]\gt ; to the of -V .
On the Reconstruction of the Stature of Prehistoric KARL PEARSON , F.R.S. , Universit , College , London .
* Received June 6 , \mdash ; Read June 16 , 1898 .
[ PLATES AND 4 ] CONTENTS .
PA ( 1 ) Introduction . . . . . . . .
170 ( 2 ) Theory one organ . . . . .
170 ( 3 ) heory for any 1lumber of organs 172 Racial variations . . . .
175 ( 5 ) Data available for stature formulae . . . . . .
178 ( 6 ) , ( 7 ) Discussiou and generalisation of results , . . . .
182 ( 8 ) Reconstruction of as corpse from bones containing their animal matter 186 ( 9 ) Reconstruction of ature in life from bones containing no animal matter . . .
190 ( 10 ) GcneraI formulae for the reconstruction of stalure . . . . .
196 ( 11 ) Reconstruction of Aino stature . . . . . . .
198 ( 12 ) Statu and long bones of anthropomorphous Apes . . . . .
201 ( 13 ) Probable stature of Palaeolithic Man . . . . .
205 ( 14 ) Probable of Neolithic Man\mdash ; ( a ) Greab Britain . . . . .
206 ( b ) France and BeIgium . . . . .
206 ( c ) Special cases and groups . . . . .
207 ( 15 ) Other races .
DoImen builders and Guanches . . . . .
209 ( 16 ) Probable stature of the Naqada Race . . . . .
211 Probable stature of otohistoric Races\mdash ; Round Barrow Bribish . . . . .
212 ( b ) Romano-British . . . . .
( c ) Romauo-Gauls . . . . .
213 ( d ) Row Grave Germans . . . . .
214 ( 18 ) Probable stature of the Anglo-Saxons . . . . .
216 ( 19 ) Probable stature of the Franks . . . . .
217 have received constant aid and assistance in the laborious arithmetical work of this paper Mr. LESLIE BRAIILEY-MOORE .
To Miss ALIOE LEE I owe the fundamental tablt on p. 180 , and the preparation of the test tables on pp. 188 and 189 . .
G. U. YuLN has prepared the diagrams and repeatedly assisted me with suggestion and criticism .
VOL. CXCII.\mdash ; A. PAGE ( 20 ) Probable of the rench of the Middle Ages . . . . .
219 ( 21 ) On the sbature of giants and dwarfs .
General considerations and regression culves .
220 for ginnt and statures . . . . .
225 ( 23 ) races . . . . .
228 Probable staturo of Bushmon . . . . . . . .
233 ( 25 ) Probable stature of Akkas . . . . .
235 ( 26 ) Probable stature of Andamanese . . . . .
( 27 ) Probable slature of European Neolithic dwarfs . . .
238 ( 28 ) Conclusion . . . . .
240 ( 29 ) Table of and sexual ratios for divers races . . . . .
242 ( 1 .
) THE object of this paper is to show , by the use of a special case as illustration , the true limits within which it is possible to reconstruct the parts of an extinct race from a knowledge of the size of a few organs or bones , when complece measurements been or can be made for an allied and still extant race .
The illustration I have is one of considerable interest in itself , and has been considered from a riety of standpoints by a long series of investigators .
But I wish it to be considered purely as an stration of a general method .
What is here done for stature from long bones is equally applicable to other organs in Man .
We reconstruct in the same manner the dimensions of the hand from a knowledge of any of the finger bones , or the bones of the upper limbs from a of the bones of the lower limbs .
Further , we need not confine our attention to Man , but can predict , with whac often amounts to a remarkable degree of accuracy , the dimensions of the organs of one local race of any species from a knowledge of a considerable number of organs in a second local race , and of only one two organs of the first .
The importance of this result for the reconstruction of fossil or prehistoric races will be obvions .
What we need for any such reconstruction are the following data : The mean izes , the variabilities ( standard-deviations ) , and the correlations of many organs in an extant allied race as it is possible conveniently to measure .
When the correlations of the organs under consideration are high ( e.g. , the long bones in Man ) , fifty to a hundred individuals may be sufficient ; in other cases it is desirable that several hundred at least should be measured .
The like sizes or characters for as many individual organs or bones of the race should then be mcasured as it is possible to collect .
It will be found always possible to reconstruct the mean racial type with greater accuracy than to reconstruct a single individual .
An appreciation must be made of the effect of time and climate in producing changes in the dimensiolJs of the organs which have survived from the extinct race .
( 2 .
) .
the above data to exist in any particular instance , we have next to ask what is ) eoretically the best method of dealing with them .
There cannot be a doubt about the answer to be given .
If we know an organ , then the most vaiue .
an organ is that given by the regression formula for the two CONTRIBUTIONS THE THEORY OF EVOLUTION .
organs .
Let the mean sizes of A their standard deviations , their coefficient of correlation , then the most probable value of .
a given value of A is , or . . . . . . . . . . . . .
where and are constants for the pair of organs under consideration .
The probable error of such a detelmination is ) .
Now there are several points to be noticed here .
If be small , the probable error of reconstruction will be large , if the organ is to be reconstructed for a individual .
No ingenuity in constructing other formulae can in the least get over this difficulty ; it is simply an expression ' the fact that races are variable .
Any formula which professes to recon struct individuals with extreme accuracy may at once be put aside as unscientific .
On the other hand , if A be known for individuals , the corresponding rnecvn value of the unknown organ may be follnd with a probable error of , and with Increasing accul.acy as increases .
( ii .
) Anthropologists and anatomists have frequently assumed that the ratio of two organs , , is the measure to be ascertained in a reconstruction problem .
They were soon compelled to , however , that this varies , and lnave tabulated the ratio for three or fonr ranges of the A. Such a table , for example is given by M. for the ratio of stature to the of the six long bones .
He gives the ratio for three values of each long bone .
He also in a second table gives values of the ratios which are to be taken when the long bones exceed or short of certain values , i.e. , in cases of what he terms macroskely and microskely .
The regression formula shows us that : and since is never small as compared with , this ratio can never be treated as constant .
Accordingly , while a table can be constructed which will give quite good reconstruction values , by determining the mean value of fdr each value of , we see that it is theoretically an erroneous principle to start from ; no constancy of the ratio ought to be expected .
The theory of regression shows us that the most probable value of is expressible , so long as the correlation is normal ( or at least " " linear as a function of A. *'Memoires de la Societe d'Anthropologie de Paris , ' vol. 4 , pp. 347-402 .
Sir GEORGE HUNPHRY gives a table the ratio for stature in his " " Treatise on the ( 3 .
) So far we have only with the reconstruction of the most probable value of from one organ , but we may propose to find the most probable value of from tyans A. Let represent the correlation coefficient of and the organ A , the correlation coefficient of and ; the S.D. of , and of the organ the mean of B. and of ; let be the determinant 1 . . . . . .
1 and , the minor corresponding to .
Then the general theory of correlation shows us that . . .
( ii .
) is the most probable value of , and that there is a probable error in this determination .
Thus we reach again a formula of the character or , is expressible as a linear function of the organs from which its value is to be predicted .
This again supposes normal , or at least " " linear\ldquo ; COlTelation .
Now are points to be noticed here .
The linear function which will give the best value for is unique .
For example , some anthropologists have attempted to reconstruct stature by adding the lengths of femur and tibia .
The proportions in which femur and tibia are to be combined are given once for all by the formula , and they are not those of equality .
I have succeeded in proving the following general theorem , which settles this point conclusively .
Given any linear function of the organs . .
, say Skeleton Cambridge , 1858 , p. 108 .
Many others have been given by French writers , in some cases with several values cf for three ranges of stature 01 of long bono , ROLLET , etc Dr. BBDDOE has given a rule which really amounts to making a linear fullction of bis values for and are widely divergent from what I have obtained by applying the theory of correlatioo .
' Journal of the Anbhropological Institute , ' vol. 17 , 1888 , p. 205 .
CONTRIBUTIONS TO THE THEORY OF EVOLUTION .
and let be the correlation of this expression with , then will be greatest or the probable error of the determination of by ns of its correlation with such an expression will be least , will be least , when the 's are proportional to the corresponding 's of the regression forunula .
Let 2 be the standard-deviation of the quantity Then and The best value of as determined from is ( iii .
) , with a probable error67449 .
This may be taken to be any linear function of the , since so far are quite arbitrary constants , and the constant has to satisfy the condition that takes its mean when the 's take their mean values .
Now select such a value of the 's as to give the greatest value to .
By differentiating with to the 's in succession we find the of equations . . . . . .
The solutions of these equations are . .
; or , the equation to the best value of , ( iii .
) above , reduces to the regression formula ( ii .
) .
In other words , no attempt to reconstruct the organ from a linear relation to the organs will give such a good result as the ordinary regression formula .
Tbis , of course , excludes all attempts to form type ratios of *I note that what is here demonstrated is only a special case of Mr. goneral theol 'em .
See ' Roy .
Soc. Proc vol. 60 , p. 477 .
or as a method of prediction .
We may , in fact , at once dismiss all nction formulae as insufficient which are not based on the theory of corretion .
The theory as here applied , be it noted , depends on the of the proposed ] not on any special form of the distribution of variations .
( ii .
) The accuracy of a prediction will not be indefinitely increased by increasing the number of organs upon which the prediction is based .
This fundament fact of the application of the theory of correlation to prediction has already been noticed by Miss ALICE LEE and myself in the case of barometric prediction .
The choice of organs upon which to base the prediction is far more important .
Thus , to illustrate this from stature I may remark that the probable error of a prediction of male stature from radius is to a prediction from femur in the ratio of to tlJat if one takes both femur and tibia for the prediction , the probable error is only reduced to , and further , if one takes femur , tibia , humerus , and radius , we only reach .
This latter reduction is so small as to be well within the of the determination of our means , variations , and correla t , and scarcely worth making .
To pass from the radius to the femur is a real gain ; to pass from femur and humerus , say , to femur , humerus , tibia , and radius , is no sensible gain .
Hence , one or two gans W selected worth much more for prediction than a much larger number selected less carefully .
( iii .
) It is the custom of French writers , when determining stature , to predict it from several single types of bones , say from femur , tibia , humerus , and radius , and then to take the mean of these results for the true stature .
This is not the best procedure .
Suppose the regression formulae for the prediction of from separately to be Then the mean of all these results would give that is to say , has been really found a linear relationship between and the four organs in question .
But the best linear reIationship for the four organs is where the 's are the true regression coefficients .
But the slightest acquaintance with the theory of regression shows that the partial regression coefficient is as a * ' On the Distribution of Frequency ( Variation and Correlation ) of the Height ab Divers Stations ' Phil. , vol. 190 , p. 456 et seq. CONTRIBUTIONS TO THHi THEORY OF EVOLUTTON .
rule not just of the value of the total regression coefficient .
For example , if were the radius and stature , , while is a This process of taking means may ccordingly screen some most important element , like the negative value of the partial regression coefficient of the radius .
therefore , as well as from the standpoint of discovery , the regression formula for organs will give more results than the mean of the results of the regression mulae for the .
A ctical modification of this principle will be referred to below ( p. 178 ) .
( 4 .
) The theory of regression will thus enable us to determine the best value to be assigned to an unknown organ , when the values of any other gans known , supposing the ) to which belong is a member of a race or group for which , , the regression coefficients ) been ascertained .
On what principle , however , can we extend the regression for one race to a second The regression coefficients depend upon two , the variability of the organs under consideration and their .
Now the nge in variability as we pass from one race to a second has never been questioned .
It been suggested that the correlations were racial characters , but the divergences in correlations between local races are far beyond the probable rors of the observations .
* Mr. FILON and I have shown that every selection from a race .
both variation and COlTelation .
have shown in a memoir not yet published that all natural and all artificial selection also these quantities .
How then can we hope that a regression formula as applied from one local race to another will give accurate results ?
Why should the stature formula obtained from measurements on modern Frenchmen apply to palaeolithic man I think M. NOUVRIER somewhat lightly skips this difficulty in the following sentences:\mdash ; ' ' Enfin les ethniques de proportions du corps seront dans le meme cas que , les precedentes [ les variations individuelles ] .
Il a des races macroskeles et des races microskeles , comme il des individus de ces deux sorts , les variations individuelles sont bien plus grandes que les variations ethniques les plus cusees .
Or les coefficients moyens des os de grande longueur tendant a abaisser la taille et cellX des os de faible longueur tendant a l'elever , il s'ensuit qu'il sera tenu count dans un certain mesure de la macroskelie des races colnme de celle des individus dont les os seront absolument longs et de la microskelie des races comme de celle des individus ayant des os absolument courts.\ldquo ; If we admit for the moment , which I should not be prepared to do generally that the individual variations in a local race are greater than the ( ethnic variations\ldquo ; or ( iivergences between the means of local aces , M. MANOUVRIER'S conclusion by no means follows .
' Phil. TrallS , vol. 187 , pp. 266 , 280 , and ' Roy .
Soc. Proc vol. 61 , p. " " On Random Selection see ' Phil. Trans , ' , vol. 191 , p. 229 , and ' Roy .
Soc. Proc vol. 62 , p. 173 .
, on my page 171 .
See the results as to adius referrod to on p. 176 below .
The forynulae for stature reconstruction , whether obtained with a consciousness of the theory of reglession , as in the present paper , or indirectly by taking the means of small groups , as by M. MANOUVRIER , are based upon averages , and involve the standard-deviations , the variabilities of distribution of each organ .
Hence , the fact that individual variations may be greater than ethnic variations does not touch the real point at issue , for the formulae depend on the proportions of macroskely and microskely in each race , and these undoubtedly change .
The individual variation being greater than the ethnic , is not a valid argument for applying a formula based on the observation of one local race straight away to a second .
