LIBRARY UNIVERSITY OF CALIFORNIA. RECEIVED BY EXCHANGE Class TIME IN ENGLISH VERSE RHYTHM AN EMPIRICAL, STUDY OF TYPICAL VERSES BY THE GRAPHIC METHOD WARNER BROWN, A.M. Submitted in Partial Fulfilment of the Requirements for the Degree of Doctor of Philosophy, in the Faculty of Philosophy, Columbia University Reprinted from the Archives of Psychology, No. 10 NEW YORK THE SCIENCE PRESS 1908 TIME IN ENGLISH YERSE RHYTHM AN EMPIRICAL STUDY OF TYPICAL VERSES BY THE GRAPHIC METHOD BY WARNER BROWN, A.M. Submitted in Partial Fulfilment of the Requirements for the Degree of Doctor of Philosophy, in the Faculty of Philosophy, Columbia University Reprinted from the Archives of Psychology, No. 10 NEW YOKK THE SCIENCE PRESS 19O8 PRESS or THE NEW ERA PRINTINS COMPAQ LANCASTER, PA. CONTENTS PAGE. INTRODUCTION I CHAP. I. METHODS OF STUDYING VERSE RHYTHM 5 a. Simple Observation by Ear 5 6. Phonographic Repetition 7 c. Tapping 7 d. Reaction Methods 9 e. Determination of Subjective Rhythms 10 f. Experimental Method 1 1 CHAP. II. GRAPHIC METHOD AND MECHANICAL AIDS 14 CHAP. III. PITCH, INTENSITY AND DURATION 21 CHAP. IV. TIME RELATIONS. THEORIES OF THE METRISTS AND POETS . . 26 CHAP. V. TIME RELATIONS AS TREATED BY THE EXPERIMENTALISTS ... 34 CHAP. VI. EXPERIMENTAL RESULTS 38 a. Procedure 38 5. Duration of Feet 42. c. Ratio of Accented to Unaccented Syllables 52 d. Rising and Falling Rhythms 55 e. Two-Syllable and Three- Syllable Rhythms 58 f. Regularity of Feet and Syllables 64 CHAP. VII. SIGNIFICANCE OF RESULTS FOR METRICAL THEORY 70 CHAP. VIII. SIGNIFICANCE OF RESULTS FOR THEORY OF RHYTHM 74 183493 TIME IN ENGLISH VERSE RHYTHM This study treats of the time relations found in English verse rhythm. It states the defects of the current methods of studying the subject and the advantages of psychological experimentation by the aid of graphic records of the voice. Selected examples of Eng- lish verse are then examined and the various theories of metrists and psychologists on the subject criticized in the light of the ex- perimental results. INTRODUCTION No one will contradict the statement that all verse is character- ized by rhythm. About poetry there may be some dispute; at any rate highly poetic prose may not be particularly rhythmical and may at the same time be better poetry than some very regular verse. Every theory of rhythm, no matter what its other peculiari- ties, must be sufficiently broad to include verse. Music and verse are the sources of most of our more affective experiences of rhythm. Not that there are not many other rhythmical experiences in our daily life, indeed the ordinary functions of our organic existence are constantly supplying a stream of sensations which are at once recognized as rhythmical when attention is paid to them. But it is through our ears that we are best able to enjoy perceiving rhythms and with the voice that we produce them oftenest and with the greatest satisfaction. The ordinary organic rhythms go on un- perceived for the most part, and the same may be said of such semi- voluntary movements as walking or riding a bicycle. Even heard rhythms are frequently ignored, it is true, but music and still more verse receive a high degree of attention. And even if we learn to disregard music under certain circumstances, when we are occupied in conversation or otherwise preoccupied, the sound of recited verse retains its power to compel our attention. This is perhaps due to the infrequency with which recited verse is heard under modern conditions; the ancients perhaps paid as little attention to the public reciter as the modern gourmet pays to the orchestra while he dines. However this may be with regard to passive appreciation, it is certain that the act of producing or repro- ducing music or verse still better good music or poetry is keenly enjoyed by most people. 2 TIME IN ENGLISH VERSE RHYTHM The mere admission, however, that all verse is rhythmical tells very little about the nature of rhythm itself and still less about verse. It is here that all the confusion arises, for without an adequate determination of one or the other term no advance can be made. If rhythm could be defined once for all it would be feasible to apply this definition to verse. On the other hand if verse were once fully analyzed a definition of rhythm could be indirectly derived with some degree of assurance. In view of the difficulty of defining rhythm independently the other course seems the more hopeful. The task appears simple, and once accomplished there could be no more of the "inaccuracy, con- fusion, misconception, misrepresentation, mystification, and down- right ignorance" of which Poe complains so bitterly in that essay of his which is still the most penetrating analysis we have of English verse. 1 Yet simple as the task seems it has not been accomplished and indeed has scarcely been broached by a long line of students both metrists and psychologists. 2 That the net result of all the careful attention devoted to the subject of verse rhythm has been so inconclusive is due in great measure to one fundamental mistake in method a mistake which has not been recognized till recently, and which is being corrected only slowly. This mistake lies in treating rhythm objectively as an art form rather than as a form of motor expression. By this I do not mean to express a prejudice against all but the so-called "motor theories" of rhythm. And indeed I would by no means slight the study of rhythmic forms as they are perceived. But I do maintain that the study of verse forms would have gotten along much faster, if we had paid more attention to the actual performance of an individual in producing spoken verse, and less to the im- pression made upon a listener. After all it is the nature of the rhythmic action which counts. So long as the listener receives a rhythmic impression it is enough. Beyond that his analysis is at best unscientific and at worst positively misleading. He is open to all sorts of peculiar illusions arising from the nature of the material in hand. It is a well known fact that the estimation of intervals of time is ex- tremely difficult under the best of circumstances. It is peculiarly 1 Poe, E. A., " The Rationale of Verse." 2 For a bibliography of metrists see Gayley and Scott, " An Introduction to the Methods and Materials of Literary Criticism," Chap. 7, 1899; Omond, T. S., "English Metrists," 1903. For psychologists see Wallin, J. E. W., Studies from the Yale Psychol. Laboratory, 9, 1901 ; Stetson, R. H., Psychol. Rev. Mon. Sup., Vol. IV., p. 413, 1903; Triplett and Sanford, Am. Jour, of Psych., 12, 1901; Miner, J. B., Psychol. Rev. Mon. Sup., No. 21, 1903; Minor, J., " Neuhochdeutsch Metrik," 1894. INTRODUCTION 3 hard to equate intervals filled with different sensational content- lights and sounds, or sounds of different intensity or quality. But in verse or poetry the conditions are still more difficult, for not only are the intervals to be compared filled with a content which is constantly changing in sensational quality the endless variety of sounds which make up the words and phrases but it is also dis- turbed by the intellectual and emotional content. Verse can not be heard as a mere succession of sounds making up a rhythm. It also conveys ideas and has meaning. And the meaning of it can not fail to interfere with any attempt to form an estimate of the duration or intensity of the mere sounds. A comparison of the various ideas which various individuals have held will suffice to justify this indictment of conclusions reached by the unaided ear in the analysis of the sounds of verse. And even the most superficial study of the verses which I have analyzed will convince anyone that the time values of words in our verse are far different from what he himself would be led by his ear to expect. Illusions due to the variety of sensational quality of the sounds themselves and graver illusions arising from the fact that verse must have meaning to be verse at all can only be avoided by putting a mechanism in place of the impressionable listener to analyze the sounds. Such mechanisms 1 are available. They are designed to present in the form of a curve or otherwise the actual sounds produced in speaking. Such a record is of course one- sided in that it deliberately excludes all the sense content of speech. But such a one-sided treatment is what scientific procedure demands. To study a phenomenon in its entirety may occasionally yield good results but as a rule analysis must proceed first by exclusion. The problem must be made specific before it can fairly be attacked. Speech or verse rhythms in their true character present a multiplicity of difficulties like the proverbial bundle of arrows. To reduce them all at once is a task far beyond the power of the present writer; this enquiry, therefore, is concerned only with the sounds which reach the ear as one listens to verse which is spoken. No account can be taken, then, of all the wealth of feeling and idea which such sounds express and convey. No matter how much we may regret such a forcible divorce of the sense from the matter no other course is open. Doubtless such a separation may seem to some readers useless as well as arbitrary. And so far as the arts of versification and poetry are concerned it is useless. Poets will continue to write good or bad verse according to their gifts as heretofore; no formulae can tell them how to practise their art nor even serve as criteria of success or failure for them. But for the psychologist such an arbitrary analysis as I propose 4 TIME IN ENGLISH VERSE RHYTHM is by no means useless. Rhythm is one of the commonest psy- chological phenomena. Hardly an act is performed by us either mentally or physically into which rhythm does not enter as one of the obvious factors. Yet we know little of the nature of rhythm. In fact we can scarcely define it coherently. There seems to be about it an element of regularity and a recurrence in time; but wherein the regularity consists we can not definitely say nor can we specify what the conditions are under which recurrence in time takes on the additional characteristics of a rhythm. In verse. however, we have rhythm exemplified about this there can be no doubt and there is every indication that an analysis of verse will reveal certain of the essential characteristics of all rhythm. And even if such positive characteristics do not appear the analysis will serve to exclude from further consideration certain features gen- erally attributed to rhythmic phenomena but which are not found in the actual analysis. For in this case as in many others negative results are of positive value. The beginning of definition is in elimination and there are abundant indications that the definition of rhythm will profit by eliminating some of the qualifications particularly in regard to regularity of recurrence which have attached themselves to it. The fact that careful analysis of verse admittedly one of the earliest and commonest forms of rhythmic expression does not reveal certain features claimed to be essential to rhythm will prove that they are not essential. It is thus that the psychologist is justified in mutilating good verse in order, not so much to find out how it is made, but to get rid of false notions about its rhythmic structure and about the necessary conditions of rhythm in general. It is sometimes claimed that verse rhythm does not reside in the sound of the words of the verse, but in their logical and emotional content. If this were so no justification could be found for analyz- ing the sounds apart from their sense. But common observation gives the lie to such an assumption. Verse in an unknown tongue is abundantly rhythmical sometimes more so than the verse we can understand, so we are even led occasionally to the false infer- ence that foreigners chant or sing-song their verse. Moreover some of our own verse, nursery rimes as well as some very ornate poetry, is practically devoid of meaning, but maintains a strong rhythm. Many other considerations might be brought forward to substantiate the impression that the mere sounds in verse are rhythmical. But the fact is too obvious. The psychologist is justified in undertaking an analysis of spoken verse rhythms with the reasonable expectation of obtaining from it information about the nature of rhythm in general. CHAPTER I METHODS OP STUDYING VERSE RHYTHM (a) Simple Observation ~by Ear The most natural way to study speech sounds is to listen at- tentively. For that no apparatus is needed and no experimental technique required. The method is still simpler if one listen to one's own voice, for then the ear is aided by the various sensations and feelings which arise in consequence of exercising the vocal organs. 1 But self-observation even when quite satisfactory to the individual is not to be trusted scientifically and can not be accepted when any other method is at hand. Direct observation of other persons' performances is, however, involved in this case with many of the same difficulties of prejudice and preoccupied attention which beset self-observation. It is hard to form a judgment about what we hear spoken, wholly on the basis of what we hear. The observer is very apt to be confused by what he would himself say, or even by what he has said or is at the very time saying under his breath as he listens to a rhythmic utterance. The scientific objections to what is here called the observational as contrasted with the experimental method can be divided into psychological and mathematical. These objections apply to the qualitative and quantitative aspects of the results respectively. Qualitatively the psychologist should look with suspicion on pro- fessed facts of observation which can not be rehearsed, that is, re- peated at will. Such an observation is valid, in general, only in so far as it asserts that there was a phenomenon. Analysis and quali- tative determination of the phenomenon should not be attempted unless the phenomenon can be recalled and the analysis verified. Taking our own subject verse rhythm as an example, the reality of this difficulty is beyond doubt. Let us suppose that I sit listen- ing to the recitation of Kingsley's " Three Fishers." When the 1 That the sensations from the speech organs themselves are not, however, alone sufficient to give a clear impression of rhythm can be discovered by any one who will try to sing or recite verse in the presence of a great roar of sound or better still when there is a strong quick echo. Under such circum- stances it is extremely difficult to maintain regular speech rhythms. When, however, the ear is functioning normally there can be no doubt that sensations attending the utterance of the sounds are of great importance in, the total rhythmic effect. 5 6 TIME IN ENGLISH VERSE RHYTHM recitation is done I make the following memoranda: (1) The effect was rhythmical and pleasing. (2) The rhythm was anapaestic, i. e., 11 rising. " (3)1 feel that rhythmic effect is based upon the rather regular recurrence of strong accents or beats followed by unaccented syllables. (4) I find the rhythm somewhat broken. Now so far as observation (1) is concerned there can be no dif- ficulty. That is the statement of the presence of a phenomenon. But the other observations are attempts to analyze the situation, and, while they have a certain value, the psychologist knows that they are subject to contradiction both by other observers and by the same observer under slightly different circumstances. Thus if I listen to the same person reading the same verses another day, (the reading being as nearly as possible the same as before), I may very likely say the meter is dactylic, very regular, and based on groups of sounds of about equal time value. Such contradictory statements arise, of course, out of the fact that observers are subject to a vast complex of emotional states and intellectual prejudices which make their uncontrolled observations extremely unreliable. Thus I hear the rhythm as dactylic when the verse affects me so as to stimulate me; I hear it as anapa3stic when it depresses me. I notice and emphasize the stress or time factors according to my interest or prejudice in favor of one or another theory of versification. The rhythm is smooth or rough to me according to the degree, perhaps, in which I enter into the swing of the verse and make it my own; or perhaps the roughness is only the result of poor muscle tonus which prevents a good motor re- sponse. In any case such observations are not to be depended on because we have no assurance that they can ever be got again. In so far as unaided observation pretends to arrive at quantita- tive results its pretensions must fail because of lack of measurement. Scientific observations are valid quantitatively only so far as the accuracy of the measurement can be determined. My mere observa- tion of the length of time a word takes in utterance or of the rela- tive durations of the syllables composing the word gives no measure that can be relied upon. So far then as rhythm in verse depends upon quantitative factors (time and stress), uncontrolled observation gives no real means of determining those factors. Such a method must ulti- mately give way to one offering conditions of experimental control. Meanwhile it is often the only method at hand, and weak or fallaci- ous as its determinations of fact often are it is not to be despised utterly. Mere observation and emotional reaction is the favorite procedure of the poet, and the study of verse is so close to poetry METHODS OF STUDYING VERSE RHYTHM 7 that it is not surprising to find this simple method persisting long after more exact methods have made their appearance. (b) Phonographic Repetition An advance is made beyond uncontrolled observations like those of Sweet 2 or Lanier 3 by using the phonograph record as repeated by the machine. 