^' a\ o>- \. ^ "> - "/^ x> '^^•.<^'' .* I^. •x^^ %'' "^ ,/-.;.<^-', .x^^--^. -,v "O * T ^ ■ ."^^ "^^ " * B , ; ■» V- x^^ ■ a , ^ ^. ^ .-.. .; -A ■') N V o '^•- a"^ s '^ ' ° / '' ^^= v^-^ '''^. C> \' -J. ^ ' " A V^.^^^ . >-. .•^^' -^M^'^ ^^•.. A. ."a^^'^/v,'- -^^ .'N^' -^M^\ .^:^^^ ^¥»}^^ .^-^ ^^'^^x:^^ ,^ o 0^ "^.. 0^' N^^^ ■.V -^ - ' .0' :^-^ ^^. v^^ ^%'-:;:.^"-^ ^^ "^ .>«r;~ ^^ ^. '» t ^,^' *4^WA'o •%<^- ..^^ \ '> '^r. 'V "^.^ S^^ i^ .0^ 't/> .^v ^^ v-^' .' .N^" "^. ^•./' :^ ^^. ,^^' o5 -n*- '>. ^^- o .^x\' MANUAL OF MENTAL AND PHYSICAL TESTS A BOOK OF DIRECTIONS COMPILED WITH SPECIAL REFERENCE TO THE EXPERIMENTAL STUDY OF SCHOOL CHILDREN IN THE LABORATORY OR CLASSROOM GUY MONTROSE WHIPPLE, PH.D. Assistant Professor of the Science and Art of Education, Cornell University Author of "A Guide to High-School Observation," "Questions in General and Educational Psychology, "Questions in School Hygiene." WARWICK & YORK, INC. BALTIMORE, U.S.A. 1910 Copyright, 1910 BY Warwick & York, Inc. 'G(,A27ri5:;? MANUAL OF MP:N TAL AND PHYSICAL TESTS PREFACE Hitherto the literature of mental and physical tests has been scattered in numerous journals; the results obtained by different investigators have too often not been compared; indeed, in many cases where the methods have been divergent, comparison has been impossible. In consequence, there have been no recognized standards of procedure and none of performance. Neverehe- less, I believe that the time has now come for the taking of an account of stock, and for the systematization of the available materials. This conviction, which is the outgrowth of my own interest in the experimental study of mental capacities, an interest that has been with me during the past ten years, has been con- firmed by many suggestions from colleagues and friends, who have pointed out that a manual of directions for mental tests would meet a real need, and might further the cause of investiga- tion. More particularly, at the instigation of Mr. C. H. Stoelting, of Chicago, who has undertaken to supply the apparatus and materials prescribed in this volume, I began, in March, 1906, to prepare a small handbook 'of mental tests. The impossibility of adequate treatment of the subject in small compass has, however, necessitated the expansion of that early undertaking into the present work. In the introductory sections of the volume, I have sought t show the general purposes of mental tests, to lay down rules their conduct, and to explain the methods of treating data, this connection I discuss the calculation of measures of g' tendency, measures of variability, indexes of correlatio other statistical constants. In the body of the volume, I have brouglit together, f treatment, some fifty of the most promising tests. In my plan has been to sketch the development of the scribe a standard form of apparatus and method o^ explain the treatment of the data secured, and to set forth the results and conclusions thus far obtained. The tests that I have selected may not prove, ultimately, to be those of most value, but they are, I think, numerous enough, and varied enough in type, to furnish a working basis for investigations for some time to come. In the choice of materials and methods, I have sought to fol- low a middle course; on the one hand to avoid the use of costly instruments of precision and of the elaborate methodology of the psychological laboratory, and on the other hand, to avoid the inexactness of make-shift apparatus and the unreliability of casual, unsystematic observation. My idea has been to supple- ment the exposition of the standard apparatus and method of procedure by suggestions for variations of apparatus or of method, so that each test will be carefully standardized, yet will retain a sufficient degree of flexibility. Doubtless, to some readers, the instructions for the conduct of the tests will seem unnecessarilj^ lengthy and detailed; but experience has convinced me that faulty results are to be traced, in quite the majority of instances, to the neglect of some seemingly trivial detail in the arrangement of the experimental conditions; so that instructions can scarcely be made too explicit in a manual of directions in which standardization is the object. In explaining the treatment of data, my aim has been to make clear the arithmetic of the various formulas, without insisting, in every case, upon acquaintance with the mathematical reasoning upon which the formula is based. And when I speak of "the results and conclusions thus far ')btained," I speak with the intent to make clear what, I am sure, made evident more than once in the text, that this book presents, a closed chapter in the experimental investigation of mental 'ity, but a program of work to be done. icknowledgments for aid should be numerous and ungrudg- ese have been made in part in the text, but in many in- laterial assistance has, perforce, gone without explicit ment. I wish, however, to make clear my indebted- Stoelting Co., for the loan of numerous cuts, to Dr. PREFACE IX Guy L. Noyes, of the University of Missouri, for assistance in the tests of vision, to Dr. H. H. Goddard, of Vineland, N. J., for the adaptation of the Binet-Simon tests to American conditions, to my colleague, Professor I. M. Bent ley, as well as to my wife and to my mother, for the reading of proof, and to my colleagues, Professors Charles DeGarmo and E. B. Titchener, for almost daily advice and encouragement. The inscription of the book to Professor Titchener is in token of my special debt to him as a teacher and as an expositor of the scientific method of attack in the solution of the problems of mental life. Guy Montrose Whipple. Cornell University, June, 1910. TABLE OF CONTENTS INTRODUCTORY Chapter I. The Nature and Purpose of Mental Tests 1 Chapter II. General Rules for the Conduct of Tests 4 Chapter III. The Tre.\tment of Measures 9 THE TESTS Chapter I\'. Anthropometric Tests Test 1. — Height, Standing and Sitting 51 Test2.— Weight 56 Test 3.— Diameter of the Skull 60 Test 4. Girth of the Skull 66 Chapter V. Tests of Physical and Motor Capacity Test J.— Vital Capacity 70 Test e.— Strength of Grip 74 Test 7.— Strength of Back t> 79 Test ^.—Strength of Legs 82 Test 9.— Endurance of Grip • 82 Test iO.— Quickness or Rate of Movement : Tapping 100 Test 11. — Accuracy or Precision of Movement : Aiming 115 Test 12. — Accuracy, Precision, or Steadiness of Movement : Tracing 119 Test 13. — Steadiness of Motor Control : Involuntary Movement 123 xi Xll TABLK OF CONTKNTS Chapter VI. Tests of Sensory Capacity. Test ;^.— Visual Acuity 131 Test.15. — Balance and Control of Eye-muscles: Hetcrophoria 142 Test /&.— Color-blindness 148 Test 17. — Discrimination of Brightness 159 Test 75.— Auditory Acuity 166 Test /9.— Discrimination of Pitch 180 Test ^0.— Discrimination of Lifted Weights 188 Test 21. — Discrimination of Pressure 194 Test ^^.—Sensitivity to Pain 198 Tcs<;?S.— Discrimination of Dual Ciitancous Impressions 207 Cmaptek VII. Tests oe Attention and Perception ^Test ;24.— Range of Visual Attention 222 Test 25. — Visual Apprehension 244 Test .35.— Cancellation 254 Test .27.- Counting Dots 270 Test ;2S.— Reading Complicated Prose 273 Test ^5.— Simultaneous Adding 277 Test SO. — Sinudtaneous Disparate Activities 279 Chapter VIII. Tests ok Description and Report Test Si.— Description of an Object 286 TestS2.—Y\dii\\{y of Report : Anssane Test 292 Chapter IX. Tests of Association, Learning, and Memory Test 33. — Uncontrolled Association (Continuous Method) 313 Test 34. — Controlled Association: Part-Wholes, Genus-Species, and Opposites 319 Test 35. — Controlled Association: Computation 327 Test 36. — Learning: Habit-Formation in Mirror-Drawing 343 Test 37.— Learning : Substitution 350 Test 38. — Memory for Serial Impressions: 'Rote' Memory 356 Test 35.- Memory for Ideas : ' Logical' Memory 394 Chapter X. Tests o^ Suggestibility Test 40. — Suggestion by the Size-Weight Illusion 405 Test 41. — Suggestion by Progressive Weights 410 Test 4^.— Suggestion by Progressive Lines 414 Test 43. — Suggestion of Line-Lengths by Personal Influence . . . . : 419 7V.S/ 4.^.— Suggestion by Illusion of Warmth 423 TABI.K OF CONTENTS XI11 Chaptku XI. Tksts of Imagination and Invention Test 45— Ink-Blots 430 Test Ifi. — Linguistic Invention 435 Test -(7.— Word-Building 441 Test 45.— Ebbinghaus' Completion-Method : Mutilated Prose Test 445 Test 49.- — Interpretation of Fal)les 454 Chapter XII. Tests of Inteij^ectual Equipment Test 50— Size of Vocabulary 45S Test 51. — Range of Information 465 Chapter XIII. Serial Graded Tests for Developmental Diagnosis Test 52. — De Sanctis'- Graded Series 469 Test 53.- Binet-Simon Graded Tests : 1905 Series 473 Test 54.— Binet-Simon Graded Tests: 1908 Series 493 List of Materials 518 Index of Names 521 Index of Subjects 527 INDEX OF TABLES TABLE PAGE 1. Strength of Grip, in Hectograms, 50 Boj's (Whipple) 10 2. Values Derived from the Data of Table 1 ll 3. Distribution of the Heights of 12-Year-Old Boys 13 4. The Numerical Smoothing of the Distribution of Table 3 22 5. Probable Error of ?• for various Values of r and of n (Yule) 32 G. Conversion of ii-Values into r-Values in Accordance with Form- ula 33 36 7. Correlation of Right and Left-Hand Grip, by Group Averages (Whipple) 37 8. Relation of Deafness and White Color in Cats (Yule) 39 9. Corresponding Values of r and U for Formula 40 (Whipple) 40 10. Norms of Stature of American Children (Boas) 53 1 1 . Norms of Standing and Sitting Height (Smedley ) 53 12. Norms of Weight (Burk) 58 13. Norms of Weight, with Clothing (Smedley) 58 14. Diameters of the Skull and the Cephalic Index (West) 63 15. Breadth of Head by School Grade (Porter) 64 16. Skull Dimensions and Proportions of Entering Classes at JNIunich (Engelsperger and Ziegler) 65 17. Circumference of the Head (MacDonald) 67 IS. Norms of Vital Capacity (Smedley).* 71 19. Value of the Vital Index, when Weight is Taken as Unity (Kotel- mann) * 72 20. Norms of Strength of Grip (Smedley) 76 21. Types of Endurance in Dynamometer Trials (Binet and Vaschide) 92 Opposed Types of Endurance (Binet and Vaschide) 92 Sample Record of a Tapping Test (Wells) 107 Dependence of Rate of Tapping upon Age (Smedley) 109 Test Numbers for Auditory Acuity (Andrews) 170 Pitch Discrimination of 167 Children (Seashore) 184 Dependence of Discrimination of Lifted Weights on Age (Gilbert) 192 28. Pain Limen for 50 Boys and 50 Girls of each Age (Gilbert) 204 29. Topography of Esthesiometric Sensitivity (Weber) 212 30. Average Number of Letters Read Correctly in One Exposure- ' (Whipple) 232 31. Effect of Practise on the Perception of Letters (Whipple) 236 INDEX TO TABLES XVll 32. Relation of Visual Range of Attention to Age (GriffingJ 236 33. Individual Differences in Visual Apprehension (Whipple) 251 34. Effect of Practise upon Visual Apprehension (Whipple) 251 35. Effect of Letters and of Fatigue on Cancellation (Whipple) 262 36. Averages and Variations in Cancellation Tests (Whipple) 263 37. Correlations in Cancellation Tests (Whipple) 264 38. Effects of Fatigue on Cancellation (Ritter) 265 39. Relation of Average Number of Letters Cancelled to Intelli- gence (Winteler) 266 40. Effect of Different Methods of Reaction in the qrst-Test (Whipple) 268 41. Specifications for Test-Cards Used in Dot-Counting 271 42. Results of Reading Tests (Whipple) 275 43. Simultaneous Reading and Writing (Sharp) 281 44. Comparative Accuracy of Sworn and Unsworn Statements (Stern and Borst) 305 45. Effect of Time-Interval on Range and Accuracy of Report (Borst) 307 46. Dependence of Report on its Form (Stern and Borst) 308 47. Effect of Practise on Coefficients of Report (Borst) 310 48. Distribution of Terms in 'Uncontrolled' Association (Jastrow, Nevers, Calkins) 317 49. Influences which affect 'Uncontrolled' Series of Words or Draw- ings (Flournoy) 318 50. Normal Performance in the Part-Wholes Test (Norsworthy) 321 51. Normal Performance in the Genus-Species Test (Norsworthy) 322 52. Normal Performance in the Opposites Test (Norsworthy) 326 53. Correlation of Opposite Tests with other Tests (Aikins, Thorn- dike and HubbelD- 327 54. Efficiency in Addition: Five 10-Minute Periods (Schulze) 336 55. Efficiency in Addition and Multiplication (Burgerstein) 336 56. Efficiency in Computation within a School Session (Laser) 337 57. Additions made per Pupil with and without a Rest-Pause (Burgerstein-Schulze) 338 58. Effect of Pauses upon Computation (Friedrich) 339 59. Effect of Fatigue on Arithmetical Work in Evening Schools (Winch) 340 60. Effect of Practise on Speed in Mirror-Drawing (Whipple) 346 61. Substitution Test. Number of Symbols Written, Form B, Group Method (Whipple) 353 62. Substitution Test. Speed in Seconds, Form B, Individual Method (Whipple) 353 63. Substitution Test. Bright and Dull Boys. Individual Method (Whipple) 354 1 INDEX OF TABLES 8ul)stitution Test. Distribution of Gains and Losses in Speed (Whipple) 355 Use of the Footrule Method in Scoring the Memory Test (Spear- man) 367 Norms of Memory Span for Digits as Conditioned by Age (Smedley) 374 Development of Memorj^ for Digits (Smedley) 374 Memory for Letter Squares in Relation to Age and Practise (Winch) 375 Percentage of Accuracy in Memory for 2-place Numbers (Schuyten) 375 Sex Differences in Memory Span for Digits (Wissler) 376 Net Efficiency of various Memories, in Relation to Age (Pohl- mann) 376 Dependence of Memory Span for Auditory Digits on Age (Jacobs) 377 Dependence of Memory for Auditory Digits on Age (Ebbinghaus) 377 Memory for 9-Term Series of Different Kinds (Lobsien) 380 Norms for Memory of Related and of Unrelated Words (Norsworthy) 381 Dependence of Memory upon Form of Presentation (Pohlmann) 382 Relation of Memory for Auditory Digits and Intelligence (Ebbinghaus) 386 Relation of Memory for Digits and School Standing (Smedley) . 388 Memory Span for Digits in the Feeble-Minded (Johnson) 389 Comparative Memory Capacity of Normal and Feeble-Minded Children (Norsworthy) 390 Recall of Different Members of a 7-Term Series (Binet and Henri) 390 Old Homestead Test: Words Written and Underlined (Whipple) 399 Percentage of Loss in Third Reproduction (Henderson) 400 Force of Suggestion (Gilbert) 408 The Progressive-Weight Suggestion (Binet) 413 Percentage of 'Yields' to Contradictory Suggestion (Binet and Henri) 421 Suggestibility to Warmth: Resistance-Coil IMethod (Okabe and Whipple) 427 Suggestibility to Warmth as Related to Age (Guidi) 428 Average Number of 'Names' given to Ink-Blots (Kirkpatrick). 433 Scores of Seven Adults in Developing Sentences (Sharp) 438 General Results in Word-Building (Whipple) 442 Sex Differences in Word-Building (Whipple) 445 Dependence of the Completion Test on Maturity (Ebbinghaus) 451 INDEX OF TABLES XIX 94. Average Vocabulary in Relation to Scholastic Status (Kirk- patrick) 461 95. Distribution of Corrected Vocabulary-Index (Whipple) 462 96. Overestimation of the Vocabulary-Index (Whipple) _ 462 97. Dependence of Range of Information on Age (Whipple) 466 98. Dependence of Range of Information on Sex (Whipple) 466 99. Results for 25 Children in the Binet-Simon Tests (Decroly and Degand) 491 100. Critique of the 1908 Binet-Simon Tests (Decroly and Degand) 515 LIST OF ABBREVIATIONS A. G. P. Archiv fiir die gesammte Psychologic. A. J. P. American Journal of Psychology. A. P. Annee psychologique. Ar. P. Archives de Psychologic. B. J. P. British Journal of Psychology. C. C. Columbia University Contributions to Philosophy, Psychol- ogy, and Education. E. P. Die experimentelle Padagogik (after vol. 5, Zeitschrift fiir experimentelle Padagogik.) International Magazine of School Hygiene. University of Iowa Studies in Psychology. Journal of Educational Psychology. Psychologische Arbeiten. Psychological Bulletin. Psychological Review. Psychological Review Monograph Supplement. Philosophische Studien. Pedagogical Seminary. Schiller-Ziehen, Sammlung von Abhandlungen aus dem Gebiet der padagogischen Psychologic und Physiologic. Report United States Commissioner of Education. Studies from the Yale Psychological Laboratory. Zeitschrift fiir angewandte Psychologic. Zeitschrift fur Psychologic und Physiologic der Sinnesorgane. Zeitschrift fiir padagogische Psychologic. Zeitschrift fiir Schulgesundheitspflege. I. M. S. Iowa S. J. E. P. P. A. P. B. P. R. P. R. M, Ph. S. Pd. S. s. Z. u. , S. Yale S. Z. A. P. z. P. z. P. P. z. S. INTRODUCTORY chaptp:r I The Nature and Purpose of Mental Tests' When we speak of a mental test we have in mind the expei i- mental determination of some phase of mental capacity, the scientific measurement of some mental trait. The mental test in some respects resembles, in some respects differs from the typical experiment of the psychological laboratory. Like this latter, the test is superior to the casual observation of everyday life because it is purposeful and methodical: it thus l^ossesses all the merits common to experimental investigation at large, viz: the control of conditions (including the elimination of disturbing, and the systematic isolation of contributory factors), the possibility of repetition, and the possibility of subjecting the obtained results to quantitative treatment. Unlike the typical experiment of the psychological laboratory, the mental test ordinarily places little or no emphasis upon intro- spective observation by the subject, in part because of its rela- tively short duration, in part because it is frequently applied to inexperienced subjects who are incapable of aught but the mcst elementary introspection, but more especially because it is con- cerned less with the qualitative examination or structural analysis of mental processes than with the quantitative -determination of mental efficiency; because, in other words, it studies mental per- formance rather than mental content. It is also Coihmonly sim- pler in form than the psychological experiment. ' The tests mth which this volume is concerned are mainly mental tests. Since, however, the intimacy of the relation between mind and body makes it well-nigh imperative to study their interrelations, attention has been paid to the more important anthropometric measurements and to those tests of physical capacity that have most frequently been used in the search for corre- lations of psychical and physical ability. 2 NATURE AND PURPOSE OF MENTAL TESTS The purposes for which mental tests have been developetl are, of course, varied, but, roughly speaking, we may distinguish a theoretical interest on the part of laboratory psychologists, and a practical interest on the part of those who are concerned" with mind at work in everyday life. Historically, it appears that most of the tests now in use have originated in the psychological laboratory, either in the natural course of the development of experimental psychology as a system, e. //., the usual tests of sensory discrimination, or as a consequence of special attempts to study mental capacity, particularly the interrelations of various mental capacities and of mental with physical capacities. It is, we think, not too much to hope that in time the application of mental tests will bear rich fruit in this field. We may hope that the skillful study of mental functions by the test-method may supply us with a satisfactory account of the nature and interrelations of mental functions, just as the typical introspective experiment has been able to furnish an account of the structural make-up of mind. If we could, to take an in- stance, obtain an exact science of mental functions so that we could know the unit-characters of mind as the biologist knows, or expects to know, the unit-characters of plants and animals, the study of mental inheritance would be carried apprecialjl}' forward. Outside the laboratory an active and ver}' natural interest in mental tests has been exhibited by those who are busy with practical problems to the solution of which the scientific study of mind may be expected to contribute. It is, naturally, the edu- cator to whom the development of a significant and reliable system of mental tests would most appeal, since he is concerned with the development of just those capacities of mind that these tests propose to measure. There has been, unfortunately it seems to us, a disposition in some quarters to speak as if a science of mental tests was already achieved; as if, for instance, a child's native ability could now ])e measured as easily as his height, as if his suggestibility or his capacity for concentration of attention could be determined as readih' as his skull circumference or his breathing capacity. To make such assertions is surelv misleading, foi', as the study of NATURE AND PURPOSE OF MENTAL TESTS 6 the tests herein embodied will show, there is, at the present time, scarcely a single mental test that can be applied unequivocally as a psychical measuring-rod. The fact is we have not agreed upon methods of procedure; we too often do not know what we are measuring; and we too seldom realize the astounding complexity, variety, and delicacy of form of our psychical nature. Paradoxical as it may seem, these are the reasons, we believe, that render the elaboration of a scientific system of mental tests a possibility, for, if the all-too-evident lack of agreement in the results of the investigations already made is not attributable to faulty or divergent methods, or to clumsiness and ignorance,- — if, in other words, the discrepancies are inherent and ultimate, — ■ then we never can have a science of mental tests. What we need is not new tests, though they are welcome enough, but an exhaustive investigation of a selected group of tests that have already been described or proposed. In particular, we need more than anything else, at least from the point of view of application, the establishment of norms of performance for these tests, ^ — norms that are based upon investigations in which standard and prescribed methods of procedure have been fol- lowed in a rigid and undeviating manner. This book is an attempt to assist in the realization of this need . It presents a program of work, rather than a final system of results. CHAPTER II GioNKUAL Rules for tiik Conduct of Tksts The followinfi' general lulos may bo laid tlown at the outset. (1) The essential and funchmK^ntal principle underlying the conduct of scientific tests is the standardimtion of conditions. This does not mean that ex{)ensiv(^ a})i)aratus oi- instruments of precision are always necessary, but simph' that Ihe conditions under which a test is given to one jierson oi- to one g'roup of per- sons nuist l)e identically followed in iiiviuij; the same test to another |)erson or i^'rouj). We cannot always make the conditions ideal, l)ui^ we can at least try to keep them constant, if fh(^ conditions are vai'icMl, they nnist l)e varied intentionally and for a (h^finite purjios<'. (2) No detail in the 'setting' of a test is too trivial to 1)0 neglecteth This is, of course, merely a restatement of the ])i'evious principle in' another form. It is noteworthy that the lack of accordance between the results obtained by different investigators in the use of what is ostensibly the same test almost invariably turns out to be due to seemingly trivial variations in the method of adminis- tering the test. In particular, attention may be calliMl h(>r(^ to such mattei's as the time of day at which the experim(>nt is nia(l(>, the nature of the instructions that ]-)recede the test, the (Muotional attitude of the |)articipants toward the investigation, their ability exactlyto comprehend what is wanted of them (of wiiich more hereafter) and their willingness to do their best tbroughout the test. It is well to write out the preliminai-y instructions and to memorize t l\em, after fii'st making a trial in oi-der to see if they are perfectly intelligible. Thus, for instance, to say to one class of school childi-en: "Cross out all the r's on this ]iaper while I take your time with a watch," and to another class: "Cross out all the fi's on this pajuM- as fast as yi)u can" may mean the same thing to GENERAL RULES 5 the experimenter, but it will not bring the same results from the classes under investigation, because in the second case the idea of fast work has been more strongly emphasized. (3) No test should be undertaken by the examiner, E, until he is perfectly familiar with its nature, its purpose and its administration. Especially if it involves the use of apparatus, he should familiar- ize himself with the manipulations until they become automatic. (4) No test should be undertaken until the subject, S, is per- fectly clear as to what is required of him. Since most mental tests are of an unfamiliar character, something beside explicit instructions, however clearly put, is needed to enable the avei-age *S to undertake the test under proper conditions. Ordinarily, a brief period (say 1 to 5 minutes) of fore-exercise is needed to remove timidity, excitement or misunderstanding. If this prelimi- nary exercise is propei-ly arranged (especially by being based upon material not usfnl in the test proper, and by being of the same length and character for all *S's), it does not introduce a serious practise error, while it does decidedly facilitate the test. In some cases, however, as, for instance, when the facility of adaptation to the test-conditions is itself an object of investigation, the fore- exercise should be omitted. (5) E should be on the look-out for external signs of the way in which S responds to the test, i. e., for indications of readiness, of quick comprehension, of a competitive spirit, or of ennui, fatigue, distraction, shift of attention, trickiness or deceit. The record -blanks should have a space for the recording of remarks of this nature. When tests are conducted individually it is surprising how much can be gleaned in regard to S's general intelligence, it is largely upon this sort of observational record that E must depend for his estimate of this general intelligence, even though the test be supplemented by school marks, the estimates of teachers, and similar devices. (6) Most mental tests may be administered either to individ- uals or to groups. Both methods have advantages and disadvan- tages. The group method has, of course, the particular merit of economy of time; a class of 50 or 100 chiklren may take a test THE CONDUCT OF TESTS in less than a fiftieth or a hundreth of the time needed to adminis- ter the same test individually. Again, in certain comi:»arative studies, e. g., of the effects of a week's vacation upon the mental efficiency of school children, it becomes imperative that all ;$'s should take the tests at the same time. On the other hand, there are almost sure to be some /S's in every group that, for one reason or another, fail to follow instructions or to execute the test to the best of their ability. The individual method allows Eio detect these cases, and in general, by the exercise of personal super- vision, to gain, as has been noted above, valuable information concerning 14 THE TREATMENT OF MEASURES the measures serially, as one-half of the measurements may be checked off by inspection. Its primary disadvantage is that it gives little -weight to extreme deviations and may fail entirely to represent the type, yet, in many psychological observations, it is precisely these extreme deviations which are most suspicious, so that this tendency of the median to lessen the significance of extreme, measures may prove a positive advantage. In general, the longer the series or the more homogeneous the values, the more nearly does the median approximate the mean. S. The Mode. If a numl)er of measurements are distriljuted in ascending or descending order, a mode is a measure that appears more frequently than do measures just above or below it in the series. There may be several modes in a distribution, though usually there is but one, and we may therefore define the mode as the commonest single value, or the commonest condition. Many statistical arrays find a better representative value in the mode than in the average. Thus, when we speak of the ''average American citizen," we really have in mind the typical citizen, the one most frequently met with. To borrow an illustration from Rietz (4, p. 684): " If a community has 10 millionaires, but all the other citizens are in poverty, an arithmetical average might give the impression that the people of the community are in good financial condition, while really the 'aveiage citizen' is in poverty. " The primary use of the mode is therefore, to characterize a type. Strictly speaking, we may have an empirical mode, as indicated in a given array and a theoretical mode, which would be the most frequent condition in a theoretical distribution. The latter is difficult to compute and not often employed. If an array is very irregular, there is, in strictness, no mode or type at all, or at least the indicated mode has little significance. In Table 3, it is clear that the mode is 142 cm., because this measure appears 150 times, and no other measure is as frequent. MEASURES OF VARIABILITY 15 B. MEASURES OF VARIABILITY It is a common fallacy to rest content with the statement of the general tendency of measurements. Even in supposedly accurate and scientific determinations, we may find the quantita- tive expression limited to averages, e.g., "the mean temperature for September," "the average weight of 12-year old boys," etc. But it is evident that the average gives no indication of the dis- tribution of the individual measures from which it is obtained, no indication of the extent to which these measures vary or deviate from the average, no information as to how homogeneous is the material that the average represents. The September temper- ature may have been seasonable and equable or there may have been some days of frost and some days of sweltering heat. Again, if five individuals weigh 80, 65, 60, 40, and 55 kg., respectively, and five others 62, 59, 60, 51, and 58 kg., respectively, then the mean weight of either group is 60 kg., but one group is distri- buted very closely around the mean, whereas the other group exhibits such marked deviations from it that M (or any other gen- eral tendency measure) has little or no significance as a repre sentative value. From this it follows that we need not only measures of general tendency, but also measures of the variability or tendency to deviation of measurements, and that these latter are of well- nigh equal importance. There are three common measures of variability, ^ — ^the average deviation, the standard deviation, and the probable error." 1. The Average Deviation {Mean Variation) To find the average deviation we must first find the mean, M, (or median or mode); second, substract each individual measure m, algebraically from M , which gives a series of deviations, d; ^ Besides these measures, range of variability is sometimes indicated roughly by stating the maximal and minimal measurements, in conjunction with M. This gives us, at least, information as to the extremes of deviation. ^ ft is well to avoid confusion here at the outset. The average deviation (A. D.), as used by the statisticians, is identical with the mean variation (m. v.) of experimental psychology. The standard deviation {o) is called the average error by Sanford, the mean error by Merriman, and the error of mean square by others. 16 THE TREATMENT OF MEASURES third, find the average of these deviations, i. e., the mean of the variations, by summating without reference to sign and dividing by the number of cases. Hence: A.D. or m.v. = ^ ~ ^ ^ ^ -i A.D. d, + ^2 + ■■ dn n A.D.--^-' (4) n Reference to Table 1 will render this i)rocess clear: there the average right hand grip is 283; the weakest boy has a standing of 158, hence he deviates 125 units from the average; the first 28 boys rank below average and therefore exhibit minus devia- tions, the rest are above average and exhibit plus deviations; all these deviations are added without regard to sign and their sum, 3088, is divided by the number of cases, 50, yielding a mean variation of 61.4 hectograms. If the median were selected as the representative value, the variability would, of course, be computed similai-ly witii a new series of rf's. 2. The Standard Deviation {Error of Mean Square) This measure of variability is preferred by many experimenters and is practically the only one employed by statisticians, as it is thought to be more accurate than the average deviation, but it is much more laborious to compute. It is the square root of the average of the squares of the individual deviations S. D., or o I d' +d: + dl+ ... d\ J^^. (5) MEASURES OF VARIABILITY 17 If n is small, the formula is often modified by wi-iting n-1 in place of n:i Hence: (6) The application of Formula 5 is illustrated in Table ], 5th and 9th columns, whei-e the squares of the individual deviations arc shown in detail. The sum of these squares foi the right-hand grip is 287,884. This is divided by 50, giving 5757.7, the square I'oot of which is 75.8, the a desired. The S. D. of a given series is somewhat larger than its A. D. Theoi-eticall}^, and practically if the distribution he symmetrical and the observations sufficiently numei-ous, the relation is constant at (7 = 1.2533^.2). (7) Conversely, yl.Z). =0.7979 (7. (8) As shown in Table 2, the S. D. computed by Formula 7 is closely similar to that computed by Formula 6. 3. The Probable Error The probable error of a single measure (P. E.) is a measure of the limits above and below M (or other representative measure) that will include one-half of the individual measures; in other words, it is a value such that the number of measures that exceed it is the same as the number of measures that fail to reach it. 2 'For the reasons for tliis substitution, consult Merriman (p. 71). It is evident that the effect of the substitution becomes progressively less as n increases: as will be seen in Table 2, the difference between Formula 5 and Formula 6 is practically negligible when n = 50. ^ The term 'probable error' is often a source of confusion to those unfamiliar with its use in mathematics. The magnitude in question is not, of course, the most probable error, neither is it, from our point of view, an 'error' at all. For a descriptive term, we might call the probable error the median deviation since it is that deviation that is found mitlway from the representative value in either direction. 18 THE TREATMENT OF MEASURES The P. E.is appi-oximately two-thirds the S. D., or more exactly. P. E. = 0.6745 a. (9) ]^y reference to Formula 5 this becomes: P.E. = 0.6745 J^-^' (10) ^^ n or, for a small miml^er of cases (Formula 6): P.E. = 0.6745 JA-^^I". (11) ^ n- 1 In practise we may find the P. E. approximately, if the dis- tribution be assumed to be normal (see under D, below), by count- ing off one-fourth of the cases from either end of a series of measure- ments, and halving the difference between the two values thus found. P.^. = ^--~^^" > (12) Thus in Table 1, these limits lie at the 12th and a half and the 37th and a half measurements, and have, for the right-hand grip the values 222.5 and 324, respectively; hence, P. E. =324 - 222.5 -=- 2 = 50.7, — a value that is approximately the same as the values of P. E. computed by Formulas 10, 11 , and 13 (Table 2). By Formula U, P. E. = 0.6745 X 76.5 =51.6. Corresponding values are given in Table 2 for the left-hand grip as distributed in Tal)le 1. Still other values might be computed on the basis of the median instead of the mean. By combination of Formulas 7, S, and 9, we may olitain for a normal distril)ution: P.E. = 0.8453 A.D. (13) S. D.= 1.4825 P.E. (14) A.D.= 1.1843 P. E. (15) The first of these is illustrated in Table 2. GRAPHIC REPRESENTATION 19 4. The Coefficient of Variability If it is desired to compare the variability of one series of measure- ments with that of another, it will be found that, as a rule, their respective measures of variability cannot be compared directly, because they are based upon different units or at least upon dif- ferent measures of general tendency, but the relations of the two measures of variability to their respective measures of general tend- ency can be directly compared. In other words, we can com- pute two coefficients of variability (C) by dividing in each serifes a measure of variability by a representative measure, i. e., either S. D., A. D., or P. E., may be divided by either mean, median, or mode. Unless otherwise specified, it may h& assumed that S. D. is divided by M. Hence: C=^ (16) M Thus, in Table 1, for strength of right hand, C = 76.5 - 283 = .27 and for strength of left hand, C = 80.2 -^ 273 = .29, hence the latter series is slightly more variable. C. THE GRAPHIC REPRESENTATION OF MEASUREMENTS A series of measurements, as we have seen, can be expressed adequately by a single representative value only when that value is accompanied by some measure of variability. Even these two values may fail to express the series completely, since they aire, after all, only symbols for the convenient summarizing of general tendency and variability, whereas a complete numerical expression of a series of measures would imply the tabulation of all the data of the series. Such a tabulation is for the most part impracticable, or at least of little significance, because of the difficulty of grasp- ing the nature of the series by the inspection of a mass of figures. The use of the graphic method, however, supplies a most ser- viceable- and effective means of showing at a glance all of the important features in the distribution of a series of measurements and likewise of relations between series of measurements. 20 THE TREATMENT OF MEASURES 1. The Plotting of Frequencies or Graphs of Distribution The most usual form of graph for iUustrating the distiibution of a series of measurements is constructed as follows: Draw two lines, OY and OX (Fig. 1) in the form of coordinate axes, i. e., with OF perpendicular to OX. Upon the horizontal, or a:-axis, lay off convenient intervals corresponding to the units of measurement of the series to be plotted; upon the vertical, or y-axis, lay off intervals corresponding to the frequencies of the series. The choice of the scale units is largely arbitrary. The intervals of the two axes need not be the same, nor need different graphs, save for purposes of direct comparison, be plotted to the same scale. In general, a scale should be selected that will bring the surface easily into view as a whole and that will render conspicuous the features that are under consideration. Thus, if one is studying rate of increase or decrease, a scale should be selected that affords a fairly steep curve in order to emphasize its rise and fall. 'Squared' or cross- section paper (usually laid off by mm. on sheets 15 x 20 cm.) may be purchased for curve-plotting, and will be found invaluable for this work. In illustration, the numerical table of frequencies above (Table 3) is turned into a surface of frequency upon the axes just men- tioned (Fig. 1). We mark off on the x-axis, it will be seen, 18 equal intervals corresponding to the range of dimensions, 126, 128, . . . 160 cm. Uponthe?/-axis we mark off equidistant intervals for the range of frequencies from 1 to 150. We next locate the series of 18 points. The first point lies vertically above the 126 cm. mark at a distance equal to 1 of the vertical units; the second lies vertically above the 128 cm. mark at a distance equal to 5 vertical units, etc. By joining the 18 points thus located, the resulting line evidently gives in a single visual impression the distribution that was expressed numerically in Table 3. Any point in this line is fixed by stating its abscissa or distance from the y-axis, and its ordinate, or distance from the rc-axis.^ Now it would have been equally feasible to have considered the val- ues in Table 3 in terms of their deviation from the mean, median or mode, and with little or no change in the curve. Take, for simplic- * The 'curve' is sometimes so drawn as to form the tops of a series of columns erected at the intervals on the base-line, instead of by joining the single points as here described. See, for illustrations, Thorndike (21, p. 48, or 20, p. 15). GRAPHS OF DISTRIBUTION 21 ity, the mode, 142 cm., as the representative value. Erect an ordi- nate of the value of 150 at a point M on the a:-axis (Fig. 1) ; inter- vals to the right of this ordinate may now represent positive deviations ( -f 2, + 4, + 6, etc.) while those to the left represent negative deviations ( — 2, — 4, —6, etc.), as indicated in Table 3. It thus becomes possible to represent negative values graphically. Y Tl60 126 III ISOi FIG. 1. GRAPHIC REPRESENTATION OF THE DISTRIBUTION OF TABLE 3. 2. The 'Smoothing' of Distributions Ordinary measurements are subject to numerous disturbing factors; our units of measurement are often coarse; our oppor- tunities for securing data are always restricted; variable factors of one sort or another obtrude themselves, — and these disturbances produce irregularities in the resultant data. The obtained dis- tribution, in other words, does not coincide with the true distri- 22 THE TREATMENT OF MEASURES bution, i. e., with the distribution that would theoretically appear under ideal conditions. Thus, in Table 3, chance may have led to the measurement among the 1,000 cases studied, of more boys of a certain height, say 144 cm., than we should ordinarily have encountered in measuring 1,000 pupils taken at random. Or, to take an instance of a striking artificial distortion, in the census returns, people who are 39 or 41 years of age show a tend- ency to report their age as 40, so that the age of 40 has an un- naturally large frequency. Minor deviations from the theoretically expected distribution may be counteracted if we are constructing a frequency graph by 'smoothing' the curve, i. e., by drawing the connecting line in the form of a true curve rather than a broken straight line: such a curve will pass in the neighborhood of the several points which have been located by the numerical data, but will not necessarily pass exactly through these points. The result is a graph that shows how the data would presumably have been distributed if the factors which produced the distortions and irregularities were eliminated. TABLE 4 The Numerical Smoothing o/ the Distribution of Table 3 Centimeters 126 128 130 132 134 136 138 140 142 144 146 148 150 152 ■" 156 us 160 Original . . . Smoothed . 1 2 5 6 14 14 24 26 39 40 58 64 96 91 120 122 150 137 142 138 123 108 88 91 63 62 36 41 23 24 12 13 5 6 1 2 ■ In tabular work these deviations may be counteracted by a simple arithmetical process. Replace each frequency except the two extreme ones by the mean (to the nearest integer) of the given frequency and the one just below it and the one just above it; replace the two extreme frequencies by the mean (to the nearest integer) of the given frequency taken two times and the adjacent frequency taken once. If necessary, a second smoothing may be made of the values obtained by the first smoothing. The values of Table 3 do not exhibit marked irregularities as is evident from their graphic distribution in Fig. 1 : the process of Smoothing may, however, be illustrated by the treatment in Table 4. NORMAL DISTRIBUTION 23 D. NORMAL AND OTHER TYPES OF DISTRIBUTION: THE PROBA- BILITY SURFACE AND ITS APPLICATIONS 1. The Normal Frequency Surface Assume that errors of observation have been ehminated and that a large number of measurements of some psychological trait or capacity have been secured: experience has shown, and theoretical considerations likewise indicate, that as a rule these measurements will distribute themselves in the form of a symme- trical bell-shaped curve, variously known as the probability curve, the curve of error, Gauss' curve, or the normal frequency surface, — the salient characteristics of which are a maximal frequency at M with a series of positive and negative d's, from M that are symmetrically disposed on either side of it and whose frequency decreases progressively as their size increases. Such a distribution implies the operation in the conditions that underlie the feature or trait under measurement, of an indefi- nitely large number of individual factors, each of which is equally likely to be present and effective. When, however, there are limiting or restricting conditions, or when one or more factors are present oftener than mere chance would allow, the resultant distribution will tend to depart from the normal type. Thus, the chances of death at different ages are not distributed according to the normal curve, but are higher in infancy and old age than in youth and middle age. The mental ability of college students is not likely to be distributed like that of the non-college population of the same age on account of the selective influence of entrance requirements.^ In general, distributions that do not conform to the normal type are termed 'skewed' distributions, and may demand special mathematical treatment. 2. Relation of the Normal Curve to S. D. and P. E. The normal surface of frequency has interest still further because in it the significance of P. E. and of S. D. becomes clear. In fact, the latter bears to the curve a relation like that of a radius ' On the application of the normal curve to the grading of college students, see M. Meyer (11). 24 THE TREATMENT OF MEASURES to its circle. If *S. D. is small, the measurements are relatively homogeneous and the curve is steep and compact (right-hand curve in Fig. 2), whereas, if S>. D. is large, the curve is broad and of easy slope (left hand curve in Fig. 2). If M and S. D. are known, the entire curve for a normal distribution is known. If the distribution is not of the normal form, the S. D. still remains a good measure of its variability, though not completely descrip- tive of the entire distribution.^ The geometrical explanation of the P. E. is simple. In Fig. 2 we draw the ordinates ah and cd equidistant from OY and at such a distance that the area obYcd is equal to the remainder of the total area under the curve: then the abscissa Oa or Od repre- sents the value of P. E., i. e., a deviation from the mean that will include one-half the total deviations. a. O d «x O d FIG. 2. TYPICAL CURVES OF NORMAL DISTRIBUTION. 3. P. E. of M and of Other Measures Since our opportunities for securing data are limited, it follows that even averages may fail to be absolutely exact measures of the general tendency of the trait under measurement. To revert to the hypothetical data of Table 3, we were there able to obtain an M, 142.9 cm., of the height of 12-year old boys: it must be evident that if we could have measured a million boys we should feel ' This mathematical relation of S. D. to the probability curve, together witli the possibility, as is shown later, of determining many other features of th(i distribution from the relation of aS. D. and M, is one of the principal reasons why S. D. is preferred by many to the more-easily calculated A. D. THE PROBABLE ERROR 25 surer that the M then obtained was the true one, or that if we had measured only ten boys of that age we should not have felt at all sure that the average thus obtained was truly representative of the height of 12-year old boys. We need, therefore, a measure of the reliability of M, so that we may have some idea as to how far the actually obtained M is likely to differ from the ideal or true M, or, reversely, how many measurements we need to secure an M that will have any desired or assigned degree of reliability. The most common measure of the reliability of M is afforded by its P. E., for there can be a P. E. of M as well as a P. E. of a single measurement or observation. To illustrate, suppose we did measure 1,000,000 boys in 1000 groups of 1000 measurements each; if we then averaged each group we should obtain 1000 M's, each representing the central tendency of a group chosen by ran- dom sampling: we should then expect these 1000 M's to be closely similar, but not identical, and we could distribute them like a series of individual measures and determine the P. E. of this distribution. In practise, the P. E. of M is found by a formula that takes into consideration the variability of the distribution which M represents and the number of cases on which it is based. This formula is P.^..= ^^^. (17) V n That is, the P. E. of M is found by dividing the P. E. of a single measurement (Formula 9) by the square root of the number of measurements. To bring this formula into relation with A. D., we may use the approximate formula P.E.^==^^-MB=A.D. (18) y^n- 1 The consistency of a series of measurements may also be indi- cated by stating the degree of probability that will attach to the appearance of an 'error' or deviation or residual, as it is often termed, of a magnitude equal to any assigned multiple of P. E. By definition, a deviation of the magnitude of P. E. is one as Ukely to be exceeded as not; in other words, the chances are even, or 26 THE TREATMENT OF MEASURES one to one, that it is exceeded. The probabiUty of the occurrence of a deviation several times as large as P. E. is, however, very much smaller, as will be seen in the following comparisons between P, the theoretical probability and X -^ P. E., multiples of P. E., from 1 to 5.^ - P.E. P. 1 1 - 1.0 2 3 5.6 - 23.2 4 1 - - 143.3 5 1 - - 1342.2 Besides the P. E. of m and of M, we may determine, by appro- priate formulas, the P. E. of measures of variability : two of these are given below; the application to measures of relationship will be discussed later. The P. E. of the S. D. is found by the formula P.E.