S.^v*V^-'2^A^^^i^-5»^-^'=>^'^ -* * LIBRARV OF THK University of California. OTKT OK n Receive^ Accession No. 7 J f/Z ■ Oms No B, .u ^ ?r ■X' THE HISTORY OF THE ENGLISH PARAGRAPH A DISSERTATION PRESENTED TO THE FACULTY OF ARTS, LITERATURE, AND SCIENCE, OF THE UNIVERSITY OF CHICAGO, IN CANDIDACY FOR THE DEGREE OF DOCTOR OF PHILOSOPHY BY EDWIN HERBERT LEWIS CHICAGO E\)t ^tlnibersitg of ai:l)irago ^ress 1894 7d "f t^ CONTENTS. mAii^^ Preface ----.-..-. g CHAPTER I. The Mechanical Signs of the Paragraph - - - - - ii CHAPTER II. Rhetorical Theories of the Paragraph ----- 20 CHAPTER III. Paragraph-Length and Sentence-Length - - - - - 34 CHAPTER IV. Recent Investigations in Prose- Form: their Bearing on the His- tory of the Paragraph ------- 52 CHAPTER V. Alfred to Tyndale -..--. . . - 66 CHAPTER VI. Tyndale to Temple -------- 75 CHAPTER VII. Temple to De (2uincey - - - - - - - -104 CHAPTER VIII. De Quincey to Holmes - - - - - - - 137 CHAPTER IX. The Prose Paragraph : Summary - - - - - 169 Bibliography - '- - - - - - - - 179 APPENDIX. Notes on the Verse Paragraph of Middle English - - - 185 3 OF THB ^ .UNIVERSITY PREFACE. Historically considered, the word paragraph means {a) a marginal character or note employed to direct attention to some part of the text; {l>) a character similar to {a), but placed in the text itself ; (c) the division of discourse introduced by a paragraph mark or by indentation, and extending to the next paragraph mark or the next indentation ; {d) the rhetorical paragraph, that is, {c) developed to a structural unit capable of organic internal arrangement. The plan of the present essay is to discuss, in the first chapter, {a) and {b) and other mechanical signs of the paragraph ; in the second chapter to introduce {c) for the purpose of further defi- nition ; in the next seven chapters to show the historical devel- opment of [c) in English prose, first by a statement of the general development, then by a particularized account according to periods, then by a summary of this account ; lastly, in an appen- dix, to offer a few incomplete notes on the development of (c) in Middle English verse. It is a pleasure to acknowledge here my indebtedness, first to Professor W. D. McClintock, who approved the choice of subject, and made most searching and suggestive comments upon the whole course of the treatment; and to Professors F. A. Blackburn, W. C. Wilkinson, and A. H. Tolman, for many helpful criticisms. Professor L. A. Sherman, of the University of Nebraska, gener- ously furnished me with certain statistics, noted in the text by the parenthesis (Sherman). Mr. G. W. Gerwig, of Allegheny, Pa., kindly supplied me in advance with the results of his research concerning the decrease of predication, — research pursued under Professor Sherman's direction. I have quoted freely from his results, using as reference mark the parenthesis (Gerwig). In such 5 CHAPTER I. THE MECHANICAL SIGNS OF THE PARAGRAPH. The various signs of the paragraph, as they appear in English manuscripts and English books, are a legacy from classical scribes, and must be studied in the light of their origin. The paragraph {napay pathos {ypajxixrj), is the oldest mark of punctuation in Greek manuscripts. It first occurs as a horizontal stroke (sometimes with a dot over it), placed at the beginning of a line, just beneath the first two or three letters. It indicated that a sentence, or some longer division of the text, was ended in the underscored line. The mark thus distinguished the close of one section rather than the beginning of another. Instead of the horizontal mark the wedge [hnvkrj), or the hook iKopoiVLid felled fpirit, to enter into the leaft &= most exact differences of caufes : a fruite no doubt of the exceeding tranquillity, and ferenity of his minde : which being no icayes charged or incumbred, eyther tvith feares, remorfes, or fcruples, but hauing beene noted for a fnan of the purcft goodneffe, without all fiction, or affectation, that hath raigned or lived: made his minde continually prefect and en tyre: he likewife approached a degree neerer vnto Owiftianity , and becatne as Agrippa faid vnto St. Paul, Halfe a Chriftian ; holding their Religion and Law in good opinion; and not only ceafing perf edition, but gluing way to the aduancement of Chriftians. Lord Brooke and Spenser are perhaps the two greatest offenders in this matter of the confusion of the period and the paragraph. At last the Latinists came to see that their units of thought were too large to be developed in any one sentence of an unin- flected language. The later Latinists were hurried on to this conclusion by the excesses of certain of their own number. They found it impossible to read some of Clarendon's clause-heaps, always been something of a tyrant, it is folly to imply that our old authors, so scrupulous about most things, could not have controlled the punctuation of their printed books. The authors of the sixteenth century did make paragraphs in their manuscripts, for the manuscripts that we have show them. If the printer tampered with the paragraphing as he did with the punctuation, why then, there is nothing for it but to hold the author responsible for not correcting him. 46 HISTORY OF THE ENGLISH PARAGRAPH. nav, even some of Burton's defied tlieiii ; and doubtless more than one classicist began to remember passages in their own beloved Hooker that had once passed for j)r()funditv. l)nt now began to look like mere tangle. Mr. Saintsburv has ai)plied to these inextricable sentences of Clarendon and Burton the name of ''sentence-and-paragraph heap"' — a name hardly less awkward than the thing itself. This is not to be confused, by the way, with the single-sentence paragraph, which may or may not coin- cide with it — does so in early prose often, in modern very rarely. A better word for what Mr. Saintsbury means is " clause-heap," a term that he employs in his preface to Bur- ton in the recent second volume of Craik.^ In the "heap," "clause is linked on to clause till not merely the grammatical but the philosophical integer is hopelessly lost sight of in a tangle of jointings and appendages." As we said before, it is not the writers of the most hopeless clause-heaps that write the largest number of paragraphed sentences ; the " heap " belongs chiefly to the Latinists. When, however, one of Clarendon's heaps is paragraphed, the result is something disheartening. The first single-sentence paragraph in the edition of 17 12 (p. 4) has 242 words; the eighth (p. 28) has 166, and the thirteenth (p. 45) has 195. As authors like these, able men, though slow to put them- selves in touch with the people, began to perceive the hopeless- ness of their self-appointed task, they began to shorten the sen- tence, retaining the paragraph. The wide popularity of the new school of vernacular writers — if we may speak of Bunyan as belonging to any school — inspired literary men with the new desire to reach a larger public. Authors began to put them- selves in the place of their readers, and write as if to an average man. Soon the superiority of French prose began to be felt as a vehicle for the expression of the clearer, more straightforward, ' Knglish Prose Style, in Spcciiiiens of English Prose, p. xix. Hislory of Elizabethan Literature, p. 42, et al. ^Craik's English Prose, vol ii., p. 117. I PARA GRAPH- LENG TH AND SENTENCE-LENG TH. 4 7 less subtle phases of thought. From this time on, the develop- ment of the modern paragraph is a matter of degree of skill rather than of stylistic method. Such, in the rough, is the history of the paragraph in the most critical period of its history. The particulars of this period will be given in Chapters VI. and VII. § 4. One other general question may properly receive considera- tion here : whether the length of the paragraph follows any rhythmical law, as, for instance, one that renders the average length a constant quantity, in successive large groups of para- graphs. To illustrate, will two books written in the same genre of composition, by the same author, yield anything like the same paragraph averages ? This question is rendered the more interesting by the recent investigations of Professor Sherman and Mr. Gerwig, to the effect that the sentence-length, the percentage of predications to the period, and the percentage of simple sentences, each tends to be constant in successive large groups of sentences, as of 500. In his discussions of the constancy of the sentence-length, however, Professor Sherman seems to give hardly weight enough to the differences caused in an author's style by time. He mentions several instances' where the sentence-length remains unchanged by change of years ; and my own observations have furnished others equally notable, none more so than that of Swift, who varies not a whole word in twenty-eight years. But Sherman gives no exceptions. Nay, he says, " Even Carlyle showed no change for worse or better, in respect to sentence proportions, between the Edinburgh Essays and his Frederick the Great.'''' But it should likewise be said that between the Essays and Frederick the average sank (in ihe Revolution) fully one-third. In the formative period of our prose similar changes are very common. Sidney's sen- tence dropped in five years from 75 to 3S. ' University Studies, I.. No. 4, p. 349. I 48 HISTORY OF THE ENGLISH PARAGRAPH. We approach the'ii^eneral question of the constancy of the paragraph, with an author as far back as Browne. We find between the Hydrotaphia and tlie Religio a difference of seven sentences to the paragraph, inalcing the word-length of the Religio nearly thrice that of the Hydrotaphia. No constancy here. We try Taylor's Liberty of Prophesying, comparing the sec- tions, beginning with the fourth. The word-averages run : 480.75,535.71, 78S. 33, 563.14, 648.66, 518.00, 450.00,681.75, 294.57, 418.92, 193.71. The paragraph-length by sentences runs: 8.37,7.46, 14.16, 11.26, 10.83, 9-33. 9-i6, 14.00. 5.71, 9.76, 4.78. Since the sections vary from 10,000 words to 3,000, we feel that the sentence averages are not so bad as we expected. Such averages as 535, 563, 518, or 9.33, 9.16, 9.76, show at least interesting coincidences. We hardly get anything more to the purpose before Dryden. Cowley, for example, varies wildly in his essays. Dryden's Satire yields 256 words, while two combined essays, Translation and the Parallel between Poetry and Painting, yield 277 words. Defoe's Essay on Projects shows 1.90 sentences, and Crusoe 1.87, which is delightfully close ; but the sentence so shoots u{) in Crusoe as to make the word-length of the paragraph thrice as great heie as in the Essay. Swift's sentence, as we have seen, stays at 40, but Gulliver shows a paragraph of 234 words, against 185 in the Tale of a Tub. Johnson helps us in the word- length, showing 102 for the Rambler, 92 for Rasselas ; but mean- time the sentence has gone down a third. Hume is not unsatis factory. The first 105 paragraphs of the History (26,197 words) yield an average of 249 words; the next 95 paragraphs (21,578 words) yield an average of 226 words. We try the Vicar of Wake- field by chajiters, not expecting too much. The word results, omitting fractions, run : 177, 183, 289, 161, 309, 302, 181, 298, 570, 236, 106, 215, 232, 344, 209, 182, the sentence average run- ning. 27, 35. 28, 31, 26, 28, 25, 28, 25, 24, 25, 21, 22, 29, 26, 28. If we average the averages of the first eight chapters against those of the second eight, we shall have 237.96 words or 8.30 sentences PARAGRAPH-LENGTH AND SENTENCE-LENGTH. 49 for the first eight, and 262.07 words, or 10.23 sentences for the second eight. Evidently Dryden and Hume are the only men thus far on whom we can put much reliance. Coming to the present century we examine first Ivanhoe by chapters. The averages run: 129, 73, 106, 62, 56, 55, 98, 70, 61, 53. The sentence average is more stable, thus: 2.92, 2.03, 2i-^l, 2.19, 1.88, 2.16, 2.61, 2.25, 2.13, 2. II. Evidently Scott clung with some monotony to the ideal of two sentences a para- graph. The first five Essays of Elia yield the following results : 171, 134, 230, 147, 125. The averages by sentences are: 7.12, 5.00, 8.00, 5.57, 5-94- We try Irving's Sketch Book, the first five sketches. Result: 157, 140, 137, 83, 104. The averages by sentences are : 5.14, 6.25, 4.94, 3.33, 3-65. Thus far our own century is no improvement — if improvement it be called — on the eighteenth. We try Macaulav ; the History of England by volumes. Results: 258.11,251.52,325.44,336.50,306.90. This is remark able. The averages for the first two volumes are practically the same. Here the writer was governed bv something very like a rigid rhythmical law. A similar, but less strong, rhythmical sense, appears in the last three volumes. But why the sudden rise between the second and third volumes? Two reasons sug- gest themselves. As Mr. Stephen somewhere remarks, Macau- lav's fullness of knowledge began to hamper him a little in the later volumes. In other words, he became somewhat verbose from plenitude of things to say. Since, now, Macaulay wrote primarily with the paragraph unit, the diffuseness would naturally ffec5t unit first. He would naturally keep to his sentence length — or does so, at any rate — but would use more proposi- tions to amplify a given integral thought. Another reason, though perhaps rather remote, suggests itself. Volume two was finished by 1848. Four years later (July, 1852), after the mate- rials for the third volume were collected and partly written up, Macaulay broke down in health from the disease that finally ended 50 HISTORY OF THE ENGLISH PARAGRAPH. his life. For months before that time there are ominous passages in his journal and letters, complaining that the task of composi- tion is a burden, that he is no longer capable of vigorous exertion. Now I do not wish to be understood as maintaining the existence of any very close connection between paragraph-length and heart-disease. But a tired man is likely to be loquacious, if he tries to talk, and when a writer has incomplete control of his brain he is likely to be at first diffuse in his composition, and later, incoherent. I am aware that Trevelyan says of Macaulay : "The habit of always working up to the highest standard within his reach was so ingrained in his nature, that, however sure and rapid might be the decline of his physical strength, the quality of his productions remained the same as ever. Instead of writing worse, he onlv wrote less. Compact in form, crisp and nervous in style, these five little essays are everything which an article in an Encyclopaedia should be." ' The five essays referred to are : Atterbury, 1853, Biinyan, 1856, Goldsmith, 1856, Doctor Johnson, 1856, William Pitt, 1859. It can hardly be granted that these essays are as compact and crisp in style as the earlier essays. It seems to me that proof enough to invalidate Trevelyan's position on this point lies in the fact that Mr. Gerwig found that the per- centage of simple sentences in the early essays is much higher than in these later ones. It is worth while to quote Mr. Gerwig's figures, which show a steady decrease in the percentage of simple sentences. Now this decrease took place as Macaulay's physical strength failed ; the figures are therefore favorable to the theory I have set concerning the rise in paragraph-length. When Number Av. Pred. Simple Work. written. periods. per period, sent. flacaulay,^ Royal Soc. of Literature 1823 100 2.03 44 Dante 1824 100 2.15 38 Milton 1825 895 2.07 38 Machiavelli 1827 693 1.88 47 History, Essay on 1828 719 2.18 40 ' Life and Letters of Lord Macaulay, p. 664. ' University Studies, Vol. II., No. i, p. 22. PARAGRAPH-LENGTH AND SENTENCE-LENGTH. 51 Work. Dryden UArblay Addison Atterbury Bunyan Goldsmith When Number Av. Pred. Simple written. periods. per period. sent. 1828 100 2.65 29 1843 918 2.31 32 1843 I33I 2.22 32 1853 240 2.35 34 1854 245 2.19 31 1856 263 2.29 33 Average 2.17 36 We shall hardly find another author as stable in his averages as Macaulay. Carlyle shows pretty nearly the same average in two books, Sartor (166.90) and the Revolution (160.31) ; but the average in an early essay, Richter, 1827, is 250.62. Emerson's Ai7ierican Scholar yields 184.60, the Divinity. School Address 210.91 (10.10 sentences) and Self Reliance 201.12 (10.04 sentences). We may not, therefore, conclude from the small number of paragraphs we have been able to examine, that the paragraph- length is relatively as constant as the sentence-length.' But in Macaulay and proportionately in authors of regular methods, there is a general tendency toward approximate uniformity in the paragraph-averages of different sections of work. CHAPTER IV. RECENT INVESTIGATIONS IN PROSE-FORM. I'lIEIR BEARING ON THE HISTORY OF THE PARAGRAPH. § I- The recent investigations that have most bearing on the his- tory of the paragraph are those of Professor L. A. Sherman/ on the questions of literary sentence-length in English prose, the coordination, subordination, and suppression of clauses, and the new articulation of clauses. Professor Sherman has demonstrated that the English sentence has dropped about one-half its length since Shakspere's time ; he holds that in the matter of connectives, our prose has passed successively through a coordinative, a subor- dinative, and a suppressive stage ; and that it has shown very great decrease in formal predication. Manifestly each of these lines of investigation has its bearing on the development of the paragraph. The relation of the short sentence to the paragraph is a vital one, and whatever causes have produced the one have doubtless affected the other. 'J'he ques- tion of the historical use of conjunctions — especially of inter-sen- tential conjunctions — bears directly upon the history of coherence in the paragraph. The question of the decrease of predication affects the paragraph quite as vitally as these preceding questions, though not quite so apparently. For, if an author omits many predications within the sentence he has a type of mind which will tempt him to omit predications between sentences, /. e. to omit transitional sentences. Clearly the omission of transi- tional sentences affects very emphatically the coherence of the paragraph. We shall therefore examine Professor Sherman's the- ories at some length. ^Analytics of Literature, Chapters xix-xxvi. University Studies,\'o\. I., No?. 2 and 4. 52 RECENT INVESTIGATIONS IN PROSE-FORM. 53 First, then, regarding the origin and tendency of the short sentence. This sentence Sherman attributes to the introduction of conversational style into literature. The explanation seems to me correct, and the point important. Some stress, however,, must be laid on the probability, already pointed out on pages 44-46, that the final adoption of the short sentence and the para- graph was partly due to despair on the part of the periodic writers. These could not go on forever without seeing the hopelessness of trying to introduce full Latin idiom into English ;, nay, even of thinking with logical precision in a kind of sentence devoid of most of the means of coherence so richly present in the Latin sentence. But Professor Sherman has also demonstrated that as the short sentence is introduced the average sentence-length acquires a very strong tendency to become a constant quantity in succes- sive groups of, say, 500 periods or more. From this interesting fact he concludes : " The evidence seemed to indicate the operation of some kind of sentence-sense, some conception or ideal of form which if it could have its will, would reduce all sentences to procrustean regularity." ' But it seems to me that this statement implies rather more than is warranted by the mere tendency toward constancy in successive large groups of periods. Is this tendency finally to destroy the long sentence ? How are we to account for the long sentence in the midst of such an oral style as Macaulay's ? Is it due merely to a survival of classical influence ? When our prose has quite acquired conversational urbanity is the long sen- tence, whether periodic or loose, to be a thing of the past ? Perhaps the paragraph has something to do with the answer to these questions. A sentence is long or short in Macaulay according to its importance in the paragraph. A dozen clauses may be bundled together in one period to show that the whole group is no more emphatic than the neighboring ' University S/iidits, Vol. I., No. 4, p. 353. 54 ///S/OKY OF THE ENGLISH PARAGRAPH. proposition of half a dozen words. For the sake of this sense of proportion, Macaulay will make almost the same words a whole period in one paragraph, a mere clause in the next. In the best modern paragraphs I think it is, in general, true, that the distance between full stops is inversely as the emphasis of each included proposition. If this be the case, the distances between periods will not soon be reduced to approxi- mate uniformity, however much influence the oral tendency may have upon the order of words in the sentence. It will further on be seen that, while the English sentence- average has pretty steadily decreased, and while it has kept the tendencv towards constancy that was fully developed by Swift's day, yet, ivithin the limits of the given series of sentences that yields a given average, the degree of variability has steadily increased. Macaulay's long sentences are very long, as indeed Sherman has noted. It is perhaps a possible thing that the time will come when the sentence-average will no longer be a constant quantity in each author, but will be wholly regulated by the paragraph-structure. § -■ We are now ready to examine the bearing upon the para- graph of the decrease in the use of conjunctions and of the decrease in predications. But at the outset we find both of these phenomena referred to by Professor Sherman as belonging to the "analytic" or " oral " style. Before we can make it clear whether these phenomena benefit or hurt the paragraph-structure, we must know the exact meaning of these terms "analytic" and "oral," as applied to style. This we must know, even at the risk of a long and tiresome detour. Analysis means psychologically the process of abstraction — the conscious recovery of the intermediate term or terms in the process of association. Analytic thinking proceeds step by step, with full consciousness of the relation between parts ; a style that incarnates such thinking may be called analytic ; such a style is abstruse, philosophic. On the other hand a thinker may proceed RECENT INVESTIGA TIONS IN PROSE-FORM. 5 5 by relatively concrete terms ; he may not see the third term in the process of association, but may pass intuitively to his remote conclusion. In so far as a style reproduces this sort of thinking it may be called synthetic or intuitive. But it is possible and natural to use the terms, in criticism, with a force exactly opposite to the strict psychological ones. Sherman' speaks of an analytic style as sy)ioiiymous with an intuitive style. He is evidently brought to this apparent paradox by having previously spoken of the short sentence as analytic, the long one as synthetic. Yet his words on this point seem also to have some psychological implication : — "The analytical principle as observed in Channing and Macau- lay appeared to mean, Put in a simple sentence no more than can be brought before the mind pictorially or symbolically in a single view. If this meaning be yet but potential, not yet translated into successive propositions, let it be realized to the mind and expressed by instalments in some logical order, each fact or judgment, since an integral part of the whole, in a sentence by itself. But the synthetic principle amounts to an impulse to develop the whole meaning in some way within the limits of a single sentence."^ In this explanation of the analytical principle Professor Sher man evidently means that the analytic style tends to ratiocination — tends to follow the steps of the thought and express them all so as to conduct the reader by easy stages. But is not this kind of analytic manner the exact opposite of the analytic manner described by Sherman elsewhere ? "The analytic manner communicates as we have seen by points, but has nothing to do with making the points large or small, frequent or widely separated. It is the business of the reader to fill them out to a superficies of sense." ^ "Analytic or intuitive styles differ according to the leap or omission of thought '^Analytics, p. 303. ^University Studies, Vol. I., No. 4, p. 355. '^Analytics, p. 301. 56 HISTORY OF 'HIE ENGLISH PARAGRAPH. between. It is the length of the leap rather than the shortness of the periods that makes an author seem laconic. No one is conscious of Bartol's staccato quality in passages where his thought is most sustained. Channing, when he writes sen- tences as short, but with lesser gajjs of meaning, seems as smooth as Newman."' Evidently, then, the terms analysis and synthesis as applied to style are likely to create confusion. For the immediate pur- poses of this paper it seems better to substitute other terms, granting, if need be, that the new terms are not intrinsically better and are open to being called pedantic. Let us have four new terms, two corresponding to analysis and synthesis in form, two to analysis and synthesis /// thought. To the style in which the sentence of maximum frequency is short — say twenty words or less— let us assign the name Segregating. The opposite of this style, then, the style that brings its clauses together in whole blocks (as old Thomas Fuller would have said) or (as Minto has improved the expression) in flocks, will be the Aggregating style. When a style proceeds by leaps, omitting the intermediate steps, we may (speaking psycho- logically, not metaphysically, of course) call it Intuitive ; and its opposite, which omits no step, we may call Redintegrating. Nay, if this last named manner proceeds not by real and rational analogies but by mere association of contiguity, we may indulge in so large a name for it as Impartially Redintegrative. We may save the word Abstract chiefly for the style whose vocabu- lary is abstract ; and Concrete for the opposite style. According to this cumbrous, but, I hope, definite terminology, Macaulay's style would be at once segregating and redintegra- ting. Macaulay asks you to supply nothing but conjunctions ; nay, he often expands into a sentence of transition a relation that De Ouincey would get rid of with a however, and that Emerson would leave you to guess at. Landor would be intui- tive, and, except in his most sustained passages, would doubtless ^Analytics, p. 303. RECENT INVESTIGA TIONS IN PROSE-FORM. 5 7 be seafreofatins: as well. Carlyle in the French Revolution would be intuitive and segregating, in Sartor Kesartus intuitive and aggregating. De Ouincey would be redintegrating and aggregating, in spite of strong flashes of imagination now and then. So much for the word analytic ; now for the word oral.' Pro- fessor Sherman speaks of the analytic sentence as belonging to the oral style. By analytic sentence in this sense he means pri- marily the short sentence. " What makes short-period styles is the oral sentence-sense given free play as in ordinary talk."- This is easily understood and easily believed. But there are other characteristics of the oral style. We gather from one part of Professor Sherman's discussion that the oral style is analytic, in the psychological sense — that it tends to explain its subject by giving the successive steps by which the main conclusions are reached. "Though it is much more convenient to put integral thoughts in single sentences, such form manifestly handicaps every reader to whom the thoughts are new. What I may have in my mind cannot be transferred bodily to another's. I can only use a series of signs from which the reader recon- structs the fabric I have builded in m}' brain. But before he can put together a thought identical with my own, I must evidently take mine to pieces, and signify to him each part, and how it must go into place. Thus, while the attainment of the meaning to be expressed is a synthetic process, the first step in the act of expression is clearly analytic." ^ In other words, the oral style, in order to make perfectly clear to the reader the thought that has been intuitively perceived, introduces a string of inter mediate predications leading to a final and chief predication. But in another place, Professor Sherman leads us to infer 'All discussion of the nature of oral stvle must of course be inadequate until the psychologist and the physiologist settle by experiment certain elemen- tary questions regarding the actual sentence of conversation. Macaulay's is far the easier to read, even if we make allowance for certain idiosyncrasies in Carlyle's vocabulary and structure. The simple fact is, indeed, that half of Carlyle's idiosyncrasy lies in the way he evades predication by the use of significant, though odd and irregular words. Since we must settle the question for ourselves, so far as immediate use of terms is concerned, I should say that Macaulay's style has the better claim to the adjective oral. He knew that the use of many condensed expressions — clause- evasions — was likely either to retard the immediate progress of understanding, or to vitiate seriously the comprehension of the ' Atialytics, p. 309, Footnote. Here these particular participial uses are referred to as for the most part inorganic and unoral. 6o HISTORY OF THE ENGLISH PARAGRAPH. thouglu if tlie rate of reading were increased. He assumed small literary training or appreciation on the part of his audience. To put it bluntly, he wrote down to them. For a readable style must not be heavy — /. c, must not convey unnecessary notions — nor, again, can it be very weighty — i.e., convey many new notions in each sentence. But clause-evasion, while it increases ease of reading when the clause suppressed can be instantly supplied, does not permit the slow stream of thought to eddy around the idea, as De Ouincey would say, and so grasp it if new. In Macaulay the percentage of clause-evasion is not high ; according to Mr. Gerwig, the saving by " substitution of present and past participles or by the use of appositives," amounts in the Essays to 5.06 per cent. White, of Selborne, saves twice as much, Dr. Barrow nearer thrice, though both wrote longer sentences and used more predications than Macaulay. Oreely, writing far fewer simple sentences than Macaulay, yet reaches 1 7 per cent, of clause-evasions. The oral style, then, as we shall use the term, will show the segregating sentence, but the redintegrating method — short sen- tences, closely consecutive. It will show an absolutely high proportion of simple sentences, but not an absolutely high pro- portion of clauses saved. When the short sentences omit the minor steps of the logical order and there is made a strong demand on the reader's interpretative powers, we shall speak of the style as intuitive, with oral sentence-length. Emerson and Bartol would be assigned to this style. It might, indeed, be maintained that it is difficult to prove the short sentence an absolute necessity to the oral style, even in this limited sense. It might with some show of reason be asserted that the real oral unit is the short, loose clause ; that the long, loose sentence, with its succession of brief propositions, repre- sents a very common phenomenon of conversation. In ora narrative, for instance, the speaker often groups together great numbers of clauses in this way, letting his voice fall only at the end of the series. This would occur at least partly in proportion RECENT INVESTIGA TIONS IX PROSE-FORM. 6 i as the apprehension of the audience was quick. With a duller audience it niio;ht be necessary to let the voice fall after each short clause, inflection thus aiding comprehension. But we may consider this loose sentence as a species intermediate between the bold oral style and the subtly subordinative literary style. There is still another way of defining oral style — namely, to make it a relative term that alters in value with the mental powers of the audience. For in conversation our style is supposedly dictated, to a large extent, by the rate of mental response on the part of the hearer. We predicate interme- diate steps — we explain, in short — or we assume such inter- mediate steps, according to the presence or the absence of the appreciative flash in the hearer's eye. Thus, in talking to a person as well informed as we, we proceed with lightning rapidity. We not only omit predicates, both immediate and intermediate, but we indulge in all manner of contractions and elisions, many of them highly unliterary, almost illiterate ; nay, we convey as much by stress and gesture as by word. This is one kind of oral style, to be sure ; if we carry the theory far enough we can secure an oral style that is not style at all, as, for instance, in talking to a superior who will guess one's meaning from one's first word. Thus the oral style would increase in intuitiveness just in proportion to the intuitive powers of the audience. In the actual case of the history of our literature the oral style, in this relative sense of the word, has undergone certain manifest changes according to the change in audience. Begin- ning with the change from the scholastic audience of the six- teenth century to what Mr. Bagehot would call the masculine, common-sense audience of the eighteenth, the oral style would be progressively analytic — /. e., segregating and redintegrating. Proceeding from the Augustan prose to the latest subtleties of what Mr. Saintsbury would call '' marivaudage," the oral style would be progressively synthetic — /. e., intuitive and segregating for one species of it, intuitive and aggregating for another species. 62 niSrOKY OF THE ENGLISH rARAGRAPH. The most of Professor Sherman's remarks tend toward a definition of oral style as thus relative and elastic. He says, " Heaviness then is a relative term. The styles of those who, like Newman, address the educated exclusively, will not be heavy to their proper public though unintelligible to common readers.'" Hut again, exactly to the contrary of this: "To comprehend a style which condenses clauses to phrases requires as much literary preparation as to read Keats.'"" With these sentences compare his one explanation (p. 57) of the style as analytic, and his other explanation of it (p. 58) as clause-suppressive. But in his latest article on the subject he has a sentence or two which look toward calling a halt to the extension of the term to all' intuitive styles. "Some of the most polished of present stylists studiously eschew seeming better than conversational writers. The style of the future is likely to be yet more informal and easy than the best examples of this sort now extant. It will not probably abound in numerical averages as low as Bartol's or Emerson's and will be less disjointed and staccato. An informal organic sentence need not be long, but must not be weighed down with predications. Effective individual styles not hard to find in the periodical lit- erature of these days will average, perhaps, as high as twenty words of numerical length, yet show not above 1.60 predications per sentence, nor less than 65 per cent, of simple sentences." ^ This is as near as Professor Sherman comes to discussing the question of what percentage the oral style should show of implied predications and xvhat percentage of sifuple sentences to a given complex thought. The passage is at least less trustful of the intui- tive manner being properly oral than this sentence from the Analytics : "Hence the ideal style will have a maximum num- ber of intuitive sentences ; and that style is lightest that comes nearest to the first impressions of the mind." In the limited view of the oral style taken in our own discussion, "the style that ' University Studies, Vol. I., No. 4, p. 363. ^Analytics, p. 296. 