4 I am not sure that the increased accessibility of the record (due to the fact that it can be repeated at will) any more than compensates for the inevitable distortion of the mechanical re- production. In this respect there is no evident advantage over a fre- quent repetition of the same verses by the reciter. The process of reducing the speed of the repeated record, while it affords time for more accurate reactions, can not make the analysis any more reliable. This method is open to the same objection that must be made to mere listening observation in any form. The disposition and rhythmical tendencies of the listener are not measurable and too much trust is put in the testimony of the ear. Even if more than one listener records results, though this excludes individual distor- tions, it can not give anything like an objectively valid measure- ment of the sounds uttered. The main fact so often observed and carefully examined that we construct rhythms in series which are objectively uniform and that our estimates of intensity and of time duration are beset with illusions, ought to make us mistrust all observations where such possibilities of illusion enter in. (c) Tapping There is another method of recording speech stress which is closely allied to the graphic method in technique but psychologically quite unrelated to it. It may be called the semi-graphic, or better, the tapping method. It permits of the use of a drum and time record and gives a permanent and indisputable account of certain movements. In this respect it is better than bare listening. But it does not record speech rhythms. Taps or other voluntary muscular movements which are sup- posed to fall upon the arsis of a verse foot or upon any designated accent are marked on the drum by direct lever, by pneumatic trans- mission, or by electric connection. 5 It may be true as Briicke 6 claims 3 Sweet, H., " History of English Sounds," 1888. 3 Lanier, S., "The Science of English Verse," 1880. *Wallin, Yale Studies, 9, 1904; Scripture, E. W., "Elements of Experi- mental Phonetics," 1902. 5 Either a signal magnet or the well-known " spark method " may be used. "Die physiologische Grundlagen der neuhochdeutschen Verkunst," 1871. 8 TIME IN ENGLISH VERSE RHYTHM that such a movement is coordinate with the vocal stress and not subordinate to it. But his attempt at finding a constant error which might be eliminated was doomed to failure as has been abundantly shown by subsequent experiments in which this method was com- bined with the graphic under the most favorable conditions. Miyake 7 found that the spark set off by a tap movement might be located at widely different points in the voice curve under different conditions. Experiments of my own both by the spark method and with a pneumatic device for recording the taps show the same irregularities. No constant error was observable, the taps preced- ing or following the syllable with which they belonged. This is to be expected with motor functions so little connected as those of speech and hand movements. In part the disparity found between a tapped and a voice record is accounted for by the fact that we are in the habit of tapping out rhythms with the hand or by nodding the head, etc., without regard to other rhythmic impressions. Most people have such motor rhythm habits well developed, as is abundantly testified to by the common observation of people as they ''drum with their fingers" abstractedly. Such a habit being well established it be- comes difficult to institute a controlled rhythmic movement of the same muscles, for they tend to relapse into the habitual movement, The tapping rhythm under such circumstances is likely to go on without respect to the voice rhythm. At any rate it is not to be expected that it will adequately represent the latter. On the other hand the claim is justly made 8 that the tapped rhythm, which is remarkably regular and probably very nearly automatic, exercises a considerable control over vocal expression. The tapping thus becomes an objective control like an instrumental accompaniment and the proper voice rhythm is made unduly regular in obedience to this control, thereby losing its peculiar character- istics. Whether it is possible for a practised experimenter to avoid this difficulty by careful attention need not be discussed here. There is certainly no way of determining whether the internal coordina- tion is regular in the first place, and still less is there any objective measure of the control exercised over the voice rhythm by the tapping movements. When the object, however, is to examine the mechanical scansion of verses 9 the tapping method can not be objected to since extreme 7 Miyake, Studies from the Yale Psycliol. Laboratory, 10, 1902. 8 Wallin, op. cit. Hurst and McKay, University of Toronto Studies. Psycliol. Series, 1, 1900. METHODS OF STUDYING VERSE RHYTHM 9 regularity is a desideratum in mechanical scansion and beat- ing time contributes directly to this end. I must confess, however, that I fail to see what is to be gained for the study of verse by an examination of strictly scanned rhythms. They do not represent any normal rhythmical phenomenon of speech. On the other hand they do not have the characteristics of freely tapped rhythms. The latter are highly significant in the study of rhythms and nothing which has been said above is intended as derogatory of the study of them. 10 Such pure motor rhythms, devoid as they are of ideatonal content, come nearer perhaps than any other one set of phenomena to giving a reliable instance of pure rhythm. 11 A rhythm, however, which is produced by a set of movements not directly concerned in the production of speech sounds has cor- respondingly no direct connection with the rhythm of those sounds. The two things are irrelevant to each other. And desirable as it is to know all about the different forms of rhythmical expression, those of the hand as well as those of the voice, nothing can be gained and all is risked by assuming that the one can be substituted for the other. (d) Reaction Methods Taps may be made by an observer in response to accents which he hears in a series of sounds. When such a procedure is adopted all the previous difficulties are complicated by the addition of those peculiar to the reaction time experiment. The most serious ob- jection to the application of the reaction method in the case of a series of rhythmical sounds is that expectation enters in an extreme degree; and there is no way of discovering whether the observer is really reacting to what accents he hears or merely recording the beat of a rhythmical series of his own. 12 When the sounds vary considerably in intensity and energy as they do in speech there is no way of ascertaining the effect of those variations upon the reaction time. The expectation is that the reaction time will be shorter for the more intense sounds 13 but no very close correction can be made with the data at hand in such an experiment. The occurrence of the sounds being quite irregular and each preceding sound serving as a signal for the succeeding "McDougall, R., Psychol. Rev. Hon. Sup., Vol. IV. 11 Compare Meumann's discussion of the effect of content or meaning upon rhythms, Philos. Studien, 10, 393, 1894. 13 Compare, for instance, the anticipatory reactions in the ordinary " com- plication experiment." 13 Froeberg, "The Relation of the Magnitude of Stimulus to the Time of Reaction," these Archives, No. 8, 1907. 10 TIME IN ENGLISH VERSE RHYTHM one complicates the situation very seriously and inevitably leads to great variability in the reaction times. In point of fact this method when applied to actual verses is only a better form of the method of observation with which this kind of study began. In that it supplies an accurate measure of the variability of the observations it has certain advantages. 14 But what it gains in mathematical accuracy it loses in psychological reliability through the interference of the observer's own rhythmical tendencies as an indeterminate factor. 15 (e) Determination of Subjective Rhythms If, however, the object of an experimenter is to ascertain these very rhythmical tendencies of his subject he is abundantly justified in the use of a reaction method. 16 It should be understood, however, at the outset that the determination of the subjective tendencies of a subject to rhythmize certain auditory or other sensations has little or nothing to do with a determination of that subject's rhythmic utterances or movements. This point has been too often lost sight of. Psychologically it makes all the difference whether I hear the supposedly similar ticks of a clock and group them into some such rhythmical order as tick-tock, tick-took, or whether on the other hand, I amuse myself by repeating the words tick, tock in a rhyth- mical series. In the one case there is a very complicated perceptive or affective situation (psychologists differ in their analysis of it) where the observer groups certain discrete impressions into fresh units supplied by himself while he is all the while subject to illu- sions of time and accent which make him feel that the ticks he hears are unequal in force and separated by unequal intervals of time. That is the one case; perceptive or affective, or both, and difficult of analysis. In the other case I simply make certain movements which some one else or I myself call rhythmical. In this case all we need to do is to measure those movements. We do not have to enquire further how and why the movements give rise to a rhythmical im- pression. It is sufficient that they do so. To refuse to recognize the distinction here drawn is to ignore the division between de- "McDougall, Psychol. Rev. Hon. Sup., Vol. IV., 393, 1903. "There is a distinct difference between this method where the sound is heard and reacted to and the tapping method, where two movements are coordinated. 16 Bolton, T. L., Amer. Jour, of Psychol., 6, 1893; Miner, Psychol. Rev. Mon. Sup., No. 21, 1903; Ebhardt, K., Ztschr. f. Psychol, 18, 1898; Biicker, Abh. d. Sachs. Ges. d. Wiss., 17, 1896. METHODS OF STUDYING VERSE RHYTHM H ductive and analytic methods and those of induction. Any single analysis of a verse of poetry as regards its rhythmic constituents involves a working definition of rhythm in general and of verse rhythms in particular. One can not analyze these things without knowing what to look for. The inductive procedure on the contrary is perfectly direct and requires no presuppositions whatever. If the sounds of a verse are admitted to be rhythmical nothing remains but the comparatively simple description of what the verse is made of acoustically. Ad- mittedly no inductive procedure ever gives final results and no one denies the necessity of an analysis and generalization from the as- certained facts. Butffar too much effort has been wasted on vain attempts to analyze verse without any facts at all or with the most loose and inaccurate observations. Such analyses are bound to fail, and it is no wonder that the theory of versification is burdened with a mass of contradictory and ill-grounded generalization. For the most part such generalizations rest on individual observation and more or less chimerical theories of the nature of rhythm and time. The accurate records of concrete instances of rhythmical verses are so few as to be utterly lost in the verbiage of the philologists and metristsJ (/) Experimental Method The method of the experimental psychologist with its peculiar excellencies also has its own defects like all the other methods. The psychologist may indeed feel envious at times of the rapidity and seeming success of methods more systematic and theoretical than his. The experimental method at best seems to be all too slow and when one must carry it to the still further degree of considering the human elements in the obscure phenomena the restraint becomes onerous. None the less the demands of scientific exactness must be met sooner or later. A few well sifted theories may endure for a long time but all at last will be brought to the bar of human experience and made to give an account of themselves. In so far as the method of psychology is first of all that of determining the characteristics of any given particular human experience it is to psychology and psychologists that the task falls of examining and criticizing those theories which pretend to formulate any given set of experiences or aspect of an experience. 17 17 Experiments which have to do with the esthetic attitude of the subject toward a rhythmical series; with his appreciation of rhythm; his ability to discover rhythm in a regular or irregular series; or with his reactions to sup- posedly rhythmical situations fall without the scope of a study of the actual rhythmical series. Such work will be referred to in these pages only as it has suggested and led up to more strictly empirical methods. 12 TIME IN ENGLISH VERSE RHYTHM The experience of rhythm is one of those which can profitably be made the subject of theoretical and schematic discussion up to a certain point. Beyond that point nothing can take the place of a sober consideration of the facts experienced and the resulting ex- perience. The judgment of an observer is final upon the question of the existence or non-existence of a rhythm, just as it is final on the existence of any other phenomena of consciousness. Whatever the definition of rhythm, the occurrence of the state of mind can not be doubted. Others may claim that there are rhythms which go on unperceived by the particular observer and not affecting him. And from some point of view that may be true. But the psychologist is only concerned with a rhythm that is felt. Here as always it is the duty of the psychologist to take account of the positive cases irrespective of the number of negative cases. The latter may be made use of under particular conditions of experi- mentation but most of them are without significance. For a man who has had one vivid hallucinatory experience, the absence of them during the rest of his life is irrelevant; and the psychologist above all, is concerned with the particular conditions of that par- ticular experience. When difference tones are heard with two notes of a certain interval or when colored after-images are present for some time after looking at a bright white light it is the posi- tive phenomena which are made the object of investigation. It matters very little that some individuals can never hear such dif- ference tones or that most of us live through similar conditions with respect to bright lights a hundred times without being conscious of the after-images which a little attention could make vivid. The psychologist can do nothing with difference tones and after-images that are not heard and seen. His business has to do with the phe- nomena that are heard and seen. The judgment of an observer is the one indispensable requisite for the pursuit of the psychological method. So it is with the case of rhythm. Its presence must be recog- nized before it has psychological existence. And, on the other hand, its existence can not be denied when it is present according to some one's deliberate judgment. An observer's judgment must receive just as much consideration no matter how divergent it may be from the consensus of opinion or from established doctrine. In questions of esthetic appreciation the difference is most strik- ing between the ordinary unscientific method and the method of psychology. The common distinction of ' ' good ' ' and ' ' bad ' ' taste in matters of art stands in direct opposition to the eminently more psy- chological dictum, de gustibus non disputandum. The conditions METHODS OF STUDYING VERSE RHYTHM 13 under which esthetic feelings occur can be defined more or less accurately, but this can be done only by observing actual cases of their occurrence ; never by laying down rules about what ought and what ought not to be appreciated. Such rules rest on a considera- tion of the subject-matter. The only valid conclusions rest on the description of phenomena experienced. Rhythm and verse are so closely related to art forms that their theory has been exposed to much the same abuses as the theory of the corresponding art products. When poetry is defined a priori it is easy to go on to a definition of verse in the same way. In fact one definition demands the other generally, as a prop. Verse and verse rhythm, however, can not be defined a priori. Such being the case, the psychologist is bound to seek out the actual facts, and build up from them an empirical description of the phenomena. Experimentation is the method which the scien- tifically minded psychologist adopts to secure observations as free as possible from error. Fortunately it is no longer necessary to defend experimentation as a method of approaching many of the questions which were made the subject of uncontrolled observation in the past. It is sufficient to say that we employ experiments be- cause only by so doing is it possible to secure measurements under conditions permitting of repetition at will. The extension of this method to the study of verse is justified, therefore, by the failure of any other method to maintain those two requisites of scientific exactness. CHAPTER II GRAPHIC METHOD AND MECHANICAL AIDS ONCE granted that there is profit in experimentally analyzing verse with regard to its sounds apart from its