Ed. /Psych. Lib Gidffing On the Conditions of Fatigue in Reading --■Q ^^^i ^^^=5 4 3 T> 2 ^^^=s 8 8 ^^^^So 7 -j 8 THE LIBRARY OF THE UNIVERSITY OF CALIFORNIA LOS ANGELES TK, Vol.111. No. 5. REPRINTED FROM THE LIBRARY September, 1896. Psychological Review J. McKEEN CATTELL Columbia University EDITED BF AND J. MARK BALDWIN Princeton University WITH THE CO-O^Mi-RATION OF ALFRED BINET, ficoLE des Hautes-£tudes, Paris; JOHN DEWEY, University of Chicago; H. H. DONALDSON, University of Chicago; G. S. FULLERTON, University of Pennsylvania; WILLIAM JAMES, Harvard University; JOSEPH JASTROW, University of Wisconsin ; G. T. LADD, Yale University; HUGO MONSTERBERG, Harvard University; M. ALLEN STARR, College of Physicians and Surgeons, New York; CARL STUMPF, University, Berlin; JAMES SULLY, University College, London. CONTENTS. Studies from the Psychological Laboratory of the University of Iowa : On the Effects of Loss of Sleep : G. T. W. Patrick, and J. Allen Gilbert, 469. Studies from the Psychological Laboratory of Harvard University : I. The Relations of Intensity to Duration of Stimulation in our Sensations of Light: James E. Lough, 484. //. Normal Motor Automatism : Leon M. Solo- mons and Gertrude Stein, 492. On the Conditions of Fatigue in Reading: Harold Griffing and Shi HERD Ivory Franz, 513. /ij Accuracy of Observation and of Recollection in School Children: Shepherd Ivory Franz and Henry E. Houston, 531. russion and Reports : Remarks on Professor Lloyd Morgan's Method in Animal Psychology : Hiram M. Stanley, 536. Recognition : Arthur Allin, Mary Whiton Calkins, 542. The Community of Ideas of Men and Women: Amy Tanner, 548. Psychological Literature : Recent French Works {T. Lachelier.A. Fouillde, T. Halleux, Ch.Mirallid, Fr. Paulhan) : A. Binet, 551. Eucken's Der Kampf um einen geistigen Lebensin- halt: A. C. Armstrong, Jr., 556. Ethnology and Anthropology : Livingston Farrand, 558. L'Annde psychologique : H. C. Warren, 562. Leuba's The Psychology of Religious Phenomena : Willkam Romaine Newbold, 569. A New Factor in Evolution: J. McKeen Cattell, 571. Vision: C. L. Frank- lin, E. B. Delabarre, 573. Localization of Touch : Herbert Nichols, 577. Memory: H. N. Gardiner, W. G. Smith, 578. Synopsia: Mary Whiton Calkins, 581. Psychical Research: J. McKeen Cattell, 582. The Emotions: H.N, Gardiner, 583. Epistemology : C. W. Hodge, H. N. Gar- diner, 584. New Books., 587; Notes^ 588. published bi-monthly by THE MACMILLAN COMPANY, 66 FIFTH AVENUE, NEW YORK; AND LONDON. \ 1 1 ON THE CONDITIONS OF FATIGUE IN READING.^ BY HAROLD GRIFFING AND SHEPHERD IVORY FRANZ. The increasing part played by reading in the life of civilized man is a striking characteristic of modern culture. In fact, the man of to-day might be defined as a reading animal. The re- sult of this strain upon the eye has been the wide prevalence of myopia, astigmatism and kindred disorders. But the functions which the optic mechanism is called upon to perform are not abnormal ; the work of the eye differs only in degree from that for which it is fitted. If the eye were never fatigued, myopia would be rare. Yet great as is their importance, we have little exact knowl- edge of the conditions of minimum visual fatigue. Cohn-, Ja- vaP and Weber^ have treated the subject with great fulness, but their work was largely theoretical. CattelP and Sanford® have, however, investigated the subject experimentally, with special reference to the relative legibility of letters. The conditions of visual fatigue are obviously highly com- plex. They may be divided into two classes. On the one hand, we have all those conditions which pertain to the individual reader ; for example, the time of reading, the position of head and eyes, and personal peculiarities, anatomical and physiologi- cal. Opposed to these are certain purely physical conditions. Such are the size and quality of the type, the intensity and quality of the illumination, the color and quality of the paper, ^From the Psychological Laboratory of Columbia University. Read in condensed form before the International Congress of Psychology, Munich, August, 1896. 2 Cohn, T/ie Hygiene of the Rye in Schools, Eng. tr., London, 1886. ■'Javal, Annales d'Oculiste, 79-82; Revue Scientifique, 1881. * Weber, Ueber die Augenuntersuchungen iji den hoheren Schulen zii Darm- stadt^ Referat erstattet d. grossherz. Ministerial, Mdrz, 1881. ^Cattell, Philosophische Studien, III. ^Sanford, Americatt Journal of Psychology, I. i^ tt .