The validity of applying the formula for one local race to a second depends , think , upon very different considerations .
In the first place , the validity is not general .
If we endeavoured to reconstruct the radius , for example , of Aino or Naqad races from the femur or tibia by a regression formula obtained from measurements on the French , the results would , we might priori expect , be so satisfactory as for stature .
* The validity depends on our conceptions as to " " local races While the problem of local races is dealt with at length in my memoir on artificial and natural selection , and I do not want to anticipate the results there stated , it is still needful to cite here a theorem reached in that memoir .
When a sub-race is established by the selection out of a primary race of a group having organs distributed with given variabilities and given correlations about given means , we shall speak of its meant dus to a selection of these organs .
But this direct selection is shown to alter also the si es of all remaining organs of the organism , the variabilities of all those organs , and the tions among themselves of th non-directly selected as well as their correlations with the selected organs .
We shall speak of this result as *Allowing , as in my page 193 , for cartila and shrinking , I find the following formulae from the Frenoh measurements reconstruction of radius in centimebrer ; : In the case of the Ainos , the prediction is within per cent. of the observed value .
In the case of Naqada race , the prediction from the femur differs by 1 cenlim .
, or 4 per .
from its true value .
An error of 6 to 7 centims. in the prediction of stature of a local race which won ld correspond in magnitude is hardly likely to occul .
The explanation is that the radius is a much differentiated bono .
CONTRIBUTIONS TO THE THEORY OF EVOLUTION .
indirect selection .
The changes due to indirect selection are shown in the memoir referred to to be in many cases of considerable importance ; every mean , every standard deviation , every correlation may be altered but the following theorems govern the changes in the sion formulae:\mdash ; The formula of a directly selected organ on any number of other organs , whether directly .
indirectly selected , will change .
The regression formula of an indirectly selected organ on all the directly selected organs , and any number of the indirectly selected organs , does not change .
The regression formula of an indirectly selected organ on some , but not the directly selected organs , will change , unless the selection happens to be one of size only , and no of variability and correlation at the same time , in which case the formula remains unchanged .
Most local races show sensible but small differences in both variability and correlation ; if we call these diiferences quantities of the first order of small quantities : then the changes in the regression formulae between two or more indirectly selected organs will be of this order of small quantities squares and products of correlations , quantities which are themselves less than unity , or what we may term a quantity of the third order ; further , the changes in the regression formulae between an indirectly selected organ and some but not all the directly selected organs be of the first order of small quantities the correlation , or what we may term a quantity of the second order .
To sum up , then , .
it would appear that the ression formulae in general will change from local race to local race , buc that a particular set ( see above ) exist which would not be changed at all , while many others , supposing to be the chief character selected , would only be changed by of the second or third order .
It will be obvious then that a knowledge of a considerable series of regression formulae of two local races will enable us to ascertain to some extent the nature and amount of differentiation which has gone on from a common ancestral stock .
Further , if we have not sufficient data for one local race to find the variabilities and correlations of its organs , but if we can find fairly closely the mean size of its organs , then the degree of consistency of the results obtained when these means are inserted in the regression formulae for the second local race is an indication of the amount of differentiation which has taken place .
The larger the number of organs we include in a regression formula the more likely we are to embrace the directly selected organs , and so to obtain a formula which remains unchanged for the two races .
Thus we see that the extension of the stature regression formulae from one local race\mdash ; say , modern French\mdash ; to other races\mdash ; say , palaeolithic man\mdash ; must be made with very great caution .
The extension assumes that stature itself has not been * A selection of the mean sizes of two , which would aIter their relati poportions , does not of course involve a selection of correlation ; in other words , selection of mean relationship does not neoessarily connote a selection of differential relabionship .
VOL. CXCIL\mdash ; A. 2 selected , however widely changed by indirect selection , ( ii .
) that the formulae involve the directly ected organs closely correlated with , or that the selection has been principally one of and not of variability of , or correla t between , these organs .
The real test of the applicability of the is whether or not they give for another local race of which we know the stature , results in with themselves and with the known stature .
I take it that thejustification required for applying our formulae to palaeolithic man is not the statement that ethnic are less than individual intra-racial variations , but is to be drawn from the fact that our formulae , based upon measnrements on the French , give results very fairly consistent among themselves with observation for such a divergent race as the Aino .
Such results seem to indicate that racial differences in sta.ture are not the of direct selection of stature , and that the selection of the long bones has been rather a selection of their absolute and relative sizes than selection , in the first place , of their degrees of variation correlation , although these have to some extent undoubtedly Our general theorems will to some degree indicate manner in which differentiation has taken place .
Suppose there has been a selection of femur and tibia , but not of humerus and radi u. Then the regression formulae for stature on femur and tibia , for stature on femur and tibia with one or both of the other two , humerus and radius , ought to give identical results ; but these results ought to diHer from those given by the formulae for stature on humerus or on radius , or on both together .
Practically , however , we have in many cases so few bones to obtaip our from ( and these bones themselves parts of different skeletons ) , that the probable errors of these means quite obscure the deyiations in stature as obtained from various formulae and due to the influence of selection .
From this standpoint a partial practical jcation can be found for taking the mean of the divergent reconstructionls of stature given by a series of regression at any rate { or the ) when the gences are not vely a These ences l ) be due to errors in the mean lengths of the long bones , or to selection directly of one or more of the long bones , or even to some direct selection ) of ature .
But as in of these sonrces of errors we can only suppose some positive and some ative , the mean of all the may to some extent eliminate these quite unknown and unascertainable divergences see .
Generally .
owever , I should expect the stature in which two or more agree , to be more probable than the mean of several divergent formulae .
( 5 .
) On the for Stature Regression Formuloe.\mdash ; The only data available for the calculation of the correlation between stature and long bones occur in the measurements made by Dr. PVOLLET on 100 corpses in the dissecting room at Lyons .
This material has already been made use of by Miss ALICE and myself in our memoir , " " On the Relative Correlation of Civilised and Uncivilised Races , \ldquo ; so that : De Iensuration des Os Longs des Membres , ' par Dr. ETIENNE ROLLET , Lyons , 1889 .
' Roy .
Soc. Proc vol. 61 , p. 343 ONTRIBUTIONS TO THE THEORY OF VOLUTION .
all the coefficients of correlation and all the variations of the long bones have already been calculated .
I owe to Miss ALICE LEE the knowledge of the additional constants required for this farther investigation , and embodied in Tables I. and II .
below , which embrace all that is needed to fully determine the correlation of stature and long bones .
The treatment of Dr. ROLLET'S material was not to be briefly settled .
He had measured only 50 bodies of each sex , and this number included a great variety of ages .
M. MANOUVRIER in determining his table of statures has at once excluded from his calculations all the males but 24 as nile , and all the females but 25 .
Now , although the corl.elations between stature and long bones are high , it would be oite hopeless to attempt to calculate them from 25 cases ; 50 cases are hardly sufficient , 25 impossible .
It seemed , fore , necessary to include all Dr. cases , and the question now ises how far the inclusion of the senile ones will affect our results .
Taking 50 as the age which stature begins to decrease , we notice that of the 25 lowest statures recol.ded by ROLLET , 18 are of men over 50 , and of the 25 highest statures , 17 are of men over 50 .
In other words , there appear sensibly as many senile statures above as below the median stature .
Of women there are 16 years old with a stature greater than the median , and only 14 women over 50 under the median stature .
Turning to means , we notice that 24 males under 60 years had for mean centims. , and 26 males over 59 years centims. , 25 females under 60 had for mean stature centims. , and 20- females over 59 had centims. 87 females under 70 had a stature , and I3 over 70 gave centims. , an absolutely greater stature .
24 years was the minimunl age .
From this it would that whatever shrinkage may be due to old age , it is not of a very marked character in these data , or largely disappears when a body is measured after death on a table ; the senile sboop may then be largely eliminated .
But there is another point to be noted : we shall not directly makc use of the mean stature as obtained ffom ROLLET'S data , except to test how far our formulae will reproduce ROLLET'S results .
What we shall make use of from ROLLET'S data are the standard-deviations and coefficients of correlation , and these will hardly have their values sensibly influenced by such comparatively small senile changes as are to be found indicated in ROLLET'S measurements .
Accordingly our constants are calculated by including all ROLLET'S measurements , namely , on 50 of each sex .
The following results were found:\mdash ; he bones shrink with old age , like the stature , the correlation lvould not be altered .
Tho length of a bone varies with the amount of moisture in it ( see below ) , and such shrinkage is itself a possibility .
The bones of the aged will of course be included among those of extinct races , and cannot easily be eliminated .
TABLE I.\mdash ; Correlation between Stature and Long Bones .
Pairs of organs .
Stature and radius Stature and tibia Male .
Stature and umerus * .
Stature and femnr Stature and radius .
Stature and femur tibia The means , standard deviations , and correlations of femur , tibia , humerus , and radius , for ROLLET'S measurements , are in the 'Roy .
Soc. Proc vol. 61 , pp. 347-350 .
The means and variability of the remaining organs not there recorded were found to be as follows:\mdash ; TABLE II .
tibi Stature Humerus radius .
Without reproducing the full tables of the memoir referred to , it is of value to form the correlation tables , which serve as the determinants from which the regression formulae have been calculated .
It is only in the case of stature in terms of the four long bones that the numerical work proyed lengthy .
The general formula used is ( ii .
) ) p. 172 .
for Stature , Femur , Humerus , Tibia , dius , all measured in ROLLET'S manner , which will be discussed at length below .
somewhat low value of the correlation for female stature and humerns was tested by means of the formula where , and are organs , their standard deviations , and a coefficient of orrelation .
putting humerus , radius , and sbature , I found the correlation between ture and humerus radius indirectly ; it was table shows that the directly-calculated value was 7547 , a difference well within the errors of observation .
Thus the correlations as given for female hnmerus and stature and female radius and stature must be correct , i.e. , the somewhat lengthy arithmetic involved is not at fault .
TABLE V.\mdash ; Male .
( b ) , ( c ) , ( d ) , ( e ) , ( f ) , ( h ) , ( i ) , ( k ) 913 Female .
( b ) , ( c ) , ( d ) , ( e ) , ( f ) , ( g ) , ( h ) , ( i ) , ( k ) 782 ( 6 .
) Now these tables require a good deal of comment .
In the first place they must not be considered as extending beyond the range of data on which they are based , thus and are tho maximum lengths of bones measured with the cartilage attached , and in a humid state , is the tibia length excluding spine .
All the constants were worked out for the right members , except in one or two cases in which they were missing .
The stature is the stature measured on the corpse .
Further the measurements are made on the French race .
We shall proceed to generalise these formulae .
In the first place , the may appear strange that the probable elror of ( e ) is less than ( f ) , but difference is really less than the probable error of observations .
If be ated from the known values of and , \amp ; c. , we find it equals instead of 8384 the directly calculated value , while thus calculated instead of , whence instead of , which is in reement vibh bhe general theorem on p. 173 .
CONTRIBUTIONS TO THE THEORY OF EVOLUTION .
numerical factors are functions only of the standard deviations and the correlation coefficients , and will accordingly be if these be Let and be any organs and and their means , and their numbers , and their coefficient of correlation .
Suppose that any hygrometric changes , different method of measurement , of animal matter in the organs at time of measurement , etc. , cause us to measure and instead of and , and let , and .
be the resultin ; ro characters , then clearly , standing for summation:\mdash ; , or , Thus a correlation coefficient will be quite unchanged .
A ression coefficient will be changed or not according as the ratio of two standard viations is changed or not , or according as to whether sensibly differs from unity .
Now in stature or any of the long bones with which we have to corresponding to amount to 1 per cent. of the value of or , but the multipliers like and are not only quantities differing in the second order from unity , but probably very nearly equal to eac , other .
Hence it is reasonable to suppose that changes in the condition of the bones , and stature measured on the living or on the corpse , while sensibly affecting MB will produce little or no effect on the nerical constants of the regression formulae to ( k ) .
We shall find that this a priori conclusion is borne out by actual nents .
Hence we conclude that Tables V. and Vf .
may be applied bo stature measured on the living or the corpse , to bones measured humid or dry , with or without the cartilage , provided .
modifications are made in the values of the fve means .
We even go so far as to predict that provided MT be properly altered , the stature from t , ibia formulae will not be much modified , even if the tibia be measured with instead of without the spine .
The change , however , in the ression fornlulae the femur is measured in the oblique position is more likely to be of importance , and the correlation between stature and oblique femur has accordingly been worked oub .
If denote oblique femur we have:\mdash ; Male Female whence for we find : Male Female Thus the regression coefficient is not changed more than55 per cent. for males and per cent. for females , even in this case where the difference between the maximum and oblique gths of the femur has been much insisted upon as very cant with regard to stature .
Putting in the lengths of the means as found on the corpse , we have : Male . . . . . .
Female The corresponding formulae for the stature in terms of the maximum length of femur are , as we shall see later : Male ( ii .
) .
Female The extreme oblique femur lengths are for males and , and for females and .
Let us calculate the stature of these individuals directly from and indirectly from ( ii .
) , by putting for males and for females .
We find Male Female Male Female The differences here in these extreme cases are absolutely unimportant for the deterrnination of stature .
In other words , the changes in the regression equation are insignificant , when we even make such a as from oblique to maximum length .
Accordingly we have the rule , if the oblique length of femur be given , the equations for the maximum length can always be safely used if we add for the male and for female to the oblique length in centimetres before using equations of type ( ii .
) .
So far we have generalised Tables V. and VI .
, having regard to the nature and condition of the organs when measured .
We see that the regression coefficients will remain sensibly constant .
Our general considerations on pp. 177 and 178 indicate the limits under which these regression coefficients may be considered constant for different local races .