sense, the question arises how mechanical means can be adapted to the purpose, thereby eliminating the illusions and errors of observation to which unaided human ears are liable. Attempts to record the voice automatically in such a way that the record can be subsequently studied at leisure have been numberless. Many have failed, others have had a certain measure of success. But in so far as they are all attempts to trans- late the temporal sound series into a visible record in space they fall under what has come to be known as the " graphic method." The name ' ' graphic method ' ' is applied to two very different things. So far as it refers to graphic methods of presenting quantitative facts in the form of plots and various kinds of graphs it does not concern us at the moment. More strictly the term is applied to any kind of direct registration upon paper or a similar surface by a stylus or pen. The facts which are registered in this way include changes of temperature; time series; the velocity and force of movement both of bodies and muscles; the pressure of gases and liquids; wave movements in various media; and a great number of physiological cycles such as respiration, pulse, etc. Under registra- tion of waves of course comes sound registration. Another very interesting field is that of stroboscopic photography. Ordinary photography would not be classed as a graphic method, but it is often of great assistance as a secondary aid in the application of the method proper. No mere record of a static fact is included in the scope of the graphic method. It applies only to the registra- tion of changes or movements of some kind. The earliest application was in the registration of meteorological data, and in that field its successful employment has gradually spread until at the present time almost all scientific measurements of these facts are made automatically. In the Weather Bureau stations for instance measurements of temperature, barometric pres- sure, wind velocity and humidity are seldom or never read directly from the instruments. In fact many of the instruments do not permit of direct reading. Nevertheless the measurements obtained are even more accurate than the most careful direct reading. One of the most interesting examples of this type of graphic record is 14 GRAPHIC METHOD AND MECHANICAL AIDS 15 the seismograph in its different forms, serving to record the ampli- tude and directions of tremors in the earth's crust. The graphic method was applied to meteorology before the middle of the eighteenth century, particularly in the form of curves for temperature and wind pressure recorded on a continuous roll of paper (le Marquis d'Ous-en-Bray 1734), or on smoked paper (Rutherford 1774). Next to the invention of the method itself in importance stands the control of it by the introduction of a time record on the same sheet. When the revolution of the clock or other mechanism com- monly used is perfectly regular the time can be read off from the space between any two marks of the pen upon the paper; but it is generally safer to make use of a distinct time record writing a line along beside the others which are being made. Without such a time record the apparatus is, technically speaking "uncontrolled." For this indispensible contribution we are indebted to Thos. Young who employed the oscillations of a simple vibrating reed. Duhamel replaced this by a tuning fork, and later H ! elmholtz, Regnault and Foucault introduced the electrically controlled tuning fork; and finally Marcel Deprez developed the "signal magnet," which has become one of the indispensable accessories of the modern laboratory. James Watt used the graphic method to record steam pressure and its employment in the physical sciences has steadily extended in scope. The Ludwig kymograph (not the drum alone) was invented in 1847. The principle was applied by Volkmann, Helmholtz, Vierordt, and others in Germany, to the registration of the heart beat, respiration, and muscle action. In 1857 Vierordt and Tubin- gen announced the sphygmograph, and it was this that interested the Frenchman, Marey in the possibilities of such registration. To him we owe much of the later development of precise methods not only in application to physiological problems but to nearly every field of experimental investigation. Thanks to his ceaseless devo- tion the method is now available not only for the recording and presenting of facts in a compact and convincing form but for the recording of many facts too minute or too ephemeral for direct observation. In Marey 's hands a registering instrument is trans- formed into an instrument of precision, an indispensable tool. In no one field has the graphic method afforded such diverse applications as in the recording of sound. The manometric flame makes it possible to analyze all the most complex compound musical notes and on the other hand the phonograph makes it possible to reproduce sounds of all sorts with extreme accuracy. The one method makes possible an accurate science of acoustics; the other 16 TIME IN ENGLISH VERSE RHYTHM not only seems to give pleasure to great numbers of people but is a tremendous aid in the study of phonetics. Neither of the methods, however, is strictly in line with the classical form of the graphic method. Scripture 1 and others 2 have used a curve taken from the recording part of the phonograph and have studied this as a record of the sounds which can be reproduced from it. Such a record whether taken on a revolving disk as in the gramophone or cut out from the wax surface of a cylinder as in the Edison phonograph is a genuine graphic record. Such records can be studied, however, only with the microscope or after great expense of time in enlarg- ing according to the method proposed by Scripture and to some extent carried out by him. Such work is more valuable for detail of sounds than for the longer stretches of connected discourse. As early as 1$56 an attempt was made to construct a recording device for speech imitating the arrangement of the ear. This is the artificial ear or phonautograph of Scott. 3 Such an instrument con- sists essentially of a receiving horn and a diaphragm of some sort which actuates a lever writing on some surface. Various develop- ments along this line are chronicled by Scripture in his "Elements of Experimental Phonetics." 4 The friction of the lever attach- ment can be got rid of by substituting an electrical contact as in the "voice key" of Cattell 5 and Scripture. 6 A more elaborate form of the same device is found in Rousselot 's electric recorder. 7 All the instruments of this class show the same limitations that were spoken of in the case of the manometric flame and photo- graphic methods; namely, that the record is limited to only com- paratively pure sounds or notes. The more complex or obscure sounds are not at all adequately represented. Increased perfec- tion of the mechanism gives more and more detail in regard to the shape and amplitude of the sound waves, but the piling up of such an amount of detail soon makes it quite impossible to interpret such records with profit. The phonograph record is the only one of all the devices so far mentioned which is adapted to record long stretches of continuous speech and the interpretation of its record as Scripture himself admits 8 is such a problem in itself that there 1 " Elements of Experimental Phonetics," also Yale Studies, 10, 1902. 2 Marichelle, " La parole d'aprfcs le trace" plionographique " ; Lloyd, Jour. Anat. and PhysioL, 31, 1897; Marage, L'Annfa psychologique, 5, 1899. 3 Inscription automatique des sons de Pair au moyen d'une oreille artificielle. * See also Yale Studies, 7, 1899. Phil. Studien, 3, 1886. 8 Yale Studies, 3, 1895. 7 " Principes de la phone"tique expe'rimentale," 1902. 8 " Researches in Experimental Phonetics," 1906, p. 4. GRAPHIC METHOD AND MECHANICAL AIDS 17 is little hope of ever accumulating any quantity of experimental data in regard to verse in this way. The disciples of Marey and particularly the investigators under Rousselot in his laboratory of the College de France have come nearer than any one else to solving the difficulties of obtaining a legible record of speech sounds. Their methods include the record- ing of lip and tongue and palate movements by direct levers writing on a revolving drum or by electrical keys connecting with such levers. More adequate are the various devices for recording the vibrations arising from the larynx, either by a pneumatic bulb placed against the throat and connecting with a Marey tambour writing upon a drum, or by a device placed in the same way but equipped with a mechanism for interrupting an electric current.^. Most successful of all, however, is the device now in general use in slightly different forms in many laboratories by which the breath as well as the sounds of the speaker's voice are conducted directly to a tambour and there recorded on the drum. The apparatus which I have used for the analysis of verse does not differ in any essential from that described independently by Rousselot and Meyer. 10 For the particular purpose of noting rhythmic time values less delicate records are required than for the study of strictly phonetic questions. Accordingly the apparatus is made as simple as possible and the rate of revolution of the drum reduced to about 2 cm. per second so that the labor of measuring the time is reduced to a minimum; the error of measurement is still less than one one-hundredth second for each syllable measured. In my experiments on nonsense syllables the time line was given by a clock beating seconds; the records were then measured (under a lens of course) in linear units and the proper corrections made for the speed of the drum during any given second. In the later experiments with actual verse a vibrating reed was employed to interrupt the electric current through a time marker every tenth of a second. The rate of the reed was kept constant and frequently tested by means of an electric counter inserted in the circuit. The rate of the reed was never so much as one tenth second fast or slow per minute; that is, it did not gain or lose as much as one count in six hundred. The regularity of the time line thus obtained was equal as nearly as could be measured to that of a 100 vs. Koenig tuning fork writing a parallel record. The essential part of the apparatus is the tambour. It should 9 For a fuller account of these devices see Marey, " La methode graphique," 1878, p. 309; Rousselot, La parole, N. S., 1, 1, 401, 769, 1899. 19 Die neueren Sprachen, 6, 123, 1899. 18 TIME IN ENGLISH VERSE RHYTHM be as small and light as possible to insure the greatest liveliness and responsiveness. When sufficient vivacity is obtained not so much attention need be paid to the accuracy with which the details of wave forms are recorded. The principal result to be arrived at is a quick response to the first breath and to the first vowel vibra- tion and a comparatively quick return to rest when the breath and sound cease. If the initial and final points of a sound are well defined further details can be dispensed with so far as the rhythm is concerned. In the interest of lively action I have found it best to use a very small capsule for the tambour (not over f-inch in diameter and ^-inch in depth) covered by a rubber membrane of fairly light weight pretty well stretched. The writing lever is of split bamboo about five inches long, one and one half inches being in the short arm. It rests in a delicate cone bearing, and is at- tached firmly by sealing wax to a small rest (this can be made of paper) gummed to the rubber membrane. A gelatine writing tip can be used but there is less friction if the bamboo is whittled to a very fine point and no other tip added. A certain loss in delicacy of tracing is more than made up for by relieving the lever of any unnecessary weight and at the same time increasing the smooth- ness of the tip. The tube for admitting the air to the tambour can as well enter at the side as the base of the drum, and when particular attention is to be paid to vowel vibrations I have found it worth while to replace the solid metal base of the drum by a paper or mica diaphragm. This seems to act as a resonator opposite the rub- ber membrane and reinforces the vibrations of the latter. The vibrations of such a diaphragm are not, however, of sufficient ampli- tude to make an independent record. Unfortunately our knowledge of tambours is only empirical and of several which appear very much alike perhaps only one or two will give good records of sound vibrations. Fortunately, however, tambours of various sizes and proportions can be constructed and set up very easily. It is best to have one or two carefully con- structed levers with a fine cone bearing arranged on a sleeve which can be slipped onto any tambour stem. It is also convenient to be able to slip off the bearing and lever when replacing the rubber mem- brane. The latter can then be adjusted without fear of injuring the delicate lever and bearing. A fine insulated copper wire is much more convenient than the usual thread for fastening the membrane. The tube connecting mouthpiece and tambour should be as short as the convenient placing of the speaker will permit. A few inches more or less does not, however, interfere appreciably with the good GRAPHIC METHOD AND MECHANICAL AIDS 19 action of the recording apparatus. I use a tube of rather stiff rubber about two feet long and of J-inch bore. When one subject is to serve for any number of readings it is well to make a plaster cast of the lower part of his face, hollow it out around the lips, and use it as a mouthpiece. Such a mask requires a minimum of attention from the reader during the experi- ment. He simply holds it easily against his face with his left hand. The mask, however, is by no means necessary. Precisely the same results can be had by using a rubber bell of about three inches diameter such as can be obtained in various sizes in hardware stores for the use of plumbers. The only difficulty here is that the bell must be pressed by the hand into the right shape to fit the face ; most readers require a little practise before they learn to handle it just right. This apparatus is so simple, so easy to construct and use that one is surprised that it has not come into more general use in the study of various speech curves. It is as accurate as is desirable for most phonetic points and its operation requires no elaborate labora- tory equipment nor special technique. The curves themselves are comparatively easy to decipher and the labor involved in this part of the work though great is far less than that of deciphering phonograph records. Any loss in accuracy in the curve as compared with that of the phonograph is more than made up by the saving of time whereby it becomes possible to multiply the number of records and so correct the error of any one. Moreover it is very desirable to have a considerable number of records of the same words for purposes of comparison and the detection of chance variations in the enunciation. No other apparatus yields so many records taken, if desirable, at one sitting and under the same conditions. The interpretation of the curves is as stated above comparatively simple. It is, of course, laborious as in all exact measurement. Practise is required in deciphering the records of each particular speaker, for each one has his characteristic tricks of enunciation which give a distinctive form to the curve. Each consonant and consonant group has a peculiar record due to the way in which the breath is controlled in uttering it. Thus p, t, d, k, show a sharp vertical rise of the lever and a quick recovery, s and cli a gradual rise and irregular fall. The more distinctive of these consonants serve as landmarks in the record, and they must be learned for each speaker. The records can be read off roughly with the naked eye, but for the finer points some magnification is necessary. A watch maker's lens of about three inches focal length is all that is re- quired. 20 TIME IN ENGLISH VERSE RHYTHM Since some sounds make a sharper and more characteristic curve than others it is desirable to choose such phrases for study as con- tain those distinctive sounds at the critical points of the record. Selection of phrases on such a basis does not prejudice the result in any way, and at the same time forestalls any question about the accuracy of the curve as a record of the sounds. When there is any difficulty about interpreting the records the process is facilitated by getting the reader to speak the separate syllables or words involved singly, or in small groups, which can be identified at once and marked in the record. A comparison of the whole record with these fragments will often clear up obscure points in the former. No doubt this simple apparatus is capable of improvement, but at present it is the best there is for the purpose and in most respects it is entirely satisfactory. CHAPTER III PITCH, INTENSITY AND DURATION PHYSICALLY considered speech sounds are characterized by in- tensity, duration and pitch or quality. But not all of these are of equal importance for verse. The present enquiry is concerned with one of these factors only, that of duration. The others can not be left out of account for they enter intimately into the time relations of every rhythm; but the method as well as the purposes which I have outlined prevents a complete consideration of anything but the time relations. /Ijraphic experimentation yields but poor results for instance, in the case of pitch. The pitch of clear vowels when sung can be measured, 1 but the vowels of ordinary speech are so complex that no system of analysis into simple components is adequate.