er orksoe i eee & bs rece tats Naseripbestemere: ctaeatat e Pesan geoceetatshe Mahaaat ate pipietrs ma ah Tatet Soieny coyensetecten Acai et » en nao’ ppredel A Fanon F fs i i ace s Pet sive evs : at , i Ped cassancints spies sp nyte str eS) me , a “ ge Sey ean ‘ ve vg Mayes Siemaeanh eee ba se rae OY hee Nha ek este} c ee ee ey Sb ging net tee “ae 2 oe hy i weve Pies Dy te a En by ta Bagh eS iat giipin Coto ae hatte Fine sh Getty NP at 0 Fats Pee ay ¢ Am stan : ae iia cea ed Sup se Es he Ap tee ga Thee + OL A teteee es treet =: VOL. XXXII psYCHOLOGICAL REVIEW PUBLICATIONS WHOLE NO. 152 j ’ 6 RES ot % : q C ‘ F ' ‘ . ~ ter o.@r9 ' » L. (ea - ' . ‘ ‘ f a ’ nv ay + . ’ ’ S -t ’ ; i" 7 . Ae a Se Ose ie ret. o.s0e 4 a 4 =| , oe ee ae GR STi ae sp fieatl MS ; ie Yale eh I 7! a a | ; i] \ = tel +7 in ‘ , ee : ¢ ‘ i ‘ _ ; “ , ’ ¢ aA : 5; 4 Brads Fr Ye 4 Pow eae oe, . i eT Vee ily ve uy id eee i bs rn ‘ a . a? oe INTRODUCTION At least three important biological problems are connected with the phenomenon of twinning: (1) The problem of twinning as a variation in the normal method of human reproduction. (2) The problem of sex determination. (3) The problem of the relative in- fluence of hereditary and environmental factors in human develop- ment. In this study no attempt will be made to do original work on the strictly physical aspect of the problem. We shall be satisfied with stating fairly the generally accepted biological views, and pointing out some of the implications of these views. The purpose of the study is primarily psychological and bears chiefly on the following questions : .I, What is the effect of environment upon the amount of in- tellectual resemblance of twins? 2. Does the fact of twin origin and birth operate in any way to lower the intellectual level of a twin population? 3. What light do the psychological data throw upon the current biological belief that there are two distinct types of twins, fraternal and duplicate? Le er By ad al re ERD * Dat eae. eter SAE Hee ad an ea A ~ . 7 ‘ y : ¥ “ate i 4 . Tle i 1 : << .« a é F ; ‘ pct - : ang , ie i ? tte Ky e i Ps : POSURE PE ae tae ae Mant Seth oF ad A if Yd 7 + , ‘ ee: & viet Am ed ' iS we twas. eH Rie ne wil eS: ui , ; Bada Si bia \y 5 a ee rd it + ot Mooi i i) onde! 7 ae Ahi 2 “fs nb Ferd Fi) yb ibs Wass + thas Aes rats Ais ic ea 7 ‘*é Ly . rs 4” j ’ * ~* i 5 1 al * 4 > - 7 ! Us? - 4 : \ , 2) - Ks ave eo 5 - + Cy on - of ; ae Sal of atts a 419 a6ce 5 PAL ay ane 0 ne 4 es . P s i ® es a ; > i * 5 ge 5 ‘ ‘ at tt 5s fea the DI 3 yr P ae rs . “a ak ’ ’ a. , oe eed Ss See 44 brat { : a we an ' i i. Ne Pet 4 . 7 a 7 o ee * oe ‘ ~ - 7 AD ‘ wl suck , i“ in i. + : PA! Fe / ee : sop has te ee yee Pe mks) -# ese a ore ce te ; aNd wa r ‘ h / ir? | : eal OR ae rd dese ae é > rr? ’ 74 oA 2 a Oho ed os igs ons “ . ere ia cpa a ae ie ae aly att Peas is {hs GUE PBT eri eae We hs. > *¥ ‘se vay { ' + ' 7 aed eile ty , J a , bie . ‘ ‘ Lf j A ‘ 5 ‘ - ’ : t i i] J + ay | ie i : ' i ’. hl 4) i} } ‘ie ty ‘ , Hi { ) of : ' : bf eo 4 ' } f eR ‘ iy ye fr ‘ Jj i% ’ . ( 4 : it an} 7" ’ aT ' ‘ ’ Ly ae J "5 4 e 7 ‘ae , Ld ‘ > a5 ‘ i 7 a: Sa as | 5 Pere eCOLEECTION OF PALA Three possible sources of data presented themselves for con- sideration. First, it was necessary to secure data on the intellec- tual behavior of each twin when he worked alone. It was immedi- ately evident that the Stanford-Binet was the most desirable in- strument for this purpose. Second, it seemed desirable to have some kind of estimate given by some one who knew the members of the twin pair intimately. For this it was decided to obtain a teacher rating on a number of traits which are commonly accepted as primarily intellectual in nature. Third, it was desirable to sup- plement both of the above by group tests. For this purpose two types were used, The National Intelligence Test, and a modified form of the Army Beta. The latter was used to get as far away as possible from the verbal factors represented in the other tests. The following data were obtained: i, otantord-Binet tests for@s7 02... 105 pairs om, VeAbuen Estates 1OT)% tercakts deat. gO pairs mea finvebeta tests LOT, ys. Mees. ts 76 pairs 4. National Intelligence tests for...... 143 pairs In the collection of the data, every known precaution was taken to insure the validity of the results. Of these precautions the fol- lowing may be mentioned: 1. All tests were given by trained examiners. Besides having ~ studied the books and manuals on testing, each person had given a considerable number of tests under the personal supervision of a psychologist. It is believed, therefore, that the procedure was as nearly uniform as it was possible to make it. 2. In almost all cases the Stanford-Binet test was given to the two members of a twin pair by the same person, the test of the second twin following immediately upon that of the first. 3. All group tests were given to both members of a pair at the same sitting. 4. Extreme care was taken to make sure that the children were 4 CURTIS MERRIMAN actually twins. Strangely enough, two cases were found of chil- dren who were passing as twins, but were not twins. 5. The twin population tested was limited to those found in the eight grades of the elementary schools. This was done because of the inadequacy of some of our tests when used above or below cer- tain age ranges. For example, the National Test was not designed to measure beyond a level represented by the brighter children in the eighth grade. The limitation of the survey to the eight grades also made it possible to avoid certain undesirable selective factors. 6. Every possible effort was made to secure data upon every twin pair in a given school population. All the schools included in this study are co-educational. This made it as easy to locate twins of unlike sex as twins of like sex. Extreme care was taken on this point. The school principal or city superintendent cooperated by calling teachers’ meetings. Statements were made before the entire school. Diligent inquiry was made of the children themselves on the playgrounds. It was announced in the newspapers that a. search was being made for twins. It is believed, therefore, that factors which could have produced a systematic tendency to overlook cases which were not in the same grade, or which did not resemble, were pretty completely eliminated. In a later part of the study, data will be presented on the number of twins that appear in a general population, and on the relative number of like sex and unlike sex pairs. It is interesting to note that in the populations covered by this study the actual number of twins found agrees closely with the observed frequency in the general population. The same is true ~ as regards the relative number of like and unlike sex pairs. These facts give added weight to the statement that there was present no systematic tendency to overlook any cases. Before presenting the tables which give the results of the various tests, it is necessary to call attention to the test procedures em- ployed. The procedure used in the Stanford-Binet tests and the meshed of recording responses conformed strictly with the directions set forth in Terman’s “The Measurement of Intelligence.’’ All items of the tables are self explanatory except in the starred cases show- THE INTELLECTUAL RESEMBLANCE OF TWINS 5 ing different chronological ages for the two members of a pair. This resulted from the tests being given at different times. The National Intelligence Test was given and scored according to the 1920 edition of the manual. ' The procedure used in giving the Army Beta test was somewhat modified. In its original form this test was given by pantomime because it was designed for use with non-English speaking men. Inasmuch as all the twins could speak and understand the English language, the following verbal form of instructions was used. Only the verbal instructions will be shown. The charts that were used were similar to the actual test figures, and the general pro- cedure was much the same as is used with the exercises of the Na- tional Test. The content of the tests and the method of scoring were exactly the same as for the original Army Beta. BETA TEST Arrention—Pencils up. I am going to show you some drawings like examples you will have to work in this examination. I will illustrate the examples, and you are to do all the examples in each test. Always, pencils up, when I say stop. bio La ATTENTION—Pencils up. Look at this drawing, marked Test I. Here are some alleys. Watch me as I try to run through them as quickly as I can without crossing any lines. I am going to go from this arrow to this one and must find - and mark the path. Now—Look at your papers. Test I. You are to trace the path through them as fast as you can, and be sure not to cross any lines, but find the way out. Go—Stop. (2 minutes) ge Th | ATTENTION—Look at this drawing. Here are some blocks piled up. Watch me while I count them. Then I put the 3 in the square here. Watch while I count these. Then I put the 9 in the square below. Now—Look at this drawing. Test I]. You are to count the number of blocks in each drawing as if you could see them all, and put the number in the square, Go—Stop. (2% minutes) TESF IIT ATTENTION—Here are some little circles and crosses. You see the first row with the circles. Some of the spaces are empty. I fill the spaces thus O, O, O, until all the rest of the spaces are filled. (E repeats with other rows.) Now—Look at your papers. Test III. You see the crosses and the little circles. You are to fill out the blank squares at the end, as the row has been started. Do all the rows. Go—Stop. (134 minutes) 6 CURTIS MERRIMAN TEST IV ATTENTION—Look at these drawings. Here are some numbers with drawings below them. Below are rows of the same numbers with empty squares below them. Watch what I do. I put the right drawing with the number to which it belongs. Go all the way through each row. Now—Look at your papers. Test IV. You are to put the right drawing in every blank square below the number which it goes with. Go through every row to the end. Go—Stop. (2 minutes) TESTA. ATTENTION—Look at these figures. Some of the numbers on each side are the same. Some are almost the same but not quite. Watch what I do. Now—Look at your papers. Test V. If the two numbers are the same put a cross on the dotted line. If the two numbers are not the same do nothing. Go—Stop. (3 minutes) LES Lav ATTENTION—Look at this drawing. There is something missing. What is it? (Gets response and draws in figure.) (Repeat.) Now—Look at your papers. Test VI. In each picture draw in the part that is missing. Don’t try to make it pretty. Fix them all. Go—Stop. (3 minutes) TEST VII : ATTENTION—Look at this drawing. I put the figures into the square like this. They fit. Watch where the lines should be drawn to show how they fit. Now—Look at your papers. Test VII. You are to draw the lines which show how the little pieces fit into the squares. Go—Stop. (2% minutes) THe viel As this test was omitted in army use, it will not be used in this study. The teacher rating was secured by. means of a special blank of which the following is a copy: In each trait or characteristic named below compare this child with the average child of the same age. Then in the square before the name of the trait, place the figure I, 2, 3, 4, or 5. These figures are to be used with the following meanings: I—Very Superior to the average child of this age; 2—Superior to the average child of this age; 3—Average; 4—Inferior to the average child of this age; 5—Very Inferior to the average child of this age. ( ) Memory. ( ) Imagination (Ability to think about things not present to the senses.) ( ) Reasoning (Ability to see meanings or to follow a complicated train of thought. ) ( ) Judgment or common sense. C- THE INTELLECTUAL RESEMBLANCE OF TWINS 7 ( ) Resourcefulness in overcoming difficulties or attaining ends. ( ) Originality (As shown by inventiveness or by ingenuity in finding ex- planations.) ( ) Curiosity (As shown by inquisitiveness or eagerness to learn.) ( ) Mechanical Ingenuity (Ability to think out mechanical contrivances. May exist without manual dexterity.) General Intelligence. Studiousness. Interest in objective things (plants, tools, etc.) Interest in books. Breadth and Variety of Interests. we Ne ee a” a The recorded grade for the teacher rating in Intellectual Traits is the mean of the various ratings. It should be borne in mind that the rating I is highest and 5 the lowest. The test results were then tabulated according to the following form: SEX GRADE CA. BINET BINET TEACHER BETA MT M.A. 1.Q. RATING Pains! Boy. tT 5-10 6-6 112 F I 5-10 6-8 114 sets rhe 3 Ak Paite 2 Meee 2D 6-11 7-6 108 1.5 30 28 F 2b 6-11 7-4 106 DD 45 44 Table 1 shows the distribution of ages, the number who took the various types of tests and the total number of pairs studied. Appendix A gives the results of the various tests. The original scores are reported so that any one who cares to do so may make further study of the data. TABLE 1 GENERAL SUMMARY OF DaTA AGE BINET TEACHER BETA N.LT. 5 I 0) re) 0) 6 9 8 3 I Gi II 8 7 2 8 II II 7 14 9 15 12 II 21 10 9 9 8 16 II 10 7 7 17 12 15 12 12 26 13 10 10 II 21 14 9 9 8 13 15 3 3 2 10 16 2 I fe) I 18 fe) 0 (a) I Totals 105 90 76 143 THE EFFECTS OF ENVIRONMENT Galton made a general comparison of two groups of twins (Galton, Francis: Inquiries into Human Faculty, Everyman’s Li- brary, 1883, pages 155-172.) One group consisted of 35 pairs showing marked similarity in infancy, the similarity being some- times so pronounced as to cause confusion of identity. The second group was composed of 20 pairs of distinctly dissimilar twins. For the second group the environment had remained substantially the same. The excess of difference in the first case, and of resemblance in the second, was thought to give a measure of the influence of en- vironment. The persistence of similarities in the first case and of differences in the second was taken as a measure of the influence of nature. Galton quotes from many letters of parents showing how the original likeness or difference remained through child- hood to adult life. He summarizes as follows: “We may, therefore, broadly conclude that the only circum- stance, within the range of those by which persons of similar conditions of life are affected, that is capable of producing a marked effect on the character of adults, is illness or some acci- dent that causes physical infirmity. The impression that all this leaves on the mind is one of some wonder whether nurture can do anything at all, beyond giving instruction and professional training. There is no escape from the conclusion that nature prevails enormously over nurture when the differences of nur- ture do not exceed what is commonly to be found among per- sons of the same rank of society and in the same country.” (p. 172) At least two comments are in place concerning Galton’s work as briefly outlined above: (1) He secured his results by the question- naire method—a method which investigators in the field of verbal report have shown to have many sources of error. (2) In spite of the imperfections of Galton’s method, his general conclusion as to persistence of nature has been fairly widely accepted. THE INTELLECTUAL RESEMBLANCE OF TWINS 9 In 1905 Thorndike (Thorndike, E. L.: Measurement of Twins, Archives of Philosophy, Psychology, and Scientific Methods, Number One, September, 1905.) published a report of the meas- urement of resemblance of fifty pairs of twins in certain specific mental traits. The traits used and the resulting resemblances were as follows: 1. Marking A’s on page of capital letters....... t= 09 2. Marking words containing at or re ......... i 7k Beevarkine misspelled ‘words: oii. cise peiebie het s i. OO As povine addition: problems sox ek. ieciheidtons ele’ ti 275 5. Solving multiplication problems ............ To=.54 6. Writing the opposites of a set of words...... jer a8 8, Having found the resemblance for the twin population as a whole, Thorndike attacked the problem of the effects of environ- ment. His argument is as follows: “Tf now these resemblances are due to the fact that the two members 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 Io years old. Again, i 2. If similarity in training is the cause of similarity in mental traits, ordinary 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 conditions 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 resem- blance in the case of traits much subject to training, such as ability in addition or in multiplication, than in traits less subject to train- ing, such as quickness in marking 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 equal- ing that of old, and 10 CURTIS MERRIMAN 2. The greater the superiority of twin resemblance to ordinary fraternal resemblance is, and 3. The nearer twin resemblance in relatively untrained capaci- ties comes to equaling that in capacities at which the home and the school direct their attention, the more must the resemblances found be attributed to inborn traits.” For the detailed statistics, methods, and conclusions of Thorn- dike’s study the reader must go to the original monograph. The following quotation will suffice to show how Thorndike takes his stand alongside Galton on the general problem of nature and nur- ture. “The facts then are easily, simply, and completely explained by one simple hypothesis : namely, that the natures of the germ cells— the conditions of conception—cause whatever similarities and dif- ferences exist in the original natures of men, that these conditions influence body and mind equally, and that in life the differences produced by such differences as obtain between the environments of present day New York City public school children are slight.” However, neither the work of Galton nor that of Thorndike is entirely conclusive. In the first place, Galton depended upon a verbal report method. Thorndike knew the weakness of this method and made use of a series of mental tests, but tests of a kind far inferior to those at present available. Later experimental results have considerably discredited most of the tests he used, as far as the measurement of general intelligence is concerned. It is to be stated, however, that Thorndike made no claims for these | tests as measures of general intelligence. He plainly says he is re- porting only the results of the measurement of 50 pairs of twins in these specific abilities, viz., marking A’s, etc. He plainly states that he makes no claims as to what might be found for other mental functions, except as we may infer from the probably reign- ing likeness between abilities. In the second place, Thorndike doubts whether he has completely eliminated the possible effects of home and school influence. The Binet and other forms of intel- ligence tests appear to meet this requirement more satisfactorily, and are therefore used in this study. Because of these differences in approach, it could not be fore- oP ea“ THE INTELLECTUAL RESEMBLANCE OF TWINS II seen whether the results of the present study would or would not support the conclusion of Thorndike. Both because of its nature and its careful standardization the Binet test ought to throw new light on the intellectual resemblance of twins. For similar reasons the various group tests should also be of great service, especially in making it possible to deal with a large population in a relatively short time. Before presenting our results it is necessary to indicate the statistical procedures employed. All correlations were derived by the use of the Pearson product-moment formula. The tables will show the correlation coefficient, the probable error, and the number of cases (pairs) used in the calculations. The scores used in these calculations were as follows: 1. For the Stanford-Binet correlations, the Stanford-Binet I.Q. as shown in Appendix A. 2. For the Teacher correlations, the ratings as shown in Ap- pendix A. 3. For the Beta correlations, the raw Beta scores as shown in Appendix A. 4. For the N.I.T. correlations, the raw N.I.T. scores, as shown in Appendix A. Table 2 shows the coefficients of correlation based upon the Stanford-Binet I.Q. for the various groups of young pairs as com- pared with the old pairs. TABLE 2 1. All pairs, 5 to 9 years, r = 809 + .032 with 47 pairs Mlnapaire TO tOnsO years, rs..757 21.037 i aso bea Difference = .042 + .048 2. Like sex, 5 tOn GO (Years; f=, GOs. 7 anne ar edn bikes sex, 15°10 t0-16' years, ‘ric 865 cee 0272 “54.4877, Difference = .017 + .038 3. Girl-Girl, Beto 0 year sy fa .018 1 Oc ne 1G, 0 4 Gusi-Giri,, 10 to. 16:;years,'r — .o14/2.050,, 2 21, “ Difference = .101 + .056 4. Boy-Boy, Sf 80 years, T0007 els amen a 10 Bey iye. 1 10.tot years, © =5". 8008 8074 ay ar via Difference = .090 + .085 Se Umike-sex,.+8 fo7'0 ‘years, £774 1064 a8 Unlike Sex10 to.10 years, r =*.208, -=.137) " 20) * Difference = .476 + .151 12 CURTIS MERRIMAN As far as the effects of age on twin resemblance are concerned, the present study confirms the conclusions of Thorndike. For all twin pairs with ages ranging from 5 to 9 years, inclusive, the cor- relation is +.809. For pairs with ages 10 to 16 years, inclusive, the correlation is +.757. The difference .042 is scarcely more than the P.E. of either measure. The P.E. of this difference, .048, also shows that no statistical significance can be attached to the change in correlation. In the case of the like-sex pairs the change from the young to the old pairs is —.o17; in the case of the boy-boy pairs +.090; and in the case of the girl-girl pairs —.101. In each case the P.E. of difference shows clearly that the slight change in cor- relation cannot be interpreted as indicating a difference in twin resemblance due to age. The results of the measures of the unlike sex pairs are not so clear. The drop from .774 to .298 is almost five times as great as is found in any other group. Three possible ex- planations present themselves. (1) Environment may actually op- erate to cause twins to grow more and more unlike. We have just seen, however, that environment has no such effect upon the other four groups studied. It is therefore reasonable to question the validity of this explanation and to look for some other explana- tion. (2) There may be inherited differences or likenesses that de- mand maturity to make them evident. This explanation, however, fails for the same reason that the first one did. Moreover, to accept this explanation would necessitate accepting the idea of a con- stantly changing I.Q. This is a debatable point, but the writer be- lieves that the preponderance of available evidence is in favor of the theory that the I.Q. remains relatively constant. Again, to ac- cept this explanation would necessitate showing why the change is a negative change rather than a positive one. (3) The change may be explained statistically in terms of the small population. That this explanation is probably the correct one, is suggested by two facts. The difference in correlation, .476, is scarcely more than three times the P.E. of the difference, .151. The best evidence, however, comes from the Beta and N.I.T. results which are to be shown later. For these tests the sharp contrast does not occur. It therefore seems reasonable to conclude that the statistical explana- tion is the most plausible one, and that as far as the Stanford- eae * THE INTELLECTUAL RESEMBLANCE OF TWINS 13 Binet results are concerned there is no valid evidence that twin re- semblance becomes greater the longer the identical environment lasts. Table 3 shows the results of the Beta tests. TABLE 3 1. All pairs, 5 to 9 years, r = .784 + .049 with 28 pairs Allepairs, 110 to 1Oyears, r= 064 co-osd tin a Ae Difference —= .120 + .072 2. Like sex, CAtOmeOuVeal Sa ba) bac Oot mace TO Tike sex, — 10 to°16 years, f° .o42. .030. 