But the constancy of the regression coefficients is not sufficient to preserve the constancy of the linear reconstruction formulae for stature .
It would be of no service if yaried from local race to local race absolutely independently .
Now if be the mean of a not directly selected organ , and .
the means of any other organs , the constant part in a reconstruction formula will with the notation of p. 172 , be ; .
CONTRIBUTIONS TO THE THEORY OF EVOLUTION .
It is shown in the memoir on selection to which I previously referred , that this expression remains the same for all local races , and equal to its value in the original stock under precisely the same conditions ( stated on p. 177 ) the regression coefficients themselves constant .
Hence we have the same degree ofjustification in applying our whole ature reconstruction formula from one race to a second , as in applying the ression coefficients .
( 7 .
) -examining Tables V. and .
with a view to drawing one or two general conclusions before we proceed further , we notice : The probable error of the reconstruction of the stature of a single individual is never sensibly less than two centimetres , and if we have only the radius to predict from may arnount to centims. Hence no attempt to reconstruct the stature of an individual from the four chief bones can possibly exceed this of accuracy on the average , at any rate no linear formula .
* No other linear formulae will give a better , or indeed as good a result as the above .
The reconstruction of racial stature is naturally more accurate , since if we reconstruct the mean fro bones of one type , the probable error is reduced by the multiplier .
At the same time we must bear in mind that possibly a definite , if small amount of direct , selection by stature bas actually taken place in the differentiation of human races , and accordingly the values of given in Tables V. and VI .
are not absolutely true measures of the probable error of racial reconstruction , even when one or more of the long bones have not been directly selected .
A direct selection of the long bones is usually evidenced by one or more of the formulae giving discordant results .
When , as will be seen later to be usually the case , several of the formulae give results well in accordance with each other , then we may assume that centims. is an approximatet measure of the probable error of the reconstructed stature .
( ii .
) The four long bones for males the least probable error , but with sensibly equal accuracy and less arithmetic we may use \amp ; or \amp ; ; then follow fairly close , ogether \amp ; or alone ; alone is sensibly worse , and is worst of all .
It is noteworthy that is better than , and the \amp ; is sensibly as good as alone .
Turning to female stature reconstruction , we notice that the order of probable errors is ] altered .
Tibia and radius now play a more important part in the determination of stature .
The four long bones still give the best result ; \amp ; .and follow closely ; then cume \amp ; , and alone ; followed at some distance by \amp ; , and alone , but both these are now worse than alone ; ] of shall to the question of the ]inearity of ths formula , when later with the stature of giants and dwarfs , see p. It musb be remembered that we have , as a rule , a number of long bones which in part do not even belong to the same skeletons .
Tbis resuIt accordingly is the probable error of a group to whom one kind of long bones belonged , rather { han the probable error of the lacial stature as reconstructed .
VOL. CXCII .
\mdash ; A. 2 all com es , as before .
Thus while in the case of men the humerus , in the case of women the tibia is the better bone of the two to predict stature from .
A simple examination shows the emphasising of the tibia cients in the case of woman .
same for the radius coefficients , but in a still more marked degree .
Both and female show in the regression formula for the four bones a remarkable feature which they have in with the anthropomorphous apes , namely , the character or the partial ession coefficient .
The longer the lius for the same vadue of femur , , the shorter be the stature .
In this point women are more akin to the anthropomorphous apes than man , for the negative radius coefficient formula ) is nearly four times as .
The tibia also ient a most double of the male , pointing in the same direction .
( iii .
) comparison of Table V. with .
shows us that man and woman are bilicy not only differentiated from a common stock directly with regard to ture , also directly with regaxd to all other long bones .
If we use female to construct male stature , or male to reconstruct female , we get surplisingly bad results , The fact ) the formula ( k ) for female diverges in a direction from that of man , which approximates to of at least one species of anthropomorphous ape , is only of course a round-about quantitative manner of indicating , what is obvious on other grounds , a sn bstancial part of the diffel.entiation of male and female took place in that part of the history of man 's lution which preceded his differentiation from the stock common to him certain of the anthro morphous apes .
( 8 .
) Before we lnodify our formulae in Tables V. and VI .
to suit the reconstruction of stature by asurements on prehistoric and other bones , we will put the numerlcal values for into these formulae .
This will erve a double purpose , it will enable us to verify our formulae on material , and ( ii .
) it will place at the disposal of the criminal authorities the best formulae yet available for the reconscruction of the stature of an adult of whom one or more members have been found under suspicious circumstances .
FoRlfULR for the Reconstruction of the Stature as Corpse , the Maximum Lengths of , and of without Spine being measured with the artilage on and in a Humid State .
TABLE .
\mdash ; Male . .
( ) .
( ) .
( ) .
( ) .
( ) .
*The probable error in these and later tables are not reproduced ; they may be considered to be the same as in V. CONTRIBUTIONS TO THE THEORY OF EVOLUTION .
TABLE .\mdash ; Male ( continued ) .
( ) .
( ) .
( ) .
( ) .
TABLE VIII .
\mdash ; Female . .
( ) .
( ) .
( ) .
( ) .
( ) .
( ) .
( ) .
( ) .
( ) .
Should the stature of the living be required from the corpse ature , then centim. should be subtracted for the male centims. for the If a left member has been measured instead of a right , a small allowance mighb be made for this on the basis of ROLLET'S means for the lsft side , but such refinement is hardl .
of service when we look at the probable error of an individual reconstruction , i.e. , about 2 centims. We shall return to the point later as a second order error in racial reconstruction .
In order to indicate to the reader the degree of confidence he may place in the above formulae of reconstruction , and also their relative value , I give below a table of observed and reconstructed statures in the case of 20 out of cases .
The individuals , in order to avoid any bias , were taken at random as the 5th , 15th , \amp ; c. entries through ROLLET'S Tables .
The observed statul.es are recorded and the differences as obtained by the formulae . .
the heading , I give the differences which would be yielded by , MANOUVRIER'S Table .
It is formulae ( f ) , ( h ) , ( i ) , and ( k ) on which I should lay most weight , and which should be used whenever the material is available .
* For the reasons for these numbers , see p. 191 below , The first point with regard to these tables is to note how , even with only ten cases , the mean errors accord closely with their theoretical values .
For example , the mean error of is centims. for male and centims. for female when deduced from the probable errors in Tables .
and VI .
, : observed mean errors in the two cases are centims. for male and centims. for female .
The mean of the mean errors is for centims. , and for females centims. ; the obseryed values are centims. and centims. for the two sets of ten cases respectively .
We conclude at once that our mulae , and therefore certainly any other linear formulae , will 1lot give results with a error of less centims. for the individual stature .
In our case the worst error is one of 8 centims. ( about 3 inches ) in the stature of a man of 47 years of age , who must have had a remarkably long trunk in proportion to his leg and arm-lengths .
It would be impossible to have predicted his stature any closer without taking into account the correlation between stature and trunk .
The preservation of the vertebral column is comparatively rare , and at present there are absolutely no statistics on the relationship between the dimensions of part of it and living stature .
We must therefore content ourselves with a probable error of 2 centims. , and expect , but rarely , to make an error of as much as 8 centims. in the reconstructing of the stature of an individual .
We have placed in the above tables M. MANOUVRIER'S results as calculated from his 'Table-bareme .
' They give somewhat larger mean errors than our formula , which would have been probably reduced somewhat if we had excluded , as he has done , the aged .
We have seen , however ( p. 179 ) , that there seems no reason to exclude the aged women , and in the case of the seven men over 60 , he actually in three cases under-estimates their stature .
In other words , while in fonr cases his table might have better results for adult stature , in three it would have given worse results , If we allow a mean old-age shrinkage of 3 centims. \mdash ; an amount hardly justified by the adult and portions of ROLLET'S returns\mdash ; we should find that MANOUVRIER'S method would have made a total error of 17 centims. in estimating the stature of these seven old men in youth , whereas it gives a total error of 16 centims. in estimating their .
statur .
Thus there might , perhaps , be a small , bu } , it would not be a very sensible , reduction of the mean errors of the results given by MANOUyRIER'S ' Table-bareme ' had we excluded the old age cases .
What deserves special notice is that our formula ( k ) gives a better result than the mean of all the formulae , and a better result than the mean of the values obtained by method for the foor .
bones .
The next stage in our work to so modify Tables IX .
and X. that they will serve for the reconstruction of the living stature from bones of which all the ppeared , which are dry free of all rtilage .
This *This value is that given by M. fANOUVRIER himself , 'Memoires de la Societe d'Anthropologie do Paris , ' vol. 4 , p. 356 , 1892 .
CONTRIBUTIONS TO THE THEORY OF F.VOLUTION .
is either the condition in which we find the bones of a prehistoric or early race , or it is one to which they are soon reduced on being preserved in museum or laboratory .
The first question which arises is the differerlce between the mecm stature of the living and the mean stature of the corpse for both sexes .
It is impossible to measure this difference satisfactorily on a sufficiently large number of individuals , and then take the mean difference .
If we suppose ROI , individuals to be an average sample of the French race , then we must place in Tables V. and VI .
for on the left the mean heights of French men and French women .
Now there is a considerable amount of evidence to show that the mean height of is 165 centims. almost exactly .
The service of M. BERT1LLON gives tims .
, and this is the furnished by the surements for military recruiting .
* M. lIANOUVRIER takes 165 centims. as the mean height , and as by selecting only twenty of ROLLET'S he gets a mean height of about 167 .
for corpse , he concludes that 2 centims. must be deducted from the corpse length to get the living stature .
In our case all we have to do is then to put centims. the same time , BERTI 1 numbers probably include many men over 50 , and recruiting service many men ) yet fully grown ; hence it seems to me doubtful whether 2 centims. really represents the difference between .
and dead stature .
165 centims. is probably a good mean height for the whole adult population , and should be compared with ROLLET'S whole adult population , which has a mean of centims. I accordingly conclude that centims. is on the a more reasonable deduction to make in order to pass from the dead to the living stature of the general population .
In the course of my investigations , however , no use is made of this diHerence , but given its observed living value .
The value for women is far less easy to obtain , as a good series of French statistics entirely fails .
The mean given in the footnote belo is clearly only that of a special class .
MANOIJVRIER has found from 130 women , between 20 and 40 years of age , inscribed in BERTILLON'S registers the mean height , centims. , and RAHON that this is the best result yet obtained .
But the mean height of ROLLET'S material is centims. ( see my p. 180 ) , and , as we have seen , this is not sensibly increased by taking only the women in the prime oflife ( see p. 179 , above ) .
If 154 centims. were the mean living stature of ROLLET'S women , we should have to suppose a of stature in women when the corpse is measured , whereas in the case of men the corpse length is greater than the living stature .
RAHON , disregarding his own statement as to centims. being the best value , follows MANOUVRIER in deducting 2 centims. from the stature as corpse to get the living stature .
MANOUVRIER'S * ' Memoircs de la Societe d'Anthropologie de Paris , ' vol. 4 , p. 413 , 1893 .
For special classes the stature is considerably greater .
See the values centims. for male and centims. for female given in the ' Mem. Soc. d'Anthrop .
, ' vol. 3 , 1888 . .
cit. , p. 413 .
rule for deducting 2 centims. seems based partly on a comparison of BERTILLON'S measurements for men , with his own selection from ROLLET'S material , which give mean centims. and 167 centims. respectively , and partly on the measurement and reclining of six men and four women .
* Now the reader should notice that in our method of reaching the reconstruction equations , we are not concerned with the amount to be subtracted from an individual stature , but with the mean living stature of the population which ROLLET has sampled .
Now there is a quantity which has very remarkable tancy , namely , the sexual ratio for stature .
The mean male is to the mean female stature in a great variety of races and classes as 13 to If , therefore , ROLLET'S wolnen are the same class as his men , we should expect their living stature to have had a mean , that of the men centims. We have seen that from the registers of BERTILLON the mean stature of women between 20 and 40 was centims. ; these probably include a considerable number of stout tramps or , not a fair sample of those who would find their way into the Lyons Hospital .
TENON measured in 1783 60 women of the village of Mussey , and obtained a mean stature of centims. If we take the of these groups we find centims. as the mean stature for French women of the lower classes ; this differs less than 3 millims. from the result already suggested by using the sex ratio .
I am , accordingly , inclined to hold that the best that can be done at present is to take centims. as the mean stature of Frenchwomcn of the class sampled by ROLLpT .
The next stage in our work is to consider the difference in length of the long bones , as measured in the dissecting room by ROLLET and his assistants , and as they would be measured in the case of a primitive race whose bones had been exhumed , and then been preserved and dried before measuring .
ROLLET merely observes that he kept several of his bones for some months , and , the cartilage being then dry , they measured on the average 2 millims. less .
S On the strength of this , MANOUVRIER and he is followed by RAIION , add 2 millims. to the length of each prehistoric bone when reconstructing the stature .
Now I am whether this gives a really close enough result .
ROLLET measured the bones in the dissecting room , the cartilages were still on , and the animal matter in the bones , but in the case of prehistoric ancient bones this does not at all represent the state of affairs .
Nor are they merely such bones with the cartilage dry ; the cartilage , together with the animal matter , has entirely gone .
There are accordingly two allowances to be made for the cartilage .
and ( b ) for the disappearance of the animal matter and drying of the bone .
* ' Memoires de la Societe d'Anthropologie de Paris , ' vol. 4 , p. 384 , 1892 .
BOLLET'S corpse statures give a sexual ratio " " Notes manuscrites elatives a la stature de l'homme , recueillies par VILLEBNH 'Annales d'Hygiene , ' 1833 .
S ROLLET , , p. 24 .
MANOUVRIER , , p. 386 .
CONTRIBUTIONS TO THE THEORY OF UTION .