} In fact the work of Scripture 2 seems to prove beyond question that vowel sounds do not consist of a fundamental combined with resonat- ing partials but that they are unstable compounds constantly chang- ing in wave form. If this is so the task of measuring them is rendered doubly difficult. Despite Scripture's 3 elaborate arguments I do not believe that we can trust the gramophone curve as being more accurate than the reproduced sound. Just the opposite may be true. But the phonograph or the gramophone record is the most refined graphic record we have at the present time. In view of all this we are compelled to neglect for the present the pitch and quality aspects of speech sounds. Any complete account of speech rhythm must take them into account. They are perhaps more im- portant for prose rhythm (if there is such a thing) 4 than for verse. The connection between intensity and duration is almost as hard to discover experimentally but it occupies such a large place in the historical theories of verse rhythm that it must be examined more closely. We are confronted here with exceptional difficulties, 5 such that present day physics, phonetics and psychology are helpless before them. Linguistic sounds are myriad, each differing from 1 Cameron, Psychol. Rev. Hon. Sup., No. 34, 1907. 2 " Researches in Experimental Phonetics," p. 146. 8 Ibid., p. 17. * Lipsky, " Rhythm as a Distinguishing Characteristic of Prose Style," these Archives, No. 4, 1907. 6 Bourdon, L'Annte psychologique, 1897, p. 372. 21 22 TIME IN ENGLISH VERSE RHYTHM every other in loudness and quality. If we limit ourselves to the English language alone we find that every individual who uses it has his own private way of modulating, accentuating, and toning it, and so characteristically that we are less likely to confuse two human voices than we are two human faces. Now it is possible to compare the physical intensities of two sounds of the same wave-length (pitch) provided they have the same quality (overtones or partials), but we have no physical measure- ment of the intensities of sound differing in pitch and quality. This being the situation, how hopeless is the task of determining the relative intensity of sounds which differ so widely in quality as, for instance, the sounds of the words it and aim. Moreover in speech we have practically a continuous flow of sound, and the determina- tion of points of maximum intensity must be made with respect to a very small element of the total series, not with respect to large masses of continuous uniform sound. Even if we had a means of comparing "it" and "aim" taken separately and as units little advance would have been made toward the determination of the points of greatest intensity and the comparison of such intensities when found in a sentence like "it is my aim." And even suppos- ing that some way were found of comparing these intensities for a given speaker at a given time there is no way of applying an abso- lute standard by which these accents could be compared with others. There is no such thing as a standard of intensity of sound. If on the other hand we abandon the attempt to obtain objective spatial measurements of the accents in speech and fall back upon our psychological perception of their presence and relative magnitude new difficulties arise to confront us. Let any one say to himself the two words booty and hoot and try to discover the location with any exactness of the greatest stress within either word. Does it lie just as the b bursts into the vowel in booty, or in the long oo sound in either case? If the observation is made while some one else pronounces the words the results will probably be different. This is a hard thing to do. It is still harder to draw a curve representing the rise and fall of the stress up to and down from the point of greatest intensity. All this is supposing that the stress is one of loudness only. But there is small probability that it is so. More likely the accented syllable is also prolonged in time and altered in pitch. In that case we are still quite well able to tell on what "syllable" the accent falls but further than that our ear does not tell much. 6 6 The application of the tapping method (McDougall) offers a possible solution. If a pneumatic system is used for recording the taps it is possible to SPECIMENS OF THE RECORDS. vp <^ CO] PITCH, INTENSITY AND DURATION 23 As has been frequently insisted upon, speech, and particularly rhythmical speech flows almost continuously. Its rhythm is, there- fore, more like that of a singing tea-kettle or humming electric motor, than it is like the rhythm of a metronome. This distinction can not be too much emphasized. The investigation of rhythm has been confined almost exclusively to the latter type the type made up of discrete elements. Perhaps more headway would have been made in the general theory if the type which arises in a continuum had received more attention. A word of caution is needed against inferring very much about verse rhythm, which is of this continuous type, from experiments in tapping and listening to series of discrete sounds. The two things are far from alike. The reaction method of study is found in its least objectionable form when the reaction is made to a qualitatively distinct member of a rhythmic series of discrete stimuli. It does not follow, however, that it is equally applicable in the case of a continuous flow of sound. Miyake 7 found that the stress is often marked by a tap before the beginning of a sound. I have found the same. Briicke, 8 however, found that the taps fell within or near the end of an initial vowel. Meyer 9 concludes that the maximum stress falls, with the tap, just within the initial consonant sound before it passes into the following vowel. He assumes perfect coordination of tap with vocal stress. These contradictory results are easily understood from the point of view of the difficulty we feel in locating our own stress, but that the stress should precede the sound is absurd from the point of view of one who listens to the sounds. The graphic method can give little help on this point. A glance at one of the accompanying records will serve to indicate its short- comings. The vowel sounds are indicated by a low, generally flat, serrated stretch ; certain of the more explosive consonants by a sharp rise and a more or less gradual fall of the curve ; the nasals by nearly flat stretches: sibilants and aspirates by a comparatively slow upward movement, jit may be possible to measure roughly the intensity of vowel sounds by the amplitude of the vibrations as recorded. Some consonants would be measured by the amount of breath expelled; this would not apply to all of the consonants however. In any event no comparison is possible between vowels and consonants as regards record their intensity as well as the interval of time between them. Neverthe- less it is a long jump to the intensity of a sound or speech movement from the intensity of a tap coordinated with it. 1 Studies from the Yale Psychol. Laboratory, 10, 45, 1902. 8 "Die physiol. Grundlage der neuhochdeutschen Verskunst," p. 24. 9 Die neueren Sprachen, 6, 122, 1899. 24 TIME IN ENGLISH VERSE RHYTHM stress nor between different vowels. It is even doubtful whether the same consonants can be compared with each other in view of the different form of the breath curve under varying stress. On the whole I think it sensible frankly to admit the impossibility of any exact analysis of verse with respect to either loudness or intensity in the more psychologically correct connotation which embraces all .the factors that give a stressed syllable weight or im- pressiveness.1 We can determine by ear roughly the syllable stressed by a reader and I have simply made measurements from the begin- ning of one such syllable to the beginning of the next on the assump- tion that the stress wherever it lies is not far from the beginning of the syllable. That this assumption is not scientific can not be denied. The result is that what I shall have to say about the posi- tion of the stress will necessarily lack such precision as is aimed at in the case of the measurement of the durations and proportions of syllables. The graphic method is peculiarly adapted to the determination of time intervals, hence its applicability to the measurement of the duration of speech sounds. In view of the fact that our interest is primarily in the time relations of the verses which are to be studied the results which our method yields in this respect more than com- pensate for its inadequacy in recording pitch and intensity. Scripture 10 has published a few records of complete verses care- fully measured off from enlarged gramophone records. Such records are of great value but the excessive amount of time and work involved in obtaining them prohibits their employment in any great number. Stetson 11 also publishes measurements from one entire stanza of verse. ) Meyer 12 has measured all the English vowel sounds stressed and unstressed, and many of the consonants in words of different lengths and in different positions, initial, medial and terminal, j His records are taken with a modification of the pneu- matic voice recorder of Rousselot, but invented by himself. The instrument is like that used in the present investigation except that the breath is allowed to escape through a gap which is left between two sections of the connecting tube. 13 The records obtained are clear and the time values given for the different letters seem to 10 Yale Studies, 10; "Elements of Exp. Phonetics"; "Researches in Experi- mental Phonetics." 11 Psychol. Rev. Mon. Sup., Vol. IV. 13 Skrifter Kongl. Humanistiska Vetenskaps-Samfundet Upsala, 8, 1903. 13 The advantage of allowing free egress for the air is that the pen recovers very quickly from the sharp rise due to an explosive consonant. This results in a more jagged curve and the apparent advantage is, in my opinion, over- balanced by the loss of characteristic slopes. PITCH, INTENSITY AND DURATION 25 be accurate to at least one one-hundredth of a second. Such ac- curacy is not required for ordinary rhythm work. The importance of having actual measurements of the duration of verse sounds will be realized by any one who has even a super- ficial acquaintance with the long arguments of the metrists on the place of classical or quantitative verse in English. The present investigation was undertaken with the end in view of either justify- ing one or the other of the prominent views of theorists on this subject or else of dismissing all of them from the position of promi- nence which they now occupy in the discussion of verse rhythm. Before taking up the actual results of experiment it will be well to review some of the more important theories on this point of the importance of the factor of duration. OF THE { UNIVERSITY ) OF CHAPTER IV TIME RELATIONS IN THE THEORIES OF THE METRISTS AND POETS THIS is no place to go into details of literary disputes, but the opinions of some of the best writers and critics of verse deserve notice because they are the opinions of exceptional observers; and while their method was faulty their conclusions are often deserving of consideration. Verse, says Pee, is inferior or less capable music. " Verse origi- nates in the human enjoyment of equality," and man derives enjoy- ment from his perception of equality. A hypothetical history of the development of verse would begin with the spondee, " words of two syllables equally accented." This becoming monotonous would lead to the collection ' ' of two or more words formed each of two syllables differently accented (that is to say, short and long) but having the same order in each word in other terms of collecting two or more iambuses or two or more trochees. And here let me pause to assert that more pitiable nonsense has been written on this topic of long and short syllables than on any other subject under the sun. In general a syllable is long or short, just as it is easy or difficult of enunciation. The natural long syllables are those encumbered, the natural short syllables are those unencumbered, with consonants ; all the rest is mere artificiality and jargon." "Of course it is not the vowel that is long but the syllable of which the vowel is a part." "It will be seen that the length of a syllable, depending on the difficulty or facility of its enunciation, must have great variety in various syllables; but for purposes of verse we suppose a long syllable equal to two short ones, and the natural deviations from this relativeness we correct in perusal. The more closely our long syllables approach this relation with our short ones the better, ceteris paribus, will be our verse ; but if the relation does not exist of itself we force it by emphasis, which can, of course make any syllable as long as desired; or by an effort we can pronounce with unnatural brevity a syllable that is not too long. Accented syllables are, of course, always long, but when unencumbered with consonants must be classed among the unnaturally long." "In fine every long syllable must of its own accord occupy in its utterance, or must be made to occupy, precisely the time demanded for two short ones. 26 TIME RELATIONS IN THEORIES OF METRISTS AND POETS 27 The only exception to this rule is found in the caesura." 1 A very full bibliography of the "pitiable nonsense" to which our great American poet refers can now be found in Omond's "English Metrists." Much that has been written on the subject is certainly foolish; yet we find as early as 1776 that Joshua Steele was pro- posing 2 the scansion of verse by musical feet. Poe's proposal amounts practically to the same thing except that he limits his verse to notes of only two lengths. The real difference between this and the classical notion of verse is important ; for the Poe theory, which I shall hereafter call the musical theory, starts out with an assump- tion of equal time values for the feet, that is, for the rhythmic units, in the verse. The theory of verse descended from classical times and which has been fighting always for a place in English versifica- tion, did not go beyond a statement of the ratio said to hold between the long and short syllables in any single foot. Both the classical theory and that of Poe are interesting for us in that they place such great stress on the time element in the rhythm. Neither Poe nor Lanier whose "Science of English Verse" con- tains the ablest and most satisfactory exposition of the music theory of verse, says what he would measure as the length of a foot or syllable. But, as they do not provide for any vacant spaces between their syllables, I presume that they would measure as I have done from the beginning of one syllable to the beginning of the next. This is what we do in music, but in that case there is no gap. As a rule the flow of the voice is practically continuous. In fact it is often difficult to tell where to dismember two syllables and in such cases the experimenter has to rely on his ear as the court of last resort. In some words like "porridge" (Table XIV.) it is quite impossible to divide the syllables and I have had to measure the two syllables together. 3 A very few cases of actually measured verses (see tables) are sufficient to show the artificiality of such a system as that of Lanier. So far as the equality of feet and the equivalence of syllables to musical notes is concerned his theorizing is utterly unwarranted. And in this case we see how easy it is for a serious student and a good poet to make observations about time values which bear no faintest resemblance to the facts. It is seldom that the method of mere observation and of introspection can be so plainly convicted; yet doubtless many of our ordinary judgments are as far from the 1 All from the first few pages of " The Rationale of Verse." 2 " Prosodia Rationalis." 3 Cf. Meyer, Die neueren Sprachen, 6, 479, 1899. 28 TIME IN ENGLISH VERSE RHYTHM truth, in cases where there is no way of securing experimental evidence. One of the more recent exponents of the music theory is Dabney. 4 Verse, he holds, is measured by accent as music is divided into feet of equal length. "The basic principle of verse is time; measure- ments of time ; uniform measurements of time ; which measurements are represented by words." (p. 27). Another elaborate treatise is that of Raymond. 5 Rhythm in verse according to him is "an effect produced by a consecutive series of sounds, or multiples of sounds, which, in themselves, may be varied and complex; but each series of which is of like duration" (p. 53). But for Raymond as well as Dabney accent is fundamental in marking off the measures. Even so their introspection is no better than Lanier's for our figures come very far off from the equalities upon which they insist. Some question may arise at this point about the marking on the graphic record of the point of accent. It is doubtful whether such writers as Raymond mean the beginning of the accented syllable as the point of greatest stress of voice. As has been said already, the latter point can not be satisfactorily determined. On the whole I have decided to consider only the beginning of the syllables. By doing this a great amount of work is avoided work not only labori- ous but of an extremely unreliable sort. There are certain theories as we have seen which require measurements from the beginning of the syllables; if other measurements must be made from some hypothetical "Arsengipfel" the work of measuring would be doubled. But this consideration would not have affected me had it been possible to find any such Arsengipfel or top point of stress. It has not been possible to do so and I am convinced that the begin- ning of the sound is' a safer place to measure from than any other one point. The maximal point is bound to lie very close to it either before or after and the amount of error in taking this point is small compared with the amount of the irregularity of total feet. Other exponents of the equal time theory are Barham 6 and Omond. 