3 University Studies, Vol. I., No. 4, p. 361. RECENT INVESTIGATIONS IN PROSE- FORM. 6 o comes nearest to the first impressions of the mind " will never be the lightest until the popular audience becomes one of literary experts. That the increasing culture of the people perha])s tends towards such an event may be true. But meantime the oral style of the future, while not sacrificing quite so much for clearness as Macaulay's, will probably be very far more expansive than Emerson's. To focalize the discussion upon the question of the paragraph is now easy. The oral style proceeds, as we have seen, by expanding into short sentences a given integral thought. When, with Temple, the paragraph may be said to acquire unity, each paragraph comes to represent an integral thought thus internally segregated. The principle to be formulated then is : From the moment of the establishment of unity, in the development of the Eng- lish paragraph, the oral sentence-structure means decreasing the number of predications in the period and increasing the number of propositions in the paragraph, in proportio)i to the author's conception of his reader'' s intuitive power : it beijig further premised that the intuitive power of the zuriter exceeds that of the reader. We have, therefore, found from this long and diffuse discus- sion, that the oral structure, /. e. a redintegrating and segrega- ting style, is an essential feature of the best paragraph ; though we shall not deny that good paragraphs may have a large number of intuitive statements. We have now to inquire how far the omis- sion of conjunctions is consistent with such a style ; whether coherence is hurt or helped by this omission. Professor Sherman says : "As there are no conjunctions in the mind — that is, no pictorial or symbolic representations of them as ideas — the style that most nearly follows thought will omit them when possible, or where formal merely." ' There certainly can be no doubt that many of the most effective recent styles show a minimum of conjunctions. But is it quite sure, that because conjunctions do not occur to the mind as substantive images, thev are usually formal and useless ? It is a hard thing ' Analytics, p. 305. 64 HISTORY OF THE ENGLISH PARAGRAPH. — some have said an impossible thinu^ — to say how a complex thought "looks" in llic mind. Hut it is probably safe to say that the minute we try to transfer that elusive thing, a thought, into the mind's eye, it seems to take the foru) of a mental image in which the notions are in some way grouped or graded. We are conscious that some ideas are principal and some subordinate. The more clearly we preceive these inter-relations of ideas, the more analytic and logical is our thinking. To see them at all clearly in their grouping requires a quiet eye, a dispassionate mind. The moment thought is disturbed by emotion awa}' fly the deli- cate middle-shades of the picture ; we see the substantive points in the stream of thought, but we see them so strongly that we do not notice their inter-relations. Now conjunctions are the result of an effort to express these un-named relations. The sense of relation may, indeed, be so strong that a mind like De Quincey's will take a long sentence in the effort to capture a gradation that the conjunction is not equal to. But when the mind is impassioned the sense of proportion between ideas is badly disturbed. The mind cares nothing for the inter- relation of facts — it wants the facts themselves. Accordingly impassioned prose — the literature of the will — may omit con- junctions with good effect. There are also a few relations so obvious that the conjunctions which express them can safely be omitted in any prose. Such is the relation of cause and effect, which is sufficiently conveyed by juxtaposition of cause and effect in separate clauses. But for the most part prose cannot be accurate without the use of conjunctions. Prose that omits them runs the risk of over-statement or under-statement. Prose that can safely run this risk is limited to a field that forms but a small part of the best literature. Accordingly we are not surprised to see that, of the men whom Professor Sherman quotes as illustrative of the new articulation, most of those who show low percentage of conjunctions are not the ones whom we praise most as stylists. The list includes (iladstone, Lowell, Emerson, Theodore Parker, Bartol, T. !'• RECENT IN I 'ESTIGA T/ONS IN PROSE-FORM. 6 5 Munger, and Dr. Holmes. I am not saying that these men are poor stylists ; but I am saying that they could hardly have been able — with their sparing use of conjunctions — to get the exquisitely true and clear effects that Newman, Pater, and Arnold secure. But let us also confess that presence of con- junctions in quantities is no proof of subtlety ; else why should Donald G. Mitchell stand next, in Sherman's list, to Newman, and before Pater ? Nor let any man accuse me of putting down Lowell as a lesser critic than Arnold, simply because Lowell's thought is occasionally too fertile for his style to be exquisitely true and clear. We now turn to consider the oral style as affecting proportion in the paragraph. The hurt that the oral style has done on the whole to the proportion of the paragraph, is, out of all comparison, less than its beneficial effect. But in such writers as use the short sentence to a maximum degree, the emphasis of the paragraph is evenly spread over each proposition. The question \vill be dis- cussed with somewhat more fullness under the head of Macaulay. For the present it is enough to say that the asyndetic structure and the exclusive use of the short sentence are " terse and intense forms," and as such have their dangers. It is hardly enough to say, with Sherman, ' " We are not to write always in terse and intense forms. The intermediate notes are normal both to those who have as yet not passed beyond them, and upon occasion to all of us." It is much nearer the truth to say as Sherman at last does : ~ " Indeed, the ideal style is either coordinative, subordina- five, suppressive, asyndeton, and at times even, for a little perhaps, synthetic, according to selective acts of the mind that are indeter- minate, or at least not yet determined." Yes, man lives by many a generous idea that can never be put into short sentences. ' Analytics, p. 312. ^Analytics, p. 312. CHAPTER V. ALFRED TO TYNDALE. The related paragraph plays no structural part in Old Eng- lish prose. There is no conscious attempt to advance by stages. The chief merit of the prose is sequence — not exactly coherence, which assumes some logical method — but general consecutive- ness. In this quality and in the short sentence there are, however, present two prerequisites of paragraph structure. But this old prose is by no means utterly formless. In much of it there is a kind of instinctive sentence-grouping that reveals the natural tendency of the language toward the paragraph and away from the long period. If this seems fanciful to anyone let him take the unbroken text of the preface to Alfred's version of the Cura Pastoralis and try whether it be harder to determine the natural divisions of this discourse, or those of a chapter in Cap- grave or Malory. The Old English writers are, however, the originators of our isolated paragraph. By this I mean that many of their so-called chapters are so short as to illustrate the structure of the isolated paragraph. Chapters of 200 words, like many in Alfred's Bede, are not chapters in the modern sense ; they are tiny whole compositions, corresponding as inventional units to that particu- lar modern editorial paragraph which, set off by itself, is at once complete in itself and related to its neighbors. The longer pieces of prose are usually broken up by the paragraph- marks of the rubricator. Whether the author ever dictated the position of these marks it is impossible to say. There are certain autograph manuscripts that contain such marks, but the fact proves nothing. There were four distinct uses of the marks : {a) to note a logical section ; {b) to note an emphatic point ; {f) formally to distinguish sacred names ; {d) to orna- 66 ALFRED TO TYNDALE. ' 67 ment and distinguish titles, colophons, etc. Of course manu- scripts differ in the degree of success with which these points, especially the first, are attained. Some are very stupidly divided, others verv cleverly. In some the emphasis mark predominates, in others it is almost absent. In most manuscripts all four principles are apparent. The habit of marking for emphasis, whether at the beginning of a rational section or in the midst of it, was not without its influence in after days. The emphasis-tradition is in full play even in Milton. It is partly responsible for one glaring fault of the sixteenth century — that of beginning a paragraph one sentence late, so to speak ; of not noticing the turn in the dis- course till this arrives at an emphatic new point. In spite of the emphasis-principle, it is not rash to say that the Old English paragraph has in a general way good unity of subject.' Coherence and proportion and mass it has not. ALFRED. Of works by i\lfred there are but three contemporary manu- scripts, namely, the Hatton and Cotton MSS. of the translation of Gregory's Pastoral Care, and the Lauderdale MS. of the Orosiiis. The last named MS. forms the basis of Sweet's text of the Orosiiis (E. E. T. S. 79). Sweet has broken the text into para- graphs that make natural steps in the story. Out of curiosity I made a count of words, sentences, paragraphs, sentence-length and paragraph-length in the first book, according to the Lauder- dale MS. and Sweet's pointing. The result was 9862 words, 381 sentences, 66 paragraphs ; average sentence-length, 25.8 ; para- graph-length in words, 151.68, in sentences, 5.8. The limits of ' I have felt very strongly the difficulty of making other than very general statements concerning the presence or absence of unity in an author's para- graphs. It is not a very hard matter to decide whether or not a paragraph has digressions ; but it is a far harder task to obser\-e and state all the principles on which a composition may properly be divided. In the main, I have attempted to distinguish but two general types of unity — the purely logical and the rhetorical or picturesque. 68 ■ HISTORY OF THE ENGLISH PARAGRAPH. word-length in the paragraph (34-453) show a field of variability less wide than Channing's or l)e Ouincev's. If we take the sentence-length as in any sense organic (as I, for one, should be inclined to do), the figures go to show that Alfred did not know enough Latin to hurt his English stntcturc, many commenUilors to the contrary notwithstanding. To illustrate, let me quote a passage (Sweet's ed., j). 171). In translating, the king has para- phrased (rather freely) a Latin sentence of forty-six words, by a very fair English paragraph of four sentences or ninety-one words. Anno ab urbe condita CCCCLXXXIII, Mamertinis auxilia contra copias et Ap. Claudium consulem cuu! cxcrcitu misere Romani : qui tani celeriier Poenos superavif, ut ipse rex ante se victiim (juam congressum fuisse prodiderit ; qui cxiii, eiim pacem rogarct, ducentis argenti talentis miiltatus, accepit. .^t'ter f>8em |'e Romeburg getimbred Ava?s I'eower liuiule wintrum 7 Lxx(x)iii, sendon hie liim Appius Claudius )one consul mid iultunie. Eft, ]'a hie togt«dereward foron (mid heora folcura), l-a flugon Pane, swa he eft selfe ssedon, 7 his wuudredan, ptet hie ser flugon ?er hie togaedere genealseeten. For jifem fleame Hanna, Pena eyning, mid eallum hio folce wear8 Romanum to gafolgil- dum, 7 him aelc gears geseahle twa hund talentana siolfres : on (vlcre anre talentan wses Lxxx punda. In the translation of the Pastoral Care, the paragraph -mark plays no part. But the preface falls naturally into excellent sentence-groups that show something like real coherence and explicitness of reference. The translation of the BoetJiiiis is not paragraphed in the MS. There are a few sections, e. g. in Otho, a. vi., before § 2, chap. 37 ; § 4, chap. 39 ; § 2, chap. 40. The paragraphing of the Bohn text is entirely the work of the editor. The Old Enaflish version of Bede's History has until recently been ascribed with confidence to Alfred, on the authority of .^Ifric and William of Malmesbury. Dr. Thomas Miller, how- ever, in his recent edition {V.. E. T. S. 95, pp. Ivi., Ivii.) believes ALFRED TO'TYNDALE. 69 it to be of Anglian origin. Miller's text is collated from {a) the Tanner 10 of the Bodleian, {b) the Corpus Christi College, Oxford, {c) the Cambridge — Kk. 3, iS. The Cambridge alone numbers the chapters. In making a count from Book I., I have therefore followed the chaptering of the Cambridge. I have a few times departed from Miller's pointing, but not to make any sentence shorter. The QiKBstiones of Augustine, placed by Miller in Cap. 27, I have not counted, since in the MSS. they stand at the end of Book IIL The first book then contains 6898 words, — since only twenty-one of the original thirty-one chap- ters remain in the Cambridge MSS. This gives an average of 222.5 words to the paragraph, the limits of fluctuation being 49-838. There are 7.84 sentences in the paragraph. These figures, like those of the Orosius, are suggestive of nineteenth century lengths. There are few exceptions to unity in the paragraphs of the Bede. The style is less flexible and subordinating than that of the Orosius, the coordinative stage of the language being rather painfully evident. There are 506 ands in the first book, or one to every twelve words. WULFSTAN. The Homilies of Wulfstan, written in vigorous native prose, are extant in numerous manuscripts. None of the homilies shows many paragraphs — most have two or three. The general structure of the prose is logical. As Napier ' has pointed the sentences and placed the indentations, both sentence and para- graph are longer and yet compacter than those of the Orosius. yELFRIC. The manuscripts of ^Ifric's Lives of Saints are not para- graphed. The chapters of the Latin Grammar are numerically divided into sections, and there are a few other paragraph-marks. But the book is not literature and its paragraphs are not to be considered. ' Wulfstan, herausgegeben von Arthur Na})ier, Berlin, 1883. 70 HISTORY OF THE ENGLISH I'ARACh'APII. vSweet, in editing selected Homilies from the Cambridge MS. (imparagraphed), has broken the text into sections, which I have counted, again out of curiosity. The first nine homibes contain 18,854 words, 723 sentences. The sentence-length is therefore 25.8 words; the paragraph-length is 176.2 words, 6.75 sentences. Both sentence and paragraph are slightly longer than Alfred's (also punctuated by Sweet) ; but if we consider the increase in learning that occurred in the intervening century, it is surprising that the increase in the length of the sentence is so small. ^Ifric's style exhibits decided advance over his predecessors in power of graceful transition. One paragraph leads to another, and there are varied devices of explicit reference. THE ANCREN RIWLE. After a long period barren of prose, we come to the Ancren Riivle, 1220. Here we have an alert and cultivated style. The MSS. are divided systematically into books, and these into simple capital-paragraphs. The main fault of this style is the abrupt transition between these paragraphs. In the Ancren Riwlc, the English sentence-length is still untouched by Latin influence. The first 61 paragraphs (509 sentences, 12,0^9 words) yield a sentence of 23.67, a paragraph of 197.5. The sentence is the same, within a quarter of a word, as that of Macaulay's England. The paragraph is four words longer than that of Emerson's Self Reliance, eight words shorter than that of Arnold's Literary Influence of Academies. THE AYENBITE OF INWYT. Dan Michel's Ayenbite of Inwyt, 1340, is a wooden piece of prose. The paragraphing is systematic to the extreme, each section being introduced by a set phrase (usually a numerical one) and forming one step in the long list of virtues and vices. The average length of the sections is only 131 words, yet a quarter of them have no unity. The sentence average is low — onlv 21 words. It is noticeable that the paragraphing does not ALFRED TO TYNDALE. 71 follow that of the French original, if that happens to be repre- sented by MS. Cleopatra Av. Fol. 17,713. MANDEVILLE. The Voiage and Travaile vfa.?, vfv[\.tGX\ about 1356. The Eng- lish version was first printed by Pynson. The edition of 1725 (7) based upon Cotton MS., Titus c. xvi, and collated with other MSS. and printed versions, is the standard edition, and was reprinted by Halliwell-Phillips, 1839, 1866, 1883. This editor in his Additional Notes says cheeringly : " The chapters are very differently divided in various MSS. Some have no divisions at all." I have, however, counted, ten para- graphs, from Halliwell's edition of 1839, to show something of the sentence-length of this early and important piece of prose. The first ten paragraphs average 337 words, 9.5 sentences, the sentence-length being 35.48. Although this is a higher average than any we have yet found, and manifestly shows Latin influ- ence, it is still suspiciously short for the time and the man, and leads us to fear editorial tampering. WICLIF. Wiclif's Bible is divided into chapters, but not into para- graphs or "verses." Most of his other works are scantily paragraphed. In these original works his unit of composition was evidently large and its construction logical. So good is the analytic consecutiveness of his work that it would be possible to divide it into reasonable stadia. By modern principles of punc- tuation his average sentence is not excessively long ; it varies from 30 to 35 words in different essays. This moderate length is the more noticeable because the diction is freely latinized. CHAUCER. The prose works of Chaucer are four : the translation of Boethiiis, De Consolation c Philosophice ; the Melibxus and the Persoiies Tale ; the Treatise on the Astrolabe, dedicated to his son. 72 HISTORY OF THE ENGLISH PARAGRAPH. The /)W//!/// J- has been edited by Morris (E. E. T. S.) from the Brit. JMus. Add. MS. 10,340, and the Cambridge Univ. Lib. MS., I., 3, 2\. \\\ tlie Cambridge MS. the parallel oblique virgules (//) are used in place of the \ In the Additional MS. the rubri- cator has been so lavish with his vermilion, that but for his rational work in the latter part we should think him gone paragraph-mad. At first every sentence begins with the mark. Of the first 100 paragraphs only three have two sen- tences each, one has three, and ninety-six, one ! The average length of these 100 is 29.51 words. Later on, e. g., Morris, p. 164 ff., the paragraphing is good ; but, if we take the MS. as a whole, evidently most of the paragraphs are false. Some ardent Chaucerian monk thought every sentence worth emphasis ; and nine times in 100 paragraphs he puts for emphasis a second mark within the line. The two prose pieces in the Canterbury Tales are differently divided in different MSS., but always somewhat arbitrarily. The judgment shown by the rubricators of the Melibceus is particu- larly bad. The paragraphing of the Persones Tale is better, though not uniformly so. By the Harleian MS. this tale yields a paragraph-average of 8.26 sentences, — about 290 words. Chaucer's prose sentence in this tale is only a word shorter than Mandeville's. On the other hand the sentence of the Boethius is considerably shorter than Mandeville's. The Astrolabe is so far from being a piece of orderly prose, much less literature, that I have not attempted to find in it a paragraph-sense so manifestly lacking even in Chaucer's best writing. PECOCK. Reginald Pecock's Repressour of Over Much Blaming of the Clergy, was edited by Churchill Babington, and published in the Rolls series in i860. The paragraph-marks, alternately red and blue, are distributed in the Canterbury MS. with no particular skill. The modern editor, like so many other modern editors, has felt free to print his text without a trace of these marks, and ALFRED ro TYNDALE. 73 to substitute paragraphing of his own. The effort only assures us tjiat Pecock's unit was the long period. Babington's para- graphs average about 262 words, and he is not able to reduce the sentence to -less than 61. Pecock's work is one of the first exam- ples of purely controversial prose. His order of procedure is formal, and his good logic makes his work divisible into rough stages. CAPGRAVE. The MS. of Capgrave's Chronicle I have not been able to con- sult. The edition of Hingeston (Rolls series, 1858) prints a page of it in facsimile. Here the first letter in each sentence is marked by a red stroke. For this or for some other unaccountable rea- son, Hingeston tends to paragraph each sentence by itself. MALORY. Caxton'sZ^ Mort Darthure was admirably edited and reprinted in 1889, thanks to the conscientiousness of Dr. Oskar^ Sommer. The most of the chapters are unbroken, as is the long preface of Caxton. The paragraphs in the remaining chapters are indi- cated sometimes by Caxton's mark [P. 11, Fig. 36] at the head of a line after spacing in the preceding line, sometimes by the mere mark in the midst of a line, sometimes by the mark and a short space in the midst of a line. The narrative is remarkably sequent, and the chapters have a certain unity. The paragraph-marks, delicious as their glossy thickness looks to the eye of the book- lover who turns the pages, are quite as likely to serve the pur- pose of mere emphasis as to guide the eve to a new section. But it is half the pleasure of reading to watch these signboards of Caxton's naive taste, and see how unerringly he plumps you down a fat mark at the exciting moment. FABVAN. Pynson's edition (15 16) of Fabyan's Concordance of Histories, was carefully reprinted by Henry Ellis in 181 1. In this edition the paragraph- mark is used only before titles, etc.; indentation marks the paragraphs. In the first part of the book these are 74 HISTORY OF THE ENGLISH PARAGRAPH. short ; the breaks become rarer as the writer approaches the events of his own day, and when he moves out of his usual cold and formal style into some show of enthusiasm he forgets to par- agraph at all. In the early part of the work the unity of subject is almost unimpeachable ; in the latter part there is no paragraph- unitv whatever. The sentence is ponderous, and often confused with the paragraph. Sherman found its average length to be 63.02. MORE. The first English version of the Utopia was made by Ralph Robinson. Robinson's second edition (1556) has been reprinted by Arber. The \, in shape much like Caxton's, appears before each main section. Indentation is employed but sparingly, but always when the words are those of a new speaker in dialogue. The real steps are indicated by marginal notes. More's best English appears in ihc Hislorie of Richard III., where the idiomatic short sentences give something like propor- tion to the long sections. His most logical paragraphing, how- ever, occurs in the polemical tracts. Here the steps are short, the arrangement of the sentences is compact, and the whole effect animated. CHAPTER VI. TYNDALE TO TEMPLE. Before Tyndale the paragraph cannot be said to have any structural character. Those qualities which the old English prose did contribute to the modern unit, namely, the qualities of consecutiveness and loose order of propositions, came to their first culmination of development in the style of Tyndale. And in the long period from Tyndale to Temple we have the battle- ground where the principles of the modern unit were victorious in a contest with various enemies. At the beginning of this period stands our first tolerable paragrapher, Tyndale; at the end of it stands our first recognized organizer of the paragraph, Tem- ple. The old English traditions represented in Tyndale were perpetuated by one line of vernacular writers throughout the Elizabethan and Jacobean period, and opposed by another line. The contest has ended when we reach Temple, and the older tradition is victor. TYND.-^LE. The Obedience of a Christian Alan. Total paragraphs considered lOO Average words per paragraph 204.48 Average sentences per paragraph 6.45 Average words per sentence 31-72 Per cent, of single-sentence paragraphs 8 Limits of fluctuation in paragraph word-length.... 33-450 In Tyndale's New Testament, 1525 (Arber's reprint), the Caxton mark is uniformly employed, and yet nearly always it is preceded by a spaced line. The index is also used, both for emphasis and for reference. The paragraphs are shorter than those of the 1611 version, and are not subdivided. Of the Obedience I consulted Day's edition, 1572. The para- graphing is admirable for Tyndale's time, and good if it is Day's 75 76 insrOKY OF THE ENGLISH J'.IKAGKAPH. and not Tvndale's. At least 75 per cent, are loose. P^ach topic is logically expanded by illustration, defense, etc. Of the 347 paragraphs in the book about 25 per cent, are of the periodic type, and perhaps 10 per cent, follow a strictly inductive order. Parallel construction occurs frequently, owing partly to imita- tion of Hebrew models, partly to the writer's natural oratorical directness. It is often e.xtended to groups of paragraphs — each being made to correspond in arrangement with its predecessor. The following example will give an idea of the internal parallel- ism ; the order of course is periodic. " Who dried up the Red Sea? Who slew Goliath? Who did all those wonderful deeds which thou readest in the Bible? Who delivered .the Israelites evermore from thraldom and bondage, as soon as they repented and turned to God? Faith verily, and God's truth, and the trust in the promises which he had made. Read the xith of Hebrews for thy consolation." LATIMER. First tivo Sermons before Edward. (Arber reprint.) Whole number of paragraphs 116 Wliole number of words 13621 Whole number of sentences 666 Average words in paragraph 1 17.42 Average sentences in paragraph 5.74 Average words in sentence 20.45 Sentence-length of first sermon 23.76 Sentence-length of second sermon i8.-|- Paragraph-length of first sermon 242.68 (28 ^s) Paragraph-length of second sermon 77-55 (88 ^s) Average predications per sentence { 4.75 Per cent, of .simple sentences (Gerwig) -^13 Per cent, of clauses saved ( 2.78 Latimer's style, as appears partly from the sentence-length, is colloquial and vernacular. Accordingly when his paragraphs are good they are modern in tone and are really admirable. But great unevenness marks them in unity, as in length. The coher- TYNDALE TO TEMPLE. 77 ence is due, not to connectives, but to tlie impassioned rush of the thought. CRANMER. Aiisiver to Gardiner, I., The Sae)aviciit. Total paragraphs considered 100 Total words considered 13)775 Average words per paragraph 137-75 Average sentences per paragraph 3.70 Average words per sentence 37-22 Per cent, single-sentence paragraphs 17 Cranmer is our first great master of the loose sentence. It is not uncommon to find in him a loose sentence of more than 100 words, put together with much skill in avoidance of tags — a skill elsewhere unknown in his day — and with a sense of prose rhvthm highly remarkable in any day. Since his paragraph is short, this gift at the long loose sentence is often no help to his paragraph. In the best sections the rhythm extends to the whole, sentence modifying sentence as subtly as Ruskin's do. Cranmer's sequence is such as might be expected from a clear and orderly mind. The rare dislocations occur, as in the case of Tyndale, from unconscious reversion to certain fixed moral themes. The coherence is largely dependent upon modern devices — inversions, demonstratives, etc., rather than conjunc- tions. And is the one coordinate that is abused as an initial con- nective. The per cent, of initial illatives is small, probably not over five per cent. The great majority of the paragraphs are loose. ASCHAM. Toxophiliis, 1544. Arber's reprint, 1868. Total paragraphs considered 100 Total words considered 13.585 Total sentences considered 315 Average words per paragraph 135-^5 Average sentences per paragraph 3. 1 5 Average words per sentence 43.13' Per cent, of single-sentence paragraphs 9 Limits (jf fluctuation in w(^rd-length of paragraph.. 10-564 ' Sherman found 49.60, for 500 periods. yS HISTORY OF THE ENGLISH PARAGRAPH. Scholcmastcr, 1570. Arber's reprint, 1870. Total paraLfiaphs ^2Q Total words c. 47,250 Average words per paragraph c. 143.62 Per cent, of single-sentence paragraphs 18 Per cent, of simple sentences ^ it) Average predications per sentence (Gerwig) ■' 3.49 Percent, of clauses saved / 4.31 Though the ScJiolemastcr was not printed till Ascham had been dead two years, the paragraphing shows no signs of having been tampered with ; for the method in this book is the same as that in the Toxophilits, written twenty years before, namely that of paragraphing the smallest possible stadia; and the word-length is about the same in both books, although the first is dialogue. The shortness of these stadia has led Professor Sherman ' to say that out of 329 paragraphs in the Scholemaster Ascham admits 148 false ones. I suppose Professor Shernaan means that these brief paragraphs should have been grouped into larger ones. Pie further says that in at least fifty-five cases the period and the para- graph are wrongly treated as o-ne. For one I should defend a good share of these single-sentence paragraphs as either marking true stadia or good transitions. Some of the others I should agree in assigning to bad logic. The rest may be explained by remembering the emphasis-tradition which we have noticed as a legacy from Old English rubricators. Such sentences as the fol- lowing Ascham paragraphed (sometimes with the obsolete ^) not out of logical confusion, but to make them prominent, in accord- ance with a use of the paragraph then perfectly understood but now forgotten : — " This he confesseth himself, this he uttereth in many places, as those can tell best who use to read him most." "The like diligence I would wish to be taken in Pindar and Horace, an equal match for all respects." " Budseus in his commentaries roughly and obscurely, after ' Analytics, p. 291. TYNDALE TO TEMPLE. 79 his kind of writing ; and for the matter, carried somewhat out of the way in overmuch misliking the imitation of Tull)'. " PhiL Melancthon, learnedly and truly. " Camerarius largely with a learned judgment, but somewhat confusedly and with over-rough a style. " Lambucus largely, with a right judgment, but somewhat a crooked style." The trouble with Ascham's style is, however, less that his par- agraphs are too short than that his sentences are too long. A sentence of nearly fifty words leaves no room for proportion in a paragraph of 140. The naivete of Ascham's manner precludes any complex coherence, and involves great abuse of conjunctions. Sherman ' has noted 'that 61 out of 329 paragraphs in the Scholemaster begin with amis; and that^ 24 paragraphs begin with biits, six with yets. If now we observe the use of sentence-con- nectives (not clause-connectives) in 300 sentences of. the Tox- ophilus, we shall find that 168 of the 300 periods are connected by conjunctions or brief conjunctive phrases. The list is varied, and serves to make the reference very explicit, in spite of the abuse of coordinatives. Nearly all of these connectives are initial, as will appear from the list given below, where the internal sen- tence-connectives are placed in the second column. From the total number (181) 13 should be deducted for repetitions (by double connectives), to get the number of connected sentences. Sentence-Connective. Initial. Interior. But (yea but) 23 For 30 And 49 In dede 4 Also . . 6 Yet 8 2 Contrariwise i So 4 Thus I ^Analytics, p. 271. '^Analytics, p. 427. 8o inSTORY OF THE ENGLISH PARAGRAPH. Sentence-Connective. Initial. Interior. Now 7 Therefore 15 2 Then 3 i Will 6 First I 3 Again 4 Furthermore i Moreover 2 Wherefore I At the last . . i Likewise 2 _ i Nor 2 To be short . . I Ascham's rhythm in the sentence and the paragraph is monot- onous. He has plenty of balanced Euphuistic sentences, that helj) his coherence, but the balance is monotonous. And where else in the language but in the Scholcmastcr can be found an author who will write you seven consecutive paragraphs of exactly twenty- five sentences each, the group being followed by three paragraphs of just fifty sentences each ? I half suspect that Ascham (or the printer) told those groups off on his fingers. On the whole, Ascham's place is midway between the vernac- ular writers to whom the paragraph is a natural structure, and the Latinists, who have no paragraph-sense at all. HOLINSHED. Chronicle, \Sl'l- Total paragraphs considered 200 Average number words per paragraph c. 204 The marginal note is rarely used ; the old mark ^ (single-stemmed) less rarely. The paragraphing of Holinshed is monotonously regular. The narrative paragraphs are often good in unity and in concentra- tion — as for example the well-known account of Macbeth. In the descriptive passages there are numerous digressions, and the dramatic grouping of details, that marks the best narrative pas- sages, is lacking. TYNDALE TO TEMPLE. 