the ear of Dionysius so much as the double trochees (or their metrical equivalent) which are found at the end of so many of the clauses. This double trochee, or dichoree, is found in its normal form (– ᴗ – ⏒) at the end of such cola as those which terminate in: τοῖς ἀρίστοις, καὶ τὸ πλῆθος, εἰς τὸ τολμᾶν, τῇ μαχαίρᾳ, καὶ Φιλωτᾶς, καὶ τὸ χρῶμα, σκαιὸν ἐχθρόν. The metrical equivalent ᴗ ᴗ ᴗ – ⏒ occurs in such instances as: πρότερον οὕτως, ἕνεκα πρᾶξαι, κατακοπῆναι, καθικετεύων. It is interesting to observe that this final dichoree is regarded both by Cicero and by Quintilian as characteristic of the Asiatic orators.[78] Let it be added that, in the extract from Hegesias, the dichorees are not confined to the close of clauses but occur freely in other positions, while many of the sentences are short and the reverse of periodic; and it will be granted that Cicero has good ground for calling attention to the jerky, or staccato, character of the style in question. In the Orator (67. 226) the effect of Hegesias’ writing is thus described: “quam (sc. numerosam comprehensionem) perverse fugiens Hegesias, dum ille quoque imitari Lysiam volt, alterum paene Demosthenem, saltat incidens particulas.” And his manner is amusingly parodied in one of the letters to Atticus (ad Att. xii. 6): “de Caelio vide, quaeso, ne quae lacuna sit in auro: | ego ista non novi; | sed certe in collubo est detrimenti satis. | huc aurum si accedit |—sed quid loquor? | tu videbis. | habes Hegesiae genus! quod Varro laudat.”[79] Two further specimens (not given by Dionysius) of Hegesias’ style will add point to Cicero’s parody. The first is preserved by Strabo (Geogr. 396): ὁρῶ τὴν ἀκρόπολιν | καὶ τὸ περιττῆς τριαίνης | ἐκεῖθι σημεῖον· | ὁρῶ τὴν Ἐλευσῖνα, | καὶ τῶν ἱερῶν γέγονα μύστης· | ἐκεῖνο Λεωκόριον· | τοῦτο Θησεῖον· | οὐ δύναμαι δηλῶσαι | καθ’ ἓν ἕκαστον. The other specimen is quoted by Photius (Bibl. cod. 250) from Agatharchides, the geographer of Cnidus: ὅμοιον πεποίηκας, Ἀλέξανδρε, Θήβας κατασκάψας, ὡς ἂν εἰ ὁ Ζεὺς ἐκ τῆς κατ’ οὐρανὸν μερίδος ἐκβάλλοι τὴν σελήνην. ὑπολείπομαι γὰρ τὸν ἥλιον ταῖς Ἀθήναις. δύο γὰρ αὗται πόλεις τῆς Ἑλλάδος ἦσαν ὄψεις. διὸ καὶ περὶ τῆς ἑτέρας ἀγωνιῶ νῦν. ὁ μὲν γὰρ εἷς αὐτῶν ὀφθαλμὸς ἡ Θηβαίων ἐκκέκοπται πόλις.[80]

It is quite clear, from his express statements, that Dionysius, in his criticisms, has in view, mainly if not entirely, the bad rhythms of Hegesias. But the passages which he quotes seem open to criticism on other grounds as well. The long extract in c. 18 contains metaphors which might well seem violent to the Greeks, who allowed themselves less licence than the moderns do in this direction (e.g. ἡ μὲν οὖν ἐλπὶς αὕτη συνέδραμεν εἰς τὸ τολμᾶν, and τοὺς δ’ ἄλλους ὀργὴ πρόσφατος ἐπίμπρα); and it is high-flown expressions of this kind which the author of the de Sublimitate has in view when he writes: τά γε μὴν Ἀμφικράτους τοιαῦτα καὶ Ἡγησίου καὶ Ματρίδος· πολλαχοῦ γὰρ ἐνθουσιᾶν ἑαυτοῖς δοκοῦντες οὐ βακχεύουσιν ἀλλὰ παίζουσιν (iii. 2). False emphasis, too, and a general desire to purchase notoriety by the cheap method of eccentric word-order, would appear to be implied in Dionysius’ own parody in c. 4 ([90] 15-19). For example, Ἀλυάττου and ἐθνῶν, though not in themselves important, are assigned prominent positions at the beginning and the end of the sentence. But the greatest of all the defects of Hegesias—especially when compared with Homer—is a certain vulgarity of tone.

