BOUGHT WITH THE INCOME FROM THE SAGE ENDOWMENT FUND THE GIFT OF fienrg W, Sage 1S91 ^...■. 2.„-2^,±X..a...7. ?.A:iU^S. MEASUREMENTS OF TWINS BY EDWAKD L. THOENDIKE FrofesioT of Educational PBychology in Teachers College Colnmbta Univereily ABCHIVES OF PHILOSOPHY, PSYCHOIiOGY AND SCIENTITIC METHODS BDIXED BT- J. McEDEX CATTEUJ AND FBESEBICK J. B. WOODBBIDGE No. 1, SEPTEMBEB, 1905 Colnubla University Costrlbntlong to Fhilosophy and Psychology, Tol. XIII. No. 3 GN295.T9 "49 ""'""''*' ""^^^ Measurements of twins by Edward L Thor olin 3 1924 029 905 449 NEW YOBK THE SCIENCE PRESS ARCHIVES OF PHILOSOPHY PSYCHOLOGY AND SCIENTIFIC METHOES Editorial oommunications should be addressed to Professor J. McKeen Cattell, Garrison, N. Y., or to Professor Fkedbkiok J. E. WooDBaiDQE, Columbia UniTersity, New York City. Subsoriptions and advertisements should be sent to The Science Press, Sub- Station 84, New York City. The subscription price is five dollars a volume, contain- ing between six and seven hundred pages. The numbers are as follows ; 1. Measurements of Twins : Edward L. Thoendike. 50 cents. 2. Avenarius and the Standpoint of Pure Experience : Wendell T. Bush. In press. 3. The Time of Perception as a Measure of Differences in Sensation : Vivian A. C Henmon. In press. 4. The Psychology of Association : Felix Arnold. In press. 5. The Diurnal Course of Human Efflciency : H. D. Maksh. In press. 6. The Pgycholoey of Reading : Walter F. Dearborn. In press. 7. The Psychology of Mentally Deficient Children : Naomi Norsworthy. In press. THE JOURNAL OF PHILOSOPHY PSYCHOLOGY AND SCIENTIFIC METHODS The contents of recent nu ncibers include : The Fourth Meeting of the American Philosophical Association. The Thirteenth Meeting of the American Psychological Association. A Philosophical Confession : Harald HOffding. A Syntaotician among the Psychologists : Basil L. Gildersleeve. The Essence of Humanism : William James. The Nature of Consciousness ; Frederick J. E. Woodbridge. Animal Psychology and Criteria of the Psychic : Egbert M. Yerkes. Inferred Conscious States and the Equality Axiom : A. H. Pierce. Radical Empiricism and Wundt's Philosophy : Charles H. Judd. How Two Minds can know One Thing : William James. Kant's Doctrine of the Basis of Mathematics : Josiah Boyce. Some Outstanding Problems for Philosophy : Cassius J. Keyser. Is Subjective Idealism a Necessary Point of View for Psychology : Stephen J. Colvin. Natural vs. Artistic Beauty : George Eebec. The Place of Affectional Pacts in a World of Pure Experience : William James. The Belational Theory of Consciousness and its Realistic Implications : W. P. Montague. A Deduction of the Law of Synthesis : H. A. Ovekstreet. The Realism of Pragmatism ; John Dewey. S a Year (36 numbers) 15 cents a Copy. LIBRARY OF PHILOSOPHY PSYCHOLOGY AND SCIENTIFIC METHODS Theory of Mental and Social Measursments : Edward L. Thorndike. $1.50. Science and Hypothesis : Henri Poincar*;. Translated by George Bruoe Halsted, with an Introduction by Josiah Royoe. In press. THE SCIENCE PKES8, Sub-station 84, New York City. MEASUREMENTS OF TWINS BY EDWAED L. THOENDIKE Professor of Educational Psycliology in Teachers Colleg© Columbia University AECHIVES OF PHILOSOPHY, PSTCHOIOGT AND SCIENTIFIC METHODS EDITED Br J. McKEEN CATTEIiL and EEEDEBICK J. E. ^VOODBEIBGE No. 1, Septembee, 1905 Colnmbia University Contributions to Philosophy and Psychology, Vol. XIII. No. 3 NEW YOEK THE SCIENCE PRESS 3 A,x^*^^o- 7 Press of The New Era Printing COMPAHV, Lancaster. Pa. 1)3 PREFACE. This research was made possible by a grant from the Esther Herrman Research Fund of the Scientific Alliance of New York and by the courteous assistance of many teachers and principals of schools in New York City. It should be dedicated, if any dedication were permissible, to 'The disinterested love of science and the intel- ligent cooperation of professional men and women.' Cornell University Library The original of tiiis book is in tine Cornell University Library. There are no known copyright restrictions in the United States on the use of the text. http://www.archive.org/details/cu31924029905449 CONTENTS. CHAPTER I. Original Nature versus Training as the Cause of the Mental Resemblance of Twins § 1. The General Argument 1 § 2. The Possible Causes op Variable and Constant Errors 3 § 3. The Resemblances op Twins and of Siblings 6 § 4. The Resemblances op Young and of Old Twins 7 § 5. The Resemblances in Traits Little and in Traits Much Subject to Training 7 § 6. The Resemblances in Mental Traits Compared with the Resem- blances in Physical Traits 8 § 7. Summary and Criticism 8 CHAPTER II. The Measurements op Mental Resemblances: Data and Methods § 8. The Original Measures 13 § 9. The Reduction op Complex Measures to Single Scores 22 § 10. The Transformation op Gross Measures into Deviation Meas- ures 23 §11. The Deviation Measures and the Calculation of the Coeffi- cients OF Correlation 25 § 12. The Sources of Possible Error 30 CHAPTER III. The Measurements of Physical Resemblances § 13. The Gross Measures 36 § 14. The Deviation Measures 40 § 15. The Measures op Resemblance 43 CHAPTER IV. The Form op Distribution op Resemblance in Twins \ 16. The Problem of the Distribution op Twin Resemblance 43 ! 17. Means op Measuring the Resemblance of a Single Pair 45 1 18. The Resemblances op Single Pairs and Their Distribution 47 i 19. The Views of Other Investigators 51 i 20. The Mode op Genesis op Twins 53 v 'i CONTENTS. CHAPTER V. The Specialization op Resemblance i 21. Means of Measubinq the Specialization op Resemblance 55 i 22. The Data on the Specialization op Resemblance 55 i 23. The Inteepeetation of the Facts 61 CONCLUSION. i 24. SUMMAEY OF RESULTS 63 INTRODUCTION This monograph will present (1) the results of precise measure- ments of fifty pairs of twins from 9 to 15 years old in six mental traits and (2) their bearing upon the comparative importance of heredity and environment as causes of human differences in intel- lectual achievement. They will be found to give well-nigh con- clusive evidence that the mental likenesses found in the ease of twins and the differences found in the case of non-fraternal pairs, when the individuals compared belong to the same age, locality and educational system, are due, to at least nine tenths of their amount, to original nature. They justify the emphasis put upon the magni- tude of heredity as a cause of the mental differences amongst men by Gallon, Pearson, Woods and others and suggest many practical applications in