AN ANTHROPOMETRICAL STUDY OF COLLEGE FRESHMEN BY MILDRED LEE ECKI B. A., Northwestern College, 1920 THESIS Submitted in Partial Fulfillment of the Requirements for the Degree of MASTER OF ARTS IN MATHEMATICS !N THE GRADUATE SCHOOL OF THE UNIVERSITY OF ILLINOIS 1921 • rrjT^'r-r ir r.-T T-L. if’ ^ . 4 X . ^ % • ! 4 ' ^ 'I ^ 2 -*^ ^1 V. \ V ' ^ * r‘-’'^r' : ■. " .- S'^~^"«'.s^;::‘/''USar .T . '■ ■'‘.Y, -; ^ f . V,’ ./ /-CW^ i'*^S^ •• - '“‘*^^^4 * M ''^' ^ V •*=, t t ' > ‘ ' - ^:V: -a % - V f V K - , *■, -r^. ■• f4;>tr y -:- :’? • i I a •K ''OrVyv*'' ' . • ,,'•. , , „., '’f Y< >43‘f r I ? 4 V V ?. ^ ‘ ^ k ^ 4 , •^> . .V.- ^ z^;.,- -'i' *• r>'^- "if^* - ' - wcvf ^v.;4J HX :#l|?tS':'^ ■ ‘^ 3 ^ ‘,V'l‘ /» 3 ®' 4 fl «1 A*J* . ....'' 1 -fJ'. •.'• . v^TOT :. ~ \ - 'r- ', -/k.***‘ YI.'^- r* c < UNIVERSITY OF ILLINOIS THE GRADUATE SCHOOL c I I92X I HEREBY RECOMMEND THAT THE THESIS PREPARED UNDER MY SUPERVISION BY- BE ACCEPTED AS FULFILLING THIS PART OF THE REQUIREMENTS FOR THE DEGREE OF V\ J In Charge of Thesis Head of Department Recommendation concurred in* Committee on Final Examination* ^Required for doctor’s degree but not for master’s Digitized by the Internet Archive in 2015 https://archive.org/detaiis/anthropometricalOOecki TABLE OF COi^iTENTS I Introduction 1 II Description of Data 1 Table I- University of Illinois Health Service '6 III Tne Frequency Distributions 4 IV Calculation of Averages 4 Table Ila- Frequency Distribution of Heights 5 Table lib- Frequency Distribution of Heights 6 Table Ilia- Frequency Distributions of Weignts 7 Table Illb- Frequency Distributions of Weignts 8 Table IV- Summation .Method for Heignts 11 Table V- Summation Method for Weights . . 13 V Calculation of the Standard Deviation 14 VI Explanation of k 15 VII Classification of Distributions According to Types .... 17 VIII Calculation of Curve of Type IV 19 Table VI- Calculation of the Normal Curve. . . 20 IX Calculation of the Curve of Type VI 32 Table VII- Calculation of Yq fof Type IV 23 Table VIll- Calculation of the Curve for Type IV .... 24 X Calculation of the Curve of Type V 35 Table IX- Calculation of Constants for Type VI 36 Table X- Calculation of the Curve of Type VI ..... . -27 XI Meaning of Correlation 38 Table XI- Calculation of Constants for Type V 39 Table XII- Calculation of the Curve for Type V 30 XII Calculation of r 31 Table XIII- Calculation of Correlation Coefficient ... 34 XIII Summary 35 9 LIST OF FIGURES Figure 1. Frequency Distribution of Heights of 1182 Freshmen Figure 2. Frequency Distribution of Heights of 638 Freshmen Figure 3. Frequency Distribution of He ight s of 1004 Freshmen Figure 4. Frequency Distribution of Heights of 991 Freshmen Figure 5. Frequency Distribution of Heights of 877 Freshmien Figure 6. Frequency Dist r ibut ion of Heights of 971 Freshman Women Figure 7. Frequency Distr ibut ion of Weights of 1182 Freshmen Figure 8. Frequency Distribution of Weights of 638 Freshmen Figure 9. Fr equency Distr ibut ion of Weights of 1004 Freshmen Figure 10. Frequency Distr ibut ion of Weights of 991 Freshmen Figure 11. Frequency Distribution of Weights of 877 Freshmen Fig’ur e 12. Frequency Distribut ion of Weights of 971 Freshman Women Figure 13. Figijre 14. Figure 15. Figure 16. Figure 17. Correlation Between Heignt and Weight for 638 Freshmen Correlation Between Height and Weight for 1004 Freshmen Correlation Between Height and Weight for 9S1 Freshmen Correlation Between Height and Weight for S77 Freshmen Correlation Between Height and Weight for 971 Freshman W cm en k--' . • A‘m‘ ’ , WM ' ' • V j •«• '• . ■ '!•• „ '■')< 2u' i! *2&Lw- ••• ' • i\ ^ ■ , c-t, : ; ', V 1 off’tti&tij V., t '«♦»' ' ' •> '»• H.V ^ , r^^£i ■• 'iir N^e'. ^ Tt( ‘ ' ft*rr * 1. ^ •■ . t?- n p«.. :# i. H ,.'i* 'r '^t - f* ■ ' ^ \ ., %i i. • V M **i3T \t; * ,.: ’ 1 j.\ i H ^ ''r tf’l , '^;n, |r;v’f . . **t 7-1 '■ •.'4'*‘' \k .^' ; XV t%. xoi^l fe-’ ► c*k> > ’■.^^■‘ < ■, V ' .' ■ ' ■■'* .‘•^‘ ■•• -M* •1; fn' ■>J^v*,>.rii ^ V‘i '■ * ff j ' ;%< X M' ' k o4- ‘'■i'^t-sT* • (i •' ^ i '^■fe::- totSaf?^ lAj 'J . ■ ,/ ;■ > • . j i.l 1 \, '■ " ■• . i 'i* ’lir ' V I’^'^ ' C ', i . • > •, ^ •■.wry3 ef , ; ^4' ^'4 S55P ««-T, v^'t' '■ . •' ' ^ *''■• .. ,' V - ■' ■ ' 1 ’ ir*'' r-,,| ' ^ e ■ r, ■ ■ ■! >, ■ A'> ->• ■ i ■; ' ' ■ ■feW- ''\ 'Tkriui I, INTRODUCTION The object of tnis study nas been to investigate the anthro- pometric measurements of students, in particular the height and weight of college freshmen. In connection with anthropometric meas- uraaients the statement is often made by well-known statisticians that such measurements follow the normal probability curve - in fact, that statement is almost an axiom in such studies. It has been my purpose to investigate whether or not that is the case, and if not, What curve will fit those measuremients more closely. In connection with such an investigation there are certain con- stants which are always of interest; these include several averages, and a measure of the spread of the data. The three common formiS of average are the arithmetic mean, the median, and the mode, while the standard deviation is the usual mieasure of the spread of the data. The calculation of these constants was the first step in this study. The question of the dependence of weight upon height is one that often arises. This too has been investigated, and the usual m.easure of the relationship (the coefficient of correlation) calcu- Idt ed. In particular, one object has been to set down on paper a sta- tistical description of certain classes of students at the University of Illinois in the years just following the war, this description to be used later, for comiparison purposes. II. DESCRIPTION OP DATA Every year the University of Illinois requires that the studeni entering that Institution for the first time take a physical examin- ation. The informxation obtained from this examination, which include^ H oi n*>d,dH4} old^ 'tils ^p9t^o e4T,. '.^m ■ mmM' ''• / “ ' ' . ■,- “ Tj. . *nCA J fickle d 5SS? ^J&X£/OX(??Cl jxi /eitiJi^i'Of^S'OO OilIf'*.40 Y41 lifted at iiga ?«otnf.A ’ . • * ^ •. V' i ' ’ V, • *^ - » jofl it Mac ^ .’’iT ic^ #5o<|^'^ ./. '^- .^Xi^sjoX* ^kotii Jtl XlJtw ©7ij;;0 m ^r.VT.; < , ."' ' I 1 .' '’i' - . f- 'At..', . . __ ... \rTy 'lfl‘ar*^:5'!' j*^ihfs0;> ^ »ilT ■ ..’Xfrved.'tf % &us\io ■ -'i ' ' '» -i. - ^ ■ ■.% , ; ’•■ t . , : • of-.:t ea-wjLr&T. tx^t . • iV Xi . i/i'? ' i 6.^' V a ■,n ei<3 }u tti .i»i « . E :/- , .. .: ■ . ^ EJE . ;. oj»b ‘1 iTf4:'t^>ii-i3^q;;' oav so ■ » ‘ A -* " ' ■ f IfiA x ae.(S^ a;feo opi '■• ^ ir-P’* ‘^•■^,'i:.fj^^-‘^'j ' • u. M'i 1 r ‘ .-iian.*» } tA r«W'T . «r^'i^ *trv.r A-ff tfi:- i'.*Bf t !■' , ■ ■ .* ir^p^ 'V' C'^*'' '■' V, (rioXixX&iTOO ’ ■>.'. ^ “ •'■■■: J ' ‘ N V.i-' 7^'‘,ca ao''twoX .Q«l‘ <»55f4-3i fit.'-'i l-So^ col t;4i .I'ijtni’ en#* - X 02 ^ ^a-yt ^ 4 .M 1 -*i *' AT'Aa m* »4i5iT=nHQJ?.i:a ■•■r ■■;■'■ , . :;s‘:'"^’js» ' ' '4?* ,■ ,.., , ,. iV®Sipi2 r;-:- fi*5i,xypoi- ^^iocUXi?'tp^ Yd'l94a,y,A;iip>'^^ *' •' , '. .»■?:/■»*(■» /','*. .Ml '. -i iir * -fji, . I ‘1 ' ..' ^' ■ , 's.' i,. i- -*U-.n.j(Ld X^olftvrfq ^-"eijiT o.sji? iui.^ ■itf ' /•.■> ,- HT' j i&tirl^tlL HOli'vc ^ffo ljrJr>ftin;!t*:%'5-'^^ I'H/ \Q^l6tfd^‘':xol ^#na'oX»U .1 -S' . • • '4 , _ -F'il^L.