The contrast drawn between Hegesias and Homer may seem overstrained, but it is eminently characteristic of Dionysius. Homer was to him the great pure fount of Greek, and his own constant desire was “antiquos accedere fontes.” Hegesias, on the other hand, typifies to him the decline in Greek literature which followed the death of Alexander, whose exploits he records with so feeble a magniloquence. And yet the curious thing is that Hegesias, who lived probably in the earlier part of the third century, aspires (as Cicero tells us) to copy Lysias. But while endeavouring thus to imitate one of the most Attic of the Attic writers, he came, by the irony of fate, to be regarded as the founder of the degenerate Asiatic school: Ἡγησίας ὁ ῥήτωρ, ὃς ἦρξε μάλιστα τοῦ Ἀσιανοῦ λεγομένου ζήλου, παραφθείρας τὸ καθεστηκὸς ἔθος τὸ Ἀττικόν (Strabo Geogr. xiv. 1. 41).[81] In the terms “Attic” and “Asiatic” there often lurks some confusion of thought, as well as no little prejudice and rhetorical animosity. But of Dionysius, as compared with Hegesias, it is clearly within the mark to say that, though he lived two centuries later, he has vastly more of the true Attic feeling for purity of style; and that, though he may himself have cherished wild dreams of turning back the tide of language, yet in league with some leading Romans of his day he did good service by showing how the best Attic models may hold out to future ages shining examples of the skill and beauty which all men should strive after in handling the language of their birth.

For Dionysius in relation to contemporary Romans, and to the struggle between Asianism and Atticism, reference may be made to Dionysius of Halicarnassus: the Three Literary Letters pp. 34-49.

G. Manuscripts and Text

The chief authorities for the text of the de Compositione are indicated in the following list of abbreviations employed in the apparatus criticus of the present edition:—

Siglorum in notulis criticis adhibitorum Index

F = cod. Florentinus Laurentianus lix. 15. saec. xii.
P = cod. Parisinus bibl. nat. 1741. saec. xi. (x.).
M = cod. Venetus Marcianus 508. saec. xv.
V = cod. Vergetii Parisiensis bibl. nat. 1798. saec. xvi.

E = Διονυσίου Ἁλικαρνασέως τοῦ περὶ Συνθέσεως Ὀνομάτων Ἐπιτομή. saec. inc.

R = Rhetor Graecus (Scholiasta Hermogenis περὶ ἰδεῶν, i. 6). saec. inc.

a = editio princeps Aldi Manutii (Aldi Manutii Rhetores Graeci, tom. i.), Venetiis. 1508.
s = editio Roberti Stephani, Lutetiae. 1547.
r = exemplum Reiskianum, Lipsiae. 1775.
Us = exemplum ab Usenero et Radermachero Lipsiae nuper editum.

The Florentine manuscript (F) contains, besides certain writings of other authors, the following works of Dionysius: (1) the essays on Lysias, Isocrates, Isaeus, and Dinarchus: and (2) the de Compositione Verborum (as far as the words πειρατέον δὴ καὶ περὶ τούτων λέγειν ἃ φρονῶ in c. 25). The Paris manuscript 1741 (P) is the famous codex which contains not only the de Comp. Verb., but also Aristotle’s Rhetoric and Poetics, Demetrius de Elocutione, Dionysius Halic. Ep. ad Amm. II., De Vet. Scr., etc. Some notes upon the manuscript are given in Demetrius on Style pp. 209-11; and the editor has examined it once more at Paris for the purposes of the present recension. The remaining manuscripts are considerably later than F and P. M belongs to the fifteenth century, and V was copied by the Cretan calligrapher Ange Vergèce (as he was called in France) in the sixteenth century. The edition of Robert Stephens is based upon V. In the Journal of Philology xxvii. pp. 83 ff., there is a careful collation, by A. B. Poynton, of “Some Readings of MS. Canonici 45” (C: sixteenth century) in the Bodleian Library, with regard to which the collator says: “Despite the care with which the work is done, the manuscript is not of much value as a presentation of the Florentine tradition, since F exists and the writer of C is rather a διασκευαστής than a copyist. The interest of the manuscript is antiquarian and bibliographical.... It is a copy made at some time in the sixteenth century, probably after 1560. It is based on the Florentine MS. with variae lectiones and marginal notes. It has not the appearance of being a mechanical copy: rather it seems to be the work of a scholar who was conversant with the MSS. of the treatise and, while he was aware of the importance of the Florentine MS., saw that in many cases it needed to be corrected.”