education and other social arts. Besides discussing this main theme, I shall note certain evidence concerning the mode of genesis of twins, the specialization of in- heritance, and the relation between physical and mental inheritance. CHAPTER I Original Nature versus Training as the Cause of the Mental Resembiances op Twins § 1. The General Argument The general argument which the reader will follow through a necessary mass of detail is as follows: We inquire concerning those causes which make one of a twin pair resemble the other. We mean by resemblance any greater like- ness than would be found in a pair of children of the same age and sex picked at random from the school population of New York City. We measure it by (A) the smallness of the difference between the measure of one twin and that of the other or by (B) the extent to which the two vary from the central type for their age and sex ia the same direction and in equal amount. For instance, suppose that all the boys in the New York school population who are exactly 12 years and 82 days old are found to be distributed with respect to circumference of head around a central type of 53 centimeters, with a median difference between any two such boys, picked by chance, of 4.2 cm. ; and suppose that John and James Smith, twins of that age, measure 57.2 cm. and 56.6 cm. Then John and James show close resemblance; for (A) their difference is only one seventh the chance difference and (B) they are respectively + 4.2 and + 3.6 from the central type for their age and sex. The second type of measure is for many reasons preferable, and will be used throughout. Measures of the general tendency to resemblance in the group of 50 twin pairs are thus obtained in the case of each mental measure- ment taken. These measures of resemblance will be presented in the form of Pearson coefficients of correlation corrected for attenua- tion by the methods described by Spearman.^ If now these resemblances are due to the fact that the two mem- bers of any twin pair are treated alike at home, have the same parental models, attend the same school and are subject in general to closely similar environmental conditions, then (1) twins should, up to the age of leaving home, grow more and more alike, and in our measurements the twins 13 and 14 years old should be much more alike than those 9 and 10 years old. Again (2) if similarity in training is the cause of similarity in mental traits, ordinary ' See American Journal of Psychology, January, 1904. 1 1 2 MEASUREMENTS OF TWINS fraternal pairs not over four or five years apart in age should show a resemblance somewhat nearly as great as twin pairs, for the home and school condition of a pair of the former will not be much less similar than those of a pair of the latter. Again, (3) if training is the cause, twins should show greater resemblance in the case of traits much subject to training, such as ability in addition or in multiplication, than in traits less subject to training, such as quick- ness in marking oS. the A's on a sheet of printed capitals, or in writing the opposites of words. On the other hand, (1) the nearer the resemblance of young twins comes to equaling that of old, (2) the greater the superiority of twin resemblance to ordinary fraternal resemblcuice is, and (3) the nearer twin resemblance in relatively untrained capacities comes to equaling that in capacities at which the home and school direct their attention, the more must the resemblances found be attributed to inborn traits. The argument is thus straightforward enough once we get abso- lutely exact and reliable measures of resemblance. Using convenient symbols, it is as follows: Let n. = the resemblance of twins. Let r«.^the resemblance of siblings. Let r9-n = the resemblance of twins 9 yrs. months to 12 yrs. months old. Let ri2-u=the resemblance of twins 12 yrs. months to 15 yrs. months old. Let rt. inunir. = ihe resemblance of twins in traits little subject to home training. Let rt.intr.^ihe resemblance of twins in traits much subject to home training. Then the influence of original nature in determining resemblances is in proportion to the extent that n. > n. ^t. in untr. --- ^t. in tr. Conversely, the influence of the environment in determining resem- blance is in proportion to the extent that rs. = ri. ''9-11 -. "2 O 1 (S « O e « a> V g 50 a o 1 a o 1 d 1 i t s < -6 < 1 -3 el 1 a s 1 1 OS S a o s 1= 1 s mi ft II as II as II II II a 1 p. o ft 1 o ft 1 ft o t4 I o s O 1 No. 1. 6 9.1 3a 42 41 43 46 12 20 8 6 6 5 3 3 3 3 5 1 6.B .5 20 MEASUREMENTS OF TWINS Sueh records will be terraed the gross complex scores. For the sake of any one who may wish to use these in any way, I print them in full (Tables). Table 5 Gross complex scores e bo a 'a 'rt d o o m g "o i E- a a i 1 1 I - a p p 3 H p 'e3 P o % p. .s I "to H H H C3 3 .p "3 't-i "bfl •a ■s-a S 1 -4J Ph O 1 o o o ID 1 » % t3 o O o & 1 P O 1 p 6 P O O o ■a ■a •a 5 J u s p p d a 2 8 2 ^ ^ ^ _ft CO a i ^ 03 .';^|a|l|s. * flj 5 a> aj ® aj .. V. .. .. C 2 S g M M M S • aj 0) OJ 5*H^H ■" '^ fe^B a K p SpSpC SiS o£paflo.ft g g S g 3 1 § 1 I ■? I -a I 20 a g 9 4b 15 24 13 15 14 15 12 9 4 12 10 47 53 58 64 9.5 2.5 10.5 .5 bg 4b 11 16 7 10 8 8 77485 36 4545439 9 1 21 a g 9 ? 13 14 27 13 13 14 14 15 12 11 10 38 38 41 39 8.6 .5 9 2 b b ? 10 12 43 15 14 12 10 11 9 12 9 34 35 39 39 7 1 10.5 .5 1 22 a b 11 5a 9 19 2 12 11 12 12 11 10 10 9 38 42 46 45 5.5 .5 2.5 .5 1 b b 5a 10 18 5 11 11 11 11 9 7 9 8 29 36 35 34 5.5 .5 4 11 11 year olds. 23 a g 1 5a 19 31 37 14 11 14 14 17 14 18 16 38 50 67 54 12 5.5 .5 1 b g 5a 20 37 20 16 15 16 16 15 15 20 19 41 46 50 49 11.5 2.S 7 2 1 24 a b 17 — — 54 18 17 16 15 13 11 17 14 36 39 48 47 11.5 1.5 12.6 .5 b b 7 15 25 35 14 13 14 13 16 13 IS 14 45 46 50 52 12 2 11.6 1.5 1 25 a b 6 7a- — — 29 24 23 18 18 16 15 11 9 45 50 55 61 14.5 1.5 11 1 3 b b 6b — — 50 20 17 18 18 17 15 12 10 35 38 45 52 10 1 7.5 1.5 1 26 a g 10 7b 18 29 91 21 21 20 20 21 19 20 17 60 64 69 67 18.5 .6 20 2 b g 7b 19 40 85 19 17 18 17 19 18 16 15 65 60 58 64 19 14 1 27 a g 5 6b 20 44 66 26 26 28 28 23 20 24 22 55 56 60 60 12 8 1 1 b g 6b 21 32 38 20 26 24 24 16 14 22 21 66 73 63 60 12.6 .5 9.5 .5 111 28 a b 4 3b — — 7 8 8 9 8 14 10 14 9 28 — — — 7.5 1.5 1 2 3 b g 5a — — 33 15 9 16 11 12 10 16 13 17 — — — 6 7 1 12 29 a b 9 4b 24 31 20 13 11 12 10 9 7 12 11 45 54 51 61 8.6 .5 6 2 b b 5a 18 33 34 16 14 15 16 18 11 20 14 47 42 50 52 11.6 1.5 7.5 .5 2 12 year olds. K 1 6a 21 34 22 10 9 13 13 7 5 7 3 47 49 50 55 5.5 .5 8 b g 6a 17 39 29 11 11 14 13 12 11 9 7 47 80 66 64 8.6 .6 7 31 a g ? 