8i STOW. Summarie of the Chronicles of England, 1561. Total paragraphs considered 200 Average words per paragraph c. 186 Average sentences per paragraph c. 3.26 Average words per sentence c. 57 Per cent, of single-sentence paragraphs 41 The paragraphing of Stow's Chronicle (1561) is inorganic and false. The author unnecessarily confounds the sentence with the paragraph. The best that can be said of the style is that its profuse use of subordinating conjunctions keeps the coherence tolerably good. LYLY. Euphues, B. I., Eitphues and his Euphcebus, Eiiphues\ind Athos, Letters. Total paragraphs considered 279 Total sentences considered 1 600 Average words per sentence (Sherman) 36.83 Average words per paragraph c. 21 1.03 Average sentences per paragraph 5.73 Per cent, of single-sentence paragraphs 33 Average predications per sentence ( 3.50 Per cent, of simple sentences (Gerwig) i 17 Per cent, of clauses saved f 10.21 The paragraphing of Euphues, B. I., and Euphues and his Euphxbus, is regular, 4.87 sentences being the average in the first case, 4.10 sentences in the second. In Euphues and Athos the average rises to 5.54 sentences, and in the diffuse style of the Letters it mounts to 10.44 sentences. Taking Euphues as Lyly's typical piece of prose, we find some improvement over Ascham in general paragraphic structure. The sentence has shortened and the paragraph lengthened. Lyly's sentence, however, is still in bondage to the colon, and we find a very high per cent, of single-sentence sections. The unity of the paragraphs is wider than Ascham's — the per cent, of false sections is reduced. But there are plenty of 82 HISTORY OF THE ENOUSU PARAORAFII. digressions. Paragraphing for emphasis is rare, although the obsolete paragraph-mark is occasionally found in the text. A common violation of unity occurs in paragraphing a sentence that introduces a speech. In the formal manner of FAiphiiism such introductory sentences are felt as transitional paragraphs. Of Lvly's coherence a good word may be said in one partic- ular. If, on one hand, the balanced structure tends to make the flow of sequence intermittent, on the other, the habitual parallel construction constantly assures the reader of the general onward movement. Other means of coherence are slighted ; particularly, initial connectives. The less said of Lyly's proportion and massing, the better. His excessive illustration spoils both. His introductions are tedious, and, though there is usually a sentence-topic, the attempt to find it is sometimes to hunt for a needle in a haystack. GOSSON. Gosson's School of Abuse shows a sentence of about 60 words and a paragraph of something like 288. The paragraph is the unit of division, since there are no chapters. Gosson's Eiiphiiistn is strong enough to keep his paragraphs good in parallel structure. But his divisions are mechanical, and the unity so defective that only the marginal notes keep the reader from floundering hopelessly. SIDNEY. Arcadia, 1590. Total paragraphs considered 200 Average words per paragraph c. 444 Average sentences per paragraph 5.92 Average words per sentence c. 75 Per cent, of single-sentence paragraphs 10 Defense of Poetry, 1595. Total paragraphs 79 Average words per paragraph 235.30 Average sentences per paragraph 6.05 Average words per sentence 38.80 TYNDALE TO TEMPLE. 83 Per cent, of single-sentence paragraphs 10 Per cent, of simple sentences (10 Average predications per sentence (Gerwigj - 3.98 Per cent, of clauses saved ' 9-27 Saintsbury's criticism of Sidney as one of the first authors of great popularity to introduce the "sentence and paragraph heap," is entirely just as regards the Arcadia. The sentence reaches enormous lengths and is endlessly jointed and rejointed to do the work of the paragraph. The number of paragraphed sentences is not high, however, — a fact noticeable in the worst offenders in the matter of the clause-heap. Unity is but indifferently observed in the Arcadia, Chapters are preceded by a list of subjects of the paragraphs ; these topics, though often felicitously put, are manifestly not matters of pre- vision. In turning from the Arcadia to the Defense, ^^ \mxxs.\o ^vi utterly different structure. The sentence drops nearly, one-half its length. Most of the tags disappear. The new style is incomparably better than the old. In his recent edition' of this work. Professor A. S. Cook apparently regards most of the original paragraphs as lacking unity. Of the original 79 paragraphs he breaks up 37 into several paragraphs each. He has likewise thrown together many others, so that the total number of paragraphs in Cook's edition is 93. Usually these changes do improve the unity, although sometimes, even when the logical unity is thus increased, a cer- tain loss is felt in the distribution of emphasis. In Sidney's edition there are 8 single-sentence paragraphs. It is a curious fact that in Cook's edition this number is more than doubled. That a modern editor should make 19 such sections as against an original 8 shows clearly enough how flexible the idea of a paragraph as a group is today, as it has always been. Praise mav be given the general sequence of Sidney's para- graph : there is no difficulty in following him, even when he 'Sidney's Defense of Poetry, ed. A. S. Cook, Boston, 1S90. 84 HISTOKY OF THE ENGLISH PAR AG R API f. rambles. Kxplicitness of reference is secured by free use of connectives ; in tlie Defense something like 35 per cent, of his sentences begin with conjunctions. Parallel construction of suc- cessive periods is frequent. WEBBE. A Discourse of English Poetrie, 1586. Arber reprint, 1870. Total paragraphs considered 75 (Arber, pp. 21-36, 52-72) Average words per paragrapli c 154 Average words per sentence c 50 50 Average sentences per paragrapli 3.10 Per cent, of single-sentence paragraphs 14 Webbe'S Discourse is paragraphed in most erratic fashion. The unity of the longer paragraphs is, strange to say, better than that of the short ones. In his task of enumerating and charac- terizing the English poets, Webbe abandons all paragraph method. A glance at the sentence-length as compared with the paragraph-length shows that Webbe marks no advance in general structure, but rather a retrogression. PUTTENHAM. The Art of English Pocsie, 1589, generally ascribed to George Puttenham, is systematically written, but its paragraphing is inorganic and without significance. The unit of composition is the short chapter. The fourteenth chapter has but 72 words; in like manner, many other chapters are so short as to form isolated paragraphs and so are left unbroken. When Puttenham does paragraph he is often moved to do so for emphasis. SPENSER. View of the Present State of Ireland, written c. 1596, pub- lished 1633. Total paragraphs considered 200 Total sentences considered 503 Average words per sentence (Sherman) 49.8 Average words per paragraph c 125.20 Average sentences per paragraph 2.51 Per cent, of single-sentence paragraphs 48 < TYNDALE TO TEMPLE. 85 Average predications per sentence f 5-44 Per cent, of simple sentences (Gerwig) - 8 Per cent, of clauses saved ( 6.74 The most divergent views are held today regarding Spenser's prose-style. One recent writer declares his prose " practically unreadable ; " ' another calls Spenser's " an excellent prose style;"- says that "it is unaffected, clear, vigorous, straightfor- ward ; " " that it is perfectly simple and by its very simplicity impressive and forcible." As is usually the case, the truth probably lies between these extremes. Spenser is far easier to read than Hooker, whom there have always been some people to read ; he is free from obscurity or serious ambiguity. On the other hand, he is by no means perfectly simple, but often very complex ; at times the most careful attention is necessary to keep the main idea of the sentence clearly in mind through the long series of clauses. Spenser belongs to the line of classicists who were beginning the experiment of extreme sentence-lengths. In this respect, therefore, and in the comparative shortness of his paragraph, he makes no advance. On the other hand, we are surprised to find how few of his verv long periods are really periodic. Evidently a punctuation by strict modern standards would reduce the sentence very greatly without hurting the syntax at all. It surprises one to see how straightforward and close-knit is the development of topic. The sequence is admirable. It is assisted by a very large proportion of conjunctions and by many initial relatives. Out of 300 sentences, 164 are connected by conjunc- tions or conjunctive phrases. The list is as follows, the number of connected sentences being found bv deducting from the total number (181) 17 for repetitions (due to use of double-connectives) : Connective. Initial. Interior. But 39 So 12 10 And 30 ' Sherman, Analytics, p. 274. ^ John W. Hales, in Craik's Eiv^lish Prose, Vol. I., p. 455. 86 HISTOKV OF THE ENGLISH PARAGRAPH. Connective. Initial. Interior. Then 5 1 1 Therefore 5 3 For 13 Indeed i 4 True it is i Also . . 9 Neither 2 i Yet 3 2 Thus 2 I \'ea I Besides 2 2 Lastly, etc 2 3 Notwithstanding i And yet 3 Nevertheless 3 Now 5 I Now then i Again . . i Or I Thus far . . i From what has been said it is easy to infer that there is little proportion in Spenser's prose. There is no skillful varying of short sentences with long. The frequent use of illatives and of subordinating conjunctives proper does, indeed, convey some sense of logical prominence given to the main proposition ; but, on the other hand, the very frequent use of initial relatives (Spenser's besetting fault) goes far to destroy the proportion thus obtained. In sequence alone, then, and in the looseness of his sentence- structure, is Spenser in the line of paragraph development. HOOKER. E(clesiastual Polity, i 594-1 6 1 8. (Gerwig.) Average predications per sentence 4.12 Per cent. . of simple sentences 12 Per cent, of clauses saved 8.73 Mr. Vernon Blackburn, in the first volume of Craik's English TVNDALE TO TEMPLE. 87 Prose, refers to Hooker as a man who " perpended every para- graph."' I cannot {)ossiblv make out what this means. It seems to me that of all cultivated men who ever wrote English, Hooker perpended paragraphs the least. The earlv editions of the Ecclesiastical Polity are, so far as I know, quite without indented paragraphs. The Stansbye edition in the Astor library (1639 ?) shows sixteen main sections in the first book, but no sub-sections. These sections average 1S75 words each. Mani- festlv a related paragraph of 45.31 sentences, the sentences being on the average 41.23 words long, is no paragraph at all. The paragraph-length — section-length — is a little shorter in the second, third, and fourth books, and decidedly shorter in the fifth — the one in which Hooker's hand is least evident. Keble, finding himself lost in these wastes of words, broke up the whole text in his edition of 1836; the paragraphs thus formed average about 260 words. Can it be that Mr. Vernon Blackburn is referring to these paragraphs when he says that Hooker per- pended each ? The real truth about Hooker's paragraphs is that he made none; that, as Minto puts it, "each sentence stands on its own bottom.'"" Not that there are no short sentences; there are as many as in Cardinal Newman.^ But the long sentences are exceedingly periodic, complex and involved. Hooker's Polity stands as the most deliberate attempt to abandon the paragraphic tendencies of the vernacular and mold English prose to the syn- tax of the Ciceronian period. HAKLUVT. Principal A'avigations^ 1598- Average predications per sentence f 4.22 Per cent, of simple sentences (Gerwig) ■ 12 Per cent, of clauses saved ' 17-54 In HdiWuyi''?, Principal Navigatio/is, &^c., of the English Natioti ' P. 467. ^ Manual, p. 220. Minto speaks, however, as if Hooker made paragraphs. 3 Cf. Sherman, University Studies, Vol. I., No. 4, p. 363. 88 HISTORY OF THE /■XGLIS/I TARAGRAPH. (ed. of 1 5()<)) the paragraph inL,^ is as iiiedianical as in Stow or the early chroniclers. The voyages are usually reported in the form of a ship's log, each day's record paragrajjhed by itself. Again, in some parts of the book the habit of " itenii/ing " reduces paragraphs on the one hand to a few words ; while in other parts there are frequent long sections, broken only by the marginal note. GREENE. The Elizabethans were perplexed as to the right way of i)ara- graphing conversation. Sidney relapsed into hopeless confusion on the subject. Greene in his earlier work observes no distinct method excej)t now and then to make a transitional step between the stadia that contain dialogue. But in Me/iapho/i (1589), the book that marks the change from Greene's exaggerated Euphu- ism to his more truly individual manner, the author holds before himself the rule of paragraphing each successive speech. Accord- ingly in this book the paragraph-length is much lowered : the average is hardly over 70 words, while in the first part ol Mamil- lia (1583) it had been about 240, and as late as Paiidosto (1588), had approached 200. Though the paragraph-length does not much rise in the pamphlets written after Menaphoii, neither does it decrease. There is no steady improvement after Maiaphoit, in the paragraphing of dialogue. The reader soon learns to expect little in point of unity in Greene's paragraphs. Each begins a new step, but too frecjuently the writer fails to see when the step ends. Again, over-illustra- tion hurts both the unity and the proportion. The coherence of Greene's paragraphs is fairly good. The movement is light and sometimes rapid, and the Euphuistic parallelism does not retard the general progress. Pro})ortion, however, is wholly missing. Greene was guilty of numerous clause-heaps, and of unnecessary single-sentence sections. The general loose structure of his sentence does not save him from the bane of his day — the excessive use of intermediate punctu- ation. TYNDALE TO TEMPLE. . 89 NASH. Nash is in nearly every point a better paragrapher than Greene — faint as this praise may be. The proportion of wholly amorphous paragraphs is small in the History of Jack Wilto)/. Nash's pamphlets are arranged with something like real orderli- ness. The units are tolerably long in most of his work : from about 260 words in the Aiiafo)ny of Absurdity the paragraph descends in the introduction to Greene's Meiiaphon to about 250, falls still nearer the 200 mark in Pierce Penilcss, and reaches its lowest point, say 160 words, in Have With You to Saffron Waldcn. There are more short sentences than in Greene, and the transition between them is more accurate. LODGE. Lodge's sentences are short, and the sequence between them is good. But the writer is utterly without paragraph-method. There is no unity in the sections of the Defense of Poetry, Music and Stage Plays. Those of the Alarum against Usurers are short and often false. The Historie of Forbonius and Prisceria has. one only merit — it paragraphs long speeches by themselves, placing between each' a short transitional paragraph. In Rosalynde the story runs wild — freshest and brightest of narratives running on heedless of whither away. Dialogue makes no difference. Para- graphs come only when the pen gives out. RALEGH. In the 1666 edition of the History, the mark §, in its early type form, divides chapters into long sections, called by Ralegh paragraphs. Each of these is subdivided by indentation. But curiously enough, Ralegh often subdivides a paragraph into what he calls sections, indicating each by reference marks, as I; or f. Thus he leaves no name for the indented paragraph, and exactly reverses the modern meaning of the words paragraph and section. Ralegh makes amazing show of systematic arrangement ; but the analysis is often arbitrary and inexact. On the one hand 90 HISTORY OF THE ENGLISH PARAGRAPH. this poor analysis, on the otlier his ullerly unwieldy and elephan- tine periods, make him an exceedingly bad paragraph ist. THK AUTHORIZED VKKSION. No English version of the Bible was broken into "verses" until 1 55 1, when Robert Stephens of Paris printed an edition with paragraphs similar to Tyndale's (already mentioned), and marginal figures indicating sub-sections of the paragraphs. The Geneva Bible, c. 1560, was the first to indent these sub-sections. The Authorized, 161 1, followed the Geneva in this respect. The main sections in the chapters were indicated by the mark ^, which, oddly enough, does not occur after the twentieth chapter of Acts. This last named fact prevents me from giving exact averages for the lengtli of these paragraphs, for while the whole number of words in both Testaments has long ago been counted by patient hands, I have not been able to get the statistics of the text minus the unparagraphed portion ; and it goes without saying that there is nothing to gain by making a count. The paragraphs of the Neiv Testame)it average, I should say, not far from 560 words, 7 verses, each. There is nothing necessary in these divisions, and they have been departed from by other edi- tions, such as the Cainbridi^e Paragraph Bilde of Dr. Scrivener, and by the Revised J'ersio//. BACON. The Advancement of Learning, 1605. Total paragraphs considered no Average words per paragraph 204.67 (Ed. of 1633.) Average words per sentence 60.03 (Ed. of Aldis Wright.) Average sentences per paragraph 3.41 Per cent, of single-sentence paragraphs 32 (Ed. of 1633.) Essays. Average words per sentence (.Sherman) 28-|- Average predications per sentence ( 3-12 Per cent, of simple sentences (Gerwig) -> 19 Per cent, of clauses saved ( 2.87 It is an immense pleasure to emerge from the mistiness of TYNDALE TO TEMPLE. 91 early Elizabethan prose into the h/me/i sicciim of the Advancement. Although it is evident that the sentences are latinized and excessively long, yet there is a very remarkable skill evident in the grouping of them. Whether consciously or not, Bacon employed all the power of a great rhetorician to reconcile the period and the paragraph. The intellectual style of the Advance- ment was favorable to the experiment ; there was no need of short sentences appealing to the will or to the picturesque imagination ; the experiment succeeded as far as it could ever succeed. The paragraphs are not too short ; they are methodical and orderlv ; the sequence is secured by explicit reference in the shape of demonstratives and conjunctions, and above all by the logic of a great dialectician. Not a paragraph lacks unity. But with It all we are conscious that the transitions are painfully formal and mechanical ; and that there is a lack of variety in structure not compensated for by brilliant rhetoric. There is no living web of discourse. The period and the paragraph have come into conflict as units, and their antagonism has for the nonce been frozen by logic into stonv civility. There is a stiff monotony in Bacon's wav of opening a paragraph w^ith, "Now as to the first point," and closing it with, "Thus much for the first point." When we turn to the Essays, we find the spell broken and the period driven off the field. The average sentence has 28 words ; a sentence epigrammatic, of course, made of mere glints and gleams of truth, but still capable of arrangement in the paragraph. The author is not, however, aiming at sequence, or even at lucidity. The changes are abrupt. Bacon felt this, and, to show the relatively isolated nature of man}' of the paragraphs, placed the old ^ before each when the lack of coherence between paragraphs was particularly noticeable. THE CHARACTER-WRITERS. JONSON. In our speaking of the history of the isolated paragraph, the character-writers. Hall, Earle, Overbury, Breton, etc., must not be left out of account. In some of them, particularly Breton, 92 HISTORY OF THE ENGLISH PARAGRAPH. much iDctliod is manifested in these short paragraphs that char- acterize various worthies and unworthies. Breton's habit is to begin each character in tlie same wav as every other, with a bold sentence summing up the main vice or virtue of tlie subject. Tliis topic is tlien rmphasi/.cd in several short sentences, by no means necessarily in logical progression. In all these writers the sentence is short, and affects the apothegmatic, Theophrastian tone. The sequence is accordingly such as might be expected between apothegms. In some of the writers Euphuistic parallel- ism is prominent. This is especially true of Overbury ; but Over- bury's general structure is stiff and sometimes incoherent. I do not take it that the early journalistic isolated paragraph owed much to these writers, though the satiric paragraph of Pope's day was not without its resemblance to these earlier, but less per- sonal pasquinades. In speaking of the isolated paragraph we must rank Ben Jonson as a really important author, though one whose influence on contemporary prose was small. Many of the detached frag- ments of the Timber are complete whole compositions in minia- ture. The sentence is short, and its structure is surprisingly simple and direct. The paragraph coherence is nearly always admirable, notwithstanding Jonson's tendency toward epigram. LORD BROOKE. Life of Sir Philip Sidney, 1652. (Brooke died, 1628.) Total paragraphs considered 200 Average words per paragraph c. 150 Average sentences per paragrapli c. 2.7 Average words per sentence c. 55.53 Average words per paragraph in first 35 paragraphs. 158.66 Average words per sentence in first 35 paragraphs. . 55-53 Per cent, of single-sentence paragraphs in 200 para- graphs 35 Fulke Greville, Lord Brooke, is one of the worst offenders in the matter of inorganic single-sentence paragraphs. His actual percentage of such is not so high as that of certain other authors, but Brooke's paragraphed sentence is a heap of clauses extremely TYNDALE TO TEMPLE. 93 awkward and involved. In the matter of unity. Brooke is better than Sidney, and worse, though less formal, than Bacon. Really false paragraphs are rare, but transitional sentences properly introductory of a new section are often given in the preceding. All in all, Brooke's style, like Bacon's, is an illustration of the futility of any compromise between the paragraph and the long period as units of discourse. BURTON. Anatomy of Melancholy, 162 1. Total paragraphs considered 100 Total words considered 38.057 Total sentences considered 948 Average words per paragraph 380.57 Average sentences per paragraph 9.48 Average words per sentence 40.14 Per cent, of single-sentence paragraphs 18 If excessive variability in stylistic averages goes to show men- tal irregularity, then Burton's paragraph-length would prove him mad. The word-length of the Anatomy varies from 39 to 2529. Evidently there is little system here. When quotations press for- ward to Burton's pen, paragraph-method is a dead letter. On the other hand, when he is framing at leisure a section of orig- inal discourse, there is visible a stylistic spirit really new. Burton's sentence average is not greater than Swift's, and as Bur- ton's long sentences are exceedingly long, it is plain that he uses counterbalancing short propositions to an extent elsewhere unknown among scholars of his day. It is indeed not far from the truth to say, with Mr. Saintsbury,' in his latest utterance, that the arrangement of Burton's sentences is often "distinctly terse and crisp." In his long sent-ences Burton takes minute pains to subor- dinate, bv conjunctions and demonstratives, the lesser clauses: often, however, with poor success. But the number of initial connectives is moderate. Out of 300 sentences, 64 begin with conjunctives, and there are 15 interior sentence-connectives : ' Craik's English Prose, Vol. II., p. 117. 94 HISTORY OF THE ENGLISH PARAGRAPH. in Aschaui there were i68 sentence-connectives in 300 sentences. The list in Burton is given below. To get the nunjber of con- nected sentences 15 is to be deducted from a total 79, for repeti- tions. Connective. Initial. Interior. .\nd 19 Besides 4 But 13 For 6 First . . . I However 2 In a wortl I In conclusion I In like sort . . 2 Nay I On the other . . i Otherwise i Or 3 Thus 2 2 So 7 Then . . i Therefore . . 7 Yea I Yet 3 I We may say, therefore, that while Burton's paragraph-struc- ture is variable in merit, and while he is guilty of many amor- phous clause-heaps, he marks an advance toward one secret of proportion (the short sentence), and toward the modern method of sentence-connection. MILTON. A reopagitica , 1644. Total paragraphs considered 33 Total words considered 17,948 ' Total sentences considered 354 Average words per paragraph ' 543-88 Average sentences per paragraph 10.73 Average words per sentence 50.70 Per cent, of single-sentence paragraphs 10 Average predications per sentence ( 4.87 Per cent, of simple sentences (Gerwig) ■< 6 Per cent, of clauses saved ( 9.31 TYNDALE TO TEMPLE. 95 Milton ridiculed the short sentence and despised the loose order. His own sentence is highly periodic and involved : he stands with Hooker and Clarendon at the extreme of the clas- sical movement. Some of his prose works, as the Defense of the People of England, are practically without paragraphs. Others, as \.h.Q Eikonoklastes, have a mechanical paragraph — used for formal enumeration of points in an argument. Such paragraphs are likely to be of one sentence, and amorphous. The Areopagitica is the best paragraphed of all the works, though not so freely paragraphed as the Reformation in England; yet few even of the sections of the Areopagitica are really units. It must be noted, however, that Milton is distinctly subject to one tradition, that of paragraphing for emphasis. Thus, in the Eikonoklastes (ed. of 1649, p. 200) there is one sentence that is broken into twelve paragraphs. These mark the various "articles" that state the conditions on which the king, Milton asserts, is ready to capitulate with God. CLARENDON. History of the Great Rebellion, published 1 704-1 707. (Clar- endon died, 1674.) Total paragraphs considered 1 00 Total words considered 21,732 Total sentences considered 290 Average words per paragraph 217.32 Average sentences per paragraph 2.90 Average words per sentence 74-94 Per cent, of single-sentence paragraphs 28 In Clarendon the classical experiment is wrecked. Other men had written long sentences — Spenser, for instance. But Spenser's clauses follow each other like cars in a railroad train ; each could be uncoupled and sent singly on its way. Claren- don's long sentence may be likened to the same train "tele- scoped ; " where framework and ornament, ribs of wood and rods of iron, are jammed together and inextricably twisted out of all resemblance to any orderly thing. Clarendon, however, did not, like Milton, utterly disregard 96 IlISrORY OF THE ENGLISH PA RAG NAP! J. the paragraph. He repeated the rash compromise of Bacon and Brooke, and failed signally. After his day the experiment was never fullv repeated. Clarendon's paragraph-length is 217 words, his sentence 74 ; manifestly a paragraph of this length could not coexist with a formless sentence of 74 words. JEREMY TAYLOR. TJic Liberty of Prophesying, 1647. Total paragraphs considered 109 Total words considered 54, y!^? Total sentences considered 1035 Average words per paragraph 502.63 Average words per sentence 52.93 Average sentences per paragraph 9.49 Per cent, of single-sentence paragraphs 6 It was good service that Coleridge rendered the fame of Taylor by showing the great bishop to be not half so unintelli- gible as some rhetoricians had held. But Coleridge, who admired Milton's prose almost as extravagantly as Landor did, would have it that Milton is clear enough ; he fails to show how much more lucid Taylor is than Milton. Tavlor's sentence is less periodic than Milton's and far less involved than Clarendon's. But his sentences are long and there is the wildest abuse of conjunctions. Often the orator himself sees that the sentence is inadequate to his unit of thought, and leaves the period in hopelessly incomplete syntac- tical shape. Taylor's chosen unit is the period, but he does not try to reconcile it with the paragraph, and is himself conscious that he is in confusion on the whole question of structure. SIR THOMA.S BROWNE. Religio Me die i, 1643. Average words per paragraph c. 340 Average sentences per paragraph io-|- Hydriotaph ia, 1658. Total paragraphs considered 107 Average words per paragraph 125.08 Average sentences per paragraph 3.78 Average words per sentence 33-09 TYNDALE TO TEMPLE. The paragraphs in both the Religio and the Hydriotaphia are numbered as "sections." The Religio has a comparatively diffuse flow, natural in an autobiographical work first written for the eyes of friends only. The Hydriotaphia, written nearly a quarter of a cen- tury later, shows more of curious care, and the condensed language of a maturer mind. The change in the style is accompanied by a shortening of the paragraph from about 340 words to 125. Browne is often assailed as a Latinizer, and in one sense the criticism is true of his syntax as it is of his vocabulary. Browne used every Latin construction to which English would lend itself; but he knew the limits, as Milton and Taylor and Hooker did not. He almost never spun a web of involutions. He knew when both sense and rhythm required a full stop. Consequently his sentence is not long, not so long even as De Quincey's ; it may be added that there are in it far fewer unnecessary connectives than in De Quincey's. The sections of the Hydriotaphia have better unitv than can be found in Elizabethan prose outside of Bacon. The sequence and coherence are not relatively so good. Yet it may be questioned whether any other writer so aphoristic has so well succeeded in keeping logical articulation between sentences. Each group of Browne's strange gems has a general hue and harmony of its own. In point, then, of unity, of sentence-length, and of logical rather than formal articulation between sentences, Browne marks an immense advance over the men with whom Coleridsre classed him as a corrupter of English. In one other respect he marks advance ; namely, in the rhythm of successive clauses, and, to a less extent, of successive sentences. Milton had a dawning sense of the necessity of an occasional short sentence as a rhyth- mical relief from the roll of the period. Browne carried this principle of variety still farther, uniting with it a sense of tone-color that has never been surpassed. HOBBES. Leviatha/i, 165 i. Total paragraphs considered 200 Average words per paragraph c. 1 16.40 98 HISTORY OF THE ENGLISH PARAGRAPH. Average sentences per paragraph c. 2.96 Average words per sentence c. 39.26 Per cent, of single-sentence paragraphs 35 Hobbes marks no improvement in the matter of sentence- length as related to the paragraph. His sentence, indeed, is only 39, but his paragraph is not large enough to hold a unit of this size. It must nevertheless be admitted that the sense of unity in Hobbes's sentence is highly developed for the time, and that the paragraphs are usually units, though not always properly amplified. But the chief virtue of these paragraphs is their precision of artic- ulation, both internal and external. The coherence is eminently good, though the massing is so poor and the formal predications so awkwardly numerous that the reader's progress is but slow. LORD HERBERT. Autobiography, written c. 1643. Total paragraphs considered 40 Total words considered 9983 Total sentences considered 132 Average words per paragraph 249-I- Average sentences per paragraph 3.30 Average words per sentence 75-60 Per cent, of single-sentence paragraphs 25 Lord Herbert's AutobiograpJiy was not published till 1764, when Walpole edited and printed it. How closely the paragraph- ing follows the MS. I do not know. Mr. Saintsbury has said of Lord Herbert, "The writer dis- plays an art, very uncommon in his time, in the alternation of short and long sentences, and the general adjustment of the paragraph.' Mr. Saintsbury is usually not far from the truth in his comments on prose structure, but surely in the present case he has overestimated Lord Herbert's paragraphic tend- dency. Out of the first 132 sentences in the Autobiography only seven fall below 20 words, and the average is 75 words, a length reached but once in the worst days of the paragraph, hitherto. These figures are enough to limit seriously the force ■ History of Elizabethan Literature, p. 439. TYNDALE TO TEMPLE. 99 of Mrl' Saintsburv's dictum. It may, however, be granted that this dictum is not without some foundation ; for Lord Herbert has a knack of making a paragraph of two or three sentences, the first very long, the second moderately short. There is an exceptional example of this in the sixth paragraph ; here the first of the two periods has 329 words, the second but 27. In the thirty-sixth paragraph there is another exceptional example, where the introductory sentence has 20 words, the second and only other has 552. Here therefore \ve find a nascent sense of paragraph rhythm, and this is really Lord Herbert's contribution to the development. In other respects he marks no advance ; his monotonously ponderous periods and enormous para- graphed sentences belong to the conflict between period and paragraph. WALTON. Life of Hooker, 1665. Total paragraphs considered 1 06 Total words considered 19.S42 Average words per paragraph 187. 19 Average sentences per paragraph 2.9 Average words per sentence 64 Per cent, of single-sentence paragraphs 25 Complete Angler, 1653. Total paragraphs considered 200 Average sentences per paragraph 1.5 Per cent, of single-sentence paragraphs 73 (dialogue) In spite of his colloquial tone, garrulous Isaac Walton belongs only too evidently to the older prose. He is not one of the Latinists, but his sentences are very long and guileless of unity. His numerous single-sentence paragraphs are nearly all clause- heaps, except in the Angler, when conversation controls them. Walton's style is childlike in its abuse of coordinating con- junctions ; it belongs in this respect almost as far back as Mande- ville ; 80 sentences out of 300 begin with aiid. The list of sen- tence-connectives from the Angler is as follows : The number of sentences connected (out of 300) is 130, or 16 less than the whole number of sentence-connectives. lOO I//S'JOKy OF THE KiXGI.ISJf PA R. 10 R. 11 11. Connective. Initial. Interior. And 8o But 25 Nay 5 Thus 2 First 3 8 Also . . 6 Well - 5 Hence i However . . i Therefore . . 2 Then i 2 On the contrary I I Indeed I So . . I Now I In the matter of unity, Walton's paragraphs are hardly defensible. Mr. Lowell's remark on Walton's poetic style, that he has "a habit of leaving the direct track of narrative on the suggestion of the first inviting by-path," ' is equally true of Walton in his prose. FULLER. Worthies of England, 1662. Total paragraphs considered 100 Total words considered 8677 Total sentences considered 370 Average words per paragraph 86.77 Average sentences per paragraph 3.70 Average words per sentence 23.45 Per cent, of single-sentence paragraphs 20 » Fuller's paragraphs are light and short, but mechanical rather than literary. He advances with regularity and order, complet- ing each step with a satisfied air of precision. But not half the paragraphs are built of fully developed sentences ; verbs are omitted with the utmost nonchalance, and the section often degenerates into a mere list of particulars. Fuller's sentence is short and pointed, though he exhibits no great skill in its adjust- ment in the paragraph. ' Lowell, Latest Literary Essays, Boston, 1892. TYNDALE TO TEMPLE. loi Fuller is not free from digressions, but, as Minto observes, he is always conscious of his digression, and takes care to return explicitly to the original topic. By far the larger number of sections state the topic first. Fuller's regularity in the use of the deductive order makes him the precursor of Johnson and Macaulav. On the whole. Fuller is distinctly in the new line of paragraph development. In sentence-length and in general method he is the most modern man of his time. SELDEN. Tabic Talk, 1689. (Selden died, 1654.) Total paragraphs considered 81 Average words per paragraph 72 . 