- ■•'»■.. ..iLV *«C5-l/!tw •''** ' ■ 09 2 various physical measurements, is recorded on individual cards for each person and is then filed in the offices of the University Health Service Department (See Table I). The date of exaniinat ion and the classification and age of the student are also recorded on this card, so that it is possible to get the heights and weights of men and women, taken when they were freshmen. As a result of this system, I was able to get sufficient data to separate the men into classes according to age, so that I have the measurements of 1004 men who were 18 years old when they entered the university as freshmen, of 991 Who were 19 at the same time, 6b8 who were 17, and 877 who were 20. The age recorded is the one given by the student himself and means that that was his age on his last birthday, rather than his age to the nearest birthday. Due to the fact that there are fewer women in the university, I was unable to get enough individuals to divide into classes as I did the men, ana instead, i have the heights and weights of 991 women of ages 16-26 (the range of the ages of a very large proport ic: , of the 991 women is 17-20 inclusive). This does not give as satis- factory results as one could desire, for it leaves nothing for com- parison at the present time, and the more or less heterogeneity of the group makes the results of less worth. Then as a matter of interest and comparison, I made a frequen- cy distribution of the mieasurements of the men, of ages 17 to 20 inclusive, in the present freshman class (1920-1921), It is understood that all of the mieasuremient s recorded are measurements of American born students. « * •M . " r ' - ' ■'"■. ■ *■- •"■■ [w;f££^i^ ^iitJ \.o hi ai or;^' ho«rto^ aojee'l a ^ ' ' *■ t/3£. flolrirfi ffl«3ca ^ e«?£.£^’‘8at .(I oIcJaT v'»>p) ativioa , '■ , .’" ^ jiiO 'dihtf tto b^'Xyc^' t'HiLi.., oite wt’jffjs tiKi 13 ©gjs hlto, ndi1^0t^i«a.f|t fA JEffii: 40A ?o •' c la;?! f c ft 4'd o’^ a Xj^llB ribi® 6 i ^^r9^X^ a siA .n^Citaefi \flf9.|C>w -•■-■- ‘.'V'’ _., . ■ 'Jf .' r'*' ^ B9a*.BX5 c^iiri o»j?i e tsBiacfSre dt lli’a * od?' adiV >0Or- [' lo f. :• jlJniir„»7i;gi:fra “^vik 1 /A;ii c^. o'l-*‘ [ niQ TLOfjv tT6 . a./ ,?X eityr a4‘-«- sdi it\ tT^ '^‘Oti’ti -SXl^ I' ^ - , V" '^■^'■- ' i>ii$ liij#t.iid 7C5^r?c ea? \f. . *•' . V'i ,'• ': 'R I .; 'ij;^ ?TC •o^£ told t • -^ ■ ' :•: I ,/ rA- ' *.» J J. y t!' '■ ’* '\aK* '.', S»4 t'f . i’l^^virtx) ;»i^3 f.i namow ■3>c--f/’^:! aiis'^y thiiz 3d etiQ I^X ^ f^9 «ijr.uhiriidii haxjc^no" jx-^vod bXe'^W','1^ ^ ,;5 1 :■*.■*, i^i% ana •a’i4^xei: ivatf ,fca'hd«ni‘''^acr£ ,ha >5nai 6^f,vt) ^5C-&I io|^&mb|r5.» ^ f e /X'& dc a ’dub’a ttd%^ . 'aVxitfc'l^Hi C/&-7X ei nekow jsiiJ lo y ' ^' , ^' ■■ ' ^ •' ■" i». i^ti ' ' f>«OA ‘iixvi*.:t<:a aBvx<»-£ ,>iti z/z\ ,aiiBta'*i>[x#oo ftno 64i ©dfTittlh rA^vii ■ /;^ ‘..i}. ff :, "3^ 4f'^» 10 t?aoi Xi> sxprti abk !li*> .• f -^ .. .X^xhw c'dftsX Ih' sdxpfii*-! '0i'i3‘ ' ' *"■ • ' ■ - rJ ■** ' ."'S *Jl. '„ ■■'' ,#'vi ,K I'l'^ .‘'^ ■?^^j ~ ■,L TABLE I University of Illinois Health Service FAMILY HISTORY {jJate f Racial '] ! extraction J j in the vear f Racial I J Wellj has Lextraction I , in the year of. uaU of Exam.) What relatives have had Tb? Cancer?.,.. Neurasthenia? Epilepsy?.. Other possibly inherited disease? PERSONAL HISTORY Birthplace What injuries? {Give age) What operations? (G/ait age) Age of last vaccination scar; under 10 yrs. ., 10 to 20 yrs over 20 yrs... Have done ; also (Mental work other than schooling) (physical tc'ork) Are you a waiter, dish-washer or cook? Where? Present general health Appetite Sleep Tea and Cof. Tob Ale Drugs., .hrs. Disease Measles Rubella Mumps Chicken-pox Whooping Cgh- Scarlet Fever Typhoid Fever- Typhoid Vac. Diphtheria Meningitis Malaria Smallpox Smallpox Vac._ Pneumonia Asthma Pleurisy Rheumatism Amygdalitis Chorea Influenza Otitis Media Gonorrhea Syphilis Constipation Dysentery Appendicitis Neurasthenia Poliomyelitis Tuberculosis Glasses Age PHYSICAL EXAMINATION ^Gen. DevL; exc., good, fair, poor. W.; thin, av., obese. lbs ins. Build: stocky, medium, slender. Head: length ^.'9. cms., width (rT".. ems Eyes: blue, gray, dark gray, greenish, hazel, dark. Hair: fair (flaxen, reddish, light brown), brown, dark brown, black. ‘Skin: type. T . . funder 15 mm. Hcne\ ^Vac.: R. L. arm, leg; pitted, keloidal, smooth; ! 15-20 mm. Temp.. Lover 20 mm. ‘■Teeth: 876543211234567 8) Remarks., 87654321 1234567 8 j "^Thyroid F. H.. I Present Res.-^ Past I Adolescent Epi., .Exp. '■"Lymph N.:C Ax Ing "Chest: norm "Lungs: norm In.sp "Heart: rate recumb , erect , norm V. C cc. “B. P. (max.), recumb mm., erect mm. B. P. (min.) "Abdomen: norm., rigid, relax. "Hernia: "Palpable: Liv , Spl , R. Kid , L. Kid _. , Other "Knee jerk: R , L "Penis: norm., circum. ^''Testes: R L "Varic: R L ^^Urine: Col Sp. gr , R , Alb , S (Kyphosis "Vertebral column-J^ Lordosis "Feet: Long arches ) R Anterior arches ) R ^Scoliosis ) L ) L "Other joints: "Nose: Nor., Spur., Div., C.C.Rh., Turg. Rh., Hyp. Rh., Atrop. Rh. "Adenoids: L., S. ^Chr. Pharyn. "Tonsils: Nor., Ab., Bur., Proj. Path "Larynx: "Ears: Nor., Cer., T. T. Chr. S., Wch Speh Whisp "Eyes: Lids: nor Muscles; nor Fundus; nor ;Col. vis.: nor Refraction : O.D O.S Defects Treated (t) Corrected (c) 10 11 12 13 14 15 16 17 Wl9 20 21 22 23 24 25 26 27 28 29 30 31 32 Examiner. 4 III, THE EREQUEHCY DISTRIBUTIONS When the height and weighs of tnese 4481 students had been re- corded according to the classifications mentioned above, I was able to make twelve frequency distributions, that is, distributions of height and of weight for 17 year , 18 year , 19 year , 2u year ,men^ for women, and for the men of one freshman class (Tables II , III). For heights I have used a class interval of one inch, the frequencies as given in the tables representing the number of students whose height lies in the interval of .5 Inch on each side of the class as given; for example, a man is recorded as 68 inches tall, if he is in reality between the limits 67.5 and 68.5 inches. The central points such as 68 are called the mid-ordinates, since ail the frequencies are assumed to be concentrated at these mid-points of the intervals Of the base. When heights were given as the limits of intervals (e.g. ,67.5), the unit was aiviaed, with one half given to each mid- ordinate just above and below tne real value. The weights were dis- tributed in class intervals of 5 pounds, in order to make fewer classes to compute, and because with intervals of 1, tne dlstributioi would nave been m.ore or less discontinuous and by no means smooth. As it is, in several of the distributions the frequencies do not run smoothly, but show frequent jutting points in tne frequency polygons (See Figs, 7,8,9,10,11). The multiples of 5 were taken as the mid- ordinates in the case of weights and when the latter were given as the limits Of intervals (e. g. , 152%5), the unit was divided as in heights. IV. CALCULATION OF AVERAGES Although frequency distributions help to make any series of » i- 1 } ii.£. r ;fri' - . # V J f A K ’" ■'■' ' *; '^’JT '■< ‘ w - ♦ - , • • , o rP- •J,r ** '♦ •, -i t 3 ' nov V « • ^ ♦ » ' , . « ; . o 4. -.■r ...-^X,: »» •Cc » . ^ A tU W C ;.‘C£ t - - ^■■. ■ . 04 . t' ii. tH'.^r ’'.1' . r." i, '■df.l.i;: I r t . » I . - < c V , .-.w i V. : f jsrr- i : u r' /' . X';. !,v ff J'#- If ^ V ‘ • ' ~'r.'\?A 1 A.j'.: .11 '■: .' • . q iOI- "-^Tp i ‘ 1 J £ 1-V . i 04 .UA . Ai: 5 TABLE Ila Frequency Distributions of Heights of College Freshmen Freshman Class 1920-21 Women 18 years Height Actual Grad. by Grad. by Actual Grad. by Grad. by Actual Grad. by in Freq. Normal Type IV Freq. Normal Type IV Freq. Normal inches Curve Curve Curve 55 1.0 0.1 0.3 56 1.0 1.0 57 1.0 3.0 3.0 58 11.0 9.5 9.0 59 0.5 0. 