6a 15 26 30 15 13 16 13 12 8 13 11 45 50 62 53 8.5 1.5 9 11 b g 6a 12 18 25 15 14 18 18 15 11 15 12 66 61 63 64 11 2 7 1 11 32 a B 4 7a 13 19 47 15 8 14 8 23 5 18 12 63 49 57 62 13 1 15 1 3 5 5 2 b g 7a 12 27 46 22 22 18 12 16 14 21 19 48 45 54 59 14 1 15 1 4 33 a b 4 3b 8 22 8 15 12 15 13 12 10 9 5 29 28 34 40 5.5 .5 1 2 12 b b 3b 11 14 5 10 8 12 12 9 5 11 10 27 24 32 30 3 1 3 2 1 34 a b 4 6b 3 5 17 9 6 6 5 7 4 7 6 19 30 32 31 8.5 .5 10 4 b b 4b 9 12 6 15 13 14 13 9 9 10 9 24 26 25 37 5 2 5 2 36 a e 7 6a 16 29 39 19 19 22 21 14 13 10 12 44 48 63 66 10 9.5 .5 2 b g 6a 18 29 42 18 17 17 17 14 14 12 11 53 60 63 56 7.6 .5 8 1 1 36 a b 5b 17 24 20 4 3 6 6 10 7 8 4 48 65 49 57 8.6 .6 6 1 b g 7a 12 16 21 12 10 11 9 8 5 8 6 38 43 44 41 13 2 1 37 a b 7 5a 12 17 7 1 5 3 3 2 2 1 22 32 29 30 4 2.5 .6 3 11 b b 4b 16 22 4 14 14 14 14 10 9 10 10 38 46 46 44 2 1 2 38 a b 6 5a 11 14 6 17 17 15 11 14 10 9 7 28 36 29 32 6.5 .5 2 2 1 b b 6a 12 25 8 17 15 14 12 14 13 14 11 30 30 40 32 8 4 2 39 a e 11 5b 17 31 33 20 18 22 20 20 16 22 16 38 43 48 53 14 1 13.6 2.5 111 b g 5b 13 22 30 17 14 18 13 15 14 15 13 41 41 42 46 9 10 1 1 40 a b 10 6a 9 21 15 16 13 14 13 10 8 11 10 37 37 43 49 13.6 .5 9.6 2.5 1 1 b b 6b 12 38 48 16 14 14 14 11 8 15 13 43 44 45 48 12.5 .5 10 1 1 41 a ff 9 7a 16 40 38 16 13 13 13 14 10 16 14 55 59 60 60 146 .5811 2 b g 7a 17 42 31 14 11 18 17 14 13 16 14 38 56 60 62 12.5 .5 9 42 a b 11 5 10 18 30 22 22 16 16 13 12 20 19 44 45 52 53 9 1 9 b b 6 — — 28 17 15 18 13 10 8 18 15 44 62 59 65 10 2 13 22 MEASUREMENTS OF TWINS g a fri M •- J-i (a V A *3 ** 5 s J:L._?, ^^s»" a •O ^ _- „ ^ a S S 2 .3-3 8.a«S„S«,£»SHHHH£g O) m O) V aaaalallaaa r" i" A" -o 13 bn bfl M « gj o o Sag, ^ ^ ^ ^ ^ g a a B •S'SmSSS 3 3 3 S 3 33333. "'.-'.---"ap. aa^>.b.t.t> 13 year olds. 43 a E 2 6a 17 37 34 17 15 18 18 19 19 17 17 55 62 64 66 17 8.5 5 b g 6» 17 30 20 20 20 17 16 18 17 12 9 48 52 53 56 13 10 44 a b 4 71) 9 18 48 21 21 19 16 17 15 16 16 38 37 38 43 19 1 15.5 1.5 J } b g 7b 17 35 63 21 19 22 21 20 19 20 16 58 66 68 68 14.5 1.5 11.5 1.5 1 1 45 a b 4 7a 19 28 43 20 20 14 13 10 7 17 11 87 46 53 52 12 3 14 , , , b b 8a 18 43 72 21 21 22 21 26 23 22 21 53 63 55 58 17 2 18 1 111 46 a b 9 7a 17 32 33 15 12 9 8 9 7 13 11 34 48 66 50 12 7 4 1 b b 6b 13 23 41 14 14 12 11 11 8 9 8 41 42 42 46 7.5 .5 2 2.5 1 14 year olds. 47 a b 3 7b 19 26 74 18 16 20 17 U 10 14 8 51 46 52 61 14.5 .5 16 J „ b b 7b 15 28 60 15 14 15 12 14 8 15 10 40 44 45 44 20 14 1 i 48 a b 5 8b 16 33 42 10 9 15 14 15 13 18 18 46 49 60 60 15 10 1 11? b b 8b 16 31 41 14 13 18 17 12 12 18 18 50 48 53 54 11.5 .5 9 13 49 a B ? 4a 13 12 24 12 12 17 16 9 8 10 6 25 31 40 40 7.5 1.5 8 b b 4a 10 13 10 9 9 10 10 misund's'd. 33 29 26 42 2 3 11 1 50 a g 9 8a 12 21 29 22 20 19 16 18 16 21 17 40 40 43 42 12 12 2 b g 7b 21 30 23 18 13 19 17 16 13 19 13 34 38 50 43 12.5 .5 12 § 9. The Reduction of Complex Measures to Single Scores For further work it is convenient, and almost imperative, to turn any score involving two or more kinds of quantities, such as amount done and errors, into a single score representing capacity as fairly as may be. It would be possible to so evaluate amounts and errors as to do this with the utmost fairness, but to do so would probably take over a year's labor. I have therefore arbitrarily reduced the complex scores to single scores by the following methods, which are probably innocuous as far as any influence on later deductions goes. I have counted each wrongly marked word in the a-t and e-r tests as — 2 words marked. Wrongly marked words occur in only one out of ten of the papers. If they were counted as — 3 words marked, as — 1 word marked or not at all, there would be little or no change in the resulting coefficient of correlation. I have in the misspelled word test neglected entirely the words wrongly marked. They occur in over a third of the records, though over two words are marked wrongly by any one person only rarely. If a slight deduc- tion were made from the number correctly marked, say of one for Add Mult. A test 0pp. Word Mis. 1 2 1 2 1 2 1 2 9 23 51 36 39 28 32 86 97 7 8 MEASUREMENTS OF MENTAL RESEMBLANCE 23 each one wrongly marked, it would very slightly raise the coefficient of correlation, perhaps one half of one per cent. In addition the single score was obtained by adding the number of correctly done units to the total number done, that is by counting a unit correctly done twice as much as one done with an error. So also in multiplication. In the opposites test the single score was obtained by subtract- ing 1 from the number of correctly written opposites for each wrong opposite written. No account was taken of words skipped. In the A test 1 a and 2 a were combined into a single measure and also 1 b and 2 b. The gross complex scores are thus reduced to what may be termed a gross single score, of which the following is a sample : 2 a g 2 5a § 10. The Transmutation of Gross Measures into Deviation Measures ' 1. To estimate resemblance we have to turn these gross measure- ments into plus and minus deviations from the central tendency for the age and sex of the individual in question. 2. To prevent the older and consequently more variable children from influencing the result more than the younger,— that is to have each pair of twins weigh alike in estimating the resemblance,— we have to divide each such plus and minus deviation by the variability for its age. 3. To prevent the more variable sex from influencing the results more than the less variable and also to prevent attenuation of the coefficient of correlation by improper comparison of a boy with his twin sister, we have to divide each such plus or minus deviation also by the variability of its sex. The facts so turn out that 2 and 3 are of no great consequence, however. These procedures involve the determination of the central tend- ency and variability for each age and sex from 9 years months through 14 years 11 months (ages, were calculated only to a month, 9 years meaning all ages from 9 years days to 9 years 30 days). I had at hand measurements in each of the tests, in addition to those from the twins, from a large number of children (from 300 to 2,000). My estimates of central tendencies and variabilities are therefore far more accurate than would be possible from calcula- tions on the basis of only the 100 individuals of the twins. These additional measurements were, however, mostly class tests instead 24 MEASUREMENTS OF TWINS of individual, and were taken in the case of some of the tests by- different methods and even by different individuals. They can not be used at their face value, therefore, and the task of finding from a large but not homogeneous set of measurements the most probable central tendencies and variabilities for each age of children tested by the person and method used in the case of the twins, is intricate. In general my practice was to wse all the information I had to determine the relative tendencies of different ages, the relative variabilities of different ages, the relative abilities of the sexes and the relative variabilities of the sexes, but to give to the 100 children of the twins the entire decision as to the absolute abilities.