^ \) {\ SH HAROLD GRIPPING AND SHEPHERD I. PRANZ. the clearness of the printing, the length of the lines, and the spacing between the letters and lines. It is this latter group of conditions with which we are now concerned. (i) The Size of the Type. Weber investigated the relation of the size of type to legi- bility by finding the maximum rate of reading. He arrived at the paradoxical result that although the rate of reading de- creased for very small type it also decreased when the height of letters was over 2 mm^ By determining the time of ex- posure required for perception CattelP studied the legibility of small Latin letters of different sizes, .7, i.i, 1.8, 2.5 and 5.8 mm. The times found were 3, 1.4, i.i, .7 and .6 for one observer, and 4, 1.7, 1.3, .9 and .7 for the other. The rela- tion is approximately expressed by an hyperbolic curve. The investigations of Cattell we have extended and supple- mented by different methods. By the first method, which we will call the method of rapid reading, we found the rates at which an observer could read printed matter in large and small type. Two passages of the Bible, each containing 622 words, were used. One observer read one passage A, in large type, and another passage B, in small type, and the next observer read the same passages, reversing the order of the type, read- ing A in small type and B in large type. The order in which the experiments were made was also reversed for alternate ob- servers. The time was taken by the observer with a stop watch, but recorded without his knowing the result by one of the writers. The observers were mostly students, five being familiar with experimental psychology. The type was Roman, i. e.^ the ordinary type used in English books. The large type, of which we here give examples, was Pica, 1.8 mm. in height, the small, Peari. .9 mm. in height. In addition to these experi- ments we made some in which the time of reading was con- stant, I minute, the number of words read being determined. Below will be found the ratios of the times and of the num- ber of words read. 1 op. cit. 2 Not given bj the writer, but calculated by us from other data given. CONDITIONS OF FATIGUE IN READING. 515 Table I. — Relative Times for Large and Small Type. Observer. K A Di B S Fi F2 F3 G Hi Ha D, Av. Tx. •77 1.04 .82 .61 .90 .88 .72 — 1.08 .96 1. 01 .92 •qo Ws Wx. .88 1. 00 •91 .42 .65 •94 .80 T =;,— = ratio of time required to read large type to that required to read small type. Ws _ ratio of number of words read in one minute in small type to the number Wl read in large type. In a few additional experiments the observers read at their T natural rates. The resulting ratios ^p- for 4 observers were .87, ^ s 1. 00, .86 and .81, the average .89, being the same practically as that obtained by the other method. Thus it takes on the average about -^-^ as much time to read large type, 1.8 mm., as to read small type, .9 mm. The dif- ference in legibility would probably be much greater were it not that when the small type is read more words can be seen simultaneously. In this way we may explain Weber's paradoxical result. As the size of the letters increases be- yond a certain limit the rate of reading will necessarily de- crease ; but this does not involve an increase of fatigue, as Weber assumed. By a second method we found the relative number of words seen when exposed for J-g- sec. by Cattell's gravity chronometer.^ Phrases of three and four words were pasted on white strips of cardboard and were shown for the time desired by a falling screen. The greater part of the screen was hidden from the view of the observer by a black sheet of paper with an opening where the letters were to appear. The phrases were cut from the books mentioned, the letters being 1.8 and .9 mm. high. None of the words were of more that two syllables. The same phrases were used for large and small type. There were 54 phrases of 3 words and 54 of 4 words, half in large type and half in small. Thus there were 216 -(- 162 words in all. ^ For description of the instrument see op. cit. 5l6 HAROLD GRIPPING AND SHEPHERD I. FRANZ. The experiment was conducted as follows : The observer took his seat in a comfortable chair opposite the instrument and placed his chin upon a rest suitably adjusted, so that his eyes were slightly above the level of the letters exposed, and 30 cm. distant from them. The experimenter (one of the writers) stood behind the instrument so as to adjust the cards with the phrases. When the card was placed the observer fixated a gray cross on the black background of the movable screen directly in front of the letters, and let the screen fall by break- ing the current with