= ^'^'''^ (19) \/2 which, by reference to Formula 18, becomes P. £..04745. ,20) V2n ^ These values are computed by reference to standard tables of values of the probability integral corresponding to various multiples of P. E. A con- densed table of this sort is published by Thorndike (21, p. 149). The values given above were derived for the author by Prof, G . C. Comstock of the University of Wisconsin from Oppolzer's 10-figure table of the Gamma Integral, and are correct to the first place of decimals given. To illustrate from Thorn- dike's condensed table; the total area of the probability surface being 1000, the total area representing deviation in either direction is 500. From the table we see that a deviation or residual equal to 3 P. E. occurs in such a manner that 479 of the 500 cases are included between it and the average or median, and hence it is exceeded by 21 of 500 cases, or by 1 case in 23.8, approx- imately; 23.2 when more accurate integral tables are used. From such a scries of values the consistency of the determination may be stated in various ways. For example, if a correlationof .50 was accompanied by a P. E. of .10, it might be said that the chances would be but 1 in more than 1300 times that such a correlation would occur by mere chance. MEASURES OF CORRELATION 27 The p. E. of the coefficient of variabihty may be found approx- imately, if C is not greater than 10 per cent, by the formula P.g.,^ 67450^ (21) but more accurately, for any value of C, by the formula ^^ 0.6745Cr^^^/C ^/2n L ^00,' If.. Other Applications of the Probability Curve Since, when the distribution is normal, the surface of frequency is determined by M and S. D., we may, by reference to suitable tables, ascertain (1) the frequency of any deviation, (2) the range of deviation that will include any given percentage of w's, (3) the chances that the true M will differ from the obtained by any given amount, (4) the range of divergence of the true from the obtained M that corresponds to any given degree of improbability, and (5) in general, the degree of reliability, or unreliability, of the several measures of variability or relationship.' E. MEASURES OF CORRELATION 1 . The Meaning of Correlation Physical science discovers numerous uniformities or corre- spondences between natural phenomena which are formulated as 'natural laws': biological science, on account of the intricacy of the factors which condition vital phenomena, can discover, fop 1 All of these calculations are made in terms of a probability integral table, which indicates for any normal surface the proportionate area of the probabiUty surface that is represented by any given degree of deviation (measured in this case in terms of the *S. D. of the distribution) . Lack of space precludes the discussion and reproduction of this table, which may be found in nearly all works on probability and statistics, e. g.-, Merriman (p. 187), C. Davenport (p. 55), Scripture, New Psychology, p. 475, Thorndike (21, p. 148; 20. p. 168). Examples of the calculations mentioned may be studied in Thorn- dike (21). 28 THE TREATMENT OF MEASURES the most part, only tendencies to uniformity or tendencies to corre- spondence. Such a tendency of two or more traits or capacities to vary together is termed a correlation. Thus height and weight are obviously correlated because in general tall people are heavier than short people, but, of course, this tendency to correspondence is far from absolute, like the correspondence between the distance and speed of a body falling in vacuum or between the electrical constants, voltage, amperage and resistance as expressed in Ohm's law. Since in practically every psychological test we are searching for these tendencies toward correspondence, it is important to know how they can be measured. In not a few psychological inves- tigations correlation has been expressed merely descriptively as 'fair,' ' large, '' poor, ' etc., and these characterizations have been derived from mere inspection of arrays of data. As a matter of fact, some of these published statements of correlation are actu- ally wrong: correlations do not exist where they have been affirmed, or do exist where they have been denied. At the present time there is no excuse for such merely descriptive statements of corre- lation, since, by the use of appropriate mathematical procedure, a tendency toward correspondence may be measured and expressed by a single quantitative symbol that has as much significance and definiteness as M, S. D., or any other statistical constant. This symbol, r, which sums up the proportionality or degree of relation- ship between two factors or events, is known as the index or co- efficient of correlation. Complete positive or direct correlation between two traits is present when the existence of the one is invariably accompanied by the existence of the other, or when increase of the one is invariably accompanied by corresponding and proportional increase of the other. Complete negative or inverse correlation is present when two traits are mutually exclusive, or when increase in the one is invari- ably accompanied by a corresponding and proportional decrease in the other. A correlation is indifferent or zero if the existence or variation of one trait is totally unrelated to that of the other. In perfect positive correlation, r is unity or 1.00; in complete THE PEARSON METHOD 20 negative correlation, r is — 1.00; indifference or complete absence of correlation is 0. In actual psychological investigation, at least when functional correspondences are under investigation, we have commonly to deal with some intermediate degree of corre- lation, and r assumes, therefore, the form of a decimal lying between and 1.00 for positive and between and — 1.00 for negative correspondence.^ 2. The Computation of the Index of Correlation (a) The ' Product-Moments ' Method of Pearson The most elaborate as well as the theoretically best possible method of computing r is the standard 'product- moments' method elaborated by Bravais, Galton, and especially by Pearson. In referring to the product-moments as the best possible method, certain quaUfications must be kept in mind. It is possible, for instance, that, as Spearman contends (17), the comparison of ranks (ig-method), for reasons that will be explained later, may be more reliable and satisfactory when psychological data are under treatment. In any case it is to be remembered that the product-moments method applies primarily to the relation of those arrays whose distribution is 'normal' — in the sense already explained. The mathematics of correlation for skewed, multimodal and other complex forms of distribution have received attention, especially at the hands of European mathematicians, but it cannot be said that their results have as yet reached a stage where they are of practical usefulness to those who have to take their measurement-methods at second-hand. (Cf. also Krueger and Spearman, pp. 53-4.) Again, it is also possible that functional relations may exist of so simple a form as to be readily expressed either verbally or mathematically, which, nevertheless, will give a zero coefficient by the standard method. Take, for instance, the hypothetical relation that Spearman has adduced as an example of zero correlation. Suppose five persons are tested for vision and hearing with the following results ( in terms of feet that the test-type is read and the sound heard): Person A B C D E Vision, in feet 6 7 9 11 14 Hearing, in feet A B C D 6 7 9 11 6 11 12 10 ^ On the meaning of r, especially in procedure by the method of rank-com- parison, see the discussion between Diirr and Spearman (5; 18). 30 THE TREATMENT OF MEASURES Here, says Spearman (15, p. 77), "we get r = 0, and thus there is no corre- lation, direct or inverse." But, as Lehmann and Pedersen (9, pp. 16,17) have cleverly shown, if these values are plotted graphically, there is revealed a simple functional relation, viz: hearing is poor when sight is poor, reaches a maximum when sight is fairly good, and then declines when sight continues to improve. From this it may be seen that it is best to plot relations in graphic form whenever possible. In many cases, indeed, such a functional graph is more significant than any coefficient could possibly be, just as a curve of distribution is more significant than an M, even when coupled with its measure of deviation. By the product-moments method X r = ^^^, (23) \ n 0^02 in which the x's are the series of deviations from M in the first array and the y's the corresponding series of deviations in the second array, and in which a^ is the standard de\'iation of the first and (72 the standard deviation of the second array, and n the number of cases in either array. The various steps of the computation may be illustrated by reference to Table V for grip of right and left hands, as follows: (1) Arrange the original measurements in order of their standing or rank, as shown in Columns 1, 3, and 7. (While this is not absolutely necessary, it commonly facilitates computation, though for speedier determination of the xy values, it might be preferable to place the two arrays in the same order by individuals, e. g., as shown by the numbers in Columns 2 and 6.) (2) Compute M (or the median) for each series (283 and 273). (3) Compute and record the individual deviations (d, columns 4 and 8) for each series, retaining the algebraic signs. (4) Multiply the d of each individual in the first series (now termed his x) by the d for the same individual in the second array (now termed his y), and record the products, observing the algebraic signs (the xy values in Column 13), e. g., boy No. 30 has for his X,— 125 and for his y, — 135, hence, for his xy, — 125 X — 135 = 16740. Again, boy No. 25 has for his corresponding values, — 21 and + 35, hence for xy, — 735. ' For a fuller illustration of correlation arithmetic, together with suggestions for shortening the work, see E. Davenport, pp. 455-471. THE PEARSON METHOD 31 (5) Add the products obtained in (4) (277,371, Column 13). (6) Compute the S. D. of both series {o^ and a, in Formula 23, illustrated in the d^ columns, 5th and 9th, and explained in Formula 5): multiply them together, and multiply their product by the number of cases (75.8 X 79 X 50 = 299,410). (7) Divide the 5th by the 6th resultant for the index desired (r = 277,371 ^ 299,410 = + 0.93). The arithmetic of the Pearson method is thus simple, though somewhat tedious. The work may be materially lessened by the use of Barlow's Tables of Squares, Cubes, Square Roots, etc. New York, 1904, of Crelle's Rechentafeln (procurable through G. E. Stechert & Co., New York) , which show at a glance the products of all numbers up to 1000 X 1000, and by the use of an adding machine.* Another considerable shortening may often be effected without serious disturbance by substituting Formula 7 for Formula .5 in computing the two S. D.'s. Thus, in our illustration, this substitution (see Table .3) gives for the denominator of the fraction: 76.9 X 79.3 X 50 = .304,908.50, from which we find r = 0.91. The probable error of the coefficient of correlation as obtained by the Pearson method is calculated by the formula P.E.r = 0.6745 ^^^, (24) although some mathematicians prefer the formula P.^., = 0.6745 ^i^!^- ■ (25) Vnd + r^) It is evident that the reliability of a coefficient increases with the number of cases compared and also with the magnitude of the r obtained. The actual values of P. E.^, as computed by Formula 24 for eleven values of r from to 1 accompanying values of n from 25 to 1000, are indicated in Table 5, so that one can not only read at a glance the P. E. for a given value of r and n, but also ' The author has found the Gem adding machine (price $15, procurable through the Automatic Adding Machine Co., New York City, or through the C. H. Stoelting Co., Chicago) serviceable for work in which there is no neces- sity for printed records such as the Burroughs, Standard, Wales, and other high-priced machines afford. 32 THE TREATMENT OF MEASURES determine the value of n, i. e., the number of observations, needed to estabhsh a given degree of correlation with any assigned degree of accuracy. In our illustrative case, since ?i = 50, and r = .93, we note that P. E.^ is less than 0.0181, that 200 observations would have reduced the error to less than .0091, etc. If our cor- relation had been lower, say 0.30, the error for 50 cases would have risen to 0.0868. Since the actual error is extremely small in rela- tion to the obtained correlation, it follows that the latter has an enormous degree of reliability. TABLE 5 Prohahle Error of r for Various Values of r and of n {Yule) VALUES OFn r = r= .1 r= .2 r = .3 r= .4 r = .5 r = .6 r = .7 r = .8 r= .9 r = 1 25 .1349 .1335 .1295 .1228 .1133 .1012 .0863 .0688 .0486 .0256 .0000 50 .0954 .0944 .0916 .0868 .0801 .0715 .0610 .0486 .0343 .0181 0000 75 .0779 .0771 .0748 .0709 .0654 . 0584 .0498 .0397 .0280 .0148 .0000 100 .0674 .0668 .0648 .0614 .0567 .0506 .0432 .0344 .0243 .0128 .0000 200 .0478 .0473 .0459 .0435 .0402 .0.359 .0306 .0244 .0172 .0091 .0000 300 . 0389 .0386 .0374 .0354 .0327 .0292 .0249 .0199 .0140 .0074 .0000 400 .0337 .0334 .0324 .0307 .0283 .0253 .0216 .0172 .0121 .0064 .0000 500 .0302 .0299 .0290 .0274 .0253 .0226 . 0193 .0154 .0109 .0057 0000 600 . 0275 .0273 .0264 .0251 .0231 .0207 .0176 .0140 .0099 .0052 .0000 700 .0255 .0252 .0245 .0232 .0214 .0191 .0163 .0130 .0092 .0048 .0000 800 .0238 .0236 .0229 .0217 .0200 .0179 .0153 .0122 .0086 .0045 .0000 900 .0225 .0223 .0216 .0205 .0189 .0169 .0144 .0115 .0081 .0043 .0000 1000 .0213 .0211 .0205 .0194 .0179 .0160 .0137 .0109 .0077 .0041 .0000 In general, a correlation, like any other determination, to have claim to scientific attention must be at least twice as large as its P. E., and to be perfectly satisfactory, should be perhaps three to five times as large. Since, in our illustration, r is some 51 times as large as its P. E. its appearance by mere chance is practically zero and its reliability is practically absolute. (b) The Pearson Method Adapted to Rank-Differences The product-moments method gives full and exact weight to the d of each m from the M. If we disregard the magnitude of THE METHOD OF RANK-DIFFERENCES 33 these d's, however, and regard only the relative order or station of individual w's in each array, we may yet measure correlation by what is known as the method of rank-differences. For this method, the formula is r = 1 - ^^ , (26) c in which D is the numerical difference between each corresponding pair of ranks^ (not to be confused with d, the deviation from the mean), and in which c is the mean value of ^ D^ by mere chance. Since c = "<"''" , (27) 6 Formula 26 may be written: r = l- ^-^ . (28) n (n' - 1) For illustration, note in Table 1, Column 11, the series of D's which are squared and summated in Column 12. Boy No. 30 ranks first (weakest) in the distribution for right-hand grip and first in the order for left-hand grip, hence his D = 0. Boy No. 522 ig 3(j in the first, and 9th in the second array, hence for him D = Q and D^ = 36. Since n = 50, by Formula 27, 50 (2500- 1)= 20,791; '= 6 hence, by Formula 26, ,.1_ 209^.1-10=.90, 20,791 1 In this and the following rank method of Spearman, cases of 'ties' for a given rank are preferably divided in such a manner as to keep the total number of ranks equal in the two series. If, for instance, two S's rank 5th, they should both be assigned the rank 5.5 (to replace 5 and 6), or if three S's rank 5th, they should all three be assigned the rank 6 (to replace the 5th, 6th and 7th places in the series). - To avoid possible confusion, it may be explained that two r ecords were discarded, so that the boys' numbers run two over the fifty. 34 THE TREATMENT OF MEASURES a result in close accordance with that of the product-moments method and obtained in a small fraction of the time. (c) Spearman's Correlation 'Foot-Rule,' or R-Method. Spearman's " ' foot-rule ' for measuring correlation " (17) is another and still simpler method of comparison b}^ rank, the essential feat- ures of which are the use of D, the numerical difference of station, in place of D^, and of only those of the D's that indicate a gain in rank (since the losses must equal the gains). It is because this simplified method gives less weight to extreme measures that Spearman believes it to be actually more reliable for psychological pur- poses than the standard Pearson method. The advantages are summarized by Spearman (17, p. 104) as follows: "By using it, we obtain a precise quanti- tative value, which can be compared with that found by any other correlation under any circumstances or between any other things, either by the same or by the standard r-method; we free ourselves from various illusions, which are otherwise almost irresistible; we get a reliable estimate as to the danger of our result being merely an accidental coincidence; and we even learn how to plan out our experiments from the outset in a manner properly adapted to the object in view." In the opinion of Lehmann and Pedersen, however, Spearman's 'footrule' has only slight value, because it can measure and express only proportionality and not other uniform relationships of phe- nomena, while even when dealing with proportionality, the method may lead to totally false results, e.g., with a complete inverse relation, the value secured, instead of —1 would be —0.5 with an odd number of cases, and 1 — 1.5 X n- / (/i^- 1) with an even number of cases (9, p. 18). This method, it is important to note, yields an index, R, that is not identical with the Pearson r, though functionally related there- to, as is explained below. The formula for Spearman's R is /? = 1 - -^ , (29) c in which cj is the numerical gain in rank of an individual in the second, as compared with the first series, and in which c is the mean value of - g by mere chance. Since: c = J (30) THE SPEARMAN METHOD 35 we obtain by substitution: R^l-l^l^. (31) n^ - 1 For illustration, note in Table 1, Column 10, the series of gains (g), which yield ^ g = 106. As n = 50, by Formula 30, c = (2500 - 1) - 6 = 416.5; hence, by Formula 29, R=l - ^^A = l-.25 = .75. 416.5 To determine whether R has any claim to reliability, one may use the rather severe formula:^ P.E.,= ^^^h (32) V n To convert /^-values into r-values we may use the formula: r = sin(~^Ry, (33) or, for all cases in which R is less than .50, this may be simplified with httle loss of accuracy into: r=1.5i^. (34) For the quick and accurate conversion of R into r, Table 6' which is based on Formula 33, may be consulted. In the correlation under examination, since /2 = .75, r = .93, or precisely the value obtained by the longer Pearson method. (d) Correlation by Distribution of Selected Groups The following method is sometimes useful as a device for pre- liminary survey, but when used, as it often has been, for a final ' According to Spearman, if we know that a correlation exists and wish merely to estimate the accuracy of R, a less rigorous formula may be used. 36 THE TREATMENT OF MEASURES expression of correlation, it is inferior to the methods already described. Distribute the data for both series in order as illustrated in Table 1, and divide them into four or five groups on the basis of equal numbers of cases or of equal amounts of deviation. By inspection it is often possible to determine at this juncture whether there is sufficient evidence of a correlation to justify further TABLE 6 Conversion of R-Values into r-Values, in Accordance with Formula 33 R r R r R r R r R »• .00 .00 .20 .31 .40 .59 .60 .81 .80 .95 .01 .01 .21 .32 .41 .60 .61 .82 .81 .96 .02 .03 .22 .34 .42 .61 .62 .83 .82 .96 .03 .05 .23 .35 .43 .62 .63 .84 .83 .96 .Oi .06 .24 .37 .44 .64 .64 .84 .84 .97 .05 .07 .25 .38 .45 .65 .65 .85 .85 .97 .06 .08 .26 .40 .46 .66 .66 .86 .86 .98 .07 .11 .27 .41 .47 .67 .67 .87 .87 .98 .08 .13 .28 .43 .48 .69 .68 .88 .88 .98 .09 .14 .29 .44 .49 .70 .69 .88 .89 .99 .10 .16 .30 .45 .50 .71 .70 .89 .90 .99 .11 .17 .31 .47 .51 .72 .71 .90 .91 .99 .12 .19 .32 .48 .52 .73 .72 .90 .92 .99 .13 .20 .33 .50 .53 .74 .73 .91 .93 .99 .14 .22 .34 .51 .54 .75 .74 .92 .94 1.00 .15 .23 .35 .52 .55 .76 .75 .93 .95 1.00 .16 .25 .36 .54 .56 .77 .76 .93 .96 1.00 .17 .26 .37 .55 .57 .78 .77 .94 .97 1.00 .18 .28 .38 .56 .58 .79 .78 .94 .98 1.00 .19 .29 .39 .57 .59 .80 .79 .95 .99 1.00 calculation. For this inspection we may take the cases found in the first group of the first series and examine their distribution in the groups of the second series. Evidently, in the absence of correlation, these cases would be distributed by chance. Thus, in Table 1, let the two series be divided into 5 groups of 10 measure- ments each. The 10 measurements in the first group for right- hand grip would tend, by chance alone, to be distributed 2 in THE METHOD OF GROUP COMPARISON 37 each of the 5 groups in the second series, but as a matter of fact they are massed in or near the first group (8 in the 1st, 2 in the 2d), hence there is evidently a high degree of correlation. The distribution in the second series of the remaining groups of the first series may be similarly tested, though an examination of the first group is commonly sufficient. If the grouping is made in terms of deviation, the number of measurements found in the several groups will usually be unequal; it is then necessary to calculate the distribution of the various groups of the first series into those of the second. Suppose the two series of Table 1 are each divided into five 70- hectogram groups; the right-hand series will subdivide into groups containing 16, 15, 12, 4 and 3 measurements; the left-hand series into groups con- taining 12, 17, 12, 6 and 3 measurements, respectively. Take the 16 cases in the first group of the first series; by chance it is clear that 12/50, or 3.84 of them, would fall in the first group of the 2d series, 17/50, or 5.44 of them would fall in the second group, 3.84 in the third, 1.97 in the fourth, and .98 in the fifth. The actual distribution of the 16 cases into those five groups is 11, 5, 0, 0, 0, as compared with the chance distribution, 3.S4, 5.44, 3.84, 1.97, .98. If now we wish not only to explore the distribution of selected groups tentatively for the presence of correlation, but also to present the evidence of the correlation in compact form, we may, as is often done, prepare a table by comparative averages. To return again to the correlation of right and left-hand strength of grip, we may by this means secure the following tabular state- ment of the relations l^etween right-hand and left-hand grip.^ TABLE 7 Correlation of Right and Left-Hand Grip by Group Averages (Whipple) 1st 10 2d 10 3d 10 4th 10 5th 10 Average Right Hand Grip 194.9 183.7 229.9 226.3 271.5 269.2 317.9 291.5 402.2 Average Left Hand Grip 394.4 ^For examples of this type of correlation, consult Bagley (1), Binet and Vaschide (2). Though this method has been frequently used, a little reflec- tion will show that it is inferior to the several methods described before, because the advantage of weighing the relation of each individual measure is lost by lumping them into averages, and because, moreover, no coefficient of correla- tion is computed. If the groups were made more numerous and the data presented in the form of a graph showing the entire course of the relationship the absence of the coefficient might be less serious. 38 THE TREATMENT OF MEASURES The data from such a table may be thrown into graphic form very simply: let ordinates represent the values of lh(! one series, abscissae the values of the other, so that the one series is plotted as a function of the other: if, then, the first series ranges upward for values from low to high and the second from left to right for values from low to high, a positive correlation will be indicated by a hne running in a southwest-northeast direction, inverse correlation by a line running in a northwest-southeast direction, and zero cor- relation by a vertical or a horizontal line (depending on which series is plotted on the ordinates). In proportion as the corre- lation is complete the line assumes an oblique position. (e) Correlation of Presence and Absence Suppose we wish to correlate two traits whose presence or absence is ascertainable, but about whose degree of presence nothing can be said. Such a correlation can be computed if we first determine: a =-- number of cases in which both traits are present, b = number of cases in which the first trait is present and the second absent, c = number of cases in which the second trait is present and the first absent, d =-- number of cases in which both traits are absent. On the basis of these values, we may use the simi)le formula of Yule: ad - he ,orx r = - , {60) ad + be or, better, the following modification: . t: Vad -' \'be ,..,.. r = sm • (,ob) 2Vad + Vbc If, in this formula, we replace the sine by the cosine of its comple- ment, we secure _ - Vad - Vbe cos 2 2\'ad + \'be CORRELATION OK PRESENCE AND ABSENCE 39 which we can reduce to (37) V fee r = cos ^^ __7i. V ad + '' be The probable error, provided ab is not verj^ unequal to cd, may be taken as 1.1 P.E., V (38) As an example one may take the measure of the tendency of white cats to be afflicted with deafness (as cited by E. Davenport from Yule). From the data as presented in Table 8, we may calculate by Formula 35 that r = .91. TAI5LE 8 Relation ofDeafnesN and WIdte Color in Cats {Yule) CATS WHITE NOT WHITE Deaf G 4 Not Deaf 14 976 Now this same formula can often be used for preliminary explor- ation of series in which the degree of presence of a trait is known, and which may, therefore, be treated, if desired, by the more elaborate methods. For this purpose, assume that all measure- ments greater than M (or the median), i. e., all plus cases, signify the presence of the trait, and all minus cases its absence. In Table 1 we find, using the median, 22 cases that are plus in both series, 3 that are plus in right and minus in left-hand grip, 3 that are minus in right and plus in left-hand grip, and 22 that are minus in both. Substituting these values in Formula 37, we have Vq r =- COS -, or COS 21 6° = .93, \/ 484 + \/ 9 which gives again the same value as by the standard formula 40 THE TREATMENT OP MEASURES (f) The Method of Unhke Signs Even this procedure may be simphfied by substituting for ybc the percentage of cases with unhke signs (U), and for 1^ ad the per- centage of cases with hke signs (Ly with the result, u COS L+ [7 or, since L + U must always equal 100, and since tt = 180° formula may be condensed, if desired, to r^ cos U 1.8°. (39) this (40) Finally, since U must lie between 50 and for positive and between 50 and 100 for inverse correlations, a talkie may be pre- pared- from which the values of r may be read directly from any integer value of U. By reference to the paragraph above it will be seen that in Table 1 we have 6 cases of unlike signs in the 50, hence U = 12 and r = 0.93, as by other methods. TABLE 9 Corresponding Values of r and U for Formula J^O (Whipple) If U is greater than 50, first subtract it from 100, then prefix the mimis sign to the correlation indicated u r u r u r U r U r 1.000 10 .951 20 .809 30 .587 40 .309 1 .999 11 .941 21 .790 31 .562 41 .279 2 .998 12 .929 22 .770 32 .536 42 .248 3 .995 13 .917 23 .750 33 .509 43 .218 4 .992 14 .904 24 .728 34 .482 44 .187 5 .987 15 .891 25 .707 35 .454 45 .156 6 .982 16 .876 26 .684 36 .426 46 .125 7 .976 17 .860 27 .661 37 .397 47 .094 8 .968 18 .844 28 .637 38 .368 48 .062 9 .960 19 .827 29 .613 39 .338 49 .031 ' That is , virtually substituting the arithmetical for the geometrical mean. ^ Reproduced from an earlier article by the author (23) in which the applic- ability of the method is discussed more fully. CORRECTION OF CORRELATIONS 41 This method cannot be recommended for final determinations of important correlations because the probable error is too large, but it is a useful device for quick examination of a relation. The m?thod of unlike signs, in sharp contrast to the product-moments method, disregards entirely the amount of deviation. The author has made occasional use, for the first approximation, of an intermediate method that also disregards the actual deviations, but introduces artificial ones to be treated by the standard Pearson formula. Assume that all measurements lying between M (or the median) and the A. D. (or S. D., or P. E.) have a deviation of 1 (plus or minus according to direction of the actual deviations), and that all other measurements have a deviation of 2 (plus or minus). Com- pute Ixy, (7, and o^ precisely as if these assumed deviations were the actual ones. This method, when applied to Table 1, yielded r = .97 when a was calculated by Formula 7, and r = .87 when a was calculated by Formula 5. 3. The Correction of Obtained Correlations to their True Value (a) Correction of the Attenuation Produced by Chance Errors The real correspondence between two traits or capacities is not, as has so often been erroneously supposed, necessarily revealed by the determination of a coefficient of correlation, even by the most approved methods and with a probable error that is satis- factorily small. All measurements, as we have noted, are subject to chance errors of observation. In the determination of averages such errors tend to counterbalance one another, so that if the measurements are sufficiently numerous, the obtained M differs from the true M by an inappreciable amount. In the case of correlations, however, these errors^ are not eliminated by increas- ing the number of observations, and their presence has the effect of decreasing the size of the correlation, so that, in so far as these errors are concerned, the ' raw' or obtained correlation is too small, or, to use Spearman's term, the correlation is 'attenuated' by errors which constitute, from this point of view, constant or sys- tematic errors. ^ The phrase 'errors of observation' i s to be understood in a wide sense, to include not only errors arising from technique, instrumentation, etc., but also chance shifts in the disposition of subjects, in their attitude toward the test, etc. 42 THE TREATMENT OF MEASURES This illusory attenuation of the correlation by errors of observation seems, in fact, a principal cause of the contradictory nature of results that have hitherto been obtained ; in experiments in which such errors have been very large, a correlation has not appeared, even when present, and has, in consequence, been erroneously denied. The determination of a small correlation therefore opens two possibihties ; it may indicate actual absence of correspondence, or it may indicate merely the presence of large chance errors of observation. (Kriiger and Spearman, p. 55.) In order to correct the raw and discover the true r, it is impera- tive to secure at least two independent series of observations. The formula for correction of attenuation, or the 'expanding' formula, as it might be termed, is then applied as follows: .„ _ ikf (A,B;, A,B„ A,B„ A3.j ,,,, ^^' m(a,a,b;b;)"""^" '*^' in which AB^ = the true correlation, M = the mean, A, = the 1st series of observations of the trait A, A.J = the 2d series of observations of the trait A, B, = the 1st series of observations of the trait B, B, = the 2d series of observations of the trait B, A,Bi =^ the raw correlation of Aj and B,, AjA, = the raw correlation of Aj and A,., etc. Thus the numerator is the M of the four possible r's between the measure- ments of A and the measurements of B, while the denominator is the M of the r of the two A series and the r of the two B series. Incidentally, these last mentioned correlations, taken singly, afford an obvious coefficient of reliabihty of the two series of measurements of A and B respectively. The above formula holds for ordinary cases, but if one series of observations, say A, should be known to be much more exact and reliable than the other, then the geometrical should be substituted for the arithmetical M. In theory the denominator should always be the geometrical M, but the arithmetical M is virtually as accurate and for short series even more desirable. For the mathematical demonstration of this and the following formulas, the accuracy of which has been disputed by several writers, consult Spearman (15). The correction does not entirely eliminate the uncertainty that arises from the use of 'random samples' for investigation; that must be removed by the use of more extended series. CORRECTION OF CORRELATIONS 43 (b) Correction of the. 'Constriction' or 'Dilation' Produced by Constant Errors Attenuation is the result of the operation of chance errors, — chance in the sense that the deviation of any measurement takes place independently of the deviation of any other measurement. If, however, some influence is at work which affects all the measure- ments of one or of both series, such a constant factor or con- stant error will prove a source of disturbance that may either increase or decrease the obtained correlation. Such disturbances will result from the operation of any factor which is not strictly relevant to the correspondence under examination. If an irrelevant factor affects both of the series, it is evident that the correlation will be unduly increased or ' dilated. ' Suppose for example, that one wished to determine the correlation of pitch discrimination with the discrimination of lifted weights, and that the subjects of the exj^eriments wert of different ages. Then, since the two capacities in question both tend to improve with age, this common dependence on age will clearly tend to induce the appearance of a correlation, even if there really be none between the capacities themselves when compared under uniform conditions of age.^ If an irrelevant factor affects but one of the series, it is evident that the correlation will be unduly decreased or 'constricted,' i. e., the irrelevant influence will tend to reduce any proportionality that really exists between the two series. To quote an example from Spearman, a correlation of 0.49 was discovered between pitch discrimination and school standing, but it was likewise discovered that more than half the children had 'taken lessons,' and thus had the opportunity for special training in the obser- vation of pitches. These constant irrelevant factors may not always be excluded? ^ This undiscovered or neglected influence of age has been a very common source of error in many studies of correlation. Obviously, this particular irrelevancy may be eliminated practicallyby proper selection of subjects for the investigation, or it may be eliminated by manipulation of the results in various ways besides that here described: see, for example, Bagley. 44 THE TREATMENT OF MEASURES but their force can frequently be measured and allowed for by the following formula ;! AB, = AB^-AC BC ^2^ V a- AC) (1 - BC^) in which ABj = the true correlation between A and B, ABa = the apparent correlation between A and B. AC = the direct correlation between A and any irrelevant fac- tor, C, BC = the direct correlation between B and C. If, as is most often the case, the irrelevant factor affects but one series, this influence of 'constriction' may be excluded by the simpler formula: AB, = ^--A^^ (43) V a- Ac^) Thus, in the example mentioned, the correlation between pitch discrimination and its disturbing factor, musical training, was found b}^ computation, to be 0.61 ; hence, by Formula 43, r= Q-^Q - = 0^9 V 1 -0.61- From the above considerations, it follows that the experimenter must define with some exactness the traits that are to be examined for a possible correla- tion, and that he must not seek to establish the correlation until, by means of suitable preliminary exploration, he has discovered all the irrelevant factors that might disturb the correspondence. The mere mechanical computation of an index of correlation does not, then, demonstrate the existence of a real correlation, or at least, does not accurately and certainly define its nature. Hence, while, as we have seen, we may very hopefully look to correlational work for revelation of the functional disposition of mind, this is no royal road to the attainment of the end, but can itself be entered upon in each instance only after a preparatory survey and critical inspection of the problem in hand has afforded sufficient acquaintance with the traits and capacities that are therein concerned. One must be a psychologist as well as a statistician. ^ ^ AB, AC and BC must first be 'expanded' by Formula 41. ^ The study of the functional correlation of five well-known tests by Krviger and Spearman affords an admirable illustration of the value of such a comliina- tion of sound psychology and sound statistics. FACTORS IN INTERCORRELATIONS 45 4. The Discovery of Common Factors in Inter correlated Capacities If three or more psychological traits show intercorrelations one with another, the question may be raised as to whether the intercorrelations are not due to the presence of some common factor to which all the capacities are functionally related, whether, in other words, these correlations may not arise from a single underlying cause. If such a common or 'central' factor be assumed to be present, we may test the validity of the assumption by mathematical procedure, leaving the exact nature of the factor out of consideration for the time being. For example, if for any given capacity. A, we have obtained two independent measure- ments, A, and A,, and if for two other capacities, we have obtained the measurements B and C respectivelj^, then the correlation (AF) between the capacity A and the hypothetical common or central factor, F, may be determined by the formula : ^j,_ JW(AB,BC) ^ > ^^ m1a7a;;bc) • ^**' In illustration, Kriiger and Spearman found the following values, — correlation of pitch discrimination with the Ebbinghaus comple- tion test, 0.65, with adding, 0.66, correlation of two measurements of pitch discrimination, 0.87, correlation of the Ebbinghaus test with adding, 0.66; hence the correlation of pitch discrimination with the hypothetical central factor is the M of 0.65 and 0.66 divided by the M of 0.87 and 0.71, or 0.83.i REFERENCES ^ (1) W. C. Bagley, On the correlation of mental and motor ability in school children, in A. J. P., 12: 1901, 193-205. (2) A. Binet and N^. Vaschide, Experiences de force musculaire et de fond ohez les jeunes gargons in A. P., 4: 1897 (1898), 15-63. ^ This central factor is tentatively ascribed by these authors to some psychophysiological condition, possibly general neural plasticity. In another article Spearman speaks of "General Discrimination" or "General Intelli- gence," or the "Intellectual Function." Here, again, is revealed the instruc- tive wealth of possibilities in the application of adequate methodological treatment to psychic life. ^ Consult the List of Abbreviations for the exact titles of periodicals in these and subsequent references. 46 THE TREATMENT OF MEASURES (3) C. B. Davenport, Statistical methods, with special reference to biologi- cal variation, N. Y., 1899. Pp. 148. (2d revised edition, 1904.) (4) E. Davenport, Principles of breeding; a treatise on thremmatology, Boston, 1907. Pp. 713. (With an appendix on statistical methods by H. Rietz.) (5) E. Durr, review of Spearman (15, 16), in Z. P., 41: 1906, 450. Also, Erwiderung, in Z. P., 42: 1906, 470-2. (6) W. P. Elderton, Frequency curves and correlation, London. 1906. (7) F. Galton, Natural inheritance, London, 1889. Pp. 259. (8) F. Krueger and C. Spearman, Die Korrelation zwischen verschiedenen geistigen Leistungsfiihigkeiten, in Z. P., 44: 1907, 50-114. (9) A. Lehmann and R. Pedersen, Das Wetter und unsere Arbeit, in A. G. P., 10: 1907, 1-104, especially 15-26. (10) M. Merriman, A text-book on the method of least squares, N. Y., 1884. Pp. 194. (11) M. Meyer, The grading of students, in Science, 28: 1908, 243-250. (12) H. Rietz: see E. Davenport. (13) E. C. Sanford, A course in experimental psychology, Boston, 1895 and 1898. Pp. 449. (14) E. W. Scripture, The new psychology, London, 1897. Pp. 500. (15) C. Spearman, The proof and measurement of association between two things, in A. J. P., 15: 1904, 72-101. (16) C. Spearman, General intelligence objectively determined and meas- ured, in A. J. P., 15: 1904, 201-293. (17) C. Spearman, 'Footrule' for measuring correlation, in B. J. P., 2: 1906, 89-109. (18) C. Spearman, Entgegnung, in Z. P., 42: 1906, 467-470. (19) C. Spearman, Demonstration of formulae for true measurements of correlation, in A. J. P., 18: 1907, 161-9. (20) E. L. Thorndike, Educational psychology, N. Y., 1903. Pp. 173. (21) E. L. Thorndike, An introduction to the theory of mental and social measurements, N. Y., 1904. Pp. 210. (22) E. B. Titchener, Experimental psychology, vol. ii. Quantitative experiments. Parts 1 and IT, N. Y., 1905. (23) G. M. Whipple, A quick method for determining the index of corre- lation, in A. J. P., 18: 1907, 322-5. (24) C. Wissler, The correlation of mental and physical tests, in P. R. M. S., 3: No. 6, 1901. Pp. 62. THE TESTS CHAPTER IV Anthropometric Tests The tests embraced in this chapter have been developed pri- marily as anthropometric tests. They do not include tests of physical capacity or function (Chapter V), but simply measure- ments of bodily size or dimension. The number of such measurements that have been made and recorded runs well into the hundreds, and an extensive literature has appeared. The science of anthropometry has developed partly in connection with anthropology and sociology, partly in connection with the study of physical development, including bodily growth, hygiene, gymnastic and athletic training. In recent years, moreover, a not inconsiderable contribution has been made by psychologists, physicians, educators, and other investi- gators who have been interested in the correlation between bodily and mental traits. It is this last-mentioned phase of anthropometry that concerns us, and hence only a few important measurements that have assumed special importance in conjunction with other physical and with mental tests are here considered. The references which follow will enable the reader to study the development of anthropometry and the application of anthro- pometric tests at large. Bertillon and Galton should be con- sulted by those who are interested in the use of anthropometric measurements in the identification of criminals: Key and Hertel have given special consideration to the relation of growth to disease and to hygienic conditions. Anthropometric charts or record- books have been published by E. Hitchcock. D. A. Sargent, J. W. Seaver, W. W. Hastings, Anna Wood, L. H. Guhck, and others. 48 ANTHROPOMETRIC TESTS REFERENCES (1) A. Bertillon, Signaletic instructions; including the theory and practise of anthropometrical identification, Eng. Trans., Chicago and New York, 1896. Pp. 260 + 81 plates, charts, etc. (2) H. G. Beyer, The growth of United States naval cadets, in Proc. U. S. N. Inst., 21: No. 2, 1895. (3) F. Boas, The correlation of anatomical or physiological measurements, in Amer. Anthropologist, 4: 1894, 313. (4) F. Boas, The growth of children, in Science, 19: 1892, 250. (5) F. Boas, On Dr. W. T. Porter's investigation of the growth of the school children of St. Louis, in Science, 1: 1895, 225. ' (6) F. Boas, Growth of first-born children, in Science, 1: 1895, 402. (7) F. Boas, Growth of Toronto children, in Rept. Brit. Ass. Adv. Science for 1897, 1898, 443. (8) F. Boas, Growth of American children, in U. S., 1896-7, ii, 1555. (9) H. P. Bowditch, The growth of children, in 10th An. Rept. State Brd. Health, Mass., 1879, 33. (10) H. P. Bowditch, The growth of children, studied by Gallon's method of percentile grades, in 22d An. Rept. State Brd. Health, Mass., 1891, 479. (11) F. Burk, Growth of children in height and weight, in A. J. P., 9: 1897-8, 253. (12) A. Engelsperger and O. Ziegler, Beitrage zur Kenntnis der physischen und psychischen Natur des sechsjahrigen in die Schule eintretenden Kindes. Anthropometrisches Teil, in E. P., 1: 1905, 173-235. (13) Lucy Ernst and E. Meumann, Das Schulkind in seiner korperlichen und geistigen Entwicklung. Part 1, by L. Ernst, Leipzig, 1906. Pp. 143. (14) W. Farr, et al., Table showing the relative statures of boys at the age of 11 to 12 years under different social and physical conditions of Hfe, in Rept. Brit. Ass. Adv. Science, 1880, i, 127. (15) F. Galton, Proposal to apply for anthropological statistics from schools, in J. Anthrop. Inst., 3: 1873-4, 308. (16) F. Galton, On the height and weight of boys aged 14 in town and country schools, in Nature, 5: 1875-6, 174. (17) F. Galton, Range in height of males at each age and in several classes, in Rept. Brit. Ass. Adv. Science, 51 : 1882, 250. (18) F. Galton, Head growth in students at the University of Cambridge, in Nature, 38: 1888-9, 14. (19) F. Galton, Useful anthropometry, in Proc. Amer. Ass. Adv. Phys. Educ, 6: 1891, 51. (20) F. Gallon, Anthropometrical instruments, in Anthrop. J., 16: 2. (21) F. Galton, Anthropometric percentiles, in Nature, 31: 1884-5,223. (22) F. Galton, Finger-prints, London, 1892. Pp. 216. (23) A. Gihon, Physical measurements, in Wood's Ref. Haiid-bk. Med. Sciences, 5: 1887, 667. ANTHROPOMETRIC TESTS 49 (24) N. -A. Gratsianoff, Data for the study of physical development in childhood ( in Russian), St. Petersburg, 1889. (25) J. M. Greenwood, Heights and weights of children, in PubHc Health Ass. Kept., 1891, Concord, N. H., 17: 1892, 199. (26) E. Hartwell, Anthropometry in the United States, in Amer. Statis. Ass., 3: 554. (27) E. Hartwell, Rept. of the director of physical training, Boston Normal Schools, in School Doc, No. 8, Boston, 1894. (28) E. Hartwell, Rept. of the director of physical training, in School Doc, No. 4, Boston, 1895. (29) E. Hartwell, Bowditch's law of growth and what it teaches, reprint from 10th An. Proc Amer. Ass. Adv. Phys. Educ, Concord, N. H., 1896. (30) W. Hastings, A manual for physical measurements for use in normal schools, public and preparatory schools, etc., Springfield, Mass., 1902. Pp. 112. (31) A. Hertel, Neuere Untersuchungen uber den allgemeinen Gesund- heits-Zustand der Schiller u. Schiilerinen, in Schulgesundheitspflege, Nos. 6 and 7, 1888, 167-183, 201-215. (Contains a review of the work of Key.) (32) E. Hitchcock, Jr., Physical measurements, fallacies, and errors, in Proc. Amer. Ass. Adv. Phys. Educ, 1887, 35. (33) E. Hitchcock, Jr., A synoptic exhibit of 15,000 physical examinations, Ithaca, 1890. (34) E. Hitchcock, Comparative study of measurements of male and female students at Amherst, Mt. Holyoke, and Wellesley Colleges, in Proc. Amer. Ass. Adv. Phys. Educ, 6: 1891, 37. (35) E. Hitchcock, The results of anthropometry as derived from the measurements of the students of Amherst College, Amherst, 1892. (36) E. Hitchcock and H. Seelye, An anthropometric manual, giving the average and mean physical measurements and tests of male college students, and modes of securing them, Amherst, 1889. (37) B. Holmes, A study of child growth, being a review of the work of Dr. Wm. T. Porter, of St. Louis, in N. Y., Med. J; 60: 1894, 417. (38) W. Jackson, Jr., Graphic methods in anthropometry, in Phys. Educ, 2: 1893, 89. (39) A. Key, Schulhygienische Untersuchungen, trans, and edited, some- what condensed, from the Swedish, by L. Burgerstein, Hamburg and Leipzig, 1899. (Report of the Swedish Commission of 1882 to investigate hygienic conditions of the schools.) (40) A. MacDonald, Experimental study of children, including anthro- pometrical and psychophysical measurements of Washington school children, reprint of chs. 21 and 25 of U. S., 1897-8, Washington, 1899. (41) P. Malling-Hansen, Perioden im Gewicht der Kinder und in der Sonnenwarme, Copenhagen, 1886. (42) S. Moon, Measurements of the boys of the McDonogh School, McDonogh, Md., 1893. 50 ANTHROPOMETRIC TESTS (43) K. Pearson, Growth of St. Louis children, in Nature, 51: 1894, 145-6' (44) G. Peckhara, The growth ©f children, in Kept. Wis. Brd. Health, 1881, 28. (45) (i. Peckham, Various observations on growth, ibid., 1882, 185. , (46) W. T. Porter, On the application to individual school children of the mean values derived from anthropological measurements by the generahzing method, in Pub. Amer. Statis. Ass., n. s. 3: 1892-3, 576. ,, (47) W. T. Porter, The growth of St. Louis children, in Trans. Acad. Science, St. Louis, 6: 1894, 263. (48) W. T. Porter, Physical basis of precocity and dullness, ibicJ., 6: 1893-4, 160. - (49) W. T. Porter, The relation between growth of children and their deviation from the physical type of their sex and age, ibid., 233. (50) W. T. Porter, Use of anthropometrical measurements in schools, in Educ. Rev., 11: 1896, 126-133. (51) C. Roberts, A manual of anthropometry. (52) Sack, Physical development of the children in the middle schools of Moscow, 1892. (53) D. Sargent, Report on anthropometric measurements in Proc. Amcr. Ass. Adv. Phys.'Educ, 2: 1886, 6. (54) D. Sargent, Anthropometric apparatus, with directions for measur- ing and testing the principal physical characteristics of the human body, Cambridge, Mass., 1887. ' (55) D. Sargent, The physical proportions of the typical man. in Scrihner's Magazine, 2: 1887, 3. ■-(55) D. Sargent, The physical development of women, ibid., 5: 1889, 541. (56) D. Sargent, Strength-tests and strong men at Harvard, in J. Boston Soc. Med. Sci., No. 13, 1896-7. (57) G. Schultz, Some new anthropometrical data, in Yale Med. J., 2: 1895-6, 149. (58) J. Seaver, Anthropometry and physical examination. New Haven, 1890. (59) G. Sergi, An anthropological cabinet for pedagogic purposes, in Education, 7: 1886, 42-9. (60) C. Stratz, Der Korper des Kindes, fur Eltern, Erzieher, Aerzte. u. Kunstler, Stuttgart, 1904. (61) F. Swain, Anthropometric measurements, in Proc. .