19.3 Allowance for the Cartilage .
* Tlle thicknesses of the cartilages here cited are taken from WERNER'S Ina , .
Dissertation , 'Die Dicke der menschlichen Gelenkknorpel , ' Berlin , 1897 .
They are only discussed for the cases required for the bones as measured by ROLLET and used in my reconstruction formulae .
Fernur .
Maximum length straight from top of head to bottom of internal condyle ( F ) .
( ii .
) " " Oblique\ldquo ; length from top of head to plane in contact with both condyles ( F ) .
For both we have for articular cartilage at upper end 2 millims. , at millims. , or the total together of millims. This is more than double MANO allowance .
Humerus.\mdash ; Length from top to lowest point of internal margin of trochlea ( H ) .
At upper end we must allow millims. , and at lower millims. , altogether millims. for articular cartilage .
Tibia .
\mdash ; The spine is excluded by OLLET .
The length is from plane of upper surfaces ( margins ) to tip of internal malleolus ( T ) .
In this case the icular cartilage has only to be allowed for at the upper end , and is here 3 millims. Radius.\mdash ; The length is measured from top of to of styloid process ( R ) .
The allowance must be for articular cartilage at upper end only , and is millims. Allowance Animal in Bones .
Here unfortunately I had not the same amount of data to guide me .
The best hypothesis to go upon seemed to be that a thoroughly dry bone , free from all animal matter , would , if it were thoroughly soaked , approximate to the condition of the bones measured by ROLLET .
BROCA , who has written a very elaborate mslnoir on the effect of humidity in altering the capacity and ions of skulls , has referred incidentally to the extension of the femur by humidity .
He took three femurs , macerated in 1873 , one of the 15th century , and one of the polished stone age .
After soaking for seven days , he found an increase of millims. in the first , millims. in the second , and 1 millim. in the third .
These results , he says , compare very well with WELCKER'S , S who gives millims. for increase of length of femur with humidity .
It was somewhat difficult to make fresh expel.iments on a considerable number of * The details of this section I owe entirely to my , Professor THANE , who in this matter , as in many others , has given me most and generous assistance .
On another occasion may take into consideration the ulna and fibula , but they have nothing liko the importance for staturs of the bones here dealt with .
'Memoires d'Anthropologie de PAUL BROCA , ' vol. 4 , pp. 163 et .
seq. ; p. 195 .
S 'Ueber Wachstum Bau des menschlichen Schadels , ' p. 30 , lS62 .
WELCKER only dealt with one male femur , and soaked it for three days .
VOL. CXCIL \mdash ; A. 2 long bones of each kind , but it seemed wort , while to measure dry and thoroughly humid a bone of each type .
A bone of each type placed at my by Professor , and they were measured independently on each occasion by Mr. .
In the one or two instances in which we did not , the bone was again independently measured .
Our results were as follows:\mdash ; TABLE XI .
\mdash ; Lengths of Long Bones , Dry and Wet , in Centimetre FTBH Dry water .
} water .
The bones themselves were between 200 and 300 years old.* They were only allowed to stand two hours for the water to run ) before they were measured after soaking .
In the case 01 the final 72 hours ' , it concluded with six hours in the neighbourhood of a stove .
The column may be considered to represent the humidity of bones preserved in a museum ; the last column complete dryness .
It seems to me that the difference bebween the first and third column is what we in general have to deal with .
In this case we have a diflerence of F. T. H. R. millims. millims. millims. 7 millim. between dry and bumid bones .
The rence between this result for the femur BROCA'S is very considerable .
I think it is due to the fact that he allowed his bones to dry for 24 hours in a room before measuring .
I was much impressed by the rapidity with which bhe bones dried , and their , of course , are very unlike what they would be if containing or by animal matter .
It is clear that the extensions due to humidity are not by any means proportional to length of the bone , and it would be quite futile to attempt any percentage allowance for the extension due to this cause , the effect of which clearly differs with the diiterent structure of different parts of the same bone .
I have accordingly thought it best to subtract the above quantities from ROLLET'S means , , and , and to consider the ] so derived as giving the means of ROLLET'S material on the supposition that the bones were dry and free from animal matter .
Even so I do not tbink we shall err in over-estimating the difference between the lengths of living and dead bone .
Making allowances and ( b ) we have finally to subtract from P , OLLET'S results for * See additional note , p. 244 , CONTRIBUTIONS TO THE THEORY OF EVOLUTION .
millims. millims. millims. millims. , respectively .
Making these subtractions ( which are sensibly different from allowance of 2 millims. for each bone ) , we are in a position to find the reconstruction formulae connecting .
stature with dry bone entirely free of animal matter .
We have for the French population , if notes living mean stature , and the mean lengths of the corresponding dry bones in centimetres : TABLE XII .
If we want the mean oblique length of the femur , we must follow the rule given on p. 184 , and we find for male and for female .
M. RAHON has measured the lengths of a large collection of bones in the Faculty of Medicine of Paris and he finds:\mdash ; Femur , oblique length , 62 males , mean .
38 females , , , .
, , Humerus , maximum length , 44 males , , , .
39 females , , , 292 .
\ldquo ; My results are placed in brackets , and it is clear that for these bones the allowances for and animal matter have been very satisfactory ; there ) certainly been no over-correction , although in the case of the femur is more than thrice , and in that of the .
humerus than twice M. M. RAHON does not give the measurement of ) radins , but he does of the tibia , and in this se there is undoubbedly some source of in his result , in -he collection .
He gives : for 53 males , mean ; for 26 female ROLLET'S material for 50 of either sex giyes , male mean , and female without allowance for the lage or presence of animal matter .
A Ilowing for these , RAHON'S measurements are , male , centims and female , centi .
too These are errors much beyond those of the determinations , which have errors of about to 18 centim. , since he is using MANOUVRTER'S method must , be supposed to be the tibia in the same manner as ) LLET , i.e. , with the malleolus and without the spine .
But even supposing he had included the spine , , .
413 .
it could not make this great difference .
* That there is some substantial error is evidenced by the fact that tibias of these dimensions would give a reconstructed stature for French males of centims. instead of 165 centims. , and for French females of centims. instead of centims. RAHON himself , on the basis of MANOUVRIER'S tnethod , forms the estimates of centims. and centims. respectively , \mdash ; the latter , at any rate , a quite impossible for the French female population .
( 10 .
) We are now in a } ) osition to write down the reconstruction formulae for living stature from dry long bones ; they are the following:\mdash ; TABLE Stature from Deadt Long Bones . .
( ) .
( ) .
( ) .
( ) .
( ) .
( ) .
( ) .
( ) .
( ) .
TABLE XV.\mdash ; Female .
Living Stature from DeadLong Bones .
( ) .
( ) .
( ) .
( ) ( ) .
( ) .
( ) .
( ) .
( ) .
( ) .
Remarks .
If the femur has been measured in the oblique position and not * Dr. WAltREN found for the New Race from Egypt the mean length of spine for 85 males centim. , and for 115 femalss centim. These numbers should be introduced as an addition to in Tables .
and VI .
, when the tibia has been measured ding spine .
The word\ldquo ; dead\ldquo ; is here used to denote a bone from which all the animal tter has and which is in a dry state .
CONTRIBUTIONS TO THE THEORY OF EVOLUTION .
straight , add centim. for male and 33 centim. for female to the length before using the above formulae .
If the tibia has been measured with , and not without , the spine , subtract 96 centim. for male and centim. for female from the length before using the above formulae .
( iii .
) The above formulae have been determined from the right members ; a small error , of the second order as a rule , arises when the left is used .
The following numbers are determined from ROLLET'S measurements ; they give the amount to be added to a left bone when it is used in the formulae:\mdash ; The femur change is insignificant .
In most statements oflengths the rightness or leftness of the bone is not given , and hence , no correction can generally be made an individual .
The error will , however , be hardly sensible except in the case of the humerus and radius .
If a considerable number of bones have been probably half may be looked upon as right and haIf left , and in this case half the above corrections may be added to the average .
In any case , it is probably only the estimate based on the hulnerus and radius which need to be corrected in this manner .
Even here it is a problem how far there is a racial character in this right and leftsideness .
Results due to CALLENDER , , GARSON , HARTING , and RAYMONDAUD are cited by ROLLET ( loc. cit. , pp. 53-60 ) , but being based either on very few cases , on measurements on the living , or on unsexed material , they are not of much service for our present purpose .
Results of much greater value for racial comparison have been given by Dr. WARREN for the Naqada race ( ' Phil. Trans , vol. 189 , p. 135 et seq He tinds:\mdash ; Dr. results are for the oblique femur , and from centre to centre of the articulate surfaces in the case of tibia and radius .
Thus they are not directly comparable with the results for the French .
On the whole , if the bone is stated to be left , we may add for the humerus ] for the radius , leaving the femur and tibia unaltered .
These additions are approximately the same fur both sexes .
Several results may be noted with regard to this table : In the first place let us compare our results with those which would be given by M. MANOUVRIER'S Tableau II .
* Corresponding to our cases , ( b ) , , ( d ) he would obtain:\mdash ; Here ( f ) , ( h ) , ( i ) , and ( k ) are obtained by taking means of the results for the single bones .
Comparing the first four formulae with my first four , M. MANOUVRIER has for male a mean error of centims. against my centims. , and for the last four a mean error ) .
as against centim. His error in as deduced from the male humerus , is greater than my error from the radius even .
In the male measurements M. MANOUVRIER has a mean error of centims. against my centims. in the first four results , and one of centims. against my 42 centim , in the last four results .
But these results by no means represent the full advantage of the present theory .
An examination of the shows us the formulae give good , i.e. , consistent results except in the case of the radius .
Here it is that the greatest differentiation has taken place , very possibly owing to the direct selection of other long bones .
Our general principles ( p. 177 ) accordingly suggest that we should omit the results for this bone from our consideration .
The best formulae then to use will be , ( f ) , and ( i ) ; we shall then have a mean error centim. for ] and 06 centim. for female\mdash ; a better approximation to the true stature could not possibly be reached .
M. MANOUVRII ] by the process of means , would have deduced from the same three bones a male stature with an error of centims. and a female stature with one of 59 centim. Dr. BEDDOE'S rule would give fcr male Aino centims. , and female Aino centims. , or errors of centims. and 5 centim. ; in this case not as great as those of M. MANOUVRIER , but still seusibly greater than our , ( f ) , or ( i ) .
The accordance obtained between the formulae for reconstruction which I have given , , tables at end of Memoir .
'Journal of the Anthropological Institute , ' , 1887 , p. 205 .
CONTRIBUTIONS TO THEORY OF EVOLUTION .
and the actually observed stature in the case of such a diverse race as the Aino ought , I think , to give considerable confidence in their use .
( ii .
) I have also included in the table the results for the male Aino , calculated from female formula , and for the female Aino , calculated from the tYlale formulae .
The reader will perceive at once that sexual differences are immensely greater than racial differences\mdash ; that it would be perfectly idle to attempt to reconstruct female stature from male formulae , or vice versed .
Exactly the same order of divergences are obtained if we endeavour to reconstruct French female from male formulae , or vice versed , and we concluded that French men and French women are more differentiated from each other than French of either sex and Aino of the same sex , at any rate , in the relations between stature and the long bones .
It is noteworthy that the only instance in which the formula for one sex gives even approximately the stature of the other , is in the case of the female formula applied to the male stature by means of the length of the radius .
In this case we get a better ] than from the male formula itself Now this is peculiarly significant , for it is in the radius that the most marked differentiation between French and Aino has taken place ; and in this respect the Aino male approaches nearer to the French female than to the French male .
We must thelefore conclude that while the sexes are widely differentiated from a common stock , still in respect of radius the females of a highly civilised race like the French , and the males of a primitive race like the Aino , are even closer together than the males or the females of these two races for this special bone .
The agreement between the same in two different races , however , is generally far closer than between different sexes in highly civilised and primitive races .
( 12 .
) Having taken an reme case of divergence in man and tested the confidence that may be put in our reconstruction , it will not be without interest to see the amount of divergence in the formulae when we apply them to allied species .
Stature is , of course , a very difficult character to deal with when we are considering the anthropomorphous apes , and it would be idle to think of going beyond a round number of centimetres .
But even here the agreements and disagreements are so remarkable that they appear to furnish material on which certain quantitative statements with regard to the general lines of evolution can be based , and further they suggest that the regression formulae .
the long bones among open up quite a new method of attacking the problem of the descent of man .
Like the rest of the material in this paper , the considerations of the present paragraph must be looked upon as suggestions for new methods of research .
I have taken what material was at hand and not endeavoured to form comprehensive statistics .
The methods are strated on stature , but they are equally applicable to the regression formulae connecting any characters or organs whatever .
* I hope later to deal at ]ength with the regression formulae the long bones of man and apply them to the anthropomorphous apes , placing stature entirely on one side as a quantity difficult to measul.e .
VOL. CXCIL \mdash ; A. CONTIIIBUTIONS TO THEORY OF EVOLUTION .
Now we see that , if the gorilla be put on one side , there is no approach to accordance between the calculated and observed statures in the case either of chimpanze or crang for any of the ten formulae .
conclude therefore , that if man and the chimpanze and orang have been derived from a common stock , they must have been directly selected with regard to stature and with regard to the lengths of the four chief long bones .
In the case of the gorilla we notice , however , a remarkable accordance between the observed stature , and that calculated from the male reconstruction formulae in the case of man , when we use ] formulae ving the femur and tibia .
It would thus appear that if man and the gorill have been differentiated from a common stock , they have been directly selected in the same manner so far as femur and tibia are concerned , but in directions when we consider humerus and radius\mdash ; we here referring only to the lengths of these bones .
