]\gt ; Data for the Problem of in Data for the Problem of Evol tion in .
On the Correlation between Duration of and the Number of By Miss M. BEETON , G. U. YULE , and PEARSON , , University College , London .
Received Apri119 , \mdash ; Read June 14 , 1900 .
1 .
According to the Darwininn theory of evolution the members of a communily less to their environment are removed by deatL But this process of natural selection would not permanently modify a race , if the members thus removed were able before death to propagate their species in average numbers .
It then becomes an important question to ascertain how far ation of life is related to fertility .
In the case of many insects death can interfere only with their single chance of offspring ; they live or not for their one breeding only.* A similar statement holds good with regard to annual and biennial plants .
In such cases there might still be a correlation betwesn duration of life and fertility , but it would be of the indirect character , which we actually find in the case of men and women living beyond sixty years of age\mdash ; a long life means better physique , and bstter pbysique increased fertility .
On the other hand , there is a direct correlation of fertility and duration oi life in the case of those animals which generally survive a number of breeding seasons , and it is this correlation which we had at first in view when investigating the influence of duration of life on fertility in man .
The discovery of the indirect factor in the correlation referred to above was therefore a point of much interest .
For it seems to show that the physique fittest to survive is really the physique which is in itself ( and independently of the duration of life ) most fecund .
In continuing our study of the inheritnnce of longevity , it occurred to us that it would be possible at the same time as extracting data duration of life to extract data bearing on the size of the family .
Accordingly Miss M. Beeton , in upon family histories , made records of this additional character .
Meanwhile Mr. G. U. Yule , who had been independently at work on this point , drew my attention again to the matter in connection with a passagc in the ' Grammar of SciencC .
We agreed to lmite our the result is the following * Of course longe ] ' life may denote grcatcr nrc of male or meeting female or male , but this case we not tility , is or is not 011CC fertile .
' Boy .
Soc. Proc vol. 65 , p. 290 .
Second e. N. Filon , , and .
K. sslcr for aid in calculutions and in the of ) VOL. LXVII .
160 lliss M. Beeton , Mr. G. U. Ynle , and Prof. K. Pearson .
The data dealt with this consist of four series , the first three collected and reduced hliss M. Beeton , and the fourth series by .
G. U. Yulc .
The sources from which they were extracted are the following:\mdash ; of I.\mdash ; Taken from the ' mily , of Connecticut , ' a wellknown history of an Quaker family .
In order to complete !
thousand and more entries some few additions were made from the 'Backhonse Family , ' the history of an English north-country Quaker family .
This series ) taken to substantially represent Amcrican women more or less closely connected with one strain of , either by inheritance or ) As soon as these results were it was noticed that the average at death of mothers immensely below the average age at death of nglishwomen .
Fnrther , the maximum frequency of deaths which occurs at to itctnally greater the maximum which occurs ) to !
ither then American women of this class very early , or the wonle of the Whitney family suffer lmder some hereditary taint , , phthisis .
tiom p records .
The data for scries were a riety of histories and records most kindly placed at by Mr. Secretaly of the ociety of Friends , and ) the of the instlrance office , the Friends ' Association , ) of whom we desire to thank for their ftid .
The iect here was to avoid the selection which may ltnconsciously bc made when the data are from the records of single In these two series , as in the third , we selected the records of the Society of Friends because\mdash ; They appear to ) the luost t complete of the mily histories The of the omen givcn ; these are rarely recorded in orksc c The of fertility seems to ) less probable in strongly ions c likc the Frieuds other sscs of the tion .
In this scries the , the , and other conquitc of the of , professing to ontains all imlnense additioLl Dnta for of olntion in ] of Si-e of III.\mdash ; The reat bnlk of the data was extracted from the Whitney Family .
Here ths features loted for the women were observed the men , ) to much less marked degree .
There rather high maximum frequency of death at but not so the maximum the age at death was than we find for the .
On the whole the series is a very good one .
IV.\mdash ; Extracted from 's ' Landed Gentry .
' lt has ) stated eIsewhere thnt this is a class for such .
It higher tverage fertility than the Peerage , a lass in which there is probably comparatively little artificial restriction .
Unfortnnntely it offers no material for the age at of women .
3 .
The are the chicf resnlts ) ined from the of these series : \mdash ; I.\mdash ; Table of Gener In table the for the of the is , the mit of the grouping tho itccomp ' tables .