2 0. 3 23.0 26.0 34.0 1.5 0.2 60 1.5 1.0 1.0 58. 0 57.0 54.0 1.5 0.8 61 4.0 4.0 4.0 94.5 101.0 101.5 3.0 3.0 63 b. 5 11.0 10.0 168.0 147.5 153.0 4.0 9. 0 63 32 . 0 28.0 26.0 168.0 174.5 180.0 21.5 23. 0 64 54,0 59.0 57.0 155.0 168.0 170.0 50.0 48.0 65 107.5 104.0 104.0 147.5 131.0 138.0 88.5 87,0 66 163.5 154.0 157.5 73.5 83.0 79.0 123.5 130.0 67 190.5 190.0 195.5 40.5 43.0 41.0 165.5 162.0 68 187.0 197.0 200.0 22,5 18.0 18.0 172.5 169,0 69 175.5 171.0 169.0 5.5 6 . 0 7.0 160.0 147.5 70 125.0 124.0 119.5 0.5 2.0 3.0 ■ 98.0 107.0 71 74.0 76.0 72.0 0.5 0.4 1.0 53.5 65. 0 72 34. 5 39,0 37.0 0.1 0,3 41.0 33.0 73 18.0 16.5 17.0 1.0 13.5 14.0 74 3.5 6 . 0 7.0 0.1 4.5 5. 0 75 3. 0 2.0 3.0 0.5 1.0 76 2.0 0.4 1,0 1.5 0.4 77 0.5 0.1 0.3 1182.0 1183.2 1181.1 971.0 971. 1 “972. 1 100470 1003.9 V 6 TABLE Ilb Frequency Distributions of Heignts of College Freshmen 17 year 19 year 20 year Height Actual Graduated Actual Graduat ed Actual Graduat ed in Freq. by Normal Freq. by Normal Freq. by Normal inches Curve Curve Curve 56 2.0 ■ 57 58 59 60 1.0 0.5 1.0 1.5 0.4 61 4.0 2.0 1.0 3.0 1.5 2.0 62 6. 0 5.5 D. 0 8.0 6.5 6. 0 65 13.0 14.0 26.0 21.0 11.5 15.0 64 33. 0 30.0 40.0 45.0 33 . 5 35.0 65 40.5 53. 0 84.5 81.0 61.0 66. 0 b6 77.5 79.5 122.0 122.0 112.0 104.5 67 109.0 100.0 158.5 155.0 145.5 137.0 68 106.5 106.0 155.0 165.0 152.5 150.0 69 89.5 94.0 136.0 147.0 129.5 136. 5 70 71.0 70.0 138.5 111.0 91.0 103.0 71 51.5 44. 0 71.5 70.0 75.5 65. 0 72 24.0 23.0 28.0 37.0 35.0 34.0 73 8. 5 10.0 13.0 17.0 16.0 15.0 74 2.0 4. 0 7.0 6 . 0 1.0 5. 0 75 1.0 1.2 3.0 2.0 0.5 2.0 76 2.5 0.4 77 0.5 0.1 638. 0 636 . 7 991.0 991.0 877.0 876.8 1 TABLE Ilia Frequency Distributions of Weights of Fresnman Class 1920-21 Women 7 College Freshmen 18 year Weight Actual Grad, by Grad. by Actual Grad. by Grad. by Actual Grad. by in Freq. Normal Type VI Freq. Normal Type V Freq. Norm^al pounds Curve Curve Curve 85 N 5.0 19.5 5.5 90 1.0 4.0 0.1 30. 5 52. 0 35,0 95 4.0 8.0 1.0 55. 5 47.0 o6. 0 4.0 6, 5 100 10.0 15,5 7.0 85.0 65. 5 92.0 7.0 12.5 105 15.0 27.5 21.0 119.0 84.0 118.0 16.0 22. 5 110 56. 5 44.0 47.0 ia2,5 99.0 127.0 52.0 57.6 115 71.5 b5. 5 80.0 135,0 108.0 121.0 58.0 o6. 5 130 115.5 89,0 112.0 98.0 108.5 105.0 85.5 78.0 125 155. 5 111.5 154. 5 85. 5 101.0 65. 0 145.5 98.5 IdO 150. 5 128. 5 145.0 66.0 87.0 6b. 0 120.5 114.0 155 128.5 156. 5 159.0 58. 5 69.0 50.0 126.0 120.0 140 155.5 135.0 135.5 42.0 50. 0 56 . 0 129.5 115.5 145 101.5 119.0 103.0 25. 5 54.0 26. 0 78.5 102.0 150 bb. 0 98.0 81.0 14.5 21.0 18.0 60.5 83.0 155 51.5 74.0 61.0 14.5 13.0 15.0 45. 5 60.5 160 54.5 51.5 44.0 5. 0 7.0 9.0 58.0 40.5 165 54.5 55 . 0 50.0 5.0 5. 0 6. 0 21.0 25.0 170 20.5 19.5 20.5 6.0 1.5 4.0 12.6 14. 0 175 11.5 10.5 13.5 2.0 1.0 5. 0 7.0 7.0 180 7.5 5.5 9.0 2.0 0.2 3.0 11.0 3.5 185 b.O 2.5 5.0 0.0 0.1 1.5 1.0 1.5 190 1.0 1.0 5.0 , 1.0 1.0 1.5 0.5 195 4. 0 0.4 2.0 1,0 0.7 6. 5 0.2 200 5. 0 0.1 1.0 2.0 0.5 205 1.0 0.7 1,0 0. 5 210 1.0 215 1.0 255 240 285 1182.0 1178.0 1181.8 971.0 950. 5 969. 5 1004.0 998.2 alt . X 8 TABLE Illb Frequency Distribucion ci nyeignts of College Fresiimen 17 year 19 year 20 year Weight Actual GracLuat ed Actual Graauat ed Actual Graauat ed in Freq. by Normal Freq. by Normal Freq. by Normal pounas Curve Curve Curve 90 2.0 3.0 95 2,0 6.0 1.0 5,0 100 2.5 11.0 6 , 5 10,0 4.0 9.0 105 11.5 18.0 9.5 18.0 8. 5 16.0 110 20,0 27.0 31.5 31.0 23. 5 26.0 115 39.5 39. 0 53.0 48 . 0 30.5 39. 5 120 64.0 51.0 96. 5 68.0 63. 0 55.5 125 84.0 62. 0 85.0 89.0 88.5 72.0 130 82.5 69.0 117.0 106.0 122,5 87.0 135 89.0 71.5 128.0 116.0 107.0 96.0 140 70.5 68. 5 123.5 116.0 111.0 98.5 145 53.0 61.0 101.0 106.0 80.5 93.0 150 48.5 49.5 64.5 89.0 55.0 81.0 155 19.0 37.0 54.5 68. 0 65. 5 66. 0 160 17.0 26.0 34.5 48.0 45.0 49.0 165 8.0 17.0 40.0 31.0 29.5 34.0 170 8.0 10.0 21.0 18.0 15.0 21.0 175 5.5 5. 5 6,0 10.0 4.5 13.0 180 4.5 3.0 3.0 5.0 6.5 7.0 185 0.0 1.0 10.0 2.0 2.0 3.0 190 2,5 0.6 2.0 1.0 5.0 2.0 195 0.5 0.2 1.0 0.3 1.0 1.0 200 1.0 0.1 2.5 0.1 2.0 0.3 205 0.5 1.5 0.1 210 2.5 235 1.0 240 1.0 245 1,0 255 1.0 2^5 1.0 638.0 636.9 991.0 980.4 877.0 874.9 i 9 ODservaD ions conprenensible, tnere is a need for quantitative ex- pressions to cnaracterize the histributions. Two ways in which ordi- nary distributions may differ are: (1) in position, that is, in tne values ot the variable around wnicn they center, and (2) in the rang? of variation. Expressions which measure position are usually called averages, of which three common ones are the arithmetic mean, the m.edian, and the mode. The more closely a distribution can be fit by the normal curve the more nearly will these three measures approach coincidence. Measures of the range of variation are termed mieasures of dispersion, of wnlch the mxst useful is the standard deviation. The values of these expressions, for the distributions in the present study, have been calculated. Tne mode of a frequency distribution is the class which has the greatest frequency. since the division into classes is arbitrary, the value of the mode thus obtained can only be approximate. A m.ore accurate definition, as given oy Gr.U . lule^ is: ’’The mode is the value Of tne variable corresponding to the maximum of the ideal fre- quency curve wnicn gives tne closest possible fit to the actual dis- tribution. ” Tnis theoretical mode is obtained in several cases under tne discussions of the types of frequency curves. The emiplri- cal or approximate value of this average can be easily obtained from tne frequency polygons. The median is tnat value of tne variable of an ordinary fre- quency distr lout ion, on eacn side of wnich tnere are an equal number of observations. The m^edian for each of the given distributions was obtained by simple interpolation, which gives the value only approx- imately, since sucn a mietnod assumes that the values in each class 1. ”An Introduction to the Tneory of Statistics”, p 120. € atoi Zit-e^r & el vnma ^aXJjtstmtiexii^Ejoo ^rio^^avidedo ^t3*3 0 ziof^ a rti ,^'i CK^ ^si) eailaJOJiclAiiO ^£I>X6C^iq V '‘*A, ai (X) r-rria yA;i Baotzuc^’zi^tt x%mc 1 . . ' * 1 ni (Xj- tnj5 tT»?nao D^jl^iji eXc^^£^e? Jdy jp’ »etrX4;V i& f Xao eUi oolllBoq, axzraAa.'t) ooitiir 'li’joiBBsiqxJ .ftoii'ii^^v io *' ' «n.jr od-j OTB itc v u>o 4ol£ti^ io_ r . ■ • ^ t 1 - : cq ireo aoi7u6Xz^ath’ M xL^tolo axX4 •' * ' '^"v‘ ' It f.c^aoiiaoQ r iom lo ' C .• ■ * /*' jt' * ' ^ 'l ' . ' r\ * Oil# 4i'^r-* :at ^q. .:Y .*C^,0,Vt: ^,5^x3 • t coxal |-eii IO fci/A£ x5asr*?is^ c.t qan.itaqKa .too -ol^itdlY ifftrsoM ®47 o?- jx'i <5iX'iX.4ifi''q 9cX • o^'WX \. .eOfiO X.^'tfvMs rt i^fextjceldo at oxoir, BidT"'’'M0J^4Wdiz:X ^-i*si ;ats t^oet#ppz: to «eq;fa to aaolaaiuoi^^ 'fiLi il b9trtJt#oo e: ^^JBO fgxrxsve ai-dx ro auX^v ©tiaAioxQqc ^,>'i y - . 