-examining the results for the male formulae from the standpoint of correspondence in the femur and tibia between the gorilla and man , we see that the chimpanze co1nes nearer to } than the orang ; the lengt.hs of the femur and tibia have been modified in the former , but not to such a marked degree as the case of the latter .
Turning to the female reconstruction formulae we notice in to ( k ) for the chimpanze and orang an accordance between the observed and calculated statures which is some 3 centims. to 6 centims. better , although still very poor .
The reason for this is obvious , the stature of the woman for the same length of long bone is 3 centims. to 6 centims. shorter than that of man , and accordingly the female formulae must give slightly better results than the male formulae when applied to the anthropomorphous apes , have for the same length of bone a markedly shorter stature than man .
In the gorilla we have , over-corrected the stature so as femur and tibia are concerned by using the female formulae .
One point , however , is of very great interest : while the female formulae for bumerus , radius , or for humerus and radius give very bad results , eveu worse for the gorilla than they do for the chimpanze , yet the female formula for femur and humerus gives a sensibly and that for ] the long bones a markedly better result for the stature than the corresponding male formulae .
The difference here is not the 3 centims. to 6 centims. due to sex .
The improvement in the result when we apply the ] for all four long bones to the estimate of the stature of the gorilla is noticeable also , if to a lesser degree , in the cases of the chimpanze and orang .
We may sum up our results as follows:\mdash ; Man is apparently differentiated from the chimpanze oraog by direct selection of stature , but this direct selection pears to be small in the case of the gorilla .
If the chimpanze and orang be tl.eated as " " dwarf men heir estinlated in the } indicated on p. 224 below , the femur and tibia give statures , respectively , nearer the actual values , in fact too , but the radius and humerns give values far too great .
The stature of the gorilla as estimated from femur and tibia , in this manner now becomes far too small , Man and the gorilla appear to have followed common lines of differentiation from a common stock in the case of the femur and tibia , but the differentiation on which they have not followed common lines has not been that of radius and humerus alone , or ( k ) would have given good results .
Other organs closely correlated with stature beside t , be four long bones must have been differentially modified in the case of the chimpanze and oranlg , or ( k ) would still have given good results .
The accordance between the result given by female ( k ) and the observed stature of the gorilla , and the want of accordance in all other formulae , seems to show that woman has been principally differentiated by these four long bones from the common stock , while man has been differentiated in other organs highly correlated with stature .
For example , the differentiation in pelvis may be much greater .
So as I am able to draw a conclusion from the few data at my command , the correlation of radius and humerus with appears to be negative for the chimpanze and orang , while it is positive for the gorilla and man .
The negative character of the partial correlatio1l coefficient for the radius in ( k ) seems to be a relic of this stage of evolution , and it is much more marked in woman than in man .
The above statements must not be taken as dogmatic conclusions ; they are only suggestions of the manner in which the regression formulae can possibly be applied to the problems of evolution .
They are no more weighty than the very slender material* on which they are based .
But they may suffice to indicate how a method of quantitative inquiry be applied to ascertain more about the relationship of man to the anthropomorphous apes , so soon as a sufficient amount of data concerning the dimensions of the organs of adult apes has been collected , and reduced to numerical expression .
* In order to velify Sir G. HUMPHRY'S measurements , I have gone through the catalogues , so far as published , of the German anthropological collections , and extrncted the ements of all adult anthropomorphous apes .
Unforbunately I could only find oue adult chimpanze ; the sex was as often as not not given .
I find : GorillaOrang . .
A better agreement with the results cited , p. 202 , could nob have been expected , or wanted .
Thus our data give racial and not random characters .
CONTRIBUTIONS TO THE THEORY OF EVOLUTION .
( 13 .
) Paloeolithic Man .
I am indebted to the memoir of M. RAHON* for the details of all the individuals that are classed under this heading .
I presume that in asurin the tibia he has not included the spine , as his formulae are , ] mine , based on its exclusion .
I have further allowed for the fact that he used the oblique length of femur , while I require , maximum length .
Unfortunately we have only five cases to base our estimate upon .
nderthal Man .
We find for stature from : Mean RAHON gives ] centims. ( but I think he ought to have given centims. , as femur estimate is rect ) centims. , so that our diverges by ) centims. to 3 centims. Man from Spy .
We find for stature from : RAHON gives centims. Man from Clay at Mean 16033 The length of the femur here is doubtful , but it is said to have been between centims. and centims. If we take the mean value , the probable stature was 16680- ntims .
, and the maximum va]ue would only be centims. RAHON gives 170 centims. , using ulna as well as femur .
I have not worked out the statureulna correlation , but , if this bone is at all akin to the radius , it will give very exaggerated results for pl.imitive man .
Man of .
Here again the radius gives clearly an exaggerated The mean is * ' moires de la Societe d'Anthropologie de Paris , ' 1893 , p. 414 et seq. " " Dcr Neanderthaler \ldquo ; ' Deutscbe Anbhropologische Gesellschaft , ' 1888 .
centims. , but , neglecting , I am inclined to take the best value as centims. RAHON gives centims. MANOUVRIER ( loc. cit. , p. 391 ) is inclined from the general character of the bones to consider the stature as determined from the ulna and radius to be the better estimate , and even thinks this troglodyte may have been 165 centims. Judging , however , from other primitive races , I should expect the arm bone estimat .
to gerate the stature , and prefer my estimate of centims. of .\mdash ; All we know here is the length of the femur centims. The probable stature is accordingly centims. TOPINARD gives it as centims. , and at centims. Taking the mean of the best values for the above five cases we have : Probable stature of palaeolithic man centims. All the above cases are supposed to be males .
Considering that it is more probably the massive bones which have survived , we must hold that palaeolithic man was shorter than the modern French population , but was taller than the men of Southern Italy ( 156 centims. to 158 centims and about the mean height of the moder1l Italian male population , i.e. , centims. ( 14 .
) Neolithic Man .
Britain .
We have not very much data to build upon here .
Dr. BEDDOE* gives the length of twenty-five male and five female femora .
Converted into centimetres , we have , centimS.ence pable stature Female ( 5 ) , Dr. BEDDOE'S estimates , male centims. , and female centims. , are , I , much too high .
The sex-ratio is France and Belgium .
The following data have been drawn from RAHON ( loc. cit. , pp. 418 et seq the numbers in brackets in the left-hand corners denoting the numbers upon which the average lengths of the bones are based .
* ' Journal of the Anthropological Institute , ' vol. 17 , 1887 , p. 209 .
CONTRIBUTIONS TO THE THEORY OF EVOLUTION .
We find : STATURE of Neolithic Man .
Sexual ratio So far , then , as we have matel.ial to judge by , there appears to be no sensible difference between Continental palaeolithic and neolithic man ; they corresponded very closely to the modern Italian in stature .
On the other hand , if we compare British with Continental neolithic man , we find , judging even from femora only , a very ) sible difference in stature .
Neolithic man in Britain was taller probably than the modern Frenchman , and markedly taller than neolithic man in France .
This leads us to consider one or two special classes of neolithic bones , for it must be remembered that probably as many neolithic races existed in Europe as we find races existing in historic times .
In the first place , we have the big bones of the Cro-Magnon man , centims. , centims. These give for the stature : Mean .
which is a centimetre greater than RAHO estimate , seven less than ROLLET'S , and seventeen less than TOPINARD'S .
This man was undoubtedly tall , but cannot be taken as a type of his race .
The second skeleton gives us centims. from which we find from : Mean .
This is also taller than the average neolithic man , but much below the other skeleton .
*As carefully determined by RAHON .
cit. , p. CONTRIBUTI0NS TO THE THEORY OF VOLUTION .
Brueil .
Mureaux both less than the result we have obtained for the general erages of ] ithic man .
Probably we have here to do with local yaces , but ] } .
MANOUVRIER considers itjust possible that the very different ironmont at Brueil and Mureaux may account for diflerences .
Neither of these groups has a stature equal to that of the modern French commonalty , although the Mureaux group approaches it somewhat closely .
The modern British far exceed in stature their neolithic landsmen , and we have thus no evidence at all in fa of a giant stature for prehistoric man .
He seems to have been markedly shorter than the taller races of to-day .
Slightly taller than the Aino , he can be compared with the Italians , who appear , as we go southward , to closely represent him in stature .
( 15 .
) Other Early Races .
In this group I propose to include a number of prehistoric or protohistoric races of whom we know very little .
Their stature is considerably greater than that which we have determined for Continental neolithic man , though sensibly below thac of British neolithic man .
The data are extracted from RAHON'S memoir , and modified to suit the formlae of this investigation .
While the dolmens of India and AIgeria appear to belong to the Stone Age , those of the Caucasus belong the first Iron Age .
VOL CXCII.\mdash ; A. 2 The series from these dolmens is very small .
On the other hand the Guanch series are both very complete .
The first are drawn from the Musee Broca , and the second from the Museum d'Histoire Naturelle RAHON , , both at Paris .
Although the first series comes from a single locality , and the second from several localities , the results are in good agreement .
The following statures have been found our formulae:\mdash ; The first point to be noticed about this table is the confidence it inspires in formula ( k ) .
Whenever the series is in the least extended , formula ( k ) gives a result sensibly identical with the mean of all ten formulae .
M. RAHON'S means for the eight groups are not very divergent from mine , he gives : \mdash ; ; and He thus does not make quite such a sensible distinction between the Guanches and the Dolmen-builders as my numbers seem to indicate .
It is curious that these three groups of Dolmen-builders should stand so close together , and also comparatively close to the Guanches .
The Dolmen-builders must have been as tall as the modern French , while the Guanches were probabIy slightly taller .
Both were of greater stature than neolithic man in France , approaching more nearly the neolibhic man of Britain .
The sexual ratio in the first and third cases cannot be considered any weight , as the female daia contain only single individuals .
While .
MANOUVRIER'S male mean does not differ widely from ours , female mean is 9 centim. greater .
His range for male stature covers centims. , and for female stature centims. , as compared with our and centims. respectively .
But the amount of this range in both cases is very significant considering the large number of bones averaged .
While our formulae applied to the Aino gave very selfaccordant results except in the case of the radius , we notice here considerable divergences .
In particular , the order of the bones arranged in order ofincreasing stature , which is nearly the same in both sexes , is very different for the corresponding order for the Aino .
The Naqada people for their stature have a remarkably small humerus , and although the Aino could hardly be separated more from the French by civilisation and locality , yet they could be derived from a common stock with the French by far less direct selection of the long bones , than would be possible in the case of the French and the Naqada races .
This Egyptian race was a tall race\mdash ; not as tall as the English commonalty\mdash ; but taller than the better French classes and centims. taller than the mean of the French army .
The sexual ratio , , was less than that of the modern European ( about ) , and this is in keeping with the greater equality in size observable in primitive and early races .
On the whole it may be ) whether any two modern races would give such divergence in character as the Naqada and French .
We see not only the radius , as in the case of the Aino , but the humerus as a source of divergence , and so far as the lengths of those long bones are concerned , it would be easier to look upon the Ainos and French than upon the Naqada people and French as local races deduced from a common stock .
If they have sprung ultimately from such a stock , there has been a very significant amount of direct selection .
There is , however , an interesting point which the Naqada people share with the Ainos the judgment of stature from the radius is excessive .
This peculiarity of early and primitive races is one which the table on p. 202 shows that they share , of course in a much less marked manner , with the anthropomorphous apes .
It will later be seen to be a feature of other primitive and early peoples .
( 17 .
) Protohistoric Races .
My next group covers to some extent the ground which precedes 1000 roughly , the beginning of the MiddIe Ages .
Dr. BEDDOE gives femur measurements for the Round Barrow population of Britain , as follows : Male centims. , mean for 27 , Female , , , , 2 .
We find at once from : Stature Male centims. , Female centims. Sexual ratio * ' Journal of the opological InsGitute , ' vol. 17 , 1887 , p. 209 .
CONTRIBUTIONS TO THE THEORY OF EVOLUTION .
21.3 These values are immense reductions on Dr. for males and for females .
Even with this reduction , the Round Barrow population must still be considered a tall one , as tall as the modern Englisl ] .
It will be remembered that it was also brachycsphalic , a curious and infrequent combination in Europe .
We may next consider the Britisl ) , for whom we obtain from Dr. BEDDOE the data : Male centims. , mean for 10 , Female , , , , 4 .
Formula gives : Stat , ur Male centims. , Female centims. Sexual ratio Here again we have very sensible reductions on Dr. BEDDOE'S estimates of and We may compare these resul } for the Romano-British with those for the -Gauls , based on data } ) rovided by These give : R. . .
Whence we deduce : STATURE of -Gauls .
Male .
*PEARSON , ' The Chances of Deatl vol. 1 , " " Variation in Man and Woman p. 363 .
, .
441 .
The second mean estimate for females is determined by neglecting the single tibia measurement , and is probably the best obtainable ; it agrees closely with : Sexual Ratio A series of 12 femora dug up in Boulogne Harbour*have also been attributed to Romano-Gauls .
They give male centims. , or for the stature My estimate here is about a centimetre larger than RAHON'S .
We sensibly agree for the males in the larger series above , while for the females I should take the most probable stature to be a cenbimetre less than that given by RAHON .
We cannot compare the Romano-British with the Romano-Gauls on the basis of all bones , for we have only the results for the femur in the former case .
But if we compare the femur estimates for the two cases we see that they are sensibly the same male ainst 1 female e against .
It is , therefore , probable that the estimate of the Romano-British male is sensibly too high , and that it would have been nearer 165 centims. had we had other bones than femora to base our estimates upon .
The sexual ratio is clearly abnormal]y high .
Row-Grave Population of South Germany .
Dr. R. LEHMANN NTTSCHE has ] ished a most interesting series of measurements on the long bones found in the Row-graves of Bavaria.hese interments date from the beginning of the 5th to the end of the 7th century .