97 Average sentences per paragraph 2.17 Average words per sentence 33-58 Per cent, of single-sentence paragraphs 43 Selden's Table Talk was published thirty-five years after the author's death. The arrangement of the dicta was made by Selden's amanuensis, Milward. Milward usually groups under one heading several related remarks, something in the style of Bacon's Essays. All Selden's paragraphs are indeed relatively isolated. They are not often organic wholes, but are mentioned here as being in the line of general sentential development, the sentence being -i^t^ words. HOWELL. Mr. Joseph Jacobs, in his recent edition of James Howell's Familiar Letters, regards Howell, rather than Dryden, as the father of the short sentence. It may be admitted that Howell had a knack, more pronounced than that of Lord Herbert, of occasionally alternating an exceedingly long senteijce with a short one. But as for Howell's being the father of the short sentence it is enough to say that in the original edition of the pamphlet, England's Tears for the Present Wars, the sentence average is actually 77 words, one of the very highest in the history of our prose. Nor yet had Howell advanced otherwise to the modern 102 HISTORY OF THE ENGLISH PARAGRAPH. conception of the paragraph. He had no proj)er sense of unity or proportion. COWLEY. Essays, from sixtli edition, 1680. Cromwell \\\x'i 14 paragraphs, 12,574 words. Proposition for the Advancement of Experimental Philosophy has 51 paragraphs, 5132 words. Average words per paragraph — Cromwell 898.14 Average words per paragraph — Philosophy 100.62 Average words per sentence (two essays) c. 48.37 Average words per sentence — Cronnvell c. 38.01 Average words per sentence — Philosophy c. 54-43 Cowley's prose is transitional ; his early sentence is long and unmethodical ; his later much shorter and more homogeneous. The sections in the earlier editions (for the paragraphing in the later ones, even Grosart's, cannot be trusted) are very irreg- ular and inorganic. Cowley is not actually disorderly, but he ' has no proper sense of paragraph method. He will in one chapter paragraph a single sentence for emphasis {e. g. Essay on Liberty, ed. of 1680, p. 82 ; Essay on Solitude, p. 92) ; in tlie next chapter he will allow a paragraph to run on for six pages. Cowley is in the line of advance, but distinctly so in one par- ticular only. He spares initial connectives and depends for sequence on logical succession. Of the 57 sentences (Lumby's ed.) in the Essay on Greatness, not one begins with and, only two begin with but, and only seven with subordinating connectives, including relative adverbs. BUNYAN. The Pilgrim's Progress, 1 678-1 684. Total paragraphs considered 200 Total words considered 12,520 Average words per paragraph 62.60 Average sentences per paragraph 1.98 Average words per sentence 31 -61 Per cent, of single-sentence paragraphs 61 Average predications per sentence ( 3-91 Per cent, of simple sentences (Gerwig) i 10 Per cent, of clauses saved ( 5.92 TYNDALE TO TEMPLE. 103 The dialogue form of the Pilgrim's Progress of course deter- mines in a large measure the length of the paragraph in this work. We have in Grace Abounding an example of Bunyan's sustained prose ; but the paragraph of Grace Abounding I have not been able to examine in any early edition. We find at last in the Pilgrini's Progress a sentence which belongs to the essential paragraph structure. Bunyan has mastered the short sentence. He can vary it with longer ones — not very periodic ones — and produce effects of severe variety and of sober rhythm. The most important outcome of the age that ends with Bunyan is this short sentence. The vernacular stream that has found its way through the obstacles of the age emerges bright and strong in Bunyan. When the next period of development sets in the writers gradually bring this short sentence into the serv- ice of the longer thought-integer, and so the new unit of style \ is evolved. CHAPTER VII. TEMPLE TO DE QUINCEY. TEMPLE. Heroic Virtue, 1692. Total paragraphs considered 184. Total words considered 28,775. Total sentences considered 538. Average words per paragraph 156-30 Average sentences per paragraph 2.90 Average words per sentence 53-40 Advancement of Tiade in Ireland, 1692. Total paragraphs considered 40 Average words per paragraph 226 -f- Average words per sentence 54 ~l~ It is probably useless to dispute whether, as Mr. Saintsbury says, Temple was a follower of Dryden, or whether, as Mr. C. D. Yonge thinks, Dryden imitated Sir William. Both men were probably indebted to Jonson, Cowley, and even Bunyan, though from Sir William's sentence-length one would hardly think so. What is certain for our purposes is that Temple's first important work, the Observations on the Netherlands (1672), is far more care- fully paragraphed than the Essay of Dramatic Poesy (1667) ; and again that in ordering of matter Dryden's best work cannot compare with Temple's best. Temple's sentence is indeed too long ; it is longer than Dry- den's and more than twice as long as Fuller's. But the clauses follow the simple oral form, and for the first time in our prose we have a balance and a cadence that are not manifestly artificial. This unobtrusive balance and the parallel construction of sen- tences are an immense help structurally to the coherence of the paragraph. Of course the balance will now and then degenerate into an artificial pointedness that, by tending toward epigram, hurts 104 TEMPLE TO DE QUINCEY. 105 the sequence. But the predominating effect is that of close-knit prose. Another virtue, most important historically, marks Sir William's sentences. Though long, they rarely lack unity. Temple's coherence depends very largely on structure. Of 300 sentences in the Heroic Virtue o\\\)- 51 are joined by con- junctions. The list follows ; it will be seen that double con- nectives are not used, for the whole number of connectives is the same, within one, as the whole number of connected sentences. It is perhaps true that the brevity of the list shows French influence. Connective. Initial. Interior. And 13 Besides i i But 12 Also I For 6 In short i It is true I Likewise . . 2 Finally . . i Nor I On the other side 2 So 3 Therefore . . i Thus 3 Yet 2 Sir William's contribution may be described as an increase of coherence by structure ; and of skill in transition between para- graphs. DRYDEN. Translation, 1685. Satire, 1693. Parallel betiveen Poetry and Painting, 1695. Average words per paragraph in Essay on Satire . . . 256 -|- Average words per paragraph in Translation -\- Par- allel hetzveen Poetry and Painting 261 -|- Whole number words in Satire -\- Tratislafion -j- Parallel 49,q69 Whole number sentences in Satire -\- Translation -\- Parallel 1 300 io6 ///STOA'V OF THE ENGLISH PAKAGKArH. Whole number paragraphs in Satire -f- Translation + Parallel l8o Average words per paragraph in Satire -\- Transla- tion -\- Parallel 277.55 Average sentences per paragraph in Satire -\- Trans- lation -\- Parallel 7.22 Average words per sentence in Satire -\- Translation -\- Parallel 38.44 Per cent, of single-sentence paragraphs in Satire -\- Translation -f- Parallel 1 1 + Dramatic Poesy, 1667. (Gerwig, for 521 periods.) Average predications per sentence 4.89 Per cent, of simple sentences 6 Per cent, of clauses saved 4.88 The fame of Dryden as our first great prosaist is not enduring without challenge. Mr. Saintsbury' and Mr. Gosse^ credit Dry- den with full mastery of English prose ; on the other hand Mr. Minto^ referred to his genius as the reverse of methodical, and Mr. Sherman '' calls him almost as formless as Spenser. It is true that Dryden brought to the writing of prose a vigor unknown before ; that he exhibited a rich vocabulary of simple speech, and a general felicity of diction ; that he was not equaled in his day in power of varying the structure of the sentence and giving it flexibility and balance. On the other hand, it is not quite the whole truth to say with Saintsbury that his slovenli- ness in sentence-structure is only occasional. Dryden was singu- larly uneven in his sentence-writing, and it is safe to say that no single piece of his prose is free from impossible periods. With every deduction Dryden nevertheless remains the most potent individuality in modifying the sentence to reasonable pro- portions. He stands as a dividing line between the old sentence and the new. But as a paragraphist he is inferior to Temple. His genius is a vagrant one, and he sins incessantly against the ^Specimens of English Prose, p. xxii. '^History of Eighteenth Century Literature, p. 91. '^Manual, p. 334. ^ Analytics, p. 292. TEMPLE TO DE QUINCEY. 107 cardinal law of unity. There are indeed plenty of good para- graphs in Dryden, but the good ones are nearly always short. It is a curious fact that in spite of lack of logical severity in the analysis of the whole composition, and in spite of the very great fluctuation in his paragraph-lengths, Dryden had some sense of rhythmical proportion in distributing his matter by paragraphs. The word-length of the paragraph in Satire^ is the same within five words, as in the combined essays, Translation + Parallel. Dryden had indeed, as everyone knows, a distinct feeling for "the other harmony of prose." This rhythmical sense gives us, along with the cadence of the sentence, a feeling for parallel construction. To this his paragraphs often owe a coherence that goes far to make up for his digressiveness. He depends for coher- ence largely on the order of words, and this regard for order of words helps to make him the most forcible of the early prosaists. He does not rely on initial sentence-connectives. Out of three hundred sentences, only twelve begin with rt'//^', eighteen with but ; while subordinatives are still more sparingly employed. His sentences^ improve as his style matures; few authors show so much change. The improvement is not so marked in the matter of paragraph unity ; the sound thinking of later life does not seem much to check his fits of irrelevance. LOCKE. Essay on Human Understanding, 1690. Total paragraphs considered 200 Whole number words 40,545 Whole number sentences S14 Average words per sentence 49.8 Average words per paragraph 202.7 Average sentences per paragraph 4.07 Per cent, of single-sentence paragraphs i S -j- It must be admitted, even by those who think that Locke crossed the line where writing ceases to be Jiterature, that he ' 1 choose late essays as exhibiting Dryden's matured style. ^ Sherman puts Dryden's sentence at 45-|- ; it seems to me this must be an average from the early prefaces. io8 IflSTORV OF THE KXGIJSII PARAGRAPII. is the most orderly writer of his day, though his method is purely formal. The early editions of his works are very carefully analyzed by chapters and sections, the latter being marked §. The section usually coincides with the paragraph, l)ut not always. Many editions have marginal summaries of sections, and tabular summaries in the table of contents. In general the paragraphs have good unity. The paragraph is short relatively to the sentence, but Locke does not, like Hobbes, paragraph tiny stadia for emphasis. His failure to reach the rhetorical paragraph lies largely in the fact that his paragraph lacks proportion. He dwells on the unimportant at the expense of the important. Half his introductions are too long. Nor is his coherence so good as might be expected in a writer who has so much to say about the value of consecutive- ness in thought. He often brings illustrations from a distance and introduces them abruptly. DEFOE. Essay on Projects (1697), omitting numerical accounts. Total paragraphs considered 200 Whole number words 16,978 Whole number sentences 342 Average words per sentence 49.64 Average words per paragraph 84.89 Average sentences per paragraph 1.7 1 Per cent, of single-sentence paragraphs 62 Rohi>isoii Crusoe (17 19), 200 paragraphs. Total words 28,327 Total sentences 360 Average words per sentence 78.68 Average words per paragraph 141.63 Average sentences per paragraph 1.87 Per cent, of single-sentence paragraphs 60 If Locke is, in certain formal respects, the best paragrapher of his day, Defoe is in all respects the worst. He really knows no difference between the sentence and the paragraph ; he para- graphs for emphasis only. The sentence of Robinso)i Crusoe is nearly as long as the paragraph of the Essay on Projects. It TEMPLE TO DE QUINCEY. 109 would be hard to find another writer of such irregularities in sentence-length. Defoe's coherence in narrative is good, for his pictorial imagination is exceedingly vivid, and his diction and method thoje. of swift, lucid conversation. But in argument all this is changed. Here he neglects every device of transition and pours out his ideas in the most haphazard way. In argument he is vigor- ous enough, but his vigor is wasted by utter disregard of method. SWIFT. The Battle of the Books, 1704 (written 1698). Total paragraphs ^^ Total words 9234 Total sentences ^3- Average words per sentence 39-^0 Average words per paragraph 297.86 Average sentences per paragraj^h 74° The Tale of a Tub, 1704 (written 1698). Total paragraphs considered 100 Total words considered 1^5,577 Total sentences considered 450 Average words per sentence 40-74 Average words per paragraph i^5-77 Average sentences per paragraph 4-5o Per cent, of single-sentence paragraphs I5 Average predications per sentence (Gerwig, J 3.69 Per cent, of simple sentences for 500 V 13 Per cent, of clauses saved periods.) \ 9-23 Travels of Lemuel Gulliver, 1726. Total paragraphs considered 200 Total words considered 4^,^44 Total sentences considered " 7 1 Average words per sentence 40.00 Average words per paragraph 234.22 Average sentences per paragraph S-^S Per cent, of single-sentence paragraphs I5 The unity of Swift's paragraphs is usually all that could be desired. Now and then, however, a paragraph will be so long as to obliterate, apparently, any sign of topic. These rare para- I lo IIISTOKY OF THE ENGLISH PARAGRAPH. graphs are almost inexplicable when compared with his usual sections. Professor Cesare Lombroso would, I tear, find the eccentricity of madness in them, as he did in the inversions of the Dean's conversation. Swift's command of jjroportion by paragraph-punctuation is small. It is noticeable that the proportion of very short sen- tences (sentences under 15 words) is not large — 6.3 per cent, in the Tii/c of a Tub, 6.4 per cent, in Gulliver. The average of the sentence is constant, in works separated even by 28 years: the three books mentioned show a variation of less than a whole word in sentence average, though the paragraph-averages of dif- ferent books differ enormously. The superb coherence and emphasis of Swift's style are due largely to the straightforward, logical order of the thought, and the skillful placing of important words at the end of a sentence or paragraph. Swift is the first author to show in the paragraph much of what Wendell calls Mass. His sentences often fall at the close like taps of a steam-hammer, and sometimes the taps seem concentrated in one great blow at the end of the paragraph. Connectives he uses less than does any of his predecessors. The list from Gulliver is as follows — showing only 39 formally connected sentences out of a total of 300. Connective. Initial. Interior. But 14 Therefore . . 2 Likewise . . 4 However 6 i Whereupon 2 Besides i Also . . I Thus . . I So . . I Now 3 For 2 Indeed . . i ADDISON. Tlic Freeholder, 1715-1716. Total paragraphs considered 200 TEMPLE TO DE QUINCE Y. Ill Total words considered 34,651 Total sentences considered 898 Average words per paragraph 173-25 Average sentences per paragraph 4.49 Average words per sentence 38.58 Per cent, of single-sentence paragraphs 14 Spectator. (Gerwig, for 500 periods.) Average predications per sentence 3.67 Per cent, of simple sentences 12 Per cent, of clauses saved 3.72 Addison's unity is usually faultless. His coherence depends largely upon word-order and sentence-structure ; of 300 sentences only 13 begin with and, 16 with but. His massing, when com- pared with Swift's, is defective. In brief, the paragraph structure is easy and flowing, correct in unity, defective in emphasis. Addison's favorite paragraph is loose, with one or two introductory sentences. Deductive specimens are not infre- quent. The topic is often developed by repetition from changing points of view, — what Scott and Denney have termed the alter- nating method. The method is frequently overdone. Addison had little sense of the value of the short sentence, either as a means of emphasis, or as a way of varying paragraph rhythm. His rhythm remained a somewhat monotonous sen- tence-rhythm. Less than 4 per cent, of his sentences fall below 15 words. There is no wide variation in the number of sentences to the paragraph : thus, 44 out of 200 paragraphs have three sen- tences each. SHAFTESBURY. Characteristics, i 7 1 1 . Total paragraphs considered lOO Total words considered 15,490 Total sentences considered 578 Average words per paragraph 154.QO Average sentences per paragraph 5,78 Average words per sentence 26.80 Per cent, of single-sentence paragraphs 3 Limits of fluctuation in paragraph word-length. . . . 44-341 1 1 2 HISTORY OF THE ENGLISH PARAGRAPH. Per cent, of sentences of less than 15 words 26 Per cent, of simple sentences (Gerwig, I 28 Average predications per sentence for 650 \ 2.61 Per cent, of clauses saved periods.) ) 4.02 Had he chosen, Addison might have learned much from the well-bred style of his contemporary, Anthony Ashley Cooper; and particularly he might have learned variety. Shaftesbury used the short sentence without stint, and often with fine effect. His percentage of sentences falling under the length of 15 words is the highest before Burke. His sentence is more variable than that of any author of his own time ; and he much sur- passes in this even Bolingbroke. It is to be admitted, never- theless, that Shaftesbury was not fully master of the short sen- tence. Many of his sentences are so brief that they utterly lack unity. In several other paragraphic virtues Shaftesbury is correct, though never firmly and surely so. His unity is good, the para- graph being very short. He follows the loose order definitely enough to give his topic in the course of the first two sentences. He is coherent, making one sentence follow, without need of connective, from the preceding. He has his faults, however. His massing is such as to obscure the emphatic words. Though his sentences do not need connec- tives, there is an abuse of initial coordinatives. His list of initial connectives is as follows, showing 79 initially connected sentences out of 300 \ the list would not be much increased if the interior sentence-connectives were added : For 20 However But And 26 21 2 2 2 Thus Now Vet Nor Nor Or So that ( )n the other .side All in all Shaftesbury may be regarded as contributing the ele ment of variety in sentence-length. His paragraph-length is not so variable. TEMPLE TO DE QUINCE V. 113 BOLINGBROKE. Letter to Sir William IVyndhani, 1753. Total paragraphs considered 173 Total words considered 34, 199 Total sentences considered 981 Average words per paragraph 197.68 Average sentences per paragraph 5.67 Average words per sentence 34-86 * Per cent, of single-sentence paragraphs 5 Per cent, of sentences of less than 15 words 13 Study of History. (Gerwig, for 977 periods.) Average predications per sentence 3.65 Per cent, of simple sentences 14 Per cent, of clauses saved 3.72 It may freely be granted that Bolingbroke's style is in some respects vicious; that, as Mr. Gosse says, it is "grandiloquent, and yet ineffectual."^ These faults affect unfavorably the empha- sis of his paragraph ; and yet, after every deduction, Bolingbroke is distinctly a modern paragrapher. He knows the value of the short sentence, though he does not use it freelv enough. Only 13 per cent, of his sentences fall below the length of 15 words ; yet he alternates long propositions and short ones, with telling effect. The unity of his paragraphs is generally unassailable. He looks to the transition between sentences, and, what was then more rare, to the transition between paragraphs. He balances sentences, sometimes to windy lengths, but does not let the coher- ence seriously suffer. He carefully eschews connectives, indeed rather too carefully. Above all he depends more on the paragraph than do his pred- ecessors. He is always making sentences that are unintelligible except in the light of the larger unit. He delights, as Macaulay does, in a preliminary generalization so sweeping and so indefinite as to require a multitude of subsequent propositions to unravel ^History of Eighteenth Century Literature, p. 174. 114 HISTORY OF THE ENGLISH PARAGRAPH. tlie puzzle. He has deliberately adopted tlie paragraph unit, and it is evident that troin the study of him some of the best English paragraphists, notably Burke and Macaulay, have their cue, slight as that cue is. RICHARDSON AND FIELDING. We saw that the Elizabethans developed no fixed method of paragraphing dialogue, though Greene and Nash tend irregularly toward the modern method of setting off each speech. In eighteenth-century novels the question is still in dispute, though not in utter confusion. Richardson in general paragraphs each speech. He does not use quotation marks. Rarely he gives the dialogue in dramatic form, prefacing each speech merely with the speaker's name. Some of the paragraphs are exceedingly short, even when they form part of a monologue. Though it is asserted that Richard- son did not read French, it is hard to believe that he was not influenced by French models in paragraphing. Else how could he, a practical printer, bring himself, say, to a series of 13 para- graphs, averaging 28.38 words (as in Clarissa, Vol. 2, Letter xxiii.) ? This is certainly Marivaudage in structure, even if Pamela be not indebted to La Vie de Marianne for its plot. But Richardson wrote in letter-form, a style that seems always to produce degeneration in the paragraph-conscience. Fielding, writing under no such artificial scheme, is a better paragrapher than his predecessor. The paragraph word-length in Tom Jones is 101.86, or 2.43 sentences of 41.92 words. The percentage of paragraphed sentences is high (38 per cent.), but this fact is not due principally to dialogue ; the great novelist studied carefully the unity of his narrative paragraphs. His principle in dialogue is, I think, something like this : paragraph primarily for unity, breaking up monologue or massing dialogue if the speeches are short and the movement rapid ; when the dialogue is leisurely, paragraph each speech. An example of breaking up monologue may be seen in Mr. AUworthy's Homily to Jenny, chap, vii.. Book i. Examples of the massing together of short speeches TEMPLE TO DE QUINCE Y. 115 when the movement is impassioned or hurried, may frequently be found. Thus, B. iv., chap. 14, paragraphs 12, 13 ; B. v., chap. 4, paragraph 2 ; B. v., chap. 6, paragraph 17 ; B. vi., chap. 6, paragraph 1. JOHNSON. Ramblei\ 175 0-1752. Total paragraphs considered 94 Total words considered 9600 Total sentences considered 218 Average words per paragraph 102.13 Average sentences per paragraph 2.32 Average words per sentence 44-03 Rasselas, r759. Total paragraphs considered 58 Total words considered 5357 Total sentences considered 174 Average words per paragraph 92.36 Average sentences per paragraph 3 Average words per sentence 30.78 Rasselas -\- Rambler. Total paragraphs considered 152 Total words considered 14.957 Total sentences considered 392 Average words per paragraph 98.40 Average sentences per paragraph 2.58 Average words per sentence 38-15 Per cent, of single-sentence paragraphs 27 Per cent, of sentences of less than 15 words 9 Lives of the Poets, 1 7 7 9- 1 7 8 1 . (Gerwig, for 500 periods.) Average predications per sentence 3.23 Per cent, of simple sentences 16 Per cent, of clauses saved 7.09 Everybody knows that Johnson's style varies in different works. This variation follows a steady chronological develop- ment toward the vernacular: the Ramblei-, 1750-1752, is the most latinized of his works; the Lives, 17 79-1 781, is the least latinized. Rasselas, 1759, stands midway in development. I i6 HISTORY OF THE ENGLISH PARAGRAPH. The sentence drops one-third of its length between the Rambler and Rassclas, but the paragraph drops only one-tenth. I have no complete count for the Lives, but should guess that the sen- tence still grows considerably shorter, and that the paragraph remains approximately in statu quo. Johnson's paragraph is remarkably short. In the Rambler there are but 2.32 sentences to the paragraph ; the two rises to three in Rasselas. The fewness of the sentences per paragraph and the high percentage (27 per cent.) of paragraphed sentences are phenomena not due in either case to dialogue. Johnson was exceedingly particular that each paragraph should form an integer; beyond this he cared not how few the sentences. His favorite order is loose, with a large share of deductive paragraphs. He loves a short introductory sentence, and when the chance permits he likes to make this sentence a generalization far wider than can be substantiated from the subsequent details. In the matter of proportion by varying short sentences with long Johnson in his later work is by no means weak. Even in the earlier works the percentage of sentences of less than 15 words is considerable — 9 per cent, in Rambler and Rasselas, while the Lives shows 16 per cent, of simple sentences. As to coherence, it is common to accuse Johnson, as De Quincey did, of " plethoric and tautologic tympany of sentence ;" ' or to say with Coleridge that his antitheses are usually verbal only.- But, at least in Rasselas and the Lives, the style is after all highly coherent. The antithesis is of course elaborate, but it has the effect of parallel construction and is not seriously retarding. The directness of the thought and the skill of the balance do away with the necessity of formal connectives. Few men have used initial connectives less than Johnson did, and none has depended less upon them. Of 300 sentences in Ras- selas 25 only are joined by formal conjunctives, whether initially or internally. The list is short. And occurs but once. ^ Works, X., 128. =^ Table Talk, Nov. I, 1833. TEMPLE TO DE QUINCEY. 117 Connective. Initial. Interior. Thus 3 I But 10 However i i So I I Yet 3 I Therefore . . 3 And yet i Johnson's chief contribution to the development is this man- agement of coherence without the use of connectives. Contrast the da_y when Walton showed eighty initial amis to 300 sen- tences, and the time when Johnson wrote but one and to the same number — 300. When Johnson did use connectives, they were never formal. As Coleridge said, " You cannot alter one of them without spoiling the sense." ' Johnson likewise fixed permanently as a model the loose order, with a preference for the deductive type. JOSEPH BUTLER. The involutions of the sentence in the Analogy are often impassable, as Emerson would sav, and utterly opposed to par- agraph structure. Butler is mentioned here merely for the fact that he has a larger percentage of strictly inductive paragraphs than almost any other writer in the language. It ma}' be added that when his sentences are short they usually need the light of the whole section to make their bearing plain. HUMK. History of England, Yo\. I., 1754. Total paragraphs considered 200 Total words considered 47,775 Total sentences considered 1 200 Average words per paragraph 238.87 Average sentences per paragraph 6 Average words per sentence 39-8 1 Per cent, of single-sentence paragraphs i Limits of fluctuation in paragraph word-length .... 48-697 Per cent, of sentences of less than 15 words 5 ^ Table Talk, July 3, 1833. I iS HISTORY Of THE ENGLISH PARAGRAPH. Per cent, of simple sentences ( 12 Averaije predications per sentence (Gerwig) - 3.29 Per cent, of clauses saved / I4-7I Dr. Johnson declared that Hume's style was not English, but French. If he meant by this that Hume was careful to use le mot propre, and that the study of French models had taught him more sententiousness than was then common in an English writer, and again that Hume aimed always at lucidity, why, then, Dr. Johnson was right. But in general structure Hume is not French : his sentences and paragraphs are too long, too monotonous. Johnson's own sentence was nearer French models, in the one point of length, than Hume's was. John- son's paragraph is but half the length of Hume's. Johnson's own use of the very short sentence was better and more Gallic than Hume's : Hume shows but 5 per cent, of sentences fall ing below the length of fifteen words, while Johnson shows 9 per cent. Hume has French lucidity, but he is stately, meas- ured, cold. Johnson, unconsciously following Gallic precedent? delights in short single-sentence sections, even to the extent of 27 per cent, of his whole number. Hume disdains a paragraph of less than five sentences, and writes but one per cent, of paragraphed periods. Johnson, out of 152 paragraphs, shows many successive paragraphs of one sentence, many of two, but none of five. Hume, out of 200 paragraphs, shows many successive groups of five sentences, but none of one, none of two. Johnson shows no successive groups of more than six sentences each ; but Hume shows such groups of seven, of eight, of nine. As regards structure, therefore. Dr. Johnson's dictum hardly holds ; though no one could dispute that dictum in the matter of Hume's vocab- ulary. Hume is impeccable in paragraph unity from the point of view of subject analysis. His unity depends on the philosophic scheme, the previsedly careful articulation of framework. It is not the picturesque unity of Macaulay. In spite of occasional extreme sententiousness, and his very sparing use of sentence-connectives, Hume's coherence is always TEMPLE TO DE QUINCE Y. 119 good. The sententiousness is never left unexplained. If the reader is ever delayed it is by the balance of the sentence, but he is never seriously checked by this. In Hume the formal balance breaks in upon the sequence as waves pass beneath a boat and lap it sharply, but only to drive it onward. Hume's favorite order is loose, with a tendency to eschew initiatory sentences. The topic sentence is likely to be somewhat indefinite, becoming clear with the first amplifying sentences. To sum up : Hume represents the long paragraph adapting itself to the Johnsonian balanced sentence. His integers of style are larger than Johnson's, but less unwieldy than Gibbon's. He is retrogressive in percentage of very short sentences. STERNE. A Sentimental Journey, 1768. Total paragraphs considered ". 200 Average words per paragraph 71-37 Average sentences per paragraph * 1.95 Average words per sentence j^-So Per cent, of single-sentence paragraphs 55 Limits of fluctuation in word-length of paragraph. . . 5~2o8 Sterne is in many respects the most eccentric of our prosaists. M. Scherer would have it that he is wilfully sensational and meretricious — a literary mountebank. I should like to find some method in his madness, even at a point where he seems maddest : /. e. his habit of making a chapter of a few words. Chapter xiii., vol. ii., of TristrajH, contains one paragraph, three sentences (in dialogue) — a total of 29 words. Chapter xxvii., vol. iii., has two paragraphs, four sentences, 83 words. Chapter v., vol. v., has one paragraph, one sentence, 16 words. Chapter xxxix., vol. v., contains one paragraph, two sentences, 30 words. There are a dozen other chapters similar in length to these. All this is freak- ish enough, but is not so very odd in view of Sterne's long study of French models, from which he had learned the trick of the tiny paragraph. He chose to emphasize a thought by paragraph- ing it, as Anglo-Saxon scribes had done, long before — and it was but one bold step further, in the process of emphasis by I20 ///SIVKY OF rilE ENGLISH PAKAGKAPH. mechanical means, to make a chapter of the {paragraph as he had made a paragraph of the sentence. It is hardly to the point for a critic to complain that these chapters are logically incom- plete. Sterne was analyzing, not logically, but rhetorically; fastening attention on these small stadia simply for tlie imagina- tive suggestions involved in their pregnant brevity. I must, for one, confess to thinking the thing sometimes shrewdly done. Sterne is a lawless wight, but his recusancy has given us some things both quaint and good. There is little else of importance to note of Sterne's para- graphs. In managing dialogue he follows Fielding. HUGH BLAIR. The only reason for mentioning Blair amid so many of his betters is that he wrote popular lectures on rhetoric, in which he said a deal about proportioning the sentence, but noth- ing about the paragraph ; and one is curious to see if such men as Blair, Campbell, and Kames, personally followed paragraph law. Blair's smooth Shaftesburian style leads him securely from sen- tence to sentence ; he writes nearly six monotonous sentences to the paragraph ; he follows the loose order of procedure in the paragraph, and observes the law of unity. In brief, it is strange that such mildly correct rhetoricians as he, wrote respectable par- agraphs, but, amid the multitude of their stylistic theories, had no theory of the process. GOLDSMITH. Vicar of Wakefield, 1766. Total paragraphs considered 107 Total words considered 23,390 Total sentences considered 868 Average words per paragraph 218.59 Average sentences per paragraph 8.1 1 Average words per sentence 26.94 Per cent, of single-sentence paragraphs 8 Limits of fluctuation in word-length of paragraphs.. 25-976 Per cent, of sentences of less than 15 words 15 TEMPLE TO DE QUINCEY. 121 The Bee, and The Citizen of the World. (Gerwig, for 500 periods.) Average predications, per sentence 2.95 Per cent, of simple sentences 18 Per cent, of clauses saved 6.35 In Goldsmith we have a respectable degree of variability in sentence-length, and therefore of one chief element of propor- tion — though other sense of paragraphic proportion Gold- smith had none. The general sentence-length is low, and 15 per cent, of the sentences fall below 15 words; on the other hand there are a few periods of more than 100 words. Goldsmith's narrative sequence is perfect, little needing nor much using connectives. He has not such unity as some descrip- tive and narrative writers of the day, Fielding, for instance. He follows Fielding carelessly in the handling of dialogue. BURKE. On Conciliatio)t zvith America, 1775. Total paragraphs considered 145 Total words considered 23,907 Total sentences considered 916 Average words per paragraph 164.87 Average sentences per paragraph 6.31 Average words per sentence 26.09 Per cent, of single-sentence paragraphs 18 Limits of fluctuation in word-length of paragraph. . . 16-559 On the Sublime and Beautiful, 1756. Total paragraphs loi Average sentences per paragraph 1 1.68 ^ux'ke.''?, Sublime and Beautiful [?, divided into parts headed as sections. These are rarely broken by indentation and are so short as to constitute relatively isolated paragraphs. Relatively, because it happens that one section may grow out of another, and accordingly begin with such a word as but {e.g., part 3, § 15, part 4, § 12) or hence {e. g., part 5, § 6). In length the sections vary from five lines to as many pages, the average number of sentences being eleven. 