1 ” '"'.05 * *' ^ -ex*i -tax»;i£x^ rra 16 eXcaxtxv «?CJ ox/Xav ledi ox orf'15 . feirpjy Xoci^., vofla Jpe;^* . fl'xao.fim lAup>\ uxi 9SX 919^^- Xibltlm to dOi>3 qo tfloX2^Xi5BxC yoflCi-v acoX jx'dx tjsxo n^vig ftCj’' xo a^-od t.it cie:i?9;A ddT ^.aco'i^^v'iq^^ijq j*oiqqxj ^ji'Xiiv i^£.j ^*12- doxAs ■ tCOX 7i^XcVjT^4[ai ilc-xd f»^i litv a©5ix;o.8a itoldaeiM i, ACX^8;’'*^'yijx,# !SX q .C, imt ..T'ffs 10 are uniformly distrilDuted throughout tne interval. Using the figures of the freshman group to illustrate, tne total number of observations is 1182, he-if of which is byi. Fromi tne table of distribution for this group (Taole ila), we see tnat there are 5b9 students wnose heignt is not greater than 67.5 incnes, and 187 more whose height comes in tne interval 67.5 to 68.5. however, only 62 are required to ria,ke up the total of 591, so tnat the value of the median is 67.5 + T^.l = 67.5 + .171 187 = 67.671 inches. The fraction added must be multiplied by the class interval, which ii this case is unity. The arithmetic mean is the particular form of average which is commonly termed the mean or the average value. It is defined by the formula in which X represents the various values of a variable, f is the fre- quency of each variable, and N is the total number of observations, ■To calculate the arithmietic means of the various distributions, a modification of Mr. Hardy’s summation method, as given by W. P. Elder- 1 2 ton , was used. By m.eans of this mjethod the monients are calculated and it is these which lead to the criterion k, which shows the type of frequency curve (as they have been classified by Karl Pearson) to which the data belong. The sumimation method is shown by Table IV in which I calculated the momients for the distribution of heignts of the freshman group. Using a central term as the starting point for the summiation, we get S 3 whicn is the difference between the sums on 1. Frequency Curves and Correlation, 1917 Edition, pp 22-33. 2. For a discussion of the method of moments see Elderton, "Frequen- cy Curves and Correlation", Chap. III. TABLE IV Summation Method for The Calculation of Moments and Other Constants Example : Distribution of Heights of 1183 Freshmen 11 ■equency 1st Sum 3nd Sum 3rd Sum 4th Sum 5th Sum .5 ,5 .5 .5 .5 .5 1.5 3.0 3.5 3.0 3. 5 4.0 4.0 6.0 8.5 11.5 15.0 19.0 5.5 11,5 30.0 3l. 5 46,5 65. 5 33.0 54.0 107.5 163.5 43.5 97.5 305.0 368. 5 63.5 161.0 366. 0 734.5 95. 0 356.0 633.0 141.5 397.5 307.0 190.5 187.0 633.0 1531.0 3433 . 5 6938.5 13385.0 175.5 436. 0 938.0 1861,5 3506. 0 6356. 5 135.0 360,5 503.0 933,5 1644.5 3850.5 74.0 13b. 5 341.5 431,5 731.0 1306.0 34 . 5 61.5 106.0 180.0 399 , 5 485.0 18,0 37.0 44.5 74.0 119.5 185.5 3.5 9.0 17.5 39,5 45.5 66.0 3.0 5.5 8.5 13.0 16.0 30.5 3.0 3.5 3.0 3.5 4.0 4.5 0.5 0.5 0.5 0. 5 0.5 0.5 S 3 1561 734.5 1183 1183 o _ ^433.5 . 633 1183 1183 .69933858 3.43174380 6938 5 397.5 S 4 = ^ItIs - -TleZ = 5. 53538071 = -^1455160 ^3 = SS 3 - d(l+d) = 5.65531343 ^3 = SS^ - 3^(l+d) - d(l+d)(3+d) = 1.11594631 ^4 = S 4 S 5 - 3/^[3(l+d)+lj -/< 2 [ 6 (l+d) (3+d)-l] -d(l+d) (3+d) (3+d) ^1 = .00688530 /^g = 3.19830317 103.38991868 m: /^i ) A. ^ -m (^:3“^6Tf4^-i'0iT = • °1277189 _ 8 ( 03 -%-!) = i4. 97401766 m = i(r+2) = 18.48700883 P./^ —A (O' ' 203 -^ 01-6 V _ r(r-3)./6i VTSU-lj-^tr-S)' = (-)4. 13388736, negative because 5 is posi tiV€ a = y-^16(r-l)-/3^(r-3)2^ = 13,76545077 1 1 « t i i ! ''I I I 'i' I t V I 12 each side of the central point; in this work it is an advantage to consider tne total frequency as unity, hence the sums of the columns must be divided by M, the total number of observations, in order to keep the work consistent. Sg is d, the difference between the act- ual mean of the distribution and the arbitrary starting point. Since the starting point in the example was 190.5 , which corresponds to class 67, and since d is .699, the actual mean of the data is 67.699 inches. Using the same group (freshman class) as an illustration of the calculation of the mean for weight, we find 150,5 taken as the central point, which corresponds to class 130; summing just as in the case of heights, d was found equal to 1.187, However, instead of adding this directly to tne arbitrary point 130, it was necessary first to multiply it by 5 in order to get it in terms of the class interval wnich is 5. Thus the mean weight of the men of one fresh- man class is 135.9 pounds (Table V), The 3.rithmetic means of the various distributions for the height of men are very close together, varying between 67.7 and 68 inches. The 20 year group averaged the tallest at 67.98 inches, with the 19 year next at 67.36, then the 17 year at 67.83, and the 18 year at 67,74; the group of freshmen men averaged 67.70. The range on each side of the mean is 8 or 9, with an exception in the 19 year group in which two men were only 56 inches. The women averaged 63.31 inches. There was a very great range in several of tne weight distri- butions, for example, one of the 17 year men weighed 285 pounds, although tne distribution was by no means continuous that far. The range of continuity was usually 90 to 210 pounds. The 20 year group averaged not only tallest but heaviest with an average weight of 139 TABLE V 13 Summation Method for The Calculation of Moments and Other Constants Example : Distribution of Weights of 1183 Freshmen Frequency 1st Sum 2nd Sum 3rd Sum 4th Sum 5th Sum 1.0 1.0 1.0 1.0 1.0 1.0 4.0 5.0 6. 0 7.0 8.0 9.0 10.0 15.0 ■ 31,0 28.0 36. 0 45.0 15.0 30.0 51. 0 79. 0 115.0 160.0 36, 5 66 . 5 117.5 196.5 311.5 471. 5 71.5 138.0 256. 5 452.0 763.5 115.5 253. 5 509,0 961. 0 153.5 407.0 916.0 150.5 128.5 624. 5 2319.5 7622.0 33017.5 65016. 5 13o. 5 496.0 1695 . 0 5302.5 15395.5 41999. 0 101.5 362, 5 1199.0 3607.5 10093,0 26603. 5 66. 0 261.0 836.5 2408.5 6485.5 16510.5 51.5 195.0 575.5 1572.0 4077.0 10025.0 54.5 143. 5 380. 5 996.5 2505.0 5948.0 34. 5 89.0 237.0 616,0 1508.5 3443. 0 20.5 54.5 148.0 379.0 892. 5 1934.5 11.5 34,0 93.5 231.0 513.5 1042.0 7.5 22.5 59. 5 137.5 282.5 528.5 6.0 15.0 37.0 78.0 145. 0 246.0 1.0 9.0 22.0 41.0 67.0 101.0 4.0 8.0 13.0 19. 0 26.0 34.0 3,0 4,0 5.0 6.0 7.0 8. 0 1.0 1.0 1.0 1.0 1.0 1.0 Q- _ 3319 . 5 _ 916 _ 1 107 'zq/jot:; q _ 33017,5 _ ^ 63. 5 _ tq 007 / t tic ^2 - 1182 “ 1.18739435 S^ 1183“ HB3 “ 18.83741116 o _ 7633 . 961 ^3 TT^ TT^ = = 11.93554339 7.36143133 65016.5 . — ri8~ 471.5 116^ =55. 40439933 = 36,ft3835548 ^ = 523.99711354 ^ 2 j AA^ '4. 1 = -f = . 411815 -^3 A^A — =- = 3. 684< -: , - i X f ii:.i . : . . ■ f! :- c :r j V nj; 1 -.: f . •• ^ ^■ ■-i ,*vii ^0 « £ '■ v.- t -fcc ^ • i V- --! *i : , * xo/ 5 ? • '*■ ‘ -. ^ • ‘ ■ '-. I.i .:" . i *i .. ''■? ' -j- nox^. ;,v.n .: •- i.'-A;: rv • • *. ♦ ; , 4 ^ Vj V* , ’ "iLX^y _v*; • '.’ i *■ I,’ *' '?<»• v*. •v.-'i i . • , . f _ •"'” fc-',' V A . . , .- ^ ■' ;<• ■~' •Tf" -* - t . '••*■ 1 ■ ^ * * - "I'j. V ... t / ‘7 *; \-’ ■«■ ' • i - ‘' 5 ' .. »t ■'■;•• / : . ^ uu •’ ■ X, .' ftriti'. ♦ ''V* ', . y__T . J'v ^ / ■' - Td-'lC If-./t ' • tl ;.:r. • ■ <’■ * iv^ a X ^ -V ^ oc>a^ xX|tcic , • '1 *. A » •r \x •.•.'■ »4 is?»£iX i '■'n -‘ V£.'. . v'v”, 'Y *'9 •J . "i ’■ ♦- r ‘ .-' li>r. ,-T. ' i ; • : X6 1 ? y, 1 o;^ 0,5 T ij n-r ptt j ’ . ' : 70 7 ;.r ' •' >? -t J ^ - 7 1 U * I, ri 1. .. -JHfl ( t ft ^"\n*.. ...- J ; , • -• f * > . • . 1 ,-^ ^ : f < • I '.-'i • -' ■ 7 y ,jK lpir ■ > ' , ' . ■■ .vuu.i '■ v; 16 approximately by running a continuous curve tnrougn tne point bi- nomial + 3) . By allowing n to increase indefinitely and letting the interval between ordinates approach zero the limiting curve is the theoretical probability curve. We usually get this binomial ex- pansion by tossing coins, for we have the probability of throwing heads equal to the probability of throwing tails'^, that is, the proba- bility of the success and failure of an event is one half. The gen- eralized form of the probability curve may be obtained from the gen- eral binomial (p+q)^, where p+q=l, but p^q. The example usually used for this form is the throwing of dice, where the probability of throwing an ace, for instance, is one sixth, and the proba-bility of not throwing it is five sixths. The normal curve may be considered as a special case of this generalized form. The still more general form worked out by Pearson includes both the normal probability and the generalized probability curves as special cases. To illustrate how a frequency curve with limiited range, ana skewness, may arise, he uses the example of drawing balls from a bag. "Take n balls in a bag, of which pn are black, and qn are white, ana let r balls be drawn and the number of black be recorded. If r> pn, the range of black balls will lie between 0 and pn; the resulting frequency poly- gon will be skew and limited in range."