Thus age of mothcrs ves the frequency of the of mothers dying between .
Of conrse the age of ccrtt in rents would lie ctly 0 the boun of a gronp , but information is very rnrely forthcoming , when it is in fcw forthcoming , , the of ) ) irth a is givcn , it is very impl ) that the of ctly bisccts the year .
Thns no otionising was found in the first three tables .
the ' L$nded ' owing to the tnrc of the reco1d , Ir .
Ynle monIlt of isiug n this ppears in the table for Series .
In the ients : ) is nnit , constant which existcnce of a modal about ] ) noted in solution of the ality ; it is the nlode of the middle ] .
See 'Phil .
runs .
, vol. 186 , p. } , 'Phil .
, vol. 192 , p. 162 Miss M. Beeton , Mr. .
U. , and Prof. K. earson .
the deviation in the age at death from the mean age at death , measured in 5-year umits , must ) multiplied in order to obtain the prol ) able ation of the family from the mean family .
II .
of egression Formulae or Curves .
Size of Family , Duration of Life .
Series I. Am all ves .
Straight line : 1 ' For lives of 50 years and under .
Straight line : ndOrigin o ) ( unit of ? .
one For lives of.50 years and over .
Straight line : ?
For all lives .
Cubical ) .
Origin of unit years : ves .
traight line : 1 of ( b ) ] of 50 years and under .
Straight line : at birth : unit of one ( c ) For lives of 50 years and over .
Straight line : year .
For all lives .
Cubical parabola .
Origin of 2 at ears i$nd ulit years : III .
th For .
Straight line : 1 .
lives of years and under .
line : ( c ) For lives of 50 years and over .
ight line : For itll lives .
of unit : Data for th For lives .
Straight line : 1 .
lives of 50 years and mder .
Straight line : irth a lllit of one of years and over .
Straight : ffi lives .
Cubical of ( years lmit years : .
The of the straight lines for all these series found : once by fitting the best ille to the ) .
, the cocfficient of have been fitted ) the method of The of this system of formulae has beeIl plotted , and is ited graphically in the ttccompanying diagrams ( pp. 176\mdash ; 179 ) .
rhese diagrams suffice to give the ltire g solution of this problem to all ctness sufficient for most tctical pposes .
A exmination of these diagrams will enable the reader to follow otlr sions e more olearly than inspection of the lgebrnic formulae .
4 .
The regression straight liIle f all lives , does give satisfactory pictnre relation etween age at death of rent itnd llmber of oflspring .
see ouce that it is too steep at the ) and not steep the cnd of life .
Accordingly , starting years as the sensible limit earing period , the mothers )rokeIl up into two the regression lines ated separately for lives of 50 url , for lives of ) years over .
this qmte a fit was to the ) tiorlS .
convclticnce , the age of was also as dividing fathers .
the regression line parents living beyon bhows a quite sellsible d from the or it if takc Arllerican mothers there is 110 of this *See , ' Roy .
Soc. Proc vol. GO , p. 477 .
lve shown in a meltloirl:ot yet published ( a ) how to all types 01 es , but ] ) al.ticularly parabolas of any order , by ] ncthod of molnents ; ( b ) that method gives l.esults of the of those the of least .
P. 164 .
U. Yule , and .
K. fertility to of English mothcrs it is less rked , to trne up to 75 years .
Beyond 75 ppenrs to slight docrease .
ning to the } series for fathers see that we might possibly ) etter taken 60 than 50 as iding age , for the trend of the observations is much the same up to 60 ycars .
After this there is still a sensible trend the results , so that aged fathers are again the most fertile .
lVith the English thers this tion is , as in the case of English mothers , far less marked , adthough it is scnsible if we take fathcrs years .
Thus I think we might snm the physique in ) men and women which leads to longcvity is also associated greatsr mdity .
Of two womeJl who both beyond 50 years , the longer is likely to had ) efore 5 the larger The association is , , much greater for American nglish parents , lthough the American parents dealt with are , in the of , of Anglo-Saxon race .
Climate , mode of life , generally selcction and environnlent , seem to be differentiating in this respect the nglish and the -American .
The English Friends , shoudcl snppose , would be class very comparable -ith the AmericaIl Friends , yet their } life is longcr , their fertility greater , and there is less association ) and .