^ That is, even if I had 10,000 class measurements with the A test giving values of 50 for 9 years old, 60 for 10 years old, 70 for 11 years old, I should discard them if they made the deviation measures of the 100 twins come out all plus or all minus. I should seek such a scale of central tendencies in each test that approximately 50 per cent, of the twins were plus and 50 per cent, minus. The main thing then is to get for each sex a series of figures representing the relative abilities of different ages, and a series of figures bearing the same ratios to each other as in the former series, but such that the deviation measures of the twins reckoned from them will be about half plus and half minus. This latter series of figures was calculated for each test in a table which I shall call the test's 'scale.' A second scale was a series of figures each of which stands to any of the others in the relationship in which the variability of its age stands to that of the other's age. Thus suppose we have as our scales: Age Boys Boys Trs. Mob. Central Tendency Variability 9 5 10 ^V 12 5 15 ^V If now John and James Smith, aged 9 years 5 months, score in addition 11 and 12, and Fred and Frank Jones, aged 12 years 5 months, score 19 and 20, the deviation measures of the four are -|- 1, +2, +4 and + 5, but, since the variability of the 9 year olds is so much less, the coefficient reckoned from these (.956) would be unduly caused by the older pair. If we divide each deviation measure by its age's variability, or more conveniently multiply it by the reciprocal of its age 's variability, we have, + 200, -|- 400, + 100, + 125, with r == .822. 'This makes the obtained correlations too low, in so far as it introduces any error. Whatever error there is, is very small. MEASUREMENTS OF MENTAL RESEMBLANCE 25 A third scale gave for each age in the case of one sex a figure by which to multiply in order to make comparison fair with the other sex. Thus if the variability of girls was 80 per cent, of that of boys we should multiply each girl's deviation by 1.25 if we wished to make our deviations properly commensurate. In some of the tests it was found that the alterations of varia- bility with age and with sex were so slight as to need no especial allowance. Even in the A, a-t and e-r tests, where there was the most need of such an allowance, the effect of making it was to alter the coefficients calculated regardless of it only by .002, .032 and .005. In this report as originally written the derivations of the scales for the A test and for the other tests in essential features were given. Since even when condensed the facts of these derivations fill some 50 pages, mostly of tables, and since their only advantage to the reader would lie in relieving him from taking anything on trust and putting at his service more or less useful statistics con- cerning some 3,000 children from 8-15 years old, it has been thought best to omit these tables and the commentaries on them. § 11. The Deviation Measures and the Calculation of the Coefficients of Correlation After transmuting in accordance with the scales described in § 10, we have for each individual in each trait a measure in terms of his deviation from the central tendency of his age and sex in gross or in terms of the deviation's fractional part of the varia- bility for his age and sex (or an approximation thereto). These deviation measures are given in Table 6. To measure the probably true general tendency to resemblance of a group we need (1) to measure the general tendency to resem- blance actually shown by the figures of the deviation measures and (2) to correct this for the attenuation due to the chance inaccuracy of the deviation measures. (1) is given by the coefficients of corre- lation between twin and twin of a pair in the A test ( 1 and 2 com- bined), in the word test {a-t and e-r combined), in the misspelled word test, in addition (1 and 2 combined), in multiplication (1 and 2 combined) and in the opposites test (1 and 2 combined). (2) is possible in all cases except the misspelled word test by the proper use of the coefficients of correlation in twins for each single measure, and of the coefficients of correlation between A 1 and A 2 in the same individuals, a-t and e-r in the same individuals, etc. These correlations for the entire group of twins are given in Table 7. So also are the coefficients as corrected for attenuation. 26 MEA8VFEMENT8 OF TWINS In the word test I use for the score for the two trials together 2a-t + e-r, since the variability in the e-r test is about twice that in a-t test. The combined deviation measure for tests 1, 2 and 3 equals the following : Al + A2 + 2a-t + e-r -j- 2 mis. This gives the three sorts of test approximately equal -weights in determining the com- bined measure and gives the a-t test approximately equal weight with the e-r test. The combined measure for tests 4, 5 and 6 equals the following: Add. 1 + add. 2 -f mult. 1 + mult. 2 + 3 (opp. 1 + 2). In cases where no score for multiplication was at hand, I used 2 (add. 1 + add. 2) +3 (opp. 1 + 2). There is not a perfect correspondence of the combined score for 1, 2 and 3 with the separate scores, owing to the fact that the allow- ance made for variability in one of the separate scores had not been made when the combined score was calculated. MEASUREMENTS OF MENTAL RESEMBLANCE Table 6 Deviation measures of twins 27 -J Ol W rfi to to 0'» tj'» o'P O'lO CT*P 0*» )-iO(D(»-j05ai*iCoto D-» c^P o"P c'P cr-s* o-P u'V CT'P c'p cr» o'p (nm mtn o-o- Co' Co* D-C- Co' O'tK) « c' D-O- n D- mcp; 0*0- CfJQ (Waq aum 1 1 §5 cog CO GO si 1 1 M £§ 1 1 coco 1 1-1 1-1 si li S3 1 ^ ta CD «s 1 1 u -1 H- Co » to 1 1^ to ts, lo^^ 1 ig oo 1 1 1 „g o8 .1 SSI 1 1 tela as 1 ii tn Ol Old 1 §8 oo 1 1 OOOO tc o> 1 1 cop is ss 1 1 1 1 2-a ^CO 1 oo S5 >' CO to to • + to J I I I I I „ I I I I I I I II I I I -^■*5 to"-* OOo t^ o I I ii. i -^ ' I'll Cn Oo t~. if^ ^ ?i at i:^ *^ •— *^ --5"^ t© GO coco ^00 Ol Oi Oi "^ t« ts I III I I I „„ I rf^ J-* oa Go I I I I I I I II Kt oi ta to CO <£! t-* to ,*^ K* ta ooOi toiP- '-'-J irto^ "^^^ OoS^ ii ill So -3-J ~J-a "^s-a .