* This polygon is given by a hypergeometric series from which he obtains the differential equa- tion 1 ^ = a+x , y dx b4cx+dx^ In oraer to get this in the form. y=f(x), it is necessary to integrate 3/^X , and the form the integral takes depends upon the values b+cx+dx® of the coefficients of x in the denominator. Thus the criterion for the form in a particular case is the same as that for the nature of ' * Vt:. -7 I'j o a »: 7 ..«--l;ftf- rv: ;r. f^CLMUaOr j; t-,;. : -ri j.r n .. i v.?.‘Xi. ■ - •••--■ ■ T . r 'i V f ' . • *« . *‘ ■ • - . . _ ' .f. i£ia. T- • - i.<:U .. . -x-i ^ • * 0- ^ Am "<■ •• N *• tn:' i. \ - V - '-v. V ^ - -> ^;ji? p 5^-52, , “ vr- .rn /w - . 1 - — L M.*' ox *•. fc ’^Tt; ^ .. ■ • .. -I - ^-- ^ -X A : . -V-;: _ ,:, . V . * w ( • - •«* • » 4 - « I J . •‘ jC - I'A, ' ' ^ xOS-lL’ *. r ;. 7 7 - •“ '• '•■ , • •■ * 1*^ ^ - i . .’ ' • “ - . i-Z •1. • i.1' .. • -: . . ; A:z: ■ ~ ' ' n n.-i* ^ :o i ;: d. : '-/vv £[ ^ ■.. ' .:) Il: J . -> - t ? f ^ 10. . f - - m *ir •>..^•L »;i . - . . .• , , ^ . '4. *. V ^*V>-T StO' l-.J, * '*■■■>' --.ii T -1. iT /Z;r!-r - • ‘ * • r,^ • > ^ tS' J'^ ‘* ' - - * - ’'i.. 7 . 5j||^, , i. z ' ' . '■ ^'' ' ' - -ii -^lO . {■ rrii . Ai , ,; •: xilTi iV - X '■C'-" L: V ^ -*• ■. ^.vr,. H?* i: •■ . >J. V t '/ / '!' ' 4lo :: ^ .. ^ - ‘ ? 'wi n. t , t .ff u 4: 4>4i(4"' (|'T :.' X i': r ,. j„ .v ^ ^ J 'Xii 3^ /*i£, X re *. * ** rw , Lj.'v ::3 -' j. 'Mb'. 17 the roots, nar/.ely, the discriminant: yc^-4hd, which Pearson puts in the form / t, allowing to equal k. Then, according as k=0, ^4bd “ ^ k^l, or k1 Type VI k = 00 Type III. VII. CLASSIFICATION OF DISTRIBUTIONS ACCORDING TO TYPES Using k, then, according to the above table, I found that the distributions of heights always approached very nearly tne normal curve, in fact to all appearances as near that as to any curve, al- though according to k ana 0 ^ the distributions of the lb) year men, the v;omen, ana the freshman class groups were fit more closely by type IV. The 17 year group with its neirative k is classed v/ith Type I. To compare the fit of the two curves (the curve of Type IV and the normal curve) I calculated the equations of the two, for the heights of women and of the freshman class, ana plotted them- 1. Frequency Curves and ’ Cor r el at i on, p 3S. t . ‘ •- -,-^V i*>nt: ' iX'l XcUi -^y^od I -5 ' »*:a .3 - », . j^'uV&4|6 ,4^^' it * ■ ' •• j ■ y" ■ '•' 4 ■’^ ‘ ■ '■'JS- t ■ ■■1®' IT ^1 1:-“’'* ®' ’ ■^'•^;"^l!0 .'ft v£df.r‘ ».*^v«i^'- ' • ,,, >9 I ” • - f • _ L.’ * • ' -w — ^ -M ^ . i' ir St. "^ix.' .y • -V V I) t>.a>: ■h • *$ '\ - * .y' ’.<' *. -'vV ■; • • ^ kh: fei.' ’ ■■ 1 ■■"■!■'■ ■ ^■■. ■ , 16,^' ^ {} ■ <■: i I ,^ Ip'^-,l itTw3 (Jj^i ;■. 5 m.!,,^. 5»4« - x^i. .4^4, ■ ,i r* . *i ^ Ei W IGpijL^ ..^^, '»• •. %?■ . y ■ .-'T ' • y* IT '■ ' -1 \ ■•. -.rf- ■#«flS' »■'.*•,'»«> -- VW'f v.A 18 following the method of Elderton, pp 6y-73. The calculation of the constants for the equation of Type IV, is shown in Tables TV and VII, as worked out for the freshman group. From the graphs of the two curves for both the freshman and women, it seems that the normal curve is as close, if not a better fit than that of Type IV (Figs.l and 6). However, it may be that the e 3 '’e can not judge accurately from, figures the closeness of fit, and it miay be tnat k has a relatively large probable error such that the normal curve is really the better fit. The types represented by the distribution of weights are three* Types IV, V and VI. The 17 year, 18 year, and 20 year men are class-' ified under Type IV, wnile the 19 year and freshman groups are rep- resented by Type VI, and the w'eights of women are fit best by a curve of Type V. From* the figures (7--12) showing the distribution of weights, it is easily seen that they are not fit by the normal curve. In two cases I have drawn the curves which fromi k snould better represent the data, and from figures 7 and 12 we see that they do. The constants of the equation of Type V: were calculated and the ordinates obtained as shown in Tables XI an and the calculation of these constants and the curve is shcwmi in For each of the six groups, the frequency polygon of both 1. See Elderton , pp 78-80. 2. See Elaerton, pp 8o-85. y = 7o X XII?’ The equation of Type Vi is y = yQ(x-a )^2 Tables IX and X .2 19 height and. weight was drawn, and on the polygon was graphed the nor- mal curve for the respective data. To get the ordinates for the normal curve I used Sheppard's tables, wnich require that the devi- ations be expressed in terms of the standard deviation of the given data, tnat is, as The z which is given in the tables is equal to 1 _ 1 x2 e ^ ^ for corresponding ^ 's. Since tne equation for the N -1 normal curve is y = — ;■ ■ ■ ' ■ , it is necessary to multiply the z crysTT ® given in the tables by ^ in order to have the proper ordinates for the given distribution. An example is given by Table VI, Since for the normal curve the origin is at the mode, and the mode and the mear are the same, we have yo(= — — ) a-s the maximum and central ordinate, corresponding to the abscissa represented by the mean. VIII. CALCULATION OF CURVE OF TYPE IV. It was shown above that the distribution of heights of the freshman group was among those which were better fit by type IV. The equation of this curve is y yod -V tan e and in order to graph it, it is necessary to evaluate the constants a, m, V, and y^. These were calculated by means of the formulae on p 69 of Elderton, Each of these constants can be expressed in terms of another constant r, the value of which in terms of /3]_ and /J 3 is derived in Elderton (pp 74-76). The calculation is simple until the integral G(r,v) is met in the formula V — N e o aG(r , v) * (a) In deriving the value of y^ (See Elderton, p 74), the integral 20 TABLE VI Calculation of the Normal Curve Example: Distribution of Heights of 1004 Freshmen Men, Age 18. y = x^ 2(T3 ^0 ® V J.. - = 170.4923 ^ -v/2tt jla 2 Origin = Mean = 67.73556 ^ = 2.3493 ^ = .42566 N O' = 427.3613 Normal ^ Curve He ignt Frequency ^ z N 2 • cr y 59 1.5 3.71838 . 0003968 0.170 0.2 60 1.5 3.29272 .0017646 0.754 0.8 61 3.0 2.86706 . 0065461 3.798 5.0 62 4.0 2.44140 . 0202597 8.658 9.0 63 21.5 2.01574 .0523132 32.357 22.0 64 50.0 1.59008 .1136899 48.16 48. 0 65 88.5 1.16442 . 2035293 86.55 87.0 66 123.5 0.73876 . 3036668 129.78 130.0 67 165.5 0.31310 . 3798557 163.34 162.0 68 172.5 0.11356 . 3964193 169.41 169.0 69 160.0 0. 53823 .3451472 147.50 147.5 70 98.0 0.96388 . 2507066 107.14 107.0 71 53.5 1.38954 .1519282 64.93 65. 0 72 o • 1 — 1 1.81520 .0768127 32. 83 33. 0 73 13.5 2. 