In cases ) show men and omen are more alike , English men and more alike than the women to the women or the men to the men of the two .
This is the more remarkal ) , as the English Friends a class are metnls identical the Landed Gentry .
( ii .
) In order to eprcseIlt the haIJge in the regl.ession , hich cannot ) done by straight lines , which only enal ) ns to distingtlish the non-fecund periods of life , the -erc fitted with rabolas .
The regression at in life the ] ] ) looked upon as the tangent to the cnbiced that .
An inspection of , 4 , 7 , what lent exprcssion for these statistics .
For mothers fathers scc co1lsistently ghout lifc , ve a most excellent denlonst ] ation of the physic cters which tend to longovity ) also ( .
In the lish f the sttme ppcars i ) much less rkod degrec ; there is of inflexion in the posive .
Up to ) , the nnmber of offspring to case ith tion of lifc , hen we off , the got a trcnd .
lVith h mothcrs , } , the has small ) trend trds in old fter the lifc conl ) otcs 1mily , Data for , of with the physique to longevity , eyolld 6 as jndged the parabola , longevity is slightly ourable .
to fecundity .
The following are the valnes of the regresslon .
coefficients obtained by differentiating the Clll ) ical p and referl.lng to birth as and a year as tulit:\mdash ; Table III.\mdash ; Regression their Changc Dnration of Life .
Cubical parabola .
By simply the nnml ) .
of of lifc - , can find the of the regression at any age .
5 .
Illustr of these is nlily of an English mother at 409 gives ) , and , all which might equal have been read off o.n the diagrams .
The actlually number is considerably excess of these , i.e. , .
Ill fact , if an nglish mother to 40 , she will , the very nearly completed her family .
] American vomall ( .
But if she another ten or cnty y shc a mily of 5 or ( ii .
) Compare the strcngth of the tionship 1ctweeu duration of life and size of mily fdying ( espectively .
We find the slopc of the cubical ) the points ding to 40 to be itnd respectively .
The mcan regression for the wholc of lifc is .
the first fifty for the fifty , and rednce to .
It the of lncre n ) of *lt has been suggested that is duc to the nature of the record , there a tendency to enter only the children who survive parcnts .
Thus the the latter live the fewer would be the ing entered .
In other words , we should be under-estimating the tion between fertility and longevity .
] ; ut he Quaker birth-records include all , and { heir is does not appear any is ground and differing so sensiblv .
166 lfiss .
Beeton , .
U. Yule , and Prof. K. Pearson .
compared with the physique hich tends to ' longevity has an effect on tility of about 5 or 6 to 1 .
( iii .
) 1eismann has suggested that it may be an advantage to a species that its duration of life should be shortened .
This is not , a , conrmed for the case of man the American series : the longsr the parcnts live the greater the lnmber of their offspring .
But if we can lay any stress on the bend-in for the nglish mothers , and on the slmilar ] less marked tendency for the English fathers , we might that reproducve selection was possibly in England working against extreme longevity , although favouring parents living till 65 or 70 .
Indeed those who rnsh rapidly to brilliant but not over-stable conclusions might emphasise Weismann 's views by showing how in an old community , with much greater pressnre on the material resources there is a tendency to reduce the fertility of the long-lived parents ; lvhile in a new community , rith plenty of food and occupation for the longest-lived parents arc the most fertile !
However , all that we can safely say is that there is a marked difference veen English and American parents , and that this distinguishing characteristic is almost cqually visible if we take opposite sexes of iuch diverse classes as English Fricnds and English country gentlemen .
wouId leave to further investigations its true interpretation .
6 .
Admitting a snbstantial correlation between length of life and fertility , it is of great interest to investigate what effect , other things being equal , )roductive sclection ould have in modifying the duration of life .
The following table gives the nleall length of life of parents taken singly and of parents weighted their offspring:\mdash ; Table lIean Durntion of Life of Parents in Years .
these are ) dificrcnccs even in the case of the EngIish parents ( II aIldIV ) , but they are very differences in the case of the American parents ( I and III ) .
If we suppose no assortative mating on the })asis of characters tcnding ards longevity , therl it is easy to *omitting , for examplc , the of natural selection as evidenced possibly grcater death-rate in large families , \amp ; c. for Problcm of obtain a rough approximation to the effect of reproductive selection in modifyin the duration of life .