7 The latter 's book is particularly sane as regards every- thing but his insistence on the temporal equality of all units. Omond, however, does not insist on a strict equality of formal feet (p. 80), and admits that a syllable may be on the border line between * " The Musical Basis of Verse," 1901. 8 " Rhythm and Harmony in Poetry and Music," 1895. e Philolog. Soc. Trans., 1860-1, part 1, p. 45. 1 " A Study of Metre," 1903. TIME RELATIONS IN THEORIES OF METRISTS AND POETS 29 two feet. This, however, is a license, and only for the sake of variety. John Ruskin 8 asserts that "measured times of utterance are the basis of verse" but also that "stress accent and quality are the same"; and Coventry Patmore 9 maintains likewise that accents are separated by isochronous intervals. "Our present verse" says Sweet 10 "is based mainly on the natural stress of the language, each strong stress marking the begin- ning of a foot (bar). But the stress-groups of ordinary speech amount to nothing more than prose : to make these stress groups into metrical feet it is necessary to have them of equal (or proportionate) length, and in English verse we lengthen or shorten syllables without scruple in order to make the feet Of the requisite length." In Schipper's 11 monumental work a place is made for measured as well as accented verse in English. He does not go to such an extreme as Sweet, however, even in the case of what he holds to be measured rather than accentual verse. Milton 12 speaks of the musical delight in poetry which "consists only in apt numbers, fit quantity of syllables, and the sense variously drawn out from one verse into another." Chas. Kingsley 13 holds that English verse is not regulated by accent but by length of syllable. Gummere 14 while holding (p. 144) that quantity is a secondary factor says that (p. 137) "we do not utterly refuse to recognize it as an element of verse," and that (p. 137) "that poetry which depends for metrical effect on detailed time relations (quantity) will come nearer to music than the poetry which depends chiefly on stress relations (intensity, accent)." 15 In discussing those authorities who maintain that time relations are the fundamental thing in English verse constant reference must be made to others who place the emphasis on accent. The advocates of this other position are rather sharply divided into two classes, those who hold that accent is primary but that the time relations of the accents are significant, and those who hold that accent or accent and number alone form the unifying principle of our verse. The latter theory is the orthodox doctrine of the grammarians 8 " Introduction to English Prosody" (cited by Gayley and Scott). 9 Essay appended to his Poems. 10 " History of English Sounds," 1888, p. 102. 11 " Grundriss der Englischen Metrik," 1895; " Englische Metrik," 1881-8. 12 Preface to " Paradise Lost." 13 " Life and Letters," p. 347. ""Handbook of Poetics for Students of English Verse," 1888. 15 Cf. Gayley and Scott for further bibliography along this line. 30 TIME IN ENGLISH VERSE RHYTHM and of the common people, in spite of the insistence on time propor- tions, which we have seen, by an intellectual minority. The ordinary scansion of our verse rests upon regular alternation of accented and unaccented syllables or of "syllables differing in quantity" (Gould Brown). The emphasis here is on the regularity of the alternation. Most of us have had painful experience of this sort of scansion and no further description is desirable. Much of our best poetry is faultless in regard to such regularity. On the other hand a great deal of verse does not show any ap- preciable attempt at such regularity. It is said that Tennyson read his own verse very musically and with regard to quantity. Such a poet would be inclined to disregard alternation of syllables, since he had another principle of unity upon which to fall back. Pope's verse is ideal for alternating accent and there is a prophecy of the place he was destined to occupy in future discussion in his depreca- tion of too much music in verse as analogous to the practise of attending church for the sake of the music. 16 The commonplace seldom can have a genius for its advocate ; the ordinary scansion is therefore left without any very articulate de- fense. Poe's caricature of it will serve as a starting point in order to make the distinction clear between his own position and that of Gould Brown who seems to have been selected as the scape-goat upon whom to cast the sins of all grammarians. Not only does Poe "deny the necessity of any regularity in the succession of the feet and by consequence the syllables, but disputes the essentiality of any alternation regular or irregular of syllables long or short." Compare this with Sweet 's dictum that ' * the general tendency of the language is to alternate strong and weak stressed syllables as much as possible." Sweet (p. 31), however, agrees with Poe that mere alter- nation does not give rhythm without the additional element of regularity in time. But he considers that (p. 33) "emphasis, length, strong stress, and high pitch are naturally though not necessarily associated. ' ' Ellis 17 finds some forty-five discernible degrees of accent or emphasis springing from various combinations of "force, length, pitch, weight and silence." His position is "that English verse, though based on alternation of force is naturally governed by length and pause, is seldom or never unaccompanied by variety of pitch unknown in prose and is more than all perhaps governed by weight." 18 The last named factor relates to the importance of the 16 " Essay on Criticism," 1. 337 ff . 17 PMlolog. Soc. Trans., 1873. Philolog. Soc. Trans., 1876. TIME RELATIONS IN THEORIES OF METRISTS AND POETS 3] ideas expressed and not to the loudness of the sound. It is not clear, however, whether weight is to be interpreted as a purely psy- chological element in the rhythm or whether its presence can be detected by some other alteration in the speech than those of force, pitch, length and pause. Over against this school of compromisers stands Skeat 19 with the flat statement "that English poetry depends wholly on accent and can only be studied by considering the effects of accent. ' ' For him accent and length are distinct things and their frequent association in the same syllable are incidental matters. That there is no neces- sary connection between accentuation and duration is also held by Gurney, 20 and he holds further that an accented syllable may occupy a smaller part of the time space between ictus and ictus. Inasmuch as the last named authority still clings to the notion that verse is rhythmical in the sense of being "metrical" it follows for him that rhythm is not latent in speech but imposed from with- out (Chap. XIX). That is to say, the nature of the language does not make an even flow of words of the right duration; it is only by constraint that we get the right arrangement. This is an idea of some significance and quite in accord with Darwin's idea that music and metrical forms were developed before language in the history of the race. If, however, the Spencerian doctrine of the primacy of language is accepted, rhythm ought to arise directly in that material and there could be no necessity for ever forcing the material to fit the form. Perhaps a compromiser will say that only some words fit given rhythms and others have to be squeezed or stretched to get them in. For Abbott and Seeley 21 the names of feet denote groups of accented and unaccented syllables without regard to quantity. Ac- cent is equivalent to loud stress. The regularity of a verse depends upon its having the right number of accents, and in this respect Abbott finds most verses regular although it is sometimes necessary to throw the accent upon a weak monosyllable in order to make up the number. In this connection Coleridge's expression 22 with regard to the meter of some of his poems is significant. ' ' The meter of Christabel is not, strictly speaking, irregular though it may seem so for its being founded on a new principle, namely, that of counting in each line the accents, not the syllables. Though the latter may vary from PUlolog. Soc. Trans., 1895-8, p. 485. 20 " The Power of Sound," 1880, p. 429. '"" English Lessons for English People," 1871. 22 Preface to "Christabel," 1816. 32 TIME IN ENGLISH VERSE RHYTHM seven to twelve yet in each line the accents will be found to be only four." This "new principle" though more new in form of ex- pression than in actual practise was of great importance in the subse- quent usage of the English poets of the nineteenth century. For us one of the most instructive things about the statement is not the ''new principle" but the old with which it is contrasted. Nothing is said about regularity of time but it is assumed that regularity gen- erally is found in the alternation of long (accented) and short (un- accented) syllables for that is what is meant by counting the syllables. For Coleridge and his friends the accent was the basis of both the old and the new principles. Nothing else was even thought of. That the ordinary routine scansions will last as long as there are children and common people is the verdict of Mayor. 23 He is "in favor of scanning by feet" and attempts to put the ordinary accented feet of the grammars on a scientific basis; and if he does not succeed in his attempt to be scientific he at least makes a good bold stand in favor of the conventional English system. He thinks that we are in no danger of falling into the error of mistaking our "iambs," etc., for quantitative feet in the classical sense simply because English speaking people have no ear for quantity as dis- tinct from accent. He refuses to recognize (p. 55) any given time for two syllables. On the other hand the number of syllables to a foot is limited to three unless slurring or elision takes place. Gummere 24 maintains that the unit in verse is one accent together with one or more unaccented syllables regularity rests in the num- ber of light syllables. "In the Germanic languages and in nearly all modern poetry accent is made the principle of verse; we weigh our words, we ask how much force, not how much time they require" (p. 137). All this in spite of what has just been quoted above about the importance of time. The conclusion seems to be that equal time intervals are found but their significance is secondary. Schipper accounts for both accented and metrical verse in Eng- lish by the historical development of Anglo-Saxon and Norman and neo-classic forms side by side. The former strain gives the accentual type, the latter the metrical type, and in our contemporary verse we find both types side by side or fused into one verse. The accentual type, however, predominates. Guest 25 is more radical and seems to feel that the intrusion of 23 " Chapters on English Metre," 1886, p. 6. 24 "Handbook of Poetics," p. 169. 25 " A History of English Rhythms," 1838. TIME RELATIONS IN THEORIES OF HETRISTS AND POETS 33 measured verse into the language must not be permitted under any circumstances. He says 26 (p. 108) that it has been "said that our English rhythms are governed by accent; I, moreover, believe this to be the sole principle that regulates them. Most of our modern writers on versification are of a different opinion. The time is oc- casionally of great importance to the beauty of a verse, but never an index of its rhythm." He believes that "the rhythms that depend on accent are independent of quantity in every living language from India westward." There is, however, generally a sharper tone (though this is not necessary either) and also a lengthening of the accented (louder) syllable due to the greater effort of enunciation. Genetically considered all the English verse forms spring from the original "sectional" verse of our ancestors according to Guest. As the sectioning of the line becomes more elaborate, and the arrange- ment of forms more variable we get the different types of modern rhythm; the accent always remains, however, the basis of the rhythmic structure. We get the well-known forms iambic, dactyllic, etc., according as the accent stands in the first, second or third position in the foot. M Ed. Skeat, 1882. CHAPTER V TIME RELATIONS AS TREATED BY THE EXPERIMENTALISTS BRUCKE 1 seems to have been the first to give experimental de- terminations of the Taktgleichkeit of verse. He asserts that "in German iambs and trochees it is not the quantity, not the alterna- tion of long and short syllables, which holds together the rhythm but the repeated rising and falling of the accent at regular inter- vals" (p. 64). It is the interval between accents that really matters. The emphasis is thus kept on time relations but thrown off from the individual syllables on to the foot as a whole. There is a strong suspicion aroused that Briicke scanned his lines. Triplett and San- ford, 2 using a combined tapping and voice record, but relying on the tapping, found great regularity in the intervals between accents in nursery rhymes. The method has already been criticized. It prob- ably accounts for the results. Hurst and McKay 3 got similar re- sults, but in their case large variations were deliberately discarded ; they never read naturally but always scanned. These two considera- tions invalidate the results as far as the present point is concerned. Bolton 4 says, "In order for vocal utterances to form a rhythmic series, they must occur at regular intervals of time which can not exceed or fall much below certain limits. ' ' The different intensities, however, give unity to the series. The above statement has no ex T perimental grounds. Squire 5 (p. 541) says, " Temporalness in its connotation of regular succession is the basal principle of rhythm." However, changes of intensity can also produce a rhythm in a regular time series. The doctrine here seems to be that there must be regularity in time first and that the grouping may be caused by temporal or accentual variations of the regular series. Wallin 6 (p. 32) says that "the question of the quantitative char- acter of poetry or prose, is closed. All speech is quantitative, and the distinction, popularly and confidently posited between quan- titative and non-quantitative verse is grounded upon fallacious as- sumptions." His experiments give no warrant for such a gratu- 1 " Die physiologischen Grundlagen der neuhochdeutsehen Verskunst." 2 American Journal of Psychology, 12, 1901. 3 Toronto Studies, 1. 4 American Journal of Psychology, 6, 157, 1893. 6 American Journal of Psychology, 12, 1901. 8 Yale Studies, 9. 34 TIME RELATIONS AS TREATED BY THE EXPERIMENTALISTS 35 items attack upon a distinction which is founded upon no assump- tions whatever. Of course, no one in his senses ever doubted that speech is quantitative but there has been grave doubt whether quantity in speech gives rise to the rhythm of verse. So far as Wallin 's results show anything at all on this point they indicate that quantity is not responsible for the phenomena of rhythm. "The regularity of the intervals in a given set of records of English poetry, the intervals of which are mostly of the two syllable type, and the scansion of which is rhythmically free was about 3 per cent, higher than the corresponding set of records of English prose, the reading of which is rhythmically free (32 per cent, of variation in this case). When the scansion or reading is natural, artistic and rhythmically free the complex centroid intervals are only slightly more regular in poetry than in prose" (pp. 118, 119). But cer- tainly poetry is more rhythmical than prose. No reasons are given why the changes in quantity within a line, so far as they have any meaning at all and are not simply natural differences in time value of the sounds, should not be regarded as concomitants of the accent rather than as themselves determining the rhythm. The centroid system advocated by Wallin or Scripture is likable and sensible, but. to try to graft it onto the older time or music theory of verse rhythm is equivalent to abandoning it altogether. Wallin himself seems to recognize this when he says (p. 113), "nor are centroid intervals to be conceived as a succession of bars of invariable length, nor as feet of similar length, nor as successive quanta of time conforming to an invariable measure." The centroid system of Wallin and Scripture is the equivalent of the section plan of the metrists and of the system of monopressures, if I understand the latter. 7 The centroid is really an accented, or as some would call it, a weighted syllable. Its marks of differentiation are increased loud- ness, duration and heightened pitch. Wallin assumes that of these duration is the essential one, but he gives no reason for saying so and certainly the more obvious one is the increased loudness. Such a centroid is, in the simplest terms, a part of the sound series marked by one stress maximum. As used by Wallin the analysis of verse into such elements leaves no essential distinction between prose and verse. According to the arrangements printed by Scripture 8 for certain verses of poetry there is nothing left corresponding to a rhythm pattern for the verse. When tables are printed giving only the average number of syllables per centroid, or the average time 7 Cf. Skeat, Philolog. Soc. Trans., 1895-8. Only a few syllables can be pro- nounced on one breath pressure and this group is made the rhythmic unit. 8 " Elements." 