20 Difference = .079 + .043 3. Girl-Girl, SrtOMmMOnyearse t <9 7001--ae Tl 2) a Or Be Girl-Girl, . +10 ,to/16. years). r == .806 == 032, “ . 16. “ Difference = .187 + .116 4. Boy-Boy, Etto Or yeats, t=2.034°- .040/) yin Base Boy-Boy, FONCO BIOL YCarsatn— <7470 = g.0G0) me ante t3 | ve Difference = .087 + .093 en Uphike Sexe Sto O-years, fr == 510 ch .1470 “ ‘12. “ Unlike Sex, 10 to 16 years, r= 643 = oor “ 19 “ Difference = .124 + .172 When the methods already used in the examination of the Stan- ford-Binet data are applied to the Beta data, the same conclusions are justified. The differences between the younger and the older groups are either very small or can be explained statistically with- out necessitating the assumption that environmental factors have been operative. TABLE 4 1. All pairs, 5 to 9 years, r= .797 + .034 with 54 pairs All pairs, #141 tOl IS years, £i5—0.075 22 O17 189 Difference = .078 + .038 2. Like Sex, Ertonroryearsars—-,040 +) OT2 bi 31. pce SOX at bi tO 16, veatsnf sa) O05 oo 4.022" 2 OT Difference = .o81 + .025 3. Girl-Girl, KR toss years tesa O05) 2c..000.. i, 24 sre bitisGitl, © / 1. to 1S yearset—) 019 1.021. 937 4.“ Difference = .046 + .022 4. Boy-Boy Sto. 10 years yf ",021) 04E) vy Yip Bee: Boy-Boy, ET ALOR TOV V CALS pita i055 O27 ee tee Difference = .026 + .049 SPrmiike: sex 06 tO 10.years.Y = .76% = 000) 2s. (alike sex, 1 lvto 18. years; f= 834/25 Sodas > oR ae Difference = .0o81 + .079 14 CURTIS MERRIMAN Table 4 shows the results of the National tests. The reader will _ note that the age ranges are different from those in Tables 2 and 3. This change was made because the National Tests were not used below the third grade level. We find that the conclusions drawn from Tables 2 and 3 are fully supported by the data in Table 4; there is no evidence of any age difference in the degree of resemblance. Table 5 shows the results of the teacher ratings. TABLE 5 1. All pairs, 5 to 9 years, r = 686 + .057 with 39 pairs All paits,.) (10 to.16 yearsora— 373 = Osi Tha Difference = .313 + .099 2. Like sex, “to “Osyeats,.t Ors (= <= 20; CASES Difference = .363 + .083 Beta Like-sex f==.008 = 20177.= ¥-" 45. cases Unlike-sex Gace 732s O50) =) =. ST pcases Difference = .176 + .058 IN ad bad B Like-sex ft = .925 = .000 = > = 02 cases Unlike-sex r = 67 + .025 - - S51 cases Difference = .058 + .026 Teacher Like-sex Trs=".054 57.053) =. = 63 Cases Unlike-Sex Te 1.200 == 1025) - re Cases Difference = .388 + .114 Since it was pointed out in the proposed argument that this would be a vital point, let us examine these results more in detail. 30 CURTIS MERRIMAN In the Binet results the like-sex resemblance is .867 while the un- like-sex resemblance is .504. This is a difference of .363 + .083 in favor of the like-sex groups. The Beta resemblance shows a dif- ference of .176 + .058 in favor of the like-sex groups. The N.I.T. difference of .o58 + .026 is not as large as the others, but since the difference is in favor of the like-sex pairs, and since a small dif- ference between high r values may be quite significant, it can hard- ly be argued that the N.I.T. evidence contradicts the Binet and Beta evidence. The teacher rating results show a difference of | .388 + .114 in favor of the like-sex groups. All of the evidence thus far, therefore, supports the biological claim that there are two distinct kinds of twins. This conclusion was reached by a study of resemblance in terms of correlation coefficients. There is still another way in which the same data can be treated, viz., in terms of differences in gross scores. Largely because of the size of population available, this part of the study will be based upon the Stanford-Binet I.Q. and the National Scores. Table 9 gives the distribution of arithmetical difference in Binet I.Q. scores for the four groups studied. The data of Table 9 can be studied in two different ways. First, _ we can calculate the mean 1.Q. difference for certain types of population. The difference between these means will be a measure of the homogeneity of population. The results of this plan of at- tack are as follows: Mean 1.Q. difference, unlike-sex pairs = 9.52 + .85 Mean I.Q. difference, like-sex pairs == 6.05 + .45 Difference == 3.47 + .96 Since the difference is 3.6 times its P.E. the evidence from this direction favors the theory that we have to deal with two types of population. The data of Table 9 may also be treated graphically as shown in Figure B. It is seen at a glance that there is a decided difference in the types of curves. The siblings and unlike-sex curves resemble in a general way. Each starts with a low frequency for zero differ- ences in I.Q. This agreement seems to lend some graphic proof for the biological claim that unlike-sex twins have the two-ege ee tested i 18h] JRE ms LAN Ae tee ty (ae THE INTELLECTUAL RESEMBLANCE OF TWINS 31 Bag Pt att prrereners wnat FETT TT ET TE esa siuincs. | {| | | | | tt ttt i tt waiay + |e a ietaneriod Pps eT Lal tebe Ie teh Pn lee Te | RT Mie Chaka eas WAT | I B ie ee Ne ice Re pale mila ee ol Pee N INS eae [| ‘ FECES NSE Pepe siete Ss G6 ¢ 6 9.10 WM 2 1s 4 AS 16 17 18 19 26. 21 22 pa polar scariae es nc ai 32 33 34 355 36 37 - FIGURE B. origin in the same manner that siblings have. On the other hand the like-sex curve starts with its highest frequency for zero dif- ferences. This would seem to conform to the biological claim for “duplicate” twins. | Similar facts are revealed by the N.I.T. data, as shown in Figure C. There is one difference to be noted in the construction of the curves of Figures B and C. In Fig. B the separate scores are plot- ted. In Fig. C, the scores are grouped as indicated at the bot- tom of the figure. This was done for two reasons. In the first place the score range for the N.I.T. data is much greater than for the Stanford-Binet. The relative frequencies are therefore smaller and it becomes difficult to see the form of the curve. In the second place, it is necessary in a later part of the study to use the grouped 45. = 5d 32 CURTIS MERRIMAN TABLE 9 DIFF. ALL PAIRS LIKE-SEX UNLIKE-SEX SIBLINGS fe) 9 8 I 6 I 6 6 (o) 12 2 10 6 4 8 3 9 5 4 II 4 10 7 3 13 5 9 7 2 9 6 9 6 3 12 7 2 2 I 10 8 6 3 3 13 9 3 3 fe) 6 Io 4 2 2 13 II 4 2 2 6 12 2 fo) -: 8 13 5 3 2 6 - 14 4 I 3 3 15 2 0 2 9 16 2 2 a) 6 17 oO ‘0 fe) 14 18 I I re) 4 19 I I Oo 5 20 (6) (a) 0. 