122 IIISrOKY OF THE ENGLISH PARAGKAPJL Hut it is in his oratorv that lUiike's paragraphs are remark- able. Me exhibits here such cjuaUties as make him the best paragrapher our literature i)roduced before the present century. His unity is simple (as opposed to that of compound para- graphs) and organic. His paragraph bears the test, as Wendell has pointed out, ' of having its substance expressed in one organic sentence. For purposes of oratorical emphasis and oratorical rhythm, he has completely mastered the short sentence. His percentage of sentences of less than fifteen words is higher than the highest yet reached. Shaftesbury's was 26 per cent., Burke's is 29 per cent. " Blithe, crisp sentences" Burke is fond of using at the beginning and the end of a paragraph. Of 145 paragraphs in Coiiciliatiou, 22 per cent, begin with a sentence of less than 15 words; 11 per cent, with a sentence of less than 10 words. The effect is striking. Here are certain such terse introductions : " The proposition is peace." " My idea is nothing more." " My next objection is its uncertainty." " First, the people of the colonies are descendants of Englishmen." "The march of the human mind is slow." "My next example is Wales." "This is an assertion of fact." Genung, a good observer, has noted' in Burke the fine effect produced by putting last in a paragraph a sin- gle terse, summarizing sentence. It is in the body of the para- graph that Burke introduces his shortest periods — those of one, two, three, four, five, six words each. These come in some- times like veritable thunder-claps, enforcing the long, preceding propositions or forcing attention to those about to come. We inspect Burke's coherence. This he owes but little to formal contrivances. But is the only initial connective that appears frequently ; the oratorical mood is, perhaps, inclined to exaggerate the prominence of adversative ideas. Burke gives small heed to conjunctions, but he is explicit in his refer- ence, usually making each sentence contain some word that refers closely to the preceding sentence ; this word is very often one ^English Coinpontiou, p. 124. - Pi-nctical Rhetoric, p. 209. TEMPLE TO BE QUINCEY. 123 repeated from that preceding sentence. Again, he secures coher- ence by regular construction. His sentences rarely contain sudden and awkward change of method. No contemporary author employed parallel construction with such freedom, such variety, such subtlety of effect. At its best, the tide of his style moves with most rapid sweep, each thought starting in the same line as its neighbor, each sentence pushed on by the preceding, each falling to the point in swift succession, like waves on the beach. Now and then there is a redundance of words that quiets the movement, but does not alter its method. In this movement there is no conflict of unmanaged masses of thought, as in Tay- lor, no choppy sea of antithesis, as in Johnson at his worst. Angus speaks of sentences " each a complete thought, easily separable from the rest of the paragraph,"' as common in John- son and Burke ; but the remark is hardly just to Burke. Burke's coherence, again, is enhanced by the order of his sentences and words. The great orator had, to a degree unconimon even in the most eminent orators, the power of marshalling his proposi- tions in a specious order. His emotion never ran away with him ; he drove straight at his hearer's intellect — did so too con- stantly for his highest immediate success. There is always the impression of a convincing chain of logic. In short, Burke is the earliest great master of the paragraph, and in impassioned- prose he still remains a master of the para- graph. But for his lingering sense of the prime importance of balancing and rounding the sentence he is a nineteenth-century paragrapher, and one of the best. GIBBON. Rome,No\. I., \ 776. Total paragraphs considered 200 Total words considered 48,748 Total sentences considered 15^2 Average words per paragraph 243.74 Average sentences per paragraph 7.81 Average words per sentence 31 .21 ' Handbook of the Englis/i Tongue, § 736. 124 ///STOAT O/-' T///-: /-.XGLISI/ /\l/^ACI^.H'//. Per cent, of single-seiUencL' paragraphs o Per cent, of paragraphs of two sentences 25 Limits of fluctuation in word-length of paragraph.. . 49-484 Per cent, of sentences of less than fifteen words. ... lo Gibbon's paragraphs ]nay be said to have unity, if we admit that historical narrative tends toward a compound unit. Gibbon not infrequently subdivides his paragraphs by numerals, and often we feel that the undivided long sections contain subordinate stadia. He is retrogressive in the matter of sentence-length. Only 10 per cent, of his sentences fall below the 15-mark. His stately and sonorous periods have a harmony of their own, but it is not i)aragraph harmony. His sentences have much propor- tion, his paragraphs little. We admire the comprehensive analy- sis of the discourse into chapters and paragraphs, but we do not quite feel that the paragraph is an organism. It is a well-defined cage in which the splendid sentence is confined. His movement is not rapid, but the sequence is in general sure. Demonstratives are numerous. When an introductory pronoun would be ambiguous he adds a noun, seldom a repeated one, but rather a synonym. Inversions, so frequent in Burke, are infrequent here. Con- junctions the author utterly despises, depending on the sheer inertia of his rolling sentences to carry the thought ahead. No other writer examined shows so small a list of sentence- connectives. The abandonment of them is Gibbon's only con- tribution to the development ; and it may be questioned if the contribution is a real or a permanent one, depending as it does on balance in the sentence. Here is the list — showinir but 17 connected sentences out of 300 : Connectives. Initial. Interior. But 9 Vet 3 However . . 2 And yet i Nor I And thus i TEMPLE rO BE QUhXCEY. 125 It ma}^ be added that Gibbon's usual order is loose, but that a really deductive paragraph is rare. It is a mistake to suppose that Gibbon abounds in abstract general statements. He is, indeed, fond of the abstract noun, as Minto ' has remarked ; but he does not make sweeping generalizations in the Johnsonian manner. PA LEY. Moral a fid Political Philosophy, 1785. Total paragraphs considered 200 Total words considered I4j77I Total sentences considered , 392 Average words per paragraph 73-85 Average sentences per paragraph 1.96 Average words per sentence .... 37-68 Per cent, of single-sentence paragraphs 58 Limits of fluctuation in paragraph word-length 6-575 The averages given above from Paley are lower in paragraph- length than many parts of Paley would yield. The paragraph- length would, nevertheless, have been reduced still further but for a few cases where a single sentence, broken by Paley into several paragraphs, was counted as a single compound one. Paley is the most prominent instance among modern writers of a man who paragraphed on the theory of emphasis. His mechanical devices for securing prominence were numerous — different kinds of type, numerals, etc. But the man that takes up only mechanical means for securing emphasis, usually perishes by the same means : he loses in proportion what he gains in emphasis. Paley is a shining illustration of this fact. Minto, bv the way, who has written about Paley's method of analysis, does not, I believe, note all of his niechanical devices. Paley used double spacing to separate groups of paragraphs. Thus B. 1., chap. 7, B. ii., chaps. 4, 7, 12. Another device is the very short chapter, as B. i., chap, i, which has three paragraphs, three sentences, 76 words. Paley's coherence depends upon conjunctions more than one '- Manual, p. 4554. 126 ///STOA'Y OF THE KXGI.ISH PARAGRAPH. would expect from so i^reat a lou;ician. The construction of his sentences and the order of words helps his coherence little or nothing. SCOIT. Ivanhoe, 1820. Total paragraphs considered 551 Total words considered 39,340 Total sentences ct)nsidered 1224 Average words per paragraph 71-39 Average sentences per paragraph 2.22 Average words per sentence 32.14 Per cent, of single-sentence paragraphs 45 Limits of fluctuation in paragraph word-length 3~338 Per cent, of sentences of less than 15 worck 14 Hazlitt was not far wrong when, in criticising the early style of the author of Waverley, he said : "There is neither momentum nor elasticity in it ; I mean as to the score, or effect upon the ear."' That style gained in vigor as years went by, but, except in the most impassioned passages, the sentences continue to ramble to the last. Even the dialogue is not equal to checking the diffuse- ness. An average of 31 words to the sentence, with only 14 per cent, of sentences under 15 words, is no help to the popularity of a novelist. In Scott the paragraphing of conversation proceeds by the modern method uniformly. His narrative and descriptive paragraphs have a certain unity always, and at times reveal a very high degree of picturesque grouping. The general straightforward coherence of his para- graphs is not to be disputed. COLERIDGE. The Friend, 1809. Total paragraphs considered 100 Total words considered 29,241 Total sentences considered 777 Average words per paragraph 292.41 Average sentences per paragraph 7.77 Average words per sentence 37.6 'On the Prose Style of Poets, \ 2. TEMPLE TO DE QUINCEY. 127 Per cent, of single-sentence paragraphs 8 Limits of fluctuation in paragraph word-length 45^758 Per cent, of sentences of less than 15 words 17 Poetry, Drama, Shakespeare. (Gerwig, for 500 periods.) Average predications per sentence 3.33 Per cent, of simple sentences 19 Per cent, of clauses saved i i.io At the beginning of the present century the journalistic short sentence was becoming popular. It had not, however, crept into the work of the literary dictators, and it is a little sur- prising that Coleridge should attack with such severity as he did, in the third issue of the Friend, a form of sentence that was not influencing the great reviews. Jeffrey was writing a sen- tence of Elizabethan proportions ; De Ouincey's sentence could hardly be spoken of as having anything in common with the "fashionable Anglo-Gallican taste" that Coleridge hated and that De Quincey, on the unconscious principle of elective affinity, praised. How little real hold the very short sentence acquired maybe seen later — considerably later, to be sure — when in 1840 De Quincey was uttering his lament that "the too general tend- ency of our sentences is toward hyperbolical length."' At any rate, Coleridge resolved not to cater much to French models. In the third essav of the Friend he admits that he may have injured his own style bv solitarv, inarticulate medita- tion, and by over-admiration for the Jacobean prosaists : but he then turns to attack the short sentence. "It is true that these short and unconnected sentences are easily and instantly under- stood : but it is equally true that wanting all the cement of thought as well as of style, all the connections, and ( if you will forgive so trivial a metaphor) all the hooks-and-eyes of the memory, they are easily forgotten ; or rather, it is equally impos- sible that they should be remembered." The practical — or impractical — result of this philosophizing appears in the style of the Friend. Here is Brandl's com- ^ Essay on Style. 128 I/ISTOh'Y OF THE KXGLISH J'AA'AGA\l/'J/. nient. "He reveled also in abstract expressions, and built up tlio most involved periods in tlie attempt to forestall every variety of objection. The paragraphs are so perversely arranged that the point is difficult to find; and the arrangement of chapters lacks all order."' The perverseness of the paragraphs comes from an attempt — not a victorious attempt — to follow the intricate order of the thought as it occurred in the writer's mind : hence also the large percentage of imperfectly developed induc- tive paragraphs. Some qualification must be made of the statement that Cole- ridge's sentences are involved. There are splendid exceptions in quantities, where he actually succeeds in performing dii^cult evolu- tions without ambiguity or obscurity. Again, Coleridge is not without some command of the short sentence. Of 777 sentences, 17 per cent, average less than 15 words. He can, when he needs it as a foil to a long and difficult period, use the disintegrating sentence with an oral force and directness like Emerson's. He tends, indeed, to put his paragraph-topic in a short sentence, for emphasis. Coleridge is " sequacious," even when he rambles ; seer though he is, he omits no step; his style is not only redintegrating, but, at times, almost impartially so — as if narcotism had touched his selective faculty. He uses more "hooks-and-eyes" than any writer of his time, more, I presume, than any great English lit- terateiir of the century. Of 300 sentences in the Friend, 100 are formally connected — up to that day a higher proportion than that of any man after Walton. The list of formal connectives is as follows, the initial connectives being double the interior in number : Connective. Initial. Interior. For 12 Again 2 Therefore . . 11 But 26 In short i Then 5 ^ Life of Coleridge, trans. T.ady Eastlake, p. 300. TEMPLE TO DE QUINCEY. I 29 Connective. At least And And yet Now Too Indeed Thus Accordingly It is true Nor On the contrary . . , On the other hand . Hitherto Yet Consequently In other words . . . . Lastly However Add to So Moreover Likewise First (etc.) Further Initial. 5 I 3 I 2 Interior. 2 3 5 2 JEFFREY. Alison on Taste : revised form, Eiicyclopcedia Britannica, 1824. Total paragraphs considered 100 Total words considered 27,608 Total sentences considered 545 Average words per paragraph 276.08 Average sentences per paragraph 5.45 Average words per sentence 50.65 Per cent, of single-sentence paragraphs 3 Limits of fluctuation in paragraph word-length 54-665 Per cent, of sentences of less than 15 words 6 In spite of its undeniable verboseness, Jeffrey's style was con- sidered brilliant and sprightly. How such a verdict could be passed on a style whose average sentence is fifty words, with only 6 per cent, of very short sentences to vary the monotony, is I ^,o HISTORY OF I'HE ENGLISH PARAGRAPH. hard for a modern reader to see. The secret lies in the compar- ative absence of periodicity. Jeffrey's huge sentences are mere groups of clauses. Many clauses are oppositional ; these are often set off by dashes. Jeffrey went as far in the direction of aggre- gating loose clauses as Macaulay went in the direction of segre- gating them. Otherwise, in the case of these two men, one style is almost as modern as the other. Jeffrey's length of paragraph is not far from Macaulay's. As a structural unit Jeffrey's lacks emphasis, from neglect of the short period : Macaulay's lacks gradation of emphasis, from his neglect of the moderately long period. Jeffrey makes clauses out of periods ; Macaulay makes periods out of clauses. Jeffrey's usual paragraph order is loose. His subject is often delayed, however, by verbose introductions. He has no sense of the importance of the first sentence and the last. His coherence is good but not graceful. There is occasional abuse of coordinate conjunctions. LAMB. Essays of Elia , 1822. Total paragraphs considered 87 Total words considered 14)386 Total sentences considered 529 Average words per paragraph 165.35 Average sentences per paragraph 6.08 Average words per sentence 27. IQ Per cent, of single-sentence paragraphs 15 Limits of fluctuation in paragraph word-length 15-726 Per cent, of sentences of less than 15 words 41 The gentle Elia was in his own day the uncomplaining target of much windy criticism as to his mechanology. Lamb's sentence and Lamb's paragraph were short, and therefore a source of worry to De Ouincey, who complained that "the most felicitous passages always accomplished their circuit in a few sentences ;'" and again that Lamb had no proper sense of the epic; — that "the solemn planetary wheelings of the Paradise Lost were not to his taste. "^ Though this could hardly be denied, a few essays of Lamb ' Works, v., p. 234. ' Works, V., p. 236. TEMPLE TO DE QUINCE Y. 131 show that he really had some command of the long paragraph : such are, " T/ic Sanity of True Genius,'' and, " On the Genteel Style of Writing^ But still. Lamb is likely to digress when he attempts a long section. Indeed, he usually avoids the long section, preferring to digress by paragraphs, — and so charmingly that we would not have him do otherwise. The unity of the short paragraphs is usually a rhetorical unity. He sometimes uses the short section purely for emphasis, and in all cases he is shy of logical division. Indeed, Professor Hunt represents Taine as maintaining that "Lamb aimed to destroy the great aristocratical style as it sprang from methodical analyses and court conventions."' If this remark refers to the passage given below," from the Histoirc^ it is not quite exact. These words of Taine about the grand aristocratic style were written of the romantic school, and vvith reference to poetry. Lamb's name happens to stand near in the context, but it is Lamb the author of John IVoodvil, Lamb the devotee of the sixteenth century. ^English Prose and Prose Writers, p. 367. -Speaking of " I'ecole romantique anglaise," Taine says : lis avaient rompu violemment avec la tradition, et sautaient par-dessus toute la culture classique pour aller prendre leurs modeles dans la Renaissance et le moyen age. L'un d'eux, Charles Lamb, comme Sainte-Beuve, avait decouvert et restaure le seizi- eme siecle. Les dramatisles les plus incultes, Marlowe par example, leur paraissaient admirables, et ils allaient chercher dans les recueils de Percy et de Warton, dans les vieilles ballades nationales et dans les anciennes poesies •etrangeres, I'accent naif et primitif qui avait manque a la litterature classique, et dont la presence leur semblait la marque de la verite et de la beaute. Par- dessus toute reforme, ils travaillaient a briser le grand style aristocratique et oratoire, tel qu'il etait ne de I'analyse methodique et des convenances de cour. Ils se proposaient "d'adapter aux usages de la poesie le langage ordinaire de la conversation, tel qu'il est employe dans la moyenne et la basse classe," et de remplacer les phrases etudiees et la vocabulaire noble par les tons naturels et les mots jplebeiens. A la place de I'ancien moule, ils essayaient la stance, le sonnet, la ballade, le vers blanc, avec les rudesses et les cassures des poetes primitifs. lis reprenaient ou arrangeaient les metres et la diction du treizieme et du seizieme siecle. Charles Lamb ecrivait une tragedie d'archeologue qu'on cut pu croire contemporaine du regne d'Klisabeth, etc. ^ Htstoii'e cie la Litterature Anglaise, Paris, 1887. Tome quatrieme, p. 286. 132 HISTORY OF THE ENGLISH PARAGRAPH. For all this, il is entirely true that Lamb was not devoted to logical analysis in prose. Lamb's use of the short sentence was incomparably freer, and as Mr. Pater might have said, " blither," than that of any of his predecessors. In sentence-length, indeed, he exhibits all the variability of insanity. His sentences fretted De Quincey : " Lamb had no sense of the rhythmical in prose composition. Rhythmus, or pomp of cadence, or sonorous ascent of clauses, in the structure of sentences, were effects of art as much thrown away upon him as the voice of the charmer upon the deaf adder."' Some of Lamb's "sentences and periods" made the author of the English Mail CoacJi " shriek with anguish of recoil." Doubtless to De Quincey the most abhorrent of these " sentences and periods " were those of two or three words, verb to be supplied from the reader's store of predicates. One can imagine the Opium Eater thrown into hysteria by Lamb's way of setting forth the bachelors of the South Sea House: "Hence they formed a sort of Noah's ark. Odd fishes. A large monastery. Domestic retainers in a great house, kept more for show than use." For my own part, I confess to being, just at this minute, in the mood to like this indefensible sentence-making. How the device flashes the conceits upon us ! We catch the first delicious over- emphasis of discovery — the very conception and birth of quaint fancies in the mind of a humorous genius. In spite of now and then a long but harmless parenthesis. Lamb knew the value of the paragraph structure — knew it better than Coleridge did, or De Quincey. Hardly one of his shorter sections but is an artistic whole. The order is loose. The mass is often perfect — the topic striking the eye instantly, and the paragraph ending with words that deserve emphasis. What shall we say of his coherence ? Coleridge, speaking in 1833, doubtless thought of Elia as one of " those modern books in which, for the most part, the sentences in a page have the same connection with each other that marbles have in a bag; they ' Works, v., 235. TEMPLE TO DE QUINCEY. 133 touch without adhering.'" But where would be Lamb's charm if his sentences were a third longer, and thick with "hooks-and- eyes"? The fact is that Lamb's style, on any subject Lamb would have been willing to touch, would be easier to follow than Coleridge's, no matter how far afield the whimsical Elia might wander. For there are no long intervals between Lamb's propo- sitions, no involved restrictions of those propositions, no neces- sity of supplying anything except a few obvious verbs and the sense of a few freakish vocables. LANDOR. Imaginary Conversations {Sovereigns and Statesmen). Total paragraphs considered 200 Total words considered 17)697 Total sentences considered 696 Average words per paragraph 88.48 Average sentences per paragraph 3.48 Average words per sentence * 25.43 Per cent, of single-sentence paragraphs 34 Per cent, of sentences of less than 15 words 22 Any statistics drawn from the Conversations are of course modified by the dialogue form. This explains the large number of paragraphed sentences, and the brevity of the paragraph. Though the Conversations yield Landor's most brilliant style, we shall base what we have to say of the structure more upon the pieces of continuous prose than upon these dialogues, which are so good in dramatic ?J^os as sometimes to seem anything but characteristically Landorian. Landor is uneven in the matter of unity He can keep severely to one topic, but he often forgets. He will begin an important paragraph on, say, Laura's decreasing coldness towards Petrarca, and. after illustrating this point by a remarkably inapposite account of the lady being kissed at a ball by Charles of Luxem- burg, will proceed to tell you in the same paragraph of Petrarca's travels and visits in the following summer.- Generally, however, Landor's frequent digressions proceed by whole paragraphs. ' 7 able Talk, July 3, 1833. ^See Works, VIII., 438. 134 HISTORY OF THE ENGLISH PAKAGRAPII. In llie matter of proportion i.andor has very considerable merits, thout^h bv no means the highest. He pays little attention to {)rop()rti<)n bv l)ulk ; but he uses the semicolon and the period with great skill to secure right distribution of emphasis. Here, however, the principle of euphony often interferes. No author ever surpassed Landor in such tricks of melody as introducing at the end of a resounding period a very brief colon clause for cadence.' These skillful variations sonjetimes misplace the thought eiiiphasis. When, however, the two principles coincide in the application, the effect is perfect. The felicitous combination occurs oftener in the short than in the long paragraphs. In the longer ones we sometimes feel that the writer is caring nothing for precision — only for the infinite variety of prose modulation which he himself describes — that "amplification of harmonies, of which even the best and most varied poetry admits but few." Landor's style is intuitive and segregating ; the incoherence of it is its weakest point. Mr. Sidney Colvin somewhat, but not greatly, over-states the case when he says: "The best skeleton type of a Landorian sentence is that which we quoted some pages back on Lord Byron : ' I had avoided him ; I had slighted him ; he knew it ; he did not love me ; he could not.' No conjunctions, no transitions ; each statement made by itself, and their \^sic\ connection left to be discovered by the reader .... But whether to the sequence of propositions in an argument, or the sequence of incidents in a narrative, Landor's style is less adapted."^ Mr. Leslie Stephen^ speaks of Landor rounding off transitions grace- fully. I cannot quite make out what this means, unless it means transition in melody. The rest of the passage in Stephen forms a good comment on Landor's coherence, and not less directly on his unity: "He is so desirous to round off his transitions gracefully, that he obliterates the necessary indications of the main divisions of the subject. When criticising Milton or Dante, ' A friend reminds me, in this connection, of Swinburne's fondness for end- ing a stanza with a short line. '^Landor, English Men of Lellcj-s, p. 223. ^ Hours in a Library, 3d series, p. 245, London, 1879. TEMPLE TO DE QiTNCEY. 135 he can hardly keep his hand off the finest passages in his desire to pare away superfluities. Treating himself in the same fashion, he leaves none of those little signs, which, like the typographical hand prefixed to a notice, are extremely convenient, though strictly superfluous. It is doubtless unpleasant to have the hard framework of logical divisions showing too distinctly in an argu- ment, or to have a too elaborate statement of dates and places and external relations in a romance. But such aids to the memory may be removed too freely. The building may be injured by taking away tfie scaffolding." His coherence is often helped by parallel construction ; but, again, the movement is a little retarded by the perfect balance of the sentences, as we have seen in older authors. In his later reading of Landor, Mr. Lowell " began to be not quite sure whether the balance of his sentences, each so admirable by itself, did not grow wearisome in continuous reading, — whether it did not hamper his freedom of movement, as when a man poises a pole upon his chin."' The minor breaks in Landor's coherence are usually due, not to false logic, but to a habit of vague reference and allusion. Landor assumed a high degree of literary information and appreciation on the part of his audience. He felt himself to be writing for the few. The chosen guests who were to "sup late" at his feast would be willing, for the sake of the elect camaraderie, to dispense with overmuch table-service. IRVING. Sketch Book, 1820. Total paragraphs considered 129 Total words considered 14,220 Total sentences considered 532 Average words per paragraph ' 1 10.23 Average sentences per paragraph 4.12 Average words per sentence 26.73 Per cent, of single-sentence paragraphs 17 Irving is in his way a skillful paragrapher. No matter how ' Latest Literary Essays, p. 45 ; Works, Boston, 1892. 13^ HISl'OKY OF THE ENGLISH PARAGKAPIJ. great the license of his subject, he always gives an impression of unity. He follows the loose order almost exclusively, kee{)ing his statement of details closely within the limits prescribed by his opening sentence. His transitions are faultless, the number of connectives being greater, however, than the placing of words requires. About one-quarter of his sentences are shorter than 15 words, and nearly one-half (41 per cent.) are under 20 words. He adapts the short sentence to the smooth and graceful manner of Addison. He does not, indeed, ever succeed in flashing out a complex thought in a telling and emphatic way; but as a type of the urbane, leisurely, correct manner, he is exemplary. CHAPTER VIII. DE QUINCEY TO HOLMES. Although we have included in the preceding chapter several writers of the nineteenth century, all of these, with the possible exception of Landor, belong properly, in structure, to the eight- eenth. De Quincey's stands as a dividing style between the two periods. The new period differs from the old, not in kind but in degree. In the nineteenth century the paragraph is organized as in the eighteenth, but acquires greater concentration. The emphasis of the short sentence is more keenly felt and more effectually employed. The unity is more organic. The coher- ence depends less and less on formal connectives. The question of mass receives its first serious attention. DE QUINCEY. Opium Eater, 1822. Total paragraphs considered 89 Total words considered 31.634 Total sentences considered 815 Average words per paragraph 355-42 Average sentences per paragraph 9.16 Average words per sentence 38.81 ' Per cent, of single-sentence paragraphs 14 Per cent, of sentences of less than 15 words 13 Limits of fluctuation m paragraph word-length. .. . 13-1441 Average predications per sentence ( Gerwig, } 3.69 Per cent, of simple sentences for 500 (-14 Per cent, of clauses saved periods.) ) 5.49 When we ask ourselves whether De Quincey's paragraphs are units we find it necessary to limit the word unity more closely than usual. Classical unity, severe, selective, exclusive, he rarely shows. On the other hand his essays were preceded by the most ' Sherman finds 33-|-. My own count is from the second American edition, purporting to give the original text. 137 I3i> ///SVVA'V OF THE EXGI.ISH PAA',u;A'A/>J/. careful analvsis, and there is no doul)! that he considered each paragrajih with regard to unity. We niav sav of liis longer para- graphs that the best show unity in somewhat wide variety, while in all cases he returns consciously, from digressions within the para- graph, to the topic. As a rule his long and numerous digres- sions proceed bv whole paragraphs. In the matter of proportion he is deficient. He expands the unimportant at the expense of the important. His use of the short sentence is usually half-hearted. No author who writes but 14 per cent, of simple sentences can obtain the highest effects in paragraph-structure. De Quincey, for purposes of rhythm, will give you numerous terse clauses within the sentence, but he fails to distribute the emphasis of his paragraph justly by means of the terse period. There are some exceptions to this general dictum, however. In Levana and Our Ladies of Sorrow we have most effective emphasis- proportion ; nothing could be finer. De Quincey's coherence is notoriously good. Mr. Stephen puts the general verdict thus : '" He is careful to show you the minutest details of his argumentative mechanism. Each step in the process is elaborately and separately set forth ; you are not assumed to know anything, or to be capable of supplying any links for yourself; it shall not even be taken for granted without due notice that things which are equal to a third thing are equal to each otner ; and the consequence is, that few people venture to question processes which seem to be so plainly set forth, and to advance by such careful development." ' Few authors are so redintegrating. The criticism which De Quincey applies to a certain style, quoting a French expression from Archbishop Huet, is applicable to his own style ; he had that flux de bouche which "places the reader at the mercy of a man's tritest remembrances from his own school-boy reading."- Let me again quote Mr. Stephen, from the same page as before. " He is utterly incapable of concentration. He is, from the ^ Hours in a Library, p. 364, London, 1874. = Works, X., 236-237 (Edinburgh ed.). DE QUINCEY TO HOLMES. 139 very principles on which his style is constructed, the most diffuse of writers. Other men will pack half-a-dozen distinct proposi- tions into a sentence, and care little if they are somewhat crushed and distorted in the process. De Ouincey insists upon putting each of them separately, smoothing them out elaborately, till not a wrinkle disturbs their uniform surface, and then presenting each of them for our acceptance with a placid smile. His very credit- able desire for lucidity of expression makes him nervously anxious to avoid any complexity of thought. Each step of his argument, each shade of meaning, and each fact in his narrative, must have its own separate embodiment ; and every joint and connecting link must be carefully and accurately defined. The clearness is won at a heavy price." The means by which this unusual "sequaciousness" is secured are many. First, of course, De Quincey rarely states a truth in its intuitive form, or at any rate rarely without explaining that form afterwards. Thus he uses a large number of clauses to elaborate a given idea. Then he employs with great art the devices of sentence-structure that lend coherence. No author uses parallel structure more freely and subtilely, shifting the mode just before it becomes mannerism. He inverts sentences and clauses constantly — hardly any writer more. Besides having at command all these structural contrivances he is opulent in con- nectives. Of 300 sentences in the Opium Eater 75 are formally joined. The list is as follows : Connective. For , However True it is Accordingly Nay Therefore Hence And Thus But Or nitial. Interior. 6 4 24 I I I I . . 3 I 3 I 12 I I40 HISTORY OF THE ENGLISH PARAGRAPH. Connective. Vet Also On the eonlrary So Moreover Nevertheless . . . Now then Thereupon Indeed Ini ial. Interior. 5 Everyone knows that De Quincey had much to say about prose rhythm. His theory stands midway between a theory of rhythm in the period and a theory of rhythm in the paragraph as a whole. To his remarks about the cumulative effect of the rhythmi/s of succeeding clauses (quoted in our section on Lamb) may be added the following, in which the writer is thinking of melody, quite as much as of sequence in thought : " It is in the relation of sentences, in what Horace terms their '■junctiira,' that the true life of composition resides. The mode of their nexus — the way in which one sentence is made to arise out of another, and to prepare the opening for a third, — this is the great loom in which the textile process of the moving intellect reveals itself and prospers."' Again, speaking of Kant's elephantine period: "Parts so remote as the beginning and end of such a sentence can have no sensible relation to each other : not much as regards their logic, but none at all as regards their more sensuous qualities, — rhythmus, for instance, or the continuity of meta- phor." = De Quincey himself exemplified his own theories of melody. In the short paragraph of his impassioned prose he has some- thing that may be called an organic paragraph rhythm. Such a paragraph will begin with a short cadence or two, followed by a longer one, and will end in a reverberating roll of dactyls, cretics, tribrachs, anapaests, what not. Much more rarely it will begin with a long, swinging cadence, followed by a shorter and a shorter, till the whole movement comes down to a short stop as ■ Works, X., 258. - Works, X., 259, DE QUINCE Y TO HOLMES. 