It has been shown that if there ) .
no ssortative mating the average deviation , , of array of ffsprlng from the mean of the general population of offspring due to parents deviating and from the mcallS of the general populations of parents is given by : where , and are coefficients of thd o , the standard deviations in offspring and parents for the character question .
Vhen that character is longevity our data are not yet complete , but two of us have shown the value of , for father and son , i.e. , the regression coefficient for inheritance of tion of life , is about if the sons die having lived at least 21 years .
We not yet completed our for the inheritance of the duration of life ) the case ( i ) of minors , or ( u ) in the case of the female line , although we have nearly reached the requisite amount of material .
Hence the following statements must be taken as tentative and suggestive only .
We will assume to be the regression coefficient for both sexes , and for all ages of ninors or .
In this case if the mean of the umweighted and of the weighted fathers , of the unweighted and of the weighted mothers , we should expect increased tion of life in the offspring to reproductive selection of for the Americalls for the nglish Thus the increased duration of life would be about 2 ycars per generation from the American data , and bout 9 to months per generation from the English data .
The result for the American series shows us how itll especially low tion of life , due possibly in this to some family will be rapidly raised by cve selection , if there be no * ' Phil. 'Roy .
Soc. Proc vol. 65 , p. 297 .
Ihe Landed vould appear to be than the Peerage to our present material .
It is by no means ntit this is the trno view of the case .
We ]lave seen that the American 14 omeu ximnm mortality in earlv middl life , and only a secondary } maximum of the table by J. P. , F.B.S. , for the ycars 1728-57 ' A Collectiou of } Yearly Bills of Mortality from 1657 to 1758 inclusive , ' London , 1759 ) occut .
S about 41 , and there is no of at 70 at all .
Lmerican nppear to resemble London of elJtu ) back .
168 Prof. K. Pearson .
factor of evolution .
Tho ults on the other show us a small ) dency in reproducve selection to prolong lration of life .
three generations to a , we might expect the ation of life to ) raised ) ycars in a centnry this factor of cvolntion .
In making this statcment snpposing that parents are not a short-lived selection ont of tho general aduIt population .
There scems no reason why they ) , and we have some statistics to show they are not .
for the ' ' ' Landed Gentry ' we have that for fathers and sons or 25 years and upwards , the age at death of the father is ) greater than that of the on .
Further , from data for the ociety of Friends , Miss Beeton has , found the average age at of in general to be , and the average ago of mothers death to be , sensibly the same .
In the ) for 1871 to 1880 ivcn 1 ) the Begistrar-Gencral , the expectation of life of women in general at20 years of age is given as years , or the average duration of life is years .
This is only slightly greater than our avcrage for nglish mothcrs bove , , and substantially less than our for mothers hted n their offspring , i.e. , ycars .
Again , the general population of males of 20 had ( 1871-80 returns ) an av erage life of years , which is 1lot comarable with our 'Landed Gentry'h ' sons }ving 2 with an average life of ycars , that of their fathers , years .
We do not think , therefore , that parcntage , in maternity , corresponds shortening in the expectatiojl of life .
Thus reproductive selection appears to increase in the expectation of life .
increased expectation of life is ally considered to have come in to existence during our centnry .
to better sanitary conditions , eater care of the sick iind invalided , \amp ; c. , \amp ; c. Its exact estimation is latter of some difficulty .
) find F. .
P. Neison , working 011 the 's returns ) cfore 1841 , Table expectations of life from 10 yearf , .
For males of 20 and 25 , his durations ; for females of 20 and and .
These not tantially less than the Registrareneral 's returns for to which gives males females and respcctively .
fact , the show reduction .
If we stick to the as given for three differe1tt periods , CSllmal ) more with other have the following * .
Proc vol. ) , 29 Ibe averagc age at death of musl in our case the expcctation of life of 1$ omcn of , for there are few marriages nd we } tables included all of sterile ltnions .
' Contributions to Vitnl , 184 of xpectation of Life .
Here there is an )irth for males , but very small increases between the first last periods at 20 and 25 .
For females there is an immeIlse illcrease at )irthaIld sensible increase the other .
Possibly a good of may ) due to more exact returns for the ages of women being llow ) If we take the earliest Table of the ) ilities of Life , that deduced ) J. P. , F.R.S. , for the ills of mortality in Londoll for to 1757 , and printed in the work cited 011 we find the number of deaths of 1000 persons given for each of life , male and female being combined .