6 21 2 I I 9 22 fa) re) oO 4 23 I I fe) 6 24 oO oO 0) fo) 25 I oO I oO 26 Oo Oo Cy) ag 27 ce) fo) fe) 4 28 oO fe) te) mie 29 O fe) ) I 30 0 oO fo) 3 31 I oO I) 32 0 Oo 0) 33 fo) oO 0) 13 34 I .e) 1) 40 oO re) 0) 41 to 50 o oO oO 2 105 67 38 234 form of the data, and for convenience the same grouping is used at this time. . If this distinction in type of curve is a real one it should be shown by the tests that have been devised for homogeneity and curve fitting. The best test for this purpose that the writer has been able to find is found in the work of Karl Pearson and W. Palin Elderton who have published a formula and complete tables for testing curves where two separate populations are concerned. The formula and its method of use can be seen from the following oan THE INTELLECTUAL RESEMBLANCE OF TWINS 33 DISTRIBUTION OF DIFFERENCE IN WN. J.T. 92 LIKE- SEX PAIRS ——-— 5! UNLIKE-SEX PAIRS o-s 6-20 21-35 36-60 €1-€0 81-100 (01-120 121-170 LIKE-SEX 7 16.3 42.4 26.1 10.8 INLIKE - + ILT 23.5 23.5 176 FIGURE C quotation: “If N and N’ be the sizes of two samples and the cor- responding frequencies: Pits alas eu eatene Tien aa wire ie ete APE AY S| Miele ha aS ee ey fc where f,, f’, are the frequencies falling in the p™ category, then if wes ae Bee cope is Nanci 09 ( ieee is) be calculated, the probability, P, that the observed or a greater di- vergence between the two series would arise from sampling the same population is obtained by determining P from X by my method of testing “goodness of fit.’ This method was first pub- lished in Phil. Mag. Vol. 50, p. 157, 1900. The shortest method of actually determining P is by aid of Palin Elderton’s tables for P 34 CURTIS MERRIMAN with argument X? issued in Biometrika, Vol. I, p. 155, 1902.” (Karl Pearson, Biometrika, Vol. 10, p. 92.) The following pages (35-36) show the application of the Pear- son formula and Elderton tables to the Stanford-Binet and Na- tional Intelligence Test data. Since the S-B data show a probability of 27 to 1 that the popu- lations are different, and since the N.I.T. data show a probability of 40 to 1 that they are different, it is clear that the biological claim for two distinct types of twins is Ley supported by the | data on intellectual resemblances. It was stated at the beginning of this study that the neoblen would be limited to intellectual resemblances. It is pertinent to ask, however, at this point what would be the result if physical resem- blances were studied by the same methods that we have been using. Thorndike gives on page 43 the correlations for a number of such measurements but he gives only the correlations for the twin group as a whole. He makes no distinction between like-sex and unlike-sex groups. His gross measures are given, however, in table 8 on page 37. The writer took these measures and calculated the resemblance in height and cephalic index. The results are as fol- lows: HEIGHT CEPHALIC INDEX a. 39 twin pairs c= o778 ee b. 9 unlike-sex pairs r = 609 + .176 if) ——yedOle toma c. 30. like-sex: pairs... r — 821 2 042 r = 800 + .045 Note: a = Thorndike’s computations; b and c = the writer’s. In b and c, 0745 (ite) P, E. = ————————— 0n account of small population. Vn — 3 It is hard to explain the difference in resemblance by any other theory than that of a real difference on the basis of likeness of sex. It must be admitted of course that the small population of the un- like-sex group and the resulting large P.E. reduces the value of the results somewhat but they certainly are in substantial accord with the results of the study of intellectual resemblances. Up to this point, the study has proceeded in terms of the entire group of like-sex versus unlike-sex pairs. The conclusions are quite definitely in favor of the two distinct classes. We can not be en- 35 THE INTELLECTUAL RESEMBLANCE OF TWINS ‘JUdIOJIp aie suoljejndod om} yey} I 0} saoueyo Zz A[YySno1 10 cbt Lto' = g saqey, uoyopyy Aq aousy AA +fQ 11 = GO VOU ea Chee eC eee tsgvoo: X ,NN = 2X pes > ate ( Ba oe! ) T eae CN N) pts sXe ( ) (545)/eN/J—N/J) 690700" $g0000° go00000" 10L100° $61000° $6S000° (8) == (£) + (L) 2UN/J—N/J) be6ozgoo | Szzb£ooo: | rboScoo0" | ocoszgofo: | g$LggSoo: | +#z6L19z0" | (ZL) st e(9) iN/J—N/J gglo" — 5g10° 100° osL1* — 99bor gigi (9) = (5) — (*) IN/ ggLor £gzo" gzSor b6gz" ggtz: gs 18 (5) = gf + (z) N/j 0000" grro: L6So° bbor’ be rf- glLr- “#(¥) = L9~ (1) $o1 St 5 14 9 gi of bb (£) = (%) + (1) gt * I zt II any: wl (z) XaS-o41[U) Lo j fo) £ + L It am (1) X9S-O4I'] NOILVINdOd HNVS AHL WOU SHTANVS auxV SNOILNGINLSIG ADNANAAAIG LANIG-GYOANVLS OML YAHLAHM ANINUALAC OL SNOLVINOIVO CURTIS MERRIMAN 36 ‘JUIIIIP aie suorepndod ay} yey} I 0} saoueyd oF AjYysno1 JO bzo' = g saqe yy, uoylopyy Aq souay AA + II‘'QI = EOVeVEOO els ec C6EEr"Loo’ XK WNN I a | ae ee ) ( Ase 3 ) T — ee Lie ae Gee 29) ) | | E6£€rEo0" | | | | | | | | (6) | = (8) (§445)/2.N/J—N/J)| 91? g£ooo" | 5gggzooo- | cogZ$o0o" | og6b1100° | zz1vZO00" | ozgro000" | 61g6g000" | 14L60000"| (g) | =(£)+-(4) :UN/J—N/J)| 91?g£000" | gSgogooo | cobofzoo: | 13961600" | 6z£gS%oo: | g£$Sgoo0" | 6gLog$£or | 6ozSozo0"} (4) | = ,(9) iN/J—N/J g610°-— +gt0°"— ogto"— 6560°— LLgo-— 95to° Lggr- £Sto: | (9) | =(S)—(+) UN/ J 9610" t6£o" ggsor gli" bolt" waft: aX a gliree=| (5) | == 15--(z) N/J 0000" gOIo’ goro" Lito" Lgor: gogz" 6fzr° 6zgI° (vy) | = 26+(1) a Jt I £ % 8 61 gf 1$ I (€) | =(z)+(1) 1$ ey I t £ 9 6 rat ZI 9 (t) | xas-aytTUG 76 J fe) I I Tt OI +z 6¢ $1 (1) XOS-O¥I'T abe Mies. [a olt-1e1--| ocr-iors | Poorzrg— | 08-19 es eed ES BS a eed | NOILV1NdOd ANVS AHL WOU SHTAWNVS AuYV SNOILNGINLSIG ADNANAAHIG “L T'N OML YHHLAHM ANINAALAG OL SNOILVTNOTVO THE INTELLECTUAL RESEMBLANCE OF TWINS 37 tirely certain of this, however, until we have made a more intensive study of the resemblance of the various pairs. Thorndike’s method of doing this is fully discussed on pages 45-46 of his monograph. “Our problem is to measure accurately the resemblance found in each pair of twins, and so to ascertain the form of distribution of the group with respect to resemblances. Suppose, for instance, that of forty twins, we found twenty to resemble each other prac- tically perfectly, the coefficients being: 92, 93, 94, 94, 94, 95, 95; 95, 95, 95, 96, 96, 96, 97, 97, 97, 97, 98, 99, and 99, and the other twenty to resemble each other as follows: 17, 18, 24, 29, 32, 37, 37, 38, 40, 41, 41, 41, 42, 42, 44, 45, 46, 46, 51, and 62. It would be clear that there were two distinct types of twins. To measure accurately the resemblance of an individual pair is, however, very difficult. The most serviceable measure which I am able to devise for a single trait is the Pearson coefficient, using each individual twice in the calculation. That is, if the deviation measures are: First) member:of pair aw fw wiles: 6, eecond membenot paiticsie:a ! ) Ratitige botkriics bye sa te) ulti oth sides —_ ,_ —_—-_ —->='—s ———— ply y N N N =D? Whence o,7~— 1/¢e* N = 1/o” oD op" Substituting r = Ir — —— Plies : ( Difference in scores)? orinspecialcaser == Yo — —W———— 2 times o? 39 40 CURTIS MERRIMAN The following computation will show the contrast between the re- sults according to the method used by Thorndike and the one here proposed. Suppose we have two twin pairs whose Stanford-Binet scores are for the first pair 97 and 100, and for the second pair 99 and rot. Let us suppose also that the mean score for a twin population is 98 and the standard deviation is 15. The compara- tive results will then be: (1) By Thorndike method, r first pair = —.80 and r 2nd pair = +.60 (2) By proposed method, r first pair = +.98 and r 2nd pair = +.99 It is quite evident that a very substantial difference will be made in the distribution tables when such results are assembled. In the above instances, the —.80 and the +-.60 will go into nearly oppo- site parts of the table or curve, while the +.98 and +-.99 