141 with a clash of cymbals. The first movement may be illustrated by the following paragraph, from the Vision of Sudden Death: "The moments were numbered; the strife was finished; the vision was closed. In the twinkling of an eye our flying horses had carried us to the termination of the umbrageous aisle ; at the right angles we wheeled into our former direction ; the turn of the road carried the scene out of my eyes in an instant, and swept it into my dreams forever." In the longer paragraphs — the best ones of the impassioned style — there is most dexterous variation of cadence, the altei - nation of long and short going on till the music merges in one long rolling surge, only to emerge at the end as in lapping waves. Such is the harmony in the description of Our Lady of Sighs. On the whole, however, there is no deliberate harmonic organization of the long paragraph as a paragraph. De Quincey's finest effects of melody, as indeed of his thought, are effects of suspense. He is never really rapid in mental move- ment, or at least not forcibly rapid ; but he delights in the evoca- tion of a vivid train of images (face to face with an impending conclusion) in a way to reproduce the lightning-like, multiform impressions of the mind when under excitement. Similarly his rhythm may be held back. Thus, in one of the last paragraphs of the Vision of Sudden Death, he gets a peculiar effect of sus- pense by ending thus, "But the lady — " and beginning the next paragraph with a repetition of the same words. In the second section of the Dream Fugue — the section ends in the midst of a sentence — the last sentence advances by soft monosyllables — on tiptoe, so to speak ; it stops with a comma, and the next section drops into the swinging rhvthm once more. Thus: — "and afterwards, but when I know not, nor how. Sweet funeral bells from some incalculable distance," etc. One other witness to De Quincey's rhythmical sense should be mentioned. He studiously avoids repeating the same number of sentences in succeeding paragraphs. Thus he has no succes- sive groups of three, or four, or five, or six sentences ; and there 142 n/STORV OF THE ENGLISH PARAGRAPH . is in the Opium Eahr only one case of a succession of (two) groups of seven and one case of a succession of (three) single- sentence paragraphs. MACAULAY. Essays : Milton, Machiavelli, Dryden, History.' Total paragraphs considered 325 Total words considered 67,158 Average words per paragraph 206 . 67 Average words per sentence c. 23 . 05 Average sentences per paragraph c. 8 .g6 Average predications per sentence (Gerwig, j 2.17 Per cent, of single-sentence paragraphs from ' i Per cent of simple sentences 5604 I 36 Per cent, of clauses saved periods.) / 5 . 06 History of England. 'total paragraphs 333^^ Total words 974>55o Average words per paragraph 291 .g6 Average sentences per paragraph 12.44 Average words per sentence , (Sherman) 23 . 43 Per cent, of single-sentence paragraphs 2 Per cent, of simple sentences (Gerwig, from / 34 Average predications, per sentence 40,000 periods.) \ 2.30 The popular impression that Macaulay is the best of para- graphers is probably not far from the truth. The great rhetori- cian bestowed unlimited pains upon his paragraphs, and no pre- ceding writer began to equal him in conscious appreciation of the importance of that structure.^ His unity is rhetorical, rather than logical ; but as such it is nearly always unimpeachable. The sections that contain real digressions are few indeed. In the matter of proportion by bulk he is nearly always admirable. He knows his principal point, and it is on this that he enlarges. His emphasis-proportion is consciously paragraphic. 'For the total number of words in the Essays (except History) and in the History of Englatid I am indebted to Professor Sherman. - Cf. Trevelyan's account of Macaulay's laboriousness. Life atid Letters of Lord Macaulay, London, 1886, p. 502. DE QUINCEY TO HOLMES. 143 He reveals very great variability in sentence-length,' and drives home his main topic and his main conclusion in simple sentences. When he masses clauses it is to relieve each of emj)hasis and show the unity of the group as amplifying some previous terse generalization. He shows such deliberate observance of this principle that he forms the first basis for the generalization made in a former chapter : in the best modern paragraphs the distance between periods is inversely as the emphasis of each included proposition. Nevertheless, in this matter of distribution of emphasis, Macaulay is not faultless. It has been the general verdict of critics that he not infrequently over-emphasizes ; that he magni- fies clauses into sentences. On the other hand, a writer so well able to give a reason for his faith as Professor Sherman, defends Macaulay's short sentence at a point where most critics would consider it least happy. Thus : "This impulse to analyze and energize, — to keep the author's meaning out of the reach of the reader save one notion at a time, leads Macaulay in his earlier compositions to go against the fashion of his day and fall foul of the semicolon as a help to thought. Hence such sentences as these are not infrequent: 'Like the former he was timid and pliable, artful and mean. But like the latter he had a country.' — ' Shallow is a fool. But his animal spirits supply, to a certain degree, the place of clever- ness.' — 'There are errors in these works. But they are errors which a writer, situated like Machiavelli, could scarcely avoid.' Professor Sherman adds in a footnote: "This method of punc- tuation is manifestly truer to the thought, and will perhaps pre- vail in time. We are naturally about as loath to give up the eighteenth-century punctuation as its natural spelling. As to 'Sherman (6^;//zw,f//)' Studies, I., 4., p. 34S) has noted Macaulav's fond- ness for groups of sentences of 17 words each. But Sherman also notes {Analytics, 284) that Macaulay"s commonest sentence-lengths are those of II, 13, 14, 15 ; and that in the essay on History the sentence of maximum frequency is 14 words {University Studies, I., 4., p. 360). Macaulay has, 011 the ullier hand, a good many sentences of more than 100 words. 144 H/S-JOKY OF THE ENGLISH PARAGRAPH. the excuse of subordinate conjunctions for making semicolon clauses, we can go back and learn something from old Homer. When a sentence is to follow as the explanation of the preceding statement, it is liis favorite practice to introduce it without a 'because' or 'since,' and thus allow the reader the satisfaction of perceiving the relation for himself. Still Homer does not slight conjunctions: he merelv avoids abusing them.'" For one, I do not see how the punctuation in these passages from Macaulav is manifestly truer to the thought than semi- colons would have been. I can hardly believe that in Macaulay's rapid antithetic thinking these contrasts could possibly have been segregated before pen touched paper. Only the habit of exaggerating contrasts for stimulifs to the reader's mind, could have permitted the dropping of the semicolon in a connection where the act throws a relatively unimportant clause into the same importance as the short topic sentence. The point about Homer must be admitted; but though Homer is fond of asyndeton for explanatory purposes, we are not sure that he could have borne it to hear one of his rhapsodes drop his voice wherever a conjunc- tion was omitted. Macaulay's coherence is dependent upon structural devices. The paragraph once accepted by the reader as a unit in the light of whose topic each sentence is to be read, Macaulay's style is indisputably sequent. True, there is no blending of colors in the picture: the sentences lie like stones in a mosaic, as Mr. Stephen puts it, or like marbles in a bag, as Coleridge would have put it. But there are no gaps in the mosaic, and though the pieces are distinct, they are numerous and rightly set. Parallel construction is almost the rule with Macaulay, and it is often mechanical and noticeable. Inversion is frequent. Connectives are few — fewer bv far than in anv man hitherto who has not been enslaved to the balanced sentence. The list runs thus, showing but 47 formally connected sentences in a total of 300. "'Some Observations on the Sentence-Length in English Prose," Univer- sitv Studies, Vol. I., No. 2, p. 126. DE QUINCEY TO HOLMES. 145 Connective. Initial, Interior. Nor 3 Yet 7 F'or 3 Therefore . . i But 15 It is true 2 Also . . I Thus I On the other hand i Too . . I However . . I At length i At last \ ^ ^ Indeed . . 7 His coherence is impaired at times by one of his methods of organization. Most of his paragraphs are loose ; but occasion- ally in the midst of one he will abruptly introduce an intuitive statement or a generalization, proceeding afterward to resolve this in redintegrating manner. Sometimes, indeed, the riddle is left unresolved : Mr. Leslie Stephen's sensibilities were much jarred by Macaulay's abrupt and unexplained contrast, to the effect that Boswell was the greatest of fools and the best of biog- raphers. This habit of introducing an enigma and then resolving it step by step gives us a type of paragraph that is pseudo-deduc- tive yet really periodic. It is a common type in the Essays. In the History we find a comparatively large number of truly periodic structures, where the writer begins his paragraph remotely and proceeds by the natural order of development to a new conclusion. Whether the order is deductive or inductive, it often happens that the very first sentence is a summary of the preceding paragraph, the transition being greatly expanded.' Macaulay had a very definite sense of paragraph rhythm, though his movement is too much staccato. He has also a keen sense of the importance of variety in paragraph-length. Here he ' On this point and that of the abrupt introduction of a general statement, see Minto's admirable analysis of Macaulay's style, Manual, p. 89 ff. I4^> HISTORY OF THE ENGLISH PARAGRAPH. is perhaps the most intelligently variable of all our prosaists. He knows how to relieve the attention b\- variety, and to drive home in a short paragraph the details accumulated in a preceding long one. His percentage of {paragraphed sentences is low, l)ut he does not hesitate to use this device to mark a brief but emphatic stadium. The question of constancy in paragraph-length has alreadv been discussed (pp. 49, 50) with reference to Macaulay, the author who in stylistic averages is perhaps the most stable of all who have written English prose. CARLVLE. Jean Paul Ricliter, 1827. Total paragraphs considered 34 Total words considered 8521 Total sentences considered 270 Average words per paragraph 250.62 Average sentences per paragraph 7.94 Average words per sentence 3 1 '56 Per cent, of single-sentence paragraphs 5 Essays. (Gerwig, for 500 periods.) Per cent, of simple sentences 18 Average predications per sentence 3.12 Per cent, of clauses saved 7.08 Sartor Re sarins, 183 3- 1834. Total paragraphs considered 100 Total words considered 16,690 Total sentences considered 476 Average words per paragraph 166.90 Average sentences per paragraph 4.76 Average words per sentence 35-o6 Per cent, of single-sentence paragraphs 12 Limits of fluctuation in paragraph word-length. . . . 27-488 French Revolution, 1837. Total paragraphs considered 100 Total words considered 16,031 Total sentences considered 671 Average words per paragraph 1 60.3 1 DE QUINCEY TO HOLMES. 147 Average sentences per paragraph 6.71 Average words per sentence 23.89 Percent, of single-sentence paragraphs 3 Per cent, of sentences of less than 1 5 words 28 Limits of fluctuation in paragraph word-length 24-374 \ Carlyle's most orderly paragraphs belong to the period of his J life when Goethe's influence over him was freshest and strongest. \ For order in the paragraph is due largely to an ascendency of !> the intellectual element over the emotional ; and Carlyle's emo- tions were never so well-tempered — or least ill-tempered — as when he saw most clearlv the mastery that Goethe had of his own nature.' Thus the Life of Schiller is sequent and orderly in a degree surprising to the reader who has of late fed on the French Revolution, i In this early time Carlyle saw life steadily and achromatically. Ijut as his egotism waxed strong with his days, as his impatience of the world increased and his hopes of reforming it decreased, he became subject to starts of. the wildest I incoherence.! In such papers as the Latter Day Pamphlets he is wholly under the influence of his habitually strongest emotions ; he raves. As Minto says, "Some pages remind us of his vivid descriptions of chaotic inundations, that hide or sweep away all guiding posts. Very seldomcan we gather from the beginning of a paragrap h what is to b e its pur]X)rt. No attempt is ijiade to keep a main subject prominent."" MirTfo finds that in Carlyle's writing of history, the case is verv different. "The arrangement is almost the perfection of clearness. When the bearing of a statement is not apparent, he is careful to make it explicit. In each paragraph the main subject is for the most part kept prominent, — his defiance of ordinary syntax giving him great facilities for a distinct foreground and background. He begins his paragraphs with some indication of their contents. Further, he is consecutive, and keeps rigidly to ' Somewhere, (I think in a letter) Carlyle likens Goethe's emotions in their number and variety to the hues of the landscape, but his intellect to the sun that irradiates and controls them all. '^ A/anual, p. 152. mS history of the en gush paragraph. the })oint." It may be said that this is high praise, and that in the case of ilic Rcvohitiou nearly every point that Minto makes should be slightly modified. The sentence-length of the early essays is moderately long- in Richtcr, 31.56. Between 1827 and 1833 Carlyle was develop- ing his own peculiar ideas of emphasis ; and the study of German models increased his sentence, which appears in Sartor as 35.06. The sentences of Sartor are full of parentheses and involutions. Of Teufelsdrock's periods the writer himself said, " Perhaps not more than nine-tenths stand straight on their legs ; the remainder are in quite angular attitudes ; a few even sprawl out helplessly on all sides, quite broken-backed and dismembered." Between 1834 and I S3 7 Carlyle came under a new influence, the French. Though his style in the Revolution cannot be called in any sense Gallic, he had at least profited by his studies ; the sentence of the Revolution is a third shorter than that of Sartor ; to be exact, it stands at 23.89. I regret that I have no figures from the Fried- rich. Sherman, as we have noted in § 4 of Chapter 3, says that " Carlyle showed no change for worse or better, in respect to sentence proportions, between the Edinburgh Essays and his Frederick the Great.'' In this case the average of Carlyle's sen- tence has again risen under study of German models ; but the sentence of the Frederick is surely a different and far better sen- tence in point of carrying power than the somewhat Johnsonian sentence of the Essays. The word-length of Carlyle's paragraph follows just the course that might be expected. In 1827 it is 250.62. In Sartor the long period becomes temporarily as prominent a unit as the paragraph, and the latter sinks to 166.90. The ensuing study of French reduces the sentence but leaves the paragraph about in statu quo (160.31). It should be added that the increase in the impassioned quality of the prose would be another reason why the early length of the paragraph would decrease. Impassioned prose cultivates short units; De Quincey's new "impassioned prose," with its long sentence and paragraph, was merely imagi- DE QUINCEY TO HOLMES. 149 native prose. The course of the single-sentence paragraph corresponds roughly with the movement of the sentence-length, increasing from 5 per cent, in Richter to 12 per cent, in Sartor, and in the Revolution dropping to 3 per cent. Minto's general remarks on the structure of Carlyle's! sentence are just — as that the sentence is an exaggeration of the loose style, — " consisting, for the most part, of two or three coordinate statements, eked out by e.xplanatory clauses either in apposition or in the 'nominative absolute' construction." But it is a most striking fact that, by the use of these devices and an enormous number of significant phrases and words, Carlyle's later style is perhaps the weightiest in the language. The amount of sup- pressed intermediate predication is unprecedented ; and when we take into account the subjects that Carlyle treated, the number of facts he was bound as an historian to express, all other intuitive styles, it seems to me, will appear in comparison with his, diffuse. It is a curious fact that Carlyle's coherence seems at first blush to depend as much on connectives as De Ouincey's. The fact is, however, that Carlyle in his later works conveys several times as many notions to the sentence as De Quincey does," and saves clauses in ways that De Quincey never dreamed of — no, not in his wildest opium dream after an evening with the "sentences and periods" of poor Lamb. Carlyle's connectives, again, are far more vital than De Ouincey's, and sometimes represent relations that De Quincey might have spun into clauses, and that Macaulay surely would so have treated. The list is from the Revolution, showing 75 formally connected sentences in 300. Connectives. Initial. Interior. So 4 2 But 10 .. • Indeed . . 5 For II Thus 3 I However 2 2 ' I wish Mr. Gerwitj had triven us the [)er cent, of clauses saved in the Revolution. The per cent, in the Essays is only 7.0S, while De Ouincey's is 5^49. 15° ///STOA'V OF '////■: ENGLISH PARAGRAPH. Connective. Initial. Interior. And 5 Likewise i Too . . 4 Or I Nevertheless i 2 Moreover . . i Neither 2 .Viui yet 2 Nay 5 Also . . 3 Accordiniclv I I 111 like manner — . . I Lastly I Then i 3 On the whole i At least : . . I Again . . i Whereupon i Carlyle has on the whole a wide variety of means for articula- tion, notably that of massing significant words at the beginning and end of sentences. He seldom repeats a word for coherence, as Macaulay and Arnold and a host of others do; by ordering his words he makes repetition unnecessary. In his historical writing Carlyle is a great master of the law of proportion, as concerns both the paragraph and the whole composition. He combines Hume's power of making a para- graph illustrate a given philosophical idea, and Macaulay's power of heightening that impression by pictorial means. He moulds his material, fuses his facts, emphasizes the salient, subordinates the unimportant. In elaborating large plans, he constantly reduces his macrocosm to microcosm to be sure of making his point; he reiterates his central truth ; he does not disdain numer- ous formal but living summaries. In the matter of distribution of emphasis by varying sen- tence-length he improved steadily. His earliest work shows about the same percentage of simple sentences as De Ouin- cey's. The Revolution, on the other hand, shows nearly 28 per cent, of sentences under fifteen words, with an unusual DE QUI NCR Y TO HOLMES. 151 tendency toward sentences of less than 8 words. With a per- centage of short sentences no greater than Burke's, Carlyle man- ages to distribute his emphasis with masterful effect. NEWMAN. Idea of a University, 1S54. Total paragraphs considered 200 Total words considered 50,896 Total sentences considered 1228 Average words per paragraph 254.48 Average sentences per paragraph 6.14 Average words per sentence 41-44 Per cent, of single-sentence paragraphs 6 Apologia. (Gerwig, for 500 periods.) Per cent, of simple sentences 16 Average predications per sentence 2.97 Per cent, of clauses saved 4. 50 Newm an's para graphs ar e the result of the most careful analysis on the part of their writer. In them unity, usually philosophical, often complex, is severely observed. The style is highly redintegrating, in spite of the aggrega- ting sentence and bookish vocabulary. But it can never be called impartially redintegrating, as one is sometimes tempted to call De Quincey's. The most careful selection of thought is made, and whatever subsidiary matter may have been generated in the act of composition is sternly repressed in the writing. In this matter we may compare Newman and De Quincey — both artistic minds. Both men are interested in the various phases of the material thev use for any given purpose, though of course Newman less than De Quincey in the sensuous qualities. But De Quincey cannot express one phase of his interest at a time ; Newman can. We find Newman not indeed depending upon connectives for coherence, but using them freely for increased accuracy. Thus Sherman found 131 initial connectives in 500 sentences' ' Analytics, p. 304. 152 HISTORY OF THE ENGLISH PARAGRAPH. — a proportion higher than Coleridge's, indeed perhaps the highest of our time. Newman's proportion by bull< is all that could be desired. His distribution of emphasis by sentence- length is faulty ; but it must be remembered that he is appeal- ing to the intellect rather than the emotions. EMERSON. Divinity School Address -\- Ameriian Scholar -\- Self-Reliancc. Total paragraphs considered 122 Total words considered 24,267 Total sentences considered 1,1 79 Average words per paragraph 198. 91 Average sentences per paragraph 9.66 Average words per sentence 20.58 Per cent, of single-sentence paragraphs 3 Average predications per sentence (Gerwig, \ 2.23 Per cent, of simple sentences for 1438 )■ 41 Per cent, of clauses saved periods.) ; 3.01 English Traits. Total paragraphs considered 200 Average sentences per paragraph 6.74 If we hold ourselves, in a definition of unity, to meaning by the word oneness of subject, we may admit Emerson's paragraphs to have unity. More than half the time, at least, every sentence bears on the point concerned. Sequence in the analytic (/. e. redintegrating) sense he had none. There is no tracking him. You are conscious that he has arrived, and from a place worth coming from, for his hands are full of gems ; but no other man can find out his way, nor can he. He was always complaining that he had no system ; speaks of his own "impassable paragraphs, each sentence an infinitely repellent particle." He has little close ordering of words for coherence, few inversions, few parallelisms of structure. Out of a desperate desire to indicate relations, he uses 49 sentence-connectives to 300 periods ; but not always do they catch and hold the true rela- tion.- Here is the list : DE QUINCEY 10 HOLMES. 153 Connective. Thus far But Thus Indeed And Finally Too Yet In fine Or Hence On the other part . Then So Therefore For First However Initial I 14 I Interior. How then, without sequence, does our author make himself clear? His statements are intuitive ; but we shall find that he has a curious alternating method of intuitive statement which amounts to resolution of the main idea. The paragraph con- tains a half-dozen intuitive sentences, each stating the main idea from a different point of view ; so that perforce some of the steps omitted in one statement are supplied in another, if only by the great variety of associations. Emerson must state the point intuitively ; but he does so under so many metaphors that he is sure somewhere to hit your experiences, your quickest road to apprehension. What of his proportion ? There is little of it, whether by bulk or by sentence-variation. He has 41 per cent, of simple sentences, and something is sure to be over-emphasized. But in the intuitive manner the lack of proportion is not so keenly felt as elsewhere. CHANNING. Self- Culture, 1838. Total paragraphs considered 60 Total words (.Siierman) 19,009 154 lIIsrORV OF THI: EXGIJSH PARAGRAPH. Total sentences (Sliennaii ) 750 Average words per paragrapli 316.81 Average sentences per paragrai>h 12.50 Average words |)er sentence 25.^5 Per cent, of single-sentence paragraphs Lenox, Napoleon, Milton. Average jjredications per sentence (Gerwig, 1 2.47 Per cent, of simple sentences 2000 \ 34 Per cent, of clauses saved periods.) f 6.55 Sherman has noted that Channing began the use of the short sentence at about the same time as Macaulay, and in nearly as great proportion. But to u]y mind Channing's emphasis-propor- tion in the paragraph is more rational, though less brilliant, than Macaulay's. Channing knew the worth of the semicolon ; Macau- lay did not. On the other hand Channing's paragraphs are too long to be well massed. Nor is the right bulk always assigned to the main ideas. We can find little fault with Channing's unity, and little with his coherence. The latter quality depends largely upon logical redintegration and upon the ordering of words. Connectives are used but sparingly. BARTOL. Radicalism and Genius: Fa titer Taylor. Total paragraphs (.Sherman) 45 Total words (Sherman) 13,385 Total sentences (Sherman) 805 Average words per paragraph 297.44 Average words per sentence 16.63 Average sentences per paragraph 17-89 Per cent, of single-sentence paragraphs Radical Problems. t Average predications per sentence (Gerwig, 1 2.10 Per cent, of simple sentences 1 500 \ 44 Per cent, of clauses saved periods.) ) 7.60 Radicalism. Words jser paragraph . 231 .64 Genius: Father Taylor. Words per paragraph 360.38 DE QUINCEY TO HOLMES. 155 I have included Dr. Bartol because he is one of the extreme examples, among reputable writers, of the frequent use of simple sentences.' His percentage of simple sentences is indeed so high in proportion to the whole number in the paragraph, that I can hardly admit that there is any right distribution of emphasis. Nor is there any proportion by bulk : the writer is as likely to pour out six sentences on an unimportant point as six on an important one. Nor have his paragraphs any necessary unity. Many are manifestly heterogeneous ; some indeed seem merely mechanical. Nor, again, can we praise the general coherence of Bartol's style. Granted that now and then, when he is driv- ing home a series of coordinate statements bearing on one sub- ject, he runs smoothly along ; at other times he proceeds by leaps and in no particular direction, like a boy from tuft to tuft in a marsh, — forever jumping, but never arriving. LINCOLN. Letter, 1863, published Century Magazine, May, 1889. Total paragraphs 12 Total words 1 659 Total sentences 91 Average words per paragraph 138.25 Average sentences per paragraph 7-6o Average words per sentence 18.23 I consider a passage from Abraham Lincoln merely to show the proper use of the very short sentence. The letter is quoted and praised by Earle, and it forms a good contrast, in point of method, to the work of the last author considered. The sentence is a little longer than Bartol's ; but the para- graph is 138 words as against Bartol's 297. Each of Lincoln's paragraphs is an organism. Each is knit together by perfect logical sequence, perfect unity. There is no modulation of emphasis, for by the nature of the subject there can be none. The letter is a challenge. Each sentence is meant to go home 'The highest average given by Mr. Gerwig is 58 per cent, of simple sen- tences — in Mr. J. A Symonds's Greek Poets. It is most extraordinary that Svmonds should also show 10 per cent, of clauses saved. 15^' HISTORY OF THE ENGLISH PARAGRAPH. like a shot. Tlit.' whole appeal is to the will, and in cases of this sort it may be of the very essence of style to eschew the fine shades of meaning that should exist in an intellectual type of discourse. DICKENS. Oli^ Curiosity Shop. Total paragraphs considered 300 Total words considered 1 5,202 Total sentences considered 639 Average words per paragraph 50.67 Average sentences per paragraph 2.13 Average words per sentence 23.78 Per cent, of single-sentence paragraphs 43 Dickens has more than once been criticised for lack of powers of construction and arrangement. Such criticisms apply often to his large plans ; but they are not just to his powers of analysis within the chapter. The unity of his narrative and descriptive paragraphs is organic and highly picturesque. There are slips at times, but again, there are whole chapters of the most subtle par- agraph-unity — of a kind that none but the great novelists can secure, a kind that no essayist dreams of. His coherence is the coherence of oral style. There are very few connectives ; their place is taken by explanatory clauses and sentences. Occasionally we feel that the style is diffuse, but obscure never — some bad grammar notwithstanding. Next to his coherence the best paragraphic quality of Dickens is his emphasis. This arises largely from his skillful ordering of words and a keen eye for the point where he should stop his sen- tence. He rambles when rambling is in order ; but no man can make a shorter cut. The extent to which he uses the short sen- tence is not excessive for a novelist : in the Old Curiosity Shop, with all the conversation included, the percentage of sentences of less than 15 words is 40 per cent. The melody of Dickens's prose is equable and flowing, with a tendency to metre now and then. He has no right feeling for the paragraph as a rhythmic whole. DE QUINCE Y TO HOLMES. 157 GEORGE ELIOT. Daniel Dcronda. Total paragraphs considered 212 Total words considered 16,233 Total sentences considered 725 Average words per paragraph ; 76-57 Average sentences per paragraph 3-42 Average words per sentence 22.39 Per cent, of single-sentence paragraphs 27 In its averages George Eliot's style approaches that of Dickens, except that the less elaborate philosophizing of the latter keeps the word-average of his paragraph down. But the sentence of the two writers is nearly the same, and George Eliot's percentage of sentences of less than 15 words is the same, within 3 per cent., as Dickens's. Of the two wTiters the balance in the matter of the short sentence is in favor of the woman, who has 43 per cent. Evidently there is here quite as much variability in the female style as in the masculine." It should be noted, how- ever, that George Eliot's short sentences tend to occur together ; the same is true of her long sentences. In the dialogue the sen- tence is short ; in the narrative it is long. We may say that George Eliot's paragraphs have unity, barring an occasional philosophical digression. We may say that they show logical coherence, excepting now and then one where a remote conclusion is introduced before it is analyzed. KINGSLEY. Alton Locke. Total paragraphs considered 200 Average words per paragraph 79-I9 Average sentences per paragraph . 3.34 Average words per sentence 23.72 Per cent, of single-sentence paragraphs 39 It is curious that the sentences of Kingsley and, Dickens should differ but a small fraction of a word and that George ' In view of Mr. Havelock Ellis's recent thesis that greater variability in mental power is shown by the male sex than bv the female, it would be an interesting study to investigate the comparative variability of masculine and feminine styles_ 158 IirSTORY OF THE ENGIJSn PARAGRAPH. Eliot's should var\- hut a single word from these two. It will be ronienihercd thai likewise Macaulav coincides, within a word, with these writers in sentence length. Again, Kingsley and George Eliot differ but tliree words in paragraph -length. Evidently the style of popular narrative and description finds 23 a favorite sen- tence ; ' while the same style when broken by conversation tends today to a paragraph of more than 50 and less than 100 words. 1 say today : but in the immediate present many good popular narrative styles are falling below the 23-mark. LOWELL. Carlylc. Total paragraphs 25 Total words 1 1,196 Total sentences 356 Average words per paragraph 447-84 Average sentences per paragraj)h M —4 Average sentence-length 31-45 Per cent, of single-sentence paragraphs 4 Dante. Total paragraphs considered 50 Average worcte per paragraph 668.30 Lessing. Total sentences 683 Average predications per sentence (Gerwig, ) 2.54 Per cent, of simple sentences for 683 r 23 Per cent, of clauses saved periods.) ) 4.76 When we come to read Lowell's noble essay on Dante we are tempted to acknowledge in his paragraphs a certain colossal unity ; at a little distance from the charm of the style we dare to speak of that unity as prolix ; later, we begin to wonder whether there is any unity at all in a paragraph of, say, 2183 words. It is hard to make out Lowell's theory of the paragraph. Apparently he had a ijiost elastic idea of the elasticity of that unit, and felt ' Why this is so remains to be determined. Indeed, the whole rjuestion of literary sentence-length must soon be minutely discussed from the point of view of the psychologist and the phvsiologist, as well as from that of the rhetorician. DE QUINCEY TO HOLMES. 