-Vmer. Ass. Adv. Phys. Educ, 3: 1887, 43. (62) G. Tarbell, On the height, weight, and relative growth of normal and feeble-minded children, in Proc. 6th An. Session Ass. Med. Officers Amer. Inst. Idiotic and Feeble Minded Persons, Phil., 1883, 188. (63) R. Thoma, Untersuchungen iibcr die Grosse u. das Gewichtder ana- tomischen Bestandtheile des menschlichen Korpers, 1882.. (64) E. B. Titchener, Anthropometry and experimental psychology, in Philos. Review, 2: 1893, 187-192. TEST 1: HEIGHT 51 (65) F. Tackermann, Anthropometric data based upon nearly 3,000 meas- urements taken from students, Amherst, 18SS. (66) G. West, Worcester school children; the growth of the body, head, and face, in Science, 21: 1893, 2-4. (67) G. West, Observations on the relation of physical development to intellectual ability, made on the school children of Toronto, Canada, in Science, n. s. 4: 1896, 156. (68) C. Wissler, Growth of boys; correlations for the annual increments, in Amer. Anthropol., n. s. 5: 1903, 81. (69) M. Wood, Anthropometric tables, arranged after the method of per- centile grades, of the measurements of 1500 Wellesley students. (70) M. Wood, Anthroporrietric tables, compiled from the measurements of 1100 Wellesley College students, 1890. (71) M. Wood, Statistical tables concerning the class of 1891 of Wellesley College, numbering 104 women. (72) M. Wood, Statistical tables, showing certain measurements of forty freshmen of Wellesley College (before and after gymnasium training), 1892. TEST I, Height, standing and sitting. — The general purpose of this test is, of course, to furnish a measurement of height as an index of physical size or growth for the sake of comparison with mental traits or with other physical traits. It is included in practi- cally every series of tests that include any physical measurements. Apparatus. — Stadiometer (Fig. 3). Small calipers (Fig. 4) or miUimeter rule. Method. — (1) For standing height, the examiner, E, should, when feasible, have the subject, S, remove his shoes, and stand on the stadiometer with the heels together and with heels, buttocks, the spine between the shoulders, and the head, all in contact with the measuring rod. The chin must not be unduly raised or depressed. E then brings down the sliding arm of the instrument until it rests squarely, but without excessive pressure, upon S's head. (2) For sitting height, let S sit erect upon the stand of the stadi- ometer with spine and head in contact with the measuring rod. Results. — (1) The best norms of stature are doubtless those calculated bj^ Boas (3)^ from studies by various investigators ' The figures in parentheses following names refer to the reference- numbers at the end of the test in which they occur. 52 ANTHROPOMETRIC TESTS of school children (45,151 boys and 43,298 girls) in Boston, St. Louis, Milwaukee, Toronto, and Oakland, Cal.i For the sake of comparison with these norms and with the norms for strength FIG. 3. STADIOMETER, OR HEIGHT STAND. Graduated in tenths of inches on one side and millimeters on the other. FIG. 4. VERNIER CALIPER, FOR EXTERNAL, INTERNAL, AND DEPTH MEASURING. Fitted with both English and metric scales and verniers for each, reading to lis of an inch and to of a millimeter. ^ The same averages converted into inches may be found in Burk, while these and other studies are summarized by MacDonald. Consult Boas for table showing the distribution of stature at each age according to the fre- quency method. Valuable tables and charts showing the distribution of height and of other anthropometric measurements by percentile grades will be found in Smedley (17). TEST 1 : HEIGHT 53 of grip, vital capacity, etc., to be quoted later, there are given herewith the norms of standing and sitting height derived from the measurement of 2788 boys and 3471 girls by Director Smedley of the Department of Child-Study and Pedagogic Investigation, Chicago (16). TABLE 10 Norms of Stature of American Children, in cm. (Boas)* Age 5.5 6.5 7.5 8.5 9.5 10.5 11.5 Boys Girls 105.90 104.88 111.58 110.08 116.83 116.08 122.04 121.21 126.91 126.14 131.78 131.27 136.20 136.62 Age 12.5 13.5 14.5 15.5 16.5 17.5 18.5 Boys Girls 140.74 142.52 146.00 148.69 152.39 153.50 159.72 156.50 164.90 158.03 168.91 159.14 171.07 *The figures in black-faced type in Tables 10-14 indicate periods in which the averages for girls exceed those for boys of the same ages. The rapid growth of puberty and early adolescence is initiated and terminated earlier in girls than in boys. TABLE 11 Norms of Standing and Sitting Height, in cm. {Smedley) STANDING HEIGHT SITTING HEIGHT AGE STANDING HEIGHT SITTING HEIGHT AQB BOYS GIBLS BOYS GIRLS BOYS GIRLS BOYS GIRLS 6.0 110.69 109.66 62.40 61.72 12.5 141.89 144.32 74.70 76.29 6.5 113.25 112.51 63.54 62.90 13.0 145.54 147.68 76.24 77.91 7.0 115.82 115.37 64.67 64.07 13.5 149.09 151.04 77.79 79.54 7.5 118.39 118. '22 65.78 65.25 14.0 151.92 153.64 79.21 80.99 8.0 120.93 120.49 66.75 66.34 14.5 154.74 156.24 80.64 82.43 8.5 123.48 122.75 67.72 67.43 15.0 158.07 156.83 82.18 83.21 9.0 126.14 125.24 68.79 68.32 15.5 161.41 157.42 83.68 83.99 9.5 128.80 127.74 69.85 69.21 16.0 164.03 158.30 85.43 84.54 10.0 130.91 130.07 70.56 70.05 : 16.5 166.65 159.18 87.17 85.09 10.5 133.03 132.41 71.26 70.89 17.0 167.85 159.26 88.16 85.20 11.0 135.11 135.35 72.10 72.23 17.5 169.04 159.34 89.14 85.30 11.5 137.19 138.30 72.93 73.58 18.0 171.23 159.42 90.30 85.51 12.0 139.54 141.31 73.80 74.93 18.5 173.41 159.50 91.46 85.72 54 ANTHROPOMETRIC TESTS From these and other statistics, the following important results may be gathered : (2) There is a period of slower growth in height in boys at 1 1 years of age, and a similar, though less marked, retardation in girls at nine years of age. (3) During the period of approximately 11 to 14 years girls are taller than boys of the same age, because the prepul^ertal acceleration of growth occurs earlier in girls. (4) Sitting-height follows the same general laws as standing- height. (5) Boys continue their growth in height later than do girls, i. e., maturity in height is not reached so early. (6) Children of purely American descent are taller than those of foreign-born parentage (Bowditch, Peckham) . (7) Children of the non-laboring classes are as a group taller than children of the laboring classes (Bowditch, Roberts). (8) According to Bowditch, large children make their most rapid growth at an earlier age than small ones, but according to Boas (1, 2) this induction is untenable. (9) The height of American-born children is modified by den- sity of population. Urban life decreases stature from five years of age (Peckham, 10, 11). (10) According to Kline, boys in the public schools are taller than boys in truant schools, save at the age of ten. Similarly, Smedley (17) found the boys in the Chicago School for incorrigibles and truants shorter than normal boys from the tenth year up. (11) Gratsianoff and Sack in Russia, and Porter (39-40), Mac- Donald and Smedley (17) in America, have concluded that bright children are taller than dull children. West (18), however, found exactly the opposite to be true, while Gilbert (6, 7) found no con- stant relation between height and mental ability. Porter and Smedley determined mental ability by the relation of grade and age, Gilbert and MacDonald by the teacher's estimate. (12) Children with abnormalities are inferior in height to children in general (MacDonald). Notes. — The upright measuring rod should be braced m such a manner that it will not be bent out of place by the pressure of *S's back. Many S's will be inclined to assume an unnatural TEST 1 : HEIGHT 55 position in this examination, especially to stretch themselves: the apparent height may be increased by as much as 20 to 30 mm. in this way. If it is not practicable to remove the shoes, height may be taken with, them on, arid the height of the heel may subsequently be determined by the use of the small calipers or millimeter rule, and then subtracted from the gross height; the resulting error will be very small. Height, as is well known, decreases slightly during the day, owing to the packing of the intervertebral cartilages and the loss of muscular tone: this loss in height during the day amounts, in the case of young men, to from 10 to 18 mm. It is therefore desirable, for accurate work, to take height measurements at approximately the same period of the daj^ It might be possible to work out empirically a corrective formula. Porter's correlation between height and mental ability raises an important question which reappears whenever we discuss the correlation between any physical trait, e. g., weight, strength, vital capacity, etc., and mental abihty. The trend of evidence is to the effect that all such correlations, where found, are largely explicable as phenomena of growth, i. e., as correlations with relative maturity (Cf. Boas, 2; Wissler, 19). This makes intelli- gible the fact that, in general, the positiveness of all such correla- tions lessens with age, and that many of them, indeed, become difficult or impossible of demonstration in adults. Thus, to take the correlation in question, a positive correlation is not, of course, to be interpreted as meaning that, taken individually, all tall boys are, ipso facto, bright boys, but that, taken collectively, those boys whose physical condition is good, whose growth is unimpaired by ill-health, faulty nutrition, etc., and who reahze to the full the possibility of physical development inherent in them (whether they will ultimately be short or tall) will be found to exhibit the best mental condition and the most rapid mental development. REFERENCES (1) F. Boas, The growth of children, m Science, 19: 1892, 250. (2) F. Boas, On Dr. W. T. Porter's investigation of the growth of the scliool children of St. Louis, in Science, n. s. 1: 1895, 225. 56 ANTHROPOMETRIC TESTS (3) F. Boas, Growth of American children, in U. S., 1896-7, ii. 1555. (4) H. P. Bowditch, The growth of children, studied by Galton's method of percentile grades, in 22d An. Rept. State Brcl. Health Mass., 1891, 479. (5) F. Burk, Growth of children in height and weight, in A. J. P., 9: 1897-8, 253. (6) J. A. Gilbert, Researches on the mental and physical development of school children, in Yale S., 2: 1894, 40-100. (7) J. A. Gilbert, Researches upon school children and college students, in Iowa S., 1: 1897, 1-39. (8) L. W. Kline, Truancy as related to the migratory instinct, in Pd. S. 5: 1898, 381-420. (9) A. MacDonald, Experimental study of school children, etc., in U. S., 1899 (chs. 21 and 25.). (10) G, Peckham, The growth of children, in Rept. Wis. Brd. Health, 1881, 28. (11) G. Peckham, Various observation on growth, ibid., 1882, 185. (12) W. T. Porter, The growth of St. Louis children, in Trans. Acad. Science, St. Louis, 6: 1894, 2(53. (13) W. T. Porter, Physical basis of precocity and dullness, ibid., 6: 1893- 4, 160. (14) C. Roberts, A manual of anthropometry. (15) Sack, Physical development of the children in the middle schools of Moscow, 1892. (16) F. W. Smedley, Rept. of the dept. of child-study and pedagogic investigation, reprint from 46th An. Rept. Brd. Educ, Chicago, 1899-1900. Also in U. S., 1902, i., 1095-1115. (17) F. W. Smedley, do.. No. 3, 1900-1901, also in U. S., 1902, i., 1115-1138. (18) G. West, Observations on the relation of physical development to intellectual abiUty, made on the school children of Toronto, Canada, in Science, n. s., 4: 1896, 156. (19) C. Wissler, The correlation of mental and physical tests, in P. R. M. S. 3: No. 6, 1901. Pp. 62. TEST 2 Weight.^ — The general purpose of determining weight is similar to that of determining height, viz : to furnish an index of physical size or growth as a basis for correlation with other tests or observa- tions. Apparatus. — Accurate scales, joreferably of the type especi- ally devised for anthropometric work, which allow readings to be rapidly and accurately taken in the metric system, with units of 50 g. or twentieths of a kilogram (Fig. 5). If avoirdupois TEST 2: WEIGHT 57 scales are, used, they should be divided into tenths of pounds rather than into ounces. Method and Treatment of Results. — For accurate measure- ments, weight should be taken without clothes. Where this is impracticable, the weight of the clothes may be deducted by sub- sequent measurement. For some comparative purposes, however, the weight of the clothes may be neglected and the figures obtained A>. i Ml'.Ui UilLlUlL M ALES. The platform side of the beams is graduated metric to 100 kilos, by 50- gram divisions, and the other side avoirdupois to 200 pounds, by tenths of a pound. from the gross weight may be taken for computation, or these fig- ures, better yet, may be corrected by arithmetical computation based upon the weights of the clothes of a limited number of S's. We may form a tolerably accurate notion of the ' clothing error' by reference to investigations upon this point. Thus, according to W. S. Christopher (3), who ascertained the weight of the ordi- 58 ANTHROPOMETRIC TESTS nary schoolroom clothing of 121 Chicago children, chiefly in the month of May, "the average weight of the clothing of all the pupils was 5.5 per cent of the gross weight" (boys, 5.8 per cent; girls, 5.2 per cent.) These figures varied little with age: obese children wore clothing lighter in proportion to their weight than that worn by others, while "the most variable element in the cloth- ing was found to be the shoes, especially the shoes worn by the boys." Only a few children wear clothing that weighs more than 7 per cent, or less than 4 per cent of their gross weight. Results. — (1) From the data of about 68,000 children in the cities of Boston, St. Louis, and Milwaukee, Burk (2) derives the norms reproduced in Table 12: the Chicago norms are reproduced in Table 13. TABLE 12 Norms of Weight, in kg. (Burk) Approx. AgC; 6.5 Boys...: .20.50 Girls 19.6921.64 10.5 11.5 12.5 13.5 1 14.5 15.5 22 . 45 24 . 72|27 . 03 29 . 66:32 . 07,34 . SSLSS . 46 43 . 18 48 . 72 54 . SSj - 23 . 81 26 03 28 . 53 31 . 52 35 . 7040 23 44 . 59 48 . 40 50 . 94 52 . 34 TABLE 1.3 Norms of Weight, in kg., with Clothing (Smedley) 11.0 12.0 14.0 I 15.0 19 . 738;21 . 613 23 . 817 26 . 336 28 18.87020.97423.01025.257 707 31 . 22334 . 151 38 . 08442 . 696 47 . 993 53 27 . 795bo . 662 34 373 38 974 44 219 48 . 161 23857.384'61.283 50.65252.386'52.923 (2) As in the case of height, girls exhibit the prepubertal in- crease in weight some two years earlier than boys, and are for the years 12 to 15 heavier than boys of the same age. (3) Growth in weight, as in height, is subject to some lessen- ing of rate at 9 years for girls and at 11 for boys. (4) Boys continue to increase in weight after girls have prac- tically attained their maximal normal weight. Girls grow most rapidly from 10 to 15 years, boys from 12 to 17 years. (5) Mean variations in weight are largest during the period of TEST 2: WEIGHT 59 fastest growth, which shows that not all individuals participate equally or evenly in the rapid growth of adolescence. (6) First-born children exceed later-born children in weight, at least during the period from 6 to 15 years, though the reverse is true of the weight at birth. The difference is slight, but very regular (Boas, 1). (7) Children of the non-laboring classes are as a group heavier than children of the laboring classes (Bowditch). (8) Children of American-born parents are heavier than those of foreign-born parents. (9) The correlation between weight and mental ability or pre- cocity is found to be positive by some investigators, negative by others, and indifferent by still others. Thus^ Porter (8) asserts very positivelj^ that "precocious children are heavier and dull children lighter than the mean child of the same age," and draws a further practical conclusion that "no child whose weight is below the average for its age should be permitted to enter a school grade beyond the average of its age, except after such a physical examination as shall make it probable that the child's strength be equal to the strain." Porter's conclusions are confirmed by Smed- ley (9) at Chicago. On the basis of the teacher's estimate of men- tal ability, Gilbert (4, 5), however, finds no constant relation be- tween weight and such ability, save that from 10 to 14 years the dull children are much heavier than the bright, while West (10), who used a similar basis, finds a negative correlation throughout. (10) Both Khne (6) and Smedley (9) fuid the mean weight of boys in truant schools to be less than that of boys in the public schools, save at the age of 10. (11) Porter concludes that the acceleration in weight preceding puberty takes place at the same age in dull, mediocre, and preco- cious children, but investigations in New York City seem to oppose this conclusion and indicate rather that puberty and pubertal growth is distinctly earlier in precocious children, i. e., that mental and physical precocity go hand in hand. (12)' Children with abnormalities are below the average in weight (MacDonald, 7). Notes. — It is not important to have scales which render possible a very fine measurement, such as fractions of an ounce, because the 60 ANTHROPOMETRIC TESTS normal weight of any individual varies from day to day and from hour to hour during the day : the daily variation is, in the case of young men, as high as 0.3 kg. The author, in a long series of observa- tions conducted at the same hour daily, found gains and losses ,of more than 1 kg. in 24* hours. It may not be amiss in this connection to point out the absurdity of attaching any sig- nificance to small gains or losses that are observed in weighings conducted at occasional and irregular intervals. Severe exercise may reduce the weight by a large amount; e.g., two hours of foot- ball practise may take off 2 or 3 kg. from a man who is not yet in training. It is well, however, for comparative purposes, to take weight measurements at approximately the same period of the day. REFERENCES (1) F. Boas, Growth of first-born children, in Science, n.s. 1 : 1895, 402-4. (2) F. Burk, Growth of children in height and weight, in A. J.P., 9: 1897-8, 253. (3) W. S. Christopher, Report on child-study, reprint from An. Rept. Brd. Educ. Chicago, 1898-1899. (4) J. A. Gilbert, Researches on the mental and physical development of school children, in Yale S., 2: 1894, 40-100. (5) J. A. Gilbert, Researches upon school children and college students, in lowaS., 1: 1897, 1-39. (6) L. W. Kline, Truancy as related to the migratory instinct, in Pd.S., 5: 1898, 381-420. (7) A. MacDonald, Experimental study of school children, etc., reprint of chs. 21 and 25 of U. S., 1899. (8) W. T. Porter, Physical basis of precocity and dullness, in Trans. Acad. Science St. Louis, 6: 1894, 263. (9) F. Smedley, Rept. dept. child-study and pedagogic investigation, No. 3, in Rept. Brd. Educ. Chicago, 1900-1901, also in U. S., 1902, i., 1115- 1138. (10) G. West, Observations on the relation of physical development to intellectual ability, made on the school children of Toronto, Canada, in Science, n.s. 4: 1896, 156. TEST 3 Diameter of the skull. — This measurement has been commonly conducted for the purpose of investigating the correlation between size of the head and general intelligence. It forms also one of the chief measurements undertaken in the Bertillon system TEST 3: DIAMETER OF THE SKULL 61 for the identification of criminals. The following directions are adapted from Bertillon's account (1). A. MEASURING THE LENGTH OF THE HEAD Instrument.— Head calipers (Fig. 6). Method. — (1) Seat S with his right side toward a window, and stand facing his left side. Hold the left tip of the calipers firmly in place at the glabella (space between the eye-brows) with the FIG. 6. HEAD CALIPERS. tip of the instrument between the thumb and forefinger, and with these resting on the adjacent parts of the forehead to prevent the compass-tip from deviating. (2) Hold the calipers in an approximately horizontal plane so that the scale is fully lighted by the window, with the right tip projecting about one cm. beyond the finger-tips of the right hand. Keep the eyes fixed upon the scale; then bring the right tip down over the back and middle of the head until it has passed the most projecting point; then m^ve the tip upward again, making sure that it is well within the hair and in constant contact with the scalp ; continue these exploring movements so as to pass the maximal point two or three times, keeping the eyes constantly fixed upon the scale to detect this point. (3) Remove the calipers and set them by tightening the set- screw at the supposed length; take care to set them accurately within 0.5 mm. 62 ANrHROPOMETRIC TESTS (4) Replace the calipers thus set and tightened, and again execute the exploring movements described in (2). If the setting is correct, the instrument will just touch the skin of the head at the maximal point, but will pass over it without undue friction and without necessitating pressure upon its arms: one millimeter too short will produce definite resistance at this point; one millimeter too long, a definite lack of friction. Practise will enable E to dis- tinguish the 'feel' of the correctly set instrument, and errors should not exceed 1 mm. B. MEASURING THE WIDTH OF THE HEAD Instrument.— Head calipers as above. Method. — Position of S, preliminary exploring movements, setting of the calipers, and subsequent verification follow the same general procedure as in the determination of the length of head. The following additional instructions are to be noted : (1) E stands behind S, and is careful to preserve an erect, symmetrical position, in order to ensure equal freedom with both elbows and a symmetrical position of the calipers. (2) Hold the calipers a short distance from each end; apply the tips first at the upper point of attachment of each ear; then raise them vertically and watch the scale to determine the point of great- est width, making several testing movements both upward and downward. (3) The true maximal diameter in most cases is not yet found, but lies in the same horizontal plane as the preliminary maximal point just determined, and about 3 cm. behind it. Hence, next move the calipers slowly back and forth two or three times in a horizontal plane and determine the true maximal point. (4) Set the instrument, as in the previous measurement, and verify the setting. In this verification, the caliper-points should describe a series of zig-zag movements, in order certainly to tra- verse the areas of maximal width (usually less than the size of a dime), which might not be traversed if the movements were cir- cular or too coarsely executed. Treatment of Results. — From the measurements of the length and width of head, the cephalic index may be computed readily by TEST 3: DIAMETER OF THE SKULL 63 multiplying the width by 100 and dividing by the length. This index is considered one of the most important of those used in anthropometry. By it, the type of head may be determined as follows : if the index is less than 75, *S is long-headed (dolichoceph- ahc); if 75-80.9, S is 'medium' (mesocephahc) ; if 81-86.9, S is broad-headed (brachycephalic) ; if 87 or over, S is excessively broad-headed (hyperbrachycephalic) . Results. — (1) Typical head measurements are those made by Boas, West, Chamberlain and others upon Worcester school chil- dren, and reported by West (7) : these are reproduced in Table 14. TABLE 14 Diavielers of the Skull and the Cephalic Index (West) AVERAGE LENGTH /ERAGE WIDTH CEPHALIC INDEX Boys Boys Boyi mm. nun 7nm. mm. % % 5 176 174 140 138 79.56 79.40 6 177 172 142 139 78.94 79.60 7 179 175 142 140 79.42 80.02 8 180 174 143 141 78.71 80.41 9 181 176 144 140 79.63 79.71 10 182 177 145 142 80.30 79.46 11 183 180 144 142 78.80 78.90 12 183 180 145 143 79.40 79.40 13 184 181 147 145 79.50 79.60 14 1 187 183 147 144 78.60 79.00 15 1 188 184 148 146 78.59 78.99 16 1 191 184 149 144 77.81 78.48 17 ■ 189 185 150 146 78.34 78.50 18 192 186 151 147 78.88 79.36 19 192 183 150 145 78.33 79.68 20 195 182 152 147 77.88 79.41 21 192 186 153 145 79.29 78.36 (2) In boys, length of head continues to increase until the age of 21; in girls, maxmial length of head is practically attained at 18. The growth, both of length and width of head, is very irregular, i.e., periods of growth alternate with periods of cessation of growth. (3) Boys' heads are longer and wider than those of girls through- out the whole period of gro^Hh, and consequently throughout life 64 ANTHROPOMETRIC TESTS (4) Width of head is greater in precocious than in dull children (Porter, 6, 7). One method of arraying data to show this prin- ciple is illustrated in Table 15, in which all the girls aged 12 and all the boys aged 10 are distributed according to their school grades. It is then seen that those children of a given age in an advanced school grade have, on the average, broader heads than those in a lower grade. TABLE 15 Breadth of Head by School Grade (Porter) BOYS AGED 10 GIRLS AGED 12 SCHOOL GRADE Cases Average Cases Average mm. mm. I . . 92 408 145.86 146.73 68 _ II 143.68 Ill 397 146.48 193 144.77 IV 170 147.21 343 144.94 V - — 217 145.50 VI - - 89 147.64 (5) Binet, somewhat similarly, finds that the head of the unin- telligent is smaller than that of the intelligent child in all dimen- sions save in vertical diameter and distance from the base of the nose to the end of the chin, though the differences are but slight and somewhat uncertain. If, however, exceptionally bright children {enfants d'elite) are compared with exceptionally dull children (enfants arrieres), differences averaging 3-4 mm. or more appear, particularly in transverse dimensions, i.e., in breadth. Excep- tionally bright children distinctly surpass average children, but the latter do not differ so much from dull children. In brief, then, exceptionally bright children are characterized by unusually wide heads. (6) According to MacDonald, "dolichocephaly increases in children as ability decreases. A high percentage of dolichocephaly is, to a certain extent, a concomitant of mental dullness." "Un- ruly boys have a large percentage of long-headedness." (7) The measurements, by Engelsperger and Ziegler, of 238 boys and 238 girls of the entering classes (average age 6 years, 4.5 TEST 3: DIAMETER OF THE SKULL 65 months) in the schools of Munich, furnish results that deviate somewhat from those just cited for American children, as is seen clearly by a comparison of Tables 14 and 16. It is of interest to note that no cases of dolichocephaly were found, but that these children were decidedly brachy cephalic. (8) Miss Lee found no correlation between the estimated skull capacity and the intellectual capacity of 60 men and 30 women. Notes. — ^Heads of unusual shape or size, irregular or deformed, should receive especial care in measurement, and a descriptive note should be appended to the record. Attempts to record the shape and size of the skull by means of the registering ' conformateur' used by hatters have usually been relinquished, because the hair interferes too much with exact de- Skull Dimensions and Proportions of Entering Classes at Munich (Engel- sperger and Ziegler) LENGTH BREADTH INDEX Mean Max. Min. Mean Max. Mm. Meso. Brachy. Hyperb. Boys 170.35 165.83 185 186 148 151 146.34 142.97 160 159 133 130 % 6.3 6.7 54^6 45.8 % 39.1 Girls 47.5 termination. This instrument might, however, be of service in preserving a rough 'picture' of heads of unusual size or propor- tions. Measurement of the head was formerly regarded as a very ob- vious means for the estimation of the size and proportions of the brain, and hence of intellectual ability, but it is easy to demon- strate that these relations obtain only in the gross. That is, brain size and form are only roughly indicated by the exterior dimensions of the head, while intelligence is conditioned primarily by the elab- orateness of the finer nerve structure and not (save in pathological cases of hypertrophy or developmental arrest) by the gross size or form of the brain. It is scarcely necessary to call attention to the absurdity, a fortiori, of the claims of phrenology. 66 ANTHROPOMETRIC TESTS REFERENCES (1) A. Bertillon, Signaletic instructions; including tlie theory and prac- tise of anthropometrical identification, Eng. tr. , Chicago and New York, 1896. Pp. 260. (2) A. Binet, Recherches sur la technique de la mensuration de la tete vivante, and four other articles on cephalometry, in A. P., 7: 1900 (1901) 314- 429. (3) A. Engelsperger and O. Ziegler, Beitriige zur Kenntniss dcr physischen und psychischen Natur des sechsjilhrigen in die Schule eintretenden Kindes. Anthropometrisches Teil in E. P., 1: 1905, 173-235. (4) Alice Lee, Study of the correlation of the human skull, in Science, n. s. 12: 1900, 946-9. (5) A. MacDonald, Experimental study of children, etc., reprint of chs. 21 and 25 of U. S., 1899. (6) W. T. Porter, The growth of St. Louis children, in Trans. Acad. Science St. Louis, 6: 1894, 263. (7) W. T. Porter, Physical basis of precocity and dullness, ibid., 160. {8"> G. West, Worcester school children; the growth of the body, head, and face, in Science, 21 : 1893, 2-4. TEST 4 Girth of the skull.— This measurement is less in favor with investigators than those just described, because of the variable factor of the hair, just mentioned. Instrument.— Anthropometric measuring tape (Fig. 7). Method. — E stands at the right of ;S, who is seated. E holds 1 llllllljllll 'l"2l" i|ii3|iiii| '" 15 ■liSBiiiB^^^^^^^ || ANTHROPOMETRIC TAPE. (iraduateid tojt^Jjths of, inches on one side and millimeters on other, and fitted \yith^,a spi'ing handle to eliminate the pensonal vqni of different exami'ne'ts.' TEST 4: GIRTH OF THE SKULL 67 the tape with the thumb and forefinger of each hand at a length approximately that of the distance to be measured. He then lifts the tape over S's head, keeping it horizontal, and applies it about the head at such a height as to pass around the largest part — over the frontal prominences and over the occipital prominences. The tension of the tape is regulated by observation of the spring- indicator. Results. — (1) Measurements of the circumference of the head of 7953 boys and 8520 girls in Washington, D.C., by MacDonald (3) ' form the basis for the results embodied in Table 17. (2) The head circumference of boys is larger than that of girls save in the case of colored children. (Colored girls have a larger circumference of head at all ages than white girls.) In American table 17 Circumference of the Head, in Inches (MacDonald) AGE BOYS GIRLS AGE BOYS GIRLS 6 20.48 20.20 13 21.01 20.95 7 20.45 19.94 14 21.21 21.18 8 20.51 20.14 15 21.45 21.28 9 20.61 20.29 16 21.67 21.38 10 20.73 20.43 17 21.87 21.55 11 20.82 20.54 18 21.91 21.60 ^ 12 20.94 20.78 children, the measurements of girls most nearly approach those of boys at 13 and 14, or at the period when the girls excel in height and weight. (3) Children of the non-laboring classes have a larger circum- ference of head than children of the laboring classes. (4) Children with abnormalities are inferior in head circumfer- ence to normal children. (5) In an examJnation of 60 juvenile delinquents, Dawson (2) found the average circumference of head less than that of normal children of the same age : in 64 per cent of the cases studied, the ^Consult MacDonald (pp. 1016ff.) for an extended discussion of the relation of circumference of head to sex, nativity, race, sociological condition and mental ability. 68 ANTHROPOMETRIC TESTS circumference was from 1.7 to 5.2 cm. less than the mean for normal children. (6) According to MacDonald, as circumference of head increases, mental ability (as reported by the teacher) increases, provided that one and the same race be under consideration. Mobius (4), simi- larly, asserts that, at least in the case of normal (gesunde) adults, mental capacity tends to exhibit correlation with skull capacity. Bayerthal (1) measured the skull circumference of 234 boys and 153 girls (ages 7.5 to 8.5 years) and related these measures with school standing by classifying both sexes into five groups, as "very good" (I), "good" (II), "good on the whole" (III), "satisfactory" (IV), and "more or less unsatisfactory " (V). The results tend to con- firm the existence of a positive correlation between skull circum- ference and general ability. Thus, the average skull-circumference s. Avere, for boys, 51.46, 50.93, 50.33, 49.60, and 49.60 cm., and for girls, 50.00, 49.83, 49.44, 49.16, and 48.84 cm., for the groups T to V, respectively. The same investigator found that in one class of 48 girls, who had been classed by their teacher into the three groups, good, average, and poor, the average skull circumference for these groups was 49.6, 49.16, and 48.75 cm., respectively. REFERENCES (1) Bayerthal, Kopfumfang unci Intelligenz im Kindesalter, in E. P., 2: 1906, 247-251. (2) G. E. Dawson, A study in youthful degeneracy, in Pd. S., 4 : 1S96, 221- 258. (3) A. MacDonald, Experimental study of school children, etc., reprint of chs. 21 and 25 of U. S., 1899. (4) Mobius, Geschlecht und Kopfgrosse, Halle, 1903. CHAPTER V Tests of Physical and Motor Capacity The title 'physical and motor capacity' is here used as a Conveni- ent and practical phrase to cover a number of tests which have often been classified under diverse rubrics, such as strength tests, motor tests, physical tests, tests of physiological condition, etc. All of the tests here described differ from the anthropometric tests of Chapter IV in that they measure not mere size or dimension, but functional, especially muscular, capacity. They differ from the tests of Chapter VI, many of which might equally well be said to measure physiological condition or capacity, e.g., the test of visual acuity, in that they are primarily tests of motor, rather than of sen- sory capacity. The first test described, that of vital capacity (often loosely termed lung capacity) is, perhaps, not so obviously a test of muscu- lar efficiency as are the four strength tests that follow. It is, how- ever, clearly a test of physical capacity dependent upon movement. The tests of quickness, accuracy, and steadiness of movement are frequently placed in a class by themselves under the rubric 'motor tests.' but they are easily subsumed under the caption here employed. Reaction-time would by many be considered a test of quickness of movement; but it is so largely dependent upon complex psycho- logical conditions, particularly upon the instructions, the direction of attention, and the type of stimulus employed, that it belongs rather to the experimental examination of action than to the meas- urement of physical capacity as such.^ These tests of physical and motor capacity have become promi- nent chiefly because of their employment in the study of the corre- ^ See an article by the writer, "Reaction-Times as a Test of Mental Abil- ity," in A. J. P., 15: 1904, 489. 70 PHYSICAL AND MOTOR CAPACITY lation of physical and mental abilitJ^ For this purpose they are commonly used in conjunction with the anthropometric tests already described and with various tests of general intelligence or mental ability to be described later. These tests have also an obvious and direct application in the stud}^ of various problems of hygiene, physical culture, etc. TEST 5 Vital capacity.— Vital capacity, also termed breathing capacity and differential capacity, is the maximal volume of air that can be FIG. S. WET SPIROMETER. Graduated in cubic inches and cubic decimeters. expired after taking a maximal inspiration. It is not identical with lung capacity, because a certain amount of air, termed the residual air, always remains in the lungs. Vital capacity is considered an important index of general physi- TEST 5: VITAL CAPACITY 71 cal condition and capacity, and has, accordingly, found a place in nearly all measurements of school children in which the physical status has been examined. It is affected by sex, age, stature, pos- ture, occupation, amount of daily physical activity, and by disease, and may be markedly increased {e.g., 300 cc, in three months) by various forms of physical exercise which demand active respiration. The ratio of vital capacity to weight is termed the vital index and is held to be of extreme significance, because it expresses the balance between bodily size and the rate and completeness with which oxidization of the blood is, or may be, effected. A high vital index is undoubtedly a preventive of auto-intoxication, gives increased resistance to disease, and is the root of endurance under effort. Thus athletic training consists primarily in the reduction of weight and the increase of breathing capacity. TABLE 18 Norms of Vital Capacity, in Cubic Centimeters (Smedley) AGK BOTS GIRLS .G. BOT9 GIRLS 6 1023 950 13 2108 1827 7 1168 1061 14 2395 2014 8 1316 1165 15 2697 2168 9 1469 1286 16 3120 2266 10 1603 1409 17 3483 2319 11 1732 1526 18 3655 2343 12 1883 1664 L_„._ Apparatus. — Spirometer, preferably of the wet type (Fig. 8) fitted with detachable wooden mouth-piece. Extra mouth- pieces. Method. — See that S's clothing is perfectly loose about his neck and chest. Instruct him to stand upright, to take as full an inspira- tion as possible, and then to blow, not too rapidly, into the spirom- eter. Also caution him to take care that no air escapes about the mouth-piece. Two or three trials may be allowed, and the best record set down. After S's record is made, discard the mouth-piece and insert a new one into the rubber tube. 72 PHYSICAL AND MOTOR CAP.\C1TY Results.— (1) The norms of vital capacity embodied in Table 18 are those established by Smedley (6) at Chicago. (2) The relation between weight and vital capacity, i.e., the vital index, presented in Table 19, is that fomid by Kotelmann (4), also given by MacDonald (5). The ratio expresses the relation in terms of kg. of weight and cc. of vital capacity. It will be seen that the weight of the body normally increases with age somewhat faster than the vital capacity. If height be similarly treated, it will be fomid, on the contrary, that vital capacity increases with age faster than it increases. (3) All investigators agree that boys have a larger vital capac- ity than girls at all ages, and that men, similarly, have a larger capac- TABLE 19 Value of the Vital Index, token Weight is Taken as Unity {Kolelniann) i 1; ' AGE INDEX 'age INDEX L AGE INDEX AGE INDEX 9 69.32 12 ! 67.51 1 15 63.18 IS 64.28 10 69.37 13 I 66.75 16 65.94 19 66.22 11 1 69.18 14 1 64.07 j 17 1 65.77 |l 20 i 65.01 ity than women. Even if we compare men and women of the same height, the former surpass the latter by about the ratio 10 : 7.5. (4) The average capacity for adults is computed b}^ Vierordt to be for men 3400 cc. (a figure noticeably lower than the Chicago norms for boys in later adolescence) and for women 2400 cc. (5) The norm is conditioned by height. For each centimeter of increase or decrease of stature above or below the mean, there is a corresponding rise or fall of the vital capacity, amounting in men to 60 cc, in women to 40 cc. This correlation with height varies somewhat at different ages. Thus, according to Wintrich, the average vital capacity for each centimeter of height is from eight to ten years, 10 cc, from 16 to 18 years, 20.65 cc, and at 50 years, 21 cc. (6) In boys, growth in vital capacity is slow and steady during the years 6 to 12, but very marked during the next four years, whereas in girls the most rapid increase is during the years 11 to 14. In both sexes, these periods of rapid growth coincide with the peri- ods of rapid growth in height and weight. TEST 5: VITAL CAPACITY 73 (7) Beyer (1), from his study of naval cadets, concludes that the maximal vital capacity is reached at 19, but other authorities place the maximum at 35, with an annual decrease of about 32 cc. there- after, up to the age of 65. (8) The most marked individual differen ces appear at the time when the period of most rapid gtowth terminates (Smedley, 7). (9) Vital capacity is proportionately reduced in men who live a sedentary life. It is also reduced by any circumstance which inter- feres with the free expansion of the thorax, such as tight clothing, tuberculosis of the lungs, visceral tumors, etc. (10) The correlation of vital capacity and mental ability is indifferent or negative according to Gilbert (2, 3), who found no con- stant relation, save that from 10 to 15 years duller children have the larger capacity. On the other hand, Smedley (6, 7) found a posi- tive correlation between school standing and vital capacity, whether he took the distribution through the grades of all pupils of a given age, or computed the average school-grade of those who stood at various precentiles of vital capacity, or compared those at and above grade with those below grade at each age. Moreover, the same investigator found that pupils in the John Worthy School (incor- rigibles, truants, etc.) were, from the age of ten up, inferio. in vital capacity to children in the other schools, and that the in- feriorit}^ became more noticeable with age. Notes. — The dry spirometer is less expensive than the wet in first cost, and is more portable, but it has the disadvantage of get- ting out of repair easily. Its readings are apt to run slightly higher than those of the wet spirometer. The mouth-piece of the ordinary spirometer forms an excellent medium for the dissemination of bacteria. For this reason the detachable mouth-pieces are imperative if hygienic conditions are to be assured. There is a certain knack in making a maximal spirometer record ; some children may exhibit it; others not. In particular, to get a good record, the expiration must be neither too fast nor too slow, and an extra effort must be made just at the end of both inspiration and expiration to utilize the available lung-capacity to the utmost. 74 ' PHYSICAL AND MOTOR CAPACITT REFERENCES (1) H. G. Beyer, The growth of United States naval cadets, in Proc. U. S. N. Inst., 21: No. 2, 1895. (2) J. A. Gilbert, Researches on the mental and physical development of school children, in Yale S., 2: 1894, 40-100. (3) J. A. Gilbert, Researches upon school children and college stu- dents, in Iowa S., 1: 1897, 1-39. (4) L. Kotelmann, Die Korperverhaltnisse der gelehrten Schliler des Johanneums in Hamburg, in Zeits. d. Konigl. Preus. statist. Bureaus, 1879. (5) A. MacDonald, Experimental study of school children, etc., reprint of chs, 21 and 25 of U. S., 1899. (6) F. Smedley, Rept. dept. child-study and pedagogic investigation, reprint from 46th An. Rept. Brd. Educ. Chicago, 1899-1900. Also in U. S., 1902, i., 1095-1115. (7) F. Smedley, do., No. 3, 1900-1901. Also in U. S., 1902, i., 1115-1138. TEST 6 Strength of grip. — This test has been used to secure an mdex of general bodily strength, to secure an index of righthandedness^ (in conjunction with Tests 10 to 12), and for comparative purposes generally. It may be modified to secure an index of endurance or fatigue (Test 9), or combined with other forms of strength measure- ment (Tests 7 and 8). Apparatus.— Improved form of Smedley's dynamometer (Fig. 9). Millimeter rule. Method. — With the millimeter rule, measure the distance from where S's thumb joins his hand to the end of his fingers. Adjust the dynamometer by whirling the inner 'stirrup' until the scale on the outer stirrup indicates one-half this distance. This should bring the second phalanx to bear against the inner stirrup, and will ordinarily prove to be the optimal adjustment; if not, it may be modified to suit S's inclinations. Then set the instrument by means of the clutch, so that the inner stirrup cannot twist while in use, and record the adjustment by reference to the scale upon the stirrup. ' The terminology of right and left-handedness is at present somewhat confused (Cf . E. Jones, in P. B., 6: April, 1909). The terms 'index of unidex- terity' and 'index of dextrality' have been used by some writers as equivalent to 'indexof righthandedness.' 'Dextrality' is here used to indicate the supe- riority of one hand (whether right or left) over the other. TEST 6: STRENGTH OF GRIP 75 Illustrate the use of the instrument to S: especially make clear that the lower pointer will register the grip, so that he does not have to continue his effort while the scale is read. Allow three trials with each hand, right and left alternately, but introduce a brief pause, say 10 sec, between each trial to avoid excessive fatigue. Have S exert his maximal grip, and in each trial encourage him to do his best. Record the amount registered at each trial; but, for ordinary purposes, use in subsequent computa- tion only the highest record for each hand. FIG. 9. DYNAMOMETER AND DYNAMOGRAPH, AFTER SMEDLEY, IMPROVED. Results. — (1) Tests of 2788 boys and 3471 girls in Chicago (9) with the Smedley dynamometer yielded the norms of Table 20. (2) Boys are uniformly stronger than girls, and men stronger than women. (3) The divergence between the sexes becomes marked at puberty, with the appearance of other sex traits. (4) Individual variation in strength is more evident in early adolescence than at any other time. (5) In his study of Washington school-children, MacDonald (6) found no correlation between strength of hand and mental ability, re PHYSICAL AND MOTOR CAPACITY but rather a dependence of strength upon sociological condition, i.e., children of the poorer classes work outside of school hours and thus develop their strength. Since from these and other causes, the percentage of dull children in such a group is liable to be large, this accounts for the indications in his results that dull children tend to surpass bright or average children in strength. Yet Schuyten (7) concludes that the children of well-to-do parents are stronger than the children of poor parents. TABLE 20 Norms of Strength of Grip, in kg. (Smedley) BOYS GIRLS Rt. Hand Lt. Hand , Rt. Hand Lt. Hand ■ 6 8 9.21 10.74 12.41 14.34 16.52 18.85 21.24 24.44 28.42 33.39 39.37 44.74 49.28 8.48 10.11 11.67 13.47 15.59 17.72 19.71 22.51 26.22 30.88 36.39 40.96 45.01 ' 8.36 1 9.88 11.16 12.77 14.65 16.54 18.92 21.84 24.79 27.00 28.70 29.56 29.75 7.74 9.24 10.48 9 11 97 10 13.72 11 12 13 14 15 15.52 17.78 20.39 22.92 24.92 16 26.56 17 27.43 18 27.66 (3n the other hand. Miss Carman (3), from measurements of the grip of 1507 boys and girls aged 10 to 19 years, found that bright children exceeded dull children by an average of 3 kg. with the right, and 1 kg. with the left hand. In Chicago (9, 10) the existence of a positive correspondence between strength of grip and class-standing was shown by three different methods, viz: by the distribution of 12-year old pupils by grades, by comparing the grip of those at and above grade with the grip of those below grade at each age, and by computing the aver- age number of school grades that had been made by the various TEST 6: STRENGTH OF GRIP 77 percentile groups (in strength), after sex and age had been ehmin- ated. Schuyten, who estimated inteUigence by school grade in rela- tion to age, also found that those who are most intelligent are strongest. (6) Dawson (5) found that juvenile delinquents have a mean strength of grip slightly less than normal children and that 56 per cent of them are inferior to the normal by from 1.32 to 11.82 kg. Similarly, boys in the school for incorrigibles and truants at Chi- cago are, at every age from 9 to 17 and with either hand, less strong than normal boys and this discrepancy increases very decidedly with age, e.g., from 96.8 per cent of the norm at the age of 9 to 63.2 per cent of the norm at the age of 17. (7) The index of righthandedness, i.e., the percentage of strength of the left hand compared with the right, will be found to range, for any ordinary group of school children, between 91 and 96 per cent. (8) Dextrality, i.e., superiority of one hand over the other, is evi- dent when the child enters school, but becomes increasingly evident as maturity approaches, and especially at puberty, so that a height- ened difference in the strength of the hands may be regarded as one of the characteristic indications of pubertal change. (9) There is a positive correlation between dextrality and intellec- tual ability (Smedley, 9), i.e., dull pupils are more nearly ambi- dextrous than average, and average than bright, pupils, while the John Worthy schoolboys are still more nearly ambidextrous than the dull pupils of the regular schools. (10) The degree of dextrality is greatest in the strongest children and least in the feeble, so that the latter may be said, as it were, to have two left hands (Binet and Vaschide, 1). (11) If the test is taken under stimulating conditions, such as competition, personal encouragement, public announcement of records, etc., Binet and Vaschide found that the average grip was increased about 3 kg., or so much that the left hand surpassed the previous record of the right hand made without such incitement. Similarly, Schuyten (8) found that ennui, or loss of interest in suc- cessive tests, is sufficient to obscure the fatigue-effect of a school session. (12) The exertion of maximal strength is commonly accom- 78 PHYSICAL AND MOTOR CAPACITY panied by characteristic poses, attitudes, facial contortions, grima- ces, etc., which are, in general, evidences of the escape of uncontrol- led energy through various motor paths. There appears to be an inverse relation between the strength and efficiency of the subject and the number and extent of these waste movements; these are correspondingly more evident when the muscles tire and aS^ is unable to accomplish what he is attempting. In particular, a sort of fool- ish laugh is characteristic of this muscular inefficiency.^ Notes.