a Morse key. He then wrote down what he thought he had seen. A dozen or more practice trials were made before beginning the experiments proper. The observer was, of course, ignorant of the phrases that were to be given. Care was taken not to have a phrase already given in one type repeated immediately in another. Of eleven observers six com- pleted only half of the series. We give below the results for the different observers. Table II. — Percentages of Words Seen; Large and Small Type. Three-Word Phrases. Four-Word Phrases. Observer, S. L. s 1 S. L. l=- H. .22 •56 •39 13 •44 •29 C. .46 •75 .61 59 •75 •79 T. G. .29 •75 •39 23 .60 .38 I. F. .60 •95 •63 80 .88 .88 H. G. .46 .81 .56 66 .96 .69 P. .46 •91 •50 45 •85 •53 L. .10 • 54 .18 iS •32 •56 R. G. .76 •79 .96 48 .68 .70 S. .12 .47 •25 12 •39 •31 A. .68 • 78 .87 55 .69 •79 S. F. •43 •85 •51 59 .81 •73 Average •53 1 .60 Vertical columns S and L give percentages of words seen for small and large type (.9 and 1.8 mm. high). Vertical column =- give ratios in per cent, or the relative legibility X of small and large type. With the observers whose initials are given in block tj'pe the full set of ex- periments (108) were made, only 39 being made on the others. In taking the average the values of A for these five might be weighted. This would change the averages somewhat. CONDITIONS OF FATIGUE IN READING. 5^7 From the above table we see that on the average but little more than one half as many words were seen in small type as in large type. Individual variations are great, but these vari- ations are probably not due to an appreciable extent to individ- ual differences in the relative legibility of large and small type. For good observers the same difference in legibility would give different values of I. This theoretical conclusion is verified by the experiments. By arranging the observers in two groups according to the per- centages seen, the values of I is for the better observers in all cases lower than that of any of the four poorest observers. A few experiments were made with 21 two-word phrases printed in very large type (4+ mm). The percentages of^ words seen correctly by three observers, together with the averages of the same observers for 1.8 mm. type as found from the table above given are as follows : Large Very large P. .88 ".93 L. .43 -64 S. .43 -7° Thus the legibility as shown by this method appears to in- crease regularly with the size. But since the number of words brought within the field of distinct vision decrease with the size, the relation is quite complex. A few phrases (15) of two words each were used with the others. The percentages for two, three and four-word combi- nations were found to vary but little with the number of words. From the table it will be seen that the values of l were about the same for phrases of three words as for those of four words, the averages for phrases of 2, 3 and 4 words in small type be- ing .42, .41 and .43. In the above experiments the paper was not exactly the same for large and small type, being slightly grayish for the small type and of a more yellowish tint for the large. To eliminate this source of error, phrases of four words in large and small type were printed on the same white paper. From 200 experiments (800 words), 100 on S. F. and lOO on H., we found the following percentages of words seen : 5l8 HAROLD GRIPPING AND SHEPHERD I. PRANZ. H. S. F. .12 •83 .90 L •37 .92 = A The values of A correspond quite closely with those pre- viously found for the same observers, .88 for F. and .29 for H. A modification of the preceding method was used b}^ deter- mining the time words composed of letters of different sizes had to be exposed in order to be seen. This we will call the time of exposure method. The same apparatus was used as before, the time of exposure varying with the extent of opening of the screen. This time can be determined to about .150", a being .001 sec. The words were of not less than 5 letters, nor over 2 syllables, on white paper. The type, as here shown, was six point and 'eleven point,' .8 and 1.6 mm. high. On ac- count of the preliminary practice necessary there were but three observers, two being the writers. The experiments were con- ducted in the same general way as those just described. The experimenter tried first very small times, increasing the time until the stimulus was perceived approximately 50% of the time. Then other words were shown which the observer had not seen. As the percentage seen tends to increase very rapidly from o to 100 (theoretically 99+ ), it was generally easy to determine at one sitting the time required either directly or by estimation from the percentage seen. The times of ex- posure found thus are now given in thousandths of a second. Table III. — Times of Exposure for Different Sizes of Type to be Seen. L Observer. L. S. S = ^ G. 1.6 1-9 .84 f I.I 1-5 ■73 " 1-3 1-7 .76 H 2.0 2.8 •71 1 i 1.6 2.5 .64 AV. .73 L and S denote the times of exposure necessary for large and small type re- spectively, .8 and 1.6 mm. ^ or /I is the relative legibilty measured by this method. The two values of L and S for F and H are for different days. The time of exposure seems to vary in the same individual. CONDITIONS OF FATIGUE IN READING. 