36 TIME IN ENGLISH VERSE RHYTHM value of a centroid group, the only possible data for the analysis into rhythmic constituents are obscured. Surely rhythm consists of something more than the mere succession of these groups at more or less regular intervals of time. Are we to say that all rhythms are the same, both prose and verse just this simple succession? Such an answer would be far from satisfactory; the centroid theory must develop a more thorough-going explanation of the phenomena which make one rhythm different from another before it can be con- sidered as anything more than a way of summarizing certain obser- vations which do not reach to the details. Inasmuch, however, as the centroid theory gets beyond the notion that there is no rhythm without extreme regularity of time relations it marks an immense step in advance. Briicke, Bolttin, Squire and McDougall recognize an effect of intensity (or accent in some form) only as a subsidiary factor enter- ing in where there is already a uniform time series. Squire 9 and McDougall 10 even conceive of a rhythm without any accentual or at least without an intensive variety whatever. Meumann considers time and intensity as compensating fac- tors, either of which may take the place of the other. Just what this means for verse rhythm in view of the very regular association of the two in one syllable is hard to understand. There is no experimental evidence that a stressed syllable when short in time is correspondingly more intense. What evidence there is points just the other way. The apparatus adopted in the present investigation is not de- signed to measure intensities nor have I been particularly concerned with the question of their measurement. Even if we had a complete plot showing the absolute and relative intensities of all the sounds of any given series of words or for any given verse, the question of the final basis of verse rhythm would be no nearer solved than it is to-day. We now know that intensities are there and we know their serial order. It is the times by which they are separated which we do not know and it is these that we must be able to describe if we are going to give a full account of the conditions under which the particular rhythmic phenomena of verse arise. The only point with regard to the intensities in verse which it would help us to know about is their sequence in such a phrase as pa ta pa' or pa' ta pa. In such groups there appears to be a certain hierarchy of intensities. Similarly in such a line as papa' papa' papa' papa' such a hierarchy seems to exist, the second accented 9 Amer. Jour, of Psychol., 12, 509, 541, 1901. "Psychol. Rev., 9, 461, 1902. TIME RELATIONS AS TREATED BY THE EXPERIMENTALISTS 37 syllable being stronger or neavier than the first. But these are ques- tions of the detail differences of particular rhythm figures. To answer this is to tell nothing of the relative importance of duration and stress. Such questions will have to be answered if it turns out in the end that stress differences of this minor sort are the only differences that can be found between rhythms of different types or figures; meanwhile such details may be left aside while the other possible differences of the patterns are being examined. CHAPTER VI EXPERIMENTAL RESULTS (a) Procedure THE technique of the experiments has already been indicated in describing the apparatus. The subject was asked to recite the verses naturally and at the same time distinctly. None of the subjects 1 found anything in the conditions of the experiment to prevent doing this. In the case of nonsense syllables, which were always TABLE I THE IAMBIC TETRAMETER, Papa' papa' papa' papa'. At the left the reader and the number of his readings. Under " Pa," the average time value 2 of the first syllable of each foot. Under "V," the mean variation per cent. 3 Under " pa'," the second syllable. Under " foot," the average time of the foot. Under " ratio," the average quotient of the accented divided by the unaccented syllable. W54 Pa V pa' V Foot V Ratio V 1st foot 20 9.6 57 8.0 77 7.1 2.9 8.1 2d foot 20 9.6 58 7.3 78 7.0 2.9 9.1 3d foot 20 8.7 57 7.6 77 7.2 2.9 9.0 4th foot 20 9.7 101 21.2 121 H41 1st foot 25 6.0 52 4.7 77 3.8 2.1 94 2d foot 23 6.7 51 5.1 75 3.1 2.3 9.2 3d foot 24 5.5 49 4.4 73 3.2 2.1 6.7 4th foot 23 5.9 93 6.5 115 F73 1st foot 18 9.0 36 5.0 55 4.4 2.0 9.8 2d foot 16 8.1 36 5.3 52 4.9 2.2 8.5 3d foot 16 6.5 36 6.5 52 6.4 2.3 7.5 4th foot 17 6.3 71 9.5 88 B48 1st foot 20 7.0 53 4.4 73 3.4 2.7 8.5 2d foot 21 6.9 55 4.4 75 4.3 2.7 7.6 3d foot 21 4.9 52 5.0 73 3.5 2.5 7.0 4th foot 22 6.4 127 12.4 149 1 The readers in these experiments were : W, Dr. F. L. Wells, then Lecturer in Barnard College; R, A. E. Rejall, graduate student in psychology; L, D. O. Lyon, graduate student in psychology; H, H. L. Hollingworth, then Assistant in Psychology; F, S. Froeberg, sometime Assistant in Psychology; B, the author. 2 The time is given, in all the tables, in hundredths of a second. 3 In all the tables, the columns headed " V " give the mean variation ex- pressed as a per cent, of the total time. 38 EXPERIMENTAL RESULTS 39 TABLE II THE TROCHAIC TETRAMETER, pa'pa pa'pa pa' pa pa' pa. Under Pa' the first (accented) syllable. Under pa the second (unaccented) syllable. Otherwise the columns correspond to Table I. W46 Pa' V pa V Foot V Eatio V 1st foot 22 5.6 48 6.4 71 4-9 .45 8.4 2d foot 22 5.4 51 6.4 73 J t ,l .44 8.2 3d foot 22 5.5 49 7.0 72 5.4 .46 8.6 4th foot 22 7.3 88 18.3 110 H47 1st foot 34 6.0 33 4-6 67 4-2 1.03 7.1 2d foot 34 4.5 33 4.1 67 3.2 1.02 6.7 3d foot 34 4.8 33 5.7 67 4.// 1.04 7.4 4th foot 35 4.9 99 9.7 133 F40 1st foot 26 5.9 33 5.7 59 3.8 .78 7.8 2d foot 25 6.0 31 6.3 56 4.0 .81 9.4 3d foot 26 5.1 31 5.0 57 2.9 .86 8.6 4th foot 25 6.0 67 9.6 92 B 50 1st foot 28 7.0 32 6\9 60 4.8 .88 9.6 2d foot 28 7.7 32 6.8 60 5.4 .90 9J 3d foot 28 8.1 32 7.2 60 5.1 .87 ^.3 4th foot 28 7.6 102 mtf 130 read as lines of verse, the reading was interrupted after each com- plete revolution of the drum. In other cases the reading was con- tinuous for a longer period. The material consisted of nonsense syllables in tetrameter lines of different rhythms and of specimen bits of well-known English poems, This latter material was selected first on the ground of its familiarity and second on the ground of the possession of consonant combinations which could be most readily read from the record. No attempt was made to cover the entire field of English rhythm; in fact the verses chosen are almost all tetrameters. It is well to have them as comparable as possible with each other and the tetrameter is the most frequent meter in our non- heroic verse. Having determined the beginnings of all the words and syllables in any record the durations of these were measured from the begin- ning of one to the beginning of the next. The validity of this pro- cedure has been discussed above. Even if the record permitted of an exact determination of the end of any sound, which it seldom does, I am convinced that we ought still to count any gap which may occur between two sounds as being a prolongation of the pre- ceding sound. Such a gap can not be excluded from the rhythm TIME IN ENGLISH VERSE RHYTHM "e? | , + ^1 <2 I 1 C rt 00 O 00 00 fc 05 CO lO O r- t^ co P-l GO B 05 to o' o tO Tf Tfl CO CO CO O5 l^ !> CM 00 GO 00 O 00* 10 >H tq o' to' c> O5 CO 00 CO CO OCOOIO CMCMCOrJ* CMCMCMCM CMCMCMCM r>+2-&O 0000 ^S*^ e ^ e e o , M ~ l 3 ^ QC ^ '^3 -4-^ rH CM CO * EXPERIMENTAL RESULTS 41 05 ^ b> CO la M OS OS CO GO b- GO 00 b- GOOOOOCO IO to to GO IQ oq to to o ^ to to* ci 06 06 H e^tooqcs >1 1 1 -^ 00 . 00 . (M CO CO to 00 a 4 CM O 2 o o *H 2 t2 -8 * ^ o 000 -M CO Tfi i-H 5^1 CO 42 TIME IN ENGLISH VERSE RHYTHM TABLE V THE TETRAMETER LINE, Oh pa'pa pa'pa pa' pa pa'. At the bottom the first " pa " is combined with the second " pa' " as an iamb. W40 Oh V Pa' V pa V Foot V Ratio V Oh pa' pa 25 9.8 17 5.9 17 5.6 34 4.5 1.04 7.1 pa' pa 17 6.0 17 5.8 35 5.2 1.03 6.2 pa' pa 18 5.9 17 5.7 35 4.8 1.05 6.9 pa' 168 19.0 pa pa' 34 5.3 1.07 6.2 H44 Oh Pa' pa 27 5.9 26 6.9 41 6.2 67 4.2 .64 10.9 pa'pa 25 5.8 40 6.6 66 3.9 .62 9.6 pa' pa 26 7.2 41 6.0 66 3.5 .62 12.0 pa' 102 15.0 pa pa' ^ 67 3.6 .63 10.7 F 40 Oh Pa' pa 32 7.9 22 7.9 24 7.2 46 4.5 .93 12.6 pa'pa 23 6.9 24 9.1 47 4.7 .94 12.4 pa'pa 23 7.9 27 7.3 49 5.2 .86 11.3 pa' 62 8.3 pa pa' 46 3.6 .95 12.0 B 40 Oh Pa' pa 27 8.3 34 6.4 19 10.7 54 4.7 1.8 11.8 pa' pa 34 8.4 18 12.2 53 4.8 1.9 17.4 pa'pa 34 6.0 19 8.3 53 3.5 1.8 12.1 pa' 109 8.5 papa' 53 4-4 1.8 16.0 altogether and certainly it does not seem sensible to call it part of a succeeding sound. In the case of the pause at the end of a verse or couplet I have made no attempt, as a rule, to separate the last syllable and the pause proper. As a consequence the final syllable does not enter into my calculations. If the verse is read as "run on" into the next verse this last pause is significant; if the verse is read as closing a stanza or as the end of a sentence the pause at the end has little or no meaning. I have contrived in nearly every case to avoid such dead endings by having the subject read another verse or two which was not counted. The nonsense verses were read one after another as if they made a long stanza. The pauses therefore are true verse pauses. (&) Duration of Feet One of the first places to look for that regularity which is sup- posed to characterize rhythm is in the time value of the "foot," or verse unit. The question here is, is one foot in a verse of approximately the same duration as another, or is the presence of such equality only accidental in some verses, others not showing it? EXPERIMENTAL RESULTS 43 The force of an analogy with music is, of course, dependent upon the presence of such equality between the feet considered as measures. The facts which bear upon this point will be found in the tables under the heading "Foot." Glancing through Tables I. TABLE VI THE EFFECT OF CHANGING THE SYLLABLES IN IAMBIC TETRAMETERS. Records for two readings; the first day's reading above, the second below. w Cases Pa V pa'r V Foot V Ratio V 1st foot 14 1st day 17 3.8 75 5.4 92 4.5 4.4 6.1 17 2d day 18 64 79 3.6 97 4-1? 4.3 4-2 2d foot 1st day 17 6.3 77 3.6 93 3.5 4.7 7.0 2d day 18 3.5 80 5.0 97 4.5 4.3 4-1 3d foot 1st day 16 4.2 77 5.6 93 4.7 4.9 6.6 2d day 18 5.1 77 6.3 96 5.5 4.3 8.1 4th foot 1st day 16 7.6 116 6.6 132 2d day 17 4.2 129 11.0 146 W pa'ra 1st foot 15 1st day 17 4.1 65 5.2 83 3.8 3.8 8.0 17 2d 'day 19 6.0 68 4^1 87 3.7 3.6 5.7 2d foot 1st day 17 6.6 67 5.5 84 5.3 3.9 5.3 2d day 18 3.8 71 2.9 89 2.8 3.9 4.0 3d foot 1st day 17 3.6 65 7.0 81 4.6 3.8 6.9 2d day 18 6.6 68 2.8 87 2.7 3.8 5.1 4th foot 1st day 17 4.3 124 12.1 141 2d day 18 5.5 158 14.1 177 W pa'rst 1st foot 13 1st day 17 4.7 65 5.3 83 5.2 4.0 5.5 13 2d day 18 6.4 76 4-5 93 4.3 4.2 6.4 2d foot 1st day 18 5.2 67 54 84 4.0 3.8 7.7 2d day 18 5.5 77 3.7 95 3.6 4.4 4.9 3d foot 1st day 17 8.7 64 6.6 80 5.6 3.8 9.5 2d day 18 4-1 76 4.8 93 4.0 4.3 5.7 4th foot 1st day 17 5.1 119 20.3 136 2d day 18 5.8 143 23.0 161 to IX., it is evident that the nonsense feet are very regular for each of the four subjects in each of the five rhythms examined. There are minor exceptions as, for instance, in Table I., Subject H, shows a slight acceleration in his reading from the first to the third foot. The low variation of his readings here makes it quite im- possible that this should be the result of chance. 4 But on the whole *The variations are expressed throughout in the amount of the mean variation (A.D.) of the series per centum of the average reading. This gives a figure that can be used for the comparison of the variation of one series with that of another series which has a somewhat different average value. The reader should, however, be very cautious about making such comparisons. The " probable error " ( P.E. ) of one of these averages would be equal to the 44 TIME IN ENGLISH VERSE RHYTHM TABLE VII THE EFFECT OF CHANGING THE SYLLABLES IN IAMBIC TETRAMETERS H 17 Pa V pa' V Foot V Katio V 1st foot 26 6.9 60 3.4 86 3.5 2.4 8.1 2d foot 26 5.2 59 4.2 85 3.4 2.3 6.1 3d foot 26 4.9 60 4.3 87 3.7 2.3 6.9 4th foot 27 4-9 133 7.2 161 H 33 pa'r 1st foot 27 6.1 63 3.8 89 2.9 2.3 7.2 2d foot 27 6.0 61 3.3 89 2.8 2.2 6.2 3d foot 27 4-2 61 4.3 89 3.2 2.2 6.9 4th foot 27 4-7 121 9.0 148 H 32 pa'rs 1st foot 26 5.6 68 3.2 95 2.1 2.6 7.7 2d foot 25 5.1 68 3.2 93 2.4 2.6 6.7 3d foot 26 $.1 67 2.7 93 2.7 2.5 5.1 4th foot 26 5.0 130 8.4 157 H 34 pa'rst 1st foot 27 8.1 67 5.0 94 4.3 2.5 10.0 2d foot 26 5.5 67 3.7 93 2.8 2.6 7.2 3d foot 26 6.3 67 4.2 93 3.5 2.6 8./ 4th foot 26 4.2 133 7.5 159 the feet may be said to have equal duration in the nonsense verse. No such equality appears, however, in the examples of actual verse. Even in the most simple and regular cases when there can be no question raised as to the correctness of the scansion the irregulari- ties are glaring. In the case given in Table X., for instance, which can be put in this form: Subject I come from haunts of coot and hern H 70 66 54 H R 62 82 58 < M where the figures represent the times for Subjects H and R re- spectively. Not only are the feet not by any means equal but the inequalities are not subject to the same tendency for the two readers. mean variation (A.D.) divided by the square root of the number of cases. This can be obtained from the per cent, value given in the table by multiplying by the average reading and dividing by the square root of the number of cases. The number of cases is large in the case of the nonsense material, because the variability itself was under consideration. In the ordinary verse only an approximate average was required and only ten cases as a rule were taken. The apparent increase in accuracy from the greater number of cases is in fact negatived by a constant shift in the tempo of the reading so that fifty cases do not form as homogeneous a series as ten. EXPERIMENTAL RESULTS 45 TABLE VIII THE EFFECT OF CHANGING THE SYLLABLES IN IAMBIC TETRAMETERS Continued F31 1st foot 2d foot Pa 17 17 V 104 9.9 pa' 39 38 V 5.4 7.0 Foot 55 55 V 4.7 6.1 Ratio 2.3 2.3 V 10.6 10.3 3d foot 16 10.4 36 5.5 52 5.2 2.3 11-4 4th foot 16 10.3 61 7.2 77 F40 1st foot 16 9.9 pa'r 48 6.2 65 5.0 3.1 10.6 2d foot 17 6.2 47 5.9 64 5.5 2.8 8.1 3d foot 16 7.5 47 5.6 63 4-7 2.9 8.5 4th foot 16 7.7 69 7.1 85 F40 1st foot 14 7.4 pa'rs 49 8.7 62 7.1 3.5 9.9 2d foot 14 9.9 49 10.2 63 9.1 3.4 11.1 3d foot 14 9.2 48 8.3 61 7.4 3.4 10.S 4th foot 14 9.5 73 6.9 87 F41 1st foot 14 10.2 pa'rst 48 7.4 63 6.8 3.5 13.4 2d foot 14 9.5 47 6.6 61 5.8 3.4 10.6 3u foot 14 10.5 47 6.8 60 6.7 3.5 10.8 4th foot 14 11.0 73 6.5 87 TABLE IX THE EFFECT OF CHANGING THE SYLLABLES IN IAMBIC TETRAMETERS Continued B23 Pa V pa' V Foot V Ratio V 1st foot 18 5.5 40 6.7 58 4.8 2.3 10.7 2d foot 18 8.0 41 4.5 60 4.2 2.3 9.7 3d foot 18 6.7 40 6.8 57 5.1 2.2 10.4 4th foot 19 7.0 116 9.0 135 B40 pa'r 1st foot 17 6.0 44 4.9 61 3.2 2.6 9.5 2d foot 17 4.8 44 4.6 61 3.7 2.6 5.7 3d foot 17 5.5 44 4-1 61 3.0 2.5 74 4th foot 18 5.3 121 5.2 139 B39 pa'rs 1st foot 16 5.7 48 6.3 64 4.2 3.0 11.1 2d foot 16 4.6 49 5.3 64 3.7 3.1 8.3 3d foot 16 4.6 49 4-9 64 3.4 3.1 7.2 4th foot 16 5.0 124 6.6 140 B40 pa'rst 1st foot 16 5.2 53 7.5 67 6.4 3.4 9.0 2d foot 16 5.9 53 7.6 69 6.5 3.5 7.0 3d foot 15 5.1 53 7.2 69 5.5 3.5 9.0 4th foot 16 5.8 127 6.0 143 a V b' V Foot V b'/a V b'/a5 V 38 4.2 32 10.1 70 5.5 .84 12.2 1.3 9.1 24 9.5 41 6.1 66 4.7 1.74 13.8 1.5 8.7 28 S.-6 26 8.1 54 4.7 .93 12.9 1.6 14.8 16 1L1 101 4.4 117 307 274 3.9 46 TIME IN ENGLISH VERSE RHYTHM TABLE XA AVERAGES OF 10 READINGS BY H OF THE VERSE, / come from haunts of coot and hern. Under " a " and " b," the unaccented and accented syllables respectively. a b' I come from haunts of coot and hern Verse Next verse 6 Come divided by from, etc. TABLE XB AVERAGES OF 10 READINGS BY R OF THE SAME VERSE a b' ^ a V b' V Foot V b'/a V I come 32 10.8 30 9.8 62 5.7 .93 14.8 from haunts 30 9.0 52 4.8 82 5.9 1.74 7.2 of coot 22 8.5 36 7.0 58 4.8 1.63 12.7 and hern 20 8.5 54 11.8 74 Verse 276 Next verse 246 4.9 TABLE XlA AVERAGES OF 10 READINGS BY H OF THE VERSE, The stag at eve had drunk his fill a b' a V b' V Foot V Ratio V The stag 19 8.3 46 5.2 65 4-5 2.4 9.3 ac eve 13 1J,.8 40 4.0 53 5.7 3.2 18.2 had drunk 18 8.6 41 2.6 59 3.2 2.4 10.3 his fill 28 9.3 106 5.9 134 Verse 311 TABLE XlB AVERAGES OF 10 READINGS BY R OF THE SAME VERSE a b' a V b' V Foot V b/a V The stag 19 16.0 47 8.0 67 7.2 2.6 16.0 at eve 19 8.8 41 7.6 60 6.7 2.1 9.0 had drunk 38 8.7 32 6.2 70 5.0 .85 10.0 his fill 24 11.1 63 9.2 87 Verse 284 Next verse 313 4.8 Any of the other verses will serve to bring out the same fact. Only one case (Table XI.). Subject The stag at eve had drunk his fill H 65 53 59 R 67 60 70 shows anything like the regularity which occurs in the case of the EXPERIMENTAL RESULTS 47 nonsense verses. Other selections are not so regular in form and the scansion is open to some question. In some cases no attempt has been made to scan the verse and the reader is left to exercise his own ingenuity if he thinks that he can find any combination mak- ing uniform feet in a verse with separate syllables, such as those, for instance, of the second half of Table XII. TABLE XII AVERAGES OF 10 READINGS BY L. The time of each syllable is below it and below that the mv. per cent. Al most u pon the wes tern wave Verse pon 22 34 11 26 11 28 15 48 9.0 8.8 13.0 18.0 10.0 20.0 7.0 9.6 Res ted the broad bright sun 28 20 13 37 20 89 22.0 19.0 15.0 9.5 19.0 6.5 When that strange shape drove sud den iy 21 16 45 46 28 28 11 36 9.0 17.0 12.0 12.0 8.6 8.0 13.0 21.0 Be twixt us and the sun Pause 17 28 27 27 12 40 99 94 84 9.6 14.0 16.0 6.7 14.0 195 207 194 250 But such puzzles as this excepted a perusal of the tables will convince anyone that there are divisions corresponding pretty closely to the conventional feet and having somewhat the same duration even though there is nothing that could be called equality between them. Such an example as the nursery rime "Pease porridge TABLE XIII THE AVERAGE TIME FOR VERSE AND THIRD FOOT FOR EACH OF FOUR READERS IN EACH RHYTHM w H F B Verse 3d Foot Pa pa' Pa pa'r 436 96 Pa pa'rs 440 87 Pa pa'rst 442 93 Pa pa' 353 77 Pa' pa 326 72 Oh pa' pa 297 35 Pa ta pa' 345 74 Pa' ta pa 289 63 Average 366 75 Verse 3d Foot Verse 3d Foot Verse 3d Foot 419 87 238 52 310 57 450 89 277 63 322 61 438 93 273 61 332 64 438 93 271 60 348 69 340 73 247 52 370 73 334 67 264 57 310 60 328 66 236 49 296 53 367 86 286 62 312 59 397 86 261 58 264 60 386 82 264 58 319 62 48 TIME IN ENGLISH VERSE RHYTHM > ? to ^ ^o Cioq uj oo to^ s to eo eo -4 *? i i *# O lO Ci i i CO i i ^f rjl CO CO 00 ^ (M J3 00 J3 (M (Mi-t F-H i i i-H(M *% ^ ^^ oq oo 10 00 &1 CO b- CO OS CO e* to >H >i eo te**aoe<>-***>**-fefee6 ^ iri *O to >-H ^'tototo^'toeo'so^eototoioio o S 4- O CD lO O & 00 CM 6 4-CX|lO.OCilOCO(>]COOOO' i O O -g XiOOt-OO' iCMCCcOI^CDCMt-t^CO 1 . ^ ci 1 . hO > -t- -^ Ci Oi f" to^oo^stodCi^eotos^^Z"" 1 > to -4 io o ^ to to to *o to" ^4 *o "*} to >-H to to 2~-- a CO CD ^"f CO ^H Ci r-H Ci ^rMioiOiO^OCiCiCMOOCOOt^ ^ ^ GO t> GO CO GO CD CM ^ CO !