— The chief objections which have been made to the employment of the dynamometer are (1) that it is painful, particu- larly if a series of grips is taken, (2) that some aS's suffer from sweat- ing of the hands, especially when excited, and that this causes the instrument to slip in their grasp, (3) that a wrong manner of hold- ing the instrument may reduce the record, e.g., by as much as 10 kg., (4) that, owing to the large number of muscles concerned, a lack of proper coordination in their contraction may lower the record. The painf ulness of the dynamometer can be largely eliminated by proper construction; the Smedley instrument is much better than the Collin elliptical form so commonly used heretofore. Moreover, if an extended investigation is to be undertaken, inurement to the pressure is rapidly developed (Bolton and Miller). For the ascertainment of strength of grip, excessive perspiration can be avoided by simply drjdng the hands with a towel whenever necessary. The proper holding of the instrument is also largely dependent on proper construction, and in this respect, again, the Smedley instrument, with its adjustable grip, is a distinct improvement over other forms. The last objection is not to be seriously considered, first, because hand-grip is one of the most common forms of coordinated move- ment and is well organized early in childhood, and second, because experience shows that most S's can make their maximal record in three attempts at least. - ' For a description, with photographic reproductions of tlaese motor automatisms of effort, consult Binct and Vaschide. ^ If, for any reason, E considers these sources of error not eliminated, it may be necessary to select a number of >S's and coach them in the use of the dynamometer until they can either avoid the errors, or report to E when they occur. J. Claviere (4) asserts that to employ only those S's who are thus trained in the use of the instrument is an indispensable condition for successful dynamometry. TEST 6: STRENGTH OF GRIP 79 In any extended investigation, E should take steps to test the cali- l)ration of the dynamometer occasionally. For this purpose, the instrument is held securely in a vise or other support and a series of weights are hung upon the stirrup while the scale-readings are com- I)ared with the actual weighting. For many purposes it is desirable to combine strength of grip with strength of back and strength of legs by adding the data secured in these three tests. The advantages and disadvantages of using a series of grips in place of a single one are discussed in Test 9. REFERENCES (1) A. Binet and N. Vaschide, Experiences de force musculaire et de fond chez les jeunes gargons, in A. P., 4: 1897 (1898), 15-63. See also pp. 173- 199, 236-252, 295-302. (2) T. Bolton and Eleanora Miller, On the validity of the ergograph as a measurer of work capacity, in Nebraska Univ. Studies, 1904. Pp. 79 + 128. (3) Ada Carman, Pain and strength measurements of 1507 school children in Saginaw, Michigan, in A. J. P., 10: 1899, 392-8. (4) J. Claviere, Le travail intellectuel dans ses rapports avec la force musculaire mesuree au dynamometre, in A. P., 7: 1900 (1901), 206-230. (5) G. E. Dawson, A Study in youthful degeneracy, in Pd. S., 4: 1896, 221- 258. (6) A. MacDonald, Experimental study of school children, reprint of chs. 21 and 25, U. S., 1899. (7) M.-C. Sch uyten, Les variations de la force musculaire et le developpe- mcnt intellectuel des eleves; summarized by A. Binet, in A. P., 9: 1902 (1903), 448-9, from Pajdologisch Jaarboek, Ghent, 1902. (8) M.-C. Schuytcn, Comment doit on mesurer la fatigue des ecoliers, in Ar. P., 4:1904, 113-128. (9) F. Smedley, Rept. dept. child-study and pedagogic investigation, reprint from 46th An. Rept., Brd. Educ. Chicago, 1899-1900. Also in U. S., 1902, i., 1095-1115. (10) F. Smedley, do.. No. 3, 1900-1901. Also in U. S., 1902, i., 1115-1138. TEST 7 Strength of back. — This test, together with the following, has l)een extensively used, in securing an index of the general bodily strength of college students, but has not been applied in most exam- inations of school children. A fairer index of strength may, how- ever, be gained by its use in combination with strength of grip. »U PHYSICAL AND MOTOR CAPACITY Instrument. — Back and leg dynamometer (Fig. 10). Method. — S stands upon the footrest of the instrument, which £^ should then adjust by lengthening or shortening the chain, so that .S'sbody is inclined forward at an angle of about 60 degrees (Fig. 11). S should then take a full breath and give a hard lift, mostly with the back and without bending the knees. Two or three trials may be recorded, and the best record used subsequently in computation. FIG. 10. BACK AND LEI! DYNAMOMP^TER. CAPACITY, 700 KG. Results. — (1) On the use of this and the succeeding test, with quantitative results as obtained in college gymnasiums, etc., consult Hastings (3), Sargent (4, 5), Seaver (6) and other author! ties already cited under anthropometry in general. (2) Binet and Vaschide (1) found the lift (force rennle) of 37 boys aged from 12 to 14 years to average 77 kg., with a maximum of TEST 7: STRENGTH OF BACK 81 121 kg., and minimum of 56 kg. With 40 young men averaging 18 years of age, the same investigators (2) obtained for the average 146.64, for the maximum 187, and for the minimum 101.6 kg. Hastings (p. 71) publishes measurements of 5000 young men (17- 30 years) whose strength of back averages 150.9 kg., P. E. 22.1, with a minimal record of 74.5 and a maximal record of 227.3. (3) Back lift is roughly about 3.2 times the strength of the right hand. y^r '"^J T ■ ! . i 1 1 i .Vi 1 i '• T 'i M •/)■ 1 1 ^- FK;. 11. BACK AND LEfi DYNAMOMETER, AS USED FOR STRENGTH OF BACK- From D. Sargent, Anthropometric Apparatus. Note. — The adjustment of the chain may, with advantage, be based upon /S's height. For this purpose, E may work out an empir- ical table of relations between height and the length of chain neces- sary to give the required position. PHYSICAL AND MOTOR CAPACITY REFERENCES (1) A. Binet and N. Vaschide, Experiences de force musculaire et de fond chez les jeunes gargons, in A. P., 4: 1897 (1898), 15-63. (2) A. Binet and N. Vaschide, La mesure de la force musculaire chez les jeunes gens, ibid., 173-199. (3) W. Hastings. A manual for physical measurements for use in normal schools, public and preparatory schools, etc., Springfield. Mass., 1902. Pp. 112. (4) D. Sargent, Strength tests and strong men at Harvard, in J. Boston Soc. Med. Science., No. 13, 1896-7. (5) D. Sargent, Anthropometric apparatus, etc., Cambridge, Mass., 1887. Pp. 16. (6) J. Seaver, Anthropometry and physical examination. New Haven, IStO. TEST 8 Strength of legs. — This strength test is to be used m conjunction with strength of grip and strength of back. The best records in each of these three tests maj^ be added, to secure an index of general bod- ily strength. Instrument. — Back and leg dynamometer. Method. — S stands upon the foot rest of the instrument with his trunk and head erect and his chest well thro^vn out, but with the knees well bent (Fig. 12). E then adjusts the instrument so that the handle, when grasped by S, rests against his thighs. S should then take a full breath and give a hard lift, mostly with the legs, using the hands to hold the handle in place. Allow two or three trials as before. Results. — Strength of legs is commonly about 26 per cent greater than strength of back. Thus, the 5000 men whose records are embodied in Hastings' table have a mean streng-th of legs of 189.5 kg., P. E., 35.3, with a minimal record of 102.2 kg. and a maximal record of 276.8 kg. TEST 9 Endurance of grip. — The object is to test the capacity of S to exert maximal muscular exertion, not in a single effort, as in Test;^ 6, 7, and 8, but during a period of one minute : the test is thus vir- tually identical with the endurance tests commonly undertaken b}' means of the ergograph. TEST 9: ENDURANCE OF GRIP 83 Since Mosso's studies of muscular fatigue (31), the ergograph has been extensively employed, not only by physiologists, but also by psychologists and by investigators of school children. The form of the apparatus and the conditions of the test have been widely varied, and the numerous factors which affect the test have been exhaust- ively discussed. In general, the purposes for which the ergograph test has been employed may be summarized thus: (1) to study the ii' T \ ii I ■ ■! -i A i ; ^ 1 ' i ' 1 i \^ % "'"';--"■• '- -^ [^^^--^^^ FIG. 12. BACK AND LEG DYNAMOMETER, AS USED FOR STRENGTH OF LEGS. From D. Sargent, Anthropometric Apparatus. physiology of muscular contraction, (2) to detect the presence and to examine the nature and extent of muscular fatigue, (3) like the strength tests, to gain an index of physical capacity or endurance under varying conditions, e.g., as affected by stimulants, narcotics, poisons, exercise, varying diets, etc., (4) to secure an index of right- handedness, (5) to discover whether physical fatigue is general or local, (6) to discover how mental work affects physical capacity, and, in particular, whether mental fatigue is reflected in muscular fatigue 84 PHYSICAL AND MOTOR CAPACITY with such clearness that its existence and degree may be ascertained by the examination of some restricted group of the muscles, and (7), on the assumption that physical capacity does measure directly the condition of mental efficiency, to determine the so-called diurnal 'course of power.' As just intimated, the question of the applicability of the ergo- graph to these varied purposes raises a large number of problems, in particular that of the nature of fatigue. Since mental fatigue is reducible to physiological fatigue, the main point at issue is that of the precise nature of the latter. The relative fatiguability of the structures concerned in physiologica fatigue, i. e. of muscle, nerve fiber, and nerve cells, is a complex and dis- puted question. We know that the fiber, in comparison with other tissues, is extremely resistant to fatigue, either because the catabolic changes are minute, or, more probably, because they are at once compensated by ade- quate anabolism. It has long been thought that the brain and spinal cord are much more susceptible to fatigue than the muscle, and consequently, the non-fatigu- ability of the nerve fiber has led most adherents of the neurone theory to look for the locus of nerve fatigue in the cell-body, in which, as has been shown by Hodge and others, excessive fatigue is accompanied by marked histologic changes. On the other hand, Sherrington denies central fatigue to the cell-bodies, and locates it in the synapse (point where one neurone comes into functional relation with the next in the series), and assumes that this acts, like the motor end-plate (point where the neurone comes into functional relation with the muscle fiber), as a sort of safety' -fuse to prevent overwork and damage to the tissues. Sherrington's experiment indicates that if a single, afferent tract is stimulated through several afferent tracts, the muscular contraction can be continued, when one afferent tract is exhausted, by re- course to the others. Other experiments, particularly those of Mile. Jote/ko, indicate that, compared with the terminal organs, the reflex mechanism of the spinal cord is practically indefatigable. If we extend this conclusion to the higher centers of the brain, we arrive at a peripheral theory of fatigue. Prominent evidence against such a conclusion has long been found in the observation that a muscle exhausted by volitional effort will still respond to electrical stimulation, but this observation is now discredited by the experi- ments of Kraepelin, G. E. Miiller (32), R. Mliller (33), Storey (38), and others, who have shown that the electrical stimulation in this case really sets in play a different set of muscles, and that, if a single muscle, like the abductor indicis, be properly isolated, the recuperation is not present. "In view of these results and others," says Lee (25) "I am inclined to the belief that when we perform continued muscular work, our muscular system TEST 9: ENDURANCE OF GRIP 85 fatigues before our central nervous system. Moreover, the same results make it probable that the brain and the spinal cord are, like the nerve fiber, resistant, and they throw a certain measure of doubt on all supposed proofs of central fatigue" (25, pp. 179-180). The question of mental fatigue, induced by intellectual work, thus becomes perplexing. That it is a reality cannot, of course, be denied. Very likely even this type of fatigue is largely peripheral in origin, but how much is peripheral and how much central cannot as yet be stated. The usual assumption that brain fatigue is local in character seems due in part to the fact that the presence of fatigue is first indicated usually by the characteris- tic feeling of fatigue, but we may conclude that this feeling, like the corre- sponding feeling of effort, is primarily of peripheral origin, the consequence of the depressant action of certain toxic products, such as sarco-lactic acid, carbon dioxid, etc, which affect the muscular system in particular. Since these toxic products, at least when present in considerable amounts, are carried by the blood throughout the body, it follows that decided fatigue is not confined to the tissues in which it arises (as Mosso showed in his experimental transfusion of the blood of a fatigued, into the veins of a fresh dog). Those who argue that fatigue is not general, but local, at least when not excessive, usually base their conviction upon the evidence of certain experimental studies which seem to indicate that well-defined local fatigue may exist without lowering perceptibly the functional capacity of other portions of the organism. It may not be incorrect to assume that the toxic products are most detri- mental in the locality where they are produced, yet may have, in time, a dis- tinct general influence, and, similarly, that the reflex effects of fatigue, whether exciting or depressing, are partly local and partly general: the consumption of cell material, on the other hand, is more distinctly a local phenomenon (Bergstrom, 2). In understanding the nature of mental fatigue and in distinguishing between general and local fatigue, it is helpful to separate the objective fatigue (Ermudung), i. e., actual functional inefficiency, from 'weariness' {Miidigkeit) i. e., the subjective experience of ennui, loss of interest, or dis- inclination to work.^ Thus, it is weariness rather than fatigue which dis- appears when one's occupation is changed: weariness is fluctuating, uncer- tain, and largely dependent upon the general conditions under which work * Again some physiologists would differentiate fatigue and exhaustion: fatigue is, for them, a merely temporary reduction of efficiency, due to the accumulation of waste products, while exhaustion is a more serious condi- tion due to lack of adequate nutrition. For a discussion of the nature of fatigue and of other factors such as practise, ennui, warming-up, spurt, etc., that complicate the determination of the fatigue curve, consult Kemsies (21), Kraepelin (22, 23), Thorndike (40) Ellis and Shipe (14) , Bergstrom (2), Miiller (33), Loeb (27), Aars and Larguier (1), Hough (19), Bettman (3), Rivers and Kraepelin (35), Weygandt (41), Lin,dley (26). An authorita- tive summary of the problem from the point of view of the physiologist will be found in Lee (25) . 86 PHYSICAL AND MOTOR CAPACITY is being done, while fatigue increases more or less steadily and progres- sively during our waking moments. If weariness is often specific, fatigue may be more often general and operative to reduce the available energy for work (Leistutigsfdhigkeit) in any direction. These considerations make it evident that, while the relations between mental fatigue and muscular energy are still obscure, we may hope, in principle, to secure some index of the former by our measurements of the latter. If the ergograph is to be employed for FIG. 13. STOP-WATCH. exact laboratory experimentation, there is no doubt but that the instrument must be of elaborate construction and the technique equally refined. The development of the ergograph itself from the relatively simple instrument of Mosso to such a complicated appa- ratus as that devised by Bergstrom is symptomatic of this gradual refinement of method and progressive analysis of the modifying factors.^ ' Consult also Cattell (11), Franz (15), Binet and Vaschide' (5), Binet and Henri (4), Bolton and Miller (10), Hirschlaff (17), R. Miiller (33). TEST 9: ENDURANCE OP GRIP 87 But while such problems as the merits of spring vs. weight load- ing, the relative capacity of a muscle working by isometric and by isotonic contractions, or the most reliable method of isolating the working muscle may be of paramount importance for laboratory investigation, it does not seem, on this account, absolutely impos- sible, as some writers assert, to secure valuable results for compara- tive purposes from large numbers of subjects by the use of simpler apparatus and less rigorous teclmique, — provided, of course, that FIG. 14. METRONOME, WITH MERCURY CUPS FOR ELECTRIC CONTACT. the conditions of experimentation are kept as constant as possible for different subjects and for the same subjects at different times. ^ For this reason, the test which is here described is suggested as a practical substitute for the more cumbersome and complicated ergo- graph. Apparatus. — Smedley dynamometer (Fig. 9) . Stop-watch (Fig. 13). Metronome (Fig. 14). [If desired, a kymograph, (Fig. 15) ' In general, it may be expected that minor variable errors will, in a sufficiently long series of tests, be distributed according to the law of chance. Possibly, some portion of the dispute concerning the value of ergograms arises from the fact that certain experimenters have worked upon large numbers of subjects, while others have contented themselves with curves obtained from a single individual. 88 PHYSICAL AND MOTOR CAPACITY with drum support (Fig. 16) and other accessories, and a Marey tambour (Fig. 17), or the Mosso ergograph (Fig. 18).] Two methods are described: one calls for a single, continuous contraction, the other for a series of separate contractions. A, WITH CONTINUOUS CONTRACTION Method. — Set the metronome at QO,i.e., so that it beats once per second. Adjust the dynamometer to S's hand, as in Test 6. Move the friction or recording pointer of the instrument well over to the KYMOGRAPH, IN HORIZONTAL POSITION. A clock-vvoi-k mechanism, with regulating fans, in the base, rotates the drum at constant speed and at any desired rate from one revolution in ten seconds to one revolution in ten minutes. Used for making graphic records upon smoked paper. right, off the face of the scale. Instruct S that he is, at the signal 'now,' to grip as forcibly as possible, to maintain this grip with his utmost effort until told to stop at the end of one minute, and to keep his eyes fixed upon the pointer, so as to hold it as high as possilile. (This instruction is designed to act as an incentive to maximal exer- TEST 9: ENDURANCE OF GRIP 89 tion.) Let the instrument be held in the vertical plane with the right-hand edge resting on the table before which S is seated. E starts the metronome, and, when he has caught the rhythm, starts the stop-watch, at the same instant saying 'now' for S to begin. E immediately takes the first reading, and thereafter glances at the scale at every fourth beat of the metronome. In the intervals, he records, of course, the reading of the pointer just obtained, estimating to the nearest half-kilogram. If the first read- ing is secured promptly, E will have 16 readings at the end of one minute. FIG. 16. SUPPORT FOR KYMUGKAFH DRUM WHILE BLACKENING PAPER. Variations of Method. — With the aid of the kymograph described in Test 10, E may secure a graphic record of S's work, either by the use of the pneumatic tambour (the Smedley instrument is fitted for pneumatic transmission), or by the use of a simple system of levers to magnify the movement of the handle.^ The quantitative evaluation of the resulting curve may then be obtained by a series of measurements of the ordinates taken at regular dis- tances and checked by the record obtained as prescribed, or by ' For a cut showing the method of securing a dynamograph record, see MacDonald (29, p. 1184). 90 PHYSICAL AND MOTOR CAPACITY ineasurenient with a planinieter of the area oncloscd by the cui've and its base hue. Treatment of Results.— (1) For some purposes, the resuhs may be treated by simpl}^ averaging the 16 readings, but (2) it will usually be more instructive also to compare the initial with the final stages, in order to secure an index of endurance, or conversely, of fatigue. For this purpose, average the first four readings and the last four readings; subtract the latter from the former, and divide the remainder by the average of the first four readings. This may be expressed by the formula .t = " when .r = the desired index in terms of per cent, n = the M of the first, and n the M of t\\v last readings. Or, (3) more simply, one may indicate endur- FIG. 17. MAREY TAMBOUR. For securing tracings l)y pneumatic transmission. The ruWher mem- brane is not shown. ance by the relation of the average to the maximal grip. (4) Fol- lowuig Binet and \'aschide (7), the records of strong, average, and weak N's (judged by their maximal grip) may be collated and treated in three groups, in order to trace the presence of the three types of endurance (see below) . R. with separate contractions Method. — Adjust the metronome, dynamometer handle and pointer as in the first method. Inform ^' that, as the word is given, he is to make a series of 16 grips, each as forcibly as possible, and that these grips will be signalled at 4-sec. intervals. E then signals 'now' on every fourth beat of the metronome, and takes the readings as previously described. TEST 9: endurancp: of grip 01 Variations of Method. — Substitute the kymograph tracing as suggested above. If this is done, there is no reason why the rate of effort may not be increased, so as to secure 60 or 120 contractions per minute, with a correspondingly more rapid onset of fatigue. If it is desired to compare results obtained with the dynamometer with those obtained by the common form of ergographic experi- ment, it is suggestetl that E repeat the experiments made upon Chi- cago school children. For this purpose, substitute the Mosso ergo- graph for the dynamometer; use the kymograph for securing the graphic record, and the record furnished by the endless tape, multi- FIG. 18. MOSSO ERGOGRAPH, MODIFIED BY LOMBARD. plied by the weight, for the quantitative result. Adjust the weight at 7 per cent of aS's weight, and time the contractions to accord with the beats of a metronome set at 30, so as to secure 45 lifts in 90 sec. Treatment of Results. — This may follow the lines already pre- scribed. Results. — (1) The measurement of endurance by the use of the dynamometer has been tried by Binet and Vaschide, though under conditions somewhat dissimilar to those we have suggested, upon a group of boys aged 10-13 years (6) and upon a group of young men aged about 18 (7). When five (or ten) grips with each hand, alter- 92 PHYSICAL AND MOTOR CAPACITY nately, were required, these authors made out four types of endurance curve, viz : (a) a sudden drop, then fairly constant, (6) an approxi- mately stationary or constant type, which is quite common, (c) a continuous, but gradual drop, and {d) a more or less definite rise. The last is rather infrequent (it was not found, e.g., by Claviere, in tests with 15 successive grips), but is sometimes given by vigor- ous individuals, though the third type is more common for such sub- TABLE 21 Types of Endurance in Dynamometer Trials: kilograms {Binet and Vaschide) TYPE a TYPE 6 TYPE C lYPE d 1st Grip 2d Grip 3d Grip 4th Grip 5th Grip 23.00 18.45 19.00 18.60 18.20 18.70 18.60 19.20 19.40 17.80 24.12 22.50 21.17 21.33 19.80 17.33 17.70 18.67 18.67 20.67 jects. Practically 90 per cent of endurance records can, in the judgment of these writers, be classed in one of these four categories. Table 21 gives average records of five grips, made with the right hand, by groups representing these four types of endurance curve. (2) If we accept this hypothesis of 'types,' it is clear that the dynamometer yields a more reliable indication of the comparative TABLE 22 Opposed Types of Endurance, 10 Readings {Binet and Vaschide) NUMBER OF GRIP 1 2 3 4 5 6 7 8 9 10 Subject B 36 36 34 37 34 42 30 43 29 45 28 42 25 42 26 45 26 46 ^9 Subject R 45 muscular capacity when it is employed to test endurance in this way than when merely a single grip is taken, as in Test 6. To take an extreme, though actual case cited by Binet and Vaschide, it will be seen (Table 22) that, if two subjects belong to opposing types, their TEST 9: ENDURANCE OF GRIP 93 actual capacities may be completely unsuspected when but a single test is taken. If we turn to the use of the ergograph, we find the following important, though too often conflicting results. (3) Ergograph curves are affected by practise improvement, which, according to Bolton and Miller (10), results (a) from 'inure- ment,' i.e., a fairly rapid "process of hardening and toughening of the skin where it comes in contact with the apparatus and of habitu- ating the muscles to the strains which the unusual effort imposes," (6) from improved coordination in the movements concerned, par- ticularly seen in the disappearance of useless movements, (c) from improvement in the rhythmic execution of the contraction, and {d) from a slow increase in endurance proper, primarily in the nerve centers. This increase of practise, as Oseretzkowsky and Krae- pelin (34) have shown, affects both the height and the number of lifts, and gradually becomes less and less noticeable as maximal practise is attained. (4) The amount of work that can be done by the muscle is increased if the rate of lifting is increased from 30 to 60 or 120 lifts per minute (Oseretzkowsky and Kraepelin). (5) The work done is conditioned by the load lifted or tension of the spring. One can not, wdthout caution, compare ergograms made with different loads. (6) The total amount of physical work done, as measured by weight X distance, can not be regarded as a necessarily correct index of the physiological capacity of the muscle; thus, 100 lifts of 25 mm. each may not be assumed to be physiologically equal to 50 lifts of 50 mm. each (Binet and Henri, 4; Franz, 15). (7) The weight ergograph is not adapted to the measurement of muscular capacity (Binet and Henri, 4), hence "the fatigue curves obtained by Mosso and later investigators with weights do not represent the true state of the neuro-muscular system" (Franz, 15). (8) "The isotonic use of a weight or a spring for measuring mus- cular force is not justified, because two variable factors, extent and force, are introduced," so that an isometric spring (such as the dyna- mometer) should be used for all comparative experiments (Franz). (9) With improperly contrived apparatus or inexperienced sub- jects, the ergographic tracing is very liable to be affected by the play 94 PHYSICAL AND MOTOR CAPACITY of muscles other than those under examination (Binet and Henri, 4. Bergstrom, 2) . Miiller (33) considers the failure properly to isolate the muscle a fundamental defect of the ergograph. (10) In addition to these specific criticisms, more general conclu- sions of a negative character may be quoted. Thus, Bolton (8) asserts that the ergograph is not adapted for measuring the degree of fatigue in school children; Bolton and Miller (10) conclude that ergograph records "have slight validity until inurement has become thorough and coordination complete, that the ergograph is quite unadapted to the obtaining of exact statistics upon a large number of individuals, and that records taken upon unpractised subjects, both before and after operations whose influences are thought to affect muscular power, are without the slightest claim to trust- worthiness." Similar conclusions are reached by Ellis and Shipe (14), after a retrial of the methods of Keller and of Smedley also by Thorndike (40) and, with some qualifications, by Berg- strom (2). (11) The effect of physical work upon ergographic curves seems to vary with the physical condition of the individual and with the nature and duration of the exercise. Thus, Bolton (8, 9) found his ergograms decreased by a 2-hour walk, but Oseretzkowsky and Kraepelin found that a 1-hour walk caused at first a transient improvement, then a reduction, the first of which they attribute to the increased excitement of central motor tracts, and the second to the dampening influence of general muscular fatigue. Smedley(37) tested Chicago children before and after a 40-minute class exercise in the gymnasium with the result that the stronger pupils were little affected, whereas weak and nervous pupils were decidedly exhausted. From this study he concluded that the classes in physi- cal culture should be graded on a physical, instead of on an intellec- tual basis. (12) Extensive study of the effect of mental work on physical endurance has so far yielded but discordant results. Some of this work, e.g., that of Keller (20), may be thrown out of court at once as careless in plan and execution and merely illustrative of blind infat- uation for the ergograph. Typical conclusions of other investiga- tors are as follows : Larguier (24) reports that two hours of mathe- matics, and Bolton that two hours of adding, definitely increase the TEST 9: ENDURANCE OF GRIP 95 ergograph record; Claviere (13), on the other hand, reports that two hours of intense mental work produces a defmite and propor- tionate dimmution of muscular force, whereas intellectual work of medium intensity does not produce any appreciable weakening of endurance ; he further confesses his inability to determine the rela- tive fatigue-effect of various school studies. The careful ergo- graphic tests of Oseretzkowsky and Kraepelin show that work- capacity is increased after one hour of simple addition or learning of 12-place numerals, but that it is lessened if the mental work is rendered more difficult, as by adding under distraction. In an extensively quoted study, Kemsies (21) reports the results of a long series of tests upon a selected group of average, industrious boys who had been trained to the use of the instrument, from which he concludes (a) that the ergograph is a reliable indicator of true fatigue (lowered fund of energy as distinct from weariness) , (6) that subjective feelings of bodily or mental condition may not accord with real capacity, (c) that some of the pupils in the Berlin schools show, at least for the time being, signs of overwork, (d) that special attention should be paid to pupils who fatigue easily, (e) that one can determine for each study its special fatigue-value or 'ergo- graphic-index,' more particularly, that the several studies range themselves, in order from highest to lowest fatigue-index, as fol- lows: gymnastics, mathematics, foreign languages, religion, Ger- man, science and geography, history, singing and drawing. In the attempt to explain these divergences, Binet and Henri (4) suggest that we must always distinguish between mental work conducted without emotion and that conducted with emotion; they conclude that the former, if prolonged, may be expected to lessen endurance, the latter to produce a transient increase fol- lowed by a decrease. Kraepelin (23), somewhat similarly, con- cludes that, while hard mental work certainly reduces muscular energy, deviating results may appear in ergograms on account of the condition of excitement (Anregung) that normally accompanies mental work, and that may be expected to affect, either positively or negatively, the tracing which follows such work. Kraepelin fur- ther calls attention, as do Ellis and Shipe, Bergstrom, Franz, and others, to the very large normal variation in the curves of any indi- vidual, due to the operation of numerous constant and variable fac- 96 PHYSICAL AND MOTOR CAPACITY tors, often little understood. Many results are valueless (e. g., in his opinion, those of Kemsies) because of the failure properly to eliminate or evaluate these factors. (13) The investigations of Christopher (12) and Smedley (37) at Chicago indicate a thorough-going correlation between endurance and class-standing, according to the method of percentile grading, the method of distribution of 12-year old pupils, and the method of comparison of the endurance of children at and above grade with that of children below grade at each age. Again, boys in the school for incorrigible and truant children were found to exhibit, at every age, less endurance (62 per cent to 82 per cent) than normal boys of the same age. ( 14) The endurance of boys is greater than that of girls at all ages, and the difference becomes very striking during adolescence (37). (15) The development of endurance and that of vital capacity bear a decided resemblance to one another, whether pupils are examined singly or collectively (37) . (16) The diurnal 'course of power' according to the Chicago experiments may be expressed as follows: "(a) The extremes of endurance and fatigue in school are greater in the morning than in the afternoon; (6) a higher grade of power is found in the morning session in children attending two sessions daily; (c) while endurance is not as great, it is better sustained in the afternoon." Compila- tions of the ergograms of 1127 pupils place the maximum at 9 a. m. and the minimum at 12 noon. Kemsies considered the first two morning hours the best. Experiments upon adults by Lombard (28), Harley (16), Storey (39), and Marsh (30) exhibit considerable lack of agreement with one another or with the Chicago results, though Marsh summarizes them by the statement that the curve of strength efhciency seems well established for the following course : "& beginning minimum in early morning, a fairly rapid rise till 1 1, a level or slight decline till 1 p.m. ( ± 1 hour), an increase to the max- imum at 5 p.m. ( ± 1 hour), thence a fall till bed-time." (17) Kemsies concludes that Monday and Tuesday, or the first two days after any rest pause, are the best days for general efficiency, and he further concludes that vacations exert a powerful effect upon efficiency, but, since this effect can not be traced for longer than four weeks, school terms should be broken up by more frequent vacations of shorter duration. TEST 9: ENDURANCE OF GRIP 97 (18) If ergographic contractions are continued to the point of exhaustion, we have both the sum total of the height of the lifts and their number for indexes of the neuro-muscular condition. Hoch and Kraepelin (18) are of the opinion that, in this case, the height of contraction is conditioned by the state of the muscles, but the num- ber of contractions by the state of the central nervous system; the two factors should, therefore, be reported separately for their diag- nostic value. On the other hand, Lombard (28) concludes that, at least when the contraction is not faster than once per second, the amount of fatigue experienced by the central nervous system does not correspond to the number of lifts, but rather to the strength of the motor impulses discharged, so that the sum total of the height of lifts is the more accurate index of the state of the central nervous mechanism. (19) According to Lombard, endurance is increased by exercise, rest (especially sleep), food, increased atmospheric pressure, and by small doses of alcohol, but lessened by general and local fatigue, hunger, lessened atmospheric pressure, high temperature, especially with high humidity, and by tobacco. Oseretzkowsky and Krae- pelin find that coffee increases the height of lifts, and that alcohol, in quantities from 15 to 20 g., causes at first a considerable increase, especially in the number of lifts, but that this soon disappears. On the other hand. Rivers and Webber (36) have discovered that small doses of alcohol (5-20 cc.) fail to produce any appreciable modifica- tion of the ergographic record if proper precautions are taken to keep the subject in ignorance as to when alcohol is administered. The results of previous workers are therefore presumably due to the influence of other factors, particularly interest and sensory stimula- tion, and no future work on the effects of small doses of alcohol can be acceptable unless these factors are controlled. Harley (16) concludes that ''moderate smoking, although it may have a slight influence in diminishing the power of doing voluntary muscular work, neither stops the morning rise nor, when done early in the evening, hinders the evening fall." Notes. — In either form of test, E must practise his work until it becomes automatic. He must take care to keep his eyes directly over the pointer to prevent the error of parallax in reading. For this reason it will be found most convenient for S and E to sit on opposite sides of the table. 98 PHYSICAL AND MOTOR CAPACITY The stop-watch is used both to test the metronome and to check up the duration of the experiment, but ought to be virtually unnec- essary after E has practised the experiment sufficiently. In the first form of test, the pointer is apt to fall by a series of sud- den drops, or even at times to rise as *S makes a momentary recovery. E must take the reading precisely on the beat of the metronome, regardless of the position of the pointer just before or just after the beat. To hasten the acquisition of skill in conducting the second form of test, E will find it helpful to accent the spoken 'now' and to get the swing of the four-beat rhythm by mentally counting the other beats, thus: "Now, two, three, four: now, two, three, four." As soon as the utterance becomes automatic, E can give his whole attention to the readings and the recording of them, and an accurate record can be obtained from very quick and brief excursions of the pointer. Incidentally, some »S's may be found who are inclined to hold the pointer up too long : they must be cautioned against this, otherwise fatigue will set in very rapidly, REFERENCES (1) K. Aars and J. Larguier, L'effort musculaire et la fatigue des centres nerVeux, in A. P., 7: 1900 (1901), 187-205. (2) J. A. Bergstrom, A new type of ergograph, with a discussion of ergo- graphic experimentation, in A. J. P., 14: 1903, 510-540. (3) S. Bettmann, Ueber die Beeinflussung einfacher psychischer Vor- gange durch korperliche u. geistige Arbeit, in P. A., 1 : 1895, 152-208. (4) A. Binet and V. Henri, La fatigue intellectuelle, Paris, 1898. Pp. 336. (5) A. Binet and N. Vaschide, Examen critique de I'ergographe de Mosso, in A. P., 1897 (1898), 253-266. Also, Un nouvel ergographe, dit ergographe a ressort, ibid., 303-315. (6) A. Binet and N. Vaschide, Experiences de force musculaire et de fond chez les jeunes gargons, ibid., 15-63. (7) A. Binet and N. Vaschide, La mesure de la force musculaire chez les jeunes gens, ibid., 173-199. See also ibid., 236-244, 245-252, and 295-302. (8) T. Bolton, The reliability of certain methods for measuring the degree of fatigue in school children, in P. R., 7: 1900, 136-7. (9) T. Bolton, Ueber die Beziehungen zwischen Ermlidung, Raumsinn der Haut und Muskelleistung, in P. A., 4: 1902, 175-234. (10) T. Bolton and Eleanora Miller, On the validity of the ergograph as a measurer of work capacity, in Nebraska Univ. Studies, 1904, 79 + 128. (11) J. Cattell, An ergometer, in Science, n.s., 5: 1897, 909-910. TEST 9: ENDURANCE OF GRIP 99 (12) W. Christopher, Rept. on child-study, reprint from An. Rept. Brd. Educ. Chicago, 1898-99. (13) J. Claviere, Le travail intellectuel dans ses rapports avec la force musculaire mesuree au dynamometre, in A. P., 7: 1900 (1901), 206-230. (14) A. C. Ellis and Maud Shipe, A study of the accuracy of the present methods of testing fatigue, in A. J. P., 14: 1903, 496-509. (15) S. I. Franz, On the methods of estimating the force of voluntary muscular contraction and on fatigue, in Amer. J. Physiol., 4: 1900, 348-372. (16) V. Harley, The value of sugar and the effect of smoking on muscular work, in J. of Physiol., 16: 1894, 97-122. (17) L. Hirschlaff, Zur Methode und Kritik derErgographenmessungen, in Z.P.P., 3: 1901, 184-198. (18) A. Hoch and E. Kraepelin, Ueber die Wirkung der Theebestand- theile auf korperliche und geistige Arbeit, in P. A., 1: 1896, 378-488. (19) T. Hough, Ergographic studies in neuro-muscular fatigue, in Amer. J. Physiol., 5: 1901, 240-266. (20) R. Keller, Padagogisch-psychometrische Studien, in Biol. Central- blatt, 14: 1894, 24-32, 38-53, 328-336. (21) F. Kemsies, Zur Frage der Ueberbiirdung unserer Schuljugend, in Deutsche med. Wochenschrift, July 2, 1896, 433. See also his Arbeits- hygiene der Schule auf Grund von Ermiidungsmessungen, Berlin, 1898, and in S. Z., 2: 1899, Heft i., Pp. 64. (22) E. Kraepelin, Zur Ueberburdungsfrage, Jena, 1897. Pp. 49. (23) E. Kraepelin, Ueber Ermiidungsmessungen, in A. G. P., 1: 1903, 9-30. (24) J. Larguier, Essai de comparison sur les differentes methodes pro- posees pour la mesure de la fatigue intellectuelle, in A. P., 5: 1898 (1899) 190-201. (25) F. S. Lee, Fatigue, in the Harvey Lectures, Phil., 1906, 169-194. Also in J. Amer. Med. Ass., 46: 190G 1491, and in Studies in Physiology, Columbia Univ., 1902-7. (26) E. H. Lindley, Ueber Arbeit u ; i Ruhe, in P. A., 3:1900, 482-534. (27) J. Loeb, Muskelthatigkeit als Mass psychischer Thatigkeit, in Arch, f. d. ges. Physiol., 39: 1886, 592-7. (28) W. F. Lombard, Some of the influences which affect the power of voluntary muscular contractions, in J. of Physiol., 13: 1892, 1-58. (29) A. MacDonald, Experimental study of school childern, etc., reprint of chs. 21 and 25 of U. S., 1899. (30) H. D. Marsh, The diurnal course of efficiency, Columbia Univ. diss., N. Y., 1906. Pp. 99. (31) A. Mosso, Fatigue, Eng. tr., N. Y., 1904. Pp. 334. (32) G. E. MuUer, Review of A. Waller's ",The Sense of Effort," in Z. P., 4: 1893, 122-138. (33) R. Miilier, Ueber Mosso's Ergographen, etc., in Ph. S., 17: 1901, 1-29. (34) A. Oseretzkowsky and E. Kraepelin, Ueber die Beeinflussung der 100 PHYSICAL AND MOTOR CAPACITY Muskelleistung durch verschiedene Arbeitsbedingungen, in P. A., 3: 1901, 587-690. (35) W. Rivers andE. Kraepelin, Ueber Ermiidungu. Erholung, in P. A., 1 : 1896, 627-678. (36) W. Rivers, The influence of small doses of alcohol on muscular acti- vity, in P. B., 5: 1908, 49. See also Rivers and H. Webber, The influence of small doses of alcohol on the capacity for muscular work, in B. J. P., 2: 1908, 261-280. (37) F. Smedley, Rept. dept. child-study and pedagogic investigation, reprint from 46th An. Rept. Brd. Educ, Chicago, 1899-1900. Also in U. S., 1902, i., 1095-1115. (38) T. Storey, The influence of fatigue upon the speed of voluntary con- traction of human muscle, in Amer. J. Physiol., 8; 1903, 355. (39) T. Storey, (a) Some daily variations in height, weight, and strength, in Amer. Phys. Educ. Rev., 6: 1901. (b) Daily variation in the power of voluntary muscular contraction, {6i(i., 7: 1902. (c) Studies in voluntary muscular contraction, Stanford Univ. Press, 1904. (40) E. L. Thorndike, Mental fatigue, in P. R., 7: 1900, 466-482, 547-579. (41) W. Weygandt, Ueber den Einfluss des Arbeitswechsels auf fortlau- fende geistige Arbeit, in P. A., 2: 1897, 118-202. TEST 10 Quickness or rate of movement : Tapping. — This has probably been more frequently applied than any other 'motor test,' and has been thought to afford a better index of motor capacity than any other single test. Recent work with tapping, however, while not discouraging the belief that the test has value, has shown that we cannot regard speed of voluntary movement as an unequivocal and comprehensive 'index of voluntary motor ability,' because a high gross rate does not necessarily go hand in hand with high speed in other phases of motor response, and because, moreover, we do not know precisely what may be the ultimate neural or psychophysical factors that condition the rate. Aside from its use in the attempt to secure this 'index of volun- tary motor ability,' the tapping test has been employed to secure an index of righthandedness (for which purpose it may be advanta- geously combined with Tests 6, 9, 11, and 12), and to secure an index of fatigue (likewise preferably in conjunction with other tests of physical capacity) . These several indexes have been studied in various comparative investigations, more especially in estimating TEST 10: TAPPING 101 I sex and age factors in motor development and the relation of physi- cal to mental ability at large. The method has ranged from the very simple making of dots or vertical marks with pencil and paper (Binet and Vaschide) to the execution of difficult trilling movements upon telegraph keys. The apparatus here prescribed is somewhat elaborate, but experience has shown that the tapping test cannot be conducted without care- ful control of experimental conditions, and the use of a reliable recording device, such as the graphic method suppHes. Tests like rapid counting aloud or the rapid reading of digits or the reaction-time test, are not psychologically comparable to the tapping test. Again, the form of test used at Columbia University and elsewhere to measure rate of movement (making a dot as rap- idly as possible in each of 100 one-cm. squares) is not equivalent to the tapping required in most quickness tests, since a certain FIG. 19. TAPPING-BOARD. amount of precision is demanded of each movement, and that test therefore stands midway between Tests 10 and 11, as here pre- scribed. Materials. — Tapping board, 55 x 10 cm., with brass plates 10 cm. square on either end (Fig. 19). Tapping stylus, with flexible connecting wire attached. Kymograph (Fig. 15) with accessories, — paper, smoking device, shellac solution. Double time-marker (Fig. 20). Seconds' pendulum (Fig. 21) or other noiseless instrument arranged to give electric contacts once per sec. Support with level- ling screw and right-angle piece to hold time-marker. Table clamps for tapping board. Large sheet of gray or white card- board. Suitable supports and clamps for holding cardboard. Two short-circuiting keys (Fig. 22) , or simple knife switches. Stop- watch. Four dry or Leclanche cells. Flexible covered Avire with connector tips or ordinary No. 18 annunciator wire. [A swivel 102 PHYSICAL AND MO TOR CAPACITY chair adjustable in height and an ammeter or battery-tester are also convenient, though not absolutely essential.] Preliminaries. — (1) Clamp or screw the tapping board securely to the side of a table in such a manner that S may have free access to either end of the board for using either right or left hand. Arrange *S's chair so that he sits sidewise to the table with his forearm rest- ing comfortably along the tapping board and his hand directly over the metallic plate. (2) Place the kymograph in a horizontal position, screened from *S's view by the sheet of cardboard. ^ Adjust the fans or gear- p'lG. 20. TRIPLE TIME-MARKER. The double and the single time-marker are of simihir construction. wheels so that the drum makes (for a 30-sec. test) one revolution in about 40 sec. (3) Remove the drum and cover with the prepared paper by simply moistening the gummed end, taking care to draw it evenly and tightly around the drum. Blacken the paper by revolving it slowly in a smoky flame.- Replace carefully in the kymograph. * The screen is to avoid the distraction of S's attention by the operation of the apparatus. If separate tables are used for tapping board and kymo- graph, this may not be necessary, but it is commonly more convenient to assemble all the apparatus on a single table. 2 An oil stove from which the top is removed is excellent for this purpose, as the flame is very sooty and not so hot as the gas flame often employed. A simple support (Fig. 16) is used to hold thedrum. both for the smoking and for the subsequent removal of the paper. For this and other details in the TEST 10: TAPPING 103 (4) Adjust the time-marker on the support so that the pointers bear upon the drum with just sufficient pressure to make a satisfac- tory tracing. The pointer must move in a plane parallel to the plane of a tangent drawn through the point of contact. The manipulation of the apparatus may be facilitated by fastening upon the table, in front of, and parallel to the surface of the drum, a straight bit of wood somewhat longer than the drum. Let the foot of the tripod which contains the levelling screw stand away from the drum, and the other two feet bear against the wooden strip. A half turn of the levelling screw will then free the pointers from the drum, and the entire support with the time- marker may be slid along to a new position, when another half turn of the screw will quickly adjust the pointers for the next record. (5) Wire one signal-magnet in series with the tapping board stylus, short-circuiting key, and two cells of the battery. The magnet will then be set in motion by the tapping when the key is closed. (6) Wire the second magnet in series with the pendulum, second short-circuiting key, and remaining two cells of the' battery. This magnet will then be set in motion by the pendulum when its con- trolling key is closed, and will thus beat off the time-line. Method. — (1) Seat S for the use of his right hand. Instruct him to tap as rapidly as possible from the signal 'now' to the signal 'stop,' which will be given about one-half minute later. Tell him to pay attention only to his tapping. He may be allowed to exer- cise some latitude with regard to the type of movement used (short or wide excursion), unless he is inclined to adopt a very heavy whole- arm pounding movement. The most favorable movement for most ;S's is that obtained by resting the elbow on the tapping board and using both the wrist and elbow joints. (2) Start the seconds' pendulum and close the time-line circuit. use of the kymograph, consult Titchener, Experimental Psychology, vol. I, Part II, pp. 172-180. If an extended series of tests is to be made, cover the drum permanently with the regular kymograph paper, and, for the records, superpose two nar- rower strips, say 75 mmi wide. These strips are wide enough to record the right and left-hand efficiency of one *S : they can then be removed promptly for fixing, and thus the danger of injury is lessened, the ease of handling in- creased, and the blackening of the metal drum is less likely to be a source of annoyance. 