24086 .0323983 13.80 14.0 74 4.5 3.66652 .0114012 4.87 5.0 75 0.5 3.09218 .0035471 1.43 1.0 76 1.5 5.51784 .0008198 0.35 0.4 1003.837 1003.9 22 4> desired; the difference between Au^ and its succeeding value as given in tne table. The value of log H(r,v) known, F(r,v) was ob- tained from equation (c) ana substituted in the equation a ■F(r,v) obtained from equations (a) and (b) . After the constants had been evaluated. Table VIII was made and the ordinates calculated. The origin for Type IV is equal to the mean - and for the freshman group is 66.07. The mode is 1 /H'i> r-2 the mean - — which in the example cited is 67.6, and the cor- responding ordinate 202.77. Vk. CALCULATION OF THE CURVE OF TYPE VI The distribution of the xveights of the freshman group was found to approach most nearly the curve represented by the equation of Type VI: ^2 y = y^Cx-a) x ^ . The three constants, a, qg, q^ can be expressed, too, in terms of /3;j_, and r, which is derived in the discussion of Type I, pages 59-60, of Elderton. However, in getting the value for y^, it is necessary to evaluate the so-called T functions. in deriving the formula for y^, X T "p—i q— 1 e^ (1-x) dx appears, and is another one which can not be expressed in terms of elementary functions. This integral is called the fi function and is represented by the symbols /3(p,q) . There is another so-called non-integrable expression: p-1 -n I x^ dx which is terrr.ed the / function of (o) and it has beer Vo ‘ found that the l3 function can be expressed in terms of the / function This relation is expressed by the following equation:^ /3 (p Q ) »Ti£lZM ’ T (p+q) 1. Elderton, Appendix Ii, p 152-156. 23 TABLE VII Calculation of for Type IV Exaimple: Distribution of Heights of 1183 Freshmen. ^ VTT aG(r^v) f,f- _ F(r,vl e ^ G(r, v) r+1 N = 1182 r = 34.97402 a = 13.76545 V = -4.13389 F(r,v) = e_l££|^)_H(r.v) tan cj> = “ = -.118170 (f> = 6°. 74738 = -.1176256 (circular measure) For log H(r, v) : r = 34 r = 35 cj? = 6° .3894611 .3897389 4> = 7° .3894778 .3897551 cf> = 67. 74738 r = 34 log H(r,v) = .3894611 + (. 74738) [l67j - -|( . 74738) (. 25262) [25] = .3894733 4>= 6®. 74738, r = 35 log H(r,v) = .3897389 + (. 74738) fl62] - •!(. 74738) (. 25262) [2^ = .3897508 (p = 6°. 74738, r = 36 log H(r.v) = .3900011 + (. 74738) [l58] - -g(. 74738) (. 25262) [ 24 ] = .3900127 Hence 6®. 74738, r = 34.97402 log Jl(r,v) = .3894733 + (. 77402) [2775;i - . 97402) (. 02598) [rl56] = .3897438 log F(r,v) = V cj> log e + (r+1) log cos^ --3 log(r-l)+ lQ©H(r,v) = -.27303500 _ N 1 y© = ^ ^ F(rQ^v) = 161.013 log y^ = 2.20686204 TABLE VIII Calculation of tne Curve for Type IV Example : Distribution of ueignts of 1182 Eresnmen. . -V tan''^ y = yo(l+ — ) e Origin = Mean + = b7, 69925858 ~ 1.62666655 = 66.07257223 r Mode = Mean - ^ = 67.61124887 24 c * t JL a i / X a Co / (4) cn, C-LTe., mess. Co/.toX Co/.{5)x Y -.5158 1.2640 .;017 27° 11 •37”^ -.4746 -.8519 -1.8809 1.4741 0.3 -.4412 1.1946 .0772 25°48 •17« -.4155 -. 745? -1.4277 . 0365 1.0 -. 5686 1.1558 .0553 20^15 •44” -. 65o1 — . *4.. 022o . OOO'd 4. 0 — . 2959 1.0875 .0564 16° 28 •51" -.2876 -.5163 -.6?67 1. 0169 10.0 -.2252 1.0498 .0211 12° 34 •58 ” -.2196 -.3942 -.6904 1.4226 26.0 -. 1506 1.0227 .0097 8°35 •44 ” -.1494 -.2686 -.1800 1.7587 o7,0 -.0779 1,0061 .0026 4° 27 • 20 ” -.0778 -.1396 -.0486 2. 0187 104.0 -. 0055 1.0000 ,0000 0°18 1 711 -. 0055 -.0095 -.0002 2, 1971 157.5 . 0674 1.0045 .0020 6 ° 51 •15" .0673 .1207 -.0364 2.2912 195.5 .1400 1.0196 .0084 7° 58 •15” .1391 .2497 -.1559 2. 5006 200.0 .2127 1.0452 .0192 12 ° 0 • 22 ” .2095 . 5761 — . 5552 2.2278 169.0 .2853 1.0814 .0340 15° 55 •26” .2779 .4988 -.6283 2.0774 119.5 .3580 1.1281 .0524 19°41 •43” .5437 .6170 -.9680 1.8558 72.0 .4306 1.1854 .0739 23° 17 •48” .4066 .7298 -15655 1.5711 57.0 . 5053 1. 255P . 0980 26°42 •50"’ , 4662 .8369-1.8125 1. 2312 17.0 .5759 1.3517 .1244 29° 56 •14” . 5225 .9378 -2.2996 .8451 7.0 .6485 1.4206 .1525 52°57 •54” ,5755 1.0327-2,8187 .4208 3.0 .7212 1.5201 .1819 o5°47 •54” .6248 1.1214 45.5623 1.9660 1.0 .7958 1.6502 .2122 68 ° 26 •37" .6710 1 . 2043 -3 . 9256 1.8876 0.3 ie. .1118 1181.1 1.0125 .0054 6° 22 •41” ,1113 .1198 -.0997 2.5070 202. 77 An eight-place logarithmic table was used in the actual calculation of this table. t The values in circular measures were obtained to seven places. These values were worked out to five places. 35 The reason for expressing in terms of T rather than /3functicns is that tables of r functions have been made. Due to the charac- teristic property of T functions, namely, T(p+1) = pT(p), it is necessary to tabulate only those values of the function from 1 to 2 since all others can be reduced to a value in that interval, How- 1 ever, for very large values of p, there is a close approximiat ion of the form: logT(x+l) = logy'^ + (x + ^)log X - (x - 1 ^) ® This approximation was used in evaluating y^ (Table IX), After the value of log y^ v/as obtained, a table for the calcu- lation of tne curve was made, and the ordinates calculated (Table X) . The origin of this type is equal to the mean - — ; instead of using the mean expressed in pounds, it is necessary to use it in termjs of unit intervals which in the example is 27.187; this puts the origin at -71.6. The miOde is equal to the miean - which in tnis case equals 26.12. This, in terms of pounds, is 130.6, and the corresponding ordinate is 143.5, X. CALCULATION OF THE CURVE OF TYPE V The curve of Type V, expressed by the equation -p 4 y = yjj X e ^ proved to fit very closely the distribution of the weights of women (See Fig. 12). The formulae for the three constants y^, p and y are given in Elderxon ( p 78) and the proof for these formulae on page 82. TheT’ function is again met in tne calculation of Yq (for ex- planation of the T' function see tne discussion of the Curve of Type VI). In the example given (Table XI) this function was easily 1. Proof of this approximation is given in Elaerton, p 15b. 26 TABLE IX Calculation of Constants for Type VI Example : Distribution of Weignts of 11S2 Freshmen Q2 ' ~^1 y = yo(x-a) x g^i3p-A-l) = 6 r Ji - 202 Q 2 and -q]_ are given by log r(117. 592096) = .59908995 + (117. 092096) (2. 06666911) - (116.59158126) (.43429448) = 191.75471442 log T(q 3 _-q 2 -l) = log T(103. 176826) = 162.33854039 log T(q2+1^' = log T(1441527) = 10.26646036 13.415270 -117.592096 Using the approximation: log yo = 222.11534746 27 TABLE X Calculation of the Curve of Type VI Example : Distribution of Weights of 1182 Freshmen y = yo(x-a) ^ X ^ Origin = Mean - = 27.187394 - 98.786951 = 71.599557 T *T»X Q Mode = Mean - ~ ZT = 26.121887 2/^3 X X lo^X* 1.9523 f ^ .4809 -229.5758 6.4519 4: y y y 18 89.5996 2.9914 .098 0.1 19 90. 5996 1.9571 .6049 -230.1425 8.1152 .0880 1.225 1.0 20 91.5996 1.9619 .7013 -230.7031 9.4076 .8198 6. 605 7.0 21 92.5996 1.9666 .7801 -231.2577 10.4648 1.3224 21.010 31.0 22 93. 5996 1.9713 .8467 -231.8062 11.3592 1.6683 46.594 47.0 23 94. 5996 1.9759 .9045 -232.3489 12. 1344 1,9008 79.591 80.0 24 96.5996 1.9805 .9555 -332.8860 12.8185 2.0479 111.662 112.0 25 96.5996 1.9850 1.0011 -233.4174 13.4307 2. 1286 134.477 134.5 26 97.5996 1.9894 1.0424 -233.9434 13.9846 2.1565 143.404 143. 0 27 98.5996 1.9939 1.0801 -234.4639 14.4903 2.1417 138.596 139. 0 28 99 . 5996 1.9983 1.1148 -234.9792 14.9557 2.0918 123.544 123.5 29 100.5996 2.0026 1.1469 -235.4895 15.3866 2.0125 102.924 103.0 30 101.5996 2. 0069 1.1769 -235.9946 15.7879 1.9085 81.015 81.0 31 102.5996 2.0111 1.2048 -236.4948 16 . 1632 1.7838 60.788 61.0 32 103.5996 2.0154 1.2311 -236.9902 16.5159 1.6410 43.757 44.0 33 104.5996 2.0195 1.2559 -237.4808 16.8484 1.4829 30.407 30.0 34 105.5996 2.0237 1.2794 -237.9667 17. 1629 1.3116 20.494 20.5 35 106.5996 2.0278 1.3016 -2o8.4480 17.4614 1.1287 13.452 13.5 36 107.5996 2.0318 1.3228 -238.9248 17.7453 .9358 8.626 9.0 37 108.5996 2. 0358 1.3429 -239.3973 18.0160 .7340 5.420 5.0 38 109.5996 2. 0398 1 . 3622 —239 . 