He divides his material into two groups , " " Bajuvars from the Row-graves of Allach in Upper Bavaria , and Suabians and Alemanns from those of Dillingen , Gundelfingen , Schwetzheim , Memmingen and Fischen .
S The mean lengths of the long bones for these two groups are , however , in such complete accordance , that we are quite justified in following Dr. NITSCHE and combining the two groups We have then the following results after the ) roper change in the femur:\mdash ; The following table gives the reconstructed stature on the basis of the ten formulae of Tables XIV .
and .
:\mdash ; cit. , p. 439 .
" " Never Beitrage znr physischen Anthropologie der Bayern vol. 11 , pp. 205-296 , Munchen , 1895 .
Ibid. , p. 207 .
S Ibid. , p. 239 .
pp. 260 , et se CONTRIBUTIONS TO I1'HE OF VOLUTION .
For modern English we have the following resultS : Male Female \mdash ; - Male Female \mdash ; - Male Female \mdash ; - Male Female \mdash ; - Male Female \mdash ; - Male Female \mdash ; - Male Female \mdash ; - Male Female \mdash ; - Male Female \mdash ; - Male Female \mdash ; - Male Female \mdash ; - Male Female \mdash ; - Male Female \mdash ; - Male Female \mdash ; - Male Female \mdash ; - Male Female \mdash ; - Male Female \mdash ; - Male Female \mdash ; - Male Female \mdash ; - Male Female \mdash ; - Male Female \mdash ; - Male Female \mdash ; - Male Female \mdash ; - Male Female \mdash ; - Male Female \mdash ; - Male Female \mdash ; - Male Female \mdash ; - Comnlonalty .
I. Comnlonalty .
I. Comnlonalty .
I. Comnlonalty .
I. Comnlonalty .
I. Comnlonalty .
I. Comnlonalty .
I. Comnlonalty .
I. Comnlonalty .
I. Comnlonalty .
I. Comnlonalty .
I. Comnlonalty .
I. Comnlonalty .
I. Comnlonalty .
I. Comnlonalty .
I. Comnlonalty .
I. Comnlonalty .
I. Comnlonalty .
I. Male Female \mdash ; - Male Female \mdash ; - Male Female \mdash ; - Male Female \mdash ; - Male Female \mdash ; - Male Female \mdash ; - Male Female \mdash ; - Male Female \mdash ; - Male Female \mdash ; - Male Female \mdash ; - Male Female \mdash ; - Male Female \mdash ; - Male Female \mdash ; - Male Female \mdash ; - Male Female \mdash ; - Male Female \mdash ; - Male Female \mdash ; - Male Female \mdash ; - Male Female \mdash ; - Male Female \mdash ; - Male Female \mdash ; - PEARSON , Middlc classes .
II .
III .
( 1077 ) Mr. GALTON'S results were measured at his South gton LaboratoIy during Exhibition of 1884 .
My first group are from my mily data cards , without boots ; my second group are from the measurement of the ambridge Antbropometrical Committee .
ubtracting centims. for boots I. and II .
, we find : Uale . . .
Female . . .
Thus there is a sensible agreement between the results .
and III .
, while I. shows just the class distinction we might expect to find .
Comparing these results with the , -Saxon we notice an increase of about 2 .
in the female stature , while the present commonalty is about 1 centim. less than the mean male stature , and the male middle classes about 2 centims. more .
lf the Wittenham skeletons with tibia belong to a class apart , then they quite equal in stature to the modern English classes , while the Anglo-Saxon bishops were distinctly inferior .
Probably the bishops were men unsuited for fighting , and showing a owel .
degree of physical development .
The Anglo-Saxon women are not very many in number , and we have only the femora to base an estimate upon , which in sll these cases gives a less stature than the ti .
We may therefore conclude that the average Englishman of to-day is certainly not behind his Anglo-Saxon cestors ; he be very slightly taller .
The average oman is probably somewhat taller , but the paucity of data for Anglo-Saxon women hardly allows an nate of how much .
The sexual ratio , , is so high that I am co mpelled to consider the Anglo- Saxon women or possibly mixed with a Romano-British element .
The modern value is about ( 19 .
) Franks .
I have put into one group the Frankish remains belonging to both the and Carolingian periods , to be found in RAHON'S memoir , separate smaller groups giving results in close accordance .
We have then : , p. 440 , et seq. VOL. CXCII.\mdash ; A. 2 PROFESSOR KARL PEARSON , rICAL a slip in using MANOUVRIER'S table , and thus much underestimated the [ nrcel male stature .
I think it impossible to accept RAHON'S view that the modern French are sensibly of the same stature as the mediaeval French , because the slight apparent difference may be accounted for by a process of selection preserving for us only the larger bones .
It is not , as RAHON supposed , a difference ceutim .
which has to be accounted for , but one of nearly 3 centims. We have the following series for France , male and female:\mdash ; Neolithic man . . . . .
Romano-Gauls . . . . .
Franks . . . . .
French , 4th to 7th century 10th to 11th century modern . . . . .
These results would seem to indicate that the Gauls were taller than the races they superseded in tnce , that their Frankish conquerors were taller again than ; but that the stature has been sinking the last 800 years , and that the French commonalty of to-day is very close in stature to the Romano-Gauls .
This may denote a selection of stature , or it may mean that the Celtic element of the , population has superseded the Teutonic element\mdash ; an explanation in accordance with the recognised greater fertility of the Breton element in France .
We should then have an interesting stration of the manner in which reproductive selection may reverse the results of natural selection .
While it might be rash to attribute the decrease in stature which has taken place in France to any one definite cause , it is interesting to note that we do not trace the like decrease in stature in England , yet we should certainly expect to do so , if the result were due simply to a selective process by which the larger bones were preserved .
There does appear to be a like decrease in the stature of the Bavarian population , where we have compared ( p. 215 ) the Row-grave populaticn with that of Munich town recruits , which appears to be considerably above the average of recl.uits from other near districts , considerably above the corpse length centims.-itself greater than the stature of the living\mdash ; which I have found from data .
On Giants and If pass from the consideration of races with mean statures varying from about 157 centims. to 170 centims. to the consideration of individual giants and dwarfs , we very soon discover that our formulae give tures hopelessly too small in the case of *The average of the coISCl.ipts for the 1st Infanferie Brigade , which ] udes Mnnich , was only 166 centims. The average of the Baden conscripts ccntims .
CONTRIBUTIONS TO THE THEOBY OF LVOLUTION .
giants , and too large in the case of dwarfs .
This defect of the theory is the more serious in that no prehistoric bones at present discovered give us diCationS of a race wit } ) giant proportions , there are such bones which indicate the ence of varf races in neolithic Europe .
The reconstruction of individual the skeletons preserved is also of some interest , although , from the standpoint of evolution , it , so far , has nothing like the importance of the reconstruction of the aces .
Tf our formulae do not apply to giants and dwarfs , we are forced to one or other of the following conclusions :\mdash ; Dwarf and giant races must have been differentiated normal races by a sele jtion which has partially or totally changed the regression formulae .
The regression formulae are not really linear ; they are only apparently linear , because , in dealing with the normal range of stature , we have only to consider a small port , ion of the regression curve which is sensibly straight .
Both these conclusions may of course be partially true .
In order to consider the validity of one or both of these hypotheses , it seen ] all we have to do is to investigate the relation between the long bones and stature in the case of a sufficient number of giants and dwarfs .
But alas !
the total material is small , and quality of it is exceptionally bad .
The majority of giants and dwarfs probably prefer a quiet life and a normal burial , so that heir bones do not reach the museum Of the dwarfs and giants whose skeletons are to be found in museu , the majority earned their livelihood by exbibition , accordingly their living stature a character likely to be underor over-estimated for the purposes of advertisement .
If we put aside all records of the living stature , we are thrown back on the measurement of the length as corpse , or on estimates formed by anatomists of the stature from the articulated skeleton .
Unfortunately , ities d very widely as to the betwee the skeleton ( after mounting ) and the corpse \mdash ; ORFILA makes a difference of , BRIANT and HAUD of 8 centims. , and TOPINARD of centims.\mdash ; and ( b ) on the difference between t eiving stature and the corpse length ( see p. 191 ) .
Even if , based upon 23 normal subjects measured as corpse and skeleton , be correct , it could hardly be safely extended to the cases of giants and dwarfs .
Professor CUNNINGHAM , in attempting to reconstruct the stature of the Irish Giant , MAGRATH , goes sc as to discard all records of living stature , and all attempts to reconstruct stature from the articulated skeleton , and would estimate only from the length of the femur .
But this method seems to me fatal , at any for our present purpose , the very object of which is to between stature and femur ( or any other long bone ) in the case of giants .
It cannot boo * In the investigation for conscripbs Bavaria , in 1875 , 43 found , and utong tho 35 measured we a range of 115 centims. to 139 centims. also four giants , or mou with sbatures of 190 centims. and over .
'Royal Academy ions , ' vol. , 1891 , often be repeated that the idea that there is in any sense a constant proportion between stature and any long bone is .
MANOUVRIER makes this ratio decrease from dwarf to giant , and this is correct so long as we suppose the regression formula linear , for example , .
But this ratio really begins to decrease gain as we go from short people to actual dwarfs , and to increase again as we go from tall people to actual giants .
For example , we have the .
results for the ratios of long bones and stature :\mdash ; Data .
.50 normal Frenchmen .
stature \ldquo ; ' , , stature , 12 cases , stature .
It will be at once obvious that ANOU V " " Coefficients moyens ultimes\ldquo ; by no means ultimate , but that in the case of giants the coefficients actually tend to return to their values for the mean population .
This will be cient to that it is .
quite impossible to consider any method of determining stature from a presumed consbant ratio to femur as satisfactory .
But this table shows an important principle , namely , as the ratio of stature to long bone first decreases as the bone increases and then begins to increase , it is impossible to consider the regression curve as a straight line when we extend it so far as the region of dwarfs and giants .
Now this is , a priori , what might have been expected , for all distributions of zoometric frequency that I have come across seem to possess sensible skewness , and in skew correlation the regression curve is not a straight line .
Its actual form is of a somewhat complicated nature , it would be purely idle to attempt to determine the constants of it from the data for dwarfs and giants which are at present available .
Accordingly it seemed to me desirable to select some empirical curve which would , so as possible , represent the available material and give results in harmony with certain general principles .
The considerations which led me to the choice of this curve were of the following character:\mdash ; It musb sensibly coincide with the line of regression already found between statures of 155 centims. to 175 centims. It must accordingly have a point of inflexion at the mean ture , at which the should be the already determined line of regression .
Referred to this tangent and its perpendicular , the form of the curve in the neighbourhood of the origin must be .
Away from the origin , may become sensible function of and , one or both .
hope to remrn to this point in a paper on skew colrelation .
CONTRIBUTIONS TO THE OF So far as the data at my command went , the dwarfs and giants appeared to deviate from the regression line in a remarkably sylmetrical manner on opposite sides of it .
In other words , the branches of ) curve on sides of the axis of appeared to be centrally symmetrical or congrncltt .
Thus the form of the curve was reduced to .
It follows from this that the asymptotes of the curve , besides , will be given by .
The problem then turns on what are the probable asymptotes .
Now if we examine the regression formula for an A an organ , it is of the form : where and are the mean organs , and the standard deviations , and the coefficient of correlation .
Now no amount of selection of either A or or any other organs , as to size , would influence in the case of normal correlation but it would change the constant term .
Hence , if we were to take the line of regression for an extreme population of dwarfs alone , or of giants alone , it would seem quite possible that might have remained while the term , changed , But these lines of regression would be the asymptotes of the required curve .
It was thus suggested to me that the asymptotes might be parallel to the line regression of the normal population .
On examining the points corl e sponding to giant and dwarf statures plotted to long bones , this hypothesis seemed to be highly probable .
Accordingly the form of the curve finally selected to represent the extended cuyve of regression was , where the axis of is the linear line of regression ' normal stature , and the axis of is the perpendicular to it through the mean stature of the French .
* A diagram was now formed by plotting to half life-size ( centim. for 1 centim. ) the points representing giants and , and the lines of regression for the population were drawn .
The and for the point for each giant for each bone were then read off , and these formed the data from which the constants of the four curves of the above type then determined .
For this determination only giants over 200 centims. were selected .
The class of what may be termed sub-giants , with statures from 180-200 centims. , vwele put on oneside .
Such individua]s , termed appear in both the Bonn and Munich anthropological catalogues , but the " " ' there given can hardly represent the living stature ; it is very probably skeleton *Some shifting of the origin would probably have improved my resuIts , but the were not sufficient to justify such extra labour .
and considerably under the real stature .
A height , for example , of 185 centims. , 6 feet 2 inches , say , would hardly entitle a man , in England at any rate , to rank as giant .
In the next place , no notice whatever was taken of the dwarfs .
I felt that , if the curves were determined the giant a only , the test that they gave good resuIts for would be the most satisfactory one able .
As it is , I have been able , on the of the long-bone stature relations for giants , to predict the stature of dwarfs to within centims. average error .
MANOUVRIER'S coefficients moyens ultimes\ldquo ; give a mean error for these dwarfs of centims. , or times as great .
The fitting of the curves was c.onducted in the following manner .
Remembering that the curve gives the value of the mean stature for the whole series of longbones of one size , i.e. , the mean of the array of statures for a long bone of given type or , I recogn ised that the curve , and accordingly its asymptote , must pass fairly centrally the group of plotted points .
An approximate of the asymptote constant was accordingly selected , and the value of calculated from the mean of bhe observational values of and .
If this form of the curve gave , as it gellerally did , not very satisfactory results , was modified , and the new .
In this manner , for example , three approximations were made in the case of the radius .