104 PHYSICAL AND MOTOR CAPACITY (3) Start the kymograph, and at the same time give S the signal 'now.' (4) When S is fairly started, throw in the record magnet by closing its key, and at the same instant start the stop-watch. (5) At the expiration of 30 sec, break the record circuit, signal 'stop' to S, stop the kymograph and the watch, and open the time- circuit. [E must practise the whole series of operations until they FIG. 21. seconds' pendulum. run smoothly and automatically, especially the simultaneous opera- tion of watch and record key.] (6) Now adjust the pointers for a new record. Let S sit facing in the other direction, and test the left hand by the use of the plate at the other end of the tapping board, following the directions given for the test of the right hand. TEST 10: TAPPING 105 Variations of Method. — (1) The duration of the test may be lengthened to 45 sec. or longer, or shortened to 10 or 20 sec. It is, however, desirable for the sake of comparison that a standard dura- tion be employed. Thirty sec. is adequate for all ordinary pur- poses. (2) It is recommended that, whenever time permits, more than one trial be made for each hand. To follow the procedure sug- gested by Wells, E should first make five trials with the right, then five trials with the left hand. Each trial lasts 30 sec, and is fol- lowed by a rest-pause of 2.5 min., during which S should refrain from all muscular effort. The five trials of 30 sec. each constitute one 'record,' and the two records of right and left hand constitute one 'experiment.' FIG. 22. SHORT-CIRCUITING KEY (dU BOIS REYMOND). (3) To make the test comparable to the form employed by some investigators, an ordinary 'sending' telegraph key may be substi- tuted for the tapping board and stylus. But the key has the disad- vantage of imposing 'a certain restriction upon the type of move- ment, and will be found in practise to reduce the record of many *S's. (4) By using the key, E may compare S's rate of tapping with different fingers. For this purpose, it is well to fasten clown the forearm with a strap at the wrist, so as to allow movement with the fingers only. (5) Again, by using the key, a trilling movement, executed by alternate movements of the index and middle fingers, may be sub- stituted for the regulation tapping movement. Without practise, this movement is quite difficult for some aS'.s, whereas for others. 106 PHYSICAL AND MOTOR CAPACITY notably for those who have practised trilling exercises on the piano, it is comparatively easy.^ For this reason, this form of experiment is not advised, save for some exceptional purpose, e.g., testing the effect of practise upon the acquisition of a new bit of manual dex- terity. Other modifications suggest themselves, such as trilling with the 4th and 5th fingers — an exercise likely to be unfamiliar even to *S's who have 'taken lessons.' Treatment of Results. — (1) When the record has been made, use any pointed article to mark it for future identification (*S's name or number, date, hand used, etc.) ; then remove carefully for preser- vation. A simple and satisfactory method is to pour a very thin solution of shellac and wood alcohol, or of powdered resin (not over 10 per cent) in alcohol, into a saucer or shallow dish, and to pass the strip through this, smoked side up. Hang the record up to dry, and pour the solution back into a wide-mouthed bottle, where it should be kept tightly stoppered. The record will dry in a few minutes and can then be trimmed and handled with impunity. (2) The result of the test is commonly expressed simply by the total number of taps executed, but it is quite as important, if not more important, to consider changes of speed during each trial. The requisite data must be secured by the rather laborious process of counting the strokes made by the recording-magnet upon the blackened paper, and tabulating them by 5-sec. intervals, as illus- trated below. The 'total efficiency' of a 'record' (5 trials with the same hand) is the average of the sum of the taps per trial. This serves as the gross index of speed : the rates for the 6 intervals within each trial afford an opportunity for studying variations in perform- ance. The use of an electric counter, such as some investigators have employed, would eliminate this work, but the counter gives no indication of changes in speed during the trial. Moreover, the electric counter is not reliable: even with 10 or 12 cells of battery, it will miss a quick tap which the graphic method will record. It follows that all results based on the use of the coun- ter are to be looked upon with suspicion, so that the conclusions of Bagley, 1 The effect of piano practise, as the investigations of Binet and Courtier (2) and of Raif (14) show, is to improve coordination of movement, i,_ e., its regularity, smoothness, etc., but not to increase the natural capacity for speed or rate of movemeot. TEST 10: TAPPING 107 Bolton, Marsh, Kelly, Smedley, and possibly those of Bryan, Davis, Gilbert, and Dresslar should be accepted with reservation. (3) To secure an index of fatigue, E may compare the record of the first 5 (or 10) sec. with that of the last 5 (or 10) sec. by the use of the formula for determining the relative loss of efficiency given in Test 9. Wells has published extensive conclusions concerning fatigue in tapping that are based upon a differently computed index. The average number of taps executed in the 2d, 3d, 4th, 5th and 6th 5-sec. intervals are divided by the number of taps executed in the 1st 5-sec. interval. This index is somewhat misleading, in so far as a high index indicates a low degree of fatigue. If it is deemed worth while to relate the last five to the first of the six intervals to compute fatigue, it would be better, in the author's opinion, to subtract the average in question from the initial speed and divide the loss by the efficiency of the first interval. TABLE 23 Sample Record of a Tapping Test (Wells) NUMBER OP INTERVAL IST 2d 3d 4th 5th 6th total 1st trial 2d trial 3d trial 41 37 37 39 39 39 35 36 37 37 38 34 34 35 1 34 37 35 32 34 34 213 217 222 4th trial 40 41 36 37 36 36 35 36 223 5th trial 227 40.6 38.2 36.6 35.8 35.0 34.2 220.4 (4) To secure an index of righthandedness, E may compute the percentage of the left-hand to the right-hand efficiency. The fatigue-index of the right hand may also be compared with that of the left hand in a similar manner. Typical Results. — Table 23 shows a sample record of the work of a normal adult with the right hand, when near the Hmit of prac- tise. The tabulation is in accordance with that recommended when five 30-sec. trials are made. General CoNCLUSiONS.^^though the tapping test is one of the most objective that can be apphed, and although it has been tried by a large number of investigators (see the references at the end of 108 PHYSICAL AND MOTOR CAPACITY the test), the results have not been always accordant, and, with the exception of the recent work of Wells, have not been so treated as to afford real insight into the factors that underlie their appearance. The lack of accordance is to be attributed in large part to differ- ences in method of procedure. Differences in apparatus, too, have been sufficient to account for some discrepancy, as has already been pointed out. As regards method, the duration of the test, to instance a single point, has varied from 5 sec. (Binet and Vaschide, Bryan, Kirkpatrick) to 2 min. (Thompson), with intermediate dura- tions, such as 10 sec. (Bagiey), 30 sec. (Smedley), 45 sec. (Gilbert), 60 sec. (Kelly), or the test has been conducted in 5 series of 5 sec. each (Bolton).^ In so far as these divergences of method may be neglected, we may note the chief conclusions of interest concerning the tapping test, as follows. (1) In general, the maximal rate of voluntary movement varies with the individual, with sex, with maturity, with the side of the body used, with practise, with the number of trials (duration of experiment), with fatigue, with mental excitement, with the time of day, but not, within wide limits, with the amplitude of the move- ment. ^ (2) Constant mdividual differences in rate of tapping can be demonstrated without much difficulty, but we cannot at present explain them, save to say that they are conditioned in a general way by fundamental neural factors, or by these plus differences in ability to coordinate voluntary movements.^J Thus, in 10 adults tested by Wells (10 trials for each hand), the average total efficiency (taps in 30 sec.) was approximately 194, but the fastest 8 averaged 225, ' The situation here, as in most tests, shows clearly how desirable it would be to establish some standard form of test and to use it alone for all compar- ative purposes. ^ For a fuller discussion of these conditions, consult Dresslar, Bryan, and Wells. ^ To quote from Wells (19, p. 444): "What is the precise physiological significance of the maximum rate is by no means well made out. ... It seems to be generally conceded that it is limited by the refractory phase of the synapses in the motor pathways, but that does not make the tapping test a measure of the period of this refractory phase; at least, not in the earlier stages of practise. ... In the beginning, as we ordinarily have to apply the test, the factors in speed are probably those of coordination mainly, and cannot be expected to afford information about the condition of the motor pathways as given in the refractory phase." TEST 10: TAPPING 109 and the slowest 153. Since the m.v. is small (here approximately 1 to 3 per cent) these figures undoubtedly indicate persistent char- acteristic differences. In general, it may be said that, for adults, initia l right-hand rates range from 5 to 14 taps per sec. ff^ (3) The rate of tapping increases with age, at least between 6 --.and^lS^years. The slight drop at 13, upon which Gilbert comments, appears in Bryan's tables with some qualifications, but not so clearly in Smedley's results, which are reproduced herewith : it will be seen, however, that boys make no apparent gain from 13 to 14. TABLE 24 Dependence of Rate of Tapping upon Age (Smedley) NUMBER . TESTED Boys NUMBER TESTED GlKLS AGE TAPS IN 30 SECONDS TAPS IN 30 SECONDS Rt. Hand Lt. Hand Rt. Hand Lt. Hand 8 31 60 47 49 147 151 161 169 117 127 132 31 44 48 146 149 ^F>7 117 9 118 10 129 11 141 48 1 169 145 50 169 156 45 178 155 67 181 169 48 : 181 170 50 188 174 : 40 184 183 24 193 139 12 44 ' 170 50 184 40 184 37 1 191 21 ! 196 13 196 3 197 140 13 153 14 157 15 159 16 167 17 162 18 169 (4) Sex. The results of most investigators lead to the conclu- sion that boys are faster than girls, and that this sex difference increases with age. Bolton, however, has reported that "the girls are uniformly better than the boys," while Bryan found girls superior at 13, when they showed improvement and the boys little or none, — a tendency that is apparently allied to the actual crossing ^f the curves of height and weight.) More extensive experiments upon adults (10 men and 10 women) by Wells (21) now indicate that women surpass men in tapping with the right hand in the first experiment, whereas elsewhere they are inferior: the sex differences 110 PHYSICAL AND MOTOR CAPACITY found by this investigator are said to be "mainly in those features of the experiment which especially involve the affective factor in the subject's attitude; and they are manifestations of the greater responsiveness of the women to this affective element." (5) The index of righthandedness (per cent left-hand is of right- hand efficiency) was found by Wells to range from .81 to .94, aver- age .90, for adults, and by Smedley to vary with age in the case of school children, in such a manner that the average index was .82 at the age of 9, and .89 at the age of 18. It is evident, therefore, that righthandedness, so far as tapping is concerned, is more pro- nounced in childhood than in adult life. Wells also states (21) that "the right and left hands are farther apart in women," though the relationship is more variable in them than in men. (6) Righthandedness and intelligence. Smedley's conclusion that there exists a positive correlation between degree of righthand- edness and school standing, i. e., that the left-hand more nearly approaches the right-hand efficiency in the case of dull and back- ward pupils is not confirmed by the results of Bolton. (7) Warming-wp. In practically every continuous psycho- physical activity there appears a tendency to improvement due to what the Germans have termed Anregung. This 'warming-up' is a kind of momentum, not identical with practise, and its effect is to increase or heighten the activity, and thus to retard or even to obscure the appearance of indications of fatigue. In tapping, we observe fatigue within each 30-sec. trial, but a comparison of successive trials within a record will show the improvement due to warming-up. With 2.5 min. rest-pauses, Wells found this factor to be clearly present (up to the 7th trial at least) in right- hand records, but by no means so evident in left-hand records. The effect of warming-up appears to be primari y operative in increa ed immunity to fatigue, and is markedly augmented by practise, e.g., in tests continued for 20 days.| (8) Spurts. The curve of performance in tapping, as well as any psychophysical activity, is also liable to be influenced by short periods of increased activity, which, to continue the analogy of the race-track, may be termed 'spurts' (German, Antriebe). Thus, Wells' discovery that the first experiment usually excels the second in women whereas the reverse may be true in men, is re- TEST 10: TAPPING 111 f erred to a special incitement of novelty (Neuigkeitsantrieb) , which affects the women markedly Similarly, each 'record' may be affected by an initial spurt (Anfangsantrieh) or by a terminal spurt (Schlussantrieh) . These dynamogenic factors obvious" y tend to obscure the real effects of fatigue. • (9) Fatigue and the fatigue-index, (a) As just stated, the speed of tapp.ng normally declines after the 1st 5-sec. interval, until it is approximately ^ as great in the last as in the 1st interval (Wells). In 45-sec. trials, the fatigue-index (loss of last 5 sec. divided by initial 5 sec), according to Gilbert, is highest n young children (24 per cent at 8 years) and declines thence irregularly to 12.7 per cent at the age of 15.yT:'ests by the author of fifty 8th-grade gram- mar-school boys reveal a fatigue-index (ratio of loss in 3d 10 sec. to 1st 10 sec. in 30 sec. tapping) of 13.7 per cent, m.v. 4.8 per cent, for the right, and 15 per cent, m.v. 4.6 per cent, for the left hand. / (6) According to Gilbert, the fatigue-index is higher for boys than for girls, but boys tap faster throughout each trial, so that their net efficiency is higher.' (c) Kelly (10), who worked on a small number of children with a ''fatigue-counter, "found that " A"-grade pupils fatigued less than "C"-grade pupils; his index (the per cent of the last to the 1st 10 sec. in a 60-sec. trial) was for the former, with the finger 87.2 per cent, with the arm 88.0 per cent; for the 'atter, with the finger 77.0 per cent, with the arm 76.4 per cent. In the author's tests, no correlation could be discovered between fatigue-index and school standing. (d) The effect of fatigue is progressively to 'level up' individual differences in speed. In other words, individual differences are more evident in initial than in terminal intervals (Bliss, Wells). (e) Objective fatigue (slowing in rate) persists after practise, but the subjective feeling of fatigue may be eliminated thereby. (/) The fatigue induced by 30-sec. tapping is apparently com- pletely eradicated by a 3-min. rest-pause (Wells). ' It is more re*' e to interpret this higher index of boys as an expres- sion of greate nthiisiasm than to follow Havelock Ellis in his in- ference th« pleofthe"more continuous character of woman's activity ' yliss Thompson (17), from comparative tests of adults, ' urpass women both in initial rapidity and in power \ other tests show that high initial speed tends to be \tigueloss. / 112 PHYSICAL AND MOTOR CAPACITY (g) The fatigue-index of right and left hands shows only slight correlation (Wells). The author's tests, however, show a correla- tion in the case of 50 boys of .33. In some persons the left hand is less susceptible to fatigue than is the right hand, though the re- verse is the rule. (h) The subjective experience of fatigue, as has been intimated, does not accord with the objective fatigue-loss.^ (10) Practise (a) The effect of practise is to p oduce a gradual improvement in speed, with, of course, occasional losses. (6) The rise of the curve of efficiency is not, as in most activities, more rapid at the beginning than elsewhere. (c) Maxima efficiency, when two experiments are performed daily, s reached, apparently, in about 20 days. (d) Practise affects the left hand no more than the right; con- sequently the index of righthandedness is unaffected by repeti- tion of the test. (e) Practise particularly increases the rate in the later trials, i.e. it particularly affects warming-up, yet "the true practise gain is one mainly in the initial efficiency of performance, as distin- guished from the warming-up gain, which shows itself chiefly in continued efficiency of performance" (Wells). (/) An "ntermission of 10 to 14 days has no unfavorable effect upon practise gains, save that the feeling of fatigue may appear when work is resumed. (11) Diurnal rhijthm. Dresslar (7) found evidence of a diurnal rhythm with a minimum at 8 a.m. and a maximum at 4 p.m. Marsh (12 also found that afternoon records generally surpas ed those of the morning, though his figure;, do not accord very closely with those of Dresslar : Marsh also discovered that the later periods in the evening, which were not tested by Dresslar, furnished the most rapid rates of all. 1 According to Wells: "The objective fatigue phenomena which we note in the tcgt are in all probability either afatiguephenomenonin the refractory phase or a lowered efficiency of coordination, especial'- - Droduct of altered synaptic conditions; the sensations of fatigue, on th. 2. 'and, may with equal assurance be ascribed to tissue changes with- ps that take place as a result of their continued effort. In this t» '^e fatigue sensations are absolutely no indications of the a litions, and any traceable correspondence between fat 'atigue of performance must be regarded as almost whr inhi- bition." (19, p. 473). TEST 10: TAPPING 113 (12) Dependence on 'general condition.' When general well- being was ranked as good, medium, below medium, and poor, Wells was unable to discern any relation between these several conditions and tapping efficiency, while there was, in the case of susceptibility to fatigue, a tendency, if anything toward an in- verse reation, i.e., fatigue seemed to be greatest on 'good' days. Dress' ar's observation that a vigorous walk decreases while mental work increases speed of tapping has been generally confirmed by other invest gators. (13) Correlation with mental ability and social status, (a) The correlation between tapping ability and mental ability is found to be generally positive by Smedley, Gilbert, Bolton, and Kirk- patrick, to be indifferent by Bagley (also by the author), while Binet and Vaschide report a positive correlation with 12-year old pupils and an inverse correlation with 16 to 20-year old pupils. While Gilbert found a very marked positive superiority of the 'bright' children in general, the relation did not appear at ages 16 and 17. Bolton found that "good children" (apparently meaning those drawn from the better social classes) were uniformly superior in tapping to children of the poorer class, both with the right and with the left hand. The fact that the divergence is greater at 9 than at 8 years, he attributes to a general arrest of develop- ment in the poorer-class children. (6) Bolton also states that the "good" children showed a dis- tinctly greater practise-improvement — a discovery which he terms "new and significant," and which he thinks is indicative of a fun- damental difference in the ability o these two classes of children to take on new habits and profit quickly by experience. (14) Abnormal types. Smith reports his inability to discern any characteristic differences between the speed of tapping in epi- leptics and in normal individuals, or between the speed of tapping' and the rate of involuntary tremors in these cases. Wells' study of several cases of retardation in the insane (20) , however, revealed alterations of the tapping activity, both in the form of a lowered average efficiency and also in the form of improvement in rate under conditions in which a normal individual would show either no changfe '6? ^ positive loss. These changes he terms 'reversal,' by which is r#eant intra- as distinguished from inter-trial warming- 114 PHYSICAL AND MOTOR CAPACITY up, and 'transference,' by which is meant a tendency for the index of righthandedness to be lower when the right-hand record is taken after the left-hand record. (15) Dependence on the type of movement. The restriction of the tapping movement to specific joints, as has been attempted by some investigators, is difficult to accomplish in practise. However, it appears that the fastest rate is made when the movement is per- formed by the elbow joint, which is the one mainly concerned in the type of free movement here prescribed. Kelly, for instance, f^und that the speed of tapping was faster with the forearm than with the forefinger, in about the ratio 15 to 13. From this, in connection with other tests of dexterity, especially tests of mini- mal movement, he argues that children only gradually acquire dexterity and quickness of movement with the fingers, and that this passage "from fundamental to accessory," to use Burk's phrase, indicates the necessity of a general readjustment of the motor tasks required of children. REFERENCES (1) W. C. Bagley, On the correlation of mental and motor abilitj' in scliool children, in A. J. P., 12: 1901, 193-205. (2) A. Binet and J. Courtier, Recherches graphiques sur hi musique, in A. P., 2: 1895 (1896), 201-222. (3) A. Binet and N. Vaschide, (a) Epreuves de vitesse chez les jeunes gar- Qons. in A. P., 4: 1897 (1898), 64-98. (6) Experiences de vitesse chez les jeunes gens, ibid., 200-224. (4) C. B. Bliss, Investigations in reaction-time and attention, in Yale S., 1: 1893, 1-55. (5) T. L. Bolton, The relation of motor power to intelligence, in A. J. P., 14: 1903,615-631. (6) W. L. Bryan, On the development of voluntary motor ability, in A. J. P., 5: 1892, 123-204. (7) F. B. Dresslar, Some influences affecting the rate of voluntary motion, in A. J. P., 4: 1892, 514-527. (8) J. A. Gilbert, Researches on the mental and physical development of school children, in Yale S., 2: 1894, 40-100. (9) J. A. Gilbert, Researches upon school children and college students, in IowaS.,1: 1897,1-39. (10) R. L. Kelly, Psychophysical tests of normal and abnormal children; a comparative study, in P. R., 10: 1903, 345-372. (11) E. A. Kirkpatrick, Individual tests of school children, in F. R., 7: 1900, 274^280. TEST 10: TAPPING 115 (12) H. D. Marsh, The diurnal course of efficiency, Columbia Univ. diss.' N. Y., 1906. Pp. 99. (13) J. M.Moore, Studies of fatigue, in Yale S., 3: 1895,68-95. (14) O. Raif, Ueber Fingerfertigkeit beim Clavierspiel, in Z. P., 24: 1900. 352. (15) F. Smedley, Rept. dept. child-study and pedagogic investigation, No. 3, 1900-1901 (Chicago Public Schools). Also in U. S., 1902, i, 1115-1138. (16) W. G. Smith, A comparison of some mental and physical tests in their application to epileptics and to normal subjects, in B. J. P., 1 : 1905, 240-260. (17) Helen B. Thompson, The mental traits of sex, Chicago, 1903. Pp. 188. (18) F. L. Wells, A neglected measure of fatigue, in A. J. P., 19:1908, 345-358. (19) F. L. Wells, Normal performance in the tapping test before and dur- ing practise, with special reference to fatigue phenomena, in A. J. P., 19: 1908, 437-483. (20) F. L. Wells, Studies in retardation as given in the fatigue phenomena of the tapping test, in A. J. P., 20: 1909, 38-59. (21) F. L. Wells, Sex differences in the tapping test: an interpretation, in A. J. P., 20: 1909, 353-363. TEST 11 Accuracy or precision of movement: Aiming.— ^The purposes for which tests of accuracy of movement have been employed are practically the same as those cited for the tapping test, viz: to obtain an index of general voluntary motor ability with which to compare different children to compare the right with the left hand, to determine the development of motor control with age, its differentiation with sex, and to test its correlation with mental ability. These tests have been only rarely used for determining the presence of fatigue, though they have been proposed as means for the diagnosis of incipient ataxia. Tests of accuracy vary greatly in form : in fact, they virtually shade by degrees from those which prescribe a rapid accurate move- ment similar to the tapping movement of Test 10, e.^., the Colum- bia test described by Wissler (5), to those which prescribe a slow steady movement more akin to a test for steadiness (No. 13). Two types of precision test have been selected for considera- tion, the aiming test and the line drawing or 'tracing' test ,No. 12). 110 PHYSICAL AND M01\>U lArAdTV Tho common foaturo of all forms of aimiiiii; tost is the measure- ment of the extent of error made by an inilividual when he tries in a series of discrete voluntary movements of hand or arm to hit some form of mark or target. According to the particular form of movement employed, the test has been knoMTi as the ' probing test,' the 'target test.' the 'thrustinii test,' etc. These movements + 1 + 2 3 4 5 + 7 6 8 + 10 9 FIG. 23. TARGET BLANK. The numbers are added to sliow the order in whieh the oros^ses are to be struek. Cut ^ size. have ranged, to speak more specifically, from a simple vertical prob- ing movement of nnn. extent (Bryan, 2) to whole-arm aiming with a pencil at a paper target at arm's length (Thompson, 3: Whipple, 4), or making lunging thrusts with a wand, somewhat after the' fashion of a fencer, or even throwing ordinary marbles at a bull's eye target 2 m. distant (Bagley, l). The results of any such precision test will t)hviously be condi TEST 11: 117 t ioiicd by the position of the target with respect to S, by the extent and rate of the aiming movement, and likewise though the fact seems not always to have been recognized, by the individually variable improvement in accuracy which will appear if a series of ' shots' are taken at the same target. Hence, to be satisfactory, an aiming test should prescribe and standardize all these conditions : it should also admit of an exact evaluation of each aiming move- ment/J The form of test here described was devised by the author severa years ago to meet these conditions and has proved satis- factory in use. Though the error of a single stroke is large (as FIG 24. AD.IUSTABLE BASE-BOARD FOR TARGET-TEST. is certain to be the case in any form of aiming test), the average of the 30 thrusts made by the same *S' is very constant. The use of ten marks in place of a single mark, or bull's eye, removes to a large extent the improvement error just mentioned. Apparatus. — Prepared blanks containing ten crosses irregularly arranged (Fig. 23). A base board upon which the blanks are fastened, arranged to be secured upon the wall and adjusted to varying heights (Fig. 24). Metronome (Fig. 14). Pencil with tough and moderately hard lead. M Uimeter scale. 118 PHYSICAL AND MOTOR CAPACITY Preliminaries. — Fasten the board upon the wall and arrange the counterweight properly, so that the board will remain in any position from one to two meters from the floor and will not be dis- placed when struck by the pencil. Set the metronome at 69. Fasten the target-sheet upon the board, with the name-date corner n the lower right-hand corner of the board Place a demonstration target on the wall conven- iently near. Method. — Make clear to S the following directions. (1) He is to stand with his ight shoulder (for the right-hand test) squarely in front of the target, at such a distance that his pencil just strikes the target when his arm is fully extended.^ (2) He is to strike in time with the beat of the metronome. (3) Each stroke is to be a full, smooth stroke, not jerky or too short, and the pencil must, therefore, be brought back, each time, until it touches the shoulder. (4) He is to start at the first cross and make successive strokes, one at each cross in the series until the tenth is reached (see Fig. 23). This process is twice rej^eated, but in the second round, further to avoid practise, the order is from ten to one. S thus makes three shots at each mark, or thirty in all. Before conducting the test proper, let S try the experiment upon the demonstration target. It will save time if E also illustrates the process at this time. E should count out the strokes of the metronome: 'one, two, three," etc., to assist S in getting the pro- per rate. Place a fresh sheet upon the target-board and test the left hand. Treatment of Results.- -Measure the error of each thrust by the application of the millimeter scale. A pair of dividers may be helpful in this process. Average the thirty errors and compute the mean variation or standard deviation. Any 'shots' that have struck the lines of the crosses and are difficult to detect may be easily located by reversing the sheet. Results. — (1) On the basis of similar tests, other investigators have shown a gradual increase in accuracy with age, particularly during the years 5 to 8. ^ If the subjects are of approximately the same size, this distance may he marked upon the floor by a chalk line. TEST 12: TRACING 119 (2) Sex differences are slight, but on the whole boys are more accurate than girls, and men than women. (3) The author, in using the test as described, has found an error of 4 to 6 mm. in university students, while in a group of fifty St.'ti- grade boys the following results were obtained : Right hand, average 5.12, lowest 3.75, highest 8.34. Left hand, average 6.39, lowest 4.15, highest 9.27. (4) For the 50 boys just mentioned, the correlation between right and left-hand efficiency was 0.54. REFERENCED (1) W. C. Bagley, On the correlation of mental and motor ability in school children, in A. J. P., 12: 1901, 193-205'. (2) W. L. Bryan, On the developvnent of vohmtary motor ability, in A. J. P., 5: 1892, 12:3-204. (3) HelenB. Thompson, The rnental traits of sex, Chicago, 1903. Pp.188. (4) G. M. Whipple, The influence of forced respiration on psychical and physical activity, in A. J. ,P., 9: 1898, 580-571. (5) C. Wissler, The correlation of mental and physical tests, in P. R. M. S., 3: No. 6, 1901. Pp. 62. ' TEST 12 Accuracy, precision, or steadiness of movement : -Tracing. — The purposes for which tracing has been used are identical with those outlined for the preceding test, but the present test differs from the former in that the movement is continuous, analogous to that made in drawing a line — the so-called 'writing movement' of Bryan (3) or 'tracing test' of Bagley (1). Since steadiness of movement is quite as much in demand as accuracy (of the sort required in Test 11), this test is often classed as a steadiness test, rather than as an accuracy test, but it differs from steadiness tests proper in that it measures control of a voluntary movement, whereas the latter measure the extent of involuntary movement which takes place when the hand or arm is held at rest (Test 13). The technique most commonly adopted for the tracing or line- drawing test consists in passing a metallic needle or stylus along a narrow slit between metallic strips and noting by telegraph sounder, bell, electric counter, or graphic record, the number of contacts 120 PHYSICAL AND MOTOR CAPACITY made in passing along a given portion of the slit. This slit may be straight and bounded by parallel sid(\s (Bolton, 2) or by slightly .converging strips (Bryan, 3; Thompson, 4), or the slit may in por- tions be curved, as in the scrolls used by Bagley. Some tests have beenvmade at Columbia University with an irregular printed pat- tern, wffich is to be traced by the subject with a lead pencil. In any of these tests, the movement may be arranged to involve pri- marily either the. finer muscles of the hand and fingers or the larger muscles concerned ir\ the whole-aun movement. The test here describ«ed follows the method used by Bryan and by Thompson. Apparatus. — Tracing-boarc? (Fig. 25). Metallic stylus with flexible connecting wire. Telegraph sounder (Fig. 26). Two dry or other open-circuit cells. No. 18 annunciator wire. FIG. 2d. rUAClNO-BOARU. Preliminaries. — Wire the battery in series with the tracing- board, the sounder, and the stylus. FIG. 26. telb:ghai'h sounder and key. The sounder may be purchased separately, l)ut the key will be found useful for other experimental work, if only for opening and closing the cir- cuit at will. When provided with a special pointer, this sounder is used for graphic records, as in Test 13. TEST 12: TRACING 121 Method. — (1) Seat S comfortably with the tracing-board squarely before him, and the apex of the angle pointing toward him, 80 that the movement is directly toward the body in the median plane. Let >S hold the stylus as he chooses, place the tip at the opening of the strips, and then attempt to draw a line on the glass between the strips of metal without touching either one. The movement should be continuous from start to finish, made entirely free-arm (not a finger or wrist movement and without supporting the hand or arm in any way.) The rate of movement must be illus- trated as accurately as possible by E, and should be such that the full length of the strips is traversed in 9 sec. Allow S two or three preliminary trials, and endeavor to secure approximately this rate before starting the records. As soon as a click of the sounder indicates a contact, S is to stop and begin again with the left hand. Repeat until *S has made five trials with each hand, alternately. E records in each case the point on the scale at which contact is made. (2) Turn the apparatus around 180°, and in the same manner test movement away from the body. (3) Test movement from left to right and (4) from right to left with either hand, by placing the test-board so that the strips lie parallel with the edge of the table nearest >S'. Variations of Method. — (1) If it is desired not to compare movements in different directions, but merely to compare the right and left-hand efficiency of different *S's, the test may with advantage be shortened by adopting the method used by Bryan in his tests of school children, viz: set the test-board so that the strips make an angle of approximately 45° with the edge of the table, i.e., with the right-hand end of the strips turned 45° away from the body for the right-hand test, and with the left-hand end of the strips turned away to the same amount for the left-hand test. Make five tests with movement inward and five with move- ment outward with each hand. The conditions may be still further varied (2) by requiring S to stand and to hold the stylus at arm's length, (3) l)y allowing S to sit and to support his arm at the elbow, or (4) to support his hand on the base-board while executing a forearm or whole-arm movement. The last was the method followed by Miss Thompson. 122 PHYSICAL AND MOTOR CAPACITY Treatment of Results.— The simplest treatment of the data is to secure an index of precision by averaging the distances at which the several points of contact are made. For a more complex method of computing the measure of precision, the reader is referred to Bryan, pp. 180 ff. Results. — (1) There is greater tariation in the outcome of pre- cision or steadiness of movement tests than in that of rate of movement tests. (2) There is undoubtedly a more or less constant improvement in precision with age, but sufficient data are not yet at hand to determine fhe yearly increments clearly. There is certainly, how- ever, a decided gain during the years 6 to 8, while Bolton also noted impiovement from the 8th to the 9th year of age. (3) Dextrality. In general, the right hand is, of course, dis- tinctly superior to the left, but the amount of this superiority varies remarkably with age, and is, according to Bryan, less evi- dent at 15 and 16 than at 6, 9, or 12 years of age. (4) Sex. With either hand, boys are probably slightly superior to girls, for, while Bolton reports the superiority of the girls in his groups of children, Bryan's examination of some 700 children re- vealed the following relations, when the results for both hands were computed together: boys were superior to girls in 51.5 per cent of the trials, girls superior to boys in 35.3 Der cent, and the sexes equal in 13.4 per cent. Moreover, in a similar test, Thompson found men superior to women. (5) Both Bolton and Thompson found movements inward or toward the body uniformly steadier than movements outward or away from the body. (6) Correlations. Bagley found "a decidedly inverse relation between mental ability, as indicated by class standings, and motor ability, as indicated by the tracing test." But Bolton, who used a different form of test, and apparently correlated rather with the social status than with the class standing, reports that 'good' are steadier than 'poor' children. REFERENCES (1) W. C. Bagley, On the correlation of mental and motor ability in school children, in A. J. P., 12: 1901, 193-205. (2) T. L. Bolton, The relation of motor power to intelligence, in A. J. P., 14: 1903, 615-631. TEST 13: INVOLUNTARY MOVEMENT 123 (3) W. L. Bryan, On the development of voluntary motor ability, in A. J. P., 5: 1892, 123-204. (4) Helen B. Thompson, The mental traits of sex, C^hicago, 1903. Pp. 188. . TEST 13 ■of Steadiness of motor control: Involuntary movement. — The ;' general idea in this type of test is to measure the amount of invol- untary movement which appears when the finger, hand, arm, or body as a whole is held as nearly motionless as possible. Like the preceding tests, this has been frequently employed as a means of obtaining an index of motor ability (Bagley, 1; Bolton, 2; Hancock, 6), but it has also been employed for numerous other purposes, e.g., for examining the motor tendencies accompanying ideational activity (Jastrow, 7, 8; Tucker, 15), for examining the bodily expression of affective states (Titchener, 14), for examining the nature of constant tendencies toward automatic movements, or the possibility of developing such movements by training (Thompson, 13; Solomons and Stein, 10; Stein, 12), and for detecting the presence of incipient or recent chorea (Crichton- Browne, 4). This variety of purpose illustrates very forcibly the difficulty of so conducting the test as to examine any one phase alone — a difficulty which is aggravated by the fact that the tracings of involuntary movement are often affected, not only by the factors implied above, but also to a considerable extent by the direciion of the attention, by the relative position of the body and the instru- ment, and by physiological processes, especially respiration. It is, therefore, not surprising that many writers, e.g., Bagley and Bolton, have incorporated records of involuntary movement only with qualifications and without placing much insistence on their worth. The instruments most commonly employed are the ataxiagraph (described by Crichton-Browne, used by him and by Hancock and Bolton for measuring the swaying of the body as a whole), the tremograph (described by Bullard and Brackett, 3, and also used by Hancock) for testing the arm or finger, the automatograph of Jastrow or that of Stein, further elaborated by Titchener, for measuring involuntary movements of the arm, the digitalgraph m PHYSICAL AND MOTOR CAPACITY devised by Delabarre (5), described, together with others of the instruments just named, by MacDonald (9), and used for recording the tremor of the j&ngejjand several instruments, as yet umiamed, for testing either the arm or hand, according to the conditions of their use.^ Apparatus. — Brass plate, set at an angle of 45°, and pierced with a series of holes whose diameters are 32, 20, 16, 13, 11, 10, 9, 8, and 7 sixtj^-fourths of an inch, respectively (Fig. 27). Metallic needle of special design, with flexible comiecting wire. Telegraph STEADINESS TESTER. sounder (Fig. 26), with writing lever attached to the armature, and ^vith sending key (or separate short-circuiting key) . Kymo- graph (Fig. 15), \vith accessories. Stop-watch (Fig. 13). Four dry or other open-circuit cells. Insulated connecting wire. [A low table, about 65 cm. high and an adjustable chair of the typewriter chair pattern are desirable.] Preliminaries. — Wire the needle, brass plate, sounder and key in series with the battery in such a way that contact between the needle and the plate will actuate the sounder. Place smoked paper on the kymograph drum, and adjust to approximately one ' For some purposes it is probably quite as well to test the subject with- out the use of apparatus. Thus, according to Sturgis, the following consti- tutes an infallible test for chorea: "Bid the child liold up both hands open, with extended arms, the palms toward you. If that is done steadily, both liands ui)right and both alike, no finger or nand quivering, no falling back of either hancl, nothing to choose between the positions of the two, then the child has not, nor is it near (either before or after) St. Vitus' dance. You may confirm this test by another. Let ttie chila place the open hands upon yours, palm to palm. Look then at the backs of the child's hands, observe whether fingers or thumbs (.esi^ecially the latter) repose without tremor and without restraint." TEST 13: INVOLUNTARY MOVEMENT 125 revolutioji in 40 sec. Adjust the position of the sounder so that the writing lever gives a satisfactory tracing on the drum.' Method. — Seat S before the table in a comfortable position. Place the brass plate flush with the front edge of the table, in front of aS's right shoulder for the right-hand test, and in front of his left shoulder for the left-hand test. Instruct S to hold the needle in such a way that his finger tips are in contact with the expanded flange of the holder, and at command, to hold the tip of the needle within the largest hole, and to maintain this position, so far as pos- sible, without touching the brass plate during the 15 sec. allowed for the trial. *S's hand and arm must be entirely free from all sup- port or contact with his body or other object, and his forearm should form an angle of approximately 100° with his upper arm. The needle should be inserted about 6 mm. into the hole. Show S that the click of the sounder will serve as warning for him that the needle is making contact with the plate. In conducting the test, allow *S about 3 sec. for taking the position (since a certain amount of movement will appear when the needle is first inserted that will afterward be checked by»S's control), then close the short-circuiting key and at the same time start the stop- watch. At the expiration of 15 sec, open the circuit, stop the watch and the kymograph, and at once rearrange the instruments for the left-hand test. Allow 30 sec. rest, and then test the right hand and the left hand with the next smaller hole, and so on, until a hole is reached so small that S has reached the limit of his capacity, and is clearly unable to keep the needle free from contact with the plate. In most cases a few unrecorded preliminary trials will show ap- proximately the degree of *S's control, and the tests with the larger holes may be omitted with a consequent saving in time. Variations of Method. — The test may be modified by altering the relative position of the plate and S's body as follows: (a) by requiring S to stand and to hold the needle extended at arm's length, (h) by allowing S to rest his elbow upon the table with the forearm free, or (c) by supporting his forearm and wrist and test- ing the steadiness of the hand and fimgers, and (d) by extending the time to 30 or 60 sec. The results will naturally differ character- ' For an account of the manipulation of the kymograph, consult Test 10. 126 PHYSICAL AND MOTOR CAPACITY istically from those ol)tained under the conditions prescribed as standard. Treatment of Results. — By reference to the kymograph trac- ings, count the number of contacts for each hole. For compara- tive purposes, E may take the total number of contacts made in a given series of holes or the number made in that hole which most satisfactorily tests the steadiness of the subjects under in- vestigation. Results.' — (1) In all tests of involuntary movement, it is clearly seen that age is an important factor. Hancock concludes that adults have approximately 5.8 times as much control over ■their fingers as do children aged 5 to 7 years. (2) Distinct sex differences have not been established. Notes. — In rare cases, the use of the graphic method may be dis- pensed with, especially if E proves by practise very skillful in the correct counting of the rapid and irregular strokes made by the sounder when the test approaches the limit of *S's capacity, but this simplification is not recommended, because it is exceedingly^ difficult to make the count under these conditions (cf. Bolton), and moreover, there may appear very short rapid contacts that will actuate the sounder sufficiently to produce a noticeable indication on the tracing, but not sufficiently to produce a noticeable click. Again, S will occasionally make contacts of longer duration: with the graphic record, it is possible for E to measure the duration of these contacts, and, if desired, to base *S's record upon the propor- tion of the time during the 15 sec. that is occupied in contact, rather than upon the number of contacts alone. The electric counter is not recommended as a substitute for the graphic method, because, as mentioned in Test 10, it will not op- erate reliably when the contact is very brief, even although a large number of cells are used in the battery. To avoid the possibility of a similar error in the use of the sounder, the excursion of the arma- ture should be rather short, i e., the armature should be adjusted as near the fields as is possible if it is to give a clean stroke. If involuntary tremor is to be studied with special care, e.g., if E wishes to make an extended study of an individual case, a 'i A summary of the results obtained with the automatograph is given by Lstrow (8, pp. 307-336). -' n. summary oi i-ne r Jastrow (8, pp. 307-336) TEST 13: INVOLUNTARY MOVEMENT 127 more sensitive instrument should be employed. For such work, the tridimensional analyzer of Sommer (11), also described by Titchener (14), is recommended. ^^' REFERENCES (1) W. C. Bagley, On the correlation of mental and motor ability in school children, in A. J. P., 12: 1901, 193-205. (2) T. L. Bolton, The relation of motor power to intelligence, in A. J. P., 14: 1903,615-631. (3).Bullard and Brackett, Boston Med. and Surg. Jour., 2: 1888, 600. (4) J. Crichton-Browne, The nervous system and education, being part 4 in The book of health, edited by M. Morris, London, 1883 (now out of print). (5) E. B. Delabarre, Ueber Bewegungsempfindungen, 1891. See also A. P., 1: 1894 (1895), 532. (6) J. A. Hancock, A preliminary study of motor ability, in Pd.S., 3: 1894, 9-29. (7) J. Jastrow, (a) A study of involuntary movements, in A. J. P., 4: 1892, 398-407. (6) A further study of involuntary movement, ibid., 5 : 1892, 223-231. (8) J. Jastrow, Fact and fable in psychology, Boston, 1900. Pp. 370. (9) A. MacDonald, Experimental study of school children, etc., chs. 21 and 25 in U. S., 1897-8 (1899). (10) L. Solomons and G. Stein, Normal motor automatisms, in P. R., 3: 1896, 492-512. (11) R. Sommer, Lehrbuch der psychopathologischen Untersuchungs- methoden, Berlin, 1899. Pp. 399. (12) Gertrude Stein, Cultivated motor automatisms, in P. R., 5: 1898, 295-306. (13) Helen Thompson, The mental traits of sex, Chicago, 1903. Pp. 188. (14) E. B. Titchener, p]xperimental psychology, 2 vols., N. Y., 1901 and 1905. (.15) M. A. Tucker, Comparative observations on the involuntary move- ments of adults and children, in A. J. P.' 8: 1897, 394-404. CHAPTER VI Tests of Setssory Capacity Psychophysical tests of sensoiy capacity are divisible into tests of liminal sensitivity (sensitivity proper) and tests of discrimina- tive or differential sensitivity (sensible discrimination.)