^— . - ,.« w tn ta ta 00 00 to bs I I I c«t> "^ te C7T CO ^^ o "-1 t-"^ oi -J oo 1-1 O CO"--* (O ~J Oil-* I I I I „l I I MCo Cn rfk — ) I I >-* ta tOCo en i«k £^ rfi li^a t^ I— ' <© I I I to O) tS en O O O Oo Oi Oj b9 to WtO rt S CD OO O CO I-' ^ t^ 1^ 0:1 iF>- 00 00 8S CnOT ^t^ tS-Ui o'o' ^g ^h- 1 5g 1 1 1 ? ? 1 1 1 1 1 sg 1 2!l 1 00 1 1 01 01 1 1-L (O ^*^ I I Cn *> Ol CO O Ol .1 I c;! to --J Ol to I-* n5 ta ci to to 1 1 t-QO I I ■, I I I en to b-i ,1 CO06 00 „„ 1 1 , fP'CO Qo bo to CjS Cd toco coco 00 I 1^, tQ - ta Oi , I I I I I I I O Oo CO o I I I I I I o I to 1-' -^OO *~*> to i-> CO 1-1 t-* fe to 1-- OT 01 O CI'— -^to i I III I I I I I I I ill I I I I I CO la M 1-1 to coco tDt> to 1-1 -ao CO to 00 CO ta it^tzi OiUi I I cn3 g K^ ta ES ^ S- 28 MEASUREMENTS OF TWINS 5^ (S-^ lO lO O 00 lO lO iM CO T-«C1 >H CO -4s ^-4 WM r--e^ !> (:; t^ Sr^ 00-* ^ -^ COO I I I I I I III I I II II I I - I I MM M T I I I I I I "5100 Ol Oi I I I C^ i-H i-H lO I I II I I I I II 77 M II I I I I 0(^ 77 III I I I I I I I I CO r^ I CC '^ (MO I I I iM CQ CO CO rH CO CO I I "" COCO O CO 0»0 Tft to 7 1 I rfH to CO '5? I I " 1 1 1 1 0!(„M IS m fl a rt i7 CO to 1 CO .2 Sfe 15 §? I I "-it- WM ©5 t- r^ CO iM(M oo to coco co?3 IS 7 t-co rfHTP CO 00 is 1 l-HCO (MtH OO 00 CM CO I I I II M I I I II SOI^ ■* -* OO CiCS >-( 1-1 lO lO (^ Co ©4 " I I f"* o C) o t: lO CO !0 05 S I III OO =0(N 0000 COOo s^to is>ito e^i^ 1-1(34 I I I I II »« m ci 1-H oi O) Old to tO.-H -* t^ 1-t I I M 7 I W5 W5 Oi i-l Oj Ol Ol Oj 1-1 iJ^ lO t-l O "SO II i" i7 7 I SOS '31 -* ^* lo v(j ^ ■^ 1-1 r- ©:j o f- -* M O 1-1 WS -* I I 77 77 Tf XH opoo CO Co-» I I oo CO to I- i-( -J o c I I 1 CO-* 1 g?8 1 1 gs gg 1 |8 1 1 tD -4 C4C4 O C) SI? 7 1 gs? 7 gSi OO §S; OO III I -f o CI 1-1 lo lo T-t-^ to r- -^ C0 1* coi-i r-i i-( CO M M lO CI O 1-1 lO O M o t- CO (M 0 -( GO GO D O 1^00 Ci -J »Co MO CS Gto So ^S S;s SS I I I I CO Co I I I 1^ I I tats CO *.!:« co^^ Vi t-^ K*!-^ OotO Ol^ OotO CD(0 I I I I Qj& ta ts ts 04:^ to ifi. Oi QO Ol CJl I I J i J I III II ta to CO ^J Ol ■ So l-» >~t. I I , ts CO I I <— CO Ki I !„.„ i t-i !^ ^O Ol to 01^ COCO "Ol in CO I-* ^3-^ oo^i^ i-i -J to ttk ^ 1^ to 1-1 t^ ~ai~' coi— I >— to o3 -J dot- >-^ otai QTOi -^-J eo-^ isjts **0) i I .11 I I I I WK. I I J.J II II I Wh-J la-t- jMQo I-*IQ t-Oi l-itO tOh- t> I I I I I I I I I \ t h-i^o tapi oij:?- oita -t^cn oato oi^ 7" I I I I I M II ^ kjc^ iQ 1-1 1^ ta fcOKi IS CO toi- 05 CO Ol ci <:i to t^ ta *-j'iQ o co -ji- I I I I I I I I K,i 1 1 I I i3 1 1 ^ to Ota Ol Oa toes ta u-i c^ O) ft- o o Old -^ -^ Oo c*> Ol >p. 4^ Jill ^ ^ to to is >~* 1-1— J »-*K- ~4CO I „ II K* h-1 CO .tw to t© t© -- — 1^ OT 00 CC- (DO — — — .- too iSrfi. *^0 OOi --JOl 1-1 Ql I II I i. I i I -3 Ol Kt to »^ *r- ca to to to 1-1 ta Ol to -* Ol t- Ol -J h~i t^£~ -^ s~i to h- ^■fr- OO Ol h(i. ta CO OO OO OO I I CJl I I OiOi I I N .1 I Oli- -^ O ** "^ O-J CI to t^ to OO Oi ca to to -J to "j-j to to oo t-t- III M 1^1 ^ ):a Cat' toco Ol CO ta Ol h~i o& cot- MM M M J „„ I I M M I I ^ I I I J I I I I I I I M M II I M II I I II II I MM II II II II II M II II I I I II II M M I CO )-• ce "^ to to CO to -* Ol Cl rfi. ^ ^ CO Ol rf^ >^ Ol Ol CO CO Cn to to 1-1 CO cots to 1-1 CO en CO -J o i— iC*. >Ck coco CO Ol I*' CD 1-1 to coco OI5 E 30 MEASUREMENTS OF TWINS Table 7 Data for and results of correction for attenuation = the raw coefficient from the best obtainable deyiation measures. ^ " " " " trial 1 in one twin (a) with trial 1 in the other twin (6). ^=11 .i '• <' " 2 " 2 " " ' Data. COO fcoisD oocx) oo COCO ^1-^ cn-^ 3bO tObO bSbO cote coco h-ibO tObO to to tots CObO TT„-„%,+ CTh^ i-'^ bOh-; -aoo ^cD rfxK) rf^co hfi.-J^ K^co oitt^ o-^ 000 h^^ o^o -5o toto -^leigtit bitting ^^^ ^j CO -J CO coco coco -J -J CO 00 ^ -f^ CO »U h*^. CO Cn Ox COCO kt- n^ >- H-o oio tcco h-o oxoi <»H^ OH^ be^ o?S ol^ wh2 o^ liqS -Width of H-' O -JO )-' O en OS f-i h- to 00 Oi CTs I^ bO rf>. O -^ CD ■^ -^i 1CD COrf^ -^lO Oi-^ OiOx CC4^ bOO OCO •t».cjx cnot oicn rfi^-oi oxcn cncn cnC7» otoi Circumference coco tObO CO*-* yf^Oi bOH-' COO COh-" CObO --abO 00 hi^rfi- 00 hJ^Cn nfTTaci^ Oioo otcn 00 coco OCO 000 0001 ocn OOO hf^-bS oibo cocn COO ^^ ne&a 00-^ COOT vfi-ox i-'to OOOO i-'to -4^ i^CO h^O ptO COCO cO^*^ CObO COOx GncO coo co«o Sitting / ht. C" Ox 01 en CTi Cn OiOi cncn 0505 OxOx cnOi CnOx oxCn CnCn CjiOi osd Cnen oxCn Ox05 tei— ' bSCO bSCO <:OCn cO*-* C005 CO-^te ocota Finger Joint Length CO bO h-i I-* b3 CO to h-' toco to l-» h-i tO bO bO l-» tO tO tO h-* h-* tO l-* tO bO 4, 000 OCO coto enen ooi— ' ht^co rf^co oo-a --a-^ en o co i— ' en Oi o en (z> t^ en rfa- cow rf>- co i** o u^ en t*. QJ^ >f^ bobo h^^co h^^o^^ r orearm OCO bOto cnoo ocn toco oi-' oto co--a cnco oco rf».---i ooen -^o Length «j-^rf». OCO coco OLO --JCO ObO CObO I— 'CO OCO oxen 000 C0t04- enO ^Jtf^ COtO bO-^ ox 05 bOGO hfi.'J h^en coo toco ooo GO--Jr rfa-Cnto 38 MEASUREMENTS OF TWINS ■^ O CO rH -^ Ir^Os OiO:) tJH (N COCO CO CD OD lO COCO 00 »0 IM 00 (M iC COO + t^ !>• iC lO "^ I— I as lO CO -^ lO lO ■«** lO »0 CO l>. t^ CO -rj^ lO CO CO I>- O OS iHCDco t— 1>- i>-i^- COCO i>i>- c^t> i^-i^* i^-t— coco 1>-I> I>-t^ t^t^ t-CD Forearm cot- -^-"ti -rfioo csco t^co osco coco oco t-^ic -^o coa> olo oo(M Length wcoco t^c- t-.»o '^cs cot- t-^oo cot- cocz) coco it-j:— r— oo ooco Tf-^^ ° coco coco coco coco coco coco coco coco coco coco coco coco coco _,"rtHCO t>.0 COi-H Oi— < t^ Oi TtHO I>-CO COtM COi— ( "^lO COOi C-t-- t-t— i>-oo corH cot- t-t- coos lO-^ coco coco coco coco coco coco coco CO CO coco coco coco coco coco ■^ CO -^ i-H CO Oi CO W CD coco -^ Oi OO CO ':0 Oi -rf* ^ t- O rJH Tfi COCO -^ CO + T-H i-H -^ CO (M CO (M rH G- coco CO'^ COtJH T-Hi-H coco t-Cs coo lO-^ as-— ( CsCi 05G^co o-^ i-iio (Mco u:)oo cO'^ i-c^ ot-^ ooco coco coo Oi-h Sitting /ht. lOio cMco coio t-io coco iocjO lOiO lOiO lOiO lOiO lOiO lOiO lOiO lOiO ^lO lOiO k;coco oit- i-fco coco corH coo cmco -^cji cot- t-co -^t- cjsco ^COCO coo coco ooco oco C^ Id OCJi coco COi— I G<1-^ r-<(M oot— ^0000 oot- t-t- t-t- cooo CO CO oot- 0000 t-co coco OOOO t-t- n -UT Tj r^'"**'^ (M(M lOcXi COu^ COi— I '^lO "^i— ( COCO lOtM lOcO »00 COt— Cephalic Index '^co'^ -^ a oo cooo oco i-iio oo oico coc^ i-h-^ (M(M ooco COCO cot- t-t- t-t— OOCO OOCO cot- OOCO t-oo OOCO coco t-t- .