159 that if he looked to a proper alternation of emphasis b}' sentence- variation and kei)t iiji a i,^eneral flow of coherence, his paragraphic duty was done. At any rate, it is easy to praise his emphasis, varied by 23 per cent, of simple sentences and by skillful inversions that put the main idea first. And we may praise his coherence, depending as it does upon closeness of logical relation, and eschewing formal connectives. Sherman found but 59 initial conjunctions in 500 periods. For all our author's general orderliness, however, the reader must be well equipped to get the pith of Lowell's finer prose. His words are meaning-crammed, and there is no pains taken to elaborate in short oral sentences that which a college-bred man should remember or understand. Once more, we must admit that Lowell loves a digression, and will take it when the material he handles is suggestive ; he carries us with him, to be sure, but we feel that the principle of logical sequence is for the time set aside for mere association by contiguity. RUSKIN. Sesame and Lilies. Total paragraphs ISI Total words 27,120 Total sentences 814 Average words per paragraph 179.60 Average sentences per paragraph 5-39 Average words per sentence Zs-i'^ Per cent, of single-sentence paragraphs 13 Average predications per sentence (Gerwig, i 3.50 Per cent, of simple sentences for 718 '- 18 Per cent, of clauses saved periods.) \ 6.63 Ruskin early began to boast of his analytic powers, and not without reason. His works are divided and subdivided with great elaboration, the later ones more intelligently but less elabor- ately than the earlier. He usually employs the words paragraph and section synon3-mously, preferring, however, the former term- The section-mark § he often places before divisions that he calls paragraphs. He is fond of compound paragraphs, numbering l6o ///SJVA'Y OF THE ENGLISH PARAGRAPH. the main paragraj))! and indicating Ijy indentation the subdivi- sions. In liis first edition of collected works he divided the text into "paragraphs," numbering these consecutively through the volumes. The paragraphs, even of the Modern Painters, are almost never heterogeneous, although Ruskin's later changes in these early works result in breaking up a few of the sections. In the Modern Painters the sections are longer than in the Sesame a)id Lilies and later works. The sequence of Ruskin's early work is marred by dislocations rather than by digressions. Many paragraphs in the Modern Painters would be bettered much by mere re-arrangement of the sentences or groups of sentences. In the comments made in the Brantwood edition (1891), on his early works, Ruskin appreciates the bad arrangement of some of the paragraphs, and even goes so far as to declare the "terrible confusion" of others. For his coherence Ruskin relies in his earlier works much on connect- ives, but in his later works less and less. He was never afraid of and, however, and does not hesitate to begin a sentence or a paragraph with a coordinate conjunction. I doubt if any other writer uses conjunctions less conventionally and more effectively. Other means of coherence Ruskin employs with very great variety and freedom from mannerism : notably parallel structure, veiled beneath changing phrases of introduction. Of emphasis-distribution the paragraphs of the Modern Paint- ers show but little. Ruskin had an early 'notion of returning as far as he could to what he thought the better style of old English literature, especially to that of his then favorite, in prose, Richard Hooker.' ' Such a notion was hardly favorable to the develop- ment of proportion in the paragraph. I have no count for the Modern Painters, but dare estimate that the percentage of simple sentences is less than 15 per cent. In Sesame and Lilies, indeed, it is but 18 per cent. Some of the sentences of the Modern Painters, particularly in the second volume, were inexcusably ' Preface lo Sesame and Lilies, collected works, 1 871. DE QUINCEY TO HOLMES. i6i long, and destructive to proportion. Thus Ruskin, comment- ing on the original sentence in which he enunciated the chief types of unity in art, says : " Yes, I should rather think so [that the types should be considered separately]; and they ought to have been named separately, too, and very slowly ; and not upset in a heap on the floor, as they are in this terrific two-pages entence."' In another place there is lack of proportion caused by the non- chalant introduction of an important theory as a subordinate part of a sentence. The fact does not escape the reviser's eye ; he says : " This rather astounding paragraph was anciently parted from the preceding text only by a semicolon. I have fenced it, at least, with two full stops; for it is in fact the radical theorem not only of this book, but of all my writings on art.'"" The same critical and artistic discrimination that made these com- ments possible, largely removes the necessity of any such com- ments hereafter upon Ruskin's later books. In these, the units of presentation — both sentence and paragraph — are XioX. long, are not confused, are not lacking in emphasis. At his best he is one of the very best paragraph writers of this or any day. No author would better repay a minute investigation. He has not been surpassed in the art of concentrating " victoriously intricate " periods in artistic wholes ; or, to speak more accurately, of amplifying a given topic in a paragraph whose interior arrange- ment reveals the most complex proportion. HERBERT SPENCER. Philosophy of Style. Total paragraphs 68 Total words 11,983 Total sentences 404 Average words per paragraph 176.22 Average sentences per paragraph 5-94 Average words per sentence 29.66 Per cent, of single-sentence paragraphs 4 Per cent, of sentences of less than 15 words 17 • Brantwood ed., II., 129. ^Brantwood ed., II., 49. I 62 HISrOKY OF THE ENGLISH PARAGRAPH. Spencer's averages are interesting as belonging to a scientific manner, — the manner, moreover, of tlie author to whom is due the theory that economy of attention is the governing principle of stvle. We find the discourse carefully analyzed into short paragraphs. These are mostly loose in structure,' a definite con- clusion being offered in the first sentence and defended in those following. Evidently Spencer's theory of periodic structure as the more economical, stops short of the paragraph. It is interesting, again, to note that, while Spencer's sentences rather favor the periodic type, they are not long ; like the short paragraphs, they are for the untechnical reader, if not for the popular one. The variability in sentenceTength is quite as great as could be expected from a style appealing so little to the emotions: the percentage of sentences of less than 15 words is I 7 per cent. The coherence and sequence of Mr. Spencer's prose are philosophical and correct. The use of connectives is less than might be supposed. Of the connectives that he does employ Mr. Bain- pointed out as characteristic the phrases, Yet another, Onee more, for adding to a cumulation already very much extended. MATTHEW ARNOLD. The Fuiietioii of Criticism at the Present Time. Total paragraphs 34 ^ Total words io,939 Total sentences 324 Average words per paragraph 321-73 Average sentences per paragraph 9-55 Average words per sentence 33-76 'Of Spencer's use of the compound type we have spoken, chap. II., § 3. ' R/ieioric, § 161. 3 In this case, as hitherto, quotations are considered as belonging to the paragraph in which thev are introduced, and not as separate paragraphs, even when indented. This, of course, only when they are introduced as an integral part of the paragraph. Arnold usually indicates such a relation by preceding he quotation with the colon and dash (: — ). DE QUINCE Y TO HOLMES. 163 Literary Influence of Academies. Total paragraphs 37 Total words 9>883 Total sentences 28 1 Average words per paragraph 267.10 Average sentences per paragraph 7.5S Average words per sentence 35'I7 Function of Criticism-\- Literary Influence. Total paragraphs 71 Total words 20,822 Total sentences.- 605 Average words per paragraph 293.26 Average sentences per paragraph 8.52 Average words per sentence 34-41 Per cent, of single-sentence paragraphs 5 Culture and Anarciiy. Total paragraphs considered 100 Average sentences per paragraph . 6.68 Per cent, of single-sentence paragraphs 4 Essays, 500 periods. (Gerwig, for 500 periods.) Average predications per sentence 2.77 Per cent, of simple sentences 20 Per cent, of clauses saved 4.51 It is a pleasant task to re-read the Essays in Criticism to see whether the measure and proportion that x\rnold found his chief delight in praising extends in his own prose to the organization of the paragraph. The result of our reading is, on the whole, satisfactory. Arnold'^ paragraphs, while thev have not the very highest varietv in unitv, do have admirable measure and propor- tion. The paragraph is usuallv loose, with an introductory sentence of transition. A large proportion are deductive : Arnold loved to regard the paragraph as a means of illustrating a general rule — he was not particular to advance a large body of particulars and base an induction upon these. We may quote on this point his own words about another matter : " Here, as evervwhere else, 1 64 HISTORY OF THE ENGLISH PARAGRAPH. the rule, llie idea, if true, couuuends itself to the judicious, and then the examples make it clearer still to them. This is the real use of examples, and this alone is the purpose which I have meant mine to serve." ' The coherence of Arnold's paragraphs is well-nigh perfect in its way. It arises primarily from an oral structure — a close logical method, redintegrating in idea, slightly aggregating in sentence. It is true that Sherman found 137 formally connected sentences in a total of 500 ; though some of Arnold's initial con- nectives are deliberately superfluous, used to give conversational tone — I refer to such words as " well," "now." But the fact is that Arnold uses not only a goodly number of conjunctions, but also a very great variety of transitional phrases and clauses. He is always aiming at the relations of things : he would rather paint no picture at all than one without the significant half-tones, the shadings that by their cool gradations make apparent the truth of the landscape. He will not even trust you to remember, under the mere stimulus of a pronominal word, exactly what a given substantive meant; he must explicitly repeat the substantive. Then another phase of his orderly, redintegrating method should be mentioned : I remember no other English prosaist who has so mastered the art of placing words in a way to secure sentence emphasis without hurting either the just order of the thought, the just proportion of the thought, or the just idiom of the language. To be sure, he is often reduced to the device of gentle exclamation ; but with what accuracy he puts the important words first and last in the sentence ! yet with how few breaks between propositions, how little exaggeration of the inconsequential, how little violence of normal English structure ! He is not, however, quite successful in so arranging the parts of the paragraph that the chief things shall be seen first. One other method of coher- ence Arnold affects, that of parallel construction. Few writers use it more extensively. Others, as De Quincey, keep the reader less aware of its presence ; still others, as Macaulay, ' Literary Influence of Academies, p. 77. DE QUINCEY TO HOLMES. 165 thrust it more prominently before the reader's eye. Arnold usually exhibits with it his habit of repeating words for explicit- ness of reference. The Hellenists will have it that the finest measure and pro- portion are not visible, when they really exist, except on the closest scrutiny. Arnold's distribution of emphasis by sentence- length may perhaps claim some such praise in this respect as would be given to a good picture. For one, I should not guess before counting that Arnold writes 20 per cent, of simple sen- tences. His brief propositions do not come in series : the nature of his subjects and of his method never makes them superfluously emphatic and conspicuous ; and so one is likely, in a general reading, to underestimate their number and impor- tance. But they are used with the greatest discretion. Again, it should be noticed that Arnold is hardly surpassed in the art of varying emphasis within the sentence itself. Here, long periodic clauses are succeeded by short loose ones ; or, a long period may consist of a half-dozen loose propositions that a less discrimina- ting man would have signalized by full stops. C^,7 WALTER PATER. Total paragraphs 37 Total words §450 Total sentences 219 Average words per paragraph 228.37 Average sentences per paragraph 5.97 Average words per sentence 38-54 Per cent, of single-sentence paragraphs i Appreciations, 500 periods. (Gerwig, for 500 periods.) Average predications per sentence 2.74 Per cent, of simple sentences 26 Per cent, of clauses saved 13-74 In Ruskin, Newman, and certain other writers, there is to be noted a decided reaction toward the long sentence. This movement reaches in Mr. Walter Pater perhaps the limit at which the paragraph and the long period can be reconciled. Mr. Pater 1 66 HISTORY OF THE ENGLISH PARAGRAPH. is conscious of the tendency of his style towards complexity and minute qualification, and he therefore conscientiously keeps to the unity of the paragraph. What is even more noticeable, he uses a large percentage of appositional clauses and phrases, that> while thev have partly the effect of parentheses, yet avoid the multiplication of predications and connectives. It is a weighty style, a correct style, a beautiful style in its fitting of word to notion ; but it has a wholly different order of procedure from that introduced by Macaulay. The coherence, always present, but seen by the reader at some expense to his attention, depends equally upon order of words and upon connectives ; very little indeed upon parallel structure. Of 300 sentences in the Appreciations, 66 are formally connected. The proportion of ands is startling. Thus : . Connective. Initial. Interior. On the other hand I I Then .... . . 4 And 21 Yet 3 So 2 For 7 Further . . I Again , I 2 Hence i Well 3 But 12 Still 2 I So far I Too . . 2 Also . . 2 At least I Indeed . . i Now I Thus . 2 The percentage of simple sentences is such that the distri- bution of emphasis is provided for mechanically, and a tribute should be paid to the often exquisite precision with which the right clause is made to bear the paragraph stress. DE QUINCE Y TO HOLMES. 167 J. R. GREEN. History of the English People. Total paragraphs considered 200 Average sentence-length for 200 sentences 29.04 Average words per paragraph c. 456.75 Average sentences per paragraph 15-75 Per cent, of single-sentence paragraphs I have included some statistics for the style of Mr. J. R. Green, to illustrate one of the newer types of historical writing. The sentence is much longer than Macaulay's, the paragraph very much longer than Macaulay's. The single-sentence para- graph is abolished. The variety that Macaulay secured by vary- ing the length of the paragraph and its structure is lacking here. The paragraphs are not well massed. The element of variety being made little of, an attempt is made to supply its place with that of intensity and weight. There are no waste sen- tences. The short sentences are sententious, and the long ones, while admirable in accuracy, are sometimes a little heavy. The coherence is good, but it is the coherence of severe method, and depends neither on connectives nor on transitional clauses. After all, it is a noble style, though not an easy one. BARRETT WENDELL. Paragraphs (chap, iv., in Eiiglisli Composition). Total paragraphs considered 55 Total words considered 9363 Total sentences considered 365 Average words per paragraph 170.23 Average sentences per paragraph 6.63 Average words per sentence 25.65 Per cent, of single-sentence paragraphs 5 We have spoken of Professor Wendell as a recent theorizer on the paragraph. Since he has treated the subject in a literary way, shunning the pedantry of technicalities, and since he mani- festly aims at producing superior massing and , emphasis, let us see what numerical results his practice gives. The chapter on the paragraph yields a sentence of 25. Nearly 24 per cent, of all i6S HISTORY OF THE ENGLISH PARAGRAPH. the sentences fall beneath the length of 15 words. The para- graph reaches but 170 words. Evidently the theory of Mass, when put in practice, tends toward keeping the paragraph to very moderate length. To mass well a long paragraph is a most difficult task. HOLMES. The style of Dr. Holmes is typical of certain popular writing, which, though not properly intuitive, omits formal predication as often as possible, and since it is not concerned with the finer restrictions of thought, omits connectives with the greatest free- dom. Holmes delights in appositive phrases and clauses, and in verbless sentences. In 500 periods Sherman found but 5 initial connectives. My own count, from 300 sentences in the ^w/c^rra/, yields a percentage very much higher — 27 initial con- nectives in 300 periods. The list runs thus : Connective. But However And So First, secondly . . Or Thus Yet On the contrary. In short Once more On the whole . . . Too At length Initial. 8 4 I 4 3 2 Interior. CHAPTER IX. THE PROSE PARAGRAPH: SUMMARY. It is the object of this chapter, not to state in essay form, woven together of all the judgments hitherto expressed, a com- plete view of the history of the prose paragraph, but to arrange in a somewhat mechanical way the more important of the theses that I propose. CHAPTER I. 1. (Page 12.) The modern reference-mark, *\, (sixth in the printer's list of reference-marks) is probably descended, not, as held by Mr. Maunde Thompson, from the original Greek gamma, but from the Latin mark P. 2. (Page 15.) The modern so-called section-maiTv, §, is prob- ably derived, noffrom the original gamma, as held by Blass, but from the Latin P ; surely not from the combination of two //, as taught by certain text-books. The type of this mark is prob- ably of Italian origin, 1467-1473. 3. (Page 14.) Indentation is probably not due, as the popular bibliophilic tradition asserts, to the omission of printed capitals to permit the insertion of rubricated ones, but to the example of those manuscripts where it is used without reference to rubrica- tion. CHAPTER II. 4. (Page 2 2ff.) While, for purposes of pedagogy, the writing of single-sentence paragraphs should largely be discouraged, in view of the natural tendency of students toward impartial analysis, it is nevertheless not correct to say, with Earle, "that the term paragraph can hardly be applied to anything short of three sentences, though rarely a complete and satisfactory effect is produced by two." For, although there has been a pretty steady decrease, in 300 years, in the use of the paragraphed 169 I 70 IflSrORY OF THE ENGLISH PARAGRAPH. sentence, most of the eminent writers of English prose have not hesitated to use this device, not merely to mark a transition but to signalize a stadium. 5. (P. 3off.) The onlv really new phases of rhetorical theory since Bain's " six rules," are Wendell's theory of Mass, and Scott and Dennev's theory of Proportion. Wendell's theory of Mass is : "A paragraph whose unity can be demonstrated by summarizing its substance in a sentence whose subject shall be a summary of its opening sentence, and whose predicate shall be a summary of its closing sentence, is theoretically well massed." Scott and Denney's theory of proportion is perhaps sufficiently implied in the following sentence : " Statements which standing alone would properly be independent sentences, are frequently united into one sentence when they become part of a paragraph." The theory implies also the converse of this statement. 6. (P. 30-32, 167.) {a) Wendell's theory of Mass is not applicable to any large proportion of existing paragraphs, and is difficult of application except in short i)aragr'aphs. Scott and Denney's theory of Proportion is true of those writers who have a conception of the paragraph as an organic whole, — Burke, Macaulay, Arnold, for example. The principle is so strongly operative in the best prose of today that we may probably go so far as to say : in general it is true that in the best modern para- graphs the distance between periods is inversely as the emphasis of each included proposition. {!') It will follow as a corollary from the principle last enunciated, that the tendency (noted by Professor Sherman) of English prose to reduce the sentence to Procrustean regularity of length, cannot indefinitely persist. CHAPTER III. 7. (P. 37ff.) In the history of English prose there has been, for relatively the same kinds of discourse, no pronounced increase or decrease of the average total number of words per paragraph. 8. (P. 42.) The paragraph of today contains more than twice as many sentences as did that of Ascham's day. Indeed, if SUMMARY. 171 we accept Macaulay's England as a present-day norm, the past increase in sentences per paragraph in three hundred years has been far more than one hundred per cent. 9. (P. 35ff.) In a list of 73 representative English prosaists, the average word-length of the paragraph falls in the case of each of 52 authors between the limits of 100 words and 300 words. Of these 52 authors, 25 show each an average falling between the limits of 200 words and 300 words ; while 27 show each an average fallinsr between the limits of 100 and 200 words. Of these two groups it would be unwarrantable to say that either is superior to the other in paragraph structure. The first includes many authors who are superior in delicacy and variety of proportion — Arnold, Newman, Pater ; the second includes many who are superior in terse emphasis — Bolingbroke, Swift, Carlyle, Lamb. But one of the greatest masters of terse emphasis, Macaulay, belongs in the first group, and one of the greatest masters of delicate and varied proportion, Ruskin, belongs in the second. Most of the writers whose average rises above 300 words are poor paragraphers, De Ouincey and Channing being exceptions. Most of those whose average falls below 100 words are writers in whom dialogue predominates. Fuller, Defoe, and Paley being exceptions. 10. (P. 35ff.) In a list of 71 representative English pro- saists, 5 show an average number of less than 2 sentences to the paragraph ; it show an average of more than 2 and less than 3 sentences ; 11 show an average of more than 3 and less than 4 sentences ; 6 show an average of more than 4 and less than 5 sentences ; 9 show an average of more than 5 and less than 6 sentences ; 10 show an average of more than 6 and less than 7 sentences ; 6 show an average of more than 7 and less than 8 sentences ; 3 show an average of more than 8 and less than 9 ; 4 show an average of more than 9 and less than 10 ; one averages 10 + ; two average 12 + ; one averages 14 + ; one 15 +, one 174-- The favorite numbers of sentences are therefore 2 + and- 3 +> each of which occurs 11 times. Then, in order of frequency, 172 HISTORY OF THE ENGLISH PARAGRAPH. come 6+, 5 -f, 4-(-, and 7 +, 9+, 8 +, 12 +, 14 +, and 15 + and 17 +. Dialogue-writing affects this list but very little. Of the romancers, Irving shows the highest average of sentences, 4.12. 11. (1*. 43.) There has been from the earliest days of our prose a unit of invention much larger than the modern sentence, and always separated, in the iniud of the writer, from the sen- tence-unit, of whatever length. In other words English writers have thought roughly in long stages before they have analyzed such stages into smaller steps. 12. (P. 44ff.) The paragraph as we know it comes into some- thing like settled shape in Sir William Temple. It was the product of perhaps five chief influences. First, the tradition, derived from the authors and scribes of the Middle Ages, that the paragraph- mark distinguishes a stadium of thought. Second, the Latin influence, which was rather towards disregarding the para- graph as the sign of anything but emphasis — the emphasis-tradi- tion being also of mediaeval origin ; the typical writers of the Latin influence are Hooker and Milton. Third, the natural genius of the Anglo-Saxon structure, favorable to the paragraph. Fourth, the beginnings of popular writing — of what may be called the oral style, or consideration for a relatively uncultivated audi- ence. Fifth, the study of French prose, in this respect a late influence, allied in its results with the third and fourth influences. The course taken by the conflict of the second principle with the rest, resulting in the intermediate unit of the amorphous par- agraphed sentence, is summarized, pp. 44-47. 13. (P. 47ff.) Throughout the eighteenth and nineteenth centuries, there is, in authors of regular methods, such as Hume and Macaulay, a perceptible but not a strong tendency towards reducing the average length of the paragraph to approximate constancy, in successive large groups of paragraphs. The author in whom the tendency is most pronounced is Macaulay. Here the tendency is so strong as to give a difference of only six words in the average paragraph word-length of the first and second vol- umes of the History of England. SUMMARY. 173 CHAPTER IV. 14. (P. 52ff.) {li) The recent investigations of Professor L. A. Sherman, in the development of the short sentence in English prose, are of much importance in their bearing upon the history of paragraph structure ; but by referring to the short sentence as " analytic," and again, in following the course of the development, by referring to the style of such intuitive (or synthetic) authors as Emerson as "analytic," the writer leads us into temporary con- fusion. From this it seems best, for the purposes of our discus- sion, to escape by the invention of certain new terms, as : segregating, applied to a style where the sentence of maximum occurrence is short, say, twenty words or less ; aggregating, to a style where the favorite sentence is long ; redintegrating, where the method of procedure is psychologically analytic ; intuitive, where the method is psychologically synthetic — omitting the steps of approach, the intermediate predications. (/') (p. syff.) The value of Professor Sherman's conclusions regarding the ^ oral' style are slightly impaired for us by the confused terminology mentioned in (12). The consequence of his theory concerning the decrease of predication is the application of the term 'oral' alike to styles redintegrating and intuitive. It seems better to limit the term 'oral style' to one in which the short sentence is employed, but the thought is psychologically redintegrating. 15. {a) The oral style as we now understand it — produced by the expression of redintegrating thought in a segregating sentence — is the style most favorable to the paragraph structure, {b) We may indeed almost define the oral style in terms of the para- graph. Thus : From the moment of the establishment of unity, in the development of the English paragraph, the oral sentence- sense means decreasing the number of predications in the period and increasing the number of propositions in the paragraph, in proportion to the author's conception of his reader's power of interpretation. 16. (P. 63ff.) The articulation of clauses without connectives is a help to the coherence of the paragraph in only one of two I 74 HISTORY OF THE ENGLISH PAKAGRAJ'H. cases : () wliere the ph\ce of connectives is supplied by transitional phrases or clauses. There- ore it is not likely that the decrease in the use of connectives — a decrease explained by Professor Sherman in his Analytics of Lit- erature^ chapter 26, — will continue indefinitely in prose that expresses proportioned and modulated thought. CHAPTER V. 17. (P. 67.) Though the paragraph pla3's no structural part in Anglo-Saxon, it is not rash to say that the paragraj)hs indicated by the rubricator have, in general, unity of subject, the exceptions being due to causes explained in (18). t8. (p. 66.) There were four distinct uses of the paragraj)h- mark, in Anglo-Saxon prose : id) to mark a logical section; {b) to note any emphatic point ; (c) to distinguish formally sacred names ; {(i) to ornament and distinguish titles, colophons, etc. 19. (P. 7off.) {a) The Anglo-Saxon prose sentence corre- sponds in length roughly with the sentence of the nineteenth century, {li) The Anglo-Saxon prose sentence increases slowly in length, and when it becomes the Middle-English sentence, reaches, under Latin influence, a length nearly as great as that attained by the latinized sentence of Jacobean times. 20. No English writer before Tyndale has any sense of the paragraph as a subject of internal arrangement. CHAPTER VI. 21. (P. 75ff.) In Tyndale we find the earliest writer who can be said to be in any sense a good paragrapher. 22. The most important men after Tyndale in the period from Tyndale to Temple, are Bacon, Hobbes, Browne, and Fuller, in respect of unity ; Lord Herbert, Burton, and Bunyan, in distribution of emphasis by variability of sentence-length ; Bur- ton in the matter of coherence without formal connectives; Fuller in the establishment of the deductive i)aragrai)h order. CHAPTERS VIl-IX. 23. The unity of the paragraph becomes nearly unimpeacha- SUMMARY, 175 ble in such men as Addison, Shaftesbury, Bolingbroke, Johnson, Hume, Burke. Only the best paragraphers of the nineteenth century, Macaulav, for example, surpass these authors in this respect. 24. Proportion in the paragraph pretty steadily increases from Temple to Arnold, both in the way of assigning due bulk to the amplification of important ideas, and in the way of dis- tributing emphasis by varying sentence-length. The following list will illustrate the latter point, by showing in the first column the percentage of each author in the use of sentences of less than fifteen words, in the second the average sentence-length. In starred authors the percentage ' of simple sentences, usually one or two points higher than the per cent, of sentences under fif- teen words, is substituted in the first column. Temple * Dryden - Locke - Defoe * Swift - * Addison - * Shaftesbury - * Bolingbroke Johnson * Hume * Goldsmith Burke Gibbon Paley Scott - * Coleridge Jeffrey - Lamb Landor Irving * De Ouincey * Macaulav ' Mr. Gerwig's figures. Per cent, of ocntcnc sentences of less than 15 words. considers 2 704 - 6 521 8 814 S 360 13 501O - 12 500 28 650 - 14 977 9 218 - 12 500 18 500 - 29 Q16 10 1562 - 17 392 14 1224 - 19 500 6 545 - 41 529 22 696 - 24 532 14 500 - 34 40,000 Sentence - Sentences length. considered 53-40 538 38.44 1300 49.80 814 38.68 360 40.00 II71 38-58 898 26.80 578 34-86 981 38-15 218 39.81 1200 26.94 868 26.09 916 31.21 1562 37-68 392 32.14 1224 37.60 777 50.65 545 27.19 529 25-43 696 26.73 532 38.81 815 23-43 41,579 I 76 HISTORY OF THE ENGLISH PARAGRAPH. Sentences considered. Sentence - Sentences length. considered. 3 1 -56 270 41.44 1228 20.5S II79 25-35 750 16.63 805 3145 356 33-31 814 34-41 605 3«-54 219 Per cent, of sentences of less than 15 words. * Carlyle - - - 18 500 * Newman - - - - 16 500 * Emerson - - - 41 1438 * Channing - - - 34 2000 * Bartol - - - 44 1500 * Lowell - - - - 23 683 * Ruskin . - . 18 718 * Arnold - - - - 20 500 * Pater - - - 26 500 25. Coherence by parallel construction of sentences, begin- ning in crude form in the paragraphs of the sixteenth century Euphuists — I-'ylj) Nash, Lodge, and their fellows — is reduced to a flexible and strong principle in Temple, Swift, Shaftesbury. Bolingbroke, Johnson, Hume, Gibbon, and Burke. It is weak in Dryden, Locke, Defoe, Sterne, Goldsmith, Paley. \\\ the next century it continues weak in Scott, Coleridge, Jeffrey, Irving, Emerson, Carlyle ; reviving in De Quincey, Macaulay, Arnold. It is neglected by many popular writers of the present day. 26. Coherence secured by so ordering words in the sentence that the mind shall pass from one sentence to another without check, is an art little observed in the sixteenth century. In the seventeenth it is perhaps strongest in Fuller and Burton. In the eighteenth century this principle is tolerably strong in Temple, Defoe, Shaftesbury, Bolingbroke, Fielding, Sterne, Goldsmith. It is very strong in Swift and Burke. It is relatively weak in John- son, Gibbon. In the nineteenth century the principle is rela- tively active in Lamb, Macaulay, Newman, and is at its best in Carlyle, for one type, and in Arnold, for another. 27. Coherence secured by the use of connectives is in most active force in the earliest periods of our prose. From the six- teenth century till the opening of the nineteenth it declines, reaching its ebb in the balanced sentences of Gibbon. At the beginning of the nineteenth century the principle became strongly operative in the reactionary prose of Coleridge, -but has again declined. Todav there are two tendencies, one continuing the SUMMARY. 177 decline, the other emphatically but intelligently reacting. The popular prose of the last twenty years tends to drop sentence- connectives. Another stream of writing, represented by the classical prose of Arnold, uses connectives freely but vitally. The present discussion holds that the dropping of inter- sentential connectives cannot successfully be accomplished without danger to one essential prose merit — the merit of reproducing the restric- tions and modulations which must characterize good prose of the intellectual type. The table on page 178 presents in outline the progress of the usage regarding inter-sentential connectives. The table shows certain interesting facts respecting the rela- tive use of different conjunctions by different authors. Walton uses the highest number of a)ids. Swift, Johnson, Macaulay use no amis at all ; Gibbon uses but one. Pater curiously exhibits more ands than any other man since Walton ; but his use of them is not formal merely. Coleridge registers the highest percentage of buts since Spenser, while De Quincey practically e'schews this word and exhibits about as large a number of interior howevers as Coleridge of initial buts. Initial therefore is little used since Ascham, and interior therefore not extensively — Coleridge head- ing the list with eleven. 28. The favorite type of paragraph in the history of our prose has been the loose type, although certain writers, as Butler in the eighteenth and Macaulay in the nineteenth, have shown some facility in the periodic type. 29. There has been, during the eighteenth and nineteenth centuries, a general tendency to make the topic-sentence of the paragraph short, but not to reduce it to laconic brevity. 30. The better paragraphs of the nineteenth century are far more organic, far more highly organized, than the better ones of the eighteenth. 31. The paragraph structure is, in proportion to the com- pK;xity and size of the thought conveyed, more economical of attention than the long periodic sentence ; and the rise of the paragraph structure is in no small degree due to this fact. TAHLK OF CONNECTIVES. The first column under eacli autlior shows the number of initial sentence-connectives; the seccmd column, the number of connectives that, though standing within the sentence, connect sen- tences, not clauses. The basis of computation in each author is joo sentences. n u < C t/5 c 3 a c 1 4) a S 1 B s S •§ •—1 4) ."2 V a; u c '5 C u Q n "3 n u n cd c c w E li. CL, 1/1 4) £ I 2 I 2 I 2 I 2 I 2 I . 2 I 2 I 2 I 2 I 2 I 2 I 2 I 2 I 2 I 2 49 6 30 13 I "9 19 ... 80 6 13 "'6 I 5 ... 3 5 II 3 7 2 .., 21 7 2 4 also I I I 3 I for 30 6 I I ... 12 6 vea nav ... I 5 I 4 ... ... likpwisp .. 2 4 I 2 4 I I ... ... 12 10 7 2 I I 3 I I 2 I I ... ... 2 I 3 ... 2 ... I first ptc > ... I 4 3 8 2 I 4 ... ajjain 2 5 I 2 II 2 I ... I 2 *** besides 4 I 2 I I ... then . * 3 I 5 3 I I 3 4 • • 3 3 ... 4 2 I ... ... too , I I "■ niorpdvpr . . 2 I ■ 1 I I I ... add to . ,.. . I I ... ... I ' I I ... ... I in a word I 1 on thp wliolp 1 I I I f contrariwise \ on the contrary on the other (hand) (side) I T I I I I I I 2 I ... ... I I I 1 • SO fcir ... 1 I ... • •• I t 1 I in other words ■■ I or or elsp I 2 I 3 I I 2 I 3 nor 2 23 8 I ... 2 I 12 2 I 9 ... I 26 4 3 IS 7 but, vea but 39 3 ... i^ I 25 ... 14 ... 10 3 I ,..T.. "1 ... ID ... 14 2 ... 12 3 ... 8 I vet . . 2 3 I once more J 3 [" I ... ... I 2 still 1 2 I ... howev^er 2 I ... 6 I I I ... 2 2 4J24 ... I 2 2 ... I T notwithstandinp" . . I I it is true I 1 I 2 i'... i 2 at least.. . .. . 1 I I I I ffinallv -; lastly (at last) I 2 2 5 3 3 I I I 2 2 I I ... therefore 15 I I 7 ... 2 ... ... 2 3 II ... 3 wherpforp thus /and still 1 and vpl 2 3 I 2 2 2 ... 3 I 3 I I 2 I I 2 I I ... 3 2 I I I I ... 2 2 ... I I I I hence I 5 I I 3 I I well.. 6 7 4 ' ... ... ... 6 I I 4 3 3 I 5 I I I indeed I 5 7 ... 5 ... I I ... iherpunfin I T othprwisp I 1 _ ... ... 1 1 I Total sentences form- ally connected 168 164 74 1 1 130 51 i 39 25 17 100 75 47 75 47 66 31 BIBLIOGRAPHY. 179 BIBLIOGRAPHY. Note. — Most of the works used in preparing Chapter I., having been given at the end of that chapter to facilitate the reading of tlie cut, are not here repeated. I. CRITICAL WORKS QUOTED. Angus, Joseph : Handbook of the English Tongue. London. Aristotle: Rhetoric, translated by J. E. C. Welldon. London, 1886. Arnold, Matthew: Essays in Criticism. London, 1893. Bain, Alexander: English Composition and Rhetoric. New York, 1869. Blades, William : Life and Typography of William Caxton. Lon- don, 1861-1863. Brandl, Alois : Life of Coleridge, translated by Lady Eastlake. Lon- don, 1887. Carpenter, G. R. : Exercises in Rhetoric and Composition (Advanced Course). Boston, 1893. Colvin, Sidney: Landor {Y^ngimh. Men of Letters). Kew York. De Quincey, Thomas: IVorks, ed. Masson. Edinburgh, 1889. Earle, John: English Prose. New York, 1890. Genung, J. F. : The Practical Elements of i^hetoric. Boston, 1892. Gerwig, G. W. : On the Decrease of Predication and of Sentence Weight in English Prose. University Studies, published by the L^niver- sity of Nebraska, Vol. II., No. i. Gosse, Edmund: History of Eighteenth Century Literature. Lon- don, 1 89 1. Hazlitt, William : Essays, ed. Carr. London. Hill. A. S. : Principles of Rhetoric. New York, 1883. Foundations of Rhetoric. New York, 1893. Hill, D. J.: Elements of Rhetoric. New York. Hunt, T. W. : The Principles of Written Discourse. New York, 1891. Hunt, T. W. : English Prose and Prose Writers. New York, 1891. Landor, W. S. : Works. London, 1876. Lowell, J. R. : Latest Literary Essays, Boston, 1892. McElroy, J. G. R. : 77^1? Structure of English Prose. New York, 1887. Minto, William : Manual of English Prose. Boston, 1892. I So HISTORY OF THE ENGLISH PARAGRAPH. Miiller, Ivan von : Handbtich der Klassisclieti Alterthiimswissenscliaft. Miinchen, 1892. Nichol, John : Prbner of English Composition. London, i8gi. Quackenbos, G. P. : Course of Composition and Rhetoric. New York. Roberts, E. S. : ./;/ Introduction to Greek Epigraphy . Cambridge University Press, 1887. Ruskin, John : Works, Brantwood ed. London and New York, i8gi. Saintsbury, George : History of Elizabethan Literature. London, 1891. Specimens of English Prose. London, 1885. Scott, F.N. and Denney, J. V. : Paragraph-Writing. Boston, 1893. Sherman, L. A.: Analytics of Literature. Boston, 1893. Some Observations on the Senteitce-Lcngth in English Prose, University Studies, published by the University of Nebraska, Vol. L, No. 2. Oti Certain Pacts and Principles in the Developtnent of Por/n in Literattire. University Studies, Vol. I., No. 4. Sidney, Sir Philip: Defense of Poetry. Arber Reprint, 1868. The Same, ed. Cook. Boston, 1890. The Countess of Pembroke's Arcadia. Facsimile of editio princeps, ed. Sommer. London, 1891. Stephen, Leslie: Hours in a Library, ist Series, London, 1874. 