> CM ^ ^ i 1 1 1 1 1 1 | < ^ i N ?^ 05 ^ * 00 ^, K> O X TABLE 2 W M 02 o 2 s ** to oo oo ^ ^ 06 ^ to ci ^~ !> O lO CO ^ t 00 2 * 2 2>;OOOOCiOO~*hOO>HC>3 pqW>* o6otoeoto-4j^^eo PQ "+ T*H.-HCOCO1OCDCMOOCO s I Q II M . 1 1 PH r- eo eo * a PH ^COOOC>JOOC>C5501 CO 06 06 06 06 *" o^ Svjiotosvjtoz^-eo'^z^-eo CM r^ rH ^*H ^*H ^*S - 00 (M CO CD ^ CO lO * * fn . lOi iCOGOlO^t l-H I! ^ oi 06 2 s - cs cj 55S5| gj Ci CO CM ^H CO K ai,i o -o QJ 2 q % ^< 03 ^ ^- O 01 ''* 5 ^J JM S-i Q j_j QifliJ^C^CUC^^rt PQjD, OM 03-ij bc r O r d >^ M * *C "C 'C * m ^ ^ >-< >* CD aaa^ . u > 6* 1 &4i 4^' I'J 3 83 S $ o .S o 1 "s 1 i PH p p ( rH EXPERIMENTAL RESULTS 49 hot" which is cited by Poe as a verse having feet of equal duration, will serve to show how far there really is equality. The scansion is Foe's, the figures from Table XIV. Pease porridge hot, pease porridge cold, 49 38 113 48 52 106 Pease porridge in the pot, nine days old. 49 46 43 48 57 46 116 Taking account of the pauses the feet certainly maintain a pretty constant length (and this testifies to the accuracy of Poe's ear) but there is no real equality between them. The same thing can be seen in Table XV., where not even the three words Break, break, break occupy equal times, but where well-defined feet are in evidence. In some cases such "feet" are only to be discovered by a viola- tion of the conventional scansion. Table XVI. gives Where twined the path in shadow hid 89 82 79 Here an extra syllable has to be borrowed by the third foot from the last and even then it is shorter than the others. Evidently mechanical scansion will not give feet of even approximately equal lengths. But if freer scansion does reveal such feet the question arises how much irregularity can be countenanced; that is. where "regularity" begins. One experiment made upon Subject H, who can sing, throws some light on this question in so far as it reveals correspondingly large irregularities in the duration of musical notes. The results of recit- ing and singing the same words, "Blow, blow, breathe and blow, Wind of the western sea," are summarized in Table XVII. The times are given for the syllables as read and as sung; 9 then the proportional time the notes should have in the song, and lastly, the time of the syllables as read increased to the tempo or rate at which they were sung. Both reading and singing have a different tempo for the two verses. It appears that in this case at any rate, the reading while regular as regards "feet" is no more so than singing without accompaniment, 10 Certain differences are noteworthy. The reading tends more to equalize the corresponding words blow, blow. 9 The subject is thoroughly competent to sing this well-known song accept- ably and he said that the apparatus did not hamper him. 10 For the degree of regularity to be expected in musical performances see Binet and Courtier, L'Annee psychologique, 1895. 50 TIME IN ENGLISH VERSE RHYTHM TABLE XVI AVERAGES OF 13 READINGS BY H OF Where twined the path in shadow hid a b' a V b' V Foot V Ratio V Where twined 43 6.9 46 5.2 89 3.9 1.1 9.3 the path 18 6.5 64 6.1 82 4.! 3.6 8.3 in shad 26 11.5 29 5.6 55 5.5 1.1 14.8 ow hid 24 11.5 81 7.0 105 Verse 331 2.8 TABLE XVII AVERAGES FOR H, 10 CASES READING AND 20 CASES SINGING, Blow, blow, breathe and blow, Wind of the western sea Beading Time V Time^V Measure Singing 7 Reading 8 Reading Time Time Time V Measure Singing Time V Blow 113 7.6'' 173 7.3 174 189 113 7.6 173 7.3 blow 116 1.5 188 5.8 175 194 116 7.5 188 5.8 breathe 51 14 113 7.7 131 86 and 45 6.7 61 11.0 44 75 95 5.7 173 5.5 blow 92 5.0 164 10.0 174 154 92 5.0 164 10.0 Verse 416 698 Wind 62 10.0 86 11.0 of 62 5.0 the 27 5.0 25 16.0 89 7.8 170 9.7 wes 38 8.8 113 12.0 tern 31 84 46 15.0 70 6.7 158 9.7 sea 199 4.7 176 10.0 176 10.0 ea (2d measure) 112 199 112 Verse 358 616 The two words breathe and are sung in correct time, but are spoken too fast. The same is true of the two syllables of western. The" general inference to be drawn from this table is that the lack of abso- lute equality in time between the feet does not mean that there are no feet, but rather that the changes due to introducing meaning and variety into the material tend to reduce very greatly the regularity. That the time occupied by a foot is only in part dependent upon the number of consonants involved is seen in Tables VI. to IX. These records were so taken as to avoid interference from the tendency of the reader to change his rate of reading during the sitting. The individual differences of the readers are so large that it is not possible to say what the effect of adding any one consonant will be. ^Evidently it will not be great, and in some cases the time is actually shorter when the consonant group is more complex. The figures of Table XIII. again give an opportunity for rough com- 7 Theoretical time for the notes on basis of total time of verse. 8 Reading time magnified to same scale as singing time. EXPERIMENTAL RESULTS 51 parison. The large individual differences undoubtedly depend in large part upon the way in which terminal r was pronounced. In many cases the verse seems to be divided into short phrases rather than "feet" in the ordinary sense. "Pease por- ridge," Table XIV., is an example. Other examples are "Break, break," Table XV.; see also Tables XVIII., XIX. These phrases might be considered as the rhythmic elements in the verse, for they are fairly uniform in length while the feet are far less regular. Still more might the whole verses themselves be considered the units since they are still more uniform than the feet. TABLE XVIII AVEEAGES OF 35 READINGS BY H OF Each purple peak, each flinty spire Was bathed in floods of living fire 4 Feet V a b' a V b' V Foot V Katio V 2 Feet V Each pur 46 9.1 22 7.2 68 6.2 .48 15.3 pie peak 28 8.9 81 6.7 110 4.9 178 44 each flin 35 7.7 31 8.3 66 5.8 .88 12.0 ty spire 21 7.5 90 7.3 111 6.8 177 5.0 Was bathed 42 9.3 46 4.0 88 5.2 1.13 9.5 in floods 17 12.0 53 6.6 70 5.2 159 4.0 of living fire 223 6.0 355 3.9 379 4.5 TABLE XIX AVERAGES OF 10 READINGS BY H OF Double, double, toil and trouble, Fire burn and caldron bubble b a' V b V Foot V Eatio V 2 Feet 4 Feet .45 9.4 .35 13.3 155 1.4 9.0 198 353 1.8 17.7 160 cal dron 41 7.2 39 6.9 81 4.5 1.07 11.0 bub ble 17 4-4 148 J t .l 165 246 406 But there is a criterion of the "foot," or element, provided in the absolute time of the nonsense verse. There does exist in that material a unit of very definite form revealed by the most casual inspection of the voice record. This unit has rather narrow limits in point of absolute time value for any one reader. The average value of none of the first three feet in the tetrameters of Subject H, Table VII., is longer than .95 second or shorter than .85 second; and so for the other readers. 11 The word fire seems to have two syllables here. Dou ble 23 6.2 51 6.1 73 4-1 dou ble 21 8.1 61 13.2 81 9.8 toil and 54 6.4 40 9.8 93 6.6 trou ble 26 11.3 79 7.4 105 5.7 Fire ? n 62 7.9 62 7.9 burn and 62 12.2 35 10.5 98 8.9 52 TIME IN ENGLISH VERSE RHYTHM Since these well-defined units seem so essential to free spoken rhythms like the nonsense verse we are justified in looking for them in other verse as well. But in every case the longer phrases of which I have spoken differ from the "feet" of the nonsense verse in the very fact that they are longer. The phrases of "pease por- ridge," Table XIV., for instance, are about twice as long as the longest feet of the nonsense verse. It is fair to say, then, that whatever the significance 12 of these phrases they are not analogous to the ' * feet, ' ' in the strict sense, which appear in nonsense material and also in the simplest poetic verse. The absolute time for a foot for any one individual is determined partly by the rhythm and partly by the content, as can be seen by following down the columns of Table XIII. In different selections of actual verse the feet occupy times approximating the time of a foot in the same rhythm in nonsense verse. The minimum is above half a second and the maximum less than one second. Closely connected with the matter of the duration of the feet is that of the rapidity of utterance as indicated by the time of a whole verse. Here we find very great individual differences. These come out most strikingly in the nonsense material. Table VIII., for example, shows Subject F going through identical material in only two thirds of the time occupied by Subject H. In Table XIII. the rates of different readers are shown in all the rhythms. It will be seen that each has a very regular characteristic tempo relative to the others, F being fastest and H slowest. The averages at the bottom give a rough measure of these personal peculiarities. 13 y (c) Ratio of Accented to Unaccented Syllables The second place in which regularity is to be expected is in the ratio between the accented and unaccented parts of the foot. Here again we find very remarkable regularity in the case of the nonsense verse and even more remarkable irregularity in the case of the natural verse. The most generally significant fact is that each separate verse has its own ratios and each reader his own, so that while these ratios remain fairly well fixed for any one reading they 12 1 am inclined to hold that such clauses are to be interpreted as short verses. The length of a line on the printed page often obscures the true length of the verse, two actual verses being printed as one, or one divided in the middle. In the absence of a regular rime the determination of the true verse requires special investigation. "This is not the place to discuss these results from the point of view of individual differences. They are probably parallel effects associated with rapidity of respiration and heart beat, all expressive of the general organic rate of the individual. EXPERIMENTAL RESULTS 53 have no constant form. Thus in Table II. the ratios for different readers range from .45 to I. 14 In Table VIII. on the other hand the ratio (still remaining fairly constant for one set of material) changes from 2.3 to 3.5 with a change of material in the same rhythm. These facts exclude at once the idea that there is any fixed rela- tion between a "long" and a "short" syllable. Moreover when we turn to the natural verse we find that there is no trace of such a relation. Table XVIII., for instance, shows two exact reversals of normal condition in iambic verse. The "short" syllable is here twice as^lpng"7as the corresponding "long n syllable. Was and bathed on the other hand are almost equal. Again, in Table XIX., double shows the accented and theoretically "long" syllable less than half as long as the "short" syllable. It is needless to multiply par- ticular instances ; the tables are full of them. Tables VI. to IX. show the results of an experiment which is crucial upon this point. It amounts to this; when the character of the syllable is changed in any given rhythm there is a change of ratio of the parts. In Table IX.tthe effect of complicating the long syllable in iambic rhythm 15 is to lengthen the whole foot regu- larly. But at the same time the short syllable is shortened as if there were a tendency to make it compensate for the greater length of the other syllable. ( The result is a steady increase in the ratio of the parts along with the increase in total length of the foot. Table VIII. shows the same thing, save that the compensation is really effective in the lower half of the table so that the total length of the foot does not increase, but there is enough decrease in the first syllable to balance the increase in the second, and the ratio mounts rapidly. In Tables VI. and VII. the short syllable remains fairly constant and changes in the long one are reflected in a changed ratio. Without the evidence from the natural verse these tables 14 By ratio is to be understood the figure got by dividing an accented by an unaccented syllable. The latter is then always unity and the figure given is the relative length of the accented syllable; this figure will be less than unity if the accented syllable (as sometimes happens) is not so long as the unaccented. The variations given with the ratios are per centum and are only to be used in comparing different ratios. They furnish no criterion for com- paring the variability of ratios and of separate syllables or feet. 15 Iambic rhythm offers the critical case for this phenomenon because the effect is seen in the short (first) syllable which is not directly affected by the changes in the other syllable. If trochaic rhythm were taken such a shorten- ing of the unaccented syllable in response to a lengthening of the accented might be ascribed to a change in the length of the pause at the end of the foot. In the case of iambic the pause can not be connected in any way with the phenomenon. 54: TIME IN ENGLISH VERSE RHYTHM offer complete proof that no " strict music" theory of verse will hold. The syllables do not behave like the notes in music. On the contrary they are inversely related ; when one is long the other is short. A change in the rate of reading does not seem to have any very regular influence on the ratio of the syllables. Table VI. shows the changes which occur when a subject reads the same material on different days. The differences both in tempo and ratio are large compared with the variation within either of the series. The records of pa paf in Table I. may also be compared with Tables VII., VIII. and IX. taken on different days. The changes of tempo, as indi- cated in the summary below, are considerable in every case but the ratios show only a very slight tendency to increase as the rate becomes slower. Subject Time of Verse Ratio H 340 2.1 419 2.3 F 247 2.3 238 2.3 B 370 2.5 310 2.2 Table XX. gives the result of a deliberate effort to change from the normal rate of reading to a faster rate. To the listener this gives a general impression of haste and the esthetic result is bad, nevertheless the reader managed to give more than thirty repetitions with an average variation of only 3 per cent, for the verse; his performance was practically as uniform at the rapid rate as when reading normally. His own impression was that the saving in time was made at the cost of certain slight logical pauses in the verse. TABLE XX THE VERSE, She kept with care her beauties rare, READ BY R 30 TIMES AT A NORMAL TEMPO, AND 30 TIMES IN HASTENED TEMPO a b> a V b' V Foot V Ratio V Verse V She kept with care her beau ties rare norm. 30 10.0 56 7.5 85 4-9 310 2.9 240 3.1 norm, hast. a 44 32 V 9.1 10.0 b' 20 18 V 4.9 10.0 Foot 64 49 V 6.4 Ratio .47 .56 V 9.9 14.0 norm, hast. 40 32 3.3 5.5 58 42 7.5 9.6 98 74 4.7 5.6 1.4 1.3 7.2 11.0 norm. hast. 37 32 7.7 7.2 23 21 6.7 4.6 60 53 5.8 4.1 .64 .66 9.6 9.2 norm, hast. 30 25 10.0 7.3 56 38 7.5 10.0 85 63 4.9 5.0 EXPERIMENTAL RESULTS 55 The results show that this introspection is only partly valid. All the syllables are shortened and the ratios, which are themselves notably different from the accepted iambic scheme, remain prac- tically the same. It appears, however, that the greater part of the shortening comes out of the long syllable whether it is accented or not and whether or not it includes a pause. Deliberately changing the tempo does not, then, serve to make any grave difference to the rhythmic structure nor to affect seriously its regularity. 