519 From the above results it appears that the large type, 1.6 mm., requires about ^ as great a time of exposure as the small type of half the height, .8 mm. In the last two sets of experiments a few observations were made, which though not bearing on the special problems under investigation are yet of psychological interest. Observers gen- erally failed to see any of the letters making up a word when they failed to perceive the whole word. There were, however, individual differences, some persons often seeing one or two letters only. At times an observer saw combinations without sense, though he knew such combinations were not given. In the time-of-exposure experiments the observer was at times con- scious of perceiving letters without knowing what they were. Occasionally the observer had an impression that a given word was present, when the letters had not appeared distinctly. More often some letters were distinct, and he guessed the word, or else the whole word was distinct. One of the writers had a marked tendency to see again what had been given before, even when he knew that the word was not repeated. One of the ob- servers, H., seemed to be an exception to the rule that one sees all letters exposed or none at all except within very small range of time. Some days it was very difficult to find the time re- quired for this reason. But perhaps the most important phe- nomenon observed was the illusory perception of a word, the letters appearing distinct when not present. This has been al- ready noted by Cattell and also by Miinsterberg. The theo- retical importance of this lies in the support which it gives to the hallucination theory of perception. The representative pro- cesses in perception seem to attain to the sensory vividness of true hallucinations. This does not, however, appear to take place in every instance, for F. seemed at times to see some of the letters and to infer by ordinary processes of association that a certain word was present. To obtain more extended results and confirm those obtained by Cattell, by the time-of-exposure method, we determined the intensity of illumination necessary for the reading of letters of different sizes. The letters were printed in the simplest kind of type, commonly called Block. Two cards were, how- 520 HAROLD GRIFFING AND SHEPHERD I. FRANZ. ever, covered with words in Roman type, .8 and 1.6 mm. in height. The observer sat in front of a stand from a projecting piece of which was suspended a small pendulum making a vibration in Yi sec. The pendulum swung in front of a screen having an opening where the letters to be seen appeared. The letters were, of course, shown ^^ sec. The letters were posted on card- board strips and these were placed in slits. The paper was the same for the different sizes, pure white. The slits were ar- ranged so that the length of the cardboard exposed was either 15 or 3 mm., according to the size of the letters. For the two largest sizes, and also for the cards on which the words in Ro- man type were shown, the large area was used. The object was to show only one or two letters at a time, except when the Roman type was used, when a larger number was seen. A black screen in front of the pendulum with the necessary open- ing served to prevent distraction of the observer by the move- ment of the pendulum. The observer's eyes were kept at a constant distance (30 cm.) from the stimulus by means of a chin rest. The light was that of a hooded petroleum lamp found to be fairly constant, shin- ing through a square of ground glass 5x5 mm. The light emitted was approximately .02 candle power. The lamp was in a movable box sliding on wheels in iron grooves. Precau- tions were taken to avoid errors from reflected or diffused light. The letters used were in combinations of one to four words in one horizontal line. They were taken from a printer's sample book. The median plane of the observer was approximately perpendicular to the plane of the cardboard to be seen, and the lamp could be moved only in a straight line, making an angle of 45° with the plane of the cardboard. With this apparatus after the observer had remained in the dark room long enough to avoid errors from adaptation (20 to 30 min.), the experiment was made as follows: A card with letters to be exposed was placed in the slit by the experimenter (one of the writers). The observer pushed back the pendulum to a fixed support with his hand, fixated a pencil cross on the cardboard piece fastened to the pendulum directly in front of CONDITIONS OF FATIGUE IN READING. 