The method of least squares was not readily applicable to the data ( which were best not very trustworthy ) , for it involves the calculation of such expressions ss and hich , owing to the large values of involved , give far too great importance to largcst giants .
The curves imately determined were }ollowing : For the femur : .
For the : For the hnmerus : .
For the radius : .
Here the unit for both and is equal to two centims. of or of longbone .
Thus the distances 7 , and centims. of the asymptotes from the lines of regression of the normal population are really nces of 14 , , 9 and 9 centims. in actual stature or -bone length .
* The mathematical reader will bear in mind that it is only the " " snake\ldquo ; and not the other branches of the quintic curve which we require .
CONTRIBUTIONS TO THE THEORY OF EVOLUTION .
I do not suggest for a moment that these curves give a fnal solution of the of determining the stature of any individual in the range of 90 to 250 centims. from the lengths of his : long bones , but they to me to give the best results obtainable with the data at present Reduced to a nula a curve type would be of little service , .
both and are linear functions of the probable ture and the observed length of the bone .
Hence we should have a tic equation to find the probable stature from the long bone .
But if these curves be plotted once for all , we have a graphical means of at once , by simply running the eye along a line , probable stature corresponding to any given lenoth of long bone .
With care we can find the stature to 5 centim. , but as , rule to the nearest centimetre is sufficient .
As the lines of regression the normal population are given as part of our curves , it is clear that the diagrams attached to this memoir ( Plates 3 , 4 ) will also serve .
the determination to a like of exactitude of the of individuals or races the ordinary range of statures .
In view of the fact that the diagrams serve all practical purposes , I have not considered it needful to deduce from the above quintics numerical approximations for the value of the stature in terms of the lengths of the various long bones .
( 22 .
) If the reader will examine the diagrarns , he will see the twelve giants , . .
marked by small dots ; from these the curves were determined , and he notice that they strike fairly the groups .
The triplet contains three pseudo-giants , or sub-giants ; these as well as the dwarfs , , were not used in the determinction of the curves .
One rkable feature of the curves must be noted , namely , that in the region of what may be termed sub-giants and super-dwarfs , namely , from about 180 to 200 centims. and 150 to 130 centims. , a very small in the long bone makes a remarkable ] in stature .
This is specially noteworthy in the case of the radius .
Thus bebween normal individuals on the one hand and giants or dwarfs on the other , there appeal .
S to be what may be termed a region of instability , in which an insignificant change in long bone may throw individual across a considerable range of stature .
The points of flexion of our curves\mdash ; other than those at the origin\mdash ; may accordingly have a biological as well as a purely mathematical interest .
The following are all the data which have been able to collect for giants and dwarfs having any degree of probable truth .
VOL. CXCII.\mdash ; A. 2 CONTRIBUTIONS TO THE OF EVOLUTION .
nd are all taken from the very valuable memoir by K. : " " Wachstum des menschlichen Skeletes mit auf den Riesen 'Denkschriften der .
Akademie der enschaften , Math. Naturwiss .
se , ' vol. , Wien , 1872 , I pp. 1-105 .
and are here distinctly stated to be the maximum lengths , and appears to be measured without The eights are apparently those of the articulated skeletons .
L. Thi is the only giant I have ventured to retain out of Sir GEOReE HUMPHRY'S list in " " The Human Skelletol Cambridge , 1858 , p. 107 , for he indicates he measured it himself ( p. 105 ) .
I have not been able to identify his " " Giant\ldquo ; at Bonn .
His Berlin giants differ considerably from those in the Berlin , while his estimates of O'BYRNE and of the giant seem hopelessly too large .
As he gives the Musee Orfila giant 17 centims. less stature than TOPINARD , think his estimate on this occasion more probable .
and are taken the Anthropological Catalogues of the Museums at Bonn and Munich .
I not clear as to what is meant by Korperlange in these cases .
The statures are curiously small as compared with the long bones , if Korperlcinge is to ?
) thus interpreted .
Possibly it is the length of the mounted skeleton without disks .
and U. The details of these dwarfs I have taken from PALTAUF'S work : ' Ueber den Zwergwuchs in anatomischer und irztJicher Beziehung , ' WieIJ , 1891 .
This book compares unfavourably with the careful memoir of LANGER .
The measurements of the long bones of MIKOLA .
are several times over , on each occasion with different values ; the exact nature of the measurements made is not stated , and results such as those on the author 's p. 92 , depending on the most elementary arithmetic , are erroneousl given .
I taken the values which seem to give the most self-consistent results , but it is impossible to feel sure of their absolute accuracy .
SCHAAFFHAUSEN'S account of his dwarf in the ' Berichte der Niederrhein .
fur Naturkunde in Bonn , ' vols .
25 and 39 , and HIS'S account of his dwarf in ' Virchow 's , p. 104 .
All the giants and dwarfs in the above list were adults ; the ages of the four dwarfs at death were , 49 years ; , 61 years ; , 58 years ; and V , 23 years . !
: The table gives the reconstructed statures of these dwarfs as obtained from my diagram and from MANOUyRIER'S " " Coefficients moyens ultimes I have not thought it necessary to publish in the latter case the estimate from each individual bone , but have simply printed the mean of the four results and the differences from the upposed actual stature .
It will be noticed that MANOUVRIER'S estimate is in every case too small .
Of my differences , 2 are zero , are positive , and 11 negative , but the negative differences sensibly larger than the itive , so that my curves rather under than over corrected for giant and dwarf stature .
* ' ' Aus der Mitte der lateralen Condylusflache in die Incisura fibularis I have verified the dimensions given for HIS'S dwarf from ' Virchow 's Archiv fur Pathologie Anatomy , ' vol. ) , 1861 , p. 104 , et seq. CONTRIBUTIONS TO THE THEORy OF VOLUTION .
The truth of this theory can only be discussed by anatomists , and many anatomists like CUNNINGHAM and Dr. PALTAUF hold that giants and dwarfs are creations\mdash ; they are the of to which they would the name of a disease .
Such a view would any conception\mdash ; especially in the case of among the normal an atavistic fluence .
existence even to-day of dwarf races in both and Asia ought , however , to give ground for pause .
When we add to this that Professor SERGI actually considers that he has good evidence of a racial type extant in Italy , and that Professor KOLLMANN , after examining SERGI'S cranial and other eyidence , has been converted from strong disbelief to belief , when we note the three dwarfs ( stature centims. ) actually brought to light by annual conscription in Bavaria alone , and finally when we consider the neolithic dwarf skeletons discovered by NUESCH , we must undoubtedly hesitate to attribute to pathological causes cases of which come under notice .
The Afiican , , and Iian dwarfs appear as a distinct racial type as little variations of normal nIan , as a monkey of the anthropomorphous apes .
It is thus possible that the characters found in so many varfs may be the result of conflict between atavistic and normal tendencies , rather than themselves the source of dwarfdom .
At any rate , while admitbing that our curves are largely based on admittedly pathological instances cf both giants and dwarfs , it seems well worth while to consider to what results they lead us when we endeavour to reconstruct the stature of dwarf laces .
In making this application we have to bear two points in mind we must expect a wide range in our prediction of statures lying between 130 and 150 centims. , for this is the ran ge for which our curves give very unstable results .
We can only hope for a .
degree of approximation in the means .
( ii .
) Ou } curves are constructed solely from male data , because female data are practicaHy -extant .
We must accordingly endeavour to find some means of passing from male to female stature .
To this we must first devote our attention .
( b ) I take the following data for sexual ratios for the French and Aino from the material of ROLLET and KOGANEI ; for the Naqada race from Dr. WARREN'S memoir , and for the Andamanese from Sir W. H. FLOWER'S memoir , which is discussed below .
SEXUAL ratio , NfANN iu NUESCH , , p. 238 .
Ibid. Now , there appears from this table to be a very clear rule , namely , that the sexual ratio for stature is certainly not sensibly larger than the least sexual ratio for the long bones .
It would seem accordingly improbable the sexual ratio for the Andamftnese can exceed .
If we compare this result with MAN'S ements on male and 41 female Andamanese of which the statures were : , mean cenLims .
; female , mean centims. , we find , a value that of the Aino .
Sir W. H. own estimated stature give a sexual ratio of ; the fundamental formulae for a normal population ( p. 196 of this paper ) give ; MANOUVRIER'S " " Coefficients moyens ultimes\ldquo ; give , and by applying the ratios of stature to long bones as obtained from the average French population we find .
The mean of all these results is .
For the Laps fANTEGAZZA found male and female , or the sexual ratio For the Negritos del Monte , or the of Luzon in the Philippines , MARCHE and MONTANO give male centims. and female centims. , from which we find the sexual ratio of ToPlNABD gives for races under 150 tims .
a mean difference of 4 per cent. between male and female which corresponds to a sexual ratio of .
FPJTSCH found a mean difference between male and female Bushmen of 4 centims. which gives ( male ] centims. ) a sexual ratio of ; while PARRY'S observations on the naux appear to give a sexual ratio of SUTHERLAND'S .
From all this it is clear that the dwarfs have a very small sexual ratio for stature as compared with the normal population .
At first sight it might seern best to assume this sexual latio for dwarf races to be TOPINARD'S average of , but as we are going to apply our chart in connection with the sexual ratios found for the bones of the Andamanese in the table above , I doubt whether it ought to be taken greater than 1035 , say in agreement with the value obtained from FLOWER'S estimates .
Accordingly I formuiate the ] owing rule for ascertaining from the chart the probable stature of a female of dwarf race : Reduce the female long bones to male long bones by multiplying their lengths by in the case of femur and tibia , by in the case of the humerus and in the case of the radius .
Find the corresponding male statures from the chart and multiply it by 9662 reciprocal ; these the probable values of the stature as estimated from the several long bones , and their mean may be taken as the best result available .
( c ) It seems desirable to compare the results thus obtained for male and female of dwarf races with their statures otherwise estimated .
If we form a similar to that on p. 222 , but for the case of dwarfs , we have\mdash ; , bhe values given , ' Journal of Anthropological Institute , ' vol. 14 , CONTRIBUTIONS TO THE THEORY OF EVOLUTION .
Now the tendency here is clearly for the ratios to increase with decrease of stature , if we corsider only French .
Aino and the group of four dwarfs .
TOPINARD'S measurements show , however , rather a tendency in the ratios to return to their values for the mean of the normal French population , and as this was closely akin to what we found in the case of giants , we cannot aflord to disregard it in the case of dwarfs .
Sir W. H. FLOWER has reconstructed Andamese frnm their femora on this supposition , and it does not give by any means obable values of the stature .
We only to look , however , at the line of regression for the normal population to see that for statures veen 1 and 175 this hypothesis will give results , but it is conceivable that for statures above and below these limits the ratios of stature 10 the long bones obtained for the means of a normal population give results which are closer to the truth than those found the lines of regression .
on Plates 1 , dotted lines give these atios ) stature to long bones , the of and can be at once read off on this .
It will be seen thab these lines do } give such good results for the four dwarfs under 120 centims. as our curves , but possibly they may give better results for normal dwarf races from 140 to 150 centims. At any rate they do not on the surface exbibit the difficulty as to " " instability\ldquo ; to which I have previously referred .
Sir W. H. FLOWER writes of the Akka skeletons that : " " They conform in the relative icns of ) head , trunk , and , not to dwarfs , but to full-sized people of other races * The chief and great difficulty , howeyer , of adopting these lines of normal stature ratios to determine the stature of dwarf races is to fix a limit to their application .
what point are we to fall back on the ormal line of regression ?
There must be such a point , for } ) line gives excellent results for statures from 155 to 175 centims. Whereyer we do fall back upon it there will arise the sort of instability which we find in our curves , only it will be a far more arbitrary and sudden change .
For this reason I cannot consider it satisfactory to obtain the stature of of less centims. by a process which is not in any sense continuous with that used * JourIlal of the Anthropological Insbitute , ' vol. lb , p. 90 .
By " " dwarf ' in the sentencc cibod 1 think we are to nnderstand " " \ldquo ; for races of more than 155 centims. stature .
The position and character of the instability is undefined and appears to be quite arbitrary .
At the same time , I give the stature of the dwarf races with which I have dealt below on this hypothesis .
In order to apply it , I add the additional data for the female stature and long bone ratios required for and MANOUVRIER'S method , putting in the Aino for comparison:\mdash ; The reader remember that coefficients are for corpse stature and of bones when the latter contain animal matter .
Hence he first adds 2 millims. to the length of the dead bone to get the bone with animal matter , and then 2 centims. are cted by him from the corpse get the living stature .
In the case of the femur , however , he works with bone in oblique position , or with a length about millims. less in the normal individual than the maximum length .
This probably does not amount to more than 2 millims. in the case of dwarf races .
Hence , when the femur of the dwarf is giveu by its maximum length , we need not add or subtract anytbing before multiplying by the staturefemur coefficient .
We have accordin the following methods of estimating the stature of dwarf races from their long bones.\mdash ; The lines of regression for a normal population , i. e. , the formulae of p. 196 of this paper , or the heavy straight lines of our charts .
As we have already seen , this overestimates the stature of dwarfs as it underestimates that of giants .
( ii .
) The curves of yression given by the empirical formulae of p. 224 or by the heavy curves of our charts .
In the case of female dwarfs the lengths of their long bones must first be reduced to male equivalents by the rule on p. 230 , and the statures found reconverted to their female equivalents .
( iii .
) " " Coefficients moyens ultimes\ldquo ; ofRIER may be used .
These arc given on pp. 231 and 232 .
Special attention must be paid to the reductions ( discussed above ) of bones and corpse lengtb .
( iv .
) The stature arJd long bone ratios for the normal population may be used .
The values of these ratios are given on pp. and 232 , but for most practical purposes it suffices I , o use the dotted lines of the chart .