The dates of the Epitome and of the Rhetor Graecus are uncertain. But both are early and highly important authorities. The latter quotes c. 14 only of the treatise, but the quotation enabled Usener to show that the text of F agreed in the main with that of the Rhetor and of the Epitome. The result was to enhance greatly the authority of F, with which earlier editors had merely an indirect and imperfect acquaintance. But by a not unnatural reaction against the excessive attention paid to what may be called the P group (PMV: though M and V sometimes coincide with F against P), Usener is inclined too readily to follow F, or even E, when standing alone. Still, while the readings supported only by F, or E, or P should be carefully scrutinized and independently judged, the concurrent testimony of FE and any other MS. is very strong indeed.

Two passages taken almost at a venture (say, the first twenty lines of c. 12 and the last twenty of c. 19) would be enough to show that neither F nor P can be exclusively followed, and that Usener himself is often (more often than is indicated in this edition) driven to desert F, which in fact contains, in these or other places, a large number of impossible or even absurd readings.[82] Where, however, there are genuine instances of various readings (as εὐκαιροτέραις: εὐροωτέραις in the last of the passages just specified), it seems best to follow F (especially when supported by other authorities), even though the hand of an ingenious early scholar may sometimes with reason be suspected.[83]

One reason for accepting with reserve the unsupported testimony of F is that its evidence is sometimes far from sound in regard to quotations from authors whose text is well established from other sources. In the principal quotations from Pindar and Thucydides this defect is not so manifest; and it may even be claimed that its text of the Pindaric dithyramb, and of the Herodotus extract on p. 82, is distinguished by many excellent features, though not so many as Usener was at first inclined to claim in the case of the Pindar. But in the extract from the Areopagiticus of Isocrates which is given in c. 23, the text presented by F (as compared with that presented by P) seems to suggest that, in dealing with Dionysius’ own words as well as with his quotations, the transcriber may have felt entitled to make rather free alterations on his own account. In order to provide readers with the means of judging for themselves, the critical apparatus has been made specially full at this point.[84]

Usener’s text of the de Compositione deserves the highest respect: it is the last undertaking of one of the greatest philologists of the nineteenth century, and every succeeding editor must find himself deep in its debt. Its record of readings is full to exhaustiveness. In the present edition less wealth of detail is attempted (especially in regard to F and R), though all really important and typical variations have, it is hoped, been duly registered, and particular attention has been paid to the minute collation of P. But apart from the correction of misprints (as on pp. [124] 13, [132] 23, [250] 7), it is hoped that the following among other readings will commend themselves (on an examination of the sections of the Notes or Glossary in which they are defended) as superior to those adopted by Usener (and indicated here in brackets) from conjecture or on manuscript authority: [64] 11 (σοὶ omitted), [70] 5 (εὖ τί), [78] 17 (παλαιαί), [80] 13 (παιδικόν), [94] 13 (προβαῖεν), [94] 16 (σπουδάζεσθαι), [98] 20 (οἷά τινα), [106] 13 (εὖ ἢ), [132] 20 (θηρᾶν), [142] 9 (σπανίζει), etc.

H. Recent Writings connected with the de Compositione

A full bibliography, covering not only the de Compositione of Dionysius but his rhetorical and critical works generally, is given in the present editor’s Dionysius of Halicarnassus: the Three Literary Letters (published in January 1901), pp. 209-219. The following are (in chronological order) the early editors who have done most to further the study of the de Compositione: Aldus Manutius (editio princeps), Robertus Stephanus, F. Sylburg, J. Upton, J. J. Reiske, G. H. Schaefer, and F. Goeller. Much interest still attaches to C. Batteux’ publication (1788): Traité de l’arrangement des mots: traduit du grec de Denys d’Halicarnasse; avec des réflexions sur la langue française, comparée avec la langue grecque. The translation is too free and based on too poor a text to meet the needs of exact scholarship. But the Réflexions (which accompany the translation, in vol. vi. of the author’s Principes de littérature) are full of suggestive remarks. Another excellent literary study of Dionysius is that of Max. Egger: Denys d’Halicarnasse: essai sur la critique littéraire et la rhétorique chez les Grecs au siècle d’Auguste (Paris, 1902). As its title indicates, this volume takes a wide range; and it reveals that full competence in these matters which it is natural to expect from the son of Émile Egger. A short general account, by Radermacher, of Dionysius’ critical essays will be found in Pauly-Wissowa’s Realencyclopädie vol. v.