will lie near together. When it is remembered that the original scores were _ very nearly alike, the second alternative seems the more reasonable. It was noted above that Thorndike introduced a subjective factor. Since the present study is based upon a rigid use of the method as just stated, it is believed that the subjective factor is pretty com- pletely eliminated. Tables 12 and 13 show the results of the application of this formula to the Stanford-Binet data, and for comparative purposes to one item of the Thorndike physical measurements. The first. three columns give the distribution of r’s for all twin pairs, like- sex pairs, and unlike-sex pairs respectively, the calculations being based upon I.Q. differences. The last three columns give the cor- responding data for calculations made from the original cephalic index measures as given by Thorndike on pages 37-39. | The outstanding fact shown by Table 12 is the more pronounced resemblance shown by the like-sex group. Of 67 pairs in this group 45 show a resemblance falling within the .go0 to I.00 range. In the unlike-sex group 17 out of 38 lie within the same range. It should be noted also that the cases of extreme negative correlation belong to the unlike-sex group. Table 13 shows the force of this argument more clearly. This table shows the distribution of the cases fall- ing within the .go to 1.00 range. If we accept the range of .99 and 1.00 as amounting to practical identity we find that 20 out of 45 pairs of like-sex pairs are practically identical, while but 5 out THE INTELLECTUAL RESEMBLANCE OF TWINS 41 TABLE 12 Distribution of Resemblance in Individual Pairs as Measured by Teed a ALL LIKE UNLIKE ALL LIKE | UNLIKE ey rae oe R bel re habe CEPHALIC | CEPHALIC | CEPHALIC se po ee INDEX INDEX INDEX Less than —.g1 2 2 —.go to —.8I ° —.80 to —.7I ° —.70 to —.61 ° —.60 to —.s1 ° —.50 to —.41 ° I I —.40 to —.3I I I I I —30 to —.2I1 ° —.20 to —.II I I I I —.1o0 to —.olI ° —.0o to -.09 2 I I +10 to -+.19 fe) I I +.20 to -+.29 2 a) 2 I I +.30 to +.39 fe) I I +.40 to +.49 7) 2 I I +.50 to +.59 6 I 5 8 2 I +.60 to -+.69 4 3 4 +.70 to -+.79 8 4 4 2 2 +.80 to -+.89 3) 8 4 4 4 +.90 to -+1.00 62 45 17 22 18 Total Lbror Abi 675) (ae stay] Sa eg eee 8 ALL LIKE UNLIKE ALL LIKE | UNLIKE es pias ae R take Soe i ae CEPHALIC | CEPHALIC | CEPHALIC eine ped pe INDEX INDEX INDEX go gi 92 9 6 3 2 2 93 94 I I 95 9 7 2 I I .96 4 4 97 fe) 7 @ I I 98 9 5 4 5 5 -99 10 6 4 4 3 I 1.00 15 14 I 4 3 I Ba] 42 CURTIS MERRIMAN of 17 unlike-sex pairs can be so regarded. The evidence that has just been cited from the Stanford-Binet part of the tables can be, in the main duplicated with reference to the cephalic index part of the tables. One other consideration remains. Do the present data lend them- selves to such treatment that we can specify what particular like- sex pairs make up the group of duplicate twins? In our study of the Stanford-Binet I.Q. differences it was stated that the peculiar shape of the curve was produced by the presence of approximately 2s duplicate pairs. When this study was first planned it was hoped that the internal evidence of the data might enable us to identify these pairs. The writer has, however, been unable to devise any method of studying the curves or the scores that will certainly point out the duplicate pairs. It is exceedingly interesting, however, to follow up certain evidence that is contained in the verbal reports TABLE 14 Scores AND SCORE DIFFERENCES OF THE PAIRS THAT WERE REPORTED AS SIMILAR i Differ- Ratin Beta AR i Pair Ta ence Teacher differ: "| Beta Differ- N.LT iffer- Eee 1.Q. rating aye: scores hee Scores SE 5 | 104— 96 8 2.3—2.2 De Yb) ihe sce a ta 8 | 110—106 4 7b 5 O08 5.0108) pees sat Tah Ns boesbiest Io | I107—II0 3 2.5—2.4 ol 25—18 9 ROAN Ne Bice 12 | 66— 62 4 3-7—3.6 as Ph ere ceedtl > | ON Etats 13 | 93— 93 ° 3-I—3.1 0.0 7—16 9 + ao lerinte 22 | 102—I00 2 2.6—3.0 av ya Wy oatte ie a Bes e 28 | 102—102 ° 2.9—2.7 py) 48—46 2 140—196 56 29 | 104—I109 5 3-I—3.1 0.0 45—58 13 I14—122 8 30 | 103— 96 % 3-:0—3.0 | 0.0 37—44 7 48— 46 2 36 | 123—118 5 3:0—3.0 | 0.0 61—So0 II | 268—286 | 22 46 | 135—127 8 2.3—2.3 | 0.0 46—60 14 | 246—238 8 58 | 112—106 6 2.7—3.0 a 74—72 2 | 250—232 18 76 | 67— 73 6 3-4—3-4 | 0.0 54—51 3 | 126—142 | 16 77 | 85— 85 ° 2.4—2.4 | 0.0 51—63 129 5 ees ; 84 | 75— 72 3 2.7—3.1 4 4I—20 OE 94—124 | 30 86 | 88—107 19 3-I—3.1 0.0 77—80 3. | 264—254 10 88 | 10g—122 13 2.2—2.2 | 0.0 84—78 i te 94 | 104—103 I 3:0—3.0 | 0.0 80—78 2 | 296—296 ° 95 | 98— 98 ° 2.6—2.9 By 74—69 § | 298—320 | 22 100 | 80— 85 5 3-5—3+3 2 67—81 7 aed Re fe 104 | 83— 83 fe) 3.2—3.3 i 69—73 4 | I98—208 Io 107 | IOI—I07 6 1.7—1.6 I i Adal 68 gos | THE INTELLECTUAL RESEMBLANCE OF TWINS 43 of the examiners. Each examiner was asked to report whether the members of the twin pair being studied resembled each other closely enough to frequently cause confusion of identity. Unfor- tunately not every examiner made this report, so the following study of the curve locations of the “reported similar’’ pairs is not as complete as it might be. Table 14 presents a summary of certain data that will be used in this part of the study. Table 14 has been so arranged that it shows the identification number of each of the pairs that was reported similar, and in ap- propriate columns the various scores and score differences. These data can be studied from the point of view of the degree of re- semblance shown by this group as compared with either the entire like-sex pair population or the entire twin population. It can also be studied from the standpoint of curve location for the various pairs under consideration. We shall note the matter of resemblance first. The following comparison furnishes rather striking evidence for the real similari- ty of those that are reported “similar.” Correlating by 6=D? N (N’?’— 1) we have: For staniord-BinetrlL.OO. 2 ae. R = +.986 BO Sache re ating ys sso cal eatesn ao R = +.940 HUOT ATS GLANs, gMerrente Oe Wey abies se: Aateu «5 8% R = +.887 SUMIMEN OP Li tol eens ers tate oe Pel aes R = +.987 These are very high correlations. Not only are they high correla- tions, but with one exception they are materially higher than the results that were found in earlier parts of the study for the re- semblance in the entire like-sex group or the entire twin popula- tion. These figures are shown in Table 15s. This gives an excess of resemblance in favor of the “reported similar” in every case when they are compared with the total twin pairs, and in all cases except Beta when compared with total like-sex pairs. It is to be noted also that there is a steady increase in the value of r as the fraternal pairs or “supposed to be fra- ternal” pairs are dropped from consideration. 44 CURTIS MERRIMAN TABLE 15 TEACHER STANFORD BINET RATING a a ee Ee ir a Bes ara ee All twin pairs 10§ | :782'|voo. *}..g12 | 763) 841%] 149 ia Son All like-sex pairs 67 | .867 | 53 | 654 | 45 | .g08 92 | .925 “Similar” like-sex pairs 92 | .986.] 22 »|. .940 | 18] 1887 12 | .987 For the discussion of curve location the reader must recall the facts presented in Table 9 and Figures B and C. It was there shown that the plotting of the score differences produced a highly skewed curve, the skew being towards a small score difference. If we define that portion of the Stanford-Binet curve which is pro- duced by score differences of 0 to 5 inclusive as the “Binet upper level,’ and that portion of the N.I.T. curve which is produced by score differences of 0 to 20 inclusive as the “N.I.T. upper level,” we can make the following observations: 1. Of the 22 pairs reported similar, 13 pairs are on the “Binet upper level.” 