I shall refer to methods as , and .
In the latter case , not will be noticed how close these are to the male cocfficients on p. 231 , except in the case of the radius .
bonc very primive and dwarf races .
CONTRIBUTIONS TO THE THEORY OF EVOLUTION .
because Sir W. H. ] !
was the first use a ratio of stature and long bone for the mean population for the reconstruction of stature , but because he has emphasised the fact that , for dwarf races , it does appear to give fairly good results .
( 24 .
) My material is very sparse .
Sir GEORGp HUMPHRy , in his work on " " The Human Skeleton gives ( p. 106 ) the mean long-bone lengths for three presumably male Bushmen .
I find : \mdash ; ESTIMATED Stature of Bushmen .
Now it is clear that neither the chart , nor MANOUVRIER'S " " Coefficients moyens \ldquo ; ( M ) , make in this case much alteration on the estimate given by my normal regression formula ( k ) for all four long bones .
But the value given by Fl is 6 centims. less .
Sir GEORGE HUMPHRY gives the average stature of these three Bushmen as centims. He does not , however , state where his data are taken from .
Curiously enough , his stature coincides exactly with the value TOPINARD says BARROW has assigned to the Bushmen .
I cannot think that this was the stature in life of the individuals whose bones are averaged by HUMPHRY .
gives the average stature of six Bushmen he measured as 144 centims. , and I should hesitate to place the mean stature of the above three below 145 centims. to 150 centims. At has been used by ORFILA , .
GEORGE HUMPHRY , and others , we have seen , quito incorrect results for races from 155 centims. in stature .
TOPINARD , ' Anthropologie generale , ' p. 461 .
VOL. CXCIL \mdash ; A. 2 the same time ib must be remembered that the stature falls within the range within which our chart shows that a very sligbt change in the long bones makes a great difference in stature .
In case the reader should be inclined to put too great faith in Fl , I would draw attention to the fact that it underestimates by slightly over 5 centims. the known stature of the fairly short Aino race , while or give it almost ccurately and fairly closely .
The other Bushmen I have been able to ) are a male and two females in the Royal College of Surgeon 's .
Selecting the right members as those for which our formulae and curves are deduced , we have : Male , ) centims. , centims. , centims. , centims. Female 1 , , , , , , , 2 , , , , , : , \ldquo ; The following table gives the estimated statures The estimates based on the skeleton height of these three Bushmen are : Male , female , and female centims. The mean error made by is , by , and by centims. But it must be noticed that the last gives in one instance less than the estimated from the skeleton\mdash ; a result which is in itself very improbable .
A consideration of the.values here given seems to show that with the mean length of bones given by HUMPHRY the mean stature couId not possibly have been the centims. he states .
For whatever estimate we take of the Female 1 , she must have been wibh bones no longer , at least 10 centims. taller CONTRIBUTIONS TO THEORY OF VOLUTION .
HUMPHRY'S mere ] male .
Taking our fonr nla , les and two we get estimated statures for male female Bushmen of bout 1 and 142 centims. , which I expect are very far from the truth .
( 25 .
) a paper by Sir W. H. FLOWER in the ' of the Anthropological Institute , ' vol. 18 , 1889 , entitled : " " Description of two ] etons of Akkas , a Pygmy from Central \ldquo ; the following data are given ( p. 14 ) : F. H. .
R. Male centims , centiuls .
centims. centims. Female , , , , , , , , In the following table the reconstructed statures are given on the same four hyputheses as we have considered in the case of Bushmen . .
W. H. FLOWER estimates the height ) both bout 4 feet , or centims. He gives as the estimate of stature the female skeleton .
We could hardly want better than are given by .
Fl ives also good results , while appears to err in excess .
* refers to an Akka voman of 136 contims .
statul.o , who must therefore have been cousiderably taller than the above woman .
PROFESSOR KARL PEARSON , MATHEMATICAL ( 26 .
) Stature .
The stature of the Andamanese is a peculiarly difficult one to estimate .
They are taller than Bushmen and Akkas , and fall more markedly into the unstable range of our chart curves .
The measurements of a very considerable number of long bones have been given by Sir W. H. FLOWER in two papers in the of the Anthropological Institute , ' vol. 9 , ls79 , and vol. 14 , 1885 .
I take the following mean values from the latter paper Male . . .
emale Constructing as in the previous cases a table of stature as estimated ] four methods we find : Now it will be observed that and give sensibly the same result : 154 centims. for the male ; that for the female , , owing to our having first to increase the female bones to reduce them to male lengths , gives a higher result than , for we have got into the unstable ran ge of the curves , and the stature-reducing factor afterwards applied does rot undo the excess .
There is not much , therefore , to choose between , and for the Andamanese .
They give results 4 centims. greater CONTRIBUTIONS TO THE THEORY OF EVOLUTION .
than Fl in the case of males , and 3 centims. greater in the case of females .
From them we should conclude that the stature of was given by male felnale .
IAN , measured 48 male and 41 female living nanese , gives the stature as , , and female centims. Sir W. H. FLOWER esti n the ature from his skeletons at male centims. , and female centims. This is mnch less than MAN'S determination of the living stature .
MANTEGAZZA , who possesses a skeleto ] of an Andamanese , gives its height at centims. , and KOLLMANN considers its living stature to have been 150 centims. The femur in this case is centim long , which would correspond in a normal Frenchnlan to a stature of 161 centims. I must state that I feel inclined to put entirely on one side estimates of stature based on the height of the articulated or unarticulated skeleton , they appear invariably to underrate the lliving stature , and often by very large amounts .
Even if we suppose the AndamaIlese to have the relative proportions of full-sized people ( e.g. , use Fl ) , we obtain statures considerably above Sir W. H. FLOWER'S estimates .
On the other hand MAN'S measurements , which give results much in excess of the latter , fall considerably short of the results we obtain from M. They even fall short of Fl , and in the case of females markedly short of ib. If we consider that FLOWER'S skeletons and MAN'S individuals belong to the same roup , then it must be confessed that our estimates are unsatisfactory .
The hypothesis Fl gives the least divergent result , but it cannot be considered a particnlarly good one .
It will be seen at once that it is the inferior members in each limb which give the stature estimates .
If we confined our attention to femur and humerus , then and ( b ) would give for males and for felnales , results better in accordance with MAN'S measurements than Fl for all four bones , or than Fl male femur and humerus only .
When we consider the immense importance of these dwarf races for the problem of evolution , the main result of our investigation is obvious ; there ought to be an borate investigation\mdash ; such KOGANEI has made for the Aino\mdash ; on the long bones of skeletons and the stature of living individuals , of some extant dwarf race .
These races are rapidly becoming extinct , and the possibility of making such a investigation is yearly diminishing .
Yet it is only by a careful comparison of the regression formulae for and normal races that it seems ) me ) ossible that we shall be able quantitatively , and therefore definitively , to fix the ) of dwarf and normal races in the course of evolution .
* See Sir .
H. on " " Pygmy Races , " " ' Journ. of Anthropological Institutc , ' vol. 18 , 1889 , p. 73 .
NUESCH , , p. 129 .
ader must bear miud that nearly all tho ucutSb ivulved in , to ( 27 .
) Neolithic In the recelJtly published work by NUESCH , 'Die prahistorische Niederlassung beim ' 1896 , is a memoir by KOLLMANN , entitled , " " Die menschlichen Skelete , besonders uber die fossilen menschlichen Zwerge This publica ion for the first time showed us that there existed in neolithic Europe , alongside a normal with a stature of about 163 centims. , dwarf race , very similar to the pygmy of which we still find traces extant in Africa and Asia .
At any rate the discovery in the same group of graves of four skeletons , or rather fragments of skeletons which must have belonged to who were pygmies , and not " " pathological ' dwarfs , points very strongly in this direction .
KOLLMANN , who gives a most interesting discussion of these neolithic pygmies .
the measurements .
H. 1 .
Female centims. .
2 .
, , or male .
, , . .
T. R. . . . .
3 . . . .
35 , , centims. centims. 4 .
, , ormale .
, , , , , , Of these : 1 , female , is an adult ; 2 , female or male , is that of a young person 16 to 18 years old , and , according to , probably , but not certainly , female ; 3 , female , and 4 , female or male , are dults , but as we see the sex of the latter appears donbtful .
Proceeding , as in the earlier cases , we find : sbruct stature , arises from the fact that the regression coefficients for long bones and stature are known for only , and that we have nothing else to go upon .
Had we endeavoured to reconstruct ono long bone from a sccond , wo should have had far more exact material to determine the differential ovolulion of local races .
If include the non-adult and suppose the whole series female , we have : centims. centims. Without the non-adult , we have : centims. centims. The two possible males give : centilns .
centims. The adult male gives : centims. centims. M. Fl .
centims. centims. M. Fl .
centims. entims .
M. centims. centims. M. Fl .
centims. centims. The single nlale here is about identical with the means obtained by the different methods on p. 236 for the male Andamanese , and the adult felnales give a result somewhat less than that of the female Andamanese as reconstructed from their long bones , but in close accordance with MAN'S measurements of living Andamanese stature .
The dimensions are somewhat larger than those of Bushmen , or Akkas , or Negritos .
We seeln , therelore , justified in assuming a neolithic pygmy race in Europe having a stature about the same as that of the Andamanese .
Whether the .utual stature of this race was for the female nearer to 144 centims. or 141 centims. ( F1 ) it seems to me impossible to ascertain definitely until we have more trustworthy and extensive measurements than yet exist of the living stature of extant pygmy races .
( 28 .
) Conclusion .
The formulae and curves for the reconstruction of stature which are given in this memoir , must by no means be taken as final .
No scientific investigation be final ; it merely represents the most probable conclusions which can be drawn from the data at the disposal of the writer .
A wider ran ge of facts , or more refined analysis , experiment , and observation will always lead to new formulae and new theories .
This is the essence of scientific progress .
] , therefore , which is claimed for this paper is that it exhibits a better theory of the reconstruction of stature than any whicI ) has so far existed\mdash ; it might not be too much to say that nothing which can be called a theory has hitherto existed ; ( ii .
) that it determines the constants of the formulae given by that theory as well as the existing data allow of ; ( iii .
) that it gives values for the probable statures of istoric races , which have far less divergence among themselves , whatever be the bone or combination of bones CONTRIBUTIONS TO THE THEORY EVOLDTION .
used , than those suggested by previous investigators ; and lastly ( iv .
) that it indicates what additional data ought to be sought for , and to some extent what is the inner meaning of divergent results , for the great problem of racial differentiation by natural selection .
* Of the general conclusions reached by the author , perhaps two deserve restating and emphasising here .
In the first place , although there were individual tall men among the neolithic populations , whose bones have so far been unearthed , yet neolithic man as a whole was short .
Of course , it is possible that a tall neolithic type , i. e. , one with a stature greater than 168 centims. , may yeb be discovered\mdash ; witness the discovery within the last two years of a neolithic dwarf .
But failing its appearance , the question arises , where and how did the tall Anglo-Saxon and Scandinavian develop ?
what extent is this tallness racial , to what extent due to environment ?
The apparently greater stature of British over Continental neolithic man deserves special consideration from anthropologists .
Secondly , granting that the modern populations in the same district are taller than the neolithic populations , there still ppears in both France and Southern Germany some regression of the modern stature on that of the ancient FraIlks , Bajuvars , and Allelnans .
I difler from both RAHON and LEHMANN-NlTSCHE in that the difference is too great to be accounted for as a process of natural selection applied to the long bones .
AHON has made a slip in his arithmetic , and LEHMANN-NITSCHE compares the Row Grave population with the most favourable element of Munich town recruits .
If the divergence could be accounted for by selection applied to the bones , why is not a similar divergence to be found in the case of Anglo-Saxons and modern English ?
I tbink an explanation must be sought elsewhere .
One suggestion is , that as the physical struggle lor existence has been lessened , reproductive selection has had more play , and the greater fertility of an older pre-Germanic element in the populations of both Southern Germany France led to a return of stature to its more ancient value .
In the case of Anglo-Saxons and Scandinavians in England there was very probably a more complete destruction of the earlier populations .
Whatever may be the real reason for this apparent degeneration , it seems most desirable that there should be a systematic measurement of long bones dug up anywhere in our own country , this whether they belong to prehistoric or historic times .
Stature is quite as marked a racial character as cephalic index , or any other skull measurement , and its high correlation with the long bones admits even in the present state of our data of its reconstruction with very considerable accuracy , if only a sufficient representation , say twenty to forty .
bones , of an ancient population has been measured .
It is only by the gradual accumulation of such data we can * The influence of directed as distinguished from random selection on size , variation , correlation , and sion has been theoretically developed in a not yet .
Having ) fully discussed in my college lectures of this ssion , much of the ecent work of my departmenI , like the present memoir , touohes on it .
VOL. CXCII.\mdash ; A. 2 I hope for light on the manner in which our own populatioa has developed and is developing .
* ( 29 .
) The following table some of the numerical results reached , and further includes , for the purposes of comparison , the stature of certain modern aces as given by various authorities .
No stress whatever is Iaid on the latter values , which have often been determined by doubtful observers from very small series .
They are merely given here in order to show the.qenercrl position of the reconstructed races in the order of racial statures .
TABLE of Stature and Sexual Ratio for Divers Races .
* For example , no one can say at present what was the stature of Englishmen from .
1000 to 1700 , and yet 1 collections of bones exist , which would suffice to answer this problem .
, for examl ) , considers sex ttio for 73 series in " " Etude snr la ille consideree .
scxc . .
et les races ' Rcvne d'Anbhrol ) olgie , ' 1876 , p. 34 , but he merely gives llleans ' grouped results and does not tell us the details for the individual series .