121 the letters to be seen, and then let the pendulum swing forward, observing the letters as they were shown. As the pendulum swung back it was caught by the observer with the left hand and fixed with a catch. He then moved the lamp nearer with the right hand. At first this was done by the experimenter, but with less convenience and economy of time. This was re- peated until the observer was quite certain he could perceive the letters correctly when exposed but once. The distance of the light from the letters was then read off on a scale. The square of the reciprocal of this distance represents the relative intensity of the illumination. The readings were, of course, taken by the experimenter. For this purpose we used the light from a small candle inside a blackened box shining through a cylindrical tube. Two or three determinations were generally made at one sitting for each of the variables under investigation, including several in addition to the type. Variations in the results made it necessary to average the records of some days separately, as given in the second horizontal columns for F and H. We give below the average values of T, the illumination threshold^ for reading in terms of one candle-meter (CM. ) , or the light of a standard candle at a perpendicular distance of one meter. Table IV. — Illumination Thresholds for Different Sizes of Type. Observer. N H = .9 H= 1.6 H=3.li 1 H =6.0 h= . h = 1.6 Av MV Av MV Av .042 MV Av MV .001 Av .36 MV .04 Av MV G lO 27 .02 .12 .01 .003 .014 .14 .01 F 6 .24 .02 .08 .01 .028 .007 .010 .001 .22 .02 .12 .02 (< 3 •17 •03 •045 .004 .018 .002 .008 .001 •13 .01 .05 .00 H 5 .077 .014 •035 .007 .014 .001 .003 .000 •19 .01 .07 .00 3 •19 .02 .09 .003 •043 .003 .009 .001 •35 •03 •13 .02 H=height of Gothic letters in mm. h=height of Roman letters in mm. N= number of determinations upon which average is based, Av= average. MV=mean variation. /Icos0 ^ Calculated bj the formula T= d2 where A is the candle power of the light, d the distance of the light from the object, and fl the angle made by the normal to the surface. 522 HAROLD GRIFFING AND SHEPHERD I. FRANZ. A graphical representation of the results is shown in the ac- companying figure. The ordinates give the intensity of illumi- nation in candle-meters, and the abscissas the height of the letters in tenths of millimeters. lb 31 Fig. I. hO The curves resemble rectangular hyperbolas, the values of the variables corresponding roughly to the equation, (s— k) i=ki, k and kj, being constants depending upon the individual. As- suming such an equation w^e may infer that after the size of the type has reached a certain limit the increase of size is in direct proportion to the decrease of illumination. The fatigue coeffi- cient increases slowly until the size of the type decreases to about 2-3 mm., after which its increase is more and more rapid. The lowest limit to the size of type in common use should be 1.5 mm. The same conclusion may be drawn from the experi- ments of Cattell already mentioned. (2) Thk Quality of the Type. On theoretical grounds it may be assumed that the legibility of letters decreases with increasing complexity of structure. From this point of view German type is open to serious criti- CONDITIONS OF FATIGUE IN READING. 5^3 cism, and even our Roman type might evidently be much im- proved. Some of our letters have unnecessary features and they are as a rule much more complex in structure than those printed in the so-called Block type. Many letters, such as c and e, are easily confused, and there are decided differences ot legibility. These differences are, indeed, slight and difficult to determine. By finding the percentage of times each letter was seen when exposed for in the last date stamped o ttKTVSESITY of CAUFuKi^i^-^ AT LOS ANGELSS LIBRARY ^PHIET BINDER ^^::^ Syrocuse, N. Y. Slocklon, Calif. ETduc-Psych. Library BF ,,i'/C SOUTHERN REGIONAL LIBRARY D 000 432 887