3d Series, London, 1879. Taine, H. A. : Histoire de la Litterature Anglaise. Paris, 1887. Thompson, E. Maunde : Hatidbook of Greek and Latin Palceography. London, 1893. Trevelyan, G. O. : Life and Letters of Macaulay. London, 1886. Wattenbach, Wilhelm : Anleitung zur Lateinischen Palceographie. Leipzig, 1886. IL Editions Employed in the Study of the Prose Paragraph. Addison, Joseph : Works. London, 1721. yElfric : Selected Hotnilies. Ed. Sweet. Oxford, 1885. Alfred: Orosius. E. E. T. S. 1879. Ed. Sweet. Ancren Riwle. Camdefi Society. Ed. Morton. London, 1853. Arnold, Matthew: Works. London, 1891. Ascham, Roger: Toxophilus. Arber reprint, 1868. Scholemaster. Arber reprint, 1870. BIBLIOGRAPHY. l8l Bacon, Francis: Advancement of Learning. Oxford, 1633. The Saine, ed. Aldis Wright. Oxford, 1891. Bartol, C. A, : Radical Probiejus. Boston, 1874. Beda : Ecclesiastical History. E. E. T. S. 45. Ed. Miller. Bentley, Richard : Epistles of Phalaris. 2nd ed. London, 1699- Blair, Hugh : Lecttcres on Rhetoric. London, iSig. Browne, Sir Thomas : Works. Bohn ed. London, 1852. Bunyan, John : The Pilgrim's Progress. Facsimile of editio pri^iceps,. London : Elliot Stock. Burke, Edmund: Works. Boston, 1889. Burton, Robert : Anatomy of Melancholy. London, 1676. Butler, Joseph : Works. Edinburgh, 1804. Carhle, Thomas : Sartor Resartits. London : Chapman & Hall. French Revolution. London, 1837. Essays. Boston, 1881. Channing, W. E. : Works. Boston, 1886. Chaucer, Geoffrey : Prose Works. E. E, T. S. Extra Series, v. and xvi. Ed. Morris and Skeat. Coleridge, S. T. : The Friend. Bohn ed. London, 1867. Cooper, Anthony : Second Earl of Shaftesbury : Characteristics. London, 1757. Cowley, Abraham : Essays. London, 1680. Essays. Ed. Lumb)'. Cambridge, 1887. Cranmer, Thomas : Works. Parker Society. Ed. Cox, from ed. of 1580. Defoe, Daniel : Robinson Crusoe. Facsimile of editio princeps. London, 1883. Defoe, Daniel : Essay on Projects. Ed. H. ]\Iorley. London. De Quincey, Thomas : Opium Eater. Boston, 1851. Works. Boston, 1874. Dickens, Charles : Works. New York, 188 1. Dryden, John : C^;/ .S"(?//rd', in Translation of Juvenal. London, 1726. IForks. Ed. Scott, Saintsbury. Edinburgh, 1882. Eliot, Sir John : Works. Ed. Grosart (from MS.), 1879. Eliot, George : Daniel Deronda. New York. Emerson, R. \V. : Miscellanies. Boston, 1883. Fabyan, Robert : Concordance of Histories. Ed. Ellis. London, 1811. Fielding, Henry : Tom fones. Ed. L. Stephen. London, 1882. 1 82 I/IS7VKY OF THE ENGLISH PARAGRAPH. Fuller, Thomas : W'ortJiifS of Ens;laiuL Editio princeps. London, 1662. Gibbiui, I'".(hvaid : J/istory of t lie Koiiiaii Etnpitr. Oxford, 1827. Goldsmith, Oliver : Hear of Wakefield. Ed. Gibbs. London, 1885. Gosson, Stephen : School of Abuse. Arber reprint, 1868. Green, J. R. : History of the Enf^lish People. New York, 1885. Greene, Robert: W^orks, ed. Grosart (Huth Library, 1881-86). Greville, Fulke, Lord Brooke: Life of Sidney. Ed. Grosart, 1870. Herbert, Edward, Lord : Autobiography. Ed. Dircks. London, 1888. Hobbes, Thomas : Leviatliaii. Ed. Molesworth. London, 1839. Holinshed, Raljih : Chronicles. London, 1574. Hooker, Richard: Ecclesiastical Polity. London, 1639 (?) Howell, James ; England's Tears for the Present Wars. Editio princeps, 1644. Hume, David : Histoty of England. Oxford, 1826. Hyde, Edward, Lord Clarendon : Histofy of the Great Rebellion. Oxford, 1 7 12. Irving, Washington: Sketch Book. New York, 1891. Jeffre\-, Francis : Contributions to the Edinburgh Review. New York, 1879. Johnson, Samuel : Works. Oxford, 1825. Jonson, Benjamin : Timber, or Discoveries jnade upon Men and Matter. Ed. Schelling. Boston, 1892. Kingsley, Charles: Alton Locke. New York, 1887. Lamb, Charles: Essays of Elia. Ed. Rhys. London, 1890. Landor, W. S.: Imaginary Conversatio7is. Boston, 1876. Latimer, Hugh: Seven Sermons. Arber reprint, 1869. Lincoln, Abraham : Letter. Century Magazine, May, 1889. Locke, John : Essay on Hitman Lhulcrstanding. Oxford, 1824. Lowell, J. R.: Among niy Books. Boston, 1881. Lyly, John : Eupjiues. Arber reprint, 1868. Macaulay, T. B.: Histojy of Engla?id. London, 1879. Essays. New York, Dr. fohnson. Encyc. Brit., 8th ed. Mandeville, John : Voia^e and Travaile. Ed. Halliwell. London, 1839. Michel of Northgate : Ayenbite of Inwyt. E. E. T. S. 23. Ed. Morris. Milton, John : Areopagitica. Arber reprint, 1868. BIBLIOGRAPII y. 183 Nash, Thos.: Works, Huth Library, ed. Grosart, 1883-4. Newman, Cardinal J. H.: Idea of a University . London, 1891. Overbury, Sir Thomas : Characios. Pater, Walter : Appreciations, with An Essay on Style. London, 1890. Plato and Platonism. London, 1893. Palev, William : Moral and Political Philosophy. Philadelphia, 1831. Pecock, Reginald : The Repressour of Overmuch 'Blaming of the Clergy. Ed. Churchill Babington. London, i860. Puttenham, George: The Arte of English Poesie. Arber reprint, i86g. Richardson, Samuel: Pamela. Ed. L. Stephen. London, 1883. Ruskin, John : Sesame and Lilies. London, 1891. St. John, Henry, \'iscount Bolingbroke : Letter to Sir IVilliam Wyndham. London, 1777. Scott, Sir Walter : h'anhoe. Edinburgh, 1871. Selden, John: Table Talk. Ed. H. Morley. London. Sidney, Sir Philip : Arcadia. Facsimile of originaT quarto. Ed. Summer. London, 1891. Defense of Poesy. Arber reprint, 1868. Spencer, Herbert: Philosophy of Style. Ed. Scott. Boston, 1892. Spenser, Edmund : View of the Present State of L?-e land. Ed. Gro- sart, 1882. Sterne, Laurence: A Sentimental foiir-ney. London, 1783. Stow, John : Chronicle. London, 1575. Swift, Tonathan : A Tale of a Tub and Battle of the Books. Editio princeps. London, 1704. Gulliver s Travels. Ed. Sheridan and Nichols. London, 1803. Taylor, Jeremy : The Liberty of Prophesyiiig. Ed. Heber. London, 1828. Temple, Sir William : Works. London, 1814. Thoreau, Henry D.: Excursions. Boston. Tyndale, William : The Obedience of a Christian Man. Parker Soc, Ed. Walter. Cambridge, 1848. The Same. Day's ed., 1572. Walton, Sir Isaac : The Complete Angler. Facsimile of editio princeps. London, 1876. Life of Hooker. Editio princeps. London, 1665. t84 II/STOKY of THE ENGLISH PAKAGKAP!/. Webbe, William ; A Discourse of English Poetrie. Arber reprint, 1870. Wendell, Barrett: Ettf^iish Composition. New York, 1892. Wilson, Sir Thomas : Tht^ A?tc of Rhctorique. Editio princeps. London, 1553. Owned by F. I. Carpenter, Esq. Wiclif, John : The English Works of Wiclif Hitherto Unpublished. E. E. T. S. 74. Ed. Matthew. Wordsworth, William : Prefaces. Ed. George. Boston, 1892. APPENDIX. NOTES ON THE VERSE PARAGRAPH IN MIDDLE ENGLISH. In this dissertation as presented in June, 1894, was included a final chapter of notes on the development of the paragraph in English verse. The following pages give such of those notes as pertained to the Mid- dle-English period. The rest of the original chapter is not printed, but reserved to form the basis of a fuller treatment at a later day. This unprinted material includes statistics of the paragraph-length of the blank verse of Milton, Cowper, Wordsworth, Tennyson, and Browning ; but the statistics would be but mere lumber here without a more adequate discussion of the aesthetic question involved than was possible for me to make. What Professor Corson has done for the "stanza" of Milton's blank verse should be done for the long poems of all the authors just mentioned. Careful consideration ought also to be given to the funda- mental question whether originally the logical unit, the sense unit, had in literature any strong influence in the development of the rhythmical unit, the stanza. As a preliminary study I have tried to learn whether the paragraph-mark had any metrical significance in our older poetry. The paragraph-mark does not occur in the oldest Anglo-Saxon poems. Neither is its place supplied by the colored initial, although colored initials do occur at the beginning of the main divisions. With the close of the twelfth century, however, we find both initial and mark used evidently with some metrical significance. To learn how far the use extends we examine about twenty authors, noting just where the scribes put paragraph-marks in the MSS. TJic Poema Morale. The Poema Morale (i 170 A. D., Zupitza ; 1 200-1225, Ten Brink) is written in rhymed septenars. These fall into strophes of four lines each, each strophe being introduced by a rubricated initial. At least this is strictly true of the Digby MS. (Bodleian A. 4.); the same regularity does not characterize the Trinity College MS. used by Morris, for here a rubricated letter often appears, apparently without significance, in the midst of a strophe. 185 i86 APPENDIX. The OniiuliDii, c. 1 200. The hfteen-syllable lines of the Ormuluin are written in the MS. continuously as in prose, the metrical point being placed at the end of the fourth foot of each verse. The Orf/nilum has, however, the para- graph-mark (see cut, p. II, Fig. 12). The length of the paragraph is exceedingly variable, depending entirely upon the scribe's rather arbitrary ideas of the logical divisions and of the emphatic points. The Holt- White edition gives only the longer logical divisions, disregarding very many of the MS. marks. In the Holt-White edition, beginning with the "Dedication," the first 60 paragraphs run as follows with respect to number of short lines:' 156, 28, 66, 30, 55, 8, 106, 30, 58, 20, 88, 58, 206, 162, 168, 12, 34, 16, 46, 19, 36, 8, 26, 148, 188, 343. 52, 200, 48, 54, 82, 56, 352, 172, 46. 56, 93, 80, 82, 20, 116, 92, 156, 68, 114, 54, 90, 124, 102, 32, 144, 148, 82, 152, 456, 158, 328, 612, 200, 228. Nothing in these figures points to a strophic grouping ; nor does any- thing in the verses themselves, although occasionally short passages are repeated with studied effort at musical effect. Laya7nonsBtitt, c. 1205. Layamons Brut is, in all MSS., almost without paragraphing. Both MSS. used by Madden muster together 14 marks, for the whole 30,000 lines. The signs occur too rarely to have either metrical or structural meaning, and are merely equivalent to marginal index-figures, pointing out important things. In MS. Cott. Otho, c. xiii., initials are used to mark divisions, but the divisions are too long to be considered as para- graphs. The Story of Genesis and Exodus, c. 1250. The song known as the Story of Genesis and Exodus is preserved in a unique MS. in the library of Corpus Christi College, Cambridge. The MS. is divided by red initials into loi short paragraphs, the brevity of which is in keeping with the light and easy movement of the poem. The average number of lines in the paragraph is 40.4, but there is great variability in the individual sections. I can see no signs of any strophic arrangement in the rhymed couplets of this poem. The same rhyme is, however, occasionally continued through several verses. The paragraphs are respectively of the following numbers of lines: 12, 16, 6, 58, 20, 16, 28, 8, 34, 14, 14, 4, 14, 14. 10, 50, 14. 22, 14, 8, 12, 20, 20, 12, 14, 10, 6, 22, 24, 10, 4, 4, 4, 16, 6, 16, 16, 2, 2, 4, 12, 6, 40, 16, 22, ' Ormin's long line is printed l)y White as a couplet. APPENDIX. 187 8, 104, 12, 10, 6, 72, 12, 42, 42, 28, 86, 14, 26, 36, 12, 58, 16, 34, 8, 24, 8, 4, 16, 18, 60, 6, 20, 34, 20, 46, 1 16, 14, 26, 6, 8. 24, 26, 4, 10, 6, 28, 16, 26, 14, 4, 52, 36, 14, 28, 22, 8, 10, 8, 10, 62, 26, 28, 26, 40, 80, 54, 80, 46, 14, 18, 14, 4, 16, 12, 18, 10, 66, 26, 14, 82, 8, 42, 6, 6, 4, 12, 6, 18, 30, 22, 16, 14, 8, 10, 30, 8, 12, 18, 20, 12, 26. 20, 34, 8, 16, 34, 10, 10, 4, 8, 6, 42, 18, 30, 38, 24, 32, 4, 4, 2, 2, 2, 2, 2, 2, 2, 2, 2, 18, 16, 4, 38, 44, 16, 16, 4, 12, 8, 30, 16, 6, 18, 16, 10, 16, 26, 44, 16, 14, 12, -6, 36, 12, 28, 14, 14, 10, 8, 28, 10, 8. King Horn, c. 1280. King Horn, according to the Cambridge University Library MS. (Gg. 4.27.2) used by Lumby (E. E. T. S.) falls into seven divisions, separated by rubrical initials. These divisions are again broken by paragraph-marks, colored red. The following table exhibits the length of each paragraph in lines, the paragraphs being grouped in capital paragraphs, represented here by Roman numerals : I. II. III. IV. V. VI. VII 24 16 18 14 3S 20 20 52 12 20 28 18 44 24 12 12 20 14 32 12 14 4 14 20 8 12 22 12 24 20 16 6 18 44 20 . 18 12 40 34 10 8 10 18 12 26 12 18 58 70 16 38 16 70 20 10 8 32 28 58 16 26 28 42 10 30 32 Of the 66 paragraphs only 26 are indivisible by four. This fact, taken as a hint, leads us to read the text with a view to seeing whether 1 88 APPENDIX. or not every four lines makes a stadium. The result of our readinj^ helps us to accept as at least probable the conjecture of the late Dr. Wissman, that there was an original strophic arrangement by fours. This arrange- ment seems to have been suggested to Wissman' by the occasional recurrence of the same rhvme in groups of four : e.g., 127-130; 227-230, etc. Havelok the Dane, c. 1280. Havelok the Dane (Skeat, E. E. T. S., from the unique MS., Laud. Misc. 108 Bodl. Lib.) has, if we omit Skeat's conjectured v. 46, 3000 verses, which fall into paragraphs by g6 rubrical initials. The para- graph-mark is used but once, then introducing the third section, and employed, I fancy, to avoid a capital thorn. So far as unity of subject is concerned, the paragraphing is excellently done. The paragraphs are respectively of the following numbers of lines: 26, 78,"^ 28, 16, 12, 22 20, 6, 16, 14, 20, 20, 6, 26, 16, 10, 20. 6, 34, 10, 39,^ 18, 80, 20, 84 42, 42, 16, 36, 26, 46, 22, 30, 18, 176, 14, 64, 66, 18, 10, 10, 68, 32 7, 250, 26, 14, 32, 52, 94. 8, 10, 36, 6, 34, 48, 6, 44, 4, 30, 10, 10 14. 38, 10, 20, 12, 20, 14, 18, 4, 10, 24, 22, 42, 24, 26, 6, 20, 58,24 54, 54, 16, 118, 24, 10, 20, 30, 12, 6, 72, 14, 6, 14, 16, 24. Although the large number of sections divisible by four might sug- gest the presence of strophic arrangement, none such appears on examination. The poem was not, like Horn, fitted for musical recita- tion. Gi/y of Warwick, 1 3 o - 1 3 2 5 . The various MSS. of Guy of IVarivick differ widely in the length of their main divisions. In the Auchinleck MS. the twelve-line stanzas are usually introduced by the paragraph-mark. The mark occurs only three times in the Cambridge Paper MS., namely at lines 7487, 1 1,267, 11,337. Zupitza, however, in his edition from the MS. last named, inserts the paragraph-mark many times, in order to break up the long divisions. Sir Bevis of Ha7ntoim, 1300-1325. In the Romance of Sir BcT'is of Hamtoun (Auchinleck MS.), the par- agraph-mark is placed before the third and sixth lines of the six-line ' King Horn. Utitersitchungen zur Mittelenglischeu Sprach- tind Littera- ttirgeschichte. Strassbuig, lS"6. P. 63. ^^ Skeat's conjectured line 46 is omitted. At 410, 411, the hnes are perhaps corrupt, for they do not rhyme. This fact may account for the odd number of lines, 39. APPENDIX. 189 stanzas, for the first 474 lines. With the 475th line the metre changes to the couplet, and hereafter the mark subdivides the main sections, which are marked by initials. I cannot discover that the mark has any metrical import after 474. The real paragraphs of the poem are the capital paragraphs [cf. p. 29). The first of these consists of nine six-line stanzas ; the second, of nine ; the third, of fourteen ; the fourth, of seventeen ; the fifth, of four- teen ; the sixth, of sixteen. Each group has a certain unity of its own- The rest of the paragraphs are of varied length and indifferent unity. They indicate no strophic tendency. The count runs as follows, by lines : 54, 54, 84, 102, 84, 96, 1 10, 60, 94, 32, 66, 72, 80, 52, 28, 68, 82, 44, 82, 88, 50, 52, 40, 60, 1 10, 136, 78, 188, 62, 42, 30, 184, 68, 64, 206. 108, 40, 12, 94, 10. The Bftice,^ c. 1376. Both the two important MSS. of the Bruce, the Cambridge and the Edinburgh, show the paragraph-mark ; but the paragraphing does not agree closely in the two. I cannot see that the mark has any metrical import in either MS. Pinkerton, who edited the Bmce in 1790, divided it into twenty books, instead of the long and irregular paragraphs. Jamieson, 1820, preferred a division into fourteen books; while Inness, 1866, following the MSS., divided his text into paragraphs, 150 in all. Cursor Mundi,^ 1 4th c. There is no meaning in the paragraphing of the various MSS. of the Cursor Mundt. Each successive scribe was positive that the unity of his predecessors' paragraphs was faulty, and so each placed the marks differently. Thus, Fairfax MS., 14 Bodleian, has, in the first 1000 verses, twelve capital paragraphs, seventy-one paragraphs. Cotton Vesp. A iii. Brit. Mus., has, in the first 1000 verses, two capital para- graphs and five paragraphs. Gottingen MS., theol. 107, has in the first 1000 lines no capital paragraphs, sixteen paragraphs. MS. R. 3. 8. Trin. Col., Cam., has in the first 1000 lines seven capital paragraphs, fifty paragraphs. The Legend of Celestin, c. 1360 (?) The Legend of Celestin^ (MS. Laud. L 70, fol. 118 b) is written in ' The Bruce, or The Book of the Most Excellent and Noble Prince, Robert de Broyss, King of Scots. Ed. Skeat, E. E. T. S. -Cursor Mundi. Ed. Morris, E. E. T. S. 3 Ed. Horstmann, Auglia, I., p. 67. IQO APrENDIX. strophes of five lines, ihyniiii,i< a a a b b. The paragraph-mark occurs only at the beginning of a strophe, thus serving as a metrical index. But it does not begin every strophe. In the companion piece, Susamia, the mark introduces each thirteen-line strophe. The capital para- graphs of the Celcstin include respectively the following numbers of strophes : 3, 2, 2, 2, 2, 3, 3, 2, 2, 2, 2, 2, 2, 2, 2, 3, 3, 2, 3, i, 3, 2, 2, 2, 2. I. 3. 2, 3, 3. I. 3. 3. 3. 3. 3. 3. 2, 3. I. 3. 2, 2, 3, 2, 3, I, 2, 3, I, 3, 2, 3, 2. 2, 3, 3, 2, 3, 3, 2. Joseph of Arimathic,^ c. 1350. In Joseph of Arimathic (Vernon MS. fol. 403) the alliterative verse is written like prose. The whole poem is, however, marked off into lines and half-lines by three devices: {a) small capitals; {h') para- graph-marks ; {c) metrical dots. The paragraphs are indicated by capitals. The paragraph-mark serves two purposes, namely, that of a metrical sign, and that of an emphasis mark, or index. It is noticeable that these two uses usually coincide in result, z". e., each paragraph-mark usually notes the beginning of a line, and at the same time calls attention to something important. The length of the capital paragraphs is successively as follows, no strophic tendency appearing : 3,= 16, 17, 16, 11, 10, 14, 20, 8, 20, 38, 6, 31, 16, 30, 19, 18, 18, 21, 29, 19, 32, 36, 22, 17, 29, 37, 51, 9, 41, 12, 10 12, 20. The IFars of Alexander? The IVay-s of Alexander has 27 passus, the last incomplete. Skeat reckons a total of 5677 vv. The number of verses to the passus runs thus: 213, 311, ig8, 158, 240, 336, 263, 286, 313, 288, 239, 167, 191, 264, 192, 120, 239, 240, 119, 192, 145, 191, 169, 216, 192, 144, 51 (incomplete). But it must be remembered that in all the alliterative poems lines were frequently lost in the copying ; thus the following lines appear in the Dublin MS., but not in the Ashmole : 1633, 1766, 1767, 2168, 2724, 2842, 2980, 3167, 3267, 4002. If we add these missing lines, our count of verses in the successive passus will stand : 213, 311, ig8, 158, 240, 336, 264, 288, 314, 288, 241, 168, 192, 265, 192, 120, 240, ^Joseph of Ariniatliie, otherwise called the Romance of the Seint Graal, ed Skeat, E.E.T.S. 2 The MS. is imperfect before this ^f. 3 The Wars of Alexander, an alliterative romance, translated chiefly from the Historia Alexandri Magttt de Pnv/iis, ed. Skeat, E.E.T.S. APPENDIX. 191 240, 119, 192, 145, 191, 169, 216, 192, 144, 51 (incomplete). Although this change has made one of the even numbers odd (264-265) it has greatly raised the sum of evens, which (omitting the incomplete last passus) now stands 18 out of 26, or 6g per cent. This is a curious thing in verse supposed to be alliterative merely, and not strophic. We look farther — to the paragraph-marks. The paragraph-marks are distributed as follows in the Ashmole MS., the Roman numeral indicating the passus, the Arabic the number of the line in the Skeat text : I- 23, 95, 190. II. 214, 334, 406, 47S. III. 525- IV. V. 881, 905, 1024. VI. II2I. VII. 1505. VIII. 1958. IX. X. 2415, 2439, 2463. XI. 2727, 2755, 2775, 2799, 2823. XII. 2894. XIII. 3037, 3085,3180. XIV. 3252, 3299, 3420. XV. 3540, 3564, 3576. XVI. 3762. XVII. 3780. XVIII. 4163, 421 1, 42.35. XIX. 4259. XX. 4378. XXI. 4644, 4692. XXII. 4715. XXIII. 4906. XXIV. 5075, 5103. XXV. 5291. XXVI. XXVII. 5656. Therefore the numbers of lines per paragraph, by the marks of the Ashmole, are: 22, 72, 95, 24, 120, 72, 72, 47, 356, 24, 119, 97, 384, 453. 457, 24, 24, 264, 28, 20, 24, 24, 71, 143, 48, 95, 72, 47, 121, 120, 24, 12, 186, 18, 383, 48, 24, 24, 119, 266, 48, 23, 191, 169, 128, 188, 365. If now we count the initial at the beginning of each passus as taking the 19- APPENDIX. j)lace of the maik, when that is kicking, and if we add in their proper places tlic ten extra lines found in tlie Dublin MS. but lacking in the Ash- mole, the list just given will stand thus : 22, 72, 95, 24, 120, 72, 72, 47, 198, 158, 24. 120. 97, 336. 48. 216, 240, 48, 314, 96, 24, 24, 144, 121, 28, 20, 24, 24, 24, 48, 120, 24, 48, 96. 24, 48, 48, 121, 48, 72, 24, 12, 84, 102, 18, 240, 144. 48, 24, 24, 119, 192, 74, 48, 23, 191, 169, 28, 188, 192, 144. 29. The merest glance at these numbers shows that the even ones greatly predominate. This predominance suggests a possible strophic arrange- ment. The suggestion is strengthened by several curious things notice- able in the Ashmole MS. First, of the twenty-seven passus, only ten begin with the paragraph-mark. These ten are passus ii., iii., v., vi, xix., XX., xxii., xxiii., xxiv., xxv. Why only ten so begin is not plain ; of course it may be a matter of chance, but, again, there are three passus, iv., ix., xxvi., that contain no paragraph-mark at all. I cannot understand the reason of this, unless it be that the mark was inserted now and then merely as a metrical regulator ; and I grant this to be but a poor reason for the omissions. But at any rate, on the suspicion that the mark means something metrically, as it did in Joseph of Arimathie, we look for the smallest paragraph. It turns out to be one of four lines, Dublin, 2795-2799. No paragraph smaller than twelve lines occurs in the Ashmole. Using four as a divisor we discover that 73 per cent, of the paragraphs in the revised list given above, are divisible. Immediately we begin to read to see if each four lines form anything like a stadium. The first paragraph contains twenty-four lines. Its natural subdi- visions seem to be 1-3, 4-7, 8-10, 11-14, 15-18. 19-23. So far, so good. This paragraph breaks into six divisions of respectively 3, 4, 3, 4, 4, 4, lines. We further discover that the first three lines have /as the letter of alliteration, the next four have /, the next three have c {k), the next four w, the last four ;-. It would be easy here to say that the first and third divisions have lost each a line, but the sense is perfect as the text now stands. We find no other groups of lines with the same alliteration. As we continue the reading we find that by no means does every fourth line mark a stadium. We therefore double the number, and read for groups of eight. ' The result is surprisingly persuasive that there is a genuine strophic arrangement by eights. There are indeed cases where the eighth line does not end a sentence, but the great majority of these groups of eight do mark real stadia. We may at APPENDIX. 193 least conclude that there is a strong tendency in this poem to write alliterative verses in strophes of eight ; but that the rule is not without many exceptions. I have tried to test this hypothesis still further by considering the whole text as divided both by the paragraph-marks of the Ashmole, and by the initials of the Dublin. The result, however, is not so assur- ing as the paragraphing of the Ashmole alone. It is entirely possible that the original poem was divided regularly into eight-line groups, but in the present state of our knowledge it seems rash to reject so many good verses and add so many conjectural ones as would be nec- essary to render all the passus divisible by eight." William of Palcnic, d. 1350. T\\Q Romance of \Villiai7i of Palerne is preserved in a unique MS. in the library of King's College, Cambridge. The paragraphing in this MS. is done by the use of small blue and red initials. A quire is miss- ing at the very beginning of the poem, and although Skeat in his E. E. T. S. reprint of Madden's edition substitutes for the lost lines the original French, I have begun my count at the first of the English divisions. The paragraphs run as follows as to number of lines : 78, 29, 52, 9,= 28, 34, 113, 39, 49 (folio 10 lost) 54 (folio ID lost) 52,3 42, 49, 102, I 19, 38, 79, 74, 26, 24, 27, 61, 75, 162, 20, 100, 53, 49, 66, 60, loi, 65, 122, 91, 72, 63, 36, 51, 84, 119, 29, 21, 95, 92, 25, 26, 61, 51, 35, 56, 73, 61, 32, 36, 54, 93, 42, 63, 64, 75, 40, 39, 35, 21, 44, 46, 45. 40, 45. 47, 65, 32, 34, 99, 43, 48, 38, 45, 137, 65, 44, 63, 40, 43> JO, 43- 99. 84, 64, 32. 32, 54, 64, 46, 70, 39, 14, 20.= Only 55 per cent, of these numbers are even, a proportion not large ' After writing this account in August, 1893, I learned that in Englische Studien (1S92) Max Kaluza has discussed StropJiische Gliede.ru7ig in der Mi(- telenglischen rein alliterirenden Dichtung. Kaluza, proceeding from a different point of view from my own, arrives at the conclusion that the Wars was written in strophes of 24 1 To reach tliis conclusion he has to add S3 verses to Skeat's (and Stevenson's) 5677 verses. This process gives 5760 lines, which is indeed a multiple of 24, and not only of 24 but of 48 and 72, for that matter. Kaluza does not utterly ignore the paragraph-mark, though he does ignore the initials of Dublin. He merely 5ays that the mark stands only tour times in the midst of a strophe (of 24) : 3576, 3762, 5103, 5655. Surelv lie must have over- ooked 2755, to say nothing of many cases in Dublin. ^ These two paragraphs begin with small letters. 3 1 restore a line uliich Skeat thinks ma\' be lost, at 500. 194 APPENDIX. en()Ui,'li t(i point to any lej^ular slropliic arraiij^ement. The text reveals here ami there a logical unit of four lines, but the grouping was hardly a conscious one. The Destruction of Jcntsalcin. The Destruction of Jerusalem, another of the alliterative poems of this period, is ])reserved in various MSS. Cotton Caligula A II, is divitied into (juatrains by the paragraph-mark, which here becomes chielly a metrical sign. The poem is broken into sections by four divine invocations (vv. 438, 888, 1104, 1332). The Bokc of Curiasye. The Boke of Curtasye (Sloane MS. 1986, Brit. Mus.) is written in rhymed couplets. The jjaragraph-niark indicates a predominating strophic arrangement in four-line groups. In the third main section, however, the paragraphing is irregular. The number of lines in the successive strophes of the whole poem runs as follows : I. 8, 4, 4, 4, 4, 4. 4, 4, 4, 4, 4, 4, 4. 4. 4, 4, 4, 4, 4, 4, 4, 4. 4, 4- 4, 4, 4. 4. 4, 4. 4, 4, 4, 4, I. II. 8, 4, 4, 4. 4. 6, 4, 4, 6, 4, 4, 4, 4, 4, 8, 6, 6, 4, 6, 6, 4, 4, 6, 4, 8, 4, 4, 10, 4. 4, 6, 4, 4, 4, 4, 4, 4, 6, 4, 8, 4, 6. III. 10, 18, 16, 10, 6, 12, 8, 4, 12, 7, II, 14, 8, 6, 7, 7, 14, 20, 4, 8, 10, 10, 22, 10, 18, 6, 10, 6, 22, 8, 54, 20, 12, 19, 43, 26. The Vision of William concernjn}:; Piers the Plowman, 1363. The mark is occasionally used in the MSS. of Pie/s the Plowman with purely metrical meaning. In such cases it is employed, as in foseph of Arimatine, to divide a long line into two short ones. The mark is also used in all the MSS. with its usual force, but without any strophic signiticance. The paragraphing of the various MSS. differs. To illustrate : the Vernon MS. has 28 paragraphs in the second passus (208 lines) while the MS. Phillips 8321 has 10 para- graphs for the same passus (252 lines). In the third passus (281 lines) Vernon has 55 paragraphs. For the same passus (349 lines) Laud. 851 has 58 paragraphs. Of the three MSS. mentioned, the Vernon is the most justly divided, the Phillips least. The MS. of Trin. Col. Cam. (1315.17) has perhaps the shortest paragraphs of all. Laud., Rawlin- son (Poet. 38. Bodl.), Trin. Col., and others, have breaks between para- graphs, a device which of course adds to the beauty and legibility of the page. M.S. Bodley 814, Oxford, is of early date, but shows no inter- paragraphic breaks. APPENDIX. ")5 The Dcstntction of Ti-oy. The alliterative romance of the Destruction of Troy (K. E. T. S. 39 and 36, Panton & Donaldson) is from a unique MS. in the Hunteriaii Museum of the University of Glasgow. The paragraph-mark is not used in this MS. and the divisions indicated by initials are rather too long to be called paragraphs. These divisions are however often broken up into paragraphs by spacing and the insertion of an explana- tory phrase, c. g. "The (Insuare of Jason to Medea." The Staciouiis of Rome, c. 1460. In the Staciouns of Rome the real paragraphs are indicated by rubrical initials. About one-half of these capital paragraphs are subdivided, and very skillfullv, by the marks — alternately red and blue. I fail to discover anv strophic arrangement beyond the rhymed cou[)let. The following table exhibits the length, in lines, of the sub-paragraphs, each brace equaling a capital paragraph. \ 6 ^0 10 \ 6 ^ 6 ( 8 ( 10 \ 18 C 12 \ 22 \' 12 \ 20 \' 16 (,' 16 8 ^6? 6] 8 I 14 J \ X I I I i U6 r 8 } . 8 20 8 i( 8 i' 8 \ 20 \ 12 ^ 14 \ 10 \ 4 \ 12 <| 8 18 ^ ] 4 ? 14 ■( } I 12} 't } 2 10 ( 6 4 4 4 4 4 2 2 4 \ 16 \ 12 { 6 \ 4 \' 14 ( 12 \' 4 s 24 ^ 18 1; 6 ( 8 ( 10 1; 10 '( I I 1 I (10 1 1 16 4 2 4 4 2 22 6 I X I I J8j8 ( 12 \ 10 \ 8 I I I A forte Arthnre. In the Thornton MS. the Morte ArtJiure is broken into 80 para- graphs by initials. The length of these paragraphs by lines is respec- tively as follows : 25, 52, 3,8, 50, 65, 57, 16, 33, 45. 25, 108, 7, 32, 16, 1 96 APPEA'DIX. 40, 15. 51. U, 43. 4S6, 395, 20, S, 42, 37, n, (;2, 14, 26, 22, 20, 8, 26, 27. 33. -^'. 63, 28, 167, 40, 12, 29, 45, 48, 37, 24, 143, 84, 11, 26, 65, 46, 27, 12, 31, 52, 44, 22, 26, 30, 132, 118, 31, 16, 88, 189, T,},, 27, 24, 55, 13, 17, 26, 22, 28, 35, 53, 42, 107, 85. Althousjfh the 80 ]iaragraphs average 54.3 lines, tlierc is wide lluctua- tidii in length — 8-486. Of the 80 paragraphs 24 are nuniericallv ilivisible into strophes of four lines. We examine these to see if the division is anything more than a numerical one. By altering occasionally the punctuation of Perry (E. E. T. S.) we reach the following subdivisions, each of which may be said to form a minor stadium. t V. 62-v. 77 = S2 = cS-f8+4+4+4+S+S+4+4- T[ V. 28S-V. 308=16 = 4-1-4-1-4+4. If V. 522-v. 553 = 32 = 8+S+44-S-f-4. \ V. 554-v. 569=16 = 4+84-4- t V. 570-v. 609 = 40 = 8-1-4-1-4+4+8+4+4 +4, Tl V. 1617-V. 1636 = 20 = 4+12+4. \ V. 1637-V. 1644 = 8 = 8. Tf V. igi2-v. 1919 = 8 = 4+4. t V. 2290-v. 2329 = 40 = 4+4+4+4+8+6+4+6. \ V. 2330-v. 2341 = 12 = 8+4. t V. 2416-V. 2463 = 48 -4+4+4+4+6+10+8+8. .^ V. 2525-v. 2668=143 = 4+4+4+4+4+4+4+4+8+4+10+6+4+6 +7+4+8+6+4+4+4+4+10+4+8+4+6. t V. 2669-v. 2752 = 84 = 4+6+4+4+6+6+4+8+4+6+5+10+12+5. ^ V. 2990-v. 3001 = 12=12. 11 V. 3033-v. 3084 = 52 = 4+4+4+4+6+8+4+4+6+8. t V. 3085-v. 3128 = 44 = 4+4+4+4+6+4+8+6+4. \ V. 3488-v. 3503=16=16. H V. 3504-v. 3591=88 = 6+6+8+4+8+8+6+8+4+4+4+6+4+4+4 +4- \ V. 3841-V. 3864 = 24 = 4+4+4+4+4+4- Tf V. 3998-v. 4025 = 28=12+4+12. t V. 4263-v. 4348 = 85=13+16+4+4+4+4+4+3+12+6+10+5. It is evident hat there is a strong tendency toward a strophe of four lines. But the poet does not hesitate to alternate with this quatrain groups of six, or eight, or even sixteen verses, or again certain irregular groups of odd numbers. The paragraphs that are not evenly divisible by four show about the same proportion of four-line groups. For example the ^[ 2525-2668 = 144 — i '= 143 gives the following ^ Ferry has unquestionably missed his count at 2592. APPENDIX. 197 j^'roups : 4, 4, 4, 4, 4, 4, 4, 4, 8, 4, 10, 6, 4, 6, 7, 4, 8, 6. 4, 4, 4, 4, 10, 4, 8, 4. 6. Chaucer. The MSS. of Chaucer vary greatly in their paragraphing. The Ellesmere MS. uses both initials and paragraph-marks. In the Pro- logue the initials are used almost exclusively. Later. on the initial seems to be used as marking a more important division than the ^, but it is not plain that the capital paragraphs form organized wholes of which the ^ marks subdivisions. The Petworth MS. likewise shows both initial and *', the former less rarel}' than the latter, and less rarely than the initial in Ellesmere. On the other hand the Hengwrt, Cam- bridge, Corpus, and Lansdowne MSS. use the ^ regularly and the initial rarely. In Ellesmere it often happens that where an initial stands in the text, the ^ occurs opposite to the initial in the margin, and precedes a marginal note. The following table shows the distribution of paragraph-marks and initials, in the Prologue and the Knighfs Tale, for the three MSS., P^llesworth, Petworth, and Lansdowne. No particular value is claimed for the table, except as it shows how nearly all the scribes pitched on the important points, while differing widely concerning the minor subdivisions. Except in the case of Ellesmere, wh'ere initials are marked C, I have not distinguished between capital and \. The figures are the numbers of lines in the Furnivall six-text edition. ^LL ESMEKE. Petworth. Lansdowne; Ellesmeke. Petworth. Lansdowne c I I c 331 331 331 c 19 c 361 361 361 c 35 c 379 379 379 c 43 43 43 c 388 388 388 51 51 c 411 411 411 73 c 445 445 445 c 79 79 c 477 477 477 c TGI lOI lOI c 529 529 529 c 118 118 118 c 542 542 542 c 165 165 163 545 545 . . c 208 208 208 c 567 567 . . c 270 270 270 c 587 587 587 c 285 285 285 c 623 623 623 c 309 309 309 c 669 669 669 lyS Arri:Nnix. Ellksmekk. I'kiwokiii. I.ANSDOWNl!. Kllesmkkic. Pkiwoktji. I.ANsno c 715 715 715 13^7 . . 747 747 743 c 1347 1347 1347 769 769 769 c 1355 '355 1355 7X3 . . 784 1361 . . 7SS 788 1380 . . 1 380 Sio 810 1399 1 393 i;th. I.ANDSOWNE. Kl.l.ES.MEKl' 2947 3041 2963 3043 c 2967 2967 3057 c 29S7 2987 3067 c 3017 3075 3021 3090 3027 3097 3035 Peiwokth. Lansdowne. ^075 i-o.e ^^^^^^2^i^ RETURN CIRCULATION DEPARTMENT TO— #^ 202 Main Library LOAN PERIOD 1 HOME USE 2 3 4 5 6 ^^CFi\/ED ^^P 1 9 m4 C/ RCUUTfON SFP 1 1 2005 ALL BOOKS MAY BE RECALLED AFTER 7 DAYS 1-monfh loans may be renewed by calling 642-3405 1-year loans may te recr,src-r- r;v bringinc) the books to the Circulation Desk ^ DUE AS STAMPED BE LOW OCT 1 8 1984 UNIVERSITY OF CALIFORNIA, BERKELEY FORM NO. DD6, 60m, 1/83 BERKELEY, CA 94720 General Library University of California , ,, Berkeley -^