-H OO t- I— I cooo coo O 1— I 00 ■^ coo oco O 1— ( (MCO OO O tM lO COOO .-iTt^G^ coo coco GOOO OCO t-CO COiO OOi "^-^ COi-H COtJh OiM Olt- 00 CO ooco t-t- t-t- OOOO t-t- OOOO t-t- coco t- 00 coco ooco t-t- Width of Head coco oOt-h t^co coco oco COlO i:DIO lOiC TjHiO Tfrfi ■^TJH LO"^ "^-^ coco f-— 1 I — 1 I — I ' — ' r— 1 1 — 1 I— 1 1 — I 1 — I 1 — 1 T^ i~H i~H T~l rH i^ i~H i~H 1— I r—^ iH f~i i— i i-H CO-^ CJSO T— iCi COOO lOt- t-O lOCO t-(M CM CO "^CO Ot- - coco t-t- t-t- t-t- cooo coco t-t- cot- t-t- t-t- ■^ t^ S^ lO O CO »0 CM (M CO (M 00 O (M t- COO t- lO 00 t- (M iM r- Oi TTPTO-lif t-(M Tj^LO (MO (M^ t-t- ot- COCO "^CO Oil- 00(M C^tO C<1G<1 fri-H iieignt co(^ ^t:^ ^^ ^^ ^^ ^^ ^^ ^^^ ^^^ ^:^ ^^ ^^ ^^ rHrH r-irH rHrH rHrH rHrH i—lr—i J~i i—{ i~( j—{ r-t j—i rHrH t-^ r-i i—ii—i rHrH 00 Eye Color ^""^ ^ '^ 2S ^"^^ SS SS '^'^ £5S5 ^^^ ^^ co(M cooo lOO •' vj GN rH CM CM Gn CM CO CO ,—t ^ HairColor ■^■^ -^-^ "2 -= -^ -^-^ Countenance :m S o s o and General "3 "^ o, 9 "T^ Appearance +j '"* m ^ ^ c a e a rO ^ K M J2-0 H H oJ c3 CS 03 -OTS qaqa TlqH Sex r^ tlD tuD £iC tiO W) W)pC bO&O-Q^ rO^ tUDbO tiObO^bOpDpD ^J= SB S Si -o-^ -flnfl a o *1 JS bo -QpO &JD bo 50 so MEASUREMENTS OF PEIYSIGAL RESEMBLANCE 39 aqcn; Cfqaq c^a^crqcraqoq CfQOQ era' o'o' o'crotscrcr^crq o^o^crqoq Sex s ^(D ^^^<^^ '^^ Countenance ^=LMBBS-BBgB^ aod General J^ s Appearance at )Z^ S.S-__^o'h2 an cr'^g^to o-c^o'ct'c-ci-'^^'" Hair Color 33 CTCT- (t g g^ m o- g-fi. o- o" cro- c c » 3 O h. to bs i-j w -to ^ M !_' h- » CnCn Cn05 CjiCji *i^ 00 ^f^rf^ Oih^ **CO ^f^^4^ i4^4^ COOO COtO COCO ^^h** Heiffllt CnOS -tOtO •— »0 coco OiOO Ol— ' CO-tO ashf^ coco CiOi OCO CnbS t^si— ' o ^ en en OS <£i OS o rf^ oo as oi o en -a coo om ^4^ oo o 4^ i— ' too --700 GOOD ^3-a -^"^J ■•*». toco coeo h*».k*^ tooo ■::;J'S Heigat bitting O 00 t—* QO i-i CO O CO OS en I— ' CO tf^ h*^ O^ O hf^ en I—* O CO 1— ' CO o en ° •. i4i. oi "^ 55 ►&- >t>. en en >i^ rf^. ^;x h^i. vf^ ^i^ ^^ Co i^ coco if; ^^ ■"* Oi-» Oioo 001—' -^Go ^^rfi. oo ^*-i coen -Jos oih^ Mw enco oo tt^ Width of ^ Head en en 4^ 4i- hf^ en CO O^ ^LJ. ^^x Cn Cn >fc>. n^ ht- h^^ ><^ v^- ^|l. rf^ CO i*>. 05 CO ji^ »i^ ^ OO OiOO COl— ' COCO COCO OtO tOtO COCn -^ICn OiCO -^1—' - CO cooo coo t-j en oo >f^ -^ coen coco * >f^ CO Ci coo coco GO rfi. 0000 cooo cooo GO -J 4i-en cnCn o^ o^ cncn cncn cnCn cncn encn encn cncn Circumference ^£^cn toto co^ ^^ coco ^co oo CO to toco coco o^:^ boto ^^ of Head 01 kP* en -q toto oo oi izn o co ■<( -^i oo COO O to oo COO coen -J-o ~a-J -Q-^ -^Oi -7 -J -a-' 0001 CO 00 CO CO to en ^p>. CO '^ C5--3 --j-o -:roi ^*-i -^30 ^Oi -^roi osos -^lOi ~:;J;;3..-,t- x-a cOhfa. co>^ enOD ento oco coco toco coco enco oo^^ Finger Joint Length I— 'I—' I— ih-l h- 'I— ' 1— '1— 1 )— *l— ' I— 'I—' I— 'I—* I— 'h- 1 H- 'I— ' I— 'I— 'H- COrfi. COrf*. h^^tO >P^CO coco coco COtO '-'to l^CO WCO + OH- ^ hJ I— I CO to as H- o OS o -a CO -^1—' coo o o to CO rf>. coco coco ►t^ CO coco coco coco coco coco coco Cico coco 000s 000 --:i— J coco co-j e;icn CftO r^r^M --JO ^ OS ^f^ ii^ y(^ ooto 'J >f^ --J -*^ COCO coco cooo coco 05 CO coco coco coco coco Forpnrm )^ -;;-i coco Cnen + -3Cn oen OS to coto crs^j^ ocs co-3 osos cocrs cnOio 40 MEASUREMENTS OF TWINS § 14. Deviation Measures The scales of probable central tendencies from which deviation measures were reckoned are summarized in Table 9. For the derivations of these scales space is lacking. They are not so reliable as I should wish them to be, but the resulting effect on the calculations of resemblance involves only a trifling variable error. No allowance was made for sex or age differences in variability except in the ease of height. Each deviation measure for height was transmuted into the fractional part which it was of the variability of the age and sex to which it belonged. The measures of variability used were those given by Boas in the Report of the U. S. Commissioner of Education, 1896-7, pp. 1555-1556. The deviation measures are given in Table 10. Table 9 Scales of proiable central tendencies for times in physical traits . Forearm Length Band G 690 Years 9 Months 0-2 Height B G Width of Head BandG 144.5 Circum. ot Head Band G 518 Finger Joint LengI B and G 1 2 1 and 58 65 123 3 519 5-6 1267 1259 6 520 66 124 9 522 58.5 66.5 125 10 145.0 523 3 525 59 67 126 5-6 1316 1309 6 526 9 527 59.5 67.5 127 11 145.5' 528 60 68 128 3 530 61 129 5-6 1362 1364 6 146.0 9 531 69 130 12 3 532 61.5 69.5 131 5-6 1405 1423 6 146.5 62 70 132 9 533 71 133 13 147.0 3 534 63 134 5-6 1458 1485 6 147.5 64 72 136 9 148.0 535 73 137 14 148.5 74 138 3 536 75 139 5-6 1521 1533 149.0 6 65 140 9 537 66 141 For height sitting take .523 X height. 705 730 755 780 810 MEASUREMENTS OF PHYSICAL RESEMBLANCE 41 For cephalic index take (A) 813 for all or (B) 79 for 9 years, 80 for 10 years, 81 for 11 years, 82 for 12 years and 83 for 13 and 14 years. By (A) r=.738; by (B) r = .741. For height sitting/height take 755 for all (the gross measure for this ratio vyas gotten, for the sake of convenience, by dividing the height in inches by the height sitting in centimeters). For separate arm lengths ( 1 and 2 ) take .5 X scale for 1 + 2. For height, height sitting and forearm length the scales used were by interpolation made into scales fitted to single months. Table 10 Deviation measures in physical traits ■a ■s d (d pd a. Q bOCtO 1 fa bo n S 2 1+2 1 i 2 1+2 1 b b 116 138 17 17 45 25 —25 —23 32 25 4 8 2 4 1 6 1 25 7 20 7 45 14 2 g ■ s ■ — 76 — 5 40 47 — 65 — 85 —S3 —40 —14 —10 —59 —52 —1 —3 —1 - 4 — 1 —11 — 3 — 7 + 6 —18 + 3 5 b b —103 — 32 —25 — 9 60 90 —17 — 1 24 28 4 0.5 —5 -4 —3 —3 — 8 — 7 —11 —10 -14 —17 —25 —27 6 g g — 38 3 8 12 — 52 — 2 3 13 —18 —16 —20 —16 2 2 2 2 1 7 — 1 6 13 7 b g — 6 —292 17 —41 13 —117 36 24 0' —42 —23 —22 2 —5 2 — i 4 — 6 — 3 —37 4 —39 1 -76 8 b b 276 214 43 31 40 75 40 46 6 10 9 14.5 4 5 4 5 13 21 22 23 35 44 9 b g 41 114 17 22 — 40 — 45 —36 -47 — 6 17 —12 8 —1 —5 —5 — 1 —10 19 8 16 4 35 12 10 g b — 97 — 70 19 28 — 37 — 77 —12 —43 —20 —22 —33 —38.5 —2 1 —3 —7 — 5 —29 —10 — 9 -53 —19 11 b b 65 — 3 —10 —24 — 88 — 98 —68 —75 — 1 25 26 3 1 4 3 7 4 18 7 14 7 32 14 12 b b 38 5 8 10 — 9 - 40 — 9 —40 12 23 -31.5 —41 —3 1 — e — .?0 — 7 — 2 —18 —13 13 b b 11 24 1 14 A3 73 10 46 —15 5 — 1 —12 4 3 2 6 3 13 27 17 34 30 61 16 g g 49 231 11 — 6 — 53 — 88 —48 —50 — 1 — 1 3 —20 —6 —7 —7 —9 —13 —16 —30 —27 —30 —SO —60 —57 17 g g 55 88 — 9 — 8 — 55 — 45 —35 —25 11 8 20.5 26.5 2 3 1 2 4 3 21 — 1 16 2 37 18 b g —136 —230 —29 —52 25 — 5 23 23 — 4 —16 — 1 —6 —5 —7 —7 —20 —18 — .Z5 —17 — 3S —35 19 g g 93 