^ In the former, we measure the bare capacity of experiencing sensations, the minimally perceptible stimulus or stimulus limen, e.g., the lightest pressure that can be felt, the least intensity of tone that can be heard, etc ; in the latter, we experience different sensations and report upon their difference; we seek, in other words, to determine the minimal objective difference of stimulation that can just be mentally cognized as different, to determine the differ- ence limen, e.g., the smallest change of vibration-rate that will suffice to yield two perceptibly different tones, or the smallest dif- ference in weight that can just be recognized as a difference. These two measurements of sensory capacity, liminal and dis- criminative sensitivity, can be applied to any modality, i.e., to any sense-department, and to any attribute of sensation, i.e., to quality, intensity, extent, and duration. We may measure, for instance, in the case of the ear, liminal sensitivity in terms of in- tensity, discriminative sensitivity in terms of intensity, in terms of quality, in terms of duration, etc. We may, furthermore, deter- mine the total number of auditory sensations that can be experi- enced, i.e., modal sensitivity. In the case of the two sense-departments that possess the attri- bute of extent, visual and cutaneous sensations, we may also meas- ure the capacity to discriminate difference in the localization of two stimuli, or the limen for spatial discrimination. While, strictly speaking, this may be regarded as a more complex process than simple sensory discrimination, it may, for our purposes, be included ' For a discussion of the terminology, methods, purposes, and results of psychophysical methods, consult Titchener, E'.rpenmenfa^ Psychology, espe- cially, Vol. 2, part 2. SENSORY CAPACITY 129 as a sensory test. Indeed, with both the eye and the skin, this determination, for practical purposes of exploration of the sense- organ or of the measurement of its functional capacity, has super- seded the determination of liminal sensitivity. Thus, in visual sensation, visual acuity refers, not to the liminal sensitivity of the retina for stimulation, but, in principle, to the capacity to distinguish tlie separation of two points. Similarly, in cutaneous sensation, For a practical test of functional capacity, the determination of the 'limen of duality,' as it may be termed, by means of the esthesio- meter, has been more often employed than the simple determina- tion of liminal sensitivity to pressure. Now, the quantitative determination of sensitivity in the psy- chological laboratory has given rise to a most elaborate and re- fined methodology, and has, in fact, been the chief problem of the science of psychophysics. It is not the purpose here to discuss or duplicate these exact methods, but merely to indicate the maimer in which, for comparative purposes, one may secure an index of functional efficiency by empirical methods. It must be clearly understood that the determination of an exact stimulus or differ- (Mice limen in the psychological laboratory, with minute introspec- tive analysis of the factors that condition the process and with elaborate methodological procedure, is quite a different process from this simple determination of functional capacity for compara- tive purposes. If, for convenience, the technical terms of psycho- physics are here employed, they are employed with this qualifi- cation in mind. To make this point clearer, the procedure for the determination of discriminative capacity as herein reconmiended is not identical with any of the established psychophysical methods. E allows S at first to try various stimulus differences ranging from large to small, until *S' has acquired general familiarity with the test, and E has obtained a general notion of the *S's capacity. E may next test S more formally by applying a series of stimulus differences ranging from clear subjective chfference to subjective equality. Hethen selects a difference which seems likely to be just cognizable by the subject and applies this difference ten times, with proper reversals for time or space errors. If eight right judgments are given, he then corroborates the result by trying similarly a slightly smaller, 130 SENSORY CAPACITY and finally a slightly larger difference, to see if S gives in the former case fewer, and in the latter case more correct judgments. S knows that a difference exists, but is ignorant of its spatial or temporal position.^ We thus obtain an index of capacity, but do not determine the mean difference limen, nor even the lower limen, in the psychophysical sense. Sensory tests of this empirical sort have been employed, partly in connection with the psychology of individual and sex differences, partly in the objective study of general intelligence, partly in the exploration of sense organs for the determination of their working condition, i.e., for hygienic and diagnostic purposes. In all of these fields the emphasis is upon the examination of simple func- tional capacity, without particular reference to introspective exami- nation or analysis of the accompanying consciousness. The use of sensory tests in correlation work is particularly interesting. In general, some writers are convinced that keen discrimination is a prerequi- site to keen intelligence, while others are equally convinced that intelligence is essentially conditioned by 'higher' processes, and onlj- remotely by sensory capacity,— barring, of course, such diminution of capacity as to interfere seriously with the experiencing of sensations, as in partial deafness or partial loss of vision. While it is scarcely the place here to discuss the evolutionary significance of discriminative sensitivity, it may be pointed out that the normal capacity is many times in excess of the actual demands of life, and that it is conse- quently difficult to understand why nature has been so prolific and generous ; to understand, in other words, what is the sanction for the seemingly hyper- trophied discriminative capacity of the human sense organs. The usual 'teleological explanations' of our sensory life fail to account for this discrep- ancy. Again, the very fact of the existence of this surplus capacity seems to negative at the outset the notion that sensory capacity can be a condition- ing factor in intelligence, — with the qualification already noted. The tests which follow are selected from a large number of theo- retically possible tests, because of their i^rominence in such experi- mental studies as have been mentioned. Their classification is simply by sense-departments. Tests for the exploration of the organ, measurement of its defects, determination of acuity, liminal ' Procedure with iialf-knowlodgo' in the sense used by Kampfe, in Phil. Stud., 8: 1893, 543, and Wundt, Grundzugc d. physiol. psvch., 5th ed., i., 1902, 492. See also Titchener, ii, Pt. 2, 127. TEST 14: VISUAL ACUITY 131 sensitivity, discriminative sensitivity, in so far as they are described, are given successively for each of the main sense departments. TEST 14 Visual acuity. — The functional capacity of the eye is examined primarily, of course, for practical purposes in connection with hygienic investigation. Occasionally, it becomes desirable to determine the presence or absence of visual defect in connection with the administration of some mental test, e.g., the cancella- tion test. Visual acuity has been studied in its relation to school stand- ing, general intelligence, occupation, habitat, race, sex, as well as to bodily disturbances, such as headache, chorea, indigestion, or other optical defects, such as strabismus, total color-blindness, etc. Optical inefficiency, aside from color-blindness, may be due to amblyopia (dimness of sight not due to refractive errors or demon- strable lesion), or to asthenopia (general impairment of retinal effi- ciency due to anaemia, over-use, etc., and often yielding to proper medical treatment), but is more commonly some form of ametropia (defect in shape of the eye-ball, lens, or cornea, with resultant de- fect in refraction and in the formation of the retinal image). Ametropia may exist as presbyopia, myopia, hyperopia, or astig- matism. Presbyopia is the- long-sightedness of old age, due to the lessened elasticity of the lens. Myopia, or short-sightedness, is commonly produced by too long an eye- ball, the effect of which is to allow rays of light in distant vision to focus in front of the retina and hence to produce a blurred image when they finally impinge upon the retina. The myopic eye is thus unable by any effort clearly to see objects situated at distances of 2 m. or more away, while its 'near-point,' i. e., the nearest point at which clear vision is possible, is brought correspondingly closer, so that objects may be seen clearly when 5 or 6 cm. distant, — something which would be impossible for the normal (emmetropic) eye. Myopia is rarely congenital, but is an acquired defect, and characteristically a disease of civilization and culture. Pure myopia, as a rule, causes no eye-strain, but it is nevertheless a serious condition, because of its tendency to increase in degree, and because of the appearance in many cases of concomitant pathological disturbances of the retina, which, in extreme cases, result in actual blindness. In practise, moreover, myopia is 182 SENSORY CAPACITY rarely found pure, but complicated with astigmatism and other defects. It may be counteracted, and its progress checked, but not cured, by the use of properly fitted concave lenses, supplemented by the exercise of caution in the use of the eyes. Hyperopia, hypermetropia, or long-sightedness, moi-e exactly over-sight- edness, is commonly produced by too short an ej^e-ball, the effect of which is to intercept rays of light too soon, i.e., before they are brought to a normal focus. The hyperopic eye must consequently exert an effort of accommo- dation in order clearly to see objects at a distance, while for near work this effort must be excessive. The result is that the hyperopic eye is under constant and abnormal strain from the incessant demands upon its ciliary muscle, and that, in consequence, numerous secondary symptoms or result- ant effects appear, some of them obvious, others unexpected, many of them serious. Local symptoms appear in inflammation, redness, or soreness of the eyes, lids, or conjunctiva, and in twitching and pain within the eye-ball. Aside from these local disturbances, perhaps the most constant symptom of hyperopia is frontal or occipital headache. Characteristic also is the hold- ing of the work at some distance from the eyes,' a peering or frowning expression, and dislike of near work. Eye-strain, whether hyperopic or astigmatic, may also occasion more serious physiological disturbances, such as chorea, vomiting, nervous dyspepsia, etc.- Since the hyperopic eye can see clearly at a distance and can read (as its possessor often boasts) with the book held at some distance, the defect is often unsuspected, because the sec- ondary symptoms are not correctly interpreted. On this account, too, it becomes necessary to take special steps to detect its presence, and many of the simple distance tests that have been applied w^holesale upon school chil- dren utterly fail to diagnose it. The oculist commonly makes use of hom- atropin or some other cycloplegic to paralyze temporarily the ciliary muscle and thus prevent accommodation. Hyperopia may, however, be detected, though less accurately, by the use of suitable test-lenses, as described below. The defect is counteracted by the use of properly fitted convex lenses. Asiigmaiism is produced by an uneven radius of curvature, usually of the cornea: this surface, which should normally be approximately spherical in form, is, in astigmatism, more strongly curved in one axis or meridian than in another, so that the cornea is ellipsoidal in form, e. g., like the bowl of a spoon, or the side, rather than the end of an egg. Thus the eye is double- focussed, and it is impossible by any effort to focus an image clearly in both meridians simultaneously. In measuring astigmatism it is evident that one must assign both the degree of refractive error and the axis in which the ' In high grades of hyperopia, distinct images can not be secured even by this process, so the child may abandon the attempt to secure clearness and seek merely to increase tne size of the image by holding his book near his eyes. He may thus bo falsely rated as near-sighted by tlie casual observer. - The injurious effects of eye-strain have found a soeciai expositor in Dr. G. M. Goula (6, 7). TEST 14: VISUAL ACUITY 133 error lies, and that in correcting it, a cylindrical lens of the proper curvature must be placed before the eye at exactly the proper axis to counteract the indicated deficiency. This lens only counteracts the defect, and does not cure it. Astigmatism may be in part congenital, in part a phenomenon of growth (often attributable to the pressure upon the eye-ball of the eye-lids and contracted brows, with the result that the maximal refractive index lies at or near the vertical meridian). When present in large amount it becomes a serious obstacle to vision; when present in small amounts, as is apt to be the case in many ej^es, it is the occasion of the same phenomena of eye-strain that have been mentioned as accessory to hyperopia; astigmatic headache is particularly symptomatic, — indeed, 60 per cent of all headaches are said to be traceable to this source. It must be understood that these three defects may, and commonly do, appear in combination, particularly astigmatism with hyperopia or myopia, and that the defects may be, and commonly are, unlike in the two eyes of the same individual. Partly for this reason, the proper fitting of glasses is an art, and, like any art, requires great skill, complete familiarity with the conditions, and long practical experience. The tests which are here described make no pretence to exactitude, but are designed to deter- mine, in so far as is possible by simple methods, the existence of defects that should invariahly he referred to a specialist for further examination and treatment. For the examination of refraction the chief appliances are (1) the ophthahnometer, for the exact measurement of the degree and axis of astigmatism, (2) the ophthalmoscope, for the examina- tion of the retina, (2) the retinoscope and the skiascope, for the objective determination of refractive errors, (4) test-types and trial lenses, for actual visual tests under varying conditions. While retinoscopy is a method of great value, especially in testing young children, the test-type is, in general, the court of final appeal and constitutes the most widely used and perhaps the most valuable single means for testing visual acuity. The most varied kinds of test-type have been devised by oculists. Perhaps oldest and best known are Snellen's "Optotypi," which form the basis of the tests ordinarily used. Interesting variations are seen in Dennett's Monoyer type, Landolt's C-test, and Cohn's E-test. The simple test which is described just below is recommended by the American Ophthalmological Society, and is designed to be used in connection with Dennett's type, with the employment of but two test lenses. The second test supplements the first by detecting astigmatism. Tests 15, 16, and 17 may be added. 134 SENSORY CAPACITY A. TEST FOR AMETROPIA Apparatus.- — Dennett's Monoyer test-type (Fig. 28).^ Trial frame (Fig. 29). Two — .75 D. and two + -75 D. spherical lenses, and one blank disc. MRTVruENCXOZD dlvatB KuehSn nCTHOFMErSPA EXAtZHDWN Y E I K S F D I X P H B Z D N L T A V R H S UE M C F Z u Fi(?. 28. Dennett's monoyer test-type. About T natural size. The small figures that indicate the normal distance for each line are not shown. ' Other test-types may, of course, be employed. Landolt's or Snellen E- test (designed for illiterates) are recommended, if Dennett's is not used. Landolt's type is placed at 5 m., Snellen's at 6 m. Visual acuity is indicated in the latter by a fraction in which the numerator is 20 (feet) and the denom- inator is the number (in feet) indicated for the smallest line that can be read at the standard distance of 20 feet. TEST 14: VISUAL ACUITY 135 Preliminaries. — Place the test type on the wall or stand, on a level with S's eyes, in a strong even illumination, though not in actual sun-light.^ Seat S comfortably at a distance of 6 m. from the chart. Note any indication of soreness or inflammation of the eyes, lids, or conjunctiva. Ascertain if S has ever suffered from such inflam- mations, from habitual headaches, or watery eyes ; whether his eyes l)ecome painful, sore, or strained in doing close work; whether he FIG. 29. TRIAL FRAME. Holds two pairs of lenses, and has rack and pinion adjustment for pupil- lary distance, and for vertical, and back and forward movement of the nose- piece. has previously been examined, and if so, with what result; whether he has ever worn glasses; if they have been worn and discarded ' If conditions render daylight illumination unreliable or unsatisfactory, E must arrange artificial illumination, carefully shaded from *S. Excel- lent devices for this purpose may be purchased from dealers in optical sup- plies, e.g., No. 4274, catalog of E. B. Meyrowitz, New York City, price SIO. Whatever the source of illumination, the light must not shine in *S"s eyes. An experienced £ may compensate for inadequate illumination by plac- ing S can, with the plus lens, read the same line as before, or a smaller line than before, then the eye is hyperopic. Thus, if pre- viously the 4th line was read and now the 2d, the record will be, V. R. E. = .7 + Hy. = .9, or, if no improvement appeared, V. R. E. = .7 -f- Hy. = .7. (3) If, in the first test, vision is less than 1.0, and if, in thesecond test, vision is impaired by the convex lens, then next replace the con- vex lens by the — .75 D. lens. If vision is now improved so that a smaller line is read, say the second or the first, then the eye is myopic, and may be recorded thus; V. R. E. = .7 -|- My. = .9, or V. R. E. = .7 + My. = l.i (4) Place the solid disk before the right eye, and test the left eye similarly. Record the results for each eye separately, e.g., V. R. E. = 1. Em. V. L. E. = .6 + My. = .9. ' This test may be nullified in some instances bj- a ciliary spasm in a hyper- opic eye which may simulate myopia of almost any degree. TEST 14: VISUAL ACUITY B. TEST FOR ASTIGMATISM 13/ Apparatus. — Trial frame and lenses as above. Verhoeff's astig- matic chart (Fig. 30). ^ Preliminaries. — Place the chart on the wall, and seat S as in the previous test. Be sure that S's head is held squarely erect, [f S 30. VERHOEFP" S .ASTIGMATIC CHART. About I natural .size. has been found to be myopic or hyperopic, place in the trial frame the lenses which correct, at least partially, this defect. Method. — Place the sohd disk in the frame before the left eye. Ask S whether one or more of the radiating lines seem to him sharper and blacker than those at right angles to them. If he answers in the affirmative, astigmatism is present. This result may be con- firmed by causing S to move his head from one shoulder to the other, in which case the location of the sharpest lines should shift in a corresponding manner. The degree of astigmatism may be ' Any standard astigmatic cnarx, maj- be substituted, but Verhoeff's isj in the author's judgment, best adapted both for making evident the presence of a.stigmatism and for determining approximately its axis. 138 SENSORY CAPACITY roughly judged by the positiveness and readiness of *S's answer; its axis may be determined approximately by his designation of the blackest line or lines. Place the disc before the right, and test similarly the left eye. Since astigmatism may exist either alone or in combination with some form of ametropia, it should, when found, be recorded with the previous determination, e.g., V. R. E. = .7 + My. = .9 + As. If vision is less than .7, but no form of ametropia can be demon- strated, the defect is recorded as amblyopia, e.g., V. L. E. = .6 + Am. To summarize the two tests: emmetropia is indicated (unless strain symptoms point to concealed hyperopia) by the reading of the smallest line and subsequent blurring by the convex lens, hyper- opia by improvement or lack of impairment of vision by the plus lens, myopia by vision less than 1 ., which is improved by the concave lens, astigmatism by unequal clearness of the radiating lines, amblyopia by vision less than .7 without demonstrable refractive error. Results.' — (1) The frequency of defective visual acuity is somewhat difficult to state accurately owing to the differences in method and in degree of rigor and precision that have characterized the many investigations upon this point. In especial, a great many investigations in school systems have been made by simple distance tests without the aid of lenses, so that hyperopia, the most frequent defect, has gone practically unmeasured. The general outcome of these simple tests is quite uniform, viz: that one child in three in the public schools suffers from visual defect. Typical figures are those obtained by Welch (20) at Passaic, and Smedley (15) at Chi- cago; the lattqr reports that 32 per cent of the 2030 boys and 37 per cent of the 2735 girls examined were defective in vision. While more than half of these defects are of a minor degree, yet, as already indicated, these may be productive of immediate distress and entail serious consequences if neglected. ' For a general discussion of the examination of eyesight, with special reference to the eyesight of school children, consult Barry (1) Calhoun (3), Carter (4), Cohn (5), Gould (6,7), Hope and Browne (8), Kotelmann (9), Newsholme (10), Risley (12,13), Schmidt-Rimpler (14), Snell (16), Stilling (17) and Young (21). TEST 14: VISUAL ACUITY 139 On the other hand, examinations that have been conducted by skilled opthalmologists with some refinement of method indicate a much larger percentage of defect.^ Risley's figures indicate that it would be more correct to state that seven children in eight, than that one in three, are ametropic. As chairman of the Philadelphia committee that examined some 2500 children, he gives the refrac- tion at 8| years of age as hyperopic in 88.11 per cent of cases, emmetropic in 7.01 per cent, and myopic in 4.27 per cent: at 17.5 years as hyperopic in 66.84 per cent, emmetropic in 12.28 per cent and myopic in 19.33 per cent. (2) From the above figures we may conclude that the eye in early childhood is an incomplete eye, naturally underfocussed and poorly adapted for near work. But, as general bodily maturity approaches, the eye under optimal conditions tends to become emmetropic. Conditions of modern life, however, are not optimal for the eye, but rather encourage overuse and neglect, with the con- sequence that these, especially when added to astigmatism or other congenital defects, produce eye-strain, myopia, and other disturb- ances of vision. (3) Cohn, especially, has demonstrated that myopia is essen- tially a disease of civihzation and culture; that it is infrequent in peasants and those who lead an outdoor life, and progressively more prevalent and of higher degree as persistent study and near work continue. Thus, in gymnasia, he found that the percentage of myopia increased during six years of study in the following manner : 12.5, 18.2, 23.7, 31.0, 41.3, 55.8. Similarly, of 138 pupils at the Friedrich gymnasium who were examined twice at an interval of 18 months, he found at the first test 70 normal and 54 myopic, whereas in the second test, 14 of the 70 had become myopic, 28 of the original myopes had developed a higher degree, and in 10 per cent, serious structural changes had taken place in the retina.^ Indeed, there are specialists who assert that an absolutely perfect pair of eyes does not exist. - Diirr, in explanation of the alarming prevalence of myopia in Germany as contrasted with other countries, has sought to show its dependence upon the excessive demands of theGerman schoolsystem : he estimates that, during the years 10 to 19, the typical English boy spends in study 16,500 hours, in exercise 4500; the French boy, in stud> 19,000, in exercise 1300; and the Ger- man boy, in study 20,000, and in exercise 650. These statistics showing the relation between myopia and excessive near work could be multiplied almost indefinitely. 140 SENSORY CAPACITY (4) Smedley, on the basis of his method of correlation by grades, asserts that ''a smaller per cent of the pupils at and above grade have defective sight than those below grade." (5) Smedley further demonstrates that defective vision is extremely common in backward and troublesome children, and that this fact may be a partial explanation of the behavior of such chil- dren. Thus, according to the tests employed at Chicago, 48 per cent of the boys of the John Worthy School were subnormal in vis- ual acuity, as contrasted with 28 per cent of the boys of the same average age in other schools. Moreover, "many of the John Worthy boys had strabismus, hypermetropia, and astigmatism, concUtions which would induce asthenopia when the eyes were used in close and long application to books, and it is easy to believe," adds Smedley, "that the strain thus set up when an attempt was made to study was a factor in producing cUslike for school and subsequent truancy." (6) Van Biervliet (19) has sought to obtain a correlation be- tween the visual acuity test and intelligence, not by cUrect reference to visual acuity itself, but to the mean variation measured by a series of tests of acuity, i.e., to what he teims the capacity of attention. In brief, the method was to compute a fraction, of which the denomina- tor represented the average distance at which the test was visible, and the numerator the mean variation of the several trials. As measured by this arbitrary index, the 10 brightest and the 10 dullest of 300 university students were related, in terms of a common denominator, as 19/1000 and 62.5/1000. Binet (2), however, points out that the dull students had the better eyesight, i.e., the larger denominator, and suggests that the index be taken directly Prom the mean variation. (7) Right-euedness. While both eyes are employed for binocu- lar vision, there is some evidence that most persons 'favor' one eye, whenever, for any reason, binocular vision is not in use, e.g., in looking through a microscope or telescope. Van Biervliet (18) has measured the visual acuity of 100 persons, whose optical defects had previously been corrected, with the result that the favored eye very uniformly excels the unfavored one in visual acuity by one-ninth: he further asserts that right-handed people are right-eyed and left-handed people left-eyed, and that the same TEST 14: VISUAL ACUITY 141 sort of sensorial asymmetry can be demonstrated in audition, cutaneous discrimination, and discrimination of lifted weights. Notes. — Statistics of visual defect are rendered difficult of com- parison, not only by differences in the methods followed in the investigations, but also at times by failure to state whether exam- inations were made with or without the glasses actually worn by pupils, or, in case such statement is made, to indicate the effect upon the results of including or excluding trials made with these glasses. Cohn distinguishes between visual capacity proper (Sehleistung) and visual acuity (Sehschdrfe) , which is the efficiency when proper glasses are used. But since numbers of children are daily wearing improperly fitted glasses, one almost needs another term to indicate the vision that is had with these glasses. The focal strength of lenses was formerly indicated in the English, or inch system, but it is now more common among opticians to make use of the metric system, in which one dioptric (D) represents a lens whose focal distance is one meter, 2 D. a lens twice as strong, i.e., with a focal distance of 50 cm., etc. REFERENCES (1) W. F. Barr\-, The hygiene of the schoolroom, N. Y., 2d ed., 1904. Pp. 191. (2) A. Binet, A propos de la mesure de I'inteUigence, in A. P., 11: 1904 (1905), 69-82. (3) A. M. Calhoun, Effects of student life upon eyesight. Bureau Educ, Washington, 1881. Pp. 29. (4) R. B. Carter, Report on vision of children attending elementary schools in London, London, 1896. Pp. 16. (5) H. Cohn, (a) The hygiene of the eye, Eng. tr., London, 1886. (6) Die Sehleistung von 50,000 Breslauer Schulkindern, nebst Anleitung zur ahnlicher Untersuchungen fur Aerzte u. Lehrer, Breslau,. 1899.- Pp. 148. (c) Was haben die.Augenarzte fiir die Schulhygiene geleistet und was miis- sen sie noch leisten?, Berlin, 1904. Pp. 35. (Contains bibliography of 74 titles.) (6) G. M. Gould, The cause, nature and consequences of eyestrain, in Pop. Sci. Mo., 67: 1905, 738-747. (7) G. M. Gould, Biographic clinics, 5 vols., Phil., 1903-7. (8) E. Hope and E. Browne, A manual of school hygiene, Cambridge, Eng. , 1904. Pp. 207. " (9) L. Kotelmann, School hygiene, Eng. tr., Syracuse, 1899. Pp. 382. (10) A. Newsholme, School hygiene, Boston, 1894. Pp. 140. 142 SENSORY CAPACITY (11) Rept. com. on statistics of defective sight and hearing of public school children, in U. S., 1902, ii., 2143-2155. (12) S. D. Risley, Weak eyes in the public schools of Philadelphia: rept. com. on examination of the eyes of the children in the public schools of Phil., in Phil. Med. Times, 11 : 1880-1, 673-685. (13) S. D. Risley, School hygiene, being pp. 353-418 in vol. 2 of W. Norris and C. Oliver, System of diseases of the eye, Phil., 1897. (14) Schmidt-Rimpler, Die Schulkurzsichtigkeit u. ihre Bekampfung, Leipzig, 1890. (15) F. W. Smedley, Rept. dept. child-study and pedag. investigation, in 46th An. Rept. Brd. Educ, Chicago, 1899-1900. Also in U. S., 1902, i., 1095- 1115. (16) S. Snell, Eyesight and school life, Bristol, Eng., 1895. Pp. 70. (17) Stilling, Die Kurzsichtigkeit, ihre Entstehung und ,Bedeutung, in Abh. aus d. Gebiete d. padag. Psych, u. Physiol. 7: No. 3, Berlin, 1903. (18) J. van Biervliet, L'asymetrie sensorielle, in Bull. Acad. Royale dos Sciences, etc., de Belgiques, 34: Serie 3: 1897, 326-366. (19) J. van Biervliet, La mesure de I'intelligence, in Jour, de psj^ch., 1: 1904, 225-235. (20) G. T. Welch, Report on the examination of the eyes of the public school children of Passaic, N. J., April, 1896. Also summarized in Ref. 11. (21) A. G. Young, 7th An. Rept. State Brd. Health Maine. Augusta, 1892. Pp. 399. See especially 128-131. TEST 15 Balance and control of eye-muscles: Heterophoria. — Strictly speaking, the examination of the condition of tlie eye-muscles is a physiological test, but because this condition affects clearness of vision, it may be included here with other visual tests. Each eye-ball is supplied with six muscles. By their action in varying combination, the eye is moved freely in its bed, somewhat after the fashion of a ball-and-socket joint. Under normal condi- tions, the balance and the iimervation of these muscles are such that both eyes move in concert, i.e., the eye-movements are auto- matically coordinated for purposes of single vision and the lines of regard are restricted to movements where a common fixation point is possible. In some individuals, however, there exists more or less 'imbalance', or asymmetry of eye-movement, so that the two eyes fail to 'track,' as it were.* If we consider only the relations of the visual lines to one another and neglect paralytic affections of the muscles, we may distinguish TEST 15: HETEROPHORIA 143 between latent tendencies toward asymmetry, or heterophoria, and actual or manifest asymmetry, heterotropia or strabismus. Follow- ing the terminology of Stevens (1), we may define the possibilities as follows: orthophoria is a tending of the visual lines in parallelism when the determination is made for a point not less than 6 m. dis- tant; heterophoria is a tending of these lines in some other way under the same conditions. Heterophoria may appear (a) as esophoria, a tending of the lines inward or toward one another, (6) as exophoria, a tending of the lines outward or away from one another, (c) as hyperphoria, a tending of the right or of the left vis- ual .line in a direction above its fellow,^ or (d) as tendencies in obhque directions, viz : hyperesophoria and hyperexophoria. The tendencies just described are tendencies only, and are latent or concealed in the ordinary use of the eyes on account of the strong 'desire' for binocular vision. For their discovery, accordingly, it is necessary to resort to means for eliminating, so far as possible, this reflex or automatic correction of the latent tendency. The means most commonly employed, as illustrated in the tests that follow, is the establishment of disparate images on the two retinas. - When binocular vision is not habitually attained, the ten- dencies above described are no longer latent, but manifest, and heterotropia (strabismus or squint) is the result. Heterotropia may appear as esotropia, converging strabismus, or deviation of the visual lines inward; as exotropia, diverging strabismus, or deviation outward; as hypertropia, strabismus sursumvergens or deorsumvergens; or as compound deviations, termed by Stevens hyperesotropia and hyperexotropia. The most obvious immediate result of heterotropia is diplopia or double vision, a very annoying, but not usually a permanent symptom, because the person thus affected soon comes to neglect the bothersome image from the 'squinting' eye, and to take account only of that from the 'fixing' eye. In time, there results, usually, a limitation of the movements and of the retinal 1 The term does not imply that the line which is too high is at fault, but merely that it is higher. Hence, of course, the lack of necessity for any term to indicate that one line is lower than the other. 2 This assumption that voluntary attempts at fusion will be renounced if the two images are sufficiently disparate, is not entirely correct, and in so far, it is not always possible to make an accurate determination of heterophoria, particularly when slight, by means of the principle of diplopia. Slight heterophoria, moreover, is not to be regarded as abnormal. 144 SENSORY CAPACITY sensitivity of the squinting eye (exanopsic amblyopia), which is one of tho most interesting instances of the loss of function through disuse.^ Strabismus or heterophoria is functionally associated with ametropia; in particular, divergent displacement is more apt to be associated with myopia, and convergent displacement with hyperopia, probably as a consequence of the straining after clear vision under the hyperopic handicap. The chief instruments for the detection of muscular asymmetries are prisms of varying construction, the Macldox rod, and the sten- opaic lens. Stevens' phorometer is a device for holding and rotat- ing prisms with accuracy and under optimal conditions. The phoro-optometer is a combination of the phorometer with other instruments, such as the Maddox rod, Risley's prism, etc. Two tests are here detailed, the Maddox rod and the prism test. Both are convenient, portable, and inexpensive, but possess the dis- advantage common to all tests for heterophoria held close to the eye, viz : tliat *S does not always completely renounce the fusion-impulse. In the Maddox test, the so-called ' rod' transforms for one eye the flame of a candle into a long narrow streak of red light, while the other eye sees the candle flame naturally. Heterophoria is indicated by the lack of coincidence in these two images. The prism test, which is essentially an auxiliary test, consists in producing artificial displacement of images by means of the prisms, and measuring *S's ability to produce voluntary fusion of these displaced images. A. THE MADDOX ROD TEST Apparatus. — Maddox multiple red rod (Fig. 31). Trial frame (Fig. 29), Candle. [A set of trial prisms may be added.] Preliminaries. — Place the lighted candle on a level with aS"s eyes and 6 m. distance. Adjust the trial frame. Method. — (1) Let S close his left eye: place the Maddox rod in the frame before the right eye with the bars set horizontally. S should then perceive a long, narrow, vertical streak of red light. Then let S open his left eye and at once state whether the red .streak passes exactly through the candle flame. ' As this is particularly to be feared in tne case of children, whose eyes liave not reacned functional maturity, prompt medical attention to strabis- mus is highly imperative. TEST 15: HETEROPHORIA 145 (2) Turn the rod until the bars run vertically. S will see a hori- zontal red streak. Let him open his left eye and at once state whether the streak passes exactly through the candle flame. Results. — In the first test, the possible results are: (a) the line passes through the flame, orthophoria (Fig. 32), (6) the line passes to the right of the flame, esophoria or homonymous displace- ment (Fig. 33), (c) the line passes to the left of the flame, exophoria or crossed displacement (Fig. 34). FIG. ;31. MAUDOX MULTIPLE UOD. In the second test, the possible results are: {(i) tlie line passes through the flame, orthophoria (Fig. 35), (/>) the line passes below the flame, right hyperphoria (Fig. 36), (c) the line passes above the flame, left hyperphoria (Fig. 37). Notes. — Next to orthophoria, esophoria is the most common condition. Unequal vertical adjustment, hyperphoria, is not com- mon, save that an upward deviation of the squinting eye is almost always associated with high degrees of convergent strabismus. If the latent asymmetry is but slight, there may appear a more or less rapid corrective movement: S will then notice lack of coincidence of the line and the flame when the left eye is opened, but the two images soon fuse together. On the other hand, if the asymmetry is larger, E may determine its degree by placing prisms before the left eye and ascertaining by trial how strong a prism is needed to enable fusion to occur. If both horizontal and vertical imbalance is observed, the de- fect is hyperesophoria or hyperexophoria. This may be demon- strated, if desired, by placing the Maddox rod in an oblique position. KIGS. 82-34. ILLCSTIIATIXG ORTHOPHORIA, ESOPHORIA, AND EXOPHORIA, RESPECTIVELV. As rovoaloii by tho ?*I:uldo\- rod when used before the right eye for hori- zontnl deviation. (W- Solnveit.itz i\nd Randall). I'Ui. OO. MADDOX TEST FOR VERTICAL DEVIA- TION: ORTHOPHORIA. Hi. 3t.l. MADDOX TEST FOR VERTICAL DEVIA- TION; RIGHT HYPER- PHORIA. FIG. 37. MADDOX TEST FOR VERTICAL DEVIA- TION) LEFT HTPBR- PHORIA. TEST 15: HETEROPHORIA 147 Stevens's stenopaic lens (Fig. 38) may be substituted for the Maddox rod. A single determination then suffices for both hor- izontal and vertical displacement. In orthophoria, the candle Hame appears in the center of a dilTused disc of light; in hetero- VIG. 3S. STEVENS' STENOPAIC LENS. phoria, it is displaced to the right or left, above or below, or obli- quely, in a manner corresponding to that of the Maddox line-and- fiame test (Fig. 39). The stenopaic lens consists of a convex lens FIG. 39. HETEROPHORIA, AS REVEALED BY THE STEVENS LENS. of 13 D., covered, save for a very small opening in the center. The principle is again that of disparate images. H. TEST WITH PRISMS -Vpparatus. — Trial frame. One 2-degree, one 8-degree, and two 20-degree prisms, of the circular pattern for the trial frame.^ Candle. ' These prisms permit E to test S's ability to overcome the degrees of dis- placement that are considered standard for the three positions: their cost is about $4. For $10, however, a fairly complete set of prisms may be pur- chased, which will permit a more flexible test. 148 SENSORY CAPACITY Preliminaries. — Place the lighted candle on a level with »S's eyes and 6 m. distant. Adjust the trial frame. Method.— (1) To test abduction, or /S's ability to overcome a standard amount of displacement by rotating the eyes outward, place the 8-degree prism before one eye with the base in, i.e., to- ward the nose. (2) To test S's ability in adduction, or forcible convergence, place a 20-degree prism, with the base out, before each eije.^ (3) To test >S's ability in sursumduction (compensation for ver- tical displacement), place the 2-degree prism, with the ])ase either up or down, before one eye. Results. — With orthophoria, S should secure fusion under the conditions imposed, if not at the first trial, at least after a few trials on different days. Failure to accomplish this, or ability to over- come larger angular displacements than those cited, is indicative of heterophoria, or of other inequalities in the set of the eye-balls, e.g., declination. - Notes. — This test may, of course, be applied to cases in which either orthophoria or heterophoria is present. It may be of value in measuring »S's control of his eye-muscles, not only as a matter of optical hygiene, but also in conjunction with tests and experiments of a psychological nature, e.g., stereoscopy, binocular fusion, and visual space-perception in general. REFERENCES (1) G. T. Stevens, A treatise on the motor apparatus of the eyes, Phil., 1906. Pp. 496. (2) W. N. Suter, The refraction and motility of the eye, Phil., 1903. Pp. 390. TEST 16 Color-blindness.^ — This test continues the examination of the functional efficiency of the eye as a sense-organ. It has obvious practical import, as well as high theoretical significance in connec- tion with the theory of vision. ' The ability to overcome prisms by convergence is about 50 degrees, according to Stevens, but an exact standard cannot be stated. 2 For further details of this, and otner forms of prism test, consult Suter and Stevens. TEST 16: COLOR-BLINDNE8S 149 The retina of the normal eye is not equally sensitive to color stimuli in all portions. On the contrary, simple experiment^ will demonstrate the existence of three distinct zones : an inner, efficient zone, over which we see all colors ; a middle zone, over which we see only blues, yellows, blacks, whites, and grays, and an outer, to- tally color-blind zone, over which we see nothing but blacks, whites, and grays. Color-blindness may be regarded as an arrest of development, or in some cases as a pathological modification, of these normal zones. The description and classification of forms of color-blindness has been much obscured, if not even actually retarded, by confusion in nomenclature, especially seen in the adoption of terms based on unwarranted, preconceived theoretical notions as to the nature of color vision. If we follow the clue afforded by the distribution of color sen- sitivity on the normal retina, we should expect to encounter total color-blindness, or partial color-blindness of a red-green type. Blue-yellow blindness would seem theoretically unjustifiable, at least as an arrest of development. And in fact we do find that all recorded cases of so-called blue-yellow (or violet) blindness are of doubtful characteT, and at least pathological in nature. Total color-blindness is well authenticated, but rare. Here, too, the defect is pathological, and is accompanied by a reduction in visual acuity, by nystagmus, and other serious disorders of the visual organ. We are left, therefore, with red-green blindness as the typical and characteristic form of color deficiency. As revealed in tests, this defect consists in the inability correctly to distinguish certain tones, particularly unsaturated tones, of red and green. The colors thus affected are invariably in pairs, i.e., the individual who fails to perceive correctly a given green will also fail to perceive correctly its complementary red, and conversely. In actual vision, certain reds and certain greens appear neutral or gray, 2 while tones in which red and green are conjoined with 1 For details, consult Titchener [\2, Part I., Section 9). ' The green which appears as gray is a somewhat bluish green, lying between the 6-line and the F-line of the spectrum, and having a wave length between 500 /i /i and 490j«/i: the complementary red is a non-spectral pur- plish red. 150 SENSORY CAPACITY blue or yellow are seen as bluish or yellowish. The spectrum is thus divided into a long- waved yellow and a short-waved blue section. By dint of daily experience, however, the color-blind individual develops a capacity to recognize some reds and greens by means of secondary criteria, such as brightness (tint) and satur- ation (chroma), and familiarity with the application of color nomenclature (grass is green, cranberries red, etc.), so that the defect may exist unrecognized, either by himself or by his acquaint- ances, until chance compels the recognition or discrimination of tones to which these criteria can not be applied. Hence arises the necessity, in the administration of tests, of displaying a large number of colors of varied saturation and brightness, in order that, for any individual, some combination or series of combinations of colors may be found, in the recognition of which these criteria can not be used. Here, too, appears in part the explanation of the seeming individuality of the defect. Despite, and quite apart from, this variability, however, two very definite sub-types of red-green blinds may be distinguished, though the distinction is of theoretical rather than of practical significance : (a) those who locate the brightest part of the spec- trum, as do normal persons, in the yellow, and (5) those who locate the brightest part of the spectrum in the yellow-green region of the spectrum. For the latter, the entire blue end of the spectrum appears relatively brighter than to the normal eye. This type is relatively infrequent. Individuals who belong to the former type are called by von Kries (13) deuteranopes, and are erroneously called by many writers green-blinds: those who belcfhg to the lat- ter type are called by von Kries protanopes, and erroneously red- blinds. This confusing, if not erroneous, terminology is to be referred to the Helmholtz theory of color vision, wherein the three primary visual and retinal elements are assumed to be red, green, and vio- let. In theory, on this basis, it is evident that an eye might pos- sess all three, or but two, or but one, of these visual elements; that, in other words, an eye might possess trichromate,dich'romate or monochromate vision. Ordinary red-green blindness would, therefore, on this theory, be a form of dichromasy with the defi- ciency actually due to loss either of the red or of the green element. TEST 16: COLOR-BLINDNESS 151 This terminology ignores the fact that color-blindness invari- ably goes in pairs/ but the terminology persists and is current in nearly all popular discussions of the topic. This forms one of the most conspicuous instances of the imposition of theo- retical convictions upon the interpretation, and even upon the observation of facts. Not a few discussions of color-blindness that make pretence to scien- tific aacuracy exhibit this error, and, one may add, other errors of a more inexcusable sort. The reader may consult, for instances, a book by Abney (1), which embodies his Tyndall lectures of 1894, and a magazine article by Ayers (2). In Abney there will be found a colored frontispiece, taken from the Report of the British Association Committee on Color Vision in 1892, which purports to show the spectrum as seen by the color-blind. The spec- trum is shown in green and blue: what becomes of yellow is not explained. In Ayers's article there will be found some very pretty colored pictures of roses and Venetian scenes as observed by the color-blind, — pictures that are good examples of the illustrator's art, but absolutely false examples of color vision. Mrs. C. L. Franklin (6) has charitably applied the term "pseudo- scientific" to such writing. A more nearly correct representation of the spectrum seen by the color-blind is given by Thomson (11). Holmgren contrasted total color-blindness with partial color- blindness, and divided the latter into complete partial color-blind- ness and incomplete partial color-blindness (confusions with the green, but not with the red, test skein). This division has not been often used, but the term 'color-weakness' has been extensively employed in place of Holmgren's incomplete partial color-blind- ness, though not quite correctly, because this group, as determined by the Holmgren test, may embrace both von Kries's deuteranopes and the so-called color-weak. The most recent and authoritative study of color weakness has been conducted by Nagel (9). According to his investigations, the color-weak, with the rarest exceptions, are (in Helmholtzian terminology, after Konig) anomalous trichromates, i.e., they pos- sess all the elements of color vision, but exhibit certain anomalies, of which the following are most prominent: (a) a considerable reduction of sensitivity and discriminative capacity in the region of yellow-green and green, (6) inability to recognize colors, particu- larly red and green, when of reduced intensity, small area, or brief 1 There have been reported a few exceptions to this rule, which are diffi- cult of explanation by any theory of color-vision. 152 SENSORY CAPACITY exposure, (c) rapid retinal fatigue to colored stimuli, (d) slow recog- nition of color tones, amounting, for reds and greens, to 20-50 fold the normal time, (e) increased dependence upon brightness differ- ences, (/) very marked augmentation of simultaneous and succes- sive contrast.^ Nagel suggests a further subdivision of anomalous trichromasy into pro" tanomalous trichr6masy (a lessened excitability to red corresponding to protanopia) and deuteranomalous trichromasy (corresponding to deutera- nopia). But, as he points out, this distinction may be somewhat premature, since the facts are not yet clearly established, and since a case has recently been discovered in which anomalous trichromasy and dichromasy appear to coexist in the same eye. In theory, the color-weak are not to be identified with the color- blind. Their defect ranges all the way from forms which are to be distinguished from normal vision only by careful tests to forms which closely approximate true dichromasy or color-blindness. PV)r practical purposes, however, they must be identified with the color-blind, because they are incapable of making those color dis- criminations that the conditions of railway and marine service demand. It is, perhaps, true that some disasters may be traceable to this defect in color vision, which has escaped the detection of medical examiners who have used only the standard wool tests. Thus, in Germany, among 1778 members of railway regiments, all of whom had passed the wool test and many of whom had also passed Stilling's test, 13 dichromates and 31 anomalous trichromates of various types were discovered by the use of Nagel's test in the hands of military physicians. Baird, however, contends that statistics of railway accidents show no trace of this factor. Color-blindness may be binocular or monocular. The latter is rare, but naturally of great theoretical importance in determining the nature of color- blindness. Color-blindness is usually congenital, and then incurable. The common form, red-green blindness, is to be regarded as an arrest of development, a reduction from normal trichromasy, or reversion to a more primitive form of retina. All acquired cases, variously attributed to traumatism, neuritis, atrophy of the optic nejve, hysteria, excessive fatigue, over-indulgence in tobacco, are accompanied by lessened visual acuity, are pathological, and of relatively small concern to the theory of color-vision.