The first volume of Usener and Radermacher’s text was included in the bibliographical list mentioned above. In 1904 appeared the second volume, containing the de Compositione and some other critical writings of Dionysius (Dionysii Halicarnasei opuscula ediderunt Hermannus Usener et Ludovicus Radermacher. Voluminis sec. fasc. prior. Lipsiae, 1904). The second volume is on a par with the first, which was welcomed, as a notable achievement, in the Classical Review xiv. pp. 452-455, where also attention was drawn (p. 454 a) to a questionable emendation previously introduced by Usener into the text of the de Imitatione. This emendation is withdrawn in Usener’s second volume—a fact which may be mentioned as one proof among many that his tendency was to grow more conservative and, in particular, more attentive to the testimony of P 1741. The titles of A. B. Poynton’s articles on Dionysius are: “Oxford MSS. of Dionysius Halicarnasseus, De Compositione Verborum” (Journal of Philology xxvii. pp. 70-99), and “Oxford MSS. of the Opuscula of Dionysius of Halicarnassus” (Journal of Philology xxviii. pp. 162-185). Among other useful subsidia lately published may be mentioned: W. Kroll’s “Randbemerkungen” in Rhein. Mus. lxii. pp. 86-101, and Larue van Hook’s Metaphorical Terminology of Greek Rhetoric and Literary Criticism (Chicago, 1905). R. H. Tukey (Classical Review, September 1909, p. 188) makes the interesting suggestion that “the De Compositione belongs chronologically between the two parts of the De Demosthene.” The use of the present tense δηλοῦται, in C.V. [182] 8 may be held to countenance this view.

In some recent books of larger scope it is pleasant to notice an increased appreciation of the high value of the work done by Dionysius in the field of literary criticism. Certain of these estimates may be quoted in conclusion. R. C. Jebb, in the Companion to Greek Studies p. 137: “The maturity of the ‘Attic revival’ is represented at Rome, in the Augustan age, by the best literary critic of antiquity, Dionysius of Halicarnassus.” A. and M. Croiset Histoire de la littérature grecque v. p. 371: “Les uns et les autres [les contemporains et les rhéteurs des âges suivants] appréciaient avec raison l’érudition de Denys, la justesse de son esprit, sa finesse dans le discernement des ressemblances et des différences, la solidité de sa doctrine, son goût dans le choix des exemples. De plus, ils se sentaient touchés, comme nous et plus que nous, par la vivacité de ses admirations, par cette sorte de foi communicative, qui faisait de lui le défenseur des traditions classiques.” Wilamowitz-Moellendorff Die griechische Literatur des Altertums pp. 102 and 148: “Von unbestreitbar hohem und dauerndem Werte ist die andere Seite der rhetorischen Theorie und Praxis, die sich auf den Ausdruck erstreckt, die Stilistik.... Es ist ein hohes Lob, dass er (Dionysios von Halikarnass) im Grunde dieselbe stilistische Überzeugung vertritt wie Cicero, und wir sind ihm für die Erhaltung von ungemein viel Wichtigem zu Dank verpflichtet; seine Schriften über die attischen Redner und über die Wortfügung sind auch eine nicht nur belehrende, sondern gefällige Lekture.” J. E. Sandys History of Classical Scholarship i. p. 279: “In the minute and technical criticism of the art and craft of Greek literature, the works of Dionysius stand alone in all the centuries that elapsed between the Rhetoric of Aristotle and the treatise On the Sublime.” G. Saintsbury History of Criticism i. pp. 136, 137, 132: “Dionysius is a very considerable critic, and one to whom justice has not usually, if at all, yet been done.... A critic who saw far, and for the most part truly, into the proper province of literary criticism.... This treatise [sc. the de Compositione], if studied carefully, must raise some astonishment that Dionysius should have been spoken of disrespectfully by anyone who himself possesses competence in criticism. From more points of view than one, the piece gives Dionysius no mean rank as a critic.” S. H. Butcher Harvard Lectures on Greek Subjects pp. 236, 239: “Of his fine perception of the harmonies of Greek speech we can entertain no reasonable doubt.... We cannot dismiss his general criticism as unsound or fanciful. The whole history of the evolution of Greek prose, and the practice of the great masters of the art, support his main contention.” With these extracts may be coupled one from the Spectator of March 23, 1901: “In this treatise Dionysius reviews and attempts to explain the art of literature. It is a brilliant effort to analyse the sensuous emotions produced by the harmonious arrangement of beautiful words. Its eternal truth might make it a textbook for to-day.”

In the Notes and Glossary, as in the Introduction, references are usually given to the lines, as well as the pages, of the Greek text here printed: e.g. 80 7 = page 80 line 7 of the De Compositione.—The following abbreviations are used in referring to volumes already issued by the editor:—

D.H. = ‘Dionysius of Halicarnassus: the Three Literary Letters.’
Long. = ‘Longinus on the Sublime.’
Demetr. = ‘Demetrius on Style.’