5. In the entire twin population, 9 pairs have a Binet I.Q. score difference of o and are therefore on the Binet upper level. Eight of these are like-sex pairs, and of the 8 pairs 5 are in the group of “reported similar.” No report as to similarity was received — upon the other 3 pairs. 3. Of the 22 pairs reported similar, 12 pairs took the National. Eight of these are located on the “N.I.T. upper level.” — i, GENERAL SUMMARY OF PURPOSES, DATA, AND RESULTS Purposes. This study of the intellectual resemblance of twins has sought to answer three questions: a) What is the effect of environment upon the amount of in- tellectual resemblance of twins? b) Does the fact of twin origin and birth operate in any way to lower the intellectual level of a twin population? c) What light do the psychological data throw upon the cur- rent biological belief that there are two distinct types of twins, fraternal and duplicate? Data. Individual and group material was collected as follows: pal FOP -DINeu teSES) LOL re ae eet ye a cree IO5 pairs a eacier estimates TOPs. Fr. es ces et QO pairs PATIY VCtAtOStG TON a ctu niece sie’ 5 5 76 pairs National Intelligence tests for........ 143 pairs Treatment of Data. These data were studied from many dif- ferent angles. Young pairs were compared with old pairs. Like-sex pairs were compared with unlike-sex pairs. Boys were compared with girls, etc., etc. In making these various com- parisons four methods of treatment were used: a) Pearson correlations between various groups. b) Difference in gross scores. c) Curve plotting and fitting to determine character of popu- lation. d) Empirical study of correspondence between psychological data and judgment of friends as to the resemblance of cer- tain pairs. Findings. The results of the study are presented in the form _ of answers to the three questions asked at the outset. For con- venience, all the correlation results are assembled in Table 16. The reader will find it very helpful to refer frequently to this summary. 46 a) b) c) CURTIS MERRIMAN Environment appears to make no significant difference in the amount of twin resemblance. Table 16 shows twenty pairs of correlations on the basis of young twin pairs versus old twin pairs. Of these twenty pairs there are 15 that show either very slight changes or changes that can be explained on the basis of small population. The larger changes of the teacher rating comparisons are explained on the basis of better acquaintance with the older pairs and over emphasis of slight differences. Twins suffer no intellectual handicap. This is shown in- various ways: wf 1) Mean and median I.Q. practically same as for general population. 2) Mental level of boys same as girls. 3) Like-sex pairs same mental level as unlike-sex pairs. 4) No significant differences in variability of sexes. 5) Young pairs show slightly higher mental level but this is explained by the fact that Stanford-Binet is more difficult for older children. The data show quite conclusively that there are two dis- tinct types of twins. This is shown in various ways: 1) In every case where like-sex pairs are compared with © unlike-sex pairs, the correlation of the like-sex pairs is significantly higher. Table 16 shows in groups 2 and 5 the twenty-four correlations that provide the evidence for the statement just made. 2) When sibling data are compared with twin data, the - correlations lie much nearer the unlike-sex pair twin data than to the like-sex pair data. This is in harmony with the biological claim that genetically speaking fra- ternal twins are siblings. 3) All the curves and curve fitting tests used in the study indicate clearly a difference between like and unlike-sex pair twins. 4) The empirical study of verbal reports on “similar pairs’ tends strongly to show that curve differences are to be largely accounted for by the like-sex pairs that THE INTELLECTUAL RESEMBLANCE OF TWINS 47 show great intellectual and physical similarity, and that presumably belong to the “duplicate” type. All twin pairs Pairs, 5to gyrs. Pairs, 10 to 16 yrs. Like-sex pairs Like-sex 5 to g Like-sex Io to 16 TABLE 16 ’ SUMMARY OF CoRRELATIONS BINET r -782+ .025 .809 +.032 275] #037 867 + .020 8824 .028 865 +.027 |N 39 gI 53 30 Girl-girl pairs Girl-girl 5 to Girl-girl 10 to be Boy-boy pairs Boy-boy 5 to 9 Boy-boy Io to 16 Unlike-sex pairs Unlike-sex 5 to 9 Unlike-sex 10 to 16 | 857 £.029 915 + .026 .814+.050 877 + .030 .800 + .078 890+ .034 -504+.081 774. + 064 .298 + .137 BETA | N.I.T. | TEACHER SON ODES a EE 105| .841+.022 | 76] .891+.011 |143] .512+.053 47] -784+.049 | 28] .797+.034 | 54] .686+.057 58] .664+.054 | 48] .875+.017 | 89] .373+.081 67| .go8+.017 | 45] .925+ .009 | 92] .654+.053 2g] .921+.025 | 16] .946+.012 | 31] .788+.053 38] 842 + .036 | 29] .865 + .022 61} .568 + .083 | 40] .866 +.033 | 25] .928+.012 | 61] .645 4.071 | Ig] .709+.112 | 9g] .g65+.009 | 24] .g13+.030 21] .896 +.032 | 16] .g1g+.021 | 37] .521+.123 27] .938+.015 | 20] .g25+.018 | 31] .605 +.090 O} -934+.049 | 7] -921+.041 | 7] .534+.161 17] .747 +.080 | 13] .895+.027 | 24] .715+.089 38] .732+.056 ele 025 266 + .102 18] .§19+.147 753 +.066 z 681 +.090 20] .643 + .OgI 834 .044 .O72 4.141 16 33 9 37 16 21 -_ : iy as 10, BIBLIOGRAPHY Bateson, WILLIAM. Determination of Sex. Nature, Feb. 3, 1921. CaTTELL, J. M. Statistical Study of American Men of Science. Science, New Series, Vol. XXIV, pp. 732-742. ConkKLIN, E.G. Heredity and Environment. Coss, Mary V. Evidence Bearing on the Origin of Twins from Single | Ovum. Science, April 12, 1915, pp. 501-2. DanrortH, C. H. Is Twinning Hereditary? Journal of Heredity, Vol. VII, (1916), p. 195. Davenport, C. B. The Influence of the Male in the Production of Human Twins. American Naturalist, March-April, 1920, pp. 97-122. Davenport, C. B. Inheritance of Temperament. DaAvENPORT, ET AL. Twins. Journal of Heredity, December, 1919. Doncaster, L. The Determination of Sex. A Review of Heredity, Vol. VI, June, 1915, p. 260. Dott, E. A. Psychological Measurement of Thirteen Pairs of Feeble Minded Siblings. Training School Bulletin, May, 1918, pp. 45-47. Gatton, F. Inquiries into Human Faculty. Everyman’s Library, pp. 155- 172. . Gorpon, Kate. Report on Psychological Tests of Orphan Children. Jr. of Delinquency, Jan. I, 1919, pp. 46-56. GresELL, ARNOLD. Mental and Physical Correspondence in Twins. The Scientific Monthly, April and May, 1922, pp. 305-331 and 415-428. Haypven, C. C. A Case of Twinning in Dairy Cattle. Ohio Agricultural Experiment Station Bulletin, March-April, 1922, pp. 54-57. . Jorpon, H. E. A Note on Twinning, Jr. of Genetics, Vol. IV, 1914-15, pp. 79-81. . Newman, H. H. The Biology of Twins. University of Chicago Press, 1Q17, pp. 1-185. . Ottver, JAMes. The Hereditary Tendency to Twinning. Eugenics Review, Vol. IV, (1912-13), pp. 39-53 and 154-167. Pearson, Kart. Concerning Inheritance, etc. Biometrika, Vol. III, pp. 131-190; Vol. V, pp. 105-146. StarcH, Danret. Similarities of Brothers and Sisters in Mental Traits. Psychological Review, May 1917, pp. 235-8. THorRNDIKE, E. L. Measurement of Twins. Archives of Philosophy, Psy- chology, and Scientific Methods, Number One, September, 1905. . THompson, J. A. Heredity. Relative Number of Twins and Triplets. Science, March 18, 1921. . Fisuer, R. A. The Genesis of Twins. Genetics, Sept. 1910, pp. 489-490. . Wiccam, A. E. What Twins Tell Us About Ourselves. Physical Culture, October and November 10921. 25. 31. 