^ ' For an excellent discussion of color weakness and the use of testa, the reader is advised to consult J. Rosmanit (10). ^ For illustrative cases, consult Collin and Nagel (4). TEST 16: COLOR-.RLINDNESS 153 Color-blindness seems to have been first noted An literature in 1684, but first described accurately by Dalton, the celebrated JRnglish chemist, in 1794. The first attempt at a systematic examination of a J.arge number of cases was made by Seebeck in Berlin in 1837 by the aid of colbred papers. The first systematic examination of railway employees dates from Ib'Jo, when a serious accident in Sweden led Holmgren of the University of Upsaik'.to devise his well-known wool test and to inducp officials to adopt it.' The chief devices and methods for testing color-blindness are Holmgren's, Galton's, Thomson's, Oliver's, and other assortments of colored worsteds, Stilling's pseudo-isochromatic charts, Nagel's card test, spectroscopic examination, various contrast tests, and the use of equations of mixed colors, particularly Nagel's equation- apparatus, and Bering's apparatus,^ which enables the examiner to adjust a color equation of transmitted light that shall appear to the color-blind as uniform gray. Nagel's or Hering's apparatus is to be recommended for careful psychological tests. In addition, numerous forms of color-blindness lantern (Williams', Frieden- berg's, Oliver's, etc.) have been devised for testing railroad and marine employees by simulating the conditions of night-signalling, and soiled signal-flags have been used for similar purposes, while Henmon has proposed a discrimination-time test. Two forms of test are here described: the familiar and widely used Holmgren wool test, adopted by the American Ophthalmo- logical Society, and Nagel's new card test, which has now been specially revised for the diagnosis of color-weakness and of other variant types of defect. Both of these tests are inexpensive, com- pact, and portable. They may be employed in conjunction with one another. A. THE HOLMGREN WOOL TEST Material. — Holmgren's worsteds. Sheet of light gray or white cardboard or a similarly colored cloth. Method.— (a) Full procedure. (1) Remove the three large test skeins, pale green, red and rose, Nos. 101, 102, 103. Scatter the remaining skeins over the cloth or paper in diffuse daylight ' For other details of the history of color-blindness as well as a discussion of methods, though not brought down to date, consult Jennings (8) and Thomson (II). = See Titchener (12. Pt. II, p. 7). 154 SENSC/ilY CAPACITY only,^ Hand to S tbie green test skein, No. 101, and direct him to pick from the ^able all those skeins that resemble the test skein i.e., all the tixiits and shades of that color. Explain that there are no two ^specimens alike, and that an exact match is not required. It "^vVill do no harm to illustrate the process by selecting two or three skeins for him, provided these are afterward mixed with the pile. To save time in explanation, other *S's may be allowed to watch this demonstration. (2) If hesitation appears, or if grays, browns or reds as well as greens are selected, continue the test by use of the rose skein. No. 102. The typical color-blind will then select some blues or pur- ples, or, less often, grays or greens. (3) Finally, the red test skein. No. 103, may be used, though many red-green blinds have little difficulty with this test on ac- count of the strong saturation of the test skein. In all three tests, preserve a careful record of the skeins selected by *^ . "What colors do you see here?" Normal »S's answer correctly, though they may call the brown, dark-yellow. (If hesitation appears over this brown, E may in- (luire whether the color might be yellow, or gray, or green, or brown.) Color- weak »S's call this card red and green (or greenish or olive). Color-blind »S"s answer as in (4). Supplementary Test A This is designed for testing all difficult or obscure cases (unin- telligent or dull ;S"s, color-weaks, etc.) that have not given clear results by the preceding tests. It should be administered to all »S"s that have shown hesitation in the preceding trials, particularly when green was under test. Method.^ — (6) Display the cards of Section A. Ask *S to point out all the cards upon which red or reddish spots are seen. (7) Ask similarly for all cards on which green spots are seen. (8) Ask what cards contain no green spots. (9) Display Cards Bl and B2; SLsk S to point out the red and green spots with a pencil. If Tests 6-9 are answered correctly, »S is certainly not color- blind. 158 SENSORY CAPACITY (10) If doubt remains as to whether S is color-weak, display in rapid succession several cards from Section A, and insert in the series the four cards of Section B. Ask S to state upon which cards green spots are found. If S mentions B2 or B3, he is color- weak, i.e., an anomalous trichromate. Supplementartj Test B This is designed for the special diagnosis of protanopia aiid deu- teranopia for scientific and statistical purposes. Method. — (11) Display Card B4, and direct *S's attention to the pairs of spots indicated by asterisks. Ask S which spot in each pair is darker. , If the green is selected, 8 is 'green-blind' or deuter- anopic, i.e., the relative brightness is approximately that of nor- mal vision: if red is selected, 8 is 'red-blind,' or protanopic. Red- anomalous and green- anomalous types may be similarly differen- tiated. Red-blinds wifl also find the red spots decidedly darker than the brown on Card B3. (12) Very rare cases of blue-yellow blindness may be recognized by inability to distinguish the blue-green from the yellow-green spots on Card A6 and by the designation of all the spots on B2 as red. Notes. — To use Nagel's test successfully, it is imperative that the following cautions be observed. (1) The directions for con- ducting the main tests must be strictly followed before attempting any supplementary or variant tests. (2) E must adopt a quiet, sympathetic manner, free from any sign of irritation or impatience, especially when dealing with slow or stupid >S's, or even with those who are plainly attempting deceit. (3) During the test, E must carefully avoid informing 8, whether directly or by suggestion, of any mistakes he may make. Discussion or criticism of *S's selections is out of place. For the sake of future tests, it would be desirable not to explain >S's errors to him even after the test. To avoid the effect of possible collusions between *S and previous *S's, Tests 4 and 5 may be preceded by similar questions applied to several cards taken from Section A. If 8 has decidedly low visual acuity, this must be corrected, at least approximately, by appropriate lenses, before the color-blind- ness test is begun. TEST 17: DISCRIMINATION OF BRIGHTNESS 159 REFERENCES (1) W. Abney, Color vision, London, 1895. (2) E. Ayers, Color-blindness, with special reference to art and artists. In the Century Mag., 73: 1907, 876-889. (3) J. W. Baird, The problems of color-blindness, in P. B., 5 : 1908, 294-300. (4) Collin and W. Nagel, Erworbene Tritanopie, in Zeits. f . Sinnesphysiol. 1 1 : 1906, 74-88. (5j Dalton, Extraordinary facts relating to the vision of colors, in Trans. of the Lit. and Philos. Soc. of Manchester, 1794. (6) Mrs..C. L. Franklin, Magazine Science, in Science, n.s. 25: 1907, 746. (7) V. Hennion, The detection of color-blindness, in Jour. Phil. Psy. and Sci. Method, 3: 1906, 341-4. (8) J. E. Jennings, Color-vision and color-blindness, Phil., 1896. Pp.111. (9) W. ]^iagel, (o) Fortgesetzte Untersuchungen zur Symptomatologie u. Diagnostik der angeborenon Storungen des Farbensinns, in Zeits. f. Sinnes- physiol., 41: 1906,239-282,319-337. (b) Zur Nomenclatur der Farbensinn- storungen, ibid., 42: 1907, 65. (Consult also other articles by the same writer in this periodical.) (10) J. Rosmanit, Zur Farbensinnpriifung im Eisenbahn- und Marine- dienste, Vienna and Leipzig, 1907. Pp. 59. (11) W. Thomson, Detection of color-blindness, being pp. 315-352 in Norris and Oliver, System of diseases of the eye, vol. ii., Phil., 1897. (12) E. B. Titchener, Experimental psychology, vol. i., N. Y., 1901. (13) J. von Kries, Ueber Farbensysteme, in Z. P., 13: 1897, 241-324. (Corrective note, p. 473.) TEST 17 Discrimination of brightness. — The object of this, test is to obtain an index, for comparative -purposes, of >S's ability to dis- tinguish very small differences in brightness, or more exactly, to determine the smallest difference in brightness that >S can distin- guish under simple experimental conditions. The present test onits consideration of chromatic stimuli, and is confined to the discrinnnation of brightness, first by the use of reflected, sec- ondly by the use of transmitted light. Visual discrimination has been studied in the laboratory by many competent investigators, e.g., Anient, Aubert, Bouguer, Helm- holtz, Frobes, Kraepelin, Masson, Merkel, Schirmer, Volkmann, and others. Tests of school children by Gilbert (3) and Spearman (6) have followed simpler methods. In the laboratory, use has been made of Masson's disk, both by daylight and artificial illumination, of the episkotister, oi gray 160 SENSORY CAPACITY papers and of shadows. Toulouse (9) proposes solutions of ani- line colors in glass receptacles. Gilbert used a scries of ten pieces of cloth soaked in a red dye of graded intensity. Investigations that are most comparable with the method here proposed are those of Anient (1), Frobes (2), and Spearman, all of whom made use of gray papers, and of Gilbert, who examined school children, though Avith chromatic stimuli. A. DISCRIMINATION OF GRAYS -REFLECTED LIGHT Apparatus. — Set of 10 test-cards, each composed of two gray strips, 13 X 40 mm., on a white background, 10 x 10 cm. Exposure frame, fitted with a card-holder which may be rotates 1 through FK;. 40. AIM'AHATUS FOKTHK DISCKIMINATiOX OF UllAVS. 180°, and with a black screen, through an opening (8x8 cm.) in which the test-cards may be viewed (Fig. 40). Light gray cloth. about 70 X 160 cm., for a background. Two supports, with angle- pieces, and a horizontal rod 70 cm. long. Headrest. The cards are numbered from to 0, corresponding to 10 different pair;? of .stimuli. Each card contains one strip of the lightest or standard gray, and one strip of comparison gray. Card No. represents no difference, or ob- jective equality; Card No. 1 represents the minimal objective difference; Card No. 9 the maximal objectiye difference and is easily supraliminal for the normal eye. Each card is numbered on the back in such a way that, when looking at the face of the card with the number up, the right strip is the darker; there is also a small black mark on the extreme edge of the card on the side of the darker strip. TEST 17: DISCRIMINATION OF BRIGHTNESS 161 The grays used on these cards have been specially prepared, under the author's direction, by S. L. Sheldon, photographer, of Ithaca, N. Y., and have been carefully standardized. Each set of grays is printed from the same negative, on which the original series was formed by graded serial exposures before a sheet of milk glass set in a north window. They will not fade or change their tone, unless brought into contact with chemical fumes or solutions; but, for additional protection, they should be kept under cover when not in use, and never be handled in bright sunlight. The tones, sizes, and spatial relations of the strips, cards, and background have been selected to eliminate errors that might arise from adaptation and contrast. The size of the strips is slightly smaller than that used by Ament (18 X 45 mm.) and slightly larger than that prescribed by Titchener (10 x 40 mm.) for the demonstration of Weber's law in brightnesses. Preliminaries. — Place a small table, say 65 x 90 cm., squarely before a window where good diffuse daylight may be secured (pre- ferably a north window with full clear exposure to the sky) ; leave just enough space between the front of the table and the window for two chairs for S and E. Spread the gray cloth over the top of the table, and stretch it up vertically at the back edge by means of the supports, so as to form a continuous background of gray, with the vertical back at least 65 cm. high and about 65 cm. dis- tant from )S's eyes. Place the exposure frame in the center of the table at the opti- mal reading distance (about 35 cm., unless S has uncorrected my- opia or hyperopia), and adjust it's height so that the top of the frame is on a level with S's eyes. Adjust the headrest so that S may sit erect, squarely before the exposure frame and close to the table-edge, with his back, of course, to the window. Keep the test-cards conveniently near, but out of *S's sight. E will find it most convenient to sit at S's right. Method. — (1) Spend 5 min. in giving S practise and familiar- ity with the test. For this purpose, begin with the large-numbered cards, and pass in general toward the smaller numbers, but with- out following any rigorous order. With each card, rotate the turn- table, so that the right strip is now the darker, now the lighter: follow an irregular order, and keep S always in ignorance of the actual location of the darker strip, and of the correctness of his judgments. In each trial, S must report his judgment in terms of the right-hand strip, saying either "darker," "lighter," or "equal.' (Any doubtful cases may be classed as equal.) 162 SENSORY CAPACITY When not observing a test-card, S should rest his eyes by directing them toward the gray background. He turns his eyes to the test-card at E's ''now," and should be asked to pass a judg- ment within 5 sec. It is not necessary to record results at this point, but from this practise work, S will attain a general familiar- ity with the test, and E will form a fair idea of >S's ' critical' region. (2) Proceed now, more formally and exactly, to determine ;S's difference limen by selecting a stimulus difference which has appeared in the preliminary series to be just noticeable for him. Give this stimulus-card 10 times, 5 times with the right strip darker, 5 times with the right strip lighter, but in chance order. ^ Inform »S that he will be shown the same card 10 times, but in different positions, of which he is to be ignorant. He may answer " 1 ighter," " darker," or "equal." (Equal j udgments may be classed aswrong.) S must not be informed during the series whether his judgments are right or wrong. If S gives 8 right answers in 10, the magni- tude of the brightness difference then in use affords the desired index. (3) Confirm the result by testing *S 10 times with a slightly larger difference, and 10 times with a slightly smaller difference. Unless the tests are disturbed by the operation of such factors as fatigue, loss of interest, practice, fluctuations of attention, etc., S will give 9 or 10 correct judgments in the former, and fewer than 8 in the latter test. Variations of Method. — Test the discriminative capacity of each eye separately, as well as in conjunction. Employ the trial frame of Test 14, placing the solid disk before the untested eye. Care must be taken to avoid visual fatigue under these conditions. This variation of method is of interest in connection with recent work on psychophysical asymmetry and the relations between right-handedness, right-eyedness, right-earedness, etc. (See, for example. Van Biervliet). If means are at hand to secure effective constant illumination by artificial light, this may be tried for comparison with daylight illumination. ^ It is convenient to prepare on small slips, beforehand, a number of chance orders, and to follow one of these with each set of 10 trials. TEST 17: DISCRIMINATION OF BRIGHTNESS 163 Treatment of Results. — For comparative purposes, S may be ranked in terms of the arbitrary units afforded by the card-num- bers. For more exact quantitative expression, however, the re- sults should be expressed in terms of the brightness-differences which correspond to the card-numbers. This correspondence must be worked out by E for the papers employed. Full direc- tions for a simple and sufficiently accurate photometric determina- tion of brightness values of gray papers will be found in Titchener (Ft. I., 35 ff.). B. DISCRIMINATION OF BRIGHTNESSES TRANSMITTED LIGHT Apparatus. — Headrest. Brightness discrimination test (Fig. 41). [This is a box fitted with a high power frosted tungsten lamp, the light of which is reflected from two independently adjustable FIG. 41. APPARATUS FOR BRIGHTNESS DISCRIMINATIONS. white screens upon two oblong, translucent windows, so placed in the face of the box as to give the same dimensions and spatial relations as obtained in the case of the gray strips.] Preliminaries. — The lamp cord is to be attached to a suitable current (106-110 volts, unless special lamps are ordered). E should endeavor to conduct the test in a dark or darkened room. If a brightly lighted room must be used, the effectiveness of the illumination of the ' windows' may be increased by erecting a pro- tecting screen of cardboard or cloth around them. Method. — It is important to arrange the headrest so that S is 164 SENSORY CAPACITY directly in front of the apparatus, with his eyes on a level with the windows in the box. The distance is less important; 50 cm. will be found convenient. The degree of illumination is controlled by two levers, which move the reflecting screens, and which are {>r()vided with scales upon the upper siu-face of the box. E first sets the right-hand lever at the point which affords the maximal illumination of the right-hand window, and records the scale- reading exactly. In accordance with the methods just outlined for the discrimination of grays, E now determines the just dis- criminable difference in the setting of the two levers (when either one of them is at the maximal point). The same precautions must, of course, be taken to reverse the standards in order to correct the space error.^ Variations of Method. — Substitute a 32 C. P. ruby lamp for the frosted lamp, and determine the discriminative capacity for reds of different brightness. Other colors may be employed sim- ilarly in this apparatus. Results. — (1) Trained observers, working under conditions similar to those prescribed, can discriminate a brightness differ- ence of r2o, though this fraction is appreciably altered by changes of technique or of experimental conditions. Untrained observers have less efficiency, about sV, according to Spearman. (2) By a different method and with colored stimuli, Gilbert found that discriminative ability increases very gradually up to the age of 17, but exhibits marked irregularities at the age of 7. (3) In discrimination of shades of color, one may conclude from studies by Nichols (5), Gilbert (3), and Thompson (7), that women and girls very slightly exceed men and boys in this capacity. Luckej'-, however, concluded that no sex differences could l)e demonstrated in color discrimination. (4) Individual »S's are apt to possess a constant space error, i.e. to tend to judge the gray on one side darker; in some cases this is the right, and in others the left, but it seems impossible to correlate this asjaiiinetry with right and left-iiandedness (Spearman). (5) Gilbert found no very decidinl correlation between visual ' Since the scales are identical and the entire instrument is symmetrical, a given setting of the lever will produce the same intensity of illumination for either window. TEST 17: DISCRIMINATION OP BRIGHTNESS 165 discrimination and intelligence. Spearman's experiments upon 24 village-school children give correlations between brightness discrim- ination and common sense, school cleverness, and general intel- ligence in the neighborhood of + 0. 50. In a series with high-class preparatorj^-school boys, however, school place and brightness discrimination gave only -\- 0.13 for the 'raw' correlation. Notes. — It is imperative that the conditions under which the gray strips are observed should be kept as constant as "possible. Backgrounds, cards, and holder provide these conditions in part, and relative brightness is not affected within a fairly wide range of illumination: nevertheless, it is desirable to work in the same place, at the same time of day, and under closely similar conditions of out- door illumination, e.g., between 9 a.m. and 3 p.m. on sunshiny days, and at a north window. To ensure evenness of illumination and absence of any shadows, E should test the setting of the ex- periment l^y placing Card No. in the holder and reversing its position several times. As this card represents objective equality, any constant judgment of difference may serve to indicate un- even conditions of illumination. In working with brightness differences, and indeed, with all small differences, E must be very careful to avoid suggestion of the direction of the difference to .S, and must keep a persistent watch for all kinds of secondary criteria of judgment. If desired, one could experimentally determine the degree of objective bright- ness difference that could be overcome. by suggestion. REFERENCES (1) W. Ament, Ueber das Verhaltniss der ebenmerklicheu zu den iiber- merklichen Unterschieden, etc., in Ph. S., 16: 1900, 135-196. (2) J. Frobes, Ein Beitrag iiber die scgenannten Vergleichungen iibei'merk- licher Empfindungsunterschieden, in Z. P., .36: 1904, 344. (3) J. A. Gilbert, (o) Researches on the mental and physical development of school children, in Yale S., 2: 1894, 40-100. (6) Researches upon school children and college students, in Iowa S., 1: 1897, 1-39. (4) G. Luckey, Comparative observations on the indirect color range of children, adults, and adults trained in color, in A. J. P., 6: 1895, 489-504. (5) L. Nichols, On the sensitiveness of the eye to colors of a low degree of saturation, in Amer. J. Science, 30: 1885, 37. (6) C. Spearman, General intelligence objectively determined and meas- ured, in A. J. P., 15: 1904, 201-293. 166 SENSORY CAPACITY (7) Helen B. Thompson, The mental traits of sex, Chicago, 1903. Pp. 188. (8) E. B. Titchener, Experimental psychology, vol ii., N. Y., 1905. (9) E. Toulouse, N. Vaschide, and H. Pi^ron, Technique de psych, experi- mentale, Paris, 1904, Pp. 330. (10) J. van Biervliet, L'asym^trie sensorielle, in Bull. Acad. Royale des Sciences, etc., de Belgique, 34: S6rie iii., 1897, 326-366. TEST 18 Auditory acuity.— This test, like that of visual acuity, is pri- marily conducted for hygienic and practical purposes, especially in the examination of the physical condition of school children, and constitutes the chief auditory test. We may distinguish be- tween simple acuity tests, which are designed merely to detect the existence of lessened aural efficiency and roughly to measure its degree, and more elaborate tests of a diagnostic character, which are for the most part, not used in group investigations, but are con- fined to the work of specialists in otology or in the psychology of audition.^ Among the latter tests may be mentioned that of bin- aural pitch-difference, integrity of the tonal scale, bone vs. air-con- duction, determination of relative and absolute deafness, diagnostic speech-tests, etc. These tests are designed to investigate the functional efficiency of the various auditory structures, such as the tympanum, ossicles, cochlea, auditory nerve, and to determine the cause of the defect in hearing and the possibility of alleviating it by medical treatment. In particular, it is important, from this point of view, to differentiate between defect in the middle, and defect in the internal ear, because in the former case partial deaf- ness may often be relieved, whereas in the latter medical treatment is ordinarily of no avail. The more common and widely employed tests for acuity fall naturally into two main groups, viz : speech tests and instrumen- tal tests. Speech tests may be conducted by either vocahzed or whispered speech, and by either the method of extreme range or the method of percentage of accuracy. For instrumental tests, use is most often made of the watch, of some form of audiometer, or ' A typical illustration is given by the interesting article of Bingham (4. ) TEST 18: AUDITORY ACUITY 167 acoumeter, or of a tuning fork. The relative merits of these tests deserve brief consideration. The primary advantage of speech tests is that they measure directly the most important function of the ear — the hearing of conversational speech, whereas all instrumental tests, because they test the perception of only a limited number of auditory qualities, fail to give unequivocal indication of auditory efficiency. One may hear the watch at a considerable distance and yet be relatively deaf for speech, or conversely. Speech tests should, accordingly, be given the preference where possible. The use of speech tests is, however, rendered difficult for several reasons. (1) Speech involves a great variety and complex combination of pitches of varied intensity and clang-color, and these elements are further varied by changes in accent, emphasis, and inflection. To render speech tests avail- able, therefore, most careful study must be made of the elements of spoken and whispered speech, and lists of test-words must be prepared in the light of this analysis.^ (2) Examiners can not guarantee uniformity of enunciation and intensity of stress in conducting the test, so that the results of different E's, or even of the same E at different times, are rendered difficult of comparison. This difficulty must be met both by preliminary practise and care on £'s part, and by ranking S's relatively, in terms of the empirically determined norms for each particular test. (3)The acoustic properties of the room in which the test is held markedly affect its outcome. The method of relative ranking, coupled with the method of constant range (described below), must be used to meet this difficulty. (4) Unavoidable noises are more likely to interfere with speech tests than with tests conducted at close range, e. g., by the audiometer. To offset this, tests must be conducted in as quiet a room as possible, and doubtful cases must be retested under the most favorable conditions that can be secured. Limits of space will usually preclude the use of vocalized or conversa- tional speech, but whispered speech may be used for tests in a range of from 17 to 40 m., or about one-third that of vocalized speech. Whispering re- duces the intensity of the vowels, whereas consonants are little changed. This test serves perfectly well for the practical examination of hearing and should be employed whenever feasible. In the use of both speech and instrumental tests it has been customary to employ the method of extreme range. A range line is chalked off on the floor of the room; S is seated at one end of this range, while E moves methodically forward and backward over it, until he determines the extreme limit of ' This work has been done by Wolf (20). English number-word lists have been prepared and tested by Andrews (1). Reference to these writers will make clear why disparate words form the best speech-test material, and why numl)ers form the best type of words. Politzer's objection to numbers (11, p. 117) is answered by Bezold (2, p. 5; 3, p. 206). 168 SENSORY CAPACITY auditory capacity for the voice or instrument. The careful experiments of Andrews (1) have revealed most serious errors in this method, due to variation in the reflection of sound waves when E changes his position in the room. In other words, intensity does not decrease in any uniform man- ner as distance increases, and consequently there is no constant relation be- tween length of range and goodness of hearing. It is probable that this error has entered into practically all tests conducted indoors by this method. The only way to avoid it is to use the method of constant range, i. e., to keep the distance of the range, and thus the acoustic conditions, constant, and to measure acuity in terms of the percentage of errors made in a series of tests at this selected range. Furthermore, since there are distinct differ- ences in the audibility of different syllables, it is imperative to employ only selected lists of test-words, and to employ a sufficient niunber of them to include all the desired vocal elements. V^ 1/ FIG. 42. politzer's acoumeter. From Titchener, Experimental Psychology. Reverting to instrumental tests, we find that the watch is most widely used. Its advantages are its convenience and accessibility and its relatively short range. Its disadvantages are that, like any instrument, it fails ade- quately to test the capacity to perceive speech, that its sounds give rise to a perception of rhythm, that its ticking is so familiar that illusions of hearing are frequent, and that watches vary in the intensity and quality of their ticks. ^ ^ Various forms of acoumeter have been invented to meet the deficiencies of the watch. The instrument invented by Politzer (Fig. 42) is best known ' Statements sometimes made in books on hygiene that, if the ticking of a watch can be heard at so-and-so many inches, the subject has normal hearing, are obviously absurd. The normal range for a watch-tick is given at 2.5 to 4.5 m., but one in the author's possession has a range of 12 m. See Bezold (3) and Sanford (11, p. 55). TEST 18: AUDITORY ACUITY 169 and is extensively employed in clinical work. Its range is commonly given at 15 m., but will vary one or two meters from this, as test conditions vary. This acoumeter yields a brief tone, 512 vibs., of constant intensity. Hear- ing is tested by the method of extreme range. For description, see Politzer (11, pp. 107-8). The upright is held between the thumb and forefinger, and the small hammer is dropped upon the steel cylinder from a constant height. A small disk attached to a pin, not shown in the cut, is usr-d for Ijono-conduction and othei' diagnostic tests. FIG. 43. seashore's audiometer. Lehmann's acoumeter (Fig. 44), which is here prescribed, has the advan- tage of allowing variation of intensity, and is thus adaptable to the space limits of the ordinary laboratory.^ The acoumeter described by Toulouse, Vaschide, and Pieron (15) sub- stitutes a drop of distilled water for the metallic ball, and an aluminum disk for the receiving plate. For descriction, see Hansen and Lehmann (7). 170 SENSORY CAPACITY Many attempts have been made to devise an instrument that will permit testing at the ear itself, in order the better to rule out disturbing noises. Commonly, these devices are electrical in nature, and are planned to utilize a telephone receiver in which clicks or tones are produced in a graded series of intensities. Typical of these instruments is Seashore's audiojneter (Fig. 43), which lias been fully described by its inventor (14), and which has been extensively employed by him (15) and by others, e. g., by the Child Study Bureau at Chicago (9, 16). The results that have been obtained by all instruments of the telephone type have apparently been rendered unreliable by physical errors (particularly by variations in the sensitivity of the telephone receiver), which are difficult to eliminate.^ Timing fo7-ks may be employed for acuity tests in accordance with the method first suggested by Von Conta (19), in which a 512 vibs. fork is struck and brought before the ear to be tested, and acuity determined by the length of time it can be heard. Blake's fork (Fig. 45) is devised especially for use by such a method, and may also be used for simple diagnostic tests as de- scribed in detail below. A. WHISPERED SPEECH TEST Materials. — Meter stick. Telegraph snapper, for signalling. A number of small rubber stoppers, for ear plugs. List of 100 test- numbers arranged in ten series, as in the following Table. table 25 Test-numbers for A uditory A cuity {Andrews ) II in IV ^ VI VII VIII IX X 6 84 19 90 25 14 8 52 73 24 29 69 53 7 13 31 93 35 41 95 42 17 34 39 46 9 27 64 16 62 87 92 28 62 7 65 60 81 95 49 53 33 97 84 54 98 15 6 57 80 94 26 45 21 70 76 74 19 38 71 70 50 72 56 91 40 36 78 20 16 35 75 60 75 83 23 49 40 89 3 18 48 3 43 68 52 82 23 64 58 61 1 86 18 92 87 51 97 2 37 Preliminaries. — Select, if possible, an oblong room of average proportions and a length of at least 30 m. By rough preliminarj^ ' For an extended discussion of the tecnnique, and particularly of the calibration of this type of apparatus, consult Bruner (5). TEST 18: AUDITORY ACUITY 171 tests, establish a range in this room such that not over 90 of 100 test- words can be correctl}^ heard by a normal ear. If space will not permit this range to be established otherwise, interpose screens between E and S, or place E and *S in adjoining rooms, off a straight hne. The range may thus be cut dowTi to from 18 to 20 m., or even less. Whatever may be the arrangement that affords a suitable range, make careful note of all acoustic conditions, e.g., distance of range from walls, dimensions of rooms, exact position of E and S, disposition of large pieces of furniture in the rooms, number of doors or windows opened or closed, time of day, etc. Be sure always to work under precisely these conditions. Method. — (1) Seat S at the end of the range selected, with his right ear toward E. Carefully close the left ear by means of a rub- ber stopper inserted into the meatus. This must completely close the ear, but must not be distressingly tight. E should practise on himself beforehand. If both ears are properly stopped, the ticking of a fairly loud clock can be heard only with difficulty when 1 or 2 m. away, and an ordinary watch cannot be heard when held close to the ears. The plug of cotton often used is entirely inadequate. Inserting the moistened finger-tip into the meatus makes an effective plug, but the position is uncomfortable, and *S is likely to move the finger and thus to cause distracting noises in the stopped ear. The same objection may be made to the practise of stopping the ear by pressing in the tragus, or by closing the meatus with the fleshy part of the ball of the thumb. Direct S to close or shield his eyes during the test, and on no account to watch ^'s lips. His mouth must hkewise be closed, since hearing is altered when the mouth is opened. Give S a short preliminary series without recording results, until satisfied that he understands the conditions of the test and feels at ease. (2) For the more formal test, pronounce the 100 words (or but 50, if time is limited) in groups of 10, in the following manner: at the conclusion of one expiration, snap the sounder once as a ready sig- nal for *S: at the conclusion of the next expiration, pronounce the test-number in whispered speech with the residual air in the lungs : then snap the sounder twice to indicate that the word has been pro- nounced, and let S either speak or write down the number that he has heard (using a dash if nothing is heard). Meantime, E inter- 172 SENSORY CAPACITY polates three complete breaths, then gives the warning signal, then the test-number after the fourth breath, and so on until 10 test- numbers are given. After a brief rest, try the second 10 numbers, and similarly, the third, fourth, etc. To avoid possible error, let S, if he is writing his report, begin a new column with each ten. (3) Stop »S',s right ear and test his left ear in the same manner. (4) T(»st *S"s binaural hearing by letting him /ace E, but with precaution that he does not secure visual aid from E's lips. This test is important, because binaural hearing may not be related to monaural range, and it is the type of hearing actually used in daily life. If time is very restricted, test this form of hearing alone. Treatment of Results. — *S's acuity is determined by the per- centage of test-numbers correctly heard, in relation to the normal percentage which has been ascertained by averaging the percent- ages of all >S"s tested under the same conditions. Thus, if the nor- mal percentage be 70, and *S's be 60, his acuity is 6/7; if S's be 80, his acuity is 8/7, i.e., supra-normal. Credit may be allowed for partially correct reports, e.g., 62 for 65: such allowance is specially recommended if 50 or fewer test-numbers are used. Notes. — The sounder is used to avoid changing E's vocal 'set.' If during the test, S becomes restless or inattentive, defer its completion. It is best to test but one *S at a time: two S's may, however, be placed back to back, for testing the right ear of the one and the left of the other, if precaution is taken to ensure against commmiication or disturbance. If the room is large, and preliminary tests warrant the belief that acoustic conditions will be identical, more *S's may be tested by seating them on an arc equidistant from E. A very crude group test may be carried out by placing all the chil- dren in a room at the limit of the ordinary classroom distance. Let them all close their eyes; then order them in a whisper to perform some unusual movement, such as to place the right forefinger on the palm of the left hand. Repeat with similar commands. Note any children who fail to respond, or who do so in evident imitation of others. Give these more careful tests later. Or take smaller groups of 10, similarly placed across the classroom. Provide each with a block of paper and pencil. Try a series of 10 whispered num- ber-words, and let each write them as heard. Test carefully any who make a single error. TEST 18: AUDITORY ACUITY 173 B. ACOUMETER TEST ApparatT'S. — Lehmann acoumeter, provided with glass, copper, and cardboard receiving plates (Fig. 44). Small level. Meter stick. Preliminaries. — Select a room with a straight range of at least 10 m. Seat S at one end of this range (10 m.) with his unused ear plugged, eyes and mouth closed, as in the speech test. Place the FIG. 44. lehmann's acoumetek. (Improved by Titchener.) aeoumeter upon a table at the other end of the range, and adjust it to a true horizontal plane by means of the test-level and levelling screw. Use the steel ball and the glass receiving plate, if only one plate can be tested. Method. — (1) Conduct a series of preliminary trials to familiar- ize S with the test conditions. Give a verbal 'now' about 2 sec. before the ball is dropped. Let S report 'yes' or 'no' after each trial. Introduce a few check tests, i.e., tests in which the 'now' is spoken, but the ball is not dropped. From these trials, E can determine approximately the 'critical' height for S. E manipulates the acoumeter with his right hand, using thumb and fore- finger to press the forceps, and thumb and middle-finger to turn the milled head of the screw on which the forceps rest. The shot is picked up with the left hand and placed in the rounded cavity in the tip of the forceps. Be- neath the instrument will be found a vertical millimeter scale. A fiat disk attached to the vertical screw just grazes this scale, while the disc is divided by cross lines into quadrants, so that variations in height of \ mm. or less may easily be secured. By setting the disk at zero and working upward, the height of fall may be noted without further reading of the vertical scale, 174 SENSORY CAPACITY simply in terms of quarter-turns of the screw. E must practise this manipu- lation until it becomes automatic; special care must be taken to make a clean release of the shot, without swerving from the point just over the center of the receiving plate. (2) Conduct a series of 10 trials from a constant height, so chosen as to lie probably just above ;S's hearing capacity. If this is done, he should then report correctly all ten trials. It is understood that several check tests are added to the series of 10. (3) Reduce the height of fall by a half turn, 0.5 mm., and give anofther similar series of 10 trials, with check tests added. If S answers correctly, reduce the height by another half turn, and con- tinue in this manner, until there is found a set of the screw at which S begins to make errors. It is well to confirm the result by taking series with a still smaller fall. Treatment op Results. — The last correctly given series may be taken as the measure of S's capacity. His acuity is measured, as in the previous test, by his relation to the norm or average result determined under the same conditions. In physical terms, S's capacity can be indicated by stating the conditions, distance from the instrument, material used for receiving plate, and indicating the physical measure of the noise produced, i.e., in mg.-mm. The ball should be weighed upon sensitive scales to secure this index. Typical Results. — Lehmann's results for average »S's at 10 m. are: with glass plate 540 mg.-mm.; with copper plate, 1110 mg.- mm.; with cardboard plate, 225 mg.-mm.^ Other experimenters report lower limens, e.g., 500 mg.-mm. with the copper plate. Results obtained by analogous methods are those of Schafhautl, who found that the noise made by the fall of a cork pellet weighing 1 mg. from a height of 1 mm., upon a glass plate, could be heard 91 mm., and of Norr, who found that with small iron balls dropped upon an iron plate, the normal limen for 50 cm. distance was 1500 mg.-mm. Notes. — All work with liminal stimuli is difficult, and this is especially true in audition. In the present test, S may imagine that 1 These figures are for a 'plate' 1x1 cm. in size and 1 mm. thick. It a larger plate is used, as in the regular equipment, the limen is altered. Thus, with the glass plate, Lehmann and Hansen report 432 mg.-mm., with the area increased to 1 x 2 cm., and 16 mg.-mm. when the area was increased to 1 x 3 cm. TEST 18: AUDITORY ACUITY 175 he hears the ball drop when it does not. Check tests are demanded for this reason. Occasionally an S may be found so 'imaginative' that the test can not be successfully used. The only remedy is to try to increase his caution by informing him of his errors. C. TUNING FORK METHOD Apparatus. — Blake's fork (Fig. 45). Stop-watch. [Rubber tube.] Method. — (1) Stand directly behind *S. Sound the fork by pressing the tips of its prongs together until they touch, and then suddenly releasing them. Hold it opposite, and close to the ear to be tested, with its plane of vibration vertical. Lift the prongs FIG. 45. Blake's fork. For acuity and diagnostic tests by the temporal or ' ringing-off ' method. away from the ear occasionally, so that S can state more easily when it actually ceases to be heard. Record the time by means of the stop-watch. Repeat 5 times with each ear, or until ac- cordant times are given. Compare this time with the norm previ- ously established empirically for the fork in use. (2) For a simple diagnostic test, place the stem of the sounding 176 SENSORY CAPACITY fork between *S's teeth. If both ears are normal, S will hear the tone with equal intensity in each ear, or the tone may be subjec- tively located in the middle of the head. If, however, one ear is defective, the tone may be heard either more loudly or less loudly in the affected ear. If the tone is heard more loudly in the ear which previous tests have shown to be defective, we may expect that the location of the defect on that side is in the middle or exter- nal ear, and that it may yield to proper medical treatment. If, on the contrary, the tone is heard better in the good ear, we may expect that the defect on the other side lies in the internal ear, or in more deeply seated portions of the auditory mechanism, and that it will probably not yield to treatment. General Results and Conclusions.— (1) It is difficult to state the prevalence of defective hearing in school children, because of the arbitrary and loose nature of the tests that have been used, and the varying standards that have been set for normality of hear- ing. Thus, in New York City, a recent report indicates only 1.1 per cent defective hearing; but here the test consisted merely in the use of a few whispered words in the school room at 20 feet dis- tance. The extensive Chicago tests,^ conducted with Seashore's audiometer upon 6729 children, show that, if a pupil is classed as defective when the audiometer record is four points or more below the norm (indicating a defect such that "he would be seriously inconvenienced in detecting sounds of medium intensity"), 1080, or 16 per cent, of the number were defective in one or both ears (6.64 per cent in both, and 9.55 per cent in one ear). A defect equivalent to three or more points of the audiometer scale was found in one ear in 26.3 per cent, and in both ears in 12.3 per cent of those examined . Other examinations are summarized l)y Young (21) as follows: "Sexton, of New York, examined 575 school children, of which 13 per cent were hard of hearing; W. von Reichard, testing with the watch 1055 pupils of the gym- nasium of Riga, found 22.2 per cent with defective hearing. Weil, of Stutt- gart, tested the sense of hearing in .5905 scholars of various kinds of schools, and found it below the normal in from 10 to 30 per cent of the children, ac- cording to their social condition. Moure, of Bordeaux, found 17 per cent; See Smedley (16), Macmillan (8, also summarized in 12). TEST 18: AUDITORY ACUITY 177 Gelle, of Paris, 22 to 25 per cent; Bezold, of Munich, 25.8 per cent of pupils with hardness of hearing." See also Chrisman (6) for a summary of inves- tigations prior to 1893. (2) With regard to the partially deaf, Macmillan and Bruner (9) conclude that, in theory, there exist varying degrees of deafness, "ranging all the way from slight and temporary impairment of hearing due to a cold, to the stage of absolute and permanent silence." An examination of the children attending the public day-schools for the deaf in Chicago, however, showed a somewhat unequal division of these pupils into 5 classes, based upon the somewhat conventional and immediately practical test of the status of the pupil in hearing in his schoolwork. Thus, of 174 cases, 55 were classed as totally deaf, 33 as ''practically deaf" (hearing only intense and continuous sounds), 53 as possessing "a degree of hear- ing power" (hearing loud sounds, but not understanding vocal speech),' 25 as ''deaf for ordinary school conditions" (hearing only words spoken loudly and close to the ear), and 8 as "hearing chil- dren temporarily needing special training in articulation." (3) Differences between the two ears. Seashore (15) found decided differences in the acuity of the two ears, differences that were unknown to the *S's that exhibited them. Preyer, Fechner, and Bezold have concluded that the left ear tends to be the more acute: Bruner (5), however, as well as Miss Nelson (10), state that in both sexes the right ear is the more acute. Van Biervliet (18) asserts that inequality of hearing of the two ears is a universal fact, that the disparity is such that the poorer ear has a capacity i less than the better ear, l)ut that the right ear is the better in right-handed, the left in left-handed >S's. For practical purposes in connection with schoolroom tests, the determination of this difference is significant only when the inferi- ority of one ear is marked; in such cases, pupils should be so seated in the classroom as to bring their 'good' ear toward the teacher. (4) Seashore's tests (15) indicate that acuity improves with age u]) to 12 years: this improvement is due partly to the development of the ear, but is slightly affected by the growth in ability to under- stand and to undertake the test. ' This class offers hope of improvement in hearing by means of mechan- ical devices for the intensification of speech. 178 SENSOEY CAPACITY (5) There are no noticeable sex differences, according to Sea- shore. Lombroso concludes that men's hearing is keener than women's. (6) Seashore says there is "no indication that the bright children hear better than the dull children: there may be cases of children who are dull or are accounted dull because they do not hear well, but such cases are not common enough to be revealed clearly by our method, although there may be some indication of them." Nearly every other investigator, however, has found evidence to show that defective hearing has a positively injurious effect upon school-standing. At Chicago (16), the examination of 5706 pupils with Seashore's audiometer showed that pupils below grade have, at every age, more cases of defect than those at and above grade, and that pupils in the school for backward and troublesome boys have a greater percentage of defect than boys of the same age in other schools. At Copenhagen, Schmiegelow found that, of 79 pupils regarded by the teachers as poorly endowed mentally, 65 per cent had defective hearing. Similarly, Gelle found 75 per cent of defect in the pupils classed as poorest. Permewan, at Liverpool, averaged the distance the watch could be heard by 203 pupils when divided into three groups^ bright, average, and diiU — and obtained the figures 51 inches, 47.3 inches, and 31.25 inches for these three groups, respectively. Shermunski, at St. Petersburg, by means of the whisper test, found that, among those of normal hearing, the ratio of good to poor students was 4.19 to 1; among those whose hearing was but 5 to 5 the normal, the ratio was 2.6 to 1 ; among those whose acuity was less than |, the ratio was 1.7 to 1. (7) Racial differences. Bruner's St. Louis Exposition tests (5) indicated that the whites were clearly superior in acuity to the other races tested. The Filipinos had the poorest hearing of those tested. (8) The simplest disturbance of hearing, if allowed to continue, may lead to serious results. In general, those who test the hearing of school children should note the condition of the ear, as well as test its capacity. Discharge of matter from the ear should be a cause for reference to medical attention. (9) Children who are partially deaf should be guided, in their adoption of occupation, to avoid callings for which they are unfitted, e.g., medicine, law, music, school-teaching, stenography, telephone or telegraph work, railroad, marine or military service. (10) The ears of school children should be tested carefully at least once in two years. TEST 18: AUDITORY ACUITY 179 (11) Defective hearing, like defective vision, may exist in seri- ous degree and yet pass unnoticed by child, teacher, parents, or friends. Of the 13 per cent found defective by Sexton, only 3 per cent were themselves aware of any defect, and only one of them was known to be deaf by his teachers. Notes.