8654 18.2746 .5245 3. 346 3.0 39 110.5996 2. 0438 1.3807 -240.3292 18.5223 .3084 2.035 2.0 40 111.5996 2.0477 1.3984 -240.7889 18. 7599 . 0863 1.220 1.0 41 112.5996 2.0515 1.4154 -241.2445 18.9882 1.8589 .723 0.7 1181.8 ^oJe. 97.7214 1.9900 1.0472 -234.0071 14.0486 2, 1568 143.506 * Six decimal places were used in the original calculation. 28 evaluated both by the approximation method used for Type VI and by making use of the cnaracter ist ic property of the T function and the table of log T' (p) . The latter method is shown in the table. For this type the origin is expressed as mean - and using unit intervals, the mean for the example is 2b. 59 and the origin at 10.003, The mode is equal to rne mean which in the ex- p(p-3) ample is 32.05, and its corresponding ordinate is 125.9 (Table XII.) XI. MEANiIJU OF CORRELATION Without going into technicalities, tne meaning of correlation is easily explained. If we have two series of paired numbers, e.g. , Heights and arm lengths of each individual of a group, or prices of flour and of cotton on certain dates, or marks in two school subjects of individual pupils of a school, we are interested in the connec- tion, if any, between the two sets of figures. There may be little or no connection between two such sets of numbers as in 'cne case of ages and heights of a group of adults. Again, there may be a very close relation, as would be shown in the case where the pairs of num- bers are the radii and corresponding circumferences of a set of cir- cles. It is clear that the set of pairs of numbers representing height and arm length is different from these two extreme cases. We express the connection between arm length and height roughly by say- ing that in general a tall man has a long arm, but given the height of a man we cannot compute the length of his arm. We need something besides the words "in general" to tell us how the two sets of numiber£ are related or correlat ed . we seek then some number which will measure for us the degree of this relationship and of this correla- t ion between paired numbers. Many measures have been devised but the one in widest use is tne measure often called Pearson's coefficient i ' * ^ _____ , . . y ■ ? ^ ^ ■' J* - ■ ' ■ ■ ' ■ 0* ,-. 'C;‘>'.. *'.- V .v "I ■ ■ '*'*'vi''''tj|^*‘ «! .•y>. !;I ff'vV ^ ’Sf^'UiJXPf'Ca .«.■■ )*r W ^'i f i t . - Kp ,n^' ..^^. rjo% a I ai’xri. 'iO ‘■^'irv:/ 'Qof yiAun ?;iip it '■ ■r' 6 - • • ■ •■ ^ •■ ■ L- - -. • • ‘ ■ ' '.'^ 'V '■ ■ :m ^.rf^ i ‘‘-j T*f - f* A .1 A.>. # ♦ A I • - 4 . -^ • ik . ^ « ««>/ ■ ‘ • * ). .»>- . -L t . ‘"iM ■ ‘ ■ ; 29 TABLE XI Calculation of Constants for Type V Example : Distribution of Weights of 971 College Women. y = y, *“P X X e p = 4 + 8 = 17.652903 Pi Y = (p-2)^A^Tp-^ ~ 212.601063, the sign of y is the same as tnat of>- . p-1 T(ip-i) Since T(p) = (p-l) r(p-l) , log T(16. 6529) = log 15.6529 = 1.19459481 14.6529 = 1.16592359 13.6529 = 1.13522491 12.6529 = 1.10319008 11.6529 = 1.06643402 10.6529 = 1.02746785 9.6529 = .98465781 8.6529 = .93716168 7.6529 = .88382604 6.6529 = .82301100 5.6529 = .75227130 4.6529 = .66772372 3.6529 = .56263778 2.6529 = .42372088 1.6529 = .21824658 log T(1.6529)= 1.9545121 12.89960415 log Vq = 28.848329 K^flXv.rrDO 76 / • j* %• W# - •ca r«>< cv'^- t-ir . -^v.r,v\ r"'*... -"• •Li \ W-''‘0:' y_ 5. '0- •\ • TT et - It d.:j < > I •‘*;-nr » i, {:r\^ • • ■ :*i - - •• *- . GioX ■3 L^v.f.vKO'* ' ^ 1 ^ . (, » '.r* . . .■ f . Jxijoi * .V --; n. ^ j!;^ .rj n* .rr^ 30 TASLE XII Calculation of the Curve of Type V Example : Distribution of Weights of 971 College Women. Y y = X Y “P e X Origin = Mean - — r- = 23.585479 - 13.582313 = 10.003266 p-3 Mode = Mean - 2V p(p-2) 22.046671 X X log X -p log X 1 X (-yiog e) log y y y 17 7.003266 .84530062 -14.922010 13. 184058 . 742261 5.5241 5 . £ 18 8. 003266 .90326725 -15.945289 11.536724 1. 366316 23. 244 23. C 19 y.00o266 .95440009 -16.847932 10.255330 1.745067 55.599 56. C 20 10.003266 1.00014181 -17.655406 — 9, 230132 1.962791 91.789 92. C 21 11.003266 1.04152161 -18.385880 8.391278 2.071171 117.807 118. C 22 12.003266 1.07929943 -19.052768 — 7.692195 2. 103366 126.872 127. ( 23 lo . 003266 1.11405244 -19.666260 — 7. 100637 2.081432 120.624 121. C 24 14. 003266 1.14622934 -20.234275 6.593567 2.020487 104.831 105. C 25 15. 003266 1.17618581 -20. 76o094 — 6.3.54091 1.931144 85.338 85. C 26 16.003266 1.20420862 -21.257778 - 5.769539 1.821012 66. 224 66. C 27 17.003266 1.23053235 -21.723469 — 5.430214 1.695646 49.619 50. C 28 18.003266 1.25535130 -22.160595 5. 128595 1.559139 36 , 236 36. C 29 19.003266 1.27882824 -22.575031 — 4.858716 1.414582 25.977 26. C 30 20. 003366 1.30110090 -22.968208 — 4.615820 1.264d01 18.678 18. C 31 21.003266 1.32228683 -23.342201 — 4.396053 1.110075 12. 885 13. C 32 22. 003266 1.34248715 -23.698795 — 4. 196262 .953272 8.979 9.C 33 23.003266 1.36178950 -24.039538 — 4.013842 . 794949 6. 236 6. C 34 24.003266 1.38027034 -24.365778 — 3. 846621 .635930 4.624 4.( 35 25.003266 1.39799674 -24.678701 — 3.692776 .476852 2.998 3.C 36 26. 003266 1.41402790 -24.961697 — 3.550764 . 335868 2.167 2.C 37 27.003266 1.43141630 -25.268653 — 3.419270 . 160406 1.446 l.£ 38 28. 003266 1.44720869 -25.547435 - 3. 297168 .003726 1.00$ l.C 39 29.003266 1.46244690 -25.816433 — 3. 183485 9.848411 .705 l.C 40 30. 003266 1.47716853 -26.076313 - 3.077381 9.694635* .495 0.£ 41 31. 003266 1.49140745 -26.327671 - 2.978121 9.542537 .348 0. 969.6 12.043405 1.08074929 -19.078362 — 7.666558 2. 103409 126.885 * -10 '61 of correlation, or the product -morrient coefficient, or simply the correlation coefficient . It is universally represented by the lettei r. The correlation coefficient varies from -1 to +1. For the ex- tremes of the range the correlation is perfect . In perfectly cor- related series, given one of any pair of numbers, we can find the other by solving a linear algebraic equation. An example of this perfect correlation is that of the radii and circumferences of cir- cles in Which tne correlation coefficient is +1. If a decrease in one of the pairs of a perfectly correlated series is accompanied by an increase in the other, the coefficient is -1, If r=0 there is no tendency to this linear relationship though there may be a close con- nection of another kind. The correlation coefficient then may be described as a mieasure of the approach to linear relationship. XII. CALCULATION OF r Tne correlation coefficient, r, is defined by the formula r = where x and y are the deviations from their respective means. O' and CTy are the standard deviations of the two distributions and W is the total number of observations. In the given sets of data of heights and weights, y represents height and x represents weight, an arrange- ■ ment which is purely arbitrary, for y might just as well have stood for weight and x for height as far as results are concerned. How- ever, as it has been arranged, the columins give the frequencies of each weight int erval distributed according to height, and the rows give the frequencies of each unit of height distributed according - - - ' • ■■ ' ' '■ jl' F v‘ * ' '*♦, -^'i: V * ' ‘ it i ±. ■ ^ f' t w ,'i ■'■, V .“w •• ' “r^ ' ■■€ ' • , v: V'v. •nplv ,r / :^i 14 . , • >r; ■'.f'. . < '-K ■X.- ♦ ■ ^ ■:.«* '■ ’ '.Sehi;*/. ■> . ' -./ f .W ^ / ,- '.. ■ f, ■> f rg -^5 • ■« '.',Jb,‘'*»^ -, ■ \.,.i .cl* ) «L u A — . __ ^ A L. 1 - ^ - .'w * L .^r AJ* .. J... . c . _ "^^RH K' ■'- . ’. ■ '- 'i *. ^ *’ ’i ji *^vi|(. ■" i^-i iC-~^ -ui- tl,i -5 t Tiii'.'i f. rM»4 Air.x *'l« f '''■jk.J^t’>'.' ‘?'