16 There is, however, a slight change in the ratio of the syllables as was the case when different vocal complexes were tried in the same rhythm. In view of the great variety of ratios which appear between the syllables of the different feet in even the most simple of the verses of poetry which have been examined it is remarkable that there is such great regularity between the different readings of the same verse. In other words, no matter how irregular the verse pattern, it is persistent, maintaining its form time after time. It is reason- able to suppose that this regularity is partly a function of the natural duration of the individual words which enter into the verse. It is not true, however, that the same words have the same time value in different verses. (d) Rising and Falling Rhythms A discussion of the ratio of the syllables in a foot naturally leads to a consideration of those particular time relations within the foot which constitute the character of a particular rhythm. And first of those ; wherein do the rising rhythms with the accent on the last part differ from the falling rhythms with the accent at the beginning of the foot? Here again the distinction is clearest in nonsense material where the nature of the subject-matter exerts the least disturbing influence. Table I. shows rising, Table II. falling rhythm with the same pair of syllables differently accented. In iambic (rising) rhythm the length of the long syllables is between two and three times that of the short (unaccented) syllable. In trochaic (falling) rhythm the accented syllable is shorter than the unaccented in every case but one. The ratios in the third foot 17 are, for the different sub- jects : Subject W H F B Iambic 2.9 2.1 2.3 2.5 Trochaic .46 1.04 .86 .87 16 See discussion of variabilities below. 17 This foot is selected as typical because it is less subject to the disturbing effect of position. Tne first foot is likely to be irregular, the second and fourth are followed by pauses. 56 TIME IN ENGLISH VERSE RHYTHM The reader, H, who has the least difference between the syllables in one rhythm has the least difference in the other, and the one who has the greatest difference in the one, W, has the greatest dif- ference in the other. This being interpreted means that the second syllable was always relatively short for the one and long for the other. The ratio for iambic tends to be greater than the orthodox 2:1 of the theorists, while the ratio for the trochaic tends more to be 1:1. The individual differences are so great as to cause an overlapping of the two species. Thus the ratio of W, trochaic, is practically the reciprocal of that of H, iambic. Consequently it is not possible to draw a hard and fast line between the two rhythms on the basis of the time relations of the syllables alone. The absolute times of the third foot as given in Tables I. and II. are: Subject W H F B Iambic 77 73 52 73 Trochaic 72 67 57 60 The tendency to acceleration in the trochaic rhythm is too small to be significant. Table V. shows what remarkable results can be got by what appears to be a very insignificant change in the rhythm. The introduction of a catch syllable at the beginning and the cutting off of the final unaccented syllable at the end of the verse gives a new tempo and new set of ratios to the whole verse. The time for verse and foot, Table XIII., is less for this form than for either trochaic or iambic. The ratio is not appreciably changed by coupling the second accented syllable with its succeeding (trochaic) or preceding (iambic) syllable. For corresponding third feet the ratios are : Subject W H F B Iambic 2.9 2.1 2.3 2.5 Intermediate form 1.05 .62 .86 1.8 Trochaic .46 1.04 .86 .87 For F the rhythm is the same as trochaic. For W and B it is intermediate between trochaic and iambic. For H it is more trochaic than the natural trochee. A comparison of trochaic and iambic rhythms in actual verse is not easy, the chief reason being that their time relations are too much alike. ' ' The boast of heraldy, ' ' Table XXI., and "The stag at eve," Table XL, are as regular as any of the iambic specimens. Table XXII. gives a trochee which may be said to be more trochaic than the nonsense type because in EXPERIMENTAL RESULTS 57 TABLE XXI AVERAGES OF 15 READINGS BY H OF The boast of heraldry, the pomp of power a b' a V b' V Foot V Ratio V The boast 17 10.6 46 3.6 63 34 2.8 11.6 of her 16 8.5 45 5.6 61 4.3 2.9 13.1 aldry 73 The pomp 18 104 42 5.2 60 4-7 2.5 13.0 of power 29 8.8 92 5.2 120 5.1 Verse 377 24 TABLE XXII AVERAGES OF 12 READINGS BY H OF Sleep the sleep that knows no breaking a' b a V b V Foot V Ratio V Sleep the sleep that knows no 50 2.9 43 8.8 49 5.0 18 14.0 19 11.0 46 5.8 67 4-5 62 7.3 94 2.6 2.8 17.0 2.3 11.0 1.1 9.3 brea king Verse 24 4-2 150 8.7 174 397 TABLE XXIII AVERAGES OF 10 READINGS BY L OF Wha will be a traitor knave, etc. a b a' V b V Foot V Ratio V Wha will 68 5.3 be a 40 10.0 trai tor 23 10.0 19 13.0 43 84 1.3 17.0 knave 55 11.0 Verse 206 Wha can 35 14.0 19 19.0 54 12.0 1.8 19.0 fill a 59 6.0 cow ard's 54 7.6 grave 76 14.0 Verse 243 Wha so 34 13.0 35 10.0 69 54 .97 19.0 base as 31 7.2 15 15.0 46 ^4 2.1 14.0 be a 38 8.5 slave 59 11.0 Verse 212 Let him 24 18.0 38 10.0 62 11.0 .62 14.0 turn and 38 10.0 20 9.0 58 9.9 1.9 9.0 flee 129 13.0 Verse 249 two of its feet the ratio of the syllables is more extreme. These two feet look more like inverted iambs. Table XXIII. shows great irregularities in respect to ratio in a verse which was selected as 58 TIME IN ENGLISH VERSE RHYTHM typically trochaic. Parts of this record could not be deciphered with certainty. The tempo is fast, and the rhythm complex, with very unequal feet. Tables XVIII. and XIX. constitute final evidence of the indistinguishableness in regard to ratios of iambs and trochees. In the former those feet which are free from pauses are better trochees than iambs ; in the latter true trochees give irregular results which would look quite as well interpreted as iambs. The confusion of these two rhythms is caused by the fact that accent does not always carry with it an increase in duration, hence the ratios do not obey the accentual schemes. In Table XXIV. we have what seems to be a specimen of the anomalous rhythm of Table VIII. Its tempo is somewhat faster TABLE XXIV TIME AND VARIATION OF SYLLABLES. H, 10 readings. And cu ckoo buds of yel low hue Verse 40 19 28 54 19 22 24 80 286 14.0 7.4 5.4 7.6 8.8 11.0 14.0 24.0 Do paint the mea dows with de light Pause Verse 25 35 9 22 26 25 19 46 100 307 8.9 5.7 10.0 10.0 9.2 11.0 6.8 8.5 4.5 TABLE XXV TIME AND VARIATION OF SYLLABLES. H, 10 readings. Read with full stop at the end. The splen dor falls on cas tie walls Verse 9 7.1 And 23 14.2 54 7.9 snow 55 9.3 26 6.4 (y) 18 52 sum 4-7 43 26 23 110 343 7.5 8.1 13.6 7.8 mits old in story Verse 59 47 19 185 414 8.5 44 4^1 5.4 than the ordinary iamb of Table XXV. but except for the unex- pected length of the unaccented initial syllable the internal structure shows no remarkable peculiarities. There is nothing to indicate whether the verse should be treated as trochaic or iambic. Both these specimens exemplify the chaotic time relations of verses which are quite respectable rhythmically. (e) 2-Syllable and 3-Syllable Rhythms The number of syllables in a foot is one of the factors which distinguishes verse rhythms. The characteristics of the three- 18 Can not be separated in the record from the preceding syllable. EXPERIMENTAL RESULTS 59 syllable type are analyzed in Tables III. and IV. In respect to the time value of the foot and verse there is no marked difference be- tween iambic (2-syllable) and anapaestic (3-sy liable). The latter move faster for two of the readers and slower for the other two. Neither do the trochaic and dactylic forms show any constant dif- ference in this respect. The relation of the ratios is complicated. Taking the point of view first that either of the short syllables in 3-syllable rhythm is analogous to the short syllable in 2-syllable rhythm we find the ratio between accented and unaccented much larger in anapaestic than in iambic. As a rule this means that the short syllable is much shorter in the former case while the long one is not correspondingly shorter. Dactylic rhythm yields a ratio much smaller than trochaic, due to the fact that the accented syllable is much shorter in the dactyl. Now consider for a moment the two short syllables together as equivalent to the corresponding part in the 2-syllable rhythm. The anapaest then shows a ratio not unlike the iamb but smaller, for the two shorts of the anapaest are longer than the one of the iamb and its long is shorter. The dactyls yield a ratio almost exactly the same as that of their analog the trochee, for all of the readers except H. His ratios for the dactyl and for the trochee with an extra syllable, Table V., are the same practically, both differing from his simple trochee in the sense that the latter departs from the regular trochaic type of the other readers. That is, his trochee is the exception, not his dactyl. Considered in this way, no very clear line of distinc- tion appears even in the nonsense verse between 2-syllable and 3-syllable rhythms. 19 The distinction between the rising and falling rhythms is not so clear in the 3-syllable type as it is in the 2-syllable type. The rate is slightly faster in the dactyls than in the anapaests. 20 The relation of the two short syllables is the same for dactyls as for anapaests, the first being shorter whether accented or not. The last syllable is only very slightly longer when accented, consequently the relation between it and the initial syllable is practically the same for rising or falling rhythm. The ratio of the last syllable to the sum of the first two in the third foot is : 19 An inspection of the voice records will bring out more clearly than the figures the fact that the two short members of the 3-syllable combination are so closely connected as to be almost inseparable while they are at the same time well marked off from the rest of the foot. 20 For subject H the verse is somewhat slower in the dactyl, but the foot only very slightly so. 60 TIME IN ENGLISH VERSE RHYTHM Subject W H F B Dactyl 2 1.5 1.2 1.2 Anapaest 1.8 1.6 1.5 1.6 These particular differences exaggerate the general argument of the tables and still they are not sufficient to be significant. Neither in regard to tempo nor ratio are dactyls clearly distinguishable from anapaests. The distinction must lie outside of the time relations of the elements in the verse. The examination of the question in the case of actual verse is made very difficult by the disputes which immediately arise as to what verse is to be read as dactylic and what as anapaestic. On this account no attempt has been made to construe the feet of cer- tain of the verses. In other cases the form is sufficiently well de- termined to warrant an interpretation. The distinction between the 2-syllable and the 3-syllable feet of the same, rising or falling, rhythm is one which is easier to draw. Table XXVI. presents a case where the comparison is made easy by the occurrence of the same words at the beginning of the two verses. The dactyl departs TABLE XXVI COMPARISON OF DACTYLS WITH TROCHEES. H, 15 readings; with full stop at the end. Dactyls Time V Trochees Time V High 52 10.1 High 48 8.0 u- 21 10.0 u- 20 8.7 pon 29 6.0 pon 22 6.4 High- 35 9.1 a 18 5.8 lands 62 6.7 gol 32 6.8 and 18 8.5 den 26 6.7 low 43 4.5 throne 160 7.8 u- 20 9.0 pon 41 8.1 Tay 125 4-5 Verse 446 326 from the standard of the nonsense verse in that the accented syllable is comparatively long, while the final unaccented syllable is not proportionally so long. The trochee is not typical in that its first syllable is too long but after that the approximately equal length of the syllables is typical. The significant difference between the dactyl and the trochee lies in the greater length of the syllable pon in the former. This is evidently due to the pause which follows. That syllable in the trochee is initial in the foot and even with a very free reading bears an accent. The presence of this accent does EXPERIMENTAL RESULTS 61 not, however, make it as long as the corresponding weak syllable at the end of the dactyl. Table XV. shows at least three anapaBsts which approximate the type given by the nonsense verse. The combination "Will never come back" can be scanned by the reader to suit himself; the point of chief interest is that the strongly accented syllable ne is very short in time, being a little over one third as long as the preceding unaccented will. This verse read as iambs (the two unaccented syllables counting as one) is quite as regular as it is taken any other way and quite as near the ordinary iamb as is, for instance, "I come from haunts," Table X. TABLE XXVII AVERAGES OF 10 READINGS BY H Verse 297 Why then comes in Half verse 34 37 31 36 137 8.9 4.6 7.7 7.2 2.3 the sweet of the year 11 32 11 15 92 11.7 8.5 12.6 13.2 8.0 For the red blood reigns Half verse 22 8 44 29 62 165 10.3 2.5 4.8 5.7 7.9 3.0 in the win- ter's pale 13 4 20 41 157 15.2 13.0 9.0 3.7 5.8 Verse 400 Table XXVII. presents a mixed type, but too much disturbed by pauses to show the internal structure. Such a rhythm as that in Table XXVIII. is not consistent with any system of proportion be- tween the syllables. As this rhythm does not show definite feet under any construction it remains a puzzle. Similar difficulties appear in Tables XXIX. and XXX., yet those verses were read very rhythmically by the subject. The lines of which the analysis is given in Table XXXI. are supposed to be regularly dactylic. Scanned thus they give fairly uniform feet in rather slow time. Another and freer scansion also gives about the same regularity. The difficulty with the conventional scansion is that it makes the final syllable of several feet very short; a condition which was not found in any of the nonsense rhythms. The cause of the length of the final syllable in all the rhythms seems to be the presence in it of a pause at the end of the foot. The absence of a long syllable, i. e., of the sign of pause, argues that the foot does not really end 62 TIME IN ENGLISH VERSE RHYTHM TABLE XXVIII AVERAGES OF 10 READINGS BY H I stood on the bridge at mid night Verse 25 27 27 16 28 17 25 76 241 15.0 94 7.5 5.8 3.9 14.0 11.0 7.3 As the clocks were strik ing the hour Verse 17 17 34 33 32 23 129 285 12.0 13.0 15. o 14.0 6.5 13.0 6.7 And the moon rose o'er the ci ty Verse 18 12 37 34 36 15 18 59 229 13.0 18.0 9.0 10.0 7.0 9.5 8.5 16.0 Be hind the dark church tower Verse 11 39 15 37 44 184 330 10.0 8.0 3.2 7.1 6.1 5.2 TABLE XXIX ._. AVERAGES OF 10 READINGS BY H Sun set and eve ning star Verse 46 48 20 25 24 105 268 9.8 8.7 8.0 12.0 13.0 7.6 And one clear call for me , Verse 18 54 48 47 24 91 282 12.0 8.2 54 4.3 8.3 4.5 And may there be no moa ning of the bar Verse 20 31 25 19 28 28 19 17 20 64 271 5.4 13.0 5.8 11.0 5.6 7.8 8.8 15.0 3.9 16.0 When I put out to sea Verse 24 32 15 37 15 180 303 7.9 8.6 11.0 8.6 8.1 4.0 TABLE XXX AVERAGES OF 10 READINGS BY H The day is done and the dark ness Verse 18 24 35 73 20 17 31 47 265 5.1 6.3 8. 4 5.5 8.0 4.1 4.9 7.2 Falls (from the) wings (of the) night Pause Verse H 20 50 22 57 138 338 5,7 8.6 5.6 12.0 5.1 6.2 EXPERIMENTAL RESULTS 63 . ~5fr- CO OO ^ 90 -^ 00 + 05 <* O O 00 00 IO CO 90 hj g ^ ,-H ^H .-H CO CO "^ ^ . ^ ^ *": H 02 y Q O >O CO C^. r-H CD O5 j CO C^ CO ^O CO P-H rH ll fl 1 05 02 QJ 00 'O ^ >i ^o' 2^ ^ +J v, O CC (M fH 05C005IT- W CO"*COCO (MCOi-HO ~^$ (M -t- c3 <3 O cS -M 7l M ? 0> ^ 03 g M <3J One hour is long, another short, one second only half as long to us as the next. In the end it will be much simpler to talk of a rhythm which is actually present, perceived or per- formed, together with and inseparable from its peculiar feelings, the beats of which are not fixed in time, while its effect upon us is that of a series of equal time intervals. Time not only fails to account for the regularity of verse rhythm ; it also fails to offer a base of distinction between the different types of rhythm. Temporally anapaests are just like dactyls and neither clearly distinguished from iambs. Here the concept of number comes to the rescue only to be followed by unwelcome consequences, and the most satisfactory course is to fall back upon the swing of 6 Compare, for instance, the lengths of the familiar words " the " or " and " in different verses. SIGNIFICANCE OF RESULTS FOR THEORY OF RHYTHM 77 the rhythm itself. The different rhythms form distinct kinds of cycles. It is the perseverance of one of these types throughout a verse or stanza that establishes the rhythm. Each beat, or each swing, brings up another of the same general structure and the same total affective value. Their form depends in part upon the number and character of the syllables, but it also depends very largely upon the disposition of the individual. When one starts to read dactylic verse he finds dactyls provided the syllables can possibly fit that rhythm. He speaks so as to make the segments of equal weight. The rise and fall of his effort is regular and of the peculiar form of that rhythm. The regularity of the motor performance and the equivalence of the resulting feelings leads naturally to the introduc- tion of the impression of temporal regularity ; but that impression is really subsequent to the rhythm itself. VITA The author was born near Greensboro, Georgia, February 9, 1882 ; matriculated in the University of California 1900 ; A.B., with special mention, 1904; M.A., 1905; Assistant in the Psychological Laboratory, 1904-6. Assistant in Philosophy, Columbia University, 1906-7 ; in Psychology, 1907-8. For inspiration and encouragement he is indebted to the teaching and example of Professors Howison and Stratton of Berkeley, and of Professors Cattell, Woodworth, Dewey, Woodbridge and Thorndike, of Columbia. 79 14 DAY USE RETURN TO DESK FROM WHICH BORROWED LOAN DEPT. This book is due on the last date stamped below, or on the date to which renewed. Renewed books are subject to immediate recall. KEC'D LD JUL 3 o Mi LD 21A-40m-4,'63 (D6471slO)476B HBWW General Library University of California Berkeley LD 21-50?n-8,-32