— In testing the hearing of those who are known to be partially deaf, e.g., such a group as is mentioned in (2) above, the ordinary speech or instrumental tests are not serviceable. Use may, however, be made of the telegraph snapper mentioned in the first method, or of Blake's fork in conjunction with a 'differential tube.' The noise of the snapper can be heard by the average ear at a distance of some 150 m. or more. In testing the partially deaf S, it should be held slightly behind his ear, out of direct view, and employed like the Politzer acoumeter, i.e., by asking S to give the number of 'clicks' (2 to 5) that he hears. In very young *S's, suffi- cient indication of hearing may be obtained by watching for reflex starts of the whole body, or of some part of it. The differential tube, as used by Macmillan and Bruner (9) con- sists of a tube of soft rubber 100 cm. long, and 4 mm. internal diam- eter, fitted with hard rubber tips for insertion, one into *S's, and one into E's ear. After S has been familiarized with the sound of the fork by hearing it with the base applied to his front teeth, his ears are tested one at a time by placing the stem of the sounding fork upon the tube. On account, presumably, of the longer duration of the sound, this device may be used to detect a grade of hearing even lower than that detected by the snapper. REFERENCES (1) B. R. Andrews, Auditory tests, in A. J. P., 15: 1904, 14-56, and 16' 1905, 302-326. (2) F. Bezold, Schuluntersuchungen liber d. kindliche Gehororgan, Wies- baden, 1885. (3) F. Bezold, Funktionelle Priifung des menschlichen Gehororgans, 1897. (4) W. Bingham, The role of the tympanic mechanism in audition, in P. R., 14: 1907, 229-243. (5) F. G. Bruner, The hearing of primitive peoples: an experimental study of the auditory acuity and the upper limit of hearing of whites, Indians, Filipinos, Ainu and African pigmies, N. Y., 1908. (Reprinted from the Archives of Psych., No. 11). See especially 55-108. 180 SENSORY CAPACI'l^Y (6) O. Chrisman, The hearing of children, in Pel. S., 2: 1893, 397-441. (7) F. Hansen and A. Lehmann, Ueber unwillkiirliches Fliistern. P'ine kritische u. exp. Untersuchung der sogenannten Gedankeniibertragung, in Ph. S., 11: 1895, 471-530, especially 494 ff. (8) D. Macmillan, Some results of hearing-tests of Chicago school children, in Medicine, April, 1902. (9) D. Macmillan (and F. G. Bruner), A special report of the Dept. of Child-study and Pedagogic Investigation on children attending the public day-school for the deaf in Chicago, Chicago, 1908. Pp. 88. (10) Mabel L. Nelson, The difference between men and women in the recognition of color and the perception of sound, in P. R., 12: 1905, 271-286, especially 280 ff . f(ll) A. Politzer, Lehrbuch d. Ohrenheilkunde, Stuttgart, 1893^ (12) Rept. com. on statistics of defective sight and hearing of public school children, in U. S., 1902, ii., 2143-2155. (13) E. C. Sanford, A course'in experimental psychology, Boston, 1895 and 1898. Pp. 449. (14) C. E. Seashore, An audiometer, in Iowa S., 2: 1899, 158-163. (15) C. E. Seashore, Hearing-ability and discriminative sensibility for pitch, in Iowa S., 2: 1899, 55-64. (16) F. W. Smedley, Rept. dept. child-study and pedagogic investigation, in 46th An. Rept. Brd. Educ. Chicago, 1899-1900. Also in U. S., 1902, i. 1095-1115. (17) E. Toulouse, N. Vaschide, and H. Pieron, Technique de psych. exp6r- imentale, Paris, 1904. Pp. 330. (18) J. van Biervliet, L'asymetrie sensorielle, in Bull. Acad. Royale des Sciences, etc. de Belgique, 34: Scrie 3, 1897, 326-366. (19) Von Conta, Ein neuer Hormesser, in Archiv f . Ohrenheilkunde, 1 : 1864, 107-111. (20) O. Wolf, Sprache und Ohr, Braunschweig, 1871. Also various articles in otological journals, as Arch. f. Augen u. Ohrenheilkunde, 3: Abth. 2, 35, and Abth. 1, 125; and Zeits. f. Ohrenheilkunde, vol. 20. (21) A. G. Young, School hygiene, in 7th An. Rept. State Brd. Health of Maine, Augusta, 1892. Pp. 399. TEST 19 Discrimination of pitch.— Like other forms of sensory discrimina- tion, this has been employed to discover the relation between such sensitivity and general intelligence. It has sometimes been employed to estimate musical ability, and it has, of course, general psychological interest. With adults and with children over eight or nine years of age, the test is relatively easy to administer. TEST 19: DISCRIMINATION OF PITCH 181 The available instruments are air-blown reeds or bottles, vibrat- ing strings, as in the sonometer, and tuning forks. Experience shows that a set of finely tuned reeds may be employed only when they have 'settled' to their permanent pitch, when they are blown by an absolutely constant source of air-supply, and when their tone-color is uniform. This renders the reed-box, such as the Appunn tonometer, out of the question, save for well supplied laboratories. Gilbert's tone-tester (1), which is constructed from an adjustable reed pitch-pipe, varies as much as five vibs. in pitch with variation in the force with which it is blown. Stern's blown-bottles or tone-variators (6) necessitate a constant air-supply, and even then do not yield pitches which correspond to the attached scales. The sonometer or monochord, employed by Wissler (12) and Spearman (5), is rather unwieldy, not always constant in pitch and tone-color, and compli- cated by certain mechanical difficulties, while its pitches must be computed at each test in order to guarantee correctness of the assigned vibration-rate values. The instrument is defended, however, by Spearman (5, 243f). Wissler's method of using the monochord, in accordance with which S was obliged to manipulate the instrument, is indefensible, and his results are worthless, as far as pitch discrimination is concerned. The use of tuning forks in which the pitch of the comparison fork is varied by weights or riders (for illustration, see Titchener, 9, i., 68) also necessitates the computation of the pitch differences by counting beats, and both this and the manipulation of the riders is not easy for inexperienced E's. For these reasons, a series of carefully tuned forks, selected for uniformity of tone color, one for each pitch desired, is here recommended, after the example of Seashore (3) in his examination of the pitch-discrimination of children. The present apparatus (Fig. 46) has been described fully by the author in conjunction with Titchener (10).' Apparatus. — Set of 11 forks — one standard fork of 426§ vibs., and 10 comparison forks, whose rates are 0.5, 1, 2, 3, 5, 8, 12, 17, 23, and 30 vibs. below the standard. A resonance box on which the forks may be mounted as they are used. (Fig. 46) . Soft-tipped hammer for striking the forks. Method. — (a) Preliminary trials. Seat >S with his back to the table at which E works, and about 1 m. distant. Instruct him as follows: ''When I say 'now,' close your eyes and listen carefully to ' Since the apparatus here prescribed was completed. Seashore has pro- posed to substitute for the resonance box, specially tuned resonators (per- haps two or three in number) of the Helmholtz type. By strongly re-enfor- cing the fundamental tone, these resonators might be particularly valuable in eliminating chance differences in tone color which are likely to appear in the small forks in use here. Otherwise, these differences must be eliminated by careful selection of the forks. 182 SENSORY CAPACITY the two tones you will hear; then tell me whether the second tone is higher or lower than the first. Say 'higher' if the second tone seems pitched above the first, 'lower' if below." S's view of the apparatus, or, if he be reliable, simply instruct him to close his eyes. Place the pendulum where its oscillations will be easily visible. Method. — The general plan of procedure is identical with that outlined in the introductory pages of this chapter, and recapitu- lated in Test 20. To apply this procedure to the test with the pressure-balance, after throwing the lever down to the right, place 196 SENSORY CAPACITY the weight marked B-lOO g. on the pin marked B, at the outer end of the beam. This weight is not removed during the experiment, and constitutes the standard stimulus. Place upon the second pin, marked A, the desired increment weight — any one, or any com- bination, of the weights marked A. To apply a pressure-stimulus, move the release-lever up to the left, so as to depress the support beneath the beam of the balance. To remove the stimulus, move the same lever to the right. The increment-weights are added to the standard stimulus when they rest upon the beam at A : they are subtracted from the total pressure, at will, by depressing the increment-weight lever, which lifts them from the beam and allows only the standard stimulus, 100 g. to be operative. Thus, for example, to test the discrimination of 150 g. and 100 g., movethe release-lever down to the right, place upon the pin A the 30 g. and the 20 g. weights, and upon the pin B the 100 g. weight. Give S a warning "now," and 2 sec. later move the release-lever smoothly up to the left: allow the pressure (150 g.) to be felt for 2 sec, then move the release-lever to the right: immediately depress the increment-weight lever, and apply the second stimulus (100 g.) in the same manner, while this lever is held down. *S judges, always in terms of the second stimulus, saying " heavier," " lighter," or " equal. "^ The exact duration of the stimuli and of the interval between them is of less importance than constancy from trial to trial. To avoid local fatigue, at least 15 sec. should elapse between suc- cessive judgments. E must practise the manipulation of the instrument, and take particular precaution to move the release lever so as to avoid either too sudden application, which produces a disturbing 'bump' and vibration, or too sIoav application, which also renders the judgment more difficult.^ *S must be specially instructed to receive the stimulus passively, so far as his finger is concerned. A downward movement of 1 Equal judgments, as previously explained, are to be avoided, if possible' in the final trials with a constant stimulus-difference. ^ The author has found that some E's, bj^ a curious kind of unconscious 'sympathy,' are inclined to apply the light pressure more gently than the heavier pressure. >S's judgments will almost certainly, even without his knowledge of it, betray the operation of this secondary criterion by exhibit- ing an unexpected and impossible delicacy of discrimination. TEST 21: DISCRIMINATION OF PRESSURE 197 reaction in the finger tip converts the test, virtually, into a test of discrimination of lifted weights.^ Treatment of Data. — The calculation of the difference limen and of the discriminative sensitivity is similar to that in the pre- ceding test, save, of course, that the standard is now 100, instead of 80 g. Results. — (1) Normal capacity. The discriminative sensitivity for cutaneous pressure depends so largely upon the type of the instrument (including especially the area of the pressure stimulus and the manner of application) that the norms obtained with other instruments can not be assumed to hold good for the present form of balance. Jastrow's results indicate a constant of approximately 1/15, which is nearly equal to that for lifted weights. Merkel similarly, reports 1/14 for his pressure balance, though Griffing believes that so fine a capacity as this must be attributed to the presence of a "muscular reaction of the finger." With a standard of 100 g. applied to the palm of the left hand, Miss Thompson found limens ranging from 4 to 20 g. (2) Dependence on the standard pressure. The limen for the same S, with the same instrument and method, is constant, at least for stimuli between 50 and 2000 g. (Weber's Law). (3) Dependence on the area of stimulation. According to Kiilpe (6, p. 160), the limen is 1/lQ to 1/20 with an area of contact 1 mm. in diameter, but rises to from 1/13 to 1/16 with an area of con- tact 7 mm. in diameter.- Griffing, however, declares that "the area of stimulation does not, on the whole, affect the accuracy of discrimination for weights, but individual peculiarities appear in the results obtained." (4) '^Practise seems to aid discrimination at places not accus- tomed to pressure stimuli" (Griffing). (5) There is no constant sex difference (Dehn, 2, and Thompson, 9). (6) Dependence on length of interval. Accuracy of discrimination does not vary appreciably when the interval between application ' If it were not for the awkwardness of the position, it would, perhaps, be better to insert the finger volar side uppermost, in order more certainly to ensure against this movement of reaction in unreliable »S"s. ^ The tips of the author's balance are 8 mm. in diameter. 198 SENSOEY CAPACITY of the t^vo stimuli is prolonged to 10 sec. (Griffing) or even to 30 sec. (Weber). (7) Dependence on place stimulated. For weights of 100 g. or more, there is no appreciable difference in the discrimination of pressure on the palm of the hand, back of the hand, and the volar side of the index finger, though the last is probably more sensitive for very light weights (Griffing). (8) Constant error. Most S's show a tendency, frequently a marked tendency, to overestimate the second weight, i. e., to judge it to be heavier (Griffing). (9) Direct judgments. The impression of the standard stimulus not infrequently becomes so clear that it is carried over from one trial to another, so that, at least with large stimulus-differences, S may pass judgment when the first pressure is applied. REFERENCES (1) T. L. Bolton and Donna L. Withey, On the relation of muscle sense to pressure sense, in Nebraska Univ. Studies, 7: No. 2, April, 1907, 175-195. (2) W. Dehn, Vergleichende Prufung iiber den Haut- und Geschmack- Sinn bei Mannern u. Frauen verschiedener Stande, Dorpat, 1894. (3) J. A. Gilbert, Researches upon school children and college students, in Iowa S., 1: 1897, 1-39. (4) H. Griffing, On sensations from pressure and impact, in P. R. M. S., i : 1895, No. 1, pp. 88. (Also Columbia Univ.Contr. to Phil. Psych, and Educ, iv.) Summarized in P. R., 2: 1895, 125-130. (5) J. Jastrow, On the pressure sense, in A. J. P., 3: 1890, 54-6. (6) O. Kxilpe, Outlines of psychology, Eng. tr., London, 1895. (7) J. Merkel, Die Abhangigkeit zwischen Reiz. u. Empfindung (II), in Ph.S., 5: 1889, 245. (8) E. C. Sanford, A course in experimental psychology, Boston, 1895 and 1898. Pp. 449. (9) Helen B. Thompson, The mental traits of sex, Chicago, 1903. Pp. 188. (10) E. B. Titchener, Experimental psychology, vol. ii., Quantitative ex- periments, N. Y., 1905. (11) G. M. Whipple, New instruments for testing discrimination of bright- ness and of pressure and sensitivity to pain, in J. E. P., 1: 1910, 101-106. TEST 22. Sensitivity to pain. — The determination of the threshold or hmen for pain has been conducted for the usual comparative purposes. TEST 22: SENSITIVITY TO PAIN 199 but it has had, in addition, a peculiar interest for some investigators, because it has been assumed that the Hmen varies in a character- istic manner with sociological status. For the determination of the pain limen, use has been made both of electrical and of mechanical stimulation. Electrical stimulation (induction-coil current) has been employed chiefly by the Italian criminologists: pressure stimulation (upon the temple, palm, or finger-tip) has been employed almost exclusively by more recent investigators, and to this form of test our attention will be chiefly confined. The value of the conclusions that have been so far reached from the use of pain tests is minimized by the difficulties, not always clearly realized, which appear in their administration. These dif- ficulties, like those of most functional tests, arise primarily from the presence of a number of variable factors. The most important of these factors are: (1) dependence of the limen upon S's ability to keep pain distinct both from strong pressure and from simple discomfort, (2) dependence upon the rate of application of the stim- ulus, including the length of time elapsing between successive appli- cations, (3) dependence upon the place of stimulation, (4) depend- ence upon the area of the stimulus, (5) dependence upon the general condition and attitude of »S, his ''good- will, "degree of fatigue, amount of practise, etc., (6) dependence upon individual constitutional differences in sensitivity, including sex, age, etc. This last is, of course, the particular dependence sought for ir the results; the others, then, constitute disturbing factors and must, accordingly, be eliminated or at least eva'uated. Proofs of these several dependences are given below in the discussion of results, but the first and second of them demand consideration here because they determine the choice of apparatus and of method. (1) The dependence upon »S's judgment as to what constitutes pain has been recognized by most investigators as the primary source of difficulty in this test. It is undoubtedly true that pain is a specific sensory quality, dis- tinct from pressure and distinct from unpleasantness; yet it appears equally true that it is often difficult, even for a practised >S, to disentangle from his experience the three elements — cutaneous pressure, cutaneous pain, and discomfort. Children, to say the least, are not always competent to make such a differentiation, at least with the method of procedure that has com- monly been followed. 200 SENSORY CAPACITY To meet this difficulty, some ^'s have instructed their S's to wait for the distinct appearance of pain; others have asked them to report the first appearance of discomfort — an affective experience that might, or might not, be accompanied by pain, and which can thus scarcely be regarded as a rational index of the real pain limen. Thus MacDonald says (13, 14) : "As soon as the subject feels the pressure to be in the least disagreeable, the amount of pressure is read from the scale .... The subject some- times hesitates to say just when the pressure becomes the least disagreeable, but this is part of the experiment (!). The idea is to approximate as near as possible to the threshold of pain." Griffing, however, thinks that there is little liability to error from this source: "It is very easy to tell," he says "when the pressure begins to be uncomfortable, and the 'imagination' does not .seem to be a disturbing factor. Indeed, the pain seems often to come with greater [great?] suddenness." We may wonder, then, why he says in the preceding paragraph: "The observers were asked to speak when the instrument began to hurt at all or to be uncomfortable ; for it was found that individuals differed as to what they called 'pain.' " Perusal of the literature makes it evident, as these instances illustrate, that some E's have been measuring a ''discomfort" limen, other a ''pain" limen: doubtless, in either case, some »S's reported ''discomfort," while others reported real ''pain." Yet again, it appears that schoolboys have sometimes understood tlje test to be a measure of their endurance of pain, and have manfully asserted : "It doesn't hurt yet" when the pain limen has long been exceeded. This difficulty of identification of the pain consciousness can not be wholly avoided, but it may be met, in part by giving S a clear account of the experi- ence he is to report as pain, in part, especially in doubtful cases, by repeat- ing the test, and in part by comparison of the results of a given S with the established norms for individuals of his age, sex, and type. (2) Unless the rate of application of the pressure is constant from test to test, there is introduced a serious variable error. Roughly speaking, the limen will be higher if the pressure is applied rapidly, lower if it is applied slowly. In the use of the ordinary type of pain-tester, e. g., of Cattell's (see 14, p. 1161) or of MacDonald's (12 and 14, pp. 1155-6) algometer, the rate of application of pressure is difficult to control; moreover, the rate has never been standardized, so that different investigators have followed different rates. ^ The chief merit of the pain-balance, or balance-algometer, as employed by Gilbert or as prescribed in the present test, consists in the guar- antee that it affords of a rate of pressure increase that shall be uniform from step to step during each trial and from trial to trial. Again, the time interval between successive trials must be standardized. If a given region, say the right temple, has been tested, its sensitivity is ^ In illustration, Griffing applied pressure with the Cattell algometer at the rate of 1.4 kg. per sec, whereas Gilbert applied pressure with his own instrument at the rate of 50 g. per sec, or only ^V as fast. TEST 22: SENSITIVITY TO PAIN 201 increased for some time thereafter, yet some investigators have not hesi- tated to make a series of 5 or 6 tests upon the same spot in immediate succes- sion. On the other hand, if a given region be subjected to daily tests for several weeks, its sensitivity becomes reduced by a process of inurement. It is clear that both of these sources of error must be avoided in the deter- mination of the limen. The remaining sources of error are fully illustrated below, and are intelligible without further discussion. When all precautions have been taken, it is probable, however, that the results of this test will )je more variable and less reliable than those of other psychophysi- cal tests. ^ Apparatus. — The author's (20) pressure-pain balance (Fig. 47). Cardboard screen with suitable supports. Seconds' pendulum (Fig, 21) or other device for time-control. A low table. [Tele- graph sounder (Fig. 26), battery and wire.] Preliminaries. — Arrange the instrument and screen as in Test 21. Place the pendulum within easy range of vision, or, since the time-relations are so important, convert it into an auditory signal by the use of the telegraph sounder and battery (preferably ad- justed to give a rather faint click). Method. — Seat S comfortably so that his hand lies upon the hand-rest with his finger-tip between the pressure-tips of the instru- ment, as described in Test 21. Give him the following inscruc- tions: "I want to measure your sensitiveness to pain. There is nothing for you to be afraid of, as I will stop the moment you tell me that you notice any pain. I shall add these weights, one after the other, on the end of this bar, and I want you simply to notice what you feel in your finger-tip. The pressure will grow stronger, bit by bit. It will, perhaps, feel uncomfortable after a time, but never mind that. Wait for the first moment when it really hurts, when you feel a stinging, sore feeling, or a real ache. Do you under- stand what I mean? I don't want to know how much pain you can 'stand' without crying out; I don't want to know when it is simply 1 The comments just given make it evident that MacDonald's algometer and Cattell's algometer are inadequate instruments. It follows that the results published by Griffing, Wissler, Swift, MacDonald, and Miss Carman are of doubtful value. Gilbert's results, though obtained by a better instru- ment, are more uneven than those of any other of the tests that he under- took and, on account of the slow rate of application that he used, are not directly comparable with those obtained by the method outlined below. 202 SENSORY CAPACITY uncomfortable; I want to know when yoii first notice what you would call actual pain." Throw the release-lever up to the left, so that the support beneath the balance-beam is permanently depressed: this makes possible a continuous, but cumulative pressure upon the finger. Apply one of the large brass disc-weights, marked B-200 g., ever}- 2 sec. These discs are placed, without jar, upon the pin marked i^, at the outer end of the beam. Continue application until S reports pain, then immediatelj' remove the pressure. The total weight on the beam measures >S's pain limen. Repeat the test with the left forefinger. Variations of Method. — (1) Test other fingers of both hands. (2) Apply the stimulus weights at a slower rate, say once in 4 sec, and note the effect upon the limen. (3) Substitute a series of pressures for the cumulative, continu- ous pressure, by using the release-lever as in Test 21, and applying the pressure for 1 sec. only, after each weight is added. Results. — (1) The short series of tests available with the pres- sure-pain balance thus far has shown that, with adults, the pain limen may be expected to lie between 1600 and 2400 g., when the method of cumulative pressure is employed. (2) Dependence on rate of application, (a) "The rate at which pressure is added influences greatly the amount that is required to produce pain" (Gilbert). (b) Immediate or close repetition of stimulation causes in- creased sensitivity, but continued practise for several weeks appears to reduce it (Griffing, 8) . (3) Dependence on nature of stinndus. Sensitivity to pain pro- duced by electrical stimulation bears no noticeable relation to that produced in the same person by pressure stimulation (Griffing, 7). (4) Dependence on the area, of the stinndus. ' ' The pain threshold increases with the area of stimulation in an approximately logarith- mic proportion " (GriflSng, 8). Thus, when Cattell's algometer was applied to the palm of the hand, areas of 10, 30, 90 and 270 sq. mm. were correlated with limens of 1.4, 2.8, 4.4, and 6.6 kg., respec- tively.^ ' The area of the pressure tip in the author's balance is approximately 50 sq. nam. TEST 22: SENSITIVITY TO PAIN 203 (5) Dependence on region. The regions of the body most sensi- tive to pain (from pressure stimulation) are those over the frontal and temporal bones, while the heel, the back, and the muscular regions of the leg and the hand are distinctly less sensitive. Illus- trative limens o])tained by Griffing are : KG. KG. Top of head =1.8 Right thigh, ventral surface = 4.3 P'orehead =1.3 Left hand, volar side =6.2 Right temple ± 1.0 Right heel, plantar side = 7.0 Left temple =1.3 Back =8.0 These differences seem "to depend largely upon the thickness of the skin and the extent of the subcutaneous tissues." The left side of the body is in general somewhat more sensitive than the right. (6) Dependence on sex. Gilbert (6), MacDonald, Dehn (4), Carman (2), Swift (16), and Wissler (21) agree that women are more sensitive to pain than are men. Thompson (17) agrees to this generalization, but adds that there are more men than women with very low thresholds, i.e., that there is greater variability in men. Ottolenghi (15) and Lombroso (10), on the other hand, state that (with electrical stimulation) women are markedly less sensitive than men. The latter authority believes that this result is confirmed by the experi . ence of surgeons, who find that women possess greater endurance of pain the popular opinion that women are more sensitive to pain is due, in his view- to the greater tendency of women to express feelings of pain by tears or other- wise; he also believes that their greater longevity may be due partly to their inferior susceptibility to pain. Typical results are those of Wissler and of Gilbert: the former publishes the following averages (Cattell algometer on the ball of the right thumb); college men, 5.9 kg.; college women, 2.4 kg. • Gilbert finds that the average difference between boys and girls is about 400 g., and that this difference increases with age, until at 18 or 19 it becomes over 1 kg. (See Table 28.) (7) Dependence on age. Sensitivity to pain, in general, de- creases with age up to 18 or 19 years, and is thenceforth approxi- mately stationary, but Carman and MacDonald both find irregu- larities near the period of puberty, and Wissler finds Seniors mofe 204 SENSORY CAPACITY sensitive than Freshmen (as a class) . It is probable that the gen- eral result is disturbed by a tendency on the part of younger chil- dren to shrink from the test and to report discomfort rather than pain. Gilbert's results are embodied in Table 28.^ TABLE 28 Faitt Limen, in ky., for about 60 Boys and 50 Girls oj' each Age (Gilbert) 10 11 12 13 I 14 I 15 I 16 17 j 18 I ! i I ^ ' : I I I Boys 1.261.381.701.691.672.072.002.05:2.132.352.702.752.85 Girls 1.150.931. 181. 36il. 451. 561. 461. 701. 821. 77|1. 85 1.931.80 Average 1.211.161.441.531.561.821.731.5 1.982.062.282.342.33 (8) The range of individual difference is large in the pain test. Gilbert, for example, found that the mean variation for his groups of children ranged from 330 to 820 g. (9) Dependence on fatigue. Swift (16) concludes that fatigue increases sensitivity to pain, especially in the case of younger pupils ai d of girls, because it lowers the tone and increases the irritability of the whole system.^ Essentially similar results are reported by Vannod (18) and by Vaschide (19). The former employed an instrument ('algesiometer') analogous to v. Frey's ' hair-esthesi- ometer,' and found that the pressure needed to produce pain fell from 45 g. at 8 a.m. to 39 g. at 10 a.m., and to 29 g. at 12 m., under the influence of school work. The latter concluded that pain tests warranted him in stating (1) that mathematics and ancient lan- guages possess an especially high fatigue-value, (2) that written exercises in the form of tests produce intense intellectual fatigue, (3) that afternoon is much more fatiguing than forenoon instruc- tion, and (4) that a forenoon spent outside of the school permits ' [t should be remembered, again, that Gilbert used a very slow rate of application, so that his results may not be comparable with those obtained hy the methods we have prescribed. ^ This conclusion is based upon tests before and after a 10 days' vacation in which the "physical condition" was determined by a dynamometer — a method already shown to be of doubtful value (Tests 6 and 9). There is no evidence to indicate that check-tests were made to determine the range of variations that might have appeared under constant work-conditions. TEST 22: SENSITIVITY TO PAIN 205 a return, in most cases, to normal sensitivity. On the other hand, Binet (1), who used an adaptation of Blocq's sphygmometer, has come to diametrically opposite conclusions, and asserts that the effect of fatigue is to reduce, not to heighten, pain sensitivity. (10) Dependence on mental ability. The relation of the pain limen to mental abilitj is not clear. Carman found that bright hoys (teacher's estimate) were more sensitive than dull boys: Swift confirmed this by contrasting the best with the poorest fifth of a class, and attributes the result to the more delicate nervoiis organi- zation of bright children. MacDonald, however, says "there is no I ecessary relation between intellectual development and pain sen- sitiveness." "Obtuseness to pain seems to be due more to hardi- hood in early life." A curious and somewhat dubious correlation unearthed by Miss Carman is that boys and girls who are especially dull in mathemat- ics are more sensitive on the right than on the left temple. (11) Dependence on sociological condition, (a) Similarly un- convincing is the series of conclusions in which MacDonald (11, 13, 1 4) summarizes his correlations between pain sensitivity and socio- logical condition ; e.g., girls in private schools, who are generally of wealthy parents, are more sensitive than girls in public schools; university women are more sensitive than washerwomen, but less sensitive than business women; self-educated women are more sen- sitive than business or university women (owing, perhaps, to over- taxing their nervous systems in the unequal struggle for an educa- tion); the non-laboring classes are more sensitive than the laboring classes, etc.^ (b) The study of the pain sensitivity of the criminal is a specific sociological problem that has attracted much attention since the concept of the 'criminal type,' or of the 'instinctive criminal,' was introduced by Lombroso and his school. It has been generally stated that the typical criminal is distinctly less sensitive to pain than the average normal man, and it has frequently been added that the norrn^l insensibility of the criminal is to be largely attributed to ' The measurements from which these conclusions are drawn were made 1)}^ different investigators, by an unreliable method, and have been assem- bled apparently by mere comparison of averages and with no attempt to determine the limit of error; they might, or might not, be confirmed by more exact methods. 206 SENSORY CAPACITY this bodily insensibility. These statements are based upon certain experimental tests and upon common observations of the hardi- hood and general obtusity of the 'typical criminal.' Nevertheless, recent pain measurements indicate that the generalization is too sweeping, and that numerous exceptions occur. It may even be doubted whether the existence of a distinct criminal type has been satisfactorily established. A general summary of the pain sensitivity of criminals is given by Ellis (5, Section 8). The inadequacy of the algometer test as applied to crim- inals is discussed briefly by Miss Kellor. A typical exception to the gen- eral belief is found in Dawson's conclusion (3) that normal children are less sensitive to pain than delinquent children, probably because many of the delinquents were of neurotic type. (12) Miscellaneous correlations reported by Miss Carman are: boys with light hair and eyes are less sensitive than boys with dark hair and eyes. First-born are more sensitive than second-born boys, and the latter than later-born brothers : the same is true of girls, save on the right temple (!). These conclusions are subject to obvious criticism. REFERENCES (1) A. Binet, Recherches sur la fatigue intellectuelle scolaire et la mesure qui pent en etre faite au moyen de I'esthesiometre, in A. P., 11: 1905, 1-37, especially 32 ff. (2) Ada Carman, Pain and strength measurements of 1507 school children in Saginaw, Michigan, in A. J. P., 10: 1899, 392-8. (3) G. E. Dawson, A study in youthful degeneracy, in Pd. S., 4: 1896, 221- 258. (4) W. Dehn, Vergleichende Prlifung liber den Haut- und Geschmack- Sinn bei Mannern u. Frauen verschiedener Stande, Dorpat, 1894. (5) H. Ellis, The criminal, 3ded., London, 1907. (6) J. A. Gilbert, Researches upon school children and college students, in lowaS., 1: 1897, 1-39. (7) H. Griffing, On individual sensitivity to pain, in P. R., 3: 1896, 412-5. (8) H. Griffing, On sensations from pressure and impact, in P. R. M. S., 1: 1895, No. 1. Pp. 88 (also Columbia Univ. Contr. to Phil., Psych, and Educ, 4). Summarized in P. R., 2: 1895, 125-130. (9) Frances Kellor, Experimental sociology, N. Y., 1901. Pp. 316. (10) C. Lombroso, The sensibility of women, brief report in Mind, n.s. 1 : 1892, 582. TEST 23: ESTHESIOMETRIC INDEX 207 (11) A. MacDonald, Sensibility to pain by pressure in the hands of indi- viduals of different classes, sexes and nationalities, in A. J. P., 6: 1895, 621-2. (12) A. MacDonald, A temporal algometer, in P. R. 5: 1898, 408-9. (13) A. MacDonald, Further measurements of pain, in P. R., 6: 1899, 168-9. (14) A. MacDonald, Experimental study of children, etc., reprint chs. 21 and 25 of U. S., 1897-8, Washington, 1899. (15) S. Ottolenghi, La sensibilite de la femme, in Revue scient., Ser. 4, vols. 5: 395, and 6: 698. (16) E. Swift, Sensibility to pain, in A. J. P., 11: 1900, 312-7. (17) Helen B. Thompson, The mental traits of sex, Chicago, 1903. Pp. 188. (18) Th. Vannod, La fatigue intellectuelle et son influence sur la sensi- hilito cutance, in Rev. med. de la Suisse Romande, 27: 1896, 21. (19) N. Vaschide, Les recherches experimentelles sur la fatigue intellec- tuelle, in Revue de philos., 5: 1905, 428. (20) G. M. Whipple, New instruments for testing discrimination of briglit- ness and of pressure and sensitivity to pain, in J. E. P., 1 : 1910, 101-106. (21) C. Wissler, The correlation of mental and physical tests, in P. R. M. S., 3: No. 6, 1901. Pp. 62. TEST 23 Discrimination of dual cutaneous impressions : Esthesiometric index. — As long ago as 1834, E. H. Weber, a German physiologist, observed (50) that, if two punctiform pressures are applied simul- taneously to adjacent points on the skin, a single impression results, whereas, if the pressure points are applied at gradually increased distances, an extent can be discovered which is just sufficient to yield a perception of two points. Weber explored many regions of the skin and published extended tables of measurements of this dis- tance, which has since become known variously as the "limen for duality," or "doubleness," as the "esthesiometric index," the "space threshold," or even, less exactly, as the "index of delicacy of touch." On account of Weber's explanation of the phenomenon, which was in terms of the supposedly quasi-circular distribution of the end-organs of the sensory nerves, the experiment is often referred to as thetestof "sensory circles." Onaccountof thetypeof instru- ment employed, it is sometimes termed the "compass test." Since Weber's time the experiment has become a classic in psy- chology. Seemingly simple and definite, more careful examination 208 SENSORY CAPACITY has revealed the fact that the determination of the esthesiometric index is in reahty unusually difficult, and that the factors which underlie the observer's judgment are surprisirgly varied and subtle. For differential psychology, the chief interest in the test is found in its use by criminologists to measure "general sensibility," and by several German investigators to measure the degree of fatigue of school children. Physicians, also, have employed it for diagnostic purposes, particularly in connection with pathological conditions of the spinal cord, and it has found special favor in the psychological laboratory, both for its intrinsic interest and for the illustration of various psychophysical methods. As in the case of other tests, the chief difficulty in the use of the esthesiometric test lies in the presence of numerous sources of error, which must be fully recognized and controlled if valid results are to be secured. In general, it may be said that the esthesiometric limen will depend upon (1) the instrument employed, (2) the region of the body tested, (3) the method of procedure, including the nature of the instructions, (4) the care with which E applies the stimulus and the actual pressure employed, (5) S's degree of fatigue, (6) *S's degree of practise, (7) S's abihty to attend to the impressions and to make accurate reports, especially in the 'critical region,' and (8) upon a number of other factors, such as >S's sex, age, the condition of the circulation in the region tested, etc. The manner in which these factors affect the index will be discussed below. The instrument employed may be extremely simple, e.g., the set of needles thrust through bits of cardboard, used by Binet in his earlier tests (2), or it may be very complicated and elaborate. In general, the development of the esthesiometer since Weber's time has been in the direction of greater complexity and delicacy, with a view of affording more adequate control of the separation of the points, of the simul- taneity of their application, and of the degree of pressure exerted. It is doubtful whether much of this elaboration is needful: objective equaliza- tion of the pressure does not insure subjective equalization, and a careful E is better able to apply the points simultaneously if he works with a rela- tively simple instrument. The instrument selected, an improved form of Jastrow's esthesiometer, possesses all the requisite features. For other models, consult Blazek (7), Binet (3, 4), and Washburn (48). The models of Ebbinghaus and v. Frey are figured in Zimmermann's catalog. Spearman's instrument is described in Sommer (38) and pictured in use in Schulze (33, p. 67). TEST 23: ESTHESIOMETRIC INDEX 209 Apparatus. — Jastrow's improved esthesiometer (Fig. 48). Cardboard screen and supports. Pillow or folded towel. Method. — (a) Preliminary practise. Seat S comfortably with his right ^^orearm laid horizontally, volar side uppermost, upon a small pillow or folded towel, with the clothing arranged to expose the forearm from elbow to wrist, without impeding the circulation at the elbow. IMPROVED ESTHESIOMETER. Arrange the screen to cut off from S the view of his forearm and of the instrument. Devote from 2 to 5 min. to a preliminary practise series in order to familiarize S with the test, particularly with the perception of one and of two points. Id struct all 20: 1909, 473-529) has shown that the theoretical conclusion of p. 221 is justified in actual experimentation, though his methods are too elaborate to serve the purpose of simple functional tests. - The name Tachisloskop was first employed by Volkmann (27) TEST 24: RANGE OF VISUAL ATTENTION 223 As Dodge has remarked, "no psychological instrument is subject to greater modification in response to special experimental conditions than exposure apparatus," and it may be added that in no other experiment are the results more evidently conditioned by the form of apparatus and type of procedure employed. For these reasons, it is advisable to review briefly the development of the instrument in the light of the experimental requirements. Wundt has formulated the essentials of a good tachistoscope as follows (29): (1) The exposure must be short enough to preclude eye-movements. (2) The arrangement of the fixation-mark and of the stimulus must be such that all the constituents of the exposed object can be seen with at least approximately equal distinctness, i.e., the exposure-field must coincide with the ocular field of direct vision. (3) The exposure of all parts of the field should be simultaneous, or so nearly so that there shall be no noticeable time-differences in the illumina- tion of the various regions. (4) Retinal adaptation must be favorable, and sudden transitions from dark to light must be avoided. (5) Persistent after-images must be avoided. (6) The duration of the retinal excitation must be limited enough to pre- clude roving of attention over the exposure-field. (7) A ready-signal must be given at an appropriate time before the ex- posure. Further requirements given by Dodge^ are as follows: (8) The relative illumination of the pre-exposure, exposure, and post- exposure fields should be capable of experimental modification. (9) The exposure should be noiseless and free from distraction. (10) It should be possible to arrange for monocular or binocular observa- tion. The earliest exposure apparatus, as used by Dove, Zdllner, and Helm- holtz, and described by Helmholtz (14, p. 710), made use of the electric spark for illumination. Here, although the duration of the spark is but 0.00004 sec, the retinal excitation of the darkness-adapted eye is longer than is obtained from an exposure of the same object for 0.01 sec. in day- light. The use of a rotating disc for tachistoscopic experimentation is illus- trated by early devices described by Helmholtz (p. 514), and more fully by Exner (10), and employed, though for somewhat different purposes, by Exner and by Baxt (1). An elaboration of the rotating disc apparatus was used by Goldscheider and Miiller (11), and the principle is embodied in a recent rotating mirror tachistoscope by Wirth (29). Here the fixation-field ' See also Erdmann and Dodge (8, pp. 94 ff.) for special requirements for the investigation of reading. 224 ATTENTION AND PERCEPTION is a virtual image obtained from the revolving mirror and the exposure-field is a real image viewed through an adjustable slit in the circular mirror, which is revolved by a constant speed motor. A very simple adaptation of the rotating disc apparatus is illustrated in the weight-driven sectors used by Quantz (24) and another by the apparatus prescribed by Titchener (26, ii., p. 201), while the instrument here employed for the range test may be regarded as a modification of this form. A well-known exposure apparatus is the fall or gravity tachistoscope employed by Cattell (4), Huey (17, 18) and others. In this instrument, exposure is accomplished by the drop of a guillotine-like screen, perforated with a horizontal slit of variable width, before a card bearing the exposure- field. To render the exposure more nearly simultaneous, to afford a wider range of exposure-times, and to secure more accurate fixation, this instru- ment is now made with taller columns, with an Atwood-machine attachment and with a new form of fixation-field, as illustrated by Zeitler (31, p. 381). Monocular vision with the aid of a reading telescope is also introduced. A large demonstration form of the fall-tachistoscope is figured by Wundt. A pendulum exposure apparatus is shown in Wundt (p. 400), also in Zim- merman's catalog (1897, p. 8). The pendulum has also formed a constituent part of other tachistoscopes, e.g., as a device for interrupting a flash of light, as in Sanford's dark-box for testing legibility of various forms of type (25), or in Dodge's tachistoscope. The horizontal exposure used by Volkmann is seen again in Hylan's simple rubber-band and shutter type of tachistoscope (19, pp. 395 and 509). Another simple device is the superposition of a photographic shutter 4 cm. in diameter over the test-material (Binet, 3). The use of a transparent mirror gives the cue to the construction of Dodge's latest instrument (6, 7, also described in Judd, 206, p. 234), which embodies all the requirements above cited. This instrument is to be recom- mended for those who wish to do careful work under experimentally varied conditions. Its only disadvantage is the necessity of employing a high-power illuminant, such as the arc light or 150 C.P. stereopticon incandescent lamp. The instrument costs about $20 without lamp, discs, exposure pendulum and other accessories. The apparatus used by Erdmann and Dodge (10, 98 ff.) andbyBecher (2) is of the camera obscura type, in which ati image of the exposure material is cast by a beam of light upon a ground glass field, and the illuminating beam interrupted by suitable devices. The controversies concerning the interpretation of the results secured by these various instruments have to deal, so far as technique is concerned, with four main problems: — (a) How essential is absolute simultaneity of exposure? (6) How carefully are convergence and accommodation controlled by the fixation-point? (c) How long is, and how long should be, the actual TEST 24: RANGE OF VISUAL ATTENTION 225 duration of the retinal excitation set up by the exposure? (d) What are the optimal conditions of general and local adaptation? A brief discussion of these issues is essential to the intelligent examination of the results of tachis- toscopy. (n) Simultaneity of exposure. Erdmann and Dodge contend that abso- lute simultaneity of exposure over the entire field is a prerequisite for suc- cessful tachistoscopy, and they therefore discount experiments, especially those of Cattell, that have been performed by falling screen and disc tachi- stoscopes. Wundt, however, maintains that Cattell's fall-tachistoscope gives us virtual simultaneity when we regard, not the physical exposure, but the retinal excitation which it induces. Hylan's demonstration that a series of letters may be exposed seriatim from right to left or from left to right indifferently, so far as the resulting experience is concerned, appears to confirm Wundt's position. (6) Fixation. In the Cattell instrument, the fixation-point lies 3 mm. in front of the test-card. Sanford pointed out that this arrangement pro- duced faulty accommodation with a tendency to double-images. Erd- mann and Dodge show that at the ordinary reading-distance,' the displace- ment amounts to 0.6 mm., and this fact affords, in their opinion, the explana- tion of the relatively small number of words that could be read by Cattell's (S's in comparison with their own. As already noted, this difficulty is remedied in the improved fall-tachistoscope. In the simpler moving- screen instruments in which the fixation-point is placed on the screen, there is undoubtedly a tendency toward ocular reaction, i.e., toward fol- lowing the movement of the fixation-point, with consequent disturbance of fixation for the ensuing exposure. It should be made clear that the center of attention does not necessarily coincide with the fixation-point. Furthermore, there is, strictly speaking, no such thing as a fixation-point, and no "punctiform functional center of the retina" (Dodge, 7) on which impressions are centered when attended to. The eye-muscles of a perfectly normal *S are subject to relatively slow fluctuations of tension (fixation pseudo-nystagmus), so that what is termed the fixation point is really a fixation area. There are also slight movements of the head, due to pulse, breathing, fluctuating muscular tonicity, etc., even when elaborate forms of head-rest are provided. ', / (c) Duration of exposure. The times chosen for objective exposure have varied from 1 to 1000 cr (o- representing 0.001 sec). Thus, Cattell used from \<7 up, and placed the optimal time at lOtr. Messmer concluded that 2o- was sufficient after practise. Goldscheider and Miiller employed lOo-, Huey 15