105 40 37 37 — 3 26 —16 9 9 —27.5 —22 7 5 8 5 15 10 24 27 22 22 46 49 20 g s 231 171 38 37 — 55 — 65 —52 -45 13 14 — 1 1.5 1.5 .5 3.5 2 5 17 5 18 2 35 7 42 MEASUREMENTS OF TWINS 1 I, Cm p t gw P 3 S5 bo bo We CO 1 O M a E 2 1+2 1 1 1 2 1+2 21 I- -SIS —15 —37 ■ — 95 — 70 —27 —30.5 —22 — 9 16 —3.5 —6.5 — g.5 - 6 — ii — 6 —25 — 7 — ^ lO CD to lOtO to eo coco + + ++ ++ g!3 ++ 90uujB8ddy lOIOQ 9^3 001 nv X9g 001-06 nv 93in« nv SPECIALIZATION OF RESEMBLANCE 61 Explanation of Table 22 This table gives facts which demonstrate the specialization of resemblance, by showing the existence of great and little resemblance in the same pair. The numbers beside the Dev.'s and Dif.'s at the top of the table give the median deviation of the whole group of 100 oases from the central tendency in the trait in question and the median difference between twin and twin of the same pair in the trait in question. The entries in the table give for each pair in each trait chosen the r for that pair, the deviations from the central tendency of the two numbers of the pair and the difference between them. If then a pair has a low r and a large difference or a low r and one large deviation, the r is surely significant of a really small resemblance. If a pair has a high r and a small difference relative to the deviations, the r is surely significant of a really great resemblance. Thus we may read off the facts concerning pair No. 5 as follows. Alike in sex, closely alike in hair color, very closely alike in eye color, very much alike in general appearance (nearly or quite indistinguishable), much less alike than twins in general in height, much less alike than twins in general in height sitting, very much more alike than twins in general in head circumference and in the combination of perception tests ; or more briefly, alike in sex, 90-100, 100, 95, 97, 64, 99 and 98. In these 13 pairs, — those most alike of all the 39 in hair color, eye color and appearance, — ^we find in every case resemblances surely below 90, six cases of less resemblance than would be found in children of the same age taken at random, and fifteen eases of decidedly less resemblance than the aver- age of twin resemblance. At the same time we find in other traits as great a resemblance as that in eye color and appearance. § 23. The Interpretation of the Facts The facts presented in section 22 prove that the general tend- ency to resemblance of any two individuals can not be easily and surely estimated from a few traits. Obviously, we can not predict from the amount of resemblance in any one trait, the amount of others, except within limits, or as to a certain degree probable. They are also strong evidence that heredity is itself highly specialized. It is perhaps possible that the variations in resem- blance are due to environmental forces, that heredity works simply by giving a certain degree of all-around likeness. However, it strains all one 's biological conceptions to endow environmental forces with the ability to alter original tendencies in eye color or cephalic index, or to suppose that two twia children meet, before the ages of nine to fourteen years, with environments different enough to make original close similarities in perception, association or ratio of height sitting to total height, change into comparative unlikeness. The only objection which can rightly be made against the hypothesis that the specialization of resemblance is due to the specialization of inheritance is that it demands so much complexity and specialization of the germs. But this is, after all, a question of fact. The germs are surely complex enough to parallel the structural traits of a species, including the anatomical and physiological basis of its re- 62 MEASUREMENTS OF TWINS flexes and instincts. They may be, in addition, complex enough to parallel the traits of an individual. It seems advisable, therefore, to choose, for the present, the hypotheses (1) that original nature is highly specialized, (2) that the causes determining the original nature of one member of a twin pair are largely, but by no means altogether, identical with those determining the original nature of the other member and (3) that the causes producing likeness in one trait are far from being identical with, or inevitably accompanying, those producing likeness in other traits. Evidence of the first hypothesis has been found, I believe, by every student of heredity who has used quantitative methods and been conscious of the problem itself. The second and third hypoth- eses, which support and make more definite the first, rest in the ease of twins upon only the data of this section. It is hardly neces- sary to can attention to the fact that if in a group of pairs chosen from those most closely alike on the whole, such specialization of heredity occurs, it will a fortiori occur in less similar twin pairs and in ordinary siblings. CONCLUSION § 24. Summary of Results This investigation proves the existence of close similarity of twins in physical and mental traits and gives approximate measures of the resemblances in eight physical and six mental traits. It shows that such likenesses and differences in environment as act upon children living in New York City and attending its public schools are utterly inadequate to explain the likenesses and differ- ences found in the traits measured, and are in all probability inade- quate to explain more than a small fraction of them. The arguments concerned the lack of differences in the amount of resemblance (1) between young and old twins, (2) between traits little and traits much subject to training and (3) between mental and physical traits, and also the great increase in resemblance of twins over ordinary siblings. The resemblance of twins was found to be approximately .80 or .75 to .80 in amount. The form of distribution of twin resemblance seems to be that of a fact with a central tendency at about .80 and with a great variability, restricted towards the upper end by the physiological limit of complete identity. Such a distribution would be most easily explained by the genesis of twins as a rule from two ova and by a great reduction of the variability of contemporaneous germs and ova below that of germs and ova developed at different times. The resemblance of any pair is far from uniform as we pass from trait to trait. An almost necessary inference is that heredity is itself highly specialized, each minute feature of physical and mental make-up possessing its representative in the germs and