According to the -Gencral 's for 1881-90 of 1000 persollS born , 728 to 20 and 709 to ) , but from J. P. only to 20 to 25 .
This tremendous mortality of infancy , outh probably largely a leCtirc death-rate .
We find accordingly the expcctation of life birth to be only years ; ever , to be to year These results for , llot England in general , but possible nccs for the difference ) city coulltry , they suggcst a most stringent selection .
incrcased expcctatio ] ] of life irth of nything like years in less two llot achieved even at the Americall rate of two is it possible that the hole of the increase the 's returns expectatiolt life at ) irth for the periods lS38-54 and increase of ) .
to to repro- ductive selection , unless we rJOSC the tion ) ctween of of rents to .
On the other ( , if we confined to 25 of sexcs , ghly , of in the expectation of lifc , thd this per could easily ) ched period * We do not know how J. P. 's table dedueed , but got results by ging the 3 by viving at any of th born .
170 eeton , .
G. U. , and Prof. K. Pearson .
question .
Generally , we may that the data are not very for real purposes of comparlsoll , but that there is nothing them opposed to the suggestion that a sensible part of the duration of life of this may be due to the inheritance of longevity and the correlation of longevity with fertility .
Furtheg letermination of the inheritance of duration of life in the case } may help to throw additiollal light on the matter .
7 .
The following method of strating the influence of longevity fel.tility may serve to impress the matter on the reader:\mdash ; In Series I the longer-lived moiety of the mothers produce pet cent. of the children , the shorter-lived moiety per cent. In eries III the moiety of the fathers produce cent. of the children , and the shorter-lived moiety per cent. In Series II the moiety of the mothers produce cent. of the children , and the shorter-lived moiety per cent. In Series IV the longer-lived moiety of the fathers produce cent. of the children , and the shorter-lived moiety per cent. Thus , hile the resuIts are sensible , those for the American parents are markedly so .
In ) American and English statistics the influence of longevity on the fertility of the mother is greater than its influence on the father .
8 .
somewhat videspread view of evolution stops at the survival of the fitter discussing the mode whereby the less fit leave no , or fewer , pring than the fit .
Of course , if the mfit are exterminated before life , there is no chance of their reproducing themselves .
It has ) shown in the second paper of this f , eries that a selective -rate does exist for adults , so that the whole work of selection does not take place before the reproductive stage is reached .
But Miss Beeton 's for the correlation of duration of lifs in the of brethren dying as nlinors seem to show that the selective death-rate for children is rather less than greater than its value for Hence , for the lction or -xtermination of stock uted to its environment , we should to Iook rgely to selection in the dult s. In the present paper we have 1nade what we ) elieve to be the first quantitative determination of how a selective mortality reduces the numbers of the oflspring of the less fit relatively to the fitter .
the case of life under wild conditions , the tion between fertility of surviving probably ] ) for such life it is most impossible to get statistics of this nature ; we hroWIl * The mattel .
is still unde .
investigation , that conclusion is ubject modilication .
Of oourse , the among chiidren may largely remove those not } ] stitution , by physical or method of old into the nonData the of Evolution .
171 : upon measuring the effect in man , and thus obtaining what be considered as a minimum value of the influence under disIn the course of our investigations we have seen that the relationship Mween fertility and duration of life does not cease with the .
We thus reach the important result that characters hich up a constitution fittest to survive are also characters which ourage its fertility .
This result is of great value from the standpoint the differentiation of type , where it is absolutely necessary that the fttest to survive should also be the most fertile.* On the other hand , ye note that duration of life is a character capable of modffication by luctive selection , and we suggest that a considerable of the iecreased expectation of life observed in recent years may be due to this cause .
In the case of the American statistics , we see at once ho replace a remarkably short-lived stock by alonger-lived stock , the klk of the offspring coming from the longer-lived members .
* ' The GrammaI : of Sciencc , ' econd lition , pp. 448\mdash ; 9 .
Tiss M. Beeton , .
U. , and .
K. Pearson . .
Data for } ' of ' Jfan .
174 Miss .
Beeton , .
U. , and Prof. K. Pearson . .
Data for the ( ' of Erolution ) .
I.XYII .
Dafa Problem of in for of