32. THE INTELLECTUAL RESEMBLANCE OF TWINS 49 Two Kinds of Twins. Literary Digest, Vol. LII (May 27, 1916), pp. 206-9. SmitH. Twins. Science, Vol. XXVII, p. 451. Partial Twin. Literary Digest, Vol. XLIV, p. 588. BALLANTYNE. Antenatal Pathology. GouLp AND Pyte. Anomalies and Curiosities of Medicine. Wiper, H. H. Duplicate Twins and Double Monsters. American Journal of Anatomy, Vol. III (1904). Wiper, H. H. Palm and Sole Studies. Biological Bulletins, Vol. XXX. Liu, Frank R. Problems of Fertilization (1919). University of Chi- cago Press. Pair Pair Pair Pair Pair Pair Pair Pair Pair Pair Pair Pair Pair Pair Pair Pair Pair Pair Pair Pair n ms * SE’ SUS IETS IER EINER SIE SRI I es APPENDIX A Gross ‘Scores FoR VARIOUS ITEMS GRADE 2b 2b Ib Ib 1a 1a Ia Ia Ia Ia 1b Ib Ib Ib Ib Ib 1a 1a 2a 2a 1a 2b 1a Ia 2a 2a 2b 2b 1a 1a 2a 2b ra Ia tb Ia 2b 2b CAS 5-10 5-10 6-11 6-II 6-3 eo | 6x 6- 6- 6- 6- 6- PUM oOeFFNN HH M.A. 6- 6 6- 8 7- 6 74 6- 4 5- 4 6-10 7- 6 6- 4 6-10 5- 6 6- 0 6- 2 5-8 7-1 6-10 6-10 6- 6 7- 4 7- 6 7-11 8- 0 5- 2 4-10 6-10 6-10 7- 2 6-0 7-8 7-10 7- 2 7-0 7- 8 7- 8 6- 3 6-7 Wi i 8- 6 7-10 8- 8 1.0. 112 114 108 106 IOI 85 107 117 06 104 90 08 102 04 110 106 107 TEACHER 1.5 2.2 2.7 3-3 3.0 3.0 i 2.6 3.7 2.7 ae 3.1 2.5 2.0 2.5 2.4 3-4 2.8 At 4.5 3.1 3.4 3-4 3-3 2.8 2.5 2.9 2.7 2.6 3.1 BETA 39 45 32 30 25 18 N.LT 56 88 128 108 Pair Pair Pair Pair Pair Pair Pair Pair Pair Pair Pair Pair Pair \efiip Pair Pair Pair Pair Pair Pair Pair Pair Pasar Pair 21 113 22 23 24 25 26 27 28 29 30 31 32 33 34 114 115 116 117 118 119 120 121 122 THE INTELLECTUAL RESEMBLANCE OF TWINS a) iz] ras AEDITE AAI AYIS SHIITES TST SPSS SII Ate Meda aaa yy eS GRADE Ib Ib Ca: 7- 0 7- 0 7-10 1.Q. 86 79 TEACHER’ BETA 3.8 3.4 2.3 2.0 Patil 2.2 3.4 3.3 2.8 2.3 2.9 2.8 3-5 3-3 3.0 3.0 3-5 3-3 21 2.4 21 2.4 3.8 2.9 2 12 53 48 47 48 51 N.1.T. 143 134 144 132 76 150 m ie] ra SA AAA ye Se eS Se a SS Sf ge Mo Fac a at 4 GRADE 3a CURTIS MERRIMAN CrAG 8-11 9- 6 Oo ' CHW ADRK WMDOOUUNUNOOO ooo vo © oO DOR Be ae aL 7 NN Lon] Lon! o 7 or GSES OPP FP9 FS WwWWwWARKDOO A TEACHER 2.9 3.8 2.6 2.6 3.2 3.0 4.5 ke 2.2 ZY 2.6 3.2 3.3 3.4 3.6 a 3.0 3.1 2.1 2.1 2.9 3.3 3.0 3.0 BETA 53 41 61 50 41 4I 63 N.LT. 104 214 198 268 286 114 218 240 258 174 138 108 ete, Pair Pair Pair Pair Pair Barr Pair Pair Pair Pair Pair Pair Pair Pair - Pair Pair Pair Pair Pair Pair Pair Pair Pair Pair 132 133 134 THE INTELLECTUAL RESEMBLANCE OF TWINS n cl * HIGGS ASUS AAAS TIA de Ae TS INS S Ay de de dae yee eS SS yD Hy GRADE C.A. 9-10 9-10 Q-II Q-II 9Q- 2 9Q- 2 10- 6 10- 6 10-10 10-10 I0- 10- 10- 10- 10- 10- 10- i fo) 7 NOWW HH OAOH Be ew % * M.A. ai'oh. eer ¢ tse ee ee eee se eee 4). 0h ee se eee eee ee se eee eens see ee eee ee see ee eeeee CRY oat TE see ee ee eee te eee is G86 1.Q. TEACHER BETA 222 Be 20 3:5 3.0 2.8 2.6 3.4 3.2 2.2 74 72 39 54 47 70 250 232 218 262 207 178 192 169 133 156 171 148 147 139 189 197 72 74 109 189 137 54 CURTIS MERRIMAN SEX GRADE C.A M.A I.Q. TEACHER BETA N.LT. F 4a TO-80y fas a tex a 136 Pair 62 M 4a II-I1 8- 6 71 4.1 44 116 F 6b II-II 9-10 83 3.0 60 204 Pair 63 M 6b II-10 II- 2 04 2.9 68 214 F 6b: II-10 II- 5 06 39 50 206 Pair 64 M 3b 11-8 9-10 84 3.8 ie M 3b 11- 8 7-11 68 3.8 Pair 65 M 3a II- 5 9- 9 85 M 3a II- 5 9- 0 79 Pair 66 M 7b II- I I5- 5 139 M 7b II- I I4-II 134 ts a Pair, 67. F II- 7 8- 6 73 59 140 F “7 II- 7 8- 6 ve en 49 146 Pair 68 F 7b II-II 12-5 104 2 74 204 F 7b II-II 13- 6 113 3.9 80 . 324 Pair 69 M Sa II-II 11- 8 90 3.0 76 196 F 6b II-II 13- 0 III 7 So =. ../202 Pair 70 M 6c II- I II- 2 IOI 3.0 38 158 M 6c II- I Tise7 104 3.0 46 152 Paineyt M 6a II- 5 I2- 4 108 2.4 73 258 M 6a II- 5 12-II 113 2.8 66: 5/262 Pair 145 F 6a 6 bee Pe . eke 196 ' F 6a TELS See ee 4 185 Pair 146 M 7a Tiss?“ Vegineeey 223 F by ph Caine Pad lM ma 255 Pair 147 F sb Ti-cE er ees 135 M 5b TI=g 12> Vt oe Me a. ye 139 Pair 148 F 6a TEsgAC ha Petes Phage ut ee 274 F 6a TERE ideas 204 Pair 149 F 6b TT=: 7 oe 242 M 4b TI) TA tee 195 Pair 150 F 7 TI-2ciee eae 253 F 7 Theos) eee ee 257 Pair 151 F 6b Tints seal 202 M 7b TLSSe) ur Ves ‘ 246 Pair 152 F Sa Bio. SA wih ae 131 F sb Tits Boh ae 183 Pair 153 F 5a Df ir Pale Og yr Nte 169 M 6b VUAlT io ies 269 Pair 154 F 4a TI fiche I4I F 4a Tiss Bee ctl 1a tae 124 Pair 92 F 6a 12- 7 II- 0 87 a2 65 oan F 6b 12-17) 10-10 86 RYE: 62 Pait 173 M sb 12- I II- 2 92 3.0 62 M 5b 12- I 10-11 OI 3.0 59 A: Paina M 6a I2- 0 II- 4 04 3.1 78 200 , F 6a 12- 0 12- 6 104 pa 82 286 Pair Pair Pair Pair Pair Pair Pair Pair Pair Pair Pair Pair Pair Pair Pair Pair Pair Pair Pair Pair Pair Bar Pair Pair THE INTELLECTUAL RESEMBLANCE OF TWINS SSAA eIddie gadis eee rye eee ee te ee Re eS = GRADE 7b 7b 4a Cc, I2- tin tn “tn ta (SO °O (ONO i 6-6 Gaee: 1C at. it H to ' _ ° 4 to ' 4 o) OOO STS] fe eH OS Gorm Go BR ow Gu OV OO see ee ee eee sees oad: ela) ee eee eens ee eee ee eee see ee ee eee eee ee see ee see ee sen ee 1.Q. 95 96 67 Vie 85 85 116 82 86 82 74 79 121 134 100 75 90 TEACHER BETA 3-4 4.3 3-5 3-5 2.6 68 7O 54 55 N.1LT. 126 142 170 148 94 124 254 264 264 254 213 215 209 247 I4I 137 266 233 117 200 178 125 259 263 169 203 241 238 263 241 22 56 CURTIS MERRIMAN SEX GRADE C.A. M.A. I.Q. TEACHER BETA N.LT. | M 3b £2750) i Yeaos s tse wea ¥ 51 Pair 166 M 5a 2-72" > ahs Aik oe is 223 M Sa oe a ati aa me 235 Pair 167 F 6a T2AZ" eat: ive ‘er we 221 F 6a T2-"2ixi polity c 4 as Ag 201 Pair 168 F a B2=°5 Ll Pade Wave oe < 248 F 8b £2258 Ulweac's tats rt ai 277 Pair 169 F 7b ae el ne ah 25 274 F 7 52-935) Wty a a ae 308 Pair 170 F 7a £B-18'7) ye “ahoeres ex Bae i 281 F 7a aA iN ks ee suas rf os 305 PA a7t F 5b 1 ae ae CAE oe eke een a I5I F Sb 12-4 Aes ahs, 4y os 159 Pair 172 F 6a ioe FW Gee by 4 ba | eA F 6a Fon07 =) he, Nay ee Ae 235 Pair 173 M 8 oy Me ae ae a ae 285 © F | TO AEA a heed sere: Sek nae as Rel OA! Pair 174 F 7a 12-19 TVs tes Lee “e 219 F 7a T2505 ecu en tie ae vs 225 Pair 88 F 8a 13- 4 I4- 7 109 2.5 84 F 8a 13- 4 16- 3 122 2-57. 78 Pair 89 M 8a 13- 9 13- 6 98 2.6 60 300 F 8b 13- 9 I2- 0 87 3-4 73 300 Pair 90 F 7a 13- 2 12- 6 905 3.5 51 256 iB 7a 13- 2 9- 9 74 4.2 53 200 Pair 91 F 7a 13- 4 II- 9 88 3.2 heh 224 F 7 13- 4 13- 5 IOI 25 89 264 Pair 92 M 8a 13- 8 13- 4 07 3.8 GG 200 M 8a 13- 8 13- 9 100 2.9 62 226 Pair 93 M 8b 13- 8 16- 0 117 2.2 82 206 M 7a 13- 8 12- 8 93 2.6 85 - 280 Pair 94 F 8a 13-10 I4- 5 104 80 206 F 8a 13-10 I4- 3 103 78 2096 Pair 95 M 7a 13- 6 13- 2 98 3.0 74 298 M 7a 13- 6 M3 08 2.9 69 320 Pair 06 M 8a 14- 4* 13-10 07 2 78 330 F 7b 14- 3* 13- I 92 3.0 59 258 Pait 67 M 5a Df ae Lal Ee 3.6 ey M 5a UR Dy ans ta 3.9 44 Pair 98 F 5b 13- 2 10- 7 80 3.5 66 180 F 5b 13- 2 10- 8 81 3.2 54 202 Pair 176 °F 6a Rae aU Cees es 202 F 6a T3552) eee 240 Painet77 M 8a T3oh Oia) MCRAR ees 337 F 7b TSO ow Anos tok 222 Pair 178 M 8a 1K Ce” MPM 6 pk 266 M 8b LO=040 8 ite ac 190» Pair Pair Pair Pair Pair Pair Pair Pair Pair Pair Pair sae Pair Pair Pair Pair Pair Pair Pair Pair Pair Pair Pair Pag THE INTELLECTUAL RESEMBLANCE OF TWINS n rel va MISSES SIS NS ee ee ee ee ye ee Oe eee nh ee GRADE 8a C.A. 13-II 13-11 I3- 5 13- 13- 13- H a om mH w& Cones I woe _ oa =x = & ses 14- 14- 14- 14- 14- NON ONO 00 CA tH OO ON ON OR = aere/e Pie elss6 eee ee oe ae: 8 sees eee ee eee ee Sari Ta ee eee eee ee een ee eee ee se eee tees ee eee ee ee eee ee 13-10 ee eee eee ee eee ee enews 6.8 eee ee eee ee oe eee L.Q. TEACHER 3-4 3.4 4.2 4.0 3.3 4.0 2.4 4.4 4.4 4.3 3-5 3.4 3.3 2.8 4.2 3.5 1.8 1.5 BETA of N.LT. 273 270 304 246 269 157 295 209 219 252 264 308 216 231 232 224 304 290 58 203 204 AAS NS de a Dh ee ey af af ty CURTIS MERRIMAN 16-II eeeee eeeee eeeene eeeee eeeee eee ee oe eee eoeee se eee eeeee eee ee eeeee sens 1.Q. TEACHER’ BETA 3.7 ‘28 3.5 23 2.8 3-7 3-5 3-5 70 79 N.LT. 217 188 209 104 BA. 4 Shih iN ais: ey * Hat MY f ig frie ee ‘ie Sane) BF21 .P96v oe effect “i ele basa ” maze il LA 1 1012 00008 5433