varying more or less independently of other features of the same germ. * Special care has been taken to so arrange the material that any investigator may readily combine the measures of this research with data obtained by himself or use them for studies of individual dif- ferences, correlation and the like ; and that any critic may repeat the calculation of resemblances after any plan that he approves. Noth- ing need be accepted upon my authority except the original gross measures and the details of the estimation of central tendencies from 63 64 CONCLUSION which to calculate deviation measures. The original measures are given for each individual in Tables 5 and 8 ; the deviation measures which are the basis of all correlations and later arguments are similarly given in full in Tables 6 and 10. The derivations of central tendencies and variabilities would have been printed in full, except for the expense and the very slight gain to the critic. SPECIES AND VARIETIES: THEIR ORIGIN BY MUTATION By HUGO DE VRIES Professor of Botany in the Univeridty of Amsterdam. Edited fay DANIEL TREMBLY MACDOUGAL Assistant Director of the New York Botanical Garden. Cloth, gat top, xviii + 847 pages. Price, postpaid, $5.00 (2Js.). " Prof, de Vries may well be regarded as the foremost advocate of experimental evo- lution, the man, moreover, who gave us the mutation theory of organic evolution. The volume before us is a splendid scientific plea for the experimental study of organic Mte." —Scientific American. "It will be seen that Professor de Vries has made an important contribution to science and thought. It is a masterly work and it is attractive because of the popular style in which the experiments are described and the conclusions stated. Every student of science will welcome it." — The Scotsman. THE OPEN COURT PUBLISHING COMPANY J322 WABASH AVENUE, CHICAGO EDUCATIONAL REVIEW ANALYTICAL INDEX To Volumes 1 to 25, January, 1891, to May, 1903 By CHARLES ALEXANDER NELSON Head Reference Librarian, Columbia University Library. 8vo, doth, $3.00. By mail, $3.J2 " The fortunate libraries and individuals who possess complete sets of the Review will be grateful for this work, which now transforms a row of bound volumes into a veritable encyclopedia of education, ready for consultation upon almost every imagin- able modem, educational problem. Analysis by topics is the leading feature of this index, and the work has been intelligently performed. The work is a dictionary cata- logue (author and subject), extending to upwards of two hundred double-columned pages." — The Dial, Chicago, 111. EDUCATIONAL REVIEW PUBLISHING COMPANY COLUMBLfV UNIVERSITY, NEW YORK EDUCATIONAL PSYCHOLOGY By Professor EDWARD L. THORNDKE In this book Professor Thorndike applies to a number of social and especially educa- tional problems, the methods of exact science. The topics are treated in the light of the most recent researches and with the aid of modern statistical technique. The book thus provides those iaterested in education as a profession or as a feature of American life with a sample of scientific method in this special field as well as with important infor- mation which has hitherto been inaccessible. The attitude of the author, who is the head of the department of Educational Psychology in Teachers College, Columbia Uni- versity, and the author of numerous original contributions to dynamic psychology, is that of a candid and painstaking student of the work that has been done in this field and upholds rigorous ideals of scientific accuracy and logic. The book is so written and illustrated as to be readable and teachable. LEMCKE & BUECHNER, 8t2 Broadway, New York standard Boo ks on Psy chology, Etc. By Pfofcssof GEORGE STUART FULLERTON Professor of Philosophy, Columbia Univcrnty A System of Metaphysics Those who have been attracted by Professor Fullerton's articles from time to time will be interested in his work as a whole ; only small portions of it have so far been printed. Cloth, 8vo,'$4.00 »i;/ (postage 19c.). By EDWARD BRADFORD TITCHENER, Ph.D Sage Professor of Psychology at Cornell University. A Primer of Psychology Outlines, methads and results of modem psychology. Clotb, $1.00 net. An Outline of Psychology Third EdUion. Cloth, 8vo, $1.50 «f/'. Experimental Psychology A MANUAL OF LABORAt6rY PRACTICE. Volume I. Qualitative Experiments, in two parts. Fart I. Student's Manual. Cloth, $1.60 net- Fart II. Instructor's Manual. Cloth, $2.50 net. Volume II. Quantitative Experiments. Preparing. By JOSIAH ROYCE, Ph.D., LL.D., Harvard University. Outlines of Psychology An Elementary Treatise with Some Practical Applications. Cloth, 8vo, $1.25 net (postage 20c.). By GEORGE M. STRATTON, Ph.D., University of California. Experimental Psychology and its bearing Upon Culture. Cloth, $2.00 net. Postage 13c. Dictionary of Philosophy and Psychology Including many of the principal Conceptions of Ethics, Logics, ./^thetics, Philosophy of Religion, Mental Pathology, Anthropol- ogy, Biology, Neurology, Physiology, Economics, Political and Social Philosophy, Philology, and Education, and giving a Term- inology in English, French, German and Italian. Written by many hands and Edited by J. MARK BALDWIN, LL«D., Johns Hopkins University, with the co-operation and assist- ance of an International Board of Consulting Editors, In three volumes. Vols. I. and II. now ready, each $8 net. When net books are ordered from the publishers, carriage Is UDiformly an extra charge. THE AmeMILLAN eOMPANY, Publishers 66 Piltli avenue, New Yorli THe date shows'when tliis volume was taiken. ' ...,, T<(reoei»tMs; book copy tlt« call No. and' give ^o ,,. :•- , ■:'~~^~ ..■'[, the librariati. - ^__ ""''" ~^r r HOME USE RULES ~ All Books subject to Recall. . Books not used for instruction or research, are returnable within' 4 weeks. Voluides of periodi' ■ cats and of pamphlets are held in me library as* much as possible. ■ For special purposes they are given out for ^_ limited time. " Borrowers shouj^d not use their, library', |)rivileges f orthe benC'^ fit of other- personfs:' Books not needed during recess jperipds! , should< be returned to the library, or arrange- ments made for their • return during borrow- er's absence, if wanted. Books needed by more than one person are held on, the reserve list. Books of special, value and gifjt b6oks, when the giver wishes it^are *not allpw«d to circulate. Readers are asked to report, all cases of books marked or muti->. - lated.; -< Do hot deface books by" marks and ^rltliie<.