\ , 1 A ”'f ^ *■ ” J) ‘ * 1 y ■ ''^ 'aJKw 4 ''i|(. "" -,- fJt . '■' . ' ■ ■•" -'liF- . tit^ -^o. ^rj;. >0 xO -41 _/:,v . d ,^»ji 4:46. . ,« to weight. The total frequency distributions for tne rows and col- umns give, as is to be expected, tne original frequency distributions (cf Tables II ana Xlii). If there seems at times to be a discrepancy It is due to the fact that in making the correlation tables, tne measurements wnicn occurred on the division point between intervals were not divided as in the original distributions, but were given alternately to the class above and to the class below, i^/iany times the number of such occurrences was not even and one of the classes received one more observation than was given by the original tables. The discrepancy of several hundredths which is seen when tne original standard deviations are compared with those of the correlation tables IS due to the same cause. The method used for finding r is given by Table XIII. In that table, d represents tne deviation from any arbitrary point, taken as in the summation method given above, so that tne least possible arithmetical work is involved. In the example (19 year mien) this arbitrary origin was taken at height 68 ana at weight Ibb as was tne case in obtaining the means and standard deviations of the distri- butions in order to compare one with tne other, so in finding the standard deviations which are necessary for the calculation of the correlation coefficient, the method of obtaining the moments can be used, since as was said before the formu.la for the square cf the standard deviation is that of the second moment ) . Since x is N the deviation from tne mean, the -£fx must necessarily be 0, but us- ing d, the aeviation from an arbitrary point, the ^fd is not equal to 0 and in order to get tne standard aeviation about the mean tne f d^ ^ f d summation — ;:r- must be decreased by the square of o(=‘^=^); that is, N /T ^fd^ ,^f d^ ® y “ — W~ " value a given in the vertical column of the ,e' 't.fr ^ •f '‘if 4 ^ I ■ '> ,i’. :; 'T , <■ 7 - ' 7 : .fe I '-h* i .. A - • i C 2 t' ll C * »A, ' •; . .] . i. .'*‘ -a O' t; ' Jbi- «, 6 - ^ • *■* V ) ’* ■ ' , •' ''■ . i“ ' t^'r • ’W '“i '■ ^"'■r’’f -J . . ;• • •• * » »» .. -jv"' . '- X r . ^ ■ 4 { *■ vA-**'' ^ •• . • * -7 4 .. l. » . . /■ . 1 - ■-■ ‘.rtw : %>-• fty ■,13 . ;.; *' «• ' v;'«' ■': ■ ' l.« -V ) I - u*. ,.:;.r ■,*>‘^.'i.’,A£i. . .< . ^ ■ ", Tf-^' .', 1 ’»/f ■J - "y » ,C - '• > ' i* .' ■, « ■ ii - •: : : ,c ... >' ■•/' • 1 .zy: vr.c; ?■.■. u .v. ( w AvkX - ^12 4 £»" V ■:o - ( ’•■V«* • ■ '*• ' (t i 1 . ;,;,or ' '■i'J! tr; 1 4 ^,, ‘‘ j 4 : '7 ■' !• • 'i‘' :i . V .: _■■ ■ .t- : • -■ . . V t ..iV . 3 .,^ 1 53 table, is the value of the frequency, as given in each square in eacn row, multiplied by the respective deviation d of the weights (given in tne row just below the table). Tne quantity ad then gives tne product of the x and y deviations from an arbitrary point weight- ed by tne proper frequency, and the row and column ad must check. In order to nave the-f^xy about the mean, must be decreased by the product c^Cy as shown in the table. Then we have r = <^ad « CvCv ~¥ ^ (T cr ^x y In the examiple, r = ,5 approximately. The remaining groups show correlations whose values are near that of the example given above. The 20 year group is the only one with a correlation higher than .b; with its coefficient of .516, it seem^s significant that the most miature group averages tallest and heaviest, and also snows the greatest tendency for heaviness and tallness, lightness and shortness to be associated. In agreemient with this we see that the youngest group shows the least assoc iatior between weight and height, with a coefficient of .425. The corre- lation Of the weights and heights of the 18 year group is expressed by a coefficient of .495, while the freshman group has a slightly lower coefficient of .478. Tne heights and iveights of women showed the least correlation of all the groups, their correlation being ,411, Since there is perfect correlation when r = ±1 and since the variables are independent when r =0, we conclude that there is a fair amount of correlation between heights and weights, or that weigl: ; is to somje extent dependent upon height. 1. See 0. B. Davenport ’’Statistical Methods”, p 20,45; also Elderton TABLE XIII Calculation of Correlation Coefficient for Height, ana weight of 991 Freshmen Age 19 \Ve L a k. 't Joo '7^ //O /is /Jo ns '70 ;4r /U /50 /•a /•h nis /*> IZS /20 /fi" tio 7oo 'KtjU c/ a ad Sh 1 / z vz 'zV '5 4o V / I -7 -7 H -4 //z u 1 ( / / / S' -6 -30 ri(S -'7 /oZ 43 f X X X X 4 3 X. / 3 xi -/ 3o kSb ZTT 4V ;z. 1 z s X 4 4 2f 4 4o -4 -li>0 L4o -/o4 / / X /I 4 4 IS ;2/ 4 7 ^4 -3 -zsz 7S'4 ~IX3 34y U / / z t IS IX n Xo 4 /4 4 /zz -z -1X3 ^7 /7 / 4 14 n x4 3X 13 xs 7 7 '^1 '/ -'Sf ‘^1 -64 44 •Ki &7 1 X z 4 J /) !% z(> xi /7 // IX 7 3 / ISS 0 0 0 0 - ^7 ( / z X k /r /r !$ /3 X7> /4 4 y 3 !3i ! nL /34 z/x O' 3: 7o / 1 / z / /5" 7 /3 XI IS XX n 7 3 /3f X zii SS!> X?4 s4i 7/ / / i" /o 4 J /O IX S X 3 / 7/ 3 Z/3 43f x^f 443 — r-2 X / 3 X 4 y 4 / X? V /tx 4z 73 / 1 ; / z 1 X 13 3" (.s 3ZS -/r Z3£^ 77 X / z 1 / 7 4 4x xSX XTZ 73" / / 1 3 7 XI x4 /6g 1. tiU 3 / /o J 4 -2/ 4 3^ xs /o/ ixil^ IZ$ "1 fs 77 ^3 3 / /O 4 yy/ -ox "7 3<4> ?73 77r 3^0 ///3o d 3 / /o ,r 4/5^ V 47 si IX •7 7 -•/s 7^3 ''4> -P -7«. -3g /A no St> 7x y4 ‘Hi 3-?0 •lx Hi- lii- n 0 4s »74 v& >lx 3to ic3 37/0 - 0 *’= Syi - 01 9i> 73 ff/ y ■ ^ 2 J-y = / .o S^JL4Bi> / O j/. iff 'J^Zo i .4 -JYJ- = ^)C .5-00 3-0^" z, 7 /V ,/"' ■ *. ' ■ ', '■ '^;.' - ■ •■ 5 ?* ii jj-i ’■' ’ , '' .'•* ' ' 'I .Tf" , ^ ., y -»r**T- r.^'iSff. ---u . ^,, < -<«L !* ^ ■, -^ ^ A}-- - ,w ^-'’.-ii'ri '■?;_‘>yii,r^4yj4i •Ipifflpl i l^.l t 4 t I ^ * • >.-»i . Jn < CTu » p. ^ ' I . rf'MBf’**" ' -T- ' — , - I , J‘ ci‘ii.'ikji;;ii>i„t ^ , . i -•' w f?^-'-l V Uj.-.b':.^:^ jA \.,\ir. , ■‘ » •>- - 1 t= A. !' i ■Ul.' y i ,--U4 . ,| .^'.K:?; ‘*’i' T ^ » * 't' /A ‘ ■**;*' 'j / t -ft < « ?5 , »* - '10 . ’'»« ■>! fi, - - ■ -■ ’• J .•', , ■•yitV.vr-;, li/ 1 ifcJESTrSR?^ ■ *» \*^ 4 t-...'s p *■ ■ ■ '1 ■ ■'’»■'*' < 8 ^ < 34 « . ' ♦'.. ' '‘' . ;d’-> -^I#- ^ -’Xy^ •i." --1?' '<*•- * . ^'4 . 4 '^ ‘.i idOifi, ■ ;r.A-#' '■"A.! '>dyir^ ,TT!t: 55 XIII. SUMARY To sum up this investigation I have made a table to show in a concise way how much the means and correlation coefficients of the heights and weights of the various classes of freshmen differ. I have expressed the means with their probable errors, which give the measure of unreliability of values calculated; the probable error is a pair of values lying one above and the other below the value deter' mined, and the true value lies some\7here between these limits. The fig-ares which follow, sum up the various distributions and indicate the closeness with which anthropometric measurements are fit by the normal curve and by several other types. Group Mean Height Mean Weight Correlation Coefficient Freshman Class 67.699±. 047 135.94+.34 .478 17 year Freshmen 67.826±.064 134. 75±. 48 .425 18 year Freshmen 67.73G±.050 135. 44±. 36 .495 19 year Freshmen 67.859±.051 137.51±. 36 .500 20 year Freshmen 67.985±. 053 139,01±.40 .516 V/omen 63.313±,048 117.93+.38 .411 eta /v t t.T\ t 7L £ ( i V' . * \ 1 LQ ki. I • Me iak't fn. incA.es * HO ns tZ9 Its >y> iss /fp /fr 'Sa iss /io i(,s Ho ns /So /fs i4 : •• Xirt 1 1 i. ♦ j ■» Itr Ho nS IZO /Zjr /So ISS' /fo if^ /S^S^ /(.o ttys' /Jo /J^ /io